arXiv Daily Digest - 2026-04-21

Large Audio-Language Models (LALMs) are increasingly integrated into daily applications, yet their generative biases remain underexplored. Existing speech fairness benchmarks rely on synthetic speech and Multiple-Choice Questions (MCQs), both offering a fragmented view of fairness. We propose VIBE, a framework that evaluates generative bias through open-ended tasks such as personalized recommendations, using real-world human recordings. Unlike MCQs, our method allows stereotypical associations to manifest organically without predefined options, making it easily extensible to new tasks. Evaluating 11 state-of-the-art LALMs reveals systematic biases in realistic scenarios. We find that gender cues often trigger larger distributional shifts than accent cues, indicating that current LALMs reproduce social stereotypes.

Despite the rapid progress, LLMs for sequential decision-making (i.e., LLM agents) still struggle to produce diverse outputs. This leads to insufficient exploration, convergence to sub-optimal solutions, and becoming stuck in loops. Such limitations can be problematic in environments that require active exploration to gather information and make decisions. Sampling methods such as temperature scaling introduce token-level randomness but fail to produce enough diversity at the sequence level. We analyze LLM exploration in the classic Multi-Armed Bandit (MAB) setting and the Text Adventure Learning Environment Suite (TALES). We find that current decoding strategies and prompting methods like Chain-of-Thought and Tree-of-Thought are insufficient for robust exploration. To address this, we introduce DORA Explorer (Diversity-Oriented Ranking of Actions), a training-free framework for improving exploration in LLM agents. DORA generates diverse action candidates, scores them using token log-probabilities, and selects actions using a tunable exploration parameter. DORA achieves UCB-competitive performance on MAB and consistent gains across TALES, e.g., improving Qwen2.5-7B's performance from 29.2% to 45.5% in TextWorld. Our project is available at: https://dora-explore.github.io/.

Enterprise AI systems increasingly deploy multiple intelligent agents across mission-critical workflows that must satisfy hard policy constraints, bounded risk exposure, and comprehensive auditability (SOX, HIPAA, GDPR). Existing coordination methods - cooperative MARL, consensus protocols, and centralized planners - optimize expected reward while treating constraints implicitly. This paper introduces CAMCO (Constraint-Aware Multi-Agent Cognitive Orchestration), a runtime coordination layer that models multi-agent decision-making as a constrained optimization problem. CAMCO integrates three mechanisms: (i) a constraint projection engine enforcing policy-feasible actions via convex projection, (ii) adaptive risk-weighted Lagrangian utility shaping, and (iii) an iterative negotiation protocol with provably bounded convergence. Unlike training-time constrained RL, CAMCO operates as deployment-time middleware compatible with any agent architecture, with policy predicates designed for direct integration with production engines such as OPA. Evaluation across three enterprise scenarios - including comparison against a constrained Lagrangian MARL baseline - demonstrates zero policy violations, risk exposure below threshold (mean ratio 0.71), 92-97% utility retention, and mean convergence in 2.4 iterations.

Decoding-free reranking methods that read relevance signals directly from LLM attention weights offer significant latency advantages over autoregressive approaches, yet suffer from attention score homogenization: middle-context documents receive near-identical scores, destroying the fine-grained distinctions required for ranking. We propose HeadRank, a framework that lifts preference optimization from discrete token space into the continuous attention domain through entropy-regularized head selection, hard adjacent-level preference pairs, and a distribution regularizer that jointly sharpen discriminability in the homogenized middle zone. Depth truncation at the deepest selected layer further reduces inference to $\mathcal{O}(1)$ forward passes. Across 14 benchmarks on three Qwen3 scales (0.6B--4B) using only 211 training queries, HeadRank consistently outperforms generative and decoding-free baselines with 100\% formatting success. At 4B, 57.4\% of relevant middle-zone documents reach the top quartile versus 14.2\% for irrelevant ones -- a 43-percentage-point selectivity gap that demonstrates the effectiveness of attention-space preference alignment for listwise reranking.

The rapid expansion of the model context protocol (MCP) ecosystem enables large language model (LLM)-based agents to access a wide range of external tools via a standardized interface. However, identifying appropriate MCP servers for a specific development task remains challenging. Existing studies primarily focus on measuring the MCP ecosystem or optimizing tool invocation mechanisms, while systematic recommendation frameworks and reproducible benchmarks for real-world development tasks remain largely unexplored. To address this limitation, we formulate task-oriented MCP server recommendation as a structured retrieval-and-ranking problem that jointly considers semantic relevance and engineering constraints. We first construct Task2MCP, a task-centered dataset that systematically associates taxonomy-grounded development tasks with curated MCP servers. This dataset provides structured supervision and a reproducible evaluation environment for research on MCP tool recommendations. Building on this dataset, we propose T2MRec, a task-to-MCP server recommendation model. It models semantic relevance and structural compatibility to construct an initial candidate set. Then it improves coverage and ranking quality through centroid-based candidate expansion and constrained LLM-based re-ranking. In addition, we design and implement an interactive MCP server recommendation agent prototype that operates in conversational environments to support dynamic decision-making. The agent assists developers in efficiently evaluating and integrating tools by providing recommended MCP servers together with usage guidelines.

Personalized image aesthetics assessment (PIAA) aims to predict an individual user's subjective rating of an image, which requires modeling user-specific aesthetic preferences. Existing methods rely on historical user ratings for this modeling and therefore struggle when such data are unavailable. We address this zero-shot setting by using user profiles as contextual signals for personalization and adopting a profile-based personalization paradigm. We introduce P-MLLM, a profile-aware multimodal LLM that augments a frozen LLM with selective fusion modules for controlled visual integration. These modules selectively integrate visual information into the model's evolving hidden states during profile-conditioned reasoning, allowing visual information to be incorporated in a profile-aware manner. Experiments on recent PIAA benchmarks show that P-MLLM achieves competitive zero-shot performance and remains effective even with coarse profile information, highlighting the potential of profile-based personalization for zero-shot PIAA.

Project Yanasse presents a method for discovering new proofs of theorems in one area of mathematics by transferring proof strategy patterns (e.g., Lean 4 tactic invocation patterns) from a structurally distant area. The system extracts tactic usage distributions across 27 top-level areas of Mathlib (217,133 proof states), computes z-scores to identify tactics that are heavily used in a source area but rare or absent in a target area, matches source and target proof states via GPU-accelerated NP-hard analogy (running on a MacBook Air via Apple's MPS backend), and then asks an AI reasoning agent to semantically adapt--not symbol-substitute--the source tactics invocation pattern to the target theorem. In this first part of the study, the method is applied to the pair Probability -> Representation Theory, producing 4 Lean-verified new proofs out of 10 attempts (40%). The proofs compile with zero sorry declarations. The key finding is that tactic schemas decompose into a head (domain-gated, rarely transfers) and a modifier (domain-general, often transfers): filter upwards's head fails in representation theory (no Filter structure), but its [LIST] with ω modifier transfers cleanly as ext1 + simp [LIST] + rfl. Crucially, the underlying matching engine--deep vision lib.py--is entirely domain independent: the same optimization code for an NP-hard matching that matches chess positions by analogy matches Lean proof states by analogy, without knowing which domain it is processing. Only a relation extractor is domain-specific.

Conditional depth execution routes a subset of tokens through a lightweight cheap FFN while the remainder execute the standard full FFN at each controlled layer. The central difficulty is gate training: the gate decision must propagate through many layers before it influences the language modeling (LM) loss, so the resulting gradients are weak and noisy. Auxiliary losses are commonly stacked to stabilise training, yet the interactions among them -- particularly between a predictive auxiliary and explicit score supervision -- have not been systematically compared under controlled conditions. We evaluate two gate designs under a 157.5M-parameter decoder-only model with controller-only training, 50% full-path budget, and 3-seed runs on a fineweb-edu subset. The MLP gate (G1) maps the current hidden state to a utility score; the JEPA-guided gate (G3) adds an action-conditional predictor that forecasts, in a low-dimensional latent space, the outcome of executing full vs. cheap per token, aligned against a fixed target head. Under the standard recipe with oracle-style utility regression and pairwise rank supervision (util/rank), G3 improves early-to-mid optimisation over G1 in 3/3 seeds (lower avg LM, faster threshold hits, ~10.3x lower grad norms), with 20k-step endpoint LM within a 0.005 heuristic reference. A key finding (ablation A3): jointly removing util/rank improves best/avg LM and threshold-hit speed in 3/3 seeds for both gates, and the early-to-mid advantage of G3 over G1 disappears. We trace this to an off-policy oracle label that assumes all subsequent layers execute full, whereas gated execution routes only a fraction through full -- making util/rank net-negative under the current recipe. Removing util/rank also cuts the training FLOPs proxy from ~1.53x to ~1.07x full-only (2.87h to 1.75h on a V100-32GB, ~39%). Conclusions are scoped to the studied regime.

The rapid rise of Large Language Models (LLMs) has revolutionized various artificial intelligence (AI) applications, from natural language processing to code generation. However, the computational demands of these models, particularly in training and inference, present significant challenges. Traditional systems are often unable to meet these requirements, necessitating the integration of cloud-native and distributed architectures. This paper explores the role of cloud platforms and distributed systems in supporting the scalability, efficiency, and optimization of LLMs. We discuss the complexities of LLM deployment, including data management, resource optimization, and the need for microservices, autoscaling, and hybrid cloud-edge solutions. Additionally, we examine emerging research trends, such as serverless inference, quantum computing, and federated learning, and their potential to drive the next phase of LLM innovation. The paper concludes with a roadmap for future developments, emphasizing the need for continued research, standardization, and cross-sector collaboration to sustain the growth of LLMs in both research and enterprise applications.

We present a novel approach for claim verification from tabular data documents. Recent LLM-based approaches either employ complex pretraining/fine-tuning or decompose verification into subtasks, often lacking comprehensive explanations and generalizability. To address these limitations, we propose a Multi-Agentic framework for Claim verification (MACE) consisting of three specialized agents: Planner, Executor, and Verifier. Instead of elaborate finetuning, each agent employs a zero-shot Chain-of-Thought setup to perform its tasks. MACE produces interpretable verification traces, with the Planner generating explicit reasoning strategies, the Executor providing detailed computation steps, and the Verifier validating the logic. Experiments demonstrate that MACE achieves state-of-the-art (SOTA) performance on two datasets and performs on par with the best models on two others, while achieving 80--100\% of best performance with substantially smaller models: 27--92B parameters versus 235B. This combination of competitive performance, memory efficiency, and transparent reasoning highlights our framework's effectiveness.

Recursive architectures such as Tiny Recursive Models (TRMs) perform implicit reasoning through iterative latent computation, yet the geometric structure of these reasoning trajectories remains poorly understood. We investigate the activation manifold of TRMs during recursive unrolling and find that activations occupy an effectively linear, low-dimensional subspace whose principal directions can be tracked dynamically with cheap power iterations. This suggests that weight-sharing concentrates iterative computation along a small number of dominant eigendirections, and we find that this concentration varies sharply across computational sites. We exploit this structure through LASER (Low-Rank Activation SVD for Efficient Recursion), a dynamic compression framework that maintains an evolving low-rank basis via matrix-free subspace tracking with a fidelity-triggered reset mechanism, achieving ${\sim}60\%$ activation memory savings with no statistically significant accuracy degradation. Our analysis raises questions about how recursive architectures allocate representational capacity during implicit reasoning, and whether this concentration can be exploited to improve the efficiency and stability of latent computation.

Breast cancer diagnosis demands rapid and precise tools, yet traditional histopathological methods often fall short in intra-operative settings. Deep Ultraviolet (DUV) fluorescence imaging emerges as a transformative approach, offering high-contrast, label-free visualization of whole-slide images (WSIs) with unprecedented detail, surpassing conventional hematoxylin and eosin (H&E) staining in speed and resolution. However, existing deep learning methods for breast cancer classification, predominantly patch-based, fragment spatial context and incur significant preprocessing overhead, limiting their clinical utility. Moreover, standard attention mechanisms, such as Spatial, Squeeze-and-Excitation, Global Context and Guided Context Gating, fail to fully exploit the rich, multi-scale regional relationships inherent in DUV-WSI data, often prioritizing generic feature recalibration over diagnostic specificity. This study introduces a novel Region-Affinity Attention mechanism tailored for DUV-WSI breast cancer classification, processing entire slides without patching to preserve spatial integrity. By modeling local neighbor distances and constructing a full affinity matrix, our method dynamically highlights diagnostically relevant regions, augmented by a contrastive loss to enhance feature discriminability. Evaluated on a dataset of 136 DUV-WSI samples, our approach achieves an accuracy of 92.67 +/- 0.73% and an AUC of 95.97%, outperforming existing attention methods.

Selective State Space Models (SSMs), notably Mamba, employ diagonal state transitions that limit both memory retention and bilinear computational capacity. We propose a factorized bilinear input modulation that augments the SSM with a state-input product, interpretable as a finite-dimensional Koopman bilinear form. After introducing a shared state across channels (Coupled SSM), the modulation admits two implementations. Coupled Bilinear Input Modulation (Coupled-BIM) retains the full bilinear product at the cost of sequential computation, while Coupled Gated Modulation (Coupled-GM) linearizes it into a gate modulation that is compatible with the parallel scan. Experiments on a multiple input-delay pendulum (memory retention) and NARMA-10 (bilinear computation) reveal a clear dissociation. Coupled-GM substantially improves memory retention but not bilinear computation, while Coupled-BIM improves both. A pathway ablation confirms that the two downstream routes of the bilinear signal serve complementary roles. The improvement is statistically robust, with Coupled-BIM consistently outperforming all other variants on bilinear computation. Furthermore, only Coupled-BIM benefits from increasing the SSM state dimension, while coupling or gate modulation alone show no improvement, establishing the bilin-ear mechanism as uniquely capable of exploiting larger state spaces.

Modeling coordination among generative agents in complex multi-round decision-making presents a core challenge for AI and operations management. Although behavioral experiments have revealed cognitive biases behind supply chain inefficiencies, traditional methods face scalability and control limitations. We introduce a scalable experimental paradigm using Large Language Models (LLMs) to simulate multi-stage supply chain dynamics. Grounded in a Hierarchical Reasoning Framework, this study specifically analyzes the impact of cognitive heterogeneity on agent interactions. Unlike prior homogeneous settings, we employ DeepSeek and GPT agents to systematically vary reasoning sophistication across supply chain tiers. Through rigorously replicated and statistically validated simulations, we investigate how this cognitive diversity influences collective outcomes. Results indicate that agents exhibit myopic and self-interested behaviors that exacerbate systemic inefficiencies. However, we demonstrate that information sharing effectively mitigates these adverse effects. Our findings extend traditional behavioral methods and offer new insights into the dynamics of AI-enabled organizations. This work underscores both the potential and limitations of LLM-based agents as proxies for human decision-making in complex operational environments.

We derive explicit non-asymptotic PAC-Bayes generalization bounds for Gibbs posteriors, that is, data-dependent distributions over model parameters obtained by exponentially tilting a prior with the empirical risk. Unlike classical worst-case complexity bounds based on uniform laws of large numbers, which require explicit control of the model space in terms of metric entropy (integrals), our analysis yields posterior-averaged risk bounds that can be applied to overparameterized models and adapt to the data structure and the intrinsic model complexity. The bound involves a marginal-type integral over the parameter space, which we analyze using tools from singular learning theory to obtain explicit and practically meaningful characterizations of the posterior risk. Applications to low-rank matrix completion and ReLU neural network regression and classification show that the resulting bounds are analytically tractable and substantially tighter than classical complexity-based bounds. Our results highlight the potential of PAC-Bayes analysis for precise finite-sample generalization guarantees in modern overparameterized and singular models.

Vision-Language Models (VLMs) achieve strong cross-modal performance, yet recent evidence suggests they over-rely on textual descriptions while under-utilizing visual evidence -- a phenomenon termed ``text shortcut learning.'' We propose an adversarial evaluation framework that quantifies this cross-modal dependency by measuring accuracy degradation (Drop) when semantically conflicting text is paired with unchanged images. Four adversarial strategies -- shape\_swap, color\_swap, position\_swap, and random\_text -- are applied to a controlled geometric-shapes dataset ($n{=}1{,}000$). We compare three configurations: Baseline CLIP (ViT-B/32), LoRA fine-tuning, and LoRA Optimized (integrating Hard Negative Mining, Label Smoothing, layer-wise learning rates, Cosine Restarts, curriculum learning, and data augmentation). The optimized model reduces average Drop from 27.5\% to 9.8\% (64.4\% relative improvement, $p{<}0.001$) while maintaining 97\% normal accuracy. Attention visualization and embedding-space analysis confirm that the optimized model attends more to visual features and achieves tighter cross-modal alignment.

Large language models require continuous adaptation to new tasks while preserving safety alignment. However, fine-tuning on even benign data often compromises safety behaviors, including refusal of harmful requests, truthfulness, and commonsense reasoning. We investigate which training samples cause alignment drift through a data-centric lens. Our empirical analysis shows samples contribute unequally: high-gradient samples cause greater safety degradation and drive models toward pretrained distributions, while moderate-gradient samples enable task learning with minimal alignment loss. We propose gradient-based sample selection that filters high-gradient samples during fine-tuning. Across multiple model families on continual domain tasks, our method substantially improves alignment preservation while maintaining competitive task performance, without requiring curated safe data or architectural modifications. Our method is robust across selection ratios, task orderings, and diverse attack benchmarks.

Extracting clinically relevant information from unstructured medical narratives such as admission notes, discharge summaries, and emergency case histories remains a challenge in clinical natural language processing (NLP). Medical Entity Recognition (MER) identifies meaningful concepts embedded in these records. Recent advancements in large language models (LLMs) have shown competitive MER performance; however, evaluations often focus on general entity types, offering limited utility for real-world clinical needs requiring finer-grained extraction. To address this gap, we rigorously evaluated the open-source LLaMA3 model for fine-grained medical entity recognition across 18 clinically detailed categories. To optimize performance, we employed three learning paradigms: zero-shot, few-shot, and fine-tuning with Low-Rank Adaptation (LoRA). To further enhance few-shot learning, we introduced two example selection methods based on token- and sentence-level embedding similarity, utilizing a pre-trained BioBERT model. Unlike prior work assessing zero-shot and few-shot performance on proprietary models (e.g., GPT-4) or fine-tuning different architectures, we ensured methodological consistency by applying all strategies to a unified LLaMA3 backbone, enabling fair comparison across learning settings. Our results showed that fine-tuned LLaMA3 surpasses zero-shot and few-shot approaches by 63.11% and 35.63%, respectivel respectively, achieving an F1 score of 81.24% in granular medical entity extraction.

Fine-tuning well-aligned large language models (LLMs) on new domains often degrades their safety alignment, even when using benign datasets. Existing safety alignment techniques primarily focus on pretraining, leaving fine-tuned models vulnerable to behavioral shifts. In this work, we introduce safety token regularization (STR), a lightweight method designed to preserve safety properties during fine-tuning. Our approach identifies salient tokens from rejection templates of well-aligned models and constrains their associated logits during training, preventing the loss of critical safety behaviors. Unlike reinforcement learning or preference optimization methods, STR requires minimal additional computation and seamlessly integrates with parameter-efficient fine-tuning techniques such as LoRA. Comprehensive experiments demonstrate that our approach achieves safety performance on par with state-of-the-art methods, while preserving task-specific utility and requiring minimal implementation overhead. Furthermore, we show that safety token regularization enhances training stability and overall performance beyond safety considerations alone. This work offers a practical and readily deployable strategy for continual safety alignment in fine-tuned LLMs.

Automating medical reports for retinal images requires a sophisticated blend of visual pattern recognition and deep clinical knowledge. Current Large Vision-Language Models (LVLMs) often struggle in specialized medical fields where data is scarce, leading to models that overfit and miss subtle but critical pathologies. To address this, we introduce DREAM (Dynamic Retinal Enhancement with Adaptive Multi-modal Fusion), a novel framework for high-fidelity medical report generation that excels even with limited data. DREAM employs a unique two-stage fusion mechanism that intelligently integrates visual data with clinical keywords curated by ophthalmologists. First, the Abstractor module maps image and keyword features into a shared space, enhancing visual data with pathology-relevant insights. Next, the Adaptor performs adaptive multi-modal fusion, dynamically weighting the importance of each modality using learnable parameters to create a unified representation. To ensure the model's outputs are semantically grounded in clinical reality, a Contrastive Alignment module aligns these fused representations with ground-truth medical reports during training. By combining medical expertise with an efficient fusion strategy, DREAM sets a new state-of-the-art on the DeepEyeNet benchmark, achieving a BLEU-4 score of 0.241, and further demonstrates strong generalization to the ROCO dataset.

Digital subtraction angiography (DSA) in coronary imaging is fundamentally challenged by physiological motion, forcing reliance on raw angiograms cluttered with anatomical noise. Existing deep learning methods often produced images with two critical clinically unacceptable flaws: persistent boundary artifacts and a loss of native tissue grayscale fidelity that undermined diagnostic confidence. We propose a novel framework termed as CDSA-Net that for the first time explicitly decouples and jointly optimizes vascular structure preservation and realistic background restoration. CDSA-Net introduces two core innovations: (i) A hierarchical geometric prior guidance (HGPG) mechanism, embedded in our coronary structure extraction network (CSENet). It synergistically combines integrated geometric prior (IGP) with gated spatial modulation (GSM) and centerline-aware topology (CAT) loss supervision, ensuring structural continuity. (ii) An adaptive noise module (ANM) within our coronary background restoration network (CBResNet). Unlike standard restoration, ANM uniquely models the stochastic nature of clinical X-ray noise, bridging the domain gap to enable seamless background intensity estimation and the complete elimination of boundary artifacts. The final subtraction is obtained by removing the restored background from the raw angiogram. Quantitatively, it significantly outperformed state-of-the-art methods in vascular intensity correlation and perceptual quality. A 25.6% improvement in morphology assessment efficiency and a 42.9% gain in hemodynamic evaluation speed set a new benchmark for utility in interventional cardiology, while maintaining diagnostic results consistent with raw angiograms. The project code is available at https://github.com/DrThink-ai/CDSA-Net.

Iterative alignment methods based on purely greedy updates are remarkably effective in practice, yet existing theoretical guarantees of \(O(\log T)\) KL-regularized regret can seem pessimistic relative to their empirical performance. In this paper, we argue that this mismatch arises from the regret criterion itself: KL-regularized regret conflates the statistical cost of learning with the exploratory randomization induced by the softened training policy. To separate these effects, we study the traditional temperature-zero regret criterion, which evaluates only the top-ranked response at inference time. Under this decision-centric notion of performance, we prove that standard greedy online alignment methods, including online RLHF and online DPO, achieve constant \((O(1))\) cumulative regret. By isolating the cost of identifying the best response from the stochasticity induced by regularization, our results provide a sharper theoretical explanation for the practical superb efficiency of greedy alignment.

Recent advances in summary evaluation are based on model-based metrics to assess quality dimensions, such as completeness, conciseness, and faithfulness. However, these methods often require large language models, and predicted scores are frequently miscalibrated, limiting their reliability. Moreover, evaluating the average quality across different summaries for a single document typically requires access to multiple reference summaries. Here, we propose a general framework that generates individual and average proxy scores without relying on reference summaries, human annotations, or expensive model-based metrics. We also propose group isotonic regression binning (GIRB), a calibration method that adjusts the raw predictions to better align with ground-truth evaluation metrics. While we focus on continuous-value scenarios, such as summarization, the method is applicable to discrete-value tasks, such as question answering. Experiments on seven datasets demonstrate that our approach consistently outperforms existing baselines.

Recent advances in summarization research focus on improving summary quality across multiple criteria, such as completeness, conciseness, and faithfulness, by jointly optimizing these dimensions. However, these efforts largely overlook the challenge of controlling summary generation with respect to individual criteria, especially in the presence of their inherent trade-offs. For example, enhancing conciseness can compromise completeness, and vice versa. In this work, we address this gap by proposing a loss function that aligns model outputs with fine-grained, model-based evaluation scores (e.g., from FineSurE), enabling both improvement in summary quality and dimension-specific control. Our approach improves the overall quality of summaries while maintaining the ability to selectively prioritize one criterion over others. Experiments on three pretrained models (LLaMA, Qwen, and Mistral) demonstrate that our method achieves performance comparable to state-of-the-art summarizers, while uniquely offering strong controllability over individual quality dimensions.

Converting betting odds into accurate outcome probabilities is a fundamental challenge in order to use betting odds as a benchmark for sports forecasting and market efficiency analysis. In this study, we propose two methods to overcome the limitations of existing conversion methods. Firstly, we propose an odds-only method to convert betting odds to probabilities without using historical data for model fitting. While existing odds-only methods, such as Multiplicative, Shin, and Power exist, they do not adjust for biases or relationships we found in our betting odds dataset, which consists of 90014 football matches across five different bookmakers. To overcome these limitations, our proposed Odds-Only-Equal-Profitability-Confidence (OO-EPC) method aligns with the bookmakers' pricing objectives of having equal confidence in profitability for each outcome. We provide empirical evidence from our betting odds dataset that, for the majority of bookmakers, our proposed OO-EPC method outperforms the existing odds-only methods. Beyond controlled experiments, we applied the OO-EPC method under real-world uncertainty by using it for six iterations of an annual basketball outcome forecasting competition. Secondly, we propose a generalised linear model that utilises historical data for model fitting and then converts betting odds to probabilities. Existing generalised linear models attempt to capture relationships that the Efficient Market Hypothesis already captures. To overcome this shortcoming, our proposed Favourite-Longshot-Bias-Adjusted Generalised Linear Model (FL-GLM) fits just one parameter to capture the favourite-longshot bias, providing a more interpretable alternative. We provide empirical evidence from historical football matches where, for all bookmakers, our proposed FL-GLM outperforms the existing multinomial and logistic generalised linear models.

Multi-agent reinforcement learning (MARL) is crucial for AI systems that operate collaboratively in distributed and adversarial settings, particularly in multi-domain operations (MDO). A central challenge in cooperative MARL is determining how agents should coordinate: existing approaches must either hand-specify graph topology, rely on proximity-based heuristics, or learn structure entirely from environment interaction; all of which are brittle, semantically uninformed, or data-intensive. We investigate whether large language models (LLMs) can generate useful coordination graph priors for MARL by using minimal natural language descriptions of agent observations to infer latent coordination patterns. These priors are integrated into MARL algorithms via graph convolutional layers within a graph neural network (GNN)-based pipeline, and evaluated on four cooperative scenarios from the Multi-Agent Particle Environment (MPE) benchmark against baselines spanning the full spectrum of coordination modeling, from independent learners to state-of-the-art graph-based methods. We further ablate across five compact open-source LLMs to assess the sensitivity of prior quality to model choice. Our results provide the first quantitative evidence that LLM-derived graph priors can enhance coordination and adaptability in dynamic multi-agent environments, and demonstrate that models as small as 1.5B parameters are sufficient for effective prior generation.

Multi-role dialogue summarization requires modeling complex interactions among multiple speakers while preserving role-specific information and factual consistency. However, most existing methods optimize for automatic metrics such as ROUGE and BERTScore, which favor surface-level imitation of references rather than genuine gains in faithfulness or alignment with human preferences. We propose a novel framework that couples explicit cognitive-style reasoning with reward-based optimization for multi-role dialogue summarization. Our method first distills structured reasoning traces (e.g., step-by-step inferences and intermediate reflections) from a large teacher model and uses them as auxiliary supervision to initialize a reasoning-aware summarizer via staged supervised fine-tuning. It then applies GRPO with a dual-principle reward that blends metric-based signals with human-aligned criteria targeting key information coverage, implicit inference, factual faithfulness, and conciseness. Experiments on multilingual multi-role dialogue benchmarks show that our method matches strong baselines on ROUGE and BERTScore. Specifically, results on CSDS confirm the framework's stability in semantic consistency, while in-depth analysis on SAMSum demonstrates clear gains in factual faithfulness and model-based preference alignment. These findings underscore the value of reasoning-aware and preference-aware training for reliable dialogue summarization. Checkpoints and datasets are available at https://huggingface.co/collections/NebulaPixel/summorchestra-multirole-summary.

We evaluate five state-of-the-art open-weights coding language models -- Kimi-K2.5 (at Q3 and Q4 quantizations), GLM-5.1, Qwen3-Coder-480B, and DeepSeek-V3.2 -- on a single multi-file React Native application generation task on NVIDIA GH200 576 GB hardware. The task specifies authentication, per-user per-day counting, and web compatibility, and is evaluated on whether the generated project runs out-of-the-box and on feature-level correctness. We find that SWE-Bench rankings do not predict task performance: Kimi-K2.5 at aggressive 3-bit quantization (UD-Q3_K_XL, 480 GB) produces the most complete and specification-compliant output, outranking models with substantially higher SWE-Bench Pro scores. We document three novel deployment findings: (1) default temperature=0 in coding tools causes sampling hangs with reasoning-model architectures, (2) reasoning-model thinking traces can leak through integration tools' file-path parsers, and (3) web-platform adaptation of native-mobile APIs is a universal training-data gap across every model tested. We also map the hardware-tier structure of April 2026 open-weights coding models, identifying two architectural schools and showing that the efficiency school (10-15 B active parameters) delivers equivalent SWE-Bench results at roughly 1/7th the hardware cost of the scale school (32-40 B active parameters).

As Artificial Intelligence (AI) and Agentic AI become increasingly integrated across sectors such as education and healthcare, it is critical to ensure that Multi-Agent Education System (MAES) is explainable from the early stages of requirements engineering (RE) within the AI software development lifecycle. Explainability is essential to build trust, promote transparency, and enable effective human-AI collaboration. Although personas are well-established in human-computer interaction to represent users and capture their needs and behaviors, their role in RE for explainable MAES remains underexplored. This paper proposes a human-first, persona-driven, explainable MAES RE framework and demonstrates the framework through a MAES for clinical reasoning training. The framework integrates personas and user stories throughout the RE process to capture the needs, goals, and interactions of various stakeholders, including medical educators, medical students, AI patient agent, and clinical agents (physical exam agent, diagnostic agent, clinical intervention agent, supervisor agent, evaluation agent). The goals, underlying models, and knowledge base shape agent interactions and inform explainability requirements that guided the clinical reasoning training of medical students. A post-usage survey found that more than 78\% of medical students reported that MAES improved their clinical reasoning skills. These findings demonstrate that RE based on persona effectively connects technical requirements with non-technical medical students from a human-centered approach, ensuring that explainable MAES are trustworthy, interpretable, and aligned with authentic clinical scenarios from the early stages of the AI system engineering. The partial MAES for the clinical scenario simulator is~\href{https://github.com/2sigmaEdTech/MAS/}{open sourced here}.

Large Language Models (LLMs) show promise for automated code repair but often struggle with the complex semantic and structural correctness required. We present SynthFix, a hybrid neural-symbolic framework that improves LLM-based vulnerability repair by unifying code synthesis with compiler-informed symbolic feedback. The core of our approach is an adaptive training strategy where a neural Router Model directs code samples to either Supervised Fine-Tuning (SFT) to learn common patterns or Reward Fine-Tuning (RFT) with symbolic rewards for complex, iterative refinement. On the FixJS (JavaScript) and CodeFlaws (C) benchmarks, SynthFix achieves up to 18% relative improvement in CodeBLEU/CrystalBLEU and 32% in Exact Match over strong SFT and RFT baselines. Our results show that this adaptive combination of training strategies, which mirrors how developers alternate between pattern application and tool feedback, significantly improves the accuracy and efficiency of LLM-based vulnerability repair. Our code and data are available at https://github.com/CoderDoge1108/SynthFix.

Bitcoin transaction fees will become more important as the block subsidy declines, but fee formation is hard to study with blockchain data alone because the relevant queueing environment is unobserved. We develop and estimate a structural model of Bitcoin fee choice that treats the mempool as a market for scarce blockspace. We assemble a novel, high-frequency mempool panel, from a self-run Bitcoin node that records transaction arrivals, exits, block inclusion, fee-bumping events, and congestion snapshots. We characterize the fee market as a Vickery-Clarke-Groves mechanism and derive an equation to estimate fees. In the first-stage we estimate a monotone delay technology linking fee-rate priority and network state to expected confirmation delay. We then estimate how fees respond to that delay technology and to transaction characteristics. We find that congestion is the main determinant of delay; that the marginal value of priority is priced in fees, which is increasing in the gradient of confirmation time reduction per movement up in the fee queue; and that transactor choice of RBF, CPFP, and block conditions have economically important effects on fees.

In LLM-based code generation, multiple code candidates are often generated in parallel from the same prompt -- for example, in best-of-N sampling or multi-candidate code completion. These requests can share KV caches through a common prefix, yet the extent to which their Mixture-of-Experts (MoE) expert routing overlaps, and how this overlap varies across layers, remains insufficiently understood. We study Qwen3.5-35B-A3B-FP8 (256 routed experts, top-8) by performing tree-search-based branching generation from a shared prefix (851 completed codes, temperature 0.7) and analyzing the results with a compiler-output-based alignment (gcc -S -O0 assembly) that controls for token-identity confounds. Our findings are threefold: (1) At positions where both sequences generated the same token, Jaccard similarity reaches 0.649 (40x random), while even at positions with different tokens it remains 0.175 (11x random). (2) A layer-wise decomposition reveals a crossing pattern: same-token routing similarity exceeds different-token similarity across all layers, but dips in the middle layers (L14-20), while different-token similarity peaks in the middle layers at 14x random. (3) In tree-search code generation, 67% of successfully compiled codes concentrate in the top three assembly-equivalent groups, and 99.6% of within-group differences consist of comments and blank lines. We show that diversity in top-P search, including beam search, poses a significant challenge. These results refine the "context-independent routing" claim of prior work through layer-wise decomposition and suggest opportunities for improving search efficiency in LLM code generation.

INTRODUCTION: The proliferation of the amalgamation of IoT and edge computing has increased the demand for decentralised trust and security mechanisms capable of operating across heterogeneous and resource-limited devices. Approaches such as federated learning, Zero Trust architectures, lightweight blockchain and distributed neural models offer alternatives to centralised control. OBJECTIVES: This review examines various state-of-the-art decentralised mechanisms and evaluates their effectiveness in terms of securing IoT networks at the edge. METHODS: Thirty recent studies were analysed to compare how decentralised architectures establish trust, support secure communication and enable intrusion and anomaly detection. Frameworks, such as DFGL-LZTA, SecFedDNN and COSIER were assessed. RESULTS: Decentralised designs enhance privacy, reduce single points of failure and improve adaptive threat response, though challenges remain in scalability, efficiency and interoperability. CONCLUSION: The study identifies key considerations and future research needs for building secure and resilient trust-aware IoT edge ecosystems.

Emotional Support Conversation (ESC) plays a critical role in mental health assistance by providing accessible psychological support in real-world applications. Large Language Models (LLMs) have shown strong empathetic abilities in ESC tasks. Yet, existing methods overlook the issue of cognitive distortions in help-seekers' expressions. As a result, current models can only provide basic emotional comfort, rather than helping help-seekers address their psychological distress at a deeper cognitive level. To address this challenge, we construct the CogBiasESC dataset, the first dataset that expands existing ESC datasets by adding labels for cognitive distortions, includes their type, intensity, and safe risk level. Furthermore, we propose the Cognitive Policy-driven Large Language Model framework (CoPoLLM) to enhance LLMs' ability to diagnose and intervene cognitive distortions in help-seekers. We also analyze the safety advantages of CoPoLLM from a theoretical perspective. Experimental results show that CoPoLLM significantly outperforms 15 state-of-the-art baselines in terms of distortion diagnosis accuracy, intervention strategy effectiveness, and safety risk control.

Future spacecraft operations require autonomy that can interpret high-level mission intent while preserving safety. However, existing trajectory optimization still relies heavily on expert-crafted formulations and does not support intent-conditioned decision-making. This paper proposes an intent-aligned spacecraft guidance framework that links high-level reasoning and safe trajectory optimization through explicit intermediate abstractions, based on behavior sequences and waypoint constraints. A foundation model first predicts an intent-aligned behavior plan, a waypoint generation model then converts it into waypoint constraints, and the safe trajectory is computed via optimization. This decomposition enables scalable supervision without sacrificing safety. Numerical experiments in close-proximity operation scenarios demonstrate that the proposed pipeline achieves over 90\% SCP convergence and yields a $1.5\times$ higher rate of generating trajectories that satisfy the top intent-prioritized performance criteria than heuristic decision-making. These results support the use of intermediate behavior abstraction as a practical interface between foundation-model reasoning and safety-critical onboard spacecraft autonomy.

Fine-tuning adapts pretrained networks to new objectives. Whether the resulting depth profile of representational change reflects an intrinsic property of the model or the magnitude of gradient flow has not been tested directly. We measure this profile across 240 fine-tuning runs spanning 15 models in four architecture families (encoder and decoder transformers, a state-space model, and an RNN) at scales from 125M to 6.9B parameters. Representational change concentrates in output-proximal layers in every standard-training run except one. We apply a per-layer control that equalizes $\|ΔW\|/\|W\|$ across layers after each optimizer step. Under this control, the profile persists in some conditions and collapses in others. At 125M--350M, sequential-block architectures (BERT, OPT, GPT-2) retain the slope across tested objectives while parallel-block architectures (Pythia, CodeGen) retain it only for causal-language-modeling objectives. This architectural distinction narrows at 1.3B--1.4B, where both block types show positive equal-step slopes for CausalLM. Under standard training, profile shape is described by two additional axes: steepness tracks a training-free objective distance at initialization, and profile width is dominated by architecture. We treat the locality gradient, the depthwise slope of representational change, as a composite phenomenon whose components are scale-dependent.

We introduce RosettaSearch, an inference-time multi-objective optimization approach for protein sequence optimization. We use large language models (LLMs) as a generative optimizer within a search algorithm capable of controlled exploration and exploitation, using rewards computed from RosettaFold3, a structure prediction model. In a large-scale evaluation, we apply RosettaSearch to 400 suboptimal sequences generated by LigandMPNN (a state-of-the-art model trained for protein sequence design), recovering high-fidelity designs that LigandMPNN's single-pass decoding fails to produce. RosettaSearch's designs show improvements in structural fidelity metrics ranging between 18\% to 68\%, translating to a 2.5$\times$ improvement in design success rate. We observe that these gains in success rate are robust when RosettaSearch-designed sequences are evaluated with an independent structure prediction oracle (Chai-1) and generalize across two distinct LLM families (o4-mini and Gemini-3), with performance scaling consistently with reasoning capability. We further demonstrate that RosettaSearch improves sequence fidelity for ProteinMPNN-designed sequences on \textit{de novo} backbones from the Dayhoff atlas, showing that the approach generalizes beyond native protein structures to computationally generated backbones. We also demonstrate a multi-modal extension of RosettaSearch with vision-language models, where images of predicted protein structures are used as feedback to incorporate structural context to guide protein sequence generation. The sequence trajectories generated by our approach can be used as training data in sequence design models or in post-training and will be released along with the code and datasets upon publication.

Modeling human cognitive states is essential for advanced artificial intelligence. Existing Large Language Models (LLMs) mainly address isolated tasks such as emotion analysis or stance detection, and fail to capture interactions among cognitive dimensions defined in psychology, including emotion, thinking style, stance, and intention. To bridge this gap, we construct CognitiveBench, the first benchmark with unified annotations across the above four dimensions. Experiments on CognitiveBench show that although LLMs perform well on single dimension tasks, their performance drops sharply in joint multi-dimensional modeling. Using Gromov $δ$-hyperbolicity analysis, we find that CognitiveBench exhibits a strong hierarchical structure. We attribute the performance bottleneck to ``Cognitive Crowding'', where hierarchical cognitive states require exponential representational space, while the Euclidean space of LLMs grows only polynomially, causing representation overlap and degraded performance. To address this mismatch, we propose HyCoLLM, which models cognitive states in hyperbolic space and aligns LLM representations via Hyperbolic Guided Alignment Tuning. Results show that HyCoLLM substantially improves multi-dimensional cognitive understanding, allowing 8B parameter model to outperform strong baselines, including GPT-4o.

Collective communication incurs significant overhead in LLM workloads. Although overlapping communication with computation in application-level is a common strategy, it often requires substantial code modifications and is impractical for many workloads (e.g., tensor and expert parallelism). We present CCCL, a built-in compression-based collective communication library that supports operations such as allreduce, alltoall, and send/recv without requiring any user-side changes, thereby enabling seamless adoption in existing applications. CCCL tightly fuses compression kernels to minimize memory accesses and integrates with NCCL to eliminate the data coalescing stage, making it fast enough (up to 3x NVLink bandwidth) to sustain communication. Our evaluation shows that CCCL improves end-to-end throughput in vLLM PD disaggregation workloads by up to 10.1% and microbenchmark throughput by up to 30%.

Sparsity or complexity? In modern high-dimensional asset pricing, these are often viewed as competing principles: richer feature spaces appear to favor complexity, while economic intuition has long favored parsimony. We show that this tension is misplaced. We distinguish capacity sparsity-the dimensionality of the candidate feature space-from factor sparsity-the parsimonious structure of priced risks-and argue that the two are complements: expanding capacity enables the discovery of factor sparsity. Revisiting the benchmark empirical design of Didisheim et al. (2025) and pushing it to higher complexity regimes, we show that nonlinear feature expansions combined with basis pursuit yield portfolios whose out-of-sample performance dominates ridgeless benchmarks beyond a critical complexity threshold. The evidence shows that the gains from complexity arise not from retaining more factors, but from enlarging the space from which a sparse structure of priced risks can be identified. The virtue of complexity in asset pricing operates through factor sparsity.

We present, to our knowledge, the most comprehensive cross-model evaluation of LLM agents on offensive cybersecurity tasks, benchmarking 10 frontier models from 7 providers on all 200 challenges of the NYU CTF Bench. Building on the D-CIPHER multi-agent framework, we extend it with multi-provider backend support, a custom Kali Linux environment with over 100 pre-installed penetration testing tools, and runtime tool-discovery agents. Through a controlled factorial study, we find that the Kali Linux environment yields a +9.5 percentage-point improvement over Ubuntu, while auto-prompting and category-specific tips often degrade performance in well-equipped environments. Among models, Claude 4.5 Opus achieves the highest solve rate (59%), followed by Gemini 3 Pro (52%), with Gemini 3 Flash offering the best cost-efficiency at $0.05 per solve. Asymmetric planner/executor model assignments provide no meaningful benefit while coherent same-model configurations consistently outperform mixed-tier pairings. Our results indicate that environment tooling and model selection emerge as the strongest drivers of performance, whereas prompt engineering interventions show diminishing or negative returns in well-equipped environments. Reported performance reflects both model reasoning ability and compatibility with agent tooling and API integration.

The occurrence of cybersickness in virtual reality (VR) significantly impairs users' perception and sense of immersion. Therefore, timely detection of cybersickness and the application of appropriate intervention strategies are crucial for enhancing the user experience. However, existing cybersickness detection methods often suffer from issues such as poor detection reliability across different levels of cybersickness and unnecessary model complexity. Furthermore, while cybersickness exhibits significant inter-user variability, most existing approaches aggregate all data from users and lack user-specific solutions. In this paper, we investigate a lightweight approach for cybersickness detection incorporating an ensemble learning model and user-specific eye and head tracking data. Our experiments using the open-source dataset Simulation 2021 demonstrate that feature engineering and training set construction are critical for determining detection performance. Models trained with data from similar-content segments achieve the best results, attaining detection accuracies of 93% in the cross-user setting and 88% in the user-personalized setting, using only 23-dimensional eye and head features. Moreover, by using user-specific data, well-tuned ensemble learning models with shorter training and inference times can be feasibly applied to real-world cybersickness detection, offering superior time efficiency and outstanding detection performance. This work offers useful evidence toward the development of lightweight and user-adaptive cybersickness detection models for VR applications.

Physics-informed neural networks (PINNs) have emerged as a promising framework for solving inverse problems governed by partial differential equations (PDEs), including the reconstruction of turbulent flow fields from sparse data. However, most existing PINN formulations are deterministic and do not provide reliable quantification of epistemic uncertainty, which is critical for ill-posed problems such as data-driven Reynolds-averaged Navier-Stokes (RANS) modeling. In this work, we develop and systematically evaluate a set of probabilistic extensions of PINNs for uncertainty quantification in turbulence modeling. The proposed framework combines (i) Bayesian PINNs with Hamiltonian Monte Carlo sampling and a tempered multi-component likelihood, (ii) Monte Carlo dropout, and (iii) repulsive deep ensembles that enforce diversity in function space. Particular emphasis is placed on the role of ensemble diversity and likelihood tempering in improving uncertainty calibration for PDE-constrained inverse problems. The methods are assessed on a hierarchy of test cases, including the Van der Pol oscillator and turbulent flow past a circular cylinder at Reynolds numbers Re=3,900 (direct numerical simulation data) and Re = 10,000 (experimental particle image velocimetry data). The results demonstrate that Bayesian PINNs provide the most consistent uncertainty estimates across all inferred quantities, while function-space repulsive ensembles offer a computationally efficient approximation with competitive accuracy for primary flow variables. These findings provide quantitative insight into the trade-offs between accuracy, computational cost, and uncertainty calibration in physics-informed learning, and offer practical guidance for uncertainty quantification in data-driven turbulence modeling.

Transforming legal text into executable decision logic is a longstanding challenge in legal informatics. With the rise of LLMs, this task has gained renewed interest, but remains challenging due to requiring extensive manual coding and evaluation. We use a unique real-world dataset that pairs production-grade decision models with legal text from the Dutch Environment and Planning Act. These models power the Omgevingsloket government platform, where citizens check permit requirements for environmental activities. We study whether intermediate structured representations can improve LLM-based generation of executable decision models from legal text. We compare four input conditions: raw legal text, text enriched with semantic role labels, text enriched with input and output constraints, and text enriched with both. We evaluate along two dimensions: structural evaluation, through similarity to gold decision models with graph kernels and graphs' descriptive statistics, and outcome evaluation, through functional equivalence by executing models on pre-configured test scenarios. Our findings show that I/O constraints provide the dominant improvement (+37-54% similarity over baseline), while semantic role labels show modest improvements. Outcome evaluation shows that generated models match the gold standard on 51-53% of test scenarios, even though generated models are typically smaller and simpler. We find LLMs eliminate redundant pass-through logic that comprises up to 45-55% of nodes. Importantly, structural similarity and outcome equivalence are complementary: structural similarity does not guarantee outcome equivalence, and vice versa. To facilitate reproducibility, we publicly release our dataset of 95 production decision models with associated legal text and all experimental code.

Accurate autoregressive prediction of 3D turbulent flows remains challenging for neural PDE solvers, as small errors in fine-scale structures can accumulate rapidly over rollout. In this paper, we propose FlowRefiner, a flow matching-based iterative refinement framework for 3D turbulent flow simulation. The method replaces stochastic denoising refinement with deterministic ODE-based correction, uses a unified velocity-field regression objective across all refinement stages, and introduces a decoupled sigma schedule that fixes the noise range independently of refinement depth. These design choices yield stable and effective refinement in the small-noise regime. Experiments on large-scale 3D turbulence with rich multi-scale structures show that FlowRefiner achieves state-of-the-art autoregressive prediction accuracy and strong physical consistency. Although developed for turbulent flow simulation, the proposed framework is broadly applicable to iterative refinement problems in scientific modeling.

With an ever-growing zoo of LLMs and benchmarks, the need to orchestrate multiple models for improved task performance has never been more pressing. While frameworks like Mixture-of-Agents (MoA) attempt to coordinate LLMs, they often fall short in terms of (1) selecting relevant agents, (2) facilitating effective intra-agent communication, and (3) integrating responses efficiently. In this work, we propose Graph-of-Agents (GoA), a new graph-based framework for modeling multi-agent LLM communication. Our approach begins with node sampling, selecting only the most relevant agents by leveraging model cards that summarize each model's domain, task specialization, and other characteristics. Next, we construct edges between the selected agents by evaluating their responses against one another to determine relevance ordering. Directed message passing is then performed from highly relevant agents to less relevant ones to enhance their responses, followed by reverse message passing to refine the original responses of the more relevant agents. Finally, the updated responses are aggregated via graph-based pooling (e.g., max or mean pooling) to produce a single, unified answer. We evaluate GoA on diverse multi-domain benchmarks (MMLU, MMLU-Pro, GPQA) and domain-specific benchmarks (MATH, HumanEval, MedMCQA), with an agent pool of 6 LLMs spanning multiple domains. Surprisingly, GoA achieves superior performance using only 3 selected agents, outperforming recent multi-agent LLM baselines that utilize all 6 agents simultaneously. By adopting a graph structure, GoA offers both scalability and effectiveness through structured message passing-positioning it as a strong candidate for navigating the challenges of the ever-growing LLM zoo. Code is available at: https://github.com/UNITES-Lab/GoA.

This paper revisits momentum in the context of min-max optimization. Momentum is a celebrated mechanism for accelerating gradient dynamics in settings like convex minimization, but its direct use in min-max optimization makes gradient dynamics diverge. Surprisingly, Gidel et al. 2019 showed that negative momentum can help fix convergence. However, despite these promising initial results and progress since, the power of momentum remains unclear for min-max optimization in two key ways. (1) Generality: is global convergence possible for the foundational setting of convex-concave optimization? This is the direct analog of convex minimization and is a standard testing ground for min-max algorithms. (2) Fast convergence: is accelerated convergence possible for strongly-convex-strong-concave optimization (the only non-linear setting where global convergence is known)? Recent work has even argued that this is impossible. We answer both these questions in the affirmative. Together, these results put negative momentum on more equal footing with competitor algorithms, and show that negative momentum enables convergence significantly faster and more generally than was known possible.

Despite their substantial successes, AI agents continue to face fundamental challenges in terms of trustworthiness. Consider deep research agents, tasked with searching for information relevant to a given topic-while AI agents can perform effective information retrieval, there is little guarantee regarding the completeness of this information. Gaps in retrieved information can leave biases that mislead users even if the information they are given is correct and relevant. We introduce SeekerGym, a benchmark designed to evaluate the completeness of information retrieved by AI agents. In addition, SeekerGym also measures how well agents quantify their uncertainty in the completeness of their information; if an agent fails to retrieve all relevant information, it is useful for it to at least quantify how much might be missing. At a high level, each task in SeekerGym is a document (e.g., a Wikipedia article), and the AI agent must issue queries to retrieve passages from that document. Intuitively, the document comprehensively covers a topic, so the ability to retrieve its sections directly measures completeness of information retrieval. In addition to Wikipedia, we also consider machine learning survey papers, where the goal is to retrieve relevant sections of a survey paper. We benchmark several models and algorithms; the best approaches retrieve 42.5% of passages on Wikipedia and 29.2% on ML Surveys, leaving substantial room for improvement.

Manufacturing industries are facing increasing product variability due to the growing demand for personalized products. Under these conditions, ensuring safety becomes challenging as frequent reconfigurations can lead to unintended hazardous behaviors. Multi-agent control architectures have been proposed to improve flexibility through decentralized decision-making and coordination. However, these architectures are based on predefined task models, which limit their ability to adapt task planning to new product requirements while preserving safety. Recently, large language models have been introduced into manufacturing systems to enhance adaptability, but reliability remains a key challenge. To address this issue, we propose a control architecture that leverages the flexibility of large language models while preserving safety on the manufacturing shop floor. Specifically, the proposed framework verifies large language model-enabled task allocations by using temporal logic and discrete event systems. The effectiveness of the proposed framework is demonstrated through a case study that involves a multi-robot assembly scenario, showing that unsafe tasks can be allocated safely before task execution.

The rapid growth of scientific literature calls for automated methods to assess and predict research impact. Prior work has largely focused on citation-based metrics, leaving limited evaluation of models' capability to reason about other impact dimensions. To this end, we introduce SciImpact, a large-scale, multi-dimensional benchmark for scientific impact prediction spanning 19 fields. SciImpact captures various forms of scientific influence, ranging from citation counts to award recognition, media attention, patent reference, and artifact adoption, by integrating heterogeneous data sources and targeted web crawling. It comprises 215,928 contrastive paper pairs reflecting meaningful impact differences in both short-term (e.g., Best Paper Award) and long-term settings (e.g., Nobel Prize). We evaluate 11 widely used large language models (LLMs) on SciImpact. Results show that off-the-shelf models exhibit substantial variability across dimensions and fields, while multi-task supervised fine-tuning consistently enables smaller LLMs (e.g., 4B) to markedly outperform much larger models (e.g., 30B) and surpass powerful closed-source LLMs (e.g., o4-mini). These results establish SciImpact as a challenging benchmark and demonstrate its value for multi-dimensional, multi-field scientific impact prediction. Our project homepage is https://flypig23.github.io/sciimpact-homepage/

We present a generic algorithm for learning and approximate inference with an intuitive epistemic interpretation: iteratively focus on a subset of the model and resolve inconsistencies using the parameters under control. This framework, which we call Local Inconsistency Resolution (LIR) is built upon Probabilistic Dependency Graphs (PDGs), which provide a flexible representational foundation capable of capturing inconsistent beliefs. We show how LIR unifies and generalizes a wide variety of important algorithms in the literature, including the Expectation-Maximization (EM) algorithm, belief propagation, adversarial training, GANs, and GFlowNets. In the last case, LIR actually suggests a more natural loss, which we demonstrate improves GFlowNet convergence. Each method can be recovered as a specific instance of LIR by choosing a procedure to direct focus (attention and control). We implement this algorithm for discrete PDGs and study its properties on synthetically generated PDGs, comparing its behavior to the global optimization semantics of the full PDG.

Multi-agent large language model (LLM) architectures increasingly rely on response-level aggregation, such as Majority Voting (MAJ), to raise reasoning ceilings. However, in open environments, agents are highly susceptible to stealthy contextual corruption, such as targeted prompt injections. We reveal a critical structural vulnerability in current multi-agent systems: response-level aggregation collapses when corrupted agents form a local majority. Because voting aggregates fully-formed conclusions, it is blind to flawed intermediate logic. To overcome this systematic limitation, we propose the Token-Level Round-Robin (RR) Collaboration, where agents sequentially interleave generation within a shared auto-regressive context. We formalize this process as a discrete-time dynamical system, proving that token-level interleaving transitions aggregation from a brittle counting of final votes (a linear sum) to a dynamic, interwoven chain of logic (a non-linear operator product). Through this theoretical lens, we prove that the honest model's restorative pull can overpower adversarial corruptions, even when corrupted agents form a majority. We conduct an exhaustive empirical evaluation across diverse reasoning benchmarks and demonstrate that while MAJ collapses when corrupted agents reach a majority, RR maintains robust accuracy well beyond this critical threshold.

Navigating complex environments poses challenges for multi-agent systems, requiring efficient extraction of insights from limited information. In this paper, we introduce the Blackbox Oracle Information Learning (BOIL) process, a scalable solution for extracting valuable insights from the environment structure. Leveraging the Pagerank algorithm and common information maximization, BOIL facilitates the extraction of information to guide long-term agent behavior applicable to problems such as coverage, patrolling, and stochastic reachability. Through experiments, we demonstrate the efficacy of BOIL in generating strategy distributions conducive to improved performance over extended time horizons, surpassing heuristic approaches in complex environments.

We present RoIt-XMASA, a multilingual dataset that extends the Cross-lingual Multi-domain Amazon Sentiment Analysis to Italian and Romanian, comprising 36,000 labeled reviews across three domains (books, movies, and music) and 202,141 unlabeled samples. To address cross-lingual and cross-domain challenges, we propose a multi-target adversarial training framework that employs loss reversal with meta-learned coefficients to dynamically balance sentiment discrimination with domain and language invariance. XLM-R achieves an F1-score of 66.23% with our approach, outperforming the baseline by 4.64%. Few-shot evaluation shows that Llama-3.1-8B achieves 58.43% F1-score, revealing a meaningful trade-off between the efficiency of prompting-based approaches and the higher performance of task-specific fine-tuning.

Continuous glucose monitors (CGMs) used in diabetes care collect rich personal health data that could improve day-to-day self-management. However, current patient platforms only offer static summaries which do not support inquisitive user queries. Large language models (LLMs) could enable free-form inquiries about continuous glucose data, but deploying them over sensitive health records raises privacy and accuracy concerns. In this paper, we present CGM-Agent, a privacy-preserving framework for question answering over personal glucose data. In our design, the LLM serves purely as a reasoning engine that selects analytical functions. All computation occurs locally, and personal health data never leaves the user's device. For evaluation, we construct a benchmark of 4,180 questions combining parameterized question templates with real user queries and ground truth derived from deterministic program execution. Evaluating 6 leading LLMs, we find that top models achieve 94\% value accuracy on synthetic queries and 88\% on ambiguous real-world queries. Errors stem primarily from intent and temporal ambiguity rather than computational failures. Additionally, lightweight models achieve competitive performance in our agent design, suggesting opportunities for low-cost deployment. We release our code and benchmark to support future work on trustworthy health agents.

Safety-aligned large language models (LLMs) often generate refusal responses to harmless queries due to the over-refusal problem. However, existing methods for mitigating over-refusal cannot maintain a low refusal ratio for harmless queries while keeping a high refusal ratio for malicious ones. In this paper, we analyze how system prompts with varying safety levels affect LLM refusal behaviors when facing over-refusal queries. A key observation is that, when LLMs suffer from the over-refusal issue, non-refusal tokens remain present in the next-token candidate list, but the model systematically fails to select them, despite the generation of refusal tokens. Based on this observation, we propose a training-free and model-agnostic approach, Adaptive Contrastive Decoding (AdaCD), to mitigate over-refusal while maintaining LLM safety. First, AdaCD compares the output distributions of the LLM with or without an extreme safety system prompt to refine the refusal token distribution. Second, we introduce an adaptive contrastive decoding strategy that dynamically incorporates or removes the refusal token distribution, adaptively boosting the probability of selecting refusal or non-refusal tokens. Experimental results on five benchmark datasets show that, on average, AdaCD reduces the refusal ratio for over-refusal queries by 10.35%, yet still increases the refusal ratio for malicious queries by 0.13%. Code is available at https://github.com/OutdoorManofML/AdaCD.

The plasma environment around Mars is highly variable because it is strongly influenced by the solar wind. Accurate identification of plasma regions around Mars is important for the community studying solar wind-Mars interactions, region-specific plasma processes, and atmospheric escape. In this study, we develop a machine-learning-based classifier to automatically identify three key plasma regions--solar wind, magnetosheath, and induced magnetosphere--using only ion omnidirectional energy spectra measured by the MAVEN Solar Wind Ion Analyzer (SWIA). Two neural network architectures are evaluated: a multilayer perceptron (MLP) and a convolutional neural network (CNN) that incorporates short temporal sequences. Our results show that the CNN can reliably distinguish the three plasma regions, whereas the MLP struggles to separate the solar wind and magnetosheath. Therefore, the CNN-based approach provides an efficient and accurate framework for large-scale plasma region identification at Mars and can be readily applied to future planetary missions.

The present study aims to investigate a cluster cleaning algorithm that is both computationally simple and capable of solving the PU classification when the SCAR condition is unsatisfied. A secondary objective of this study is to determine the robustness of the LassoJoint method to perturbations of the SCAR condition. In the first step of our algorithm, we obtain cleaning labels from 2-means clustering. Subsequently, we perform logistic regression on the cleaned data, assigning positive labels from the cleaning algorithm with additional true positive observations. The remaining observations are assigned the negative label. The proposed algorithm is evaluated by comparing 11 real data sets from machine learning repositories and a synthetic set. The findings obtained from this study demonstrate the efficacy of the clustering algorithm in scenarios where the SCAR condition is violated and further underscore the moderate robustness of the LassoJoint algorithm in this context.

Model Context Protocol (MCP) is a rapidly adopted standard for defining and invoking external tools in LLM applications. The multi-layered architecture of MCP introduces new attack surfaces such as tool poisoning, in addition to traditional prompt injection. Existing defense systems suffer from limitations including high false positive rates, API dependency, or white-box access requirements. In this study, we propose CASCADE, a three-tiered cascaded defense architecture for MCP-based systems: (i) Layer 1 performs fast pre-filtering using regex, phrase weighting, and entropy analysis; (ii) Layer 2 conducts semantic analysis via BGE embedding with an Ollama Llama3 fallback mechanism; (iii) Layer 3 applies pattern-based output filtering. Evaluation on a dataset of 5,000 samples yielded 95.85% precision, 6.06% false positive rate, 61.05% recall, and 74.59% F1-score. Analysis across 31 attack types categorized into 6 tiers revealed high detection rates for data exfiltration (91.5%) and prompt injection (84.2%), while semantic attack (52.5%) and tool poisoning (59.9%) categories showed potential for improvement. A key advantage of CASCADE over existing solutions is its fully local operation, requiring no external API calls

Transformers encode structure in sequences via an expanding contextual history. However, their purely feedforward architecture fundamentally limits dynamic state tracking. State tracking -- the iterative updating of latent variables reflecting an evolving environment -- involves inherently sequential dependencies that feedforward networks struggle to maintain. Consequently, feedforward models push evolving state representations deeper into their layer stack with each new input step, rendering information inaccessible in shallow layers and ultimately exhausting the model's depth. While this depth limit can be bypassed by dynamic depth models and by explicit or latent thinking that externalizes state representations, these solutions are computationally and memory inefficient. In this article, we argue that temporally extended cognition requires refocusing from explicit thought traces to implicit activation dynamics via recurrent architectures. We introduce a taxonomy of recurrent and continuous-thought transformer architectures, categorizing them by their recurrence axis (depth versus step) and their ratio of input tokens to recurrence steps. Finally, we outline promising research directions, including enhanced state-space models and coarse-grained recurrence, to better integrate state tracking into modern foundation models.

Accurate segmentation of gastrointestinal (GI) organs in magnetic resonance enterography (MRE) is critical for diagnosing inflammatory bowel disease (IBD). However, anatomical variability, class imbalance, and low tissue contrast hinder reliable automation. This study proposes a dual-stage deep learning framework for organ-specific segmentation of GI structures from coronal MRE images to address these challenges. A publicly available MRE dataset of 3,195 coronal T2-weighted HASTE slices from 114 IBD patients was used. Initially, a DenseNet201-UNet++ model generated coarse masks for ROI extraction. A DenseNet121-SelfONN-UNet model was then trained on organ-specific patches. Extensive data augmentation, normalization, five-fold cross-validation, and class-specific weighting were applied to mitigate severe class imbalance, particularly for the appendix. The initial stage achieved strong organ localization but underperformed for the appendix; class weighting improved its DSC from 6.76% to 85.76%. The second-stage DenseNet121-SelfONN-UNet significantly enhanced segmentation across all GI structures, with notable DSC gains (cecum +23.62%, sigmoid +18.57%, rectum +17.99%, small intestine +16.06%). Overall, the framework achieved mDSC of 88.99%, mIoU of 84.76%, and mHD95 of 6.94 mm, outperforming all baselines. This framework demonstrates the effectiveness of a coarse-to-fine, organ-aware segmentation strategy for intestinal MRE. Despite higher computational cost, it shows strong potential for clinical translation and enables anatomically informed diagnostic tools in gastroenterology.

Frontier large language models generate clinically accurate outputs, but their citations are often fabricated. We term this the Provenance Gap. We tested five frontier LLMs across 36 clinician-validated scenarios for three rare neuromuscular disease pairs. No model produced a clinically relevant PubMed identifier without prompting. When explicitly asked to cite, the best model achieved 15.3% relevant PMIDs; the majority resolved to real publications in unrelated fields. We present HEG-TKG (Hierarchical Evidence-Grounded Temporal Knowledge Graphs), a system that grounds clinical claims in temporal knowledge graphs built from 4,512 PubMed records and curated sources with quality-tier stratification and 1,280 disease-trajectory milestones. In a controlled three-arm comparison using the same synthesis model, HEG-TKG matches baseline clinical feature coverage while achieving 100% evidence verifiability with 203 inline citations. Guideline-RAG, given overlapping source documents as raw text, produces zero verifiable citations. LLM judges cannot distinguish fabricated from verified citations without PubMed audit data. Independent clinician evaluation confirms the verifiability advantage (Cohen's d = 1.81, p < 0.001) with no degradation on safety or completeness. A counterfactual experiment shows 80% resistance to injected clinical errors with 100% detectability via citation trace. The system deploys on-premise via open-source models so patient data never leaves institutional infrastructure.

Large language models (LLMs) often produce confident yet incorrect responses, and uncertainty quantification is one potential solution to more robust usage. Recent works routinely rely on self-consistency to estimate aleatoric uncertainty (AU), yet this proxy collapses when models are overconfident and produce the same incorrect answer across samples. We analyze this regime and show that cross-model semantic disagreement is higher on incorrect answers precisely when AU is low. Motivated by this, we introduce an epistemic uncertainty (EU) term that operates in the black-box access setting: EU uses only generated text from a small, scale-matched ensemble and is computed as the gap between inter-model and intra-model sequence-semantic similarity. We then define total uncertainty (TU) as the sum of AU and EU. In a comprehensive study across five 7-9B instruction-tuned models and ten long-form tasks, TU improves ranking calibration and selective abstention relative to AU, and EU reliably flags confident failures where AU is low. We further characterize when EU is most useful via agreement and complementarity diagnostics.

When multiple LLM coding agents share a rate-limited API endpoint, they exhibit resource contention patterns analogous to unscheduled OS processes competing for CPU, memory, and I/O. In a motivating incident, 3 of 11 parallel agents died from connection resets and HTTP 502 errors - a 27% failure rate - despite the API having sufficient aggregate capacity to serve all 11 sequentially. We present HIVEMIND, a transparent HTTP proxy that applies five OS-inspired scheduling primitives - admission control, rate-limit tracking, AIMD backpressure with circuit breaking, token budget management, and priority queuing - to eliminate the failure modes caused by uncoordinated parallel execution. The proxy requires zero modifications to existing agent code and supports Anthropic, OpenAI, and local model APIs via auto-detected provider profiles. Our evaluation across seven scenarios (5-50 concurrent agents) shows that uncoordinated agents fail at 72-100% rates under contention, while HIVEMIND reduces failures to 0-18% and eliminates 48-100% of wasted compute. An ablation study reveals that transparent retry - not admission control - is the single most critical primitive, but the primitives are most effective in combination. Real-world validation against Ollama confirms that HIVEMIND adds under 3ms of proxy overhead per request. The system is open-source under the MIT license.

Ising machines -- special-purpose hardware for heuristically solving Ising optimization problems -- based on probabilistic bits (p-bits) have been established as a promising alternative to heuristic optimization algorithms run on conventional computers. However, it has -- until now -- been thought that Ising spins that are connected in probabilistic Ising machines cannot be updated in parallel without ruining the machine's solving ability. This has been a major challenge for using probabilistic Ising machines as fast solvers for densely connected problems. Here, we circumvent this by introducing a modified Ising spin dynamics with an added inertia term, and verify in algorithm simulations, FPGA hardware emulation, and FPGA experiments that it enables fully parallel, synchronous updates while improving rather than degrading success probability. We evaluated on various types of abstract (Max-Cut and Sherrington-Kirkpatrick-model) and application-derived (MIMO, wireless detection) dense Ising benchmark instances. Performing fully parallel updates results in a speed advantage that grows faster than linearly with the number of spins, giving rise to large time-to-solution increases for practical problem sizes. For both Max-Cut and the SK-1 model at a problem size of 200, our approach achieved an average speedup of $\approx 35\times$, with the best single-instance speedup reaching $150\times$. As an example of the practical utility of our approach in an application where speed is critical, we further show by co-designing the algorithm dynamics with the hardware implementation -- co-optimizing for solver ability and silicon resource usage -- that probabilistic Ising machines based on our approach satisfy the stringent solution quality and latency/throughput requirements for real-time MIMO detection in modern 5G cellular wireless networks while using a practically reasonable silicon area.

Coreference Resolution (CR) is a fundamental NLP task critical for long-form tasks as information extraction, summarization, and many business applications. However, CR methods originally designed for English struggle with Morphologically Rich Languages (MRLs), where mention boundaries do not necessarily align with word boundaries, and a single token may consist of multiple anaphors. CR modeling and evaluation protocols standardly assume that, as in English, words and mentions mostly align. However, this assumption breaks down in MRLs, particularly in the context of LLMs' raw-text processing and end-to-end tasks. To assess and address this challenge, we introduce {\em KibutzR}, the first comprehensive CR dataset for Modern Hebrew, an MRL rich with complex words and pronominal clitics. We deliver an annotated dataset that identifies mentions at word, sub-word and multi-word levels, and propose an evaluation protocol that directly addresses word/morpheme boundary discrepancies. Our experiments show that contemporary LLMs perform significantly worse on Hebrew than on English, and that performance degrades on raw unsegmented text. Crucially, we show an inverse performance-trend in Hebrew relative to English, where smaller encoders perform far better than contemporary decoder models, leaving ample space for investigation and improvement. We deliver a new benchmark for Hebrew coreference resolution and a segmentation-aware evaluation protocol to inform future work on other MRLs.

Magnetic Resonance Imaging (MRI) is vital for prostate cancer (PCa) diagnosis. While advanced techniques such as Hybrid Multi-dimensional MRI (HM-MRI) have enhanced diagnostic capabilities, the significant need remains for robust, automated Artificial Intelligence (AI)-based detection methods. In this study, we combine quantitative HM-MRI of tissue composition with an AI-based neural network. We propose the Hadamard-Bias Network plus ResNet18 (HBR-Net-18), a two-stage AI framework for PCa detection. In the first stage, a Hadamard U-Net-based algorithm suppresses intensity inhomogeneities (bias fields) across six parametric HM-MRI maps generated via a Physics-Informed Autoencoder (PIA). In the second stage, a Residual Network (ResNet-18) performs patch-level classification. The framework utilizes overlapping 11-by-11 patches, incorporating both 2D intra-slice and 3D inter-slice (adjacent-slice) information to improve spatial consistency. Our experimental results demonstrate that HB-Net achieves balanced sensitivity and specificity, significantly outperforming conventional radiomics-based approaches and baseline CNN models, highlighting its potential for clinical deployment.

Motivated by the challenge presented by non-Markovian objectives in reinforcement learning (RL), we present a novel framework to track and represent the progress of autonomous agents through complex, multi-stage tasks. Given a specification in finite linear temporal logic (LTL), the framework establishes a 'tracking vector' which updates at each time step in a trajectory rollout. The values of the vector represent the status of the specification as the trajectory develops, assigning true, false, or 'open' labels (where 'open' is used for indeterminate cases). Applied to an LTL formula tree, the tracking vector can be used to encode detailed information about how a task is executed over a trajectory, providing a potential tool for new performance metrics, diverse exploration, and reward shaping. In this paper, we formally present the framework and algorithm, collectively named Live LTL Progress Tracking, give a simple working example, and demonstrate avenues for its integration into RL models. Future work will apply the framework to problems such as task-space exploration and diverse solution-finding in RL.

Tokenization is the first step in every language model (LM), yet it never takes the sounds of words into account. We investigate how tokenization influences text-only LMs' ability to represent phonological knowledge. Through a series of probing experiments, we show that subword-based tokenization systematically weakens the encoding of both local (e.g., rhyme) and global (e.g., syllabification) phonological features. To quantify this effect, we introduce the syllabification-tokenization alignment distance (STAD), a metric that measures the misalignment between a model's tokenization and the natural syllable boundaries of words, and find that higher misalignment correlates with poorer phonological representations, providing a simple diagnostic for phonology-aware tokenization. To address these limitations, we propose a lightweight IPA-based fine-tuning method that infuses phonological awareness into LMs, leading to consistent improvements across three phonology-related tasks while largely preserving math and general reasoning ability, with 1.1\% and 0.9\% drops on GSM8K and MMLU, respectively.

Modern AI models are growing rapidly in size and redundancy, leading to significant storage and distribution challenges in model hubs. We present TensorHub, a tensor-centric system for reducing storage overhead through fine-grained deduplication and compression. TensorHub leverages tensor-level fingerprinting and clustering to identify redundancy across models without requiring annotations. Our design enables efficient storage reduction while preserving model usability and performance. Experiments on real-world model repositories demonstrate substantial storage savings with minimal overhead.

Benchmarking of open-source LLMs for hardware design focuses on which LLMs to use, while treating inference-time decoding configuration as a secondary concern. This work shows that it matters more how an LLM is configured than which model is selected. Benchmarking 26 open-source LLMs on VerilogEval and RTLLM with synthesis-in-the-loop evaluation, the study first maps the current capability landscape and then conducts an extensive 108-configuration hyperparameter sweep on three prominent models. The sweep reveals absolute pass-rate gaps of up to 25.5% between the best and worst settings for the same LLM, which is 5x larger than the average spread observed across various model families under their respective default configurations. Ranking all configurations by Spearman's $ρ$ across the two benchmark suites yields near-zero correlation, demonstrating that optimal configurations do not transfer. These results show that benchmarking conducted under default hyperparameters confounds model capabilities with configuration effects. Realizing the full potential of open-source LLMs for RTL generation requires architecture and benchmark aware hyperparameter selection, as enabled by the proposed methodology.

Edge applications increasingly demand custom hardware, yet Field-Programmable Gate Array (FPGA) design requires expertise that domain engineers lack. Large Language Models (LLMs) promise to bridge this gap through zero-knowledge hardware programming, where users describe circuits in natural language and an LLM compiles them to a hardware intermediate representation (IR) targeting silicon. Modeling this flow as a cascade of binary filters, this work demonstrates that IR choice, not model choice, is the dominant factor governing end-to-end success, a phenomenon termed the representation bottleneck. An evaluation of three frontier LLMs across six IRs spanning Verilog, VHDL, Chisel, Bluespec, PyMTL3, and HLS C on 202 tasks through a pipeline of compilation, simulation, FPGA synthesis on a Lattice iCE40UP5K, and LLM-based repair shows that simulation pass rates range from 3% to 88% across IRs but typically vary less than 1.25x across models within any single IR. On the resource-constrained iCE40, LLM designs achieve a higher conditional FPGA pass rate than reference solutions, 86.5% vs. 68.7%, not because they are better but because a simplicity bias makes them small enough to fit. The analysis reveals an accessibility-competence paradox: the most user-friendly IRs yield the worst LLM performance, suggesting that optimal IR selection will evolve as LLM capabilities grow.

As AI-assisted development tools proliferate, developers face a growing challenge: understanding the cost, quality, and behavioral patterns of AI interactions across their workflow. We present a unified approach to AI observability for developer productivity tools, combining real-time token tracking, configurable model pricing registries, response validation, and cost analytics into a single-pane dashboard. Our work synthesizes two complementary systems -- Workstream, a developer productivity dashboard that centralizes pull requests, Jira tasks, and AI code reviews; and an AI observability summarizer that monitors inference workloads with Prometheus-backed metrics and multi-provider LLM gateways. We describe the architectural patterns adopted, the implementation of real token tracking from provider APIs (replacing heuristic estimation), a 24-model pricing registry, response validation pipelines, LLM-powered review intelligence, and exportable reports. Our evaluation on a six-month development workflow shows the system captures per-review cost with less than 2% variance from provider billing and reduces time-to-insight for AI usage patterns by an order of magnitude compared to manual tracking.

Long-horizon large language model (LLM) agents are fundamentally limited by context. As interactions become longer, tool descriptions, retrieved memories, and raw environmental feedback accumulate and push out the information needed for decision-making. At the same time, useful experience gained from tasks is often lost across episodes. We argue that long-horizon performance is determined not by context length, but by how much decision-relevant information is maintained within a finite context budget. We present GenericAgent (GA), a general-purpose, self-evolving LLM agent system built around a single principle: context information density maximization. GA implements this through four closely connected components: a minimal atomic tool set that keeps the interface simple, a hierarchical on-demand memory that only shows a small high-level view by default, a self-evolution mechanism that turns verified past trajectories into reusable SOPs and executable code, and a context truncation and compression layer that maintains information density during long executions. Across task completion, tool use efficiency, memory effectiveness, self-evolution, and web browsing, GA consistently outperforms leading agent systems while using significantly fewer tokens and interactions, and it continues to evolve over time. Project: https://github.com/lsdefine/GenericAgent

Continual learning aims to update models under distribution shift without forgetting, yet many high-stakes deployments, such as healthcare, also require interpretability. In practice, models that adapt well (e.g., deep networks) are often opaque, while models that are interpretable (e.g., decision trees) are brittle under shift, making it difficult to achieve both properties simultaneously. In response, we propose Tree of Concepts, an interpretable continual learning framework that uses a shallow decision tree to define a fixed, rule-based concept interface and trains a concept bottleneck model to predict these concepts from raw features. Continual updates act on the concept extractor and label head while keeping concept semantics stable over time, yielding explanations that do not drift across sequential updates. On multiple tabular healthcare benchmarks under continual learning protocols, our method achieves a stronger stability-plasticity trade-off than existing baselines, including replay-enhanced variants. Our results suggest that structured concept interfaces can support continual adaptation while preserving a consistent audit interface in non-stationary, high-stakes domains.

Recent Multimodal Large Language Models (MLLMs) have demonstrated strong performance on vision-language understanding tasks, yet their inference efficiency is often hampered by the large number of visual tokens, particularly in high-resolution or multi-image scenarios. To address this issue, we propose EvoComp, a visual token compression framework that significantly reduces token count while preserving task accuracy. EvoComp introduces a lightweight encoder-only transformer-based compressor that selects the most informative and non-redundant visual tokens by jointly considering visual and textual contexts. A core challenge lies in providing effective supervision for training the compressor. To this end, we design an evolutionary labeling strategy that searches for token subsets minimizing the MLLM's output loss, while enforcing semantic diversity through vocabulary-based token grouping. We further train the compressor using a tailored loss function combining the GHM loss to mitigate class and difficulty imbalance, and a cosine similarity regularization to encourage semantic separation between retained and discarded tokens. Extensive experiments across multiple vision-language benchmarks show that EvoComp outperforms existing methods based on attention or similarity heuristics. Notably, it retains 99.3% of the original accuracy under 3x token compression and delivers up to 1.6x speedup on mobile devices.

The interpretation of implicit meanings is an integral aspect of human communication. However, this framework may not transfer to interactions with Large Language Models (LLMs). To investigate this, we introduce the task of Implicit Information Extraction (IIE) and propose an LLM-based IIE pipeline that builds a structured knowledge graph from a context sentence by extracting relational triplets, validating implicit inferences, and analyzing temporal relations. We evaluate two LLMs against crowdsourced human judgments on two datasets. We find that humans agree with most model triplets yet consistently propose many additions, indicating limited coverage in current LLM-based IIE. Moreover, in our experiments, models appear to be more conservative about implicit inferences than humans in socially rich contexts, whereas humans become more conservative in shorter, fact-oriented contexts. Our code is available at https://github.com/Antonio-Dee/IIE_from_LLM.

When users seek social support from chatbots, they disclose their situation gradually, yet most evaluations of supportive LLMs rely on single-turn, fully specified prompts. We introduce a multi-turn simulation framework that closes this gap. Support-seeking narratives from five Reddit communities are decomposed into ordered fragments and revealed turn by turn to a language model. Each response is coded with the Social Support Behavior Code (SSBC), an established multi-label taxonomy that captures the composition of support, rather than a single quality score. To ask whether support choices track the model's own construal of user distress, we use linear probes on hidden representations to estimate this internal signal without altering the generation context. Across two mid-scale models (Llama-3.1-8B, OLMo-3-7B) and more than 6,200 turns, support composition shifts systematically with estimated distress: teaching declines as estimated distress rises, a finding that replicates across architectures, while increases in affective and esteem-oriented strategies (such as validation) are suggestive but model-specific and rest on noisier annotations. Community context independently shapes behavior, tracking topic and discourse norms rather than demographic categories. These trajectory-level dynamics, invisible to single-turn evaluation, motivate multi-turn auditing frameworks for socially sensitive applications.

Task arithmetic provides an efficient, training-free way to edit pre-trained models, yet lacks a fundamental theoretical explanation for its success. The existing concept of ``weight disentanglement" describes the ideal outcome of non-interfering task composition but does not reveal its underlying cause. Crucially, what intrinsic properties of the pre-trained model ($θ_0$) or the task vectors ($τ_t$) enable this disentanglement remains underexplored. In this paper, we introduce Task-Feature Specialization (TFS), a model's ability to allocate distinct internal features to different tasks, as the fundamental principle. We first prove that TFS is a sufficient condition for weight disentanglement. More importantly, we find that TFS also gives rise to an observable geometric consequence: weight vector orthogonality. This positions TFS as the common cause for both the desired functional outcome (disentanglement) and a measurable geometric property (orthogonality). This relationship provides the key insight for our method: since the abstract TFS property is intractable to enforce directly, we can instead promote weight disentanglement by shaping its concrete geometric consequence, orthogonality. Therefore, we propose OrthoReg, a simple and effective regularization method that actively enforces an internal orthogonal structure on weight updates ($ΔW$) that constitute $τ_t$ during fine-tuning. And we theoretically prove that OrthoReg promotes disentanglement. Extensive experiments demonstrate that OrthoReg consistently and significantly enhances the performance of various task arithmetic methods. Code is available at \href{https://github.com/RL-MIND/OrthoReg}{https://github.com/RL-MIND/OrthoReg}.

Reinforcement fine-tuning improves the reasoning ability of large language models, but it can also encourage them to answer unanswerable queries by guessing or hallucinating missing information. Existing abstention methods either train models to produce generic refusals or encourage follow-up clarifications without verifying whether those clarifications identify the key missing information. We study queries that are clear in meaning but cannot be reliably resolved from the given information, and argue that a reliable model should not only abstain, but also explain what is missing. We propose a clarification-aware RLVR reward that, while rewarding correct answers on answerable queries, jointly optimizes explicit abstention and semantically aligned post-refusal clarification on unanswerable queries. Using this reward, we train Abstain-R1, a 3B model that improves abstention and clarification on unanswerable queries while preserving strong performance on answerable ones. Experiments on Abstain-Test, Abstain-QA, and SelfAware show that Abstain-R1 substantially improves over its base model and achieves unanswerable-query behavior competitive with larger systems including DeepSeek-R1, suggesting that calibrated abstention and clarification can be learned through verifiable rewards rather than emerging from scale alone.

The autonomous synthesis of deep research reports represents a critical frontier for Large Language Models (LLMs), demanding sophisticated information orchestration and non-linear narrative logic. Current approaches rely on rigid predefined linear workflows, which cause error accumulation, preclude global restructuring from subsequent insights, and ultimately limit in-depth multimodal fusion and report quality. We propose CogGen, a Cognitively inspired recursive framework for deep research report Generation. Leveraging a Hierarchical Recursive Architecture to simulate cognitive writing, CogGen enables flexible planning and global restructuring. To extend this recursivity to multimodal content, we introduce Abstract Visual Representation (AVR): a concise intent-driven language that iteratively refines visual-text layouts without pixel-level regeneration overhead. We further present CLEF, a Cognitive Load Evaluation Framework, and curate a new benchmark from Our World in Data (OWID). Extensive experiments show CogGen achieves state-of-the-art results among open-source systems, generating reports comparable to professional analysts' outputs and surpassing Gemini Deep Research. Our code and dataset are available at https://github.com/NJUNLP/CogGen.

Diffusion large language models (dLLMs) enable parallel text generation by iteratively denoising a fully masked sequence, unmasking a subset of masked tokens at each step. Existing decoding strategies rely on static confidence metrics computed at a single denoising step, ignoring temporal history and often leading to premature unmasking of unstable tokens. In this work, we theoretically establish that a token's temporal instability, quantified by the KL divergence between consecutive prediction distributions, provides a strict lower bound on its mutual information with the remaining masked context, indicating that temporally unstable tokens are inherently unsafe to unmask. Based on this insight, we propose Stability-Weighted Decoding (SWD), a training-free, plug-and-play strategy that incorporates temporal stability into token scoring and acts as a universal modulator for arbitrary score-based decoding policies. Experiments on code generation and mathematical reasoning benchmarks demonstrate that SWD consistently improves generation accuracy across representative scoring metrics and selection policies, and exhibits exceptional robustness, maintaining a significant performance lead over standard baselines across varying acceleration ratios.

Linear convergence of first-order methods is typically characterized by global optimization conditions whose constants reflect worst-case geometry of the ambient space. In high-dimensional or structured problems, these global constants can be arbitrarily conservative and fail to capture the geometry actually encountered by optimization trajectories. In this paper, we develop a trajectory-restricted framework for linear convergence based on localized geometric regularity. We introduce restricted variants of the Polyak--Łojasiewicz inequality, error bound, and quadratic growth conditions that are required to hold only on subsets of the domain. We show that classical convergence guarantees extend under these localized conditions, and in key cases, we develop new arguments that yield explicit relationships between the corresponding constants. The resulting rates are governed by geometric quantities associated with the regions traversed by the algorithm. For polyhedral composite problems, we prove that convergence is controlled by restricted Hoffman constants corresponding to the active polyhedral faces visited along the trajectory. Once the iterates enter a well-conditioned face, the effective condition number improves accordingly. Our work provides a geometric quantification for fast local convergence after active-set or manifold identification and more broadly suggests that linear convergence is fundamentally governed by the geometry of the subsets explored by the algorithm, rather than by worst-case global conditioning.

Coherent systems are representative of many practical applications, ranging from infrastructure networks to supply chains. Probabilistic evaluation of such systems remains challenging, however, because existing decomposition-based methods scale poorly as the number of components grows. To address this limitation, this study proposes the Reference-state System Reliability (RSR) method. Like existing approaches, RSR characterises the boundary between different system states using reference states in the component-state space. Where it departs from these methods is in how the state space is explored: rather than using reference states to decompose the space into disjoint hypercubes, RSR uses them to classify Monte Carlo samples, making computational cost significantly less sensitive to the number of reference states. To make this classification efficient, samples and reference states are stored as matrices and compared using batched matrix operations, allowing RSR to exploit the advances in high-throughput matrix computing driven by modern machine learning. We demonstrate that RSR evaluates the system-state probability of a graph with 119 nodes and 295 edges within 10~seconds, highlighting its potential for real-time risk assessment of large-scale systems. We further show that RSR scales to problems involving hundreds of thousands of reference states -- well beyond the reach of existing methods -- and extends naturally to multi-state systems. Nevertheless, when the number of boundary reference states grows exceedingly large, RSR's convergence slows down, a limitation shared with existing reference-state-based approaches that motivates future research into learning-based representations of system-state boundaries.

High-performance computing (HPC) systems must support fast-moving software stacks, especially in AI/ML, while preserving scheduler control, scalable startup, and production performance. Yet many HPC container solutions rely on specialized runtime stacks that weaken continuity with mainstream cloud-native workflows and require ongoing effort to sustain compatibility with the evolving upstream ecosystem. We argue that HPC should specialize the integration layer while keeping the container engine aligned with upstream container evolution. We present Sarus Suite, an upstream-aligned HPC container architecture built around an unchanged Podman engine. Sarus Suite adds the HPC-specific functionality needed for production use through complementary system layers for declarative runtime specification, scheduler-native execution, scalable shared-image access, and standards-based host capability injection. We evaluate Sarus Suite on a Cray EX GH200 system using communication-intensive HPC workloads, large scale AI training, metadata-heavy startup workloads, and container startup measurements. Across PyFR, SPH-EXA, Megatron-LM, and Pynamic, Sarus Suite matches the performance and scaling of the production Enroot+Pyxis baseline while delivering consistently faster per-node container startup. The architecture also enables direct use of upstream OCI images, including NGC-based images, and supports cloud-native multi-container workflows expressed through Kubernetes manifests. These results show that HPC-grade containers do not require an HPC-specific runtime, provided that scheduler semantics, scalable image access, and host integration are implemented in explicit system layers. This preserves upstream continuity and software agility while maintaining scheduler control, scalability, and production performance.

Conventional air traffic control divides airspace into specific regions, creating a scaling bottleneck as traffic grows. Choosing how to partition airspace is not straightforward because grid size affects workload, handoff frequency, and the capacity of whatever coordination mechanism operates within each sector. We present a three stage pipeline that automates sectorization and sector coordination while preserving human oversight. First, a two stage XGBoost classifier predicts the optimal 3D grid configuration from 23 location-agnostic traffic features, achieving 91.38% accuracy on a 65,000 sample dataset derived from Federal Aviation Administration System Wide Information Management replays. Second, a leaderless Paxos consensus protocol lets aircraft coordinate sector entries among themselves, maintaining above 96% entry success with low near mid-air collision rates across all tested configurations. Third, Bayesian Optimization with a Gaussian Process surrogate tunes eight protocol parameters per airport in 50 trials, revealing that each traffic environment requires a qualitatively different configuration. The resulting pipeline offers a practical path toward scalable, autonomous airspace management as traffic demand outpaces controller capacity.

A key challenge in distributed coalition formation within characteristic function games is determining how to allocate the calculation of coalition values across a set of agents. The number of possible coalitions grows exponentially with the number of agents, and existing distributed approaches may produce uneven or redundant allocations, or assign coalitions to agents that are not themselves members. In this article, we present the \emph{Necklace-based Distributed Coalition Algorithm} (N-DCA), a communication-free algorithm in which each agent independently determines its own coalition value calculation allocation using only its identifier and the total number of agents. The approach builds on the notion of Increment Arrays (IAs), for which we develop a complete mathematical framework: equivalence classes under circular shifts, periodic IAs, and a rotated designation scheme with formal load-balance guarantees (tight bounds). We establish a bijection between canonical representative IAs and two-colour combinatorial necklaces, enabling the use of efficient necklace generation algorithms to enumerate allocations in constant amortised time. N-DCA is, to the best of our knowledge, the only distributed coalition value calculation algorithm for unrestricted characteristic function games to provably satisfy five desirable properties: no inter-agent communication, equitable allocation, no redundancy, balanced load, and self-interest. An empirical evaluation against DCVC (Rahwan and Jennings 2007) demonstrates that, although DCVC is faster by a constant factor, this difference becomes negligible under realistic characteristic-function evaluation costs, while N-DCA offers advantages in working memory, scalability, and the self-interest guarantee.

When does an LLM controller outperform rule-based traversal for knowledge graph exploration? We study this question through RLM-on-KG, a retrieval system that treats an LLM as an autonomous navigator over an RDF-encoded mention graph for grounded question answering. Unlike GraphRAG pipelines that rely on offline LLM indexing, RLM-on-KG performs entity-first, multi-hop exploration at query time using deterministic graph construction and a fixed tool set. Our central finding is a conditional advantage: the value of LLM control depends on evidence scatter and tool-calling sophistication. The paper's core claim is LLM control versus heuristic traversal, not a generic win over GraphRAG. On GraphRAG-Bench Novel (519 questions), Gemini 2.0 Flash achieves +2.47 pp F1 over a rule-based heuristic baseline (p < 0.0001), but only +0.16 pp over a GraphRAG-local variant (not significant). With a stronger controller, Claude Haiku 4.5, the gain over heuristic grows to +4.37 pp (p < 0.001) and extends to a +2.42 pp significant improvement over GraphRAG-local (p < 0.001). The gain is largest when gold evidence is scattered across 6-10 chunks (+3.21 pp) and smallest for concentrated evidence (+1.85 pp). Cross-scale validation on MuSiQue confirms that the LLM-over-heuristic advantage transfers, with expected attenuation on smaller per-question graphs. The core architectural insight is the separation of candidate discovery from ranking: the LLM adds value through exploration breadth, while final evidence selection is best handled by pure vector re-ranking. Beyond retrieval, exploration traces provide a proposed stress-test harness for structured data quality, yielding diagnostics for coverage, connectivity, provenance, and queryability.

Modern software engineers operate across 5-10 disconnected tools daily: GitHub, GitLab, Jira, Slack, calendar applications, CI dashboards, AI coding assistants, and container platforms. This fragmentation creates cognitive overhead that interrupts deep work and delays response to critical engineering signals. We present Workstream, an open-source, local-first developer command center that aggregates pull requests, task management, calendar, AI-powered code review, historical review intelligence, repository AI-readiness scoring, and agent observability into a single interface. We describe the system architecture, a novel 5-category AI readiness scoring algorithm, a review intelligence pipeline that mines historical PR reviews for team-specific patterns, and an agent observability layer implementing the Model Context Protocol (MCP), Agent-to-Agent (A2A), and Agent Observability Protocol (AOP). Through a case study of applying the tool to its own development, we demonstrate measurable improvements in AI-readiness scores (48 to 98 on our internal scanner; 41.6 to 73.7 on the independent agentready CLI). Workstream is released as open source under the Apache 2.0 license at https://github.com/happybhati/workstream.

Scalable Vector Graphics (SVGs) function both as visual images and as structured code that encode rich geometric and layout information, yet most methods rasterize them and discard this symbolic organization. At the same time, recent sentence embedding methods produce strong text representations but do not naturally extend to visual or structured modalities. We propose a training-free, instruction-guided multimodal embedding framework that uses a Multimodal Large Language Model (MLLM) to map text, raster images, and SVG code into an aligned embedding space. We control the direction of embeddings through modality-specific instructions and structural SVG cues, eliminating the need for learned projection heads or contrastive training. Our method has two key components: (1) Multimodal Explicit One-word Limitation (mEOL), which instructs the MLLM to summarize any multimodal input into a single token whose hidden state serves as a compact semantic embedding. (2) A semantic SVG rewriting module that assigns meaningful identifiers and simplifies nested SVG elements through visual reasoning over the rendered image, exposing geometric and relational cues hidden in raw code. Using a repurposed VGBench, we build the first text-to-SVG retrieval benchmark and show that our training-free embeddings outperform encoder-based and training-based multimodal baselines. These results highlight prompt-level control as an effective alternative to parameter-level training for structure-aware multimodal retrieval. Project page: https://scene-the-ella.github.io/meol/

Pluralism alignment with AI has the sophisticated and necessary goal of creating AI that can coexist with and serve morally multifaceted humanity. Research towards pluralism alignment has many efforts in enhancing the learning of large language models (LLMs) to accomplish pluralism. Although this is essential, the robustness of LLMs to produce moral content over pluralistic values is still under exploration.Inspired by the astonishing persuasion abilities via jailbreak prompts, we propose to leverage jailbreak attacks to study LLMs' internal pluralistic values. In detail, we develop a morality dataset with 10.3K instances in two categories: Value Ambiguity and Value Conflict. We further formalize four adversarial attacks with the constructed dataset, to manipulate LLMs' judgment over the morality questions. We evaluate both the large language models and guardrail models which are typically used in generative systems with flexible user input. Our experiment results show that there is a critical vulnerability of LLMs and guardrail models to these subtle and sophisticated moral-aware attacks.

Large Language Models (LLMs) have demonstrated excellent performance in general language understanding, generation and other tasks. However, when fine-tuning for specific domain tasks, the general knowledge accumulated in the pre-training phase is often partially overwritten or forgotten due to parameter updates, which severely limits the generalization ability and transferability of LLMs. Traditional fine-tuning strategies mostly train on the entire parameter space, ignoring the heterogeneity of model parameters, that is, some parameters are extremely important for general tasks, while other parameters are more sensitive to specific tasks. To alleviate the above problems, this paper innovatively proposes a parameter element importance evaluation method, which divides parameters into "core parameters" and "non-core parameters" by distinguishing the importance of parameters for general language ability tasks and specific domain tasks, and fixes the core parameters during fine-tuning, and only fine-tunes the non-core parameters. Extensive experiments on scientific, medical and physical tasks using GPT-J and LLaMA-3 show that our method can mitigate catastrophic forgetting while enhancing the adaptability of the model.

We present E2E-WAVE, the first end-to-end learned waveform generation system for underwater video multicasting. Acoustic channels exhibit 20--46% bit error rates where forward error correction becomes counterproductive -- LDPC increases rather than decreases errors beyond its decoding threshold. E2E-WAVE addresses this by embedding semantic similarity directly into physical layer waveforms: when decoding errors are unavoidable, the system preferentially selects semantically similar tokens rather than arbitrary corruption. Combining VideoGPT tokenization (1024x compression) with a trainable waveform bank and fully differentiable OFDM transmission, E2E-WAVE achieves +5 dB (19.26%) PSNR and +0.10 (14.28%) SSIM over the strongest FEC-protected baseline in less challenging underwater channel (NOF1) while delivering real-time 16 FPS video at 128x128 resolution over 2.3 kbps channels -- impossible for conventional digital modulation. The performance gap only increases in harsher channels (BCH1, NCS1). Trained on a single channel, E2E-WAVE generalizes to unseen underwater environments without retraining, while HEVC fails at sub-5 kbps rates and SoftCast's AWGN assumptions collapse on frequency-selective channels.

Spiking neural operators are appealing for neuromorphic edge computing because event-driven substrates can, in principle, translate sparse activity into lower latency and energy. Whether that advantage survives deployment on commodity edge-GPU software stacks, however, remains unclear. We study this question on a Jetson Orin Nano 8 GB using five pretrained variable-spiking wavelet neural operator (VS-WNO) checkpoints and five matched dense wavelet neural operator (WNO) checkpoints on the Darcy rectangular benchmark. On a reference-aligned path, VS-WNO exhibits substantial algorithmic sparsity, with mean spike rates decreasing from 54.26% at the first spiking layer to 18.15% at the fourth. On a deployment-style request path, however, this sparsity does not reduce deployed cost: VS-WNO reaches 59.6 ms latency and 228.0 mJ dynamic energy per inference, whereas dense WNO reaches 53.2 ms and 180.7 mJ, while also achieving slightly lower reference-path error (1.77% versus 1.81%). Nsight Systems indicates that the request path remains launch-dominated and dense rather than sparsity-aware: for VS-WNO, cudaLaunchKernel accounts for 81.6% of CUDA API time within the latency window, and dense convolution kernels account for 53.8% of GPU kernel time; dense WNO shows the same pattern. On this Jetson-class GPU stack, spike sparsity is measurable but does not reduce deployed cost because the runtime does not suppress dense work as spike activity decreases.

Deception detection is of great significance for ensuring information security and conducting public opinion analysis, with personality factors and emotion cues playing a critical role. However, existing methods lack sample-level dynamic annotations for emotions and personality.In this paper, we propose an innovative multi-model multi-prompt annotation scheme and a strict label quality evaluation standard, and establish a multimodal joint detection dataset DDEP for deception, emotion, and personality. Meanwhile, we propose Rel-DDEP, an adaptive reliability-weighted fusion framework. Our framework quantifies uncertainty by mapping modal features to a high-dimensional Gaussian distribution space. It then performs reliability-weighted fusion and incorporates an alignment module and a sorting constraint module to achieve joint detection of deception, emotion, and personality. Experimental results on the MDPE and DDEP datasets show that our Rel-DDEP significantly outperforms the existing state-of-the-art baseline models in three tasks. The F1 score of the deception detection increases by 2.53%, that of the emotion detection increases by 2.66%, and that of the personality detection increases by 9.30%. The experiments fully verify the necessity of annotating dynamic emotion and personality labels for each sample and the effectiveness of reliability-weighted fusion.

The individuation problem for large language models asks which entities associated with them, if any, should be identified as minds. We approach this problem through mechanistic interpretability, engaging in particular with recent empirical work on persona vectors, persona space, and emergent misalignment. We argue that three views are the strongest candidates: the virtual instance view and two new views we introduce, the (virtual) instance-persona view and the model-persona view. First, we argue for the virtual instance view on the grounds that attention streams sustain quasi-psychological connections across token-time. Then we present the persona literature, organised around three hypotheses about the internal structure underlying personas in LLMs, and show that the two persona-based views are promising alternatives.

Large Language Models (LLMs) produce a controllability gap in safety-critical engineering: even low rates of undetected constraint violations render a system undeployable. Current orchestration paradigms suffer from sycophantic compliance, context attention decay [Liu et al., 2024], and stochastic oscillation during self-correction [Huang et al., 2024]. We introduce the Convergent AI Agent Framework (CAAF), which transitions agentic workflows from open-loop generation to closed-loop Fail-Safe Determinism via three pillars: (1) Recursive Atomic Decomposition with physical context firewalls; (2) Harness as an Asset, formalizing domain invariants into machine-readable registries enforced by a deterministic Unified Assertion Interface (UAI); and (3) Structured Semantic Gradients with State Locking for monotonic convergence. Empirical evaluation across two domains -- SAE Level 3 (L3) autonomous driving (AD) (n=30, 7 conditions) and pharmaceutical continuous flow reactor design (n=20, 4 conditions including a Mono+UAI ablation) -- shows that CAAF-all-GPT-4o-mini achieves 100% paradox detection while monolithic GPT-4o achieves 0% (even at temperature=0). The pharmaceutical benchmark features 7 simultaneous constraints with nonlinear Arrhenius interactions and a 3-way minimal unsatisfiable subset, representing a structurally harder challenge than the 2-constraint AD paradox. Alternative multi-agent architectures (debate, sequential checking) also achieve 0% across 80 trials, confirming that CAAF's reliability derives from its deterministic UAI, not from multi-agent orchestration per se. A Mono+UAI ablation (95%) isolates UAI as the core contribution. CAAF's reliability is invariant to prompt hints; all components use a single commodity model, enabling fully offline deployment.

For four decades, the QWERTY keyboard organized white-collar knowledge work. Typing's dominance was instrumental, not cognitively necessary. As multimodal AI achieves human-parity understanding of speech and gesture, this necessity dissolves. We introduce instrumental dissolution -- loss of institutional-default status while persisting in specialist niches. The keyboard era ends not through hardware replacement but through migration of its function into AI systems. The central contribution identifies the verification bottleneck: as AI collapses production friction, the primary constraint shifts from generation to evaluation. Knowledge workers become adversarial auditors rather than keystroke-producers. This restructures professional expertise, organizational communication, and how productive labor is recognized. Converging evidence from history, philosophy, neuroscience, technology, organizational studies, and cultural analysis supports this thesis. We map synthetic literacy -- oral input generating literate output -- as the defining feature of this transition. Under three scenarios (optimistic: 2028-2035; base: 2035-2045; pessimistic: 2045-2060), we specify disconfirmation criteria that would weaken the thesis if observed. We propose seven interface primitives operationalizing verification-centered HCI.

Subjective NLP datasets typically aggregate annotator judgments into a single gold label, making it difficult to diagnose whether disagreement reflects unclear criteria, collapsed distinctions, or legitimate plurality. We propose a \emph{schema-level diagnostic} for auditing expert-designed annotation schemas \emph{prior to} gold-label commitment, using only multi-annotator criterion judgments. The diagnostic separates two failure modes: unstable criteria with hard-to-operationalize boundaries, and systematic overlap that blurs the boundaries between mutually exclusive categories. Applied to persuasive value extraction in commercial documents, we find that disagreement is not diffuse: instability concentrates in a few criteria, while nearly half of covered sentences activate multiple categories. These signals align with where domain experts disagree, yielding an evidence-based audit for tightening guidelines, revising category structure, or reconsidering the annotation paradigm.

Static benchmarks for harmful content detection face limitations in scalability and diversity, and may also be affected by contamination from web-scale pre-training corpora. To address these issues, we propose a framework for synthesizing harmful content, leveraging persona-guided large language model (LLM) agents. Our approach constructs two-dimensional user personas by integrating demographic identities and topical interests with situational harmful strategies, enabling the simulation of diverse and contextually grounded harmful interactions. We evaluate the framework along three dimensions: harmfulness, challenge level, and diversity. Both human and LLM-based evaluations confirm that our framework achieves a high harmful generation success rate. Experiments across multiple detection systems reveal that our synthetic scenarios are more challenging to detect than those in existing benchmarks. Furthermore, a multi-faceted analysis confirms that our approach achieves linguistic and topical diversity comparable to human-curated datasets, establishing our framework as an effective tool for robust stress-testing of harmful content detection systems.

Instruction granularity is an important yet poorly controlled variable in language-guided embodied AI. Existing benchmarks typically pair each task with a single static instruction, making it difficult to study how agent behavior changes when the same task is described at different levels of detail. We introduce Mini-BEHAVIOR-Gran, a new benchmark for controlled studies of instruction granularity that extends Mini-BEHAVIOR with multiple instruction variants per task, ranging from high-level goal descriptions to step-by-step guidance. Using this benchmark, we compare four candidate metrics for cross-task granularity quantification: token count, entity count, action-verb count, and planning-width, and find that width correlates most consistently with agent performance. Using width to organize training and evaluation further reveals a non-monotonic U-shaped relationship between instruction granularity and performance, with peaks at both fine and coarse extremes. Further analysis suggests that the coarse-granularity performance rebound is associated with shallow grounding, where agents learn vision-dominant policies.

Large Language Models (LLMs) perform well on automatic program repair (APR) for high-resource programming languages (HRPLs), but their effectiveness drops sharply in low-resource programming languages (LRPLs), due to a lack of sufficient verified buggy-fixed pairs for APR training. To address this challenge, we propose HELO-APR (High-resource Enabled LOw-resource APR), a two-stage APR framework that enables cross-lingual transfer of repair knowledge from HRPLs to LRPLs. HELO-APR (1) constructs high-quality LRPL training data by synthesizing LRPL buggy-fixed pairs from HRPL counterparts, preserving defect type consistency while ensuring the synthesized code is idiomatic, and then (2) adopts a curriculum learning strategy that progressively performs HRPL repair learning, cross-lingual repair alignment, and LRPL repair adaptation, improving repair effectiveness in LRPLs. Using C++ as the source HRPL and Ruby and Rust as the target LRPLs, experiments on xCodeEval show that HELO-APR consistently outperforms strong baselines, increasing Pass@1 from 31.32% to 48.65% on DeepSeek-Coder-6.7B and from 1.67% to 11.97% on CodeLlama-7B, while improving syntactic validity by raising the average target compilation rate on CodeLlama from 49.77% to 91.98%. On Defects4Ruby, HELO-APR increases BLEU-4 from 61.20 to 66.79 and ROUGE-1 from 76.76 to 83.59 on CodeLlama-7B, indicating higher similarity to developer patches in real-world settings. Finally, we conduct ablation studies to assess the necessity of each core component. These results suggest that verified cross-lingual supervision provides a reusable approach for improving LLM-based repair in low-resource languages.

We introduce a self-play framework for semantic equivalence in Haskell, utilizing formal verification to guide adversarial training between a generator and an evaluator. The framework leverages Liquid Haskell proofs for validating equivalence and execution-based counterexamples for inequivalence, organized via a difficulty-aware curriculum. To facilitate this, we release \textbf{OpInstruct-HSx}, a synthetic dataset of $\approx$28k validated Haskell programs. Empirical experiments show that our evaluator transfers effectively to downstream tasks, achieving up to 13.3pp accuracy gain on EquiBench and consistent gains on PySecDB. Ablation studies on the SEQ-SINQ regimes indicate that while inequivalence supervision provides data volume, equivalence proofs are uniquely responsible for the model's reasoning capabilities. The entire training pipeline and dataset are publicly released on GitHub and Hugging Face respectively.

Multi-agent systems (MAS) demonstrate clear advantages in tackling complex problems by coordinating diverse agents and external tools. However, most existing orchestration methods rely on static workflows or serial agent scheduling, and are further constrained by heterogeneous interface protocols between tools and agents. This leads to high system complexity and poor extensibility. To mitigate these issues, we propose Agent-as-Tool, a unified parallel orchestration paradigm that abstracts both agents and tools into a standardized, learnable action space with protocol normalization and explicit state feedback. Building on this paradigm, we train a lightweight orchestrator, ParaManager, which decouples planning decisions from subtask solving, enabling state-aware parallel subtask decomposition, delegation, and asynchronous execution. For training, we adopt a two-stage ParaManager training pipeline. It improves robustness by incorporating supervised fine-tuning (SFT) trajectories equipped with recovery mechanisms, and further applies reinforcement learning (RL) to achieve an optimal balance among task success, protocol compliance, diversity, and reasoning efficiency. Experiments show that ParaManager achieves strong performance across multiple benchmarks and exhibits robust generalization under unseen model pools.

Large Language Models (LLMs) are increasingly used to generate narrative content, including children's stories, which play an important role in social and cultural learning. Despite growing interest in AI safety and alignment, most existing evaluations focus primarily on English, leaving the cross-lingual generalization of aligned behavior underexplored. In this work, we introduce BiasedTales-ML, a large-scale parallel corpus of approximately 350,000 children's stories generated across eight typologically and culturally diverse languages using a full-permutation prompting design. We propose a structured generator-extractor pipeline and a multi-dimensional distributional analysis framework to examine how narrative attributes vary across languages, models, and social conditions. Our analysis reveals substantial cross-lingual variability in narrative generation patterns, indicating that distributions observed in English do not always exhibit similar characteristics in other languages, particularly in lower-resource settings. At the narrative level, we identify recurring structural patterns involving character roles, settings, and thematic emphasis, which manifest differently across linguistic contexts. These findings highlight the limitations of English-centric evaluation for characterizing socially grounded narrative generation in multilingual settings. We release the dataset, code, and an interactive visualization tool to support future research on multilingual narrative analysis and evaluation.

Mobile deployment of facial age estimation requires models that balance predictive accuracy with low latency and compact size. In this work, we present MobileAgeNet, a lightweight age-regression framework that achieves an MAE of 4.65 years on the UTKFace held-out test set while maintaining efficient on-device inference with an average latency of 14.4 ms measured using the AI Benchmark application. The model is built on a pretrained MobileNetV3-Large backbone combined with a compact regression head, enabling real-time prediction on mobile devices. The training and evaluation pipeline is integrated into the NN LEMUR Dataset framework, supporting reproducible experimentation, structured hyperparameter optimization, and consistent evaluation. We employ bounded age regression together with a two-stage fine-tuning strategy to improve training stability and generalization. Experimental results show that MobileAgeNet achieves competitive accuracy with 3.23M parameters, and that the deployment pipeline from PyTorch training through ONNX export to TensorFlow Lite conversion - preserves predictive behavior without measurable degradation under practical on-device conditions. Overall, this work provides a practical, deployment-ready baseline for mobile-oriented facial age estimation.

The recently established Convolution Nuclear Norm Minimization (CNNM) addresses the problem of \textit{tensor completion with arbitrary sampling} (TCAS), which involves restoring a tensor from a subset of its entries sampled in an arbitrary manner. Despite its promising performance, the optimization procedure of CNNM needs performing Singular Value Decomposition (SVD) multiple times, which is computationally expensive and hard to parallelize. To address the issue, we reformulate the optimization objective of CNNM from the perspective of convolution eigenvectors. By introducing pre-learned convolution eigenvectors which are shared among different tensors, we propose a novel method called Inductive Convolution Nuclear Norm Minimization (ICNNM), which bypasses the SVD step so as to decrease significantly the computational time. In addition, due to the extra prior knowledge encoded in the pre-learned convolution eigenvectors, ICNNM also outperforms CNNM in terms of recovery performance. Extensive experiments on video completion, prediction and frame interpolation verify the superiority of ICNNM over CNNM and several other competing methods.

Reinforcement learning (RL) has emerged as a promising paradigm for training reasoning-oriented models by leveraging rule-based reward signals. However, RL training typically tends to improve single-sample success rates (i.e., Pass@1) while offering limited exploration of diverse reasoning trajectories, which is crucial for multi-sample performance (i.e., Pass@k). Our preliminary analysis reveals that this limitation stems from a fundamental squeezing effect, whereby probability mass is excessively concentrated on a narrow subset of high-reward trajectories, restricting genuine exploration and constraining attainable performance under RL training. To address this issue, in this work, we propose Steering Probability Squeezing (SPS), a training paradigm that interleaves conventional RL with inverse reinforcement learning (IRL). SPS treats on-policy rollouts as demonstrations and employs IRL to explicitly reshape the induced trajectory distribution, thereby enhancing exploration without introducing external supervision. Experiments on five commonly used reasoning benchmarks demonstrate that SPS can enable better exploration and improve Pass@k. Beyond algorithmic contributions, we provide an analysis of RL learning dynamics and identify an empirical upper bound on Pass@k, shedding light on intrinsic exploration limits in RL-based reasoning models. Our findings suggest that alternating between RL and IRL offers an effective pathway toward extending the exploration capacity of reasoning-oriented large language models.

As embodied AI transitions to real-world deployment, the success of the Vision-and-Language Navigation (VLN) task tends to evolve from mere reachability to social compliance. However, current agents suffer from a "goal-driven trap", prioritizing physical geometry ("can I go?") over semantic rules ("may I go?"), frequently overlooking subtle regulatory constraints. To bridge this gap, we establish Rule-VLN, the first large-scale urban benchmark for rule-compliant navigation. Spanning a massive 29k-node environment, it injects 177 diverse regulatory categories into 8k constrained nodes across four curriculum levels, challenging agents with fine-grained visual and behavioral constraints. We further propose the Semantic Navigation Rectification Module (SNRM), a universal, zero-shot module designed to equip pre-trained agents with safety awareness. SNRM integrates a coarse-to-fine visual perception VLM framework with an epistemic mental map for dynamic detour planning. Experiments demonstrate that while Rule-VLN challenges state-of-the-art models, SNRM significantly restores navigation capabilities, reducing CVR by 19.26% and boosting TC by 5.97%.

We report new results on six problems in mathematics and theoretical computer science, produced with the assistance of Bolzano, an open-source multi-agent LLM system. Bolzano orchestrates rounds of interaction between parallel prover agents and a verifier agent while maintaining a persistent knowledge base that is carried across rounds. Classified using the significance-autonomy taxonomy of Feng et al., four of the six results reach the level of publishable research, and three of the six were produced essentially autonomously by Bolzano. Our results provide evidence that LLMs can contribute meaningfully to mathematical research, complementing recent reports by Bubeck et al., Woodruff et al., and others.

Non-stationary sequences arise naturally in control, forecasting, and decision-making. The data-generating process shifts at unknown times, and models must detect the change, discard or downweight obsolete evidence, and adapt to new dynamics on the fly. Transformer-based foundation models increasingly rely on in-context learning for time series forecasting, tabular prediction, and continuous control. As these models are deployed in non-stationary environments, understanding their ability to detect and adapt to regime shifts is important. We formalize this as an in-context change-point detection problem and formally establish the existence of transformer models that solve this problem. Our construction demonstrates that model complexity, in layers and parameters, depends on the level of information available about the change-point location, from no knowledge to knowing exact timing. We validate our results with experiments on synthetic linear regression and linear dynamical systems, where trained transformers match the performance of optimal baselines across information levels. We also show that encoding and incorporating changepoint knowledge indeed improves the real-world performance of a pretrained foundation models on infectious disease forecasting and on financial volatility forecasting around Federal Open Market Committee (FOMC) announcements without retraining, demonstrating practical applicability to real-world regime changes.

Current knowledge graph (KG) construction methods are confirmatory, focusing on recovering known relationships rather than identifying novel or context-dependent nodes. This paper proposes a phenotype-driven and evidence-governed framework that shifts the paradigm toward structured hypothesis discovery and controlled KG expansion. The approach integrates graph neural networks (GNNs) for phenotype discovery, causal inference, probabilistic reasoning and large language models (LLMs) for hypothesis generation and claim extraction within a unified pipeline. The framework prioritizes relationships that are both structurally supported by data and underexplored in the literature. KG expansion is formulated as a multi-objective optimization problem, where candidate claims are jointly evaluated in terms of relevance, structural validation and novelty. Pareto-optimal selection enables the identification of non-dominated claims that balance confirmation and discovery, avoiding trivial or redundant knowledge inclusion. Experiments on heterogeneous population datasets demonstrate that the proposed framework produces more interpretable phenotypes, reveals context-dependent causal structures and generates high-quality claims that align with both data and scientific evidence. Compared to rule-based and LLM-only baselines, the method achieves the best trade-off across plausibility, novelty, validation and relevance. In retrieval-augmented settings, it significantly improves performance (Recall@5=0.98) while reducing hallucination rates (0.05), highlighting its effectiveness in grounding LLM outputs.

The LEOPs (Light-ERG-Oscillatory Potentials) dataset provides light-adapted (LA) electroretinogram (ERG) and Oscillatory Potentials (OPs) waveforms for typically developing Control, Autism Spectrum Disorder (ASD) and ASD + Attention Deficit Hyperactivity Disorder (ADHD) childhood and adolescent populations. The ERGs were recorded in the Right And Left eyes with skin electrodes using the handheld RETeval device at two sites in Australia and the United Kingdom. The LEOPs dataset includes 5309 single flash ERG and 4434 OPs waveforms as well as images selected from each participant showing the position of the skin electrode. The LEOPs dataset is constructed from recordings using a 9 step randomized flash series from $-0.37$ to $1.20$~$Td.s$, a 2 step at 113 and 446 $Td.s$ flash strengths (2500 Control, 1730 ASD and 451 ASD + ADHD samples), as well as the $85$~$Td.s$ (Light Adapted 3 $cd.s.m^{-2}$ (LA3)) equivalent International Society of Clinical Electrophysiology of Vision (ISCEV) Standard flash with 435 Control, 176 ASD and 37 ASD + ADHD waveform samples. Code for the stimulus is provided along with participant demographics, date and time of testing, and where available diagnostic scores for the ASD and ASD + ADHD groups, alongside iris color, electrode position with image files and time domain values for the ERG and summed values for the OPs. The repository contains excel file, exported JSON files on the patient level that are more suitable for machine learning tasks, images of electrode position for each recording and the protocol files for use with the RETeval.

Background: Large language models (LLMs) are increasingly proposed for diagnostic support, but few evaluations use real-world multimodal inpatient data, particularly in low and middle-income country (LMIC) public hospitals. Methods: We conducted VALID, a retrospective evaluation of 539 multimodal inpatient cases from a tertiary public hospital in South Africa. Inputs included radiology imaging (CT, MRI, CXR) and reports, laboratory results, clinical notes, and vital signs. Expert panels adjudicated 300 cases (balanced and discordant subsets) to establish ground truth diagnoses, differentials, and reasoning. Ten multimodal LLMs generated zero-shot outputs. A calibrated three-model LLM Jury scored all outputs and routine ward diagnoses across diagnostic accuracy, differential quality, reasoning, and patient safety (>10,000 evaluations). Primary outcomes were composite scores ($S_3$, $S_4$) and win rates. Results: (i) LLM performance was tightly clustered (<15% variation) despite large cost differences; low-cost models performed comparably to top models. (ii) All LLMs significantly outperformed routine ward diagnoses on average diagnostic and safety scores. (iii) Top performance was achieved by GPT-5.1, followed by Gemini models. (vi) Adding radiology reports improved performance by 6%. (v) Diagnostic and reasoning scores were highly correlated ($ρ= 0.85$). (vi) Output rates varied (65-100%) due to input constraints. Results were robust across subsets and evaluation design. Conclusions: Across a real-world LMIC dataset, multimodal LLMs showed similar diagnostic performance despite large cost differences and outperformed routine care on average safety metrics. Affordability, robustness, and deployment constraints may outweigh marginal performance differences in LMIC settings.

High-quality and diverse multimodal data are essential for improving vision-language models (VLMs), yet existing datasets often contain noisy, redundant, and poorly aligned samples. To address these problems, data filtering is commonly used to enhance the efficiency and performance of multimodal learning, but it introduces extra computational cost because filtering models are usually trained on the same data they are meant to screen. To reduce this cost, we study DOSE, which explores whether off-the-shelf pretrained models that have never seen the target data can be used to select training samples for larger and stronger multimodal models without any task-specific training. Even without fine-tuning, these models can effectively assess text quality and image-text alignment to guide data selection. Based on this, we build a joint quality-alignment distribution and apply adaptive weighted sampling to select informative samples while maintaining long-tail diversity. This approach enhances data diversity, enabling models trained on DOSE-filtered data to match or surpass those trained on the full dataset on standard VQA and math benchmarks. Extensive experiments demonstrate its effectiveness, efficiency, and scalability.

Recent work has shown that parameterizing and optimizing coordinate transformations using normalizing flows, i.e., invertible neural networks, can significantly accelerate the convergence of spectral approximations. We present the first error estimates for approximating functions using Hermite expansions composed with adaptive coordinate transformations. Our analysis establishes an equivalence principle: approximating a function $f$ in the span of the transformed basis is equivalent to approximating the pullback of $f$ in the span of Hermite functions. This allows us to leverage the classical approximation theory of Hermite expansions to derive error estimates in transformed coordinates in terms of the regularity of the pullback. We present an example demonstrating how a nonlinear coordinate transformation can enhance the convergence of Hermite expansions. Focusing on smooth functions decaying along the real axis, we construct a monotone transport map that aligns the decay of the target function with the Hermite basis. This guarantees spectral convergence rates for the corresponding Hermite expansion. Our analysis provides theoretical insight into the convergence behavior of adaptive Hermite approximations based on normalizing flows, as recently explored in the computational quantum physics literature.

Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a promising approach to improve the reasoning abilities of Large Language Models (LLMs). Among RLVR algorithms, Group Relative Policy Optimization (GRPO) and its variants have demonstrated strong performance and high training efficiency. However, GRPO-style objectives exhibit two issues on high accuracy prompts including mastered prompts (rollout accuracy =1) and majority-correct prompts (rollout accuracy in (0.5,1)). For mastered prompts, group-relative advantages vanish, yielding no training signal and unconstrained policy drift that can cause forgetting. For majority-correct prompts, the induced query weight shrinks as accuracy increases, weakening consolidation from partial correctness to mastery. To alleviate this, we propose Mastery-Consolidated Policy Optimization (MCPO), which introduces (i) a hinge-KL regularizer applied exclusively to mastered prompts to bound harmful policy drift between successive gradient steps, and (ii) a weighting mechanism that prioritizes majority-correct prompts to better allocate optimization effort. Extensive experiments across three mathematical benchmarks demonstrate that MCPO consistently improves pass@1 performance. Counter-intuitively, rather than restricting exploration, MCPO boosts pass@k metrics, indicating that mastery consolidation further catalyzes solution diversity.

Experience-driven self-evolution has emerged as a promising paradigm for improving the autonomy of large language model agents, yet its reliance on self-curated experience introduces underexplored safety risks. In this study, we investigate how experience accumulation and utilization in self-evolving agents affect safety performance across web-based and embodied environments. Notably, experience gathered solely from benign tasks can still compromise safety in high-risk scenarios. Further analysis attributes this degradation to the execution-oriented nature of accumulated experience, which reinforces agents' tendency to act rather than refuse. In more realistic settings where agents encounter both benign and harmful tasks, refusal-related experience mitigates safety decline but induces over-refusal, revealing a fundamental safety-utility trade-off. Overall, our findings expose inherent limitations of current self-evolving agents and call for more principled strategies to ensure safe and reliable adaptation.

Path planning is usually solved by addressing either the (high-level) route planning problem (waypoint sequencing to achieve the final goal) or the (low-level) path planning problem (trajectory prediction between two waypoints avoiding collisions). However, real-world problems usually require simultaneous solutions to the route and path planning subproblems with a holistic and efficient approach. In this paper, we introduce NaviFormer, a deep reinforcement learning model based on a Transformer architecture that solves the global navigation problem by predicting both high-level routes and low-level trajectories. To evaluate NaviFormer, several experiments have been conducted, including comparisons with other algorithms. Results show competitive accuracy from NaviFormer since it can understand the constraints and difficulties of each subproblem and act consequently to improve performance. Moreover, its superior computation speed proves its suitability for real-time missions.

The evolution from static ranking models to Agentic Recommender Systems (Agentic RecSys) empowers AI agents to maintain long-term user profiles and autonomously plan service tasks. While this paradigm shift enhances personalization, it introduces a vulnerability: reliance on Long-term Memory (LTM). In this paper, we uncover a threat termed "Visual Inception." Unlike traditional adversarial attacks that seek immediate misclassification, Visual Inception injects triggers into user-uploaded images (e.g., lifestyle photos) that act as "sleeper agents" within the system's memory. When retrieved during future planning, these poisoned memories hijack the agent's reasoning chain, steering it toward adversary-defined goals (e.g., promoting high-margin products) without prompt injection. To mitigate this, we propose CognitiveGuard, a dual-process defense framework inspired by human cognition. It consists of a System 1 Perceptual Sanitizer (diffusion-based purification) to cleanse sensory inputs and a System 2 Reasoning Verifier (counterfactual consistency checks) to detect anomalies in memory-driven planning. Extensive experiments on a mock e-commerce agent environment demonstrate that Visual Inception achieves about 85% Goal-Hit Rate (GHR), while CognitiveGuard reduces this risk to around 10% with configurable latency trade-offs (about 1.5s in lite mode to about 6.5s for full sequential verification), without quality degradation under our setup.

Memory simulators are used to estimate application performance on advanced memory systems, yet they may exhibit significant discrepancies compared to real hardware. This paper investigates two key questions: (1) what causes these inaccuracies, and (2) how can simulators be properly validated to ensure reliable performance predictions. We propose a methodology that evaluates memory performance from three complementary perspectives: the memory simulator, the CPU-memory interface, and the application. Our analysis reveals that these perspectives can diverge substantially, with application-level performance often decoupled from internal simulator statistics. We identify the CPU-memory interface as the primary source of these inaccuracies. To address these problems, we implement a set of corrections and enhancements that improve the fidelity of integrated simulators. We evaluate these changes across multiple widely used simulators, including Ramulator, Ramulator 2, and DRAMsim3 integrated with ZSim. The results show that correcting interface-related issues is essential to achieve simulation outcomes that closely resemble actual system performance.

Floating-point square-root computation is a power- and delay-critical operation in edge-AI, signal-processing, and embedded systems. Conventional implementations typically rely on multipliers or iterative pipelines, resulting in increased hardware complexity, switching activity, and energy consumption. This work presents E2AFS, a lightweight and fully multiplier-free floating-point square-root architecture optimized for energy-efficient computation. By reducing logic depth and minimizing switching activity, the proposed design achieves substantial improvements in hardware efficiency and performance. FPGA implementation on an Artix-7 device demonstrates that E2AFS achieves the lowest dynamic power (7.63 mW), the shortest critical-path delay (4.639 ns), and the minimum power-delay product (35.39 pJ) compared to existing ESAS and CWAHA architectures. Error evaluation using multiple accuracy metrics, together with graphical analysis, shows that E2AFS closely approximates the exact square-root function with consistently low deviation. Application-level validation in Sobel edge detection and K-means color quantization further confirms its suitability for low-power real-time edge and embedded platforms.

Generating trajectories for synthetic aperture radar (SAR)-equipped aircraft poses significant challenges due to terrain constraints, and the need for straight-flight segments to ensure high-quality imaging. Related works usually focus on trajectory optimization for predefined straight-flight segments that do not adapt to the target visibility, which depends on the 3D terrain and aircraft orientation. In addition, this assumption does not scale well for the multi-target problem, where multiple straight-flight segments that maximize target visibility must be defined for real-time operations. For this purpose, this paper presents a multi-stage planning system. First, the waypoint sequencing to visit all the targets is estimated. Second, straight-flight segments maximizing target visibility according to the 3D terrain are predicted using a novel neural network trained with deep reinforcement learning. Finally, the segments are connected to create a trajectory via optimization that imposes 3D Dubins curves. Evaluations demonstrate the robustness of the system for SAR missions since it ensures high-quality multi-target SAR image acquisition aware of 3D terrain and target visibility, and real-time performance.

Incomplete Multi-View Clustering (IMVC) faces the challenge of learning discriminative representations from fragmentary observations while maintaining robustness against missing views. However, prevalent Euclidean-based methods suffer from a geometric mismatch when modeling real-world data with intrinsic hierarchies, leading to semantic blurring where representations drift towards spatially proximal but semantically distinct neighbors. To bridge this gap, we propose HERL, a Hyperbolic Enhanced Representation Learning framework for IMVC. Operating within the Poincaré ball, HERL constructs a structure-aware latent space to enhance representation learning. Specifically, we design a dual-constraint hyperbolic contrastive mechanism optimizing: an angular-based loss to preserve semantic identity via directional alignment, and a distance-based loss to enforce hierarchical compactness. Furthermore, a hyperbolic prototype head is introduced to rectify global structural drift by aligning cross-view hierarchy-aware prototype distributions. Consequently, HERL disentangles fine-grained semantic correlations to sharpen cluster boundaries and imposes geometric constraints to rectify the data recovery process. Extensive experimental results demonstrate that HERL consistently outperforms state-of-the-art approaches.

We present Open-TQ-Metal, the first implementation of fused compressed-domain attention on Apple Silicon, enabling 128K-context inference for Llama 3.1 70B on a single 64GB consumer Mac -- a configuration impossible with all existing inference frameworks. Open-TQ-Metal quantizes the KV cache to int4 on the fly and computes attention directly on the compressed representation via custom Metal compute shaders, eliminating all intermediate dequantization matrices. Across 330 experiments spanning two model families (Gemma 4 31B and Llama 3.1 70B), the fused sdpa_int4 kernel achieves 48x attention speedup at 128K context over the dequantize-then-attend baseline, reduces KV cache memory from 40 GB to 12.5 GB (3.2x compression), and maintains identical top-1 token predictions to FP16 inference. We further provide the first cross-architecture analysis of KV cache quantization methods, revealing that the attention scale factor -- not model size -- determines whether angular quantization schemes like PolarQuant succeed or fail, with Gemma 4's attn_scale=1.0 amplifying directional error 25-100x more than Llama's standard 1/sqrt(d) scaling.

Quantitative prediction of future retinal appearance from longitudinal imaging would support clinical decisions in progressive macular disease that currently rely on qualitative comparison or scalar progression scores. Recent methods have moved toward increasing generative complexity, but whether this complexity is necessary for slowly progressing retinal disease is unclear. We tested this through a controlled comparison of five conditioning configurations sharing one architecture and training dataset, spanning standard conditional diffusion, inference-aligned stochastic training, and deterministic regression. In our evaluation, aligning the training and inference input distributions produced large gains (delta-SSIM +0.082, SSIM +0.086, both p < 0.001), while the choice among aligned frameworks did not significantly affect any primary metric. Task-entropy and posterior-concentration analyses, replicated on two fundus autofluorescence (FAF) platforms, provided a mechanistic account: the predictable component of inter-visit change is small relative to time-invariant acquisition variability, leaving stochastic sampling with little width to exploit. Guided by these findings, we developed TRU (Temporal Retinal U-Net), a deterministic direct-regression model with continuous time-delta conditioning and multi-scale history aggregation. We evaluated TRU on 28,902 eyes across three imaging platforms: a mixed-disease Optos FAF cohort (9,942 eyes), zero-shot transfer to Stargardt macular dystrophy on Optos (288 eyes) and Heidelberg Spectralis (125 eyes), and a boundary evaluation on Cirrus en-face fundus images from a glaucoma cohort (18,547 eyes). TRU matched or exceeded delta-SSIM, SSIM, and PSNR in every FAF cohort against three state-of-the-art benchmarks, and its advantage grew monotonically with available history length.

Breast cancer diagnosis through thermographic image analysis remains a critical challenge in medical AI, with classical deep learning approaches facing limitations in complex thermal pattern classification tasks. This paper presents a novel Hybrid Quantum Neural Network (HQNN) architecture that integrates quantum computing principles with classical convolutional neural networks for enhanced breast cancer classification. Our approach employs parameterized quantum circuits with multi-head attention mechanisms for quantum-aware feature encoding, coupled with classical convolutional layers for comprehensive pattern recognition. The quantum component utilizes a 4qubit variational circuit with strongly entangling layers, while the classical component incorporates advanced attention mechanisms for feature fusion. Experimental validation on breast cancer thermographic data demonstrates substantial performance improvements over state-of-the-art classical architectures, with the quantum-enhanced approach exhibiting superior convergence dynamics and enhanced feature representation capabilities. Our findings provide evidence for quantum advantage in medical image classification through classical simulation, establishing a framework for quantum-classical hybrid systems in healthcare applications. The methodology addresses key challenges in quantum machine learning deployment while maintaining computational feasibility on near-term quantum devices.

Product attribute extraction in e-commerce is bottlenecked by ontologies that are inconsistent, incomplete, and costly to maintain. We present AutoPKG, a multi-agent Large Language Model (LLM) framework that automatically constructs a Product-attribute Knowledge Graph (PKG) from multimodal product content. AutoPKG induces product types and type-specific attribute keys on demand, extracts attribute values from text and images, and consolidates updates through a centralized decision agent that maintains a globally consistent canonical graph. We also propose an evaluation protocol for dynamic PKGs that measures type and key validity, consolidation quality, and edge-level accuracy for value assertions after canonicalization. On a large real-world marketplace catalog dataset from Lazada (Alibaba), AutoPKG achieves up to 0.953 Weighted Knowledge Efficiency (WKE) for product types, 0.724 WKE for attribute keys, and 0.531 edge-level F1 for multimodal value extraction. Across three public benchmarks, our method improves edge-level exact-match F1 by 0.152 and yields a precision gain of 0.208 on the attribute extraction application. Online A/B tests show that AutoPKG-derived attributes increase Gross Merchandise Value (GMV) in Badge by 3.81 percent, in Search by 5.32 percent, and in Recommendation by 7.89 percent, supporting the practical value of AutoPKG in production.

The paper considers the computation of L1 regularization paths in a state space setting, which includes L1 regularized Kalman smoothing, linear SVM, LASSO, and more. The paper proposes two new algorithms, which are duals of each other; the first algorithm applies to L1 regularization of independent variables while the second applies to L1 regularization of dependent variables. The heart of the proposed algorithms is parametric Gaussian message passing (i.e., Kalman-type forward-backward recursions) in the pertinent factor graphs. The proposed methods are broadly applicable, they (usually) require only matrix multiplications, and their complexity can be competitive with prior methods in some cases.

Multimodal large language models (MLLMs) have shown impressive capabilities, yet they often struggle to effectively capture the fine-grained textual information within images crucial for accurate image translation. This often leads to a modality gap between visual text inputs and textual inputs/outputs for image translation. Existing methods, primarily relying on instruction fine-tuning, risk parameter redundancy of pre-trained knowledge, hindering generalization performance. To address this, we introduce modality neuron-aware fine-tuning (MNAFT), a novel approach that takes advantage of the specialized roles of individual neurons within MLLMs for enhanced image translation. MNAFT identifies language-agnostic and language-specific neurons in both vision and language modules through an instruction-driven activation analysis, evaluating their importance in various translation tasks. We then perform selective fine-tuning, updating only the parameters of language-specific and language-agnostic neurons within the selected layers relevant to the target task, while preserving the knowledge encoded in other neurons and layers. Our extensive experiments on multiple benchmarks demonstrate that MNAFT significantly outperforms state-of-the-art image translation methods, including cascaded models, standard full fine-tuning, and parameter-efficient tuning techniques. Furthermore, we provide comprehensive analysis, including visualizations of neuron activations and clustering patterns, to offer insights into the roles of different neuron groups in mediating cross-modal understanding and facilitating accurate language-specific translation.

We present MEMRES, an agentic system for Python dependency resolution that introduces a multi-level confidence cascade where the LLM serves as the last resort. Our system combines: (1) a Self-Evolving Memory that accumulates reusable resolution patterns via tips and shortcuts; (2) an Error Pattern Knowledge Base with 200+ curated import-to-package mappings; (3) a Semantic Import Analyzer; and (4) a Python 2 heuristic detector resolving the largest failure category. On HG2.9K using Gemma-2 9B (10 GB VRAM). MEMRES resolves 2503 of 2890 (86.6%, 10-run average) snippets, combining intra-session memory with our confidence cascade for the remainder. This already exceeds PLLM's 54.7% overall success rate by a wide margin.

Supervised Fine-Tuning (SFT) accelerates taskspecific large language models (LLMs) development, but the resulting proliferation of finetuned models incurs substantial memory overhead. Delta compression addresses this by retaining a single pre-trained LLM with multiple compressed delta weights. However, existing methods fail on models fine-tuned with largescale datasets. We find that larger SFT data scale amplifies delta parameter magnitude, singular values, and entropy, exacerbating compression errors. To tackle this, we propose DQRELO (Delta Compression via Quantization and Residual Low-Rank), a novel training- and data-free delta compression method. It combines coarse-grained one-bit quantization to capture the dominant structure of the delta, followed by compensated residual low-rank approximation to recover fine-grained details from the smaller residual error. Experiments on various LLMs spanning dense and MoE architectures across multiple domains under this challenging setting demonstrate that DQRELO outperforms existing methods. Moreover, we establish key design principles for delta compression through extensive empirical analysis, demonstrating how task difficulty, architecture, and layer positioning create predictable patterns that can guide optimal compression strategies in production systems.

Translation-based prompting is widely used in multilingual LLMs, yet its effectiveness varies across languages and tasks. We evaluate prompting strategies across ten languages of different resource levels and four benchmarks. Our analysis shows that no single strategy is universally optimal. Translation strongly benefits low-resource languages even when translation quality is imperfect, high-resource languages gain little, and prompt-based self-routing underperforms explicit translation. Motivated by these findings, we formulate prompting strategy selection as a learned decision problem and introduce lightweight classifiers that predict whether native or translation-based prompting is optimal for each instance. The classifiers achieve statistically significant improvements over fixed strategies across four benchmarks and generalize to unseen task formats not observed during training. Further analysis reveals that language resource level, rather than translation quality alone, determines when translation is beneficial.

Feature reconstruction techniques are widely applied for few-shot fine-grained image classification (FSFGIC). Our research indicates that one of the main challenges facing existing feature-based FSFGIC methods is how to choose the size of the receptive field to extract feature descriptors (including spatial and frequency feature descriptors) from different category input images, thereby better performing the FSFGIC tasks. To address this, an adaptive receptive field-based spatial-frequency feature reconstruction network (ARF-SFR-Net) is proposed. The designed ARF-SFR-Net has the capability to adaptively determine receptive field sizes for obtaining spatial and frequency features, and effectively fuse them for reconstruction and FSFGIC tasks. The designed ARF-SFR-Net can be easily embedded into a given episodic training mechanism for end-to-end training from scratch. Extensive experiments on multiple FSFGIC benchmarks demonstrate the effectiveness and superiority of the proposed ARF-SFR-Net over state-of-the-art approaches. The code is available at: https://github.com/ICL-SUST/ARF-SFR-Net.git.

Scarce longitudinal evidence examines LLMs' persuasiveness and humanness along time-evolving psychological frameworks. We introduce Talk2AI, a longitudinal framework quantifying psycho-social, reasoning and affective dimensions of LLMs' persuasiveness about polarizing societal topics. In a four-way longitudinal setup, Talk2AI's 770 participants engaged in structured conversations with one of four leading LLMs on topics like climate change, social media misinformation, and math anxiety. This produced 3,080 conversations over 60,000 turns. After each wave, participants reported conviction in their initial topic stance, perceived opinion change, LLM's perceived humanness, a self-donation to the topic and a textual explanation. Feedback time series showed longitudinal inertia in convictions, indicating some human anchoring to initial opinions even after repeated exposure to AI-generated arguments. Interestingly, NLP analyses revealed that both humans and LLMs relied on fallacious reasoning in 1 conversational quip every 6, countering the ``LLMs as superior systems" stereotype behind LLMs' cognitive surrender. LLMs' perceived humanness was most learnable from sociodemographic, psychological and engagement features ($R^2=0.44$), followed by opinion change ($R^2=0.34$), conviction ($R^2=0.26$) and personal endowment ($R^2=0.24$). Crucially, explainable AI (XAI) indicated: (i) the presence of individuals more susceptible to LLM-based opinion changes; (ii) psychological susceptibility to LLM-convincing consisted of having more trust in LLMs, being more agreeable and extraverted and with a higher need for cognition. A multiverse approach with mixed-effects models confirmed XAI results, alongside strong individual differences. Talk2AI provides a grounded framework and evidence for detecting how GenAI can influence human opinions via multiple psycho-social pathways in AI-human digital platforms.

Behavioral Co-Versioning remains absent from mainstream practice: while developers routinely version source code with Git, they rarely persist and query how run-time behavior evolves across revisions. This paper argues that this mismatch contributes to a blind spot in software evolution analysis and CI, where rich execution information is discarded and typically reduced to pass/fail outcomes -- despite partial test oracles, flakiness, and silent output or performance drift. We propose \textit{Behavioral Co-Versioning}, a paradigm that couples the Git history with a \textit{Behavioral Archive}: an append-only, queryable store of selected run-time observations (e.g., method I/O and performance signals) collected during test runs and keyed by commit and test context. This enables semantic diffing, behavior-aware regression localization, and retrospective auditing by querying historical executions, complementing proactive, signal-specific monitoring tools. We first outline a minimal data model and change diagnostics based on code/test/behavior fingerprints, and then demonstrate feasibility with a laptop-scale prototype that replays historical commits of a Python project, archives run-time observations in a local Parquet-backed store, and detects behavioral changes not apparent from textual diffs.

Across many scientific fields, measurements often represent the number of times an event occurs. For example, a document can be represented by word occurrence counts, neural activity by spike counts per time window, or online communication by daily email counts. These measurements yield high-dimensional count data that often approximate a Poisson distribution, frequently with low rates that produce substantial sparsity and complicate downstream analysis. A useful approach is to embed the data into a low-dimensional space that preserves meaningful structure, commonly termed dimensionality reduction. Yet existing dimensionality reduction methods, including both linear (e.g., PCA) and nonlinear approaches (e.g., t-SNE), often assume continuous Euclidean geometry, thereby misaligning with the discrete, sparse nature of low-rate count data. Here, we propose p-SNE (Poisson Stochastic Neighbor Embedding), a nonlinear neighbor embedding method designed around the Poisson structure of count data, using KL divergence between Poisson distributions to measure pairwise dissimilarity and Hellinger distance to optimize the embedding. We test p-SNE on synthetic Poisson data and demonstrate its ability to recover meaningful structure in real-world count datasets, including weekday patterns in email communication, research area clusters in OpenReview papers, and temporal drift and stimulus gradients in neural spike recordings.

Recent advances in large language models (LLMs) have shown that test-time scaling can substantially improve model performance on complex tasks, particularly in the coding domain. Under this paradigm, models use a larger token budget during inference to generate intermediate reasoning traces before producing a final answer. However, current evaluations primarily rely on competitive programming benchmarks, which may not capture the full range of reasoning abilities. In this work, we perform a systematic study of frontier reasoning models to understand their performance on real-world coding benchmarks. To gain more insights into the performance of such models, we devise a programmatic way to {\em automatically generate} coding tasks of arbitrary difficulty and structure from existing benchmarks. Using this framework, our analysis reveals that the structure of a reasoning trace, not just its contents, is a strong predictor of correctness. Motivated by this, we propose structured thought-trees as means to represent reasoning traces. To illustrate their use, we train a lightweight classifier on features extracted from thought-trees to predict trace correctness, and demonstrate that flagging and retrying structurally anomalous traces based on the extracted features yields consistent gains at lower complexity levels.

Visual Question Answering (VQA) requires models to identify the correct answer options based on both visual and textual evidence. Recent Mixture-of-Experts (MoE) methods improve option reasoning by grouping similar concepts or routing based on examples. However, unstable routing can lead to inconsistent expert selection in the same question type, while overly stable routing may reduce flexibility. To address this, we propose Concept-Guided Routing framework (CoGR-MoE), which incorporates semantics of the answer options to guide expert selection in the training phase. Next, option features are used to reweight the selected experts, producing discriminative representations for each candidate option. These option-level representations are further used for option comparison and optimized via contrastive learning. The experimental results indicate that CoGR-MoE delivers strong performance across multiple VQA tasks, demonstrating the effectiveness of our approach.

The accurate extraction of scientific measurements from literature is a critical yet challenging task in AI4Science, enabling large-scale analysis and integration of quantitative research findings. However, Large Language Models (LLMs) frequently exhibit severe hallucinations, which significantly undermine the reliability of automated scientific document understanding systems. To address this problem, we propose MeasHalu, a novel framework for mitigating scientific measurement hallucinations through enhanced reasoning and targeted optimization. We first present a fine-grained taxonomy of measurement-specific hallucinations, categorizing errors across quantities, units, modifiers, and relations. Our approach incorporates a two-stage reasoning-aware fine-tuning strategy using augmented scientific data and process-based supervision. Furthermore, we introduce a progressive reward curriculum designed to penalize specific hallucination types, significantly improving extraction faithfulness. Experimental results demonstrate that MeasHalu substantially reduces hallucination rates and improves overall accuracy on the MeasEval benchmark. This work provides a targeted solution to a key bottleneck in automated scientific knowledge extraction, facilitating more trustworthy and scalable machine-assisted scientific literature analysis.

Electroencephalography (EEG) foundation models have shown strong potential for learning generalizable representations from large-scale neural data, yet their clinical deployment is hindered by distribution shifts across clinical settings, devices, and populations. Test-time adaptation (TTA) offers a promising solution by enabling models to adapt to unlabeled target data during inference without access to source data, a valuable property in healthcare settings constrained by privacy regulations and limited labeled data. However, its effectiveness for EEG remains largely underexplored. In this work, we introduce NeuroAdapt-Bench, a systematic benchmark for evaluating test-time adaptation methods on EEG foundation models under realistic distribution shifts. We evaluate representative TTA approaches from other domains across multiple pretrained foundation models, diverse downstream tasks, and heterogeneous datasets spanning in-distribution, out-of-distribution, and extreme modality shifts (e.g., Ear-EEG). Our results show that standard TTA methods yield inconsistent gains and often degrade performance, with gradient-based approaches particularly prone to heavy degradation. In contrast, optimization-free methods demonstrate greater stability and more reliable improvements. These findings highlight the limitations of existing TTA techniques in EEG, provide guidance for future development, and underscore the need for domain-specific adaptation strategies.

Detecting AI-generated text is an important but challenging problem. Existing likelihood-based detection methods are often sensitive to content complexity and may exhibit unstable performance. In this paper, our key insight is that modern Large Language Models (LLMs) undergo alignment (including fine-tuning and preference tuning), leaving a measurable distributional imprint. We theoretically derive this imprint by abstracting the alignment process as a sequence of constrained optimization steps, showing that the log-likelihood ratio can naturally decompose into implicit instructional biases and preference rewards. We refer to this quantity as the Alignment Imprint. Furthermore, to mitigate the instability in high-entropy regions, we introduce Log-likelihood Alignment Preference Discrepancy (LAPD), a standardized information-weighted statistic based on alignment imprint. We provide statistical guarantee that alignment-based statistics dominate Fast-DetectGPT in performance. We also theoretically show that LAPD strictly improves the unweighted alignment scores when the aligned and base models are close in distribution. Extensive experiments show that LAPD achieves an improvement 45.82% relative to the strongest existing baselines, yielding large and consistent gains across all settings.

Climate research is pivotal for mitigating global environmental crises, yet the accelerating volume of multi-scale datasets and the complexity of analytical tools have created significant bottlenecks, constraining scientific discovery to fragmented and labor-intensive workflows. While the emergence Large Language Models (LLMs) offers a transformative paradigm to scale scientific expertise, existing explorations remain largely confined to simple Question-Answering (Q&A) tasks. These approaches often oversimplify real-world challenges, neglecting the intricate physical constraints and the data-driven nature required in professional climate science.To bridge this gap, we introduce ClimAgent, a general-purpose autonomous framework designed to execute a wide spectrum of research tasks across diverse climate sub-fields. By integrating a unified tool-use environment with rigorous reasoning protocols, ClimAgent transcends simple retrieval to perform end-to-end modeling and analysis.To foster systematic evaluation, we propose ClimaBench, the first comprehensive benchmark for real-world climate discovery. It encompasses challenging problems spanning 5 distinct task categories derived from professional scenarios between 2000 and 2025. Experiments on ClimaBench demonstrate that ClimAgent significantly outperforms state-of-the-art baselines, achieving a 40.21% improvement over original LLM solutions in solution rigorousness and practicality. Our code are available at https://github.com/usail-hkust/ClimAgent.

Diffusion models (DMs) have recently shown remarkable performance on inverse problems (IPs). Optimization-based methods can fast solve IPs using DMs as powerful regularizers, but they are susceptible to local minima and noise overfitting. Although DMs can provide strong priors for Bayesian approaches, enforcing measurement consistency during the denoising process leads to manifold infeasibility issues. We propose Noise-space Hamiltonian Monte Carlo (N-HMC), a posterior sampling method that treats reverse diffusion as a deterministic mapping from initial noise to clean images. N-HMC enables comprehensive exploration of the solution space, avoiding local optima. By moving inference entirely into the initial-noise space, N-HMC keeps proposals on the learned data manifold. We provide a comprehensive theoretical analysis of our approach and extend the framework to a noise-adaptive variant (NA-NHMC) that effectively handles IPs with unknown noise type and level. Extensive experiments across four linear and three nonlinear inverse problems demonstrate that NA-NHMC achieves superior reconstruction quality with robust performance across different hyperparameters and initializations, significantly outperforming recent state-of-the-art methods. The code is available at https://github.com/NA-HMC/NA-HMC.

Reinforcement Learning (RL) has achieved impressive success in post-training Large Language Models (LLMs) and Vision-Language Models (VLMs), with on-policy algorithms such as PPO, GRPO, and REINFORCE++ serving as the dominant paradigm. However, these methods discard all collected trajectories after a single gradient update, resulting in poor sample efficiency, particularly wasteful for agentic tasks where multi-turn environment interactions are expensive. While Experience Replay drives sample efficiency in classic RL by allowing agents to reuse past trajectories and prioritize informative ones, directly applying Prioritized Experience Replay (PER) to LLMs fails. The rapid policy evolution of billion-parameter models renders stored priorities stale, causing old high-priority trajectories to dominate sampling long after they have become uninformative. We propose Freshness-Aware PER, which addresses this priority staleness problem by augmenting any PER-based priority with a multiplicative exponential age decay grounded in effective sample size analysis. To the best of our knowledge, Freshness-Aware PER is the first work to successfully apply PER to LLM/VLM reinforcement learning. We evaluate on eight multi-step agentic, reasoning, and math competition tasks with 0.5B, 3B, and 7B models. Freshness-Aware PER significantly outperforms on-policy baselines, achieving +46% on NQ Search, +367% on Sokoban, and +133% on VLM FrozenLake, while standard PER without age decay consistently degrades performance. Our code is publicly available at https://github.com/Vision-CAIR/Freshness-Aware-PER.

Languages encode distinct abstractions and inductive priors, yet most large language models (LLMs) overlook this diversity by reasoning in a single dominant language. In this work, we introduce x1, a family of reasoning models that can adaptively reason in an advantageous language on a per-instance basis. To isolate the effect of reasoning-language choice, x1 is constructed without expanding the model's knowledge boundaries and is trained by contrasting linguistically distinct reasoning trajectories for the same input. Our extensive experiments demonstrate the benefits of adaptive multilingual reasoning across multilingual mathematical reasoning and culturally grounded tasks. Moreover, our results challenge a simplistic view of scaling laws: while scaling reduces cross-lingual disparities in procedural domains such as math reasoning, it does not eliminate the advantages of culture-associated languages in culturally grounded tasks, as we empirically show that such reasoning enables more efficient and accurate cultural knowledge recall. Overall, our findings establish language choice as a functional component of reasoning, with implications for building more generalist and globally competent reasoning models.

Safety alignment in large language models (LLMs) is primarily evaluated under open-ended generation, where models can mitigate risk by refusing to respond. In contrast, many real-world applications place LLMs in structured decision-making tasks, such as multiple-choice questions (MCQs), where abstention is discouraged or unavailable. We identify a systematic failure mode in this setting: reformulating harmful requests as forced-choice MCQs, where all options are unsafe, can systematically bypass refusal behavior, even in models that consistently reject equivalent open-ended prompts. Across 14 proprietary and open-source models, we show that forced-choice constraints sharply increase policy-violating responses. Notably, for human-authored MCQs, violation rates follow an inverted U-shaped trend with respect to structural constraint strength, peaking under intermediate task specifications, whereas MCQs generated by high-capability models yield near-saturation violation rates across constraints and exhibit strong cross-model transferability. Our findings reveal that current safety evaluations substantially underestimate risks in structured task settings and highlight constrained decision-making as a critical and underexplored surface for alignment failures.

Decentralized Autonomous Organizations (DAOs) are inclined explore Small Language Models (SLMs) as edge-native constitutional firewalls to vet proposals and mitigate semantic social engineering. While scaling inference-time compute (System 2) enhances formal logic, its efficacy in highly adversarial, cryptoeconomic governance environments remains underexplored. To address this, we introduce Sentinel-Bench, an 840-inference empirical framework executing a strict intra-model ablation on Qwen-3.5-9B. By toggling latent reasoning across frozen weights, we isolate the impact of inference-time compute against an adversarial Optimism DAO dataset. Our findings reveal a severe compute-accuracy inversion. The autoregressive baseline (System 1) achieved 100% adversarial robustness, 100% juridical consistency, and state finality in under 13 seconds. Conversely, System 2 reasoning introduced catastrophic instability, fundamentally driven by a 26.7% Reasoning Non-Convergence (cognitive collapse) rate. This collapse degraded trial-to-trial consensus stability to 72.6% and imposed a 17x latency overhead, introducing critical vulnerabilities to Governance Extractable Value (GEV) and hardware centralization. While rare (1.5% of adversarial trials), we empirically captured "Reasoning-Induced Sycophancy," where the model generated significantly longer internal monologues (averaging 25,750 characters) to rationalize failing the adversarial trap. We conclude that for edge-native SLMs operating under Byzantine Fault Tolerance (BFT) constraints, System 1 parameterized intuition is structurally and economically superior to System 2 iterative deliberation for decentralized consensus. Code and Dataset: https://github.com/smarizvi110/sentinel-bench

Large Language Model (LLM) agents are increasingly extended at runtime via skill packages, structured natural-language instruction bundles loaded from a well-known directory. Community install tooling and registries exist, but two gaps persist: no public tool scores skill packages against Anthropic's published format specification, and no mechanism bundles related skills with the shared context they need to remain mutually coherent. We present Skilldex, a package manager and registry for agent skill packages addressing both gaps. The two novel contributions are: (1) compiler-style format conformance scoring against Anthropic's skill specification, producing line-level diagnostics on description specificity, frontmatter validity, and structural adherence; and (2) the skillset abstraction, a bundled collection of related skills with shared assets (vocabulary files, templates, reference documents) that enforce cross-skill behavioral coherence. Skilldex also provides supporting infrastructure: a three-tier hierarchical scope system, a human-in-the-loop agent suggestion loop, a metadata-only community registry, and a Model Context Protocol (MCP) server. The system is implemented as a TypeScript CLI (skillpm / spm) with a Hono/Supabase registry backend, and is open-source.

As large language models (LLMs) evolve from conversational assistants into agents capable of handling complex tasks, they are increasingly deployed in high-risk domains. However, existing benchmarks largely rely on mixed queries and posterior evaluation, output-level scoring, which quantifies hallucination severity but offers limited insight into where and why hallucinations arise in the generation pipeline. We therefore reformulate hallucination evaluation as a diagnostic problem and propose PRISM, a controlled benchmark that disentangles hallucinations into four dimensions: knowledge missing, knowledge errors, reasoning errors, and instruction-following errors, grounded in three stages of generation (memory, instruction, and reasoning). PRISM contains 9,448 instances across 65 tasks and supports fine-grained, stage-aware diagnostic evaluation. Evaluating 24 mainstream open-source and proprietary LLMs, we uncover consistent trade-offs across instruction following, memory retrieval, and logical reasoning, showing that mitigation strategies often improve specific dimensions at the expense of others. We hope PRISM provides a framework for understanding the specific mechanisms behind LLMs hallucinations, ultimately accelerating the development of trustworthy large language models.

Native Omni-modal Large Language Models (OLLMs) have shifted from pipeline architectures to unified representation spaces. However, this native integration gives rise to a critical yet underexplored phenomenon: modality preference. To bridge this gap, we first systematically quantify modality preference of OLLMs using a newly-curated conflict-based benchmark and the modality selection rate metric. Our evaluation of ten representative OLLMs reveals a notable paradigm shift: unlike the ``text-dominance'' of traditional VLMs, most OLLMs exhibit a pronounced visual preference. To further understand the underlying mechanism, we conduct layer-wise probing and demonstrate that such modality preference is not static but emerges progressively in the mid-to-late layers. Building upon these insights, we leverage these internal signals to diagnose cross-modal hallucinations, achieving competitive performance across three downstream multi-modal benchmarks without task-specific data. Our work provides both a mechanistic understanding and a practical tool for building more trustworthy OLLMs. Our code and related resources are publicly available at: https://github.com/icip-cas/OmniPreference

Many automated market makers can be understood through the geometry of their trading orbits, the sets of states reachable from one another through swaps. In prominent designs, this geometry is captured by a simple closed-form invariant such as the constant product $xy$ in Uniswap or a weighted geometric mean $x^w y^{1-w}$ in Balancer. This paper explains why these forms arise by deriving them from three basic assumptions: validity invariance (swaps preserve the validity of states), Pareto efficiency (no state on an orbit weakly dominates another), and unit invariance (changing measurement units does not change the mechanism). Together, these force every trading orbit of a two-asset AMM to be a level set of a weighted geometric mean $x^w y^{1-w}$. Applied pairwise, the axioms extend the classification to $n$-asset pools: orbits are level sets of $\prod_i x_i^{w_i}$ with positive weights $w_i$ summing to $1$. Imposing token-relabeling symmetry then pins down the weights, recovering the constant-product form $xy$ in the two-asset case and $\prod_i x_i$ in general. The main text provides an intuitive proof sketch and discusses fees and liquidity operations. Complete proofs and a machine-checked Lean 4 formalization accompany the paper.

Designing proteins that satisfy natural language functional requirements is a central goal in protein engineering. A straightforward baseline is to fine-tune generic instruction-tuned LLMs as direct text-to-sequence generators, but this is data- and compute-hungry. With limited supervision, LLMs can produce coherent plans in text yet fail to reliably realize them as sequences. This plan-execute gap motivates ProtoCycle, an agentic framework for protein design that uses LLMs primarily to drive a multi-round, feedback-driven decision cycle. ProtoCycle couples an LLM planner with a lightweight tool environment designed to emulate the iterative workflow of human protein engineering and uses LLM-driven reflection on tool feedback to revise plans. Trained with supervised trajectories and online reinforcement learning, ProtoCycle achieves strong language alignment while maintaining competitive foldability, and ablations show that reflection substantially improves sequence quality.

We propose Physics-Informed Tracking (PIT), a video-based framework for tracking a single particle from video, where a neural network autoencoder localizes a particle as a heatmap peak (landmark) and a differentiable physics module embedded in the autoencoder constrains several landmarks over time (a trajectory) to satisfy known dynamics. The novel Physics-Informed Landmark Loss (PILL) compares this predicted trajectory back against the landmarks, enforcing physical consistency without labels. Its supervised variant (PILLS) instead compares the prediction against ground-truth position, velocity, and bounce from simulation, enabling end-to-end backpropagation. To support supervised and unsupervised learning, we use an autoencoder with a split bottleneck that separates A) tracking-related structure via landmark heatmaps from B) background noise and subsequent image reconstruction. We evaluate a replicated 26 factorial design (n = 4 replicates, 64 configurations), showing that PILLS consistently achieves sub-pixel tracking accuracy for the bilinear and physics-refined decoder outputs under both clean and noisy conditions.

Factor-based Structural Equation Modeling (SEM) relies on likelihood-based estimation assuming a nonsingular sample covariance matrix, which breaks down in small-sample settings with $p>n$. To address this, we propose a novel estimation principle that reformulates the covariance structure into self-covariance and cross-covariance components. The resulting framework defines a likelihood-based feasible set combined with a relative error constraint, enabling stable estimation in small-sample settings where $p>n$ for sign and direction. Experiments on synthetic and real-world data show improved stability, particularly in recovering the sign and direction of structural parameters. These results extend covariance-based SEM to small-sample settings and provide practically useful directional information for decision-making.

Reinforcement learning from verifiable rewards (RLVR) has demonstrated remarkable effectiveness in improving the reasoning capabilities of large language models. As models evolve into natively multimodal architectures, extending RLVR to video understanding becomes increasingly important yet remains largely unexplored, due to the diversity of video task types, the computational overhead of repeatedly decoding and preprocessing high-dimensional visual inputs, and the difficulty of reproducible evaluation across numerous sensitive hyperparameters. Existing open-source RL training frameworks provide solid infrastructure for text and image scenarios but lack systematic optimizations tailored for video modality. In this work, we present \textbf{EasyVideoR1}, a complete and efficient reinforcement learning framework specifically designed for training large vision-language models on video understanding tasks. EasyVideoR1 makes the following contributions: (1) a full video RL training pipeline with offline preprocessing and tensor caching that eliminates redundant video decoding and yields a 1.47 $\times$ throughput improvement; (2) a comprehensive, task-aware reward system covering 11 distinct video and image problem types with unified routing and modular extension; (3) a mixed offline-online data training paradigm that combines curated high-quality trajectories with on-policy exploration, benefiting the learning of more challenging tasks; (4) joint image-video training with independently configurable pixel budgets, allowing the two modalities to mutually reinforce each other; and (5) an asynchronous multi-benchmark evaluation framework covering 22 mainstream video understanding benchmarks, with reproduced accuracy closely aligned with officially reported scores.

Large reasoning models that use long chain-of-thought excel at problem-solving yet waste compute on redundant checks. Curbing this overthinking is hard: training-time length penalties can cripple ability, while inference-time early-exit adds system overhead. To bridge this gap, we propose Step-GRPO, a novel post-training framework that internalizes dynamic early-exit capabilities directly into the model. Step-GRPO shifts the optimization objective from raw tokens to semantic steps by utilizing linguistic markers to structure reasoning. We introduce a Dynamic Truncated Rollout mechanism that exposes the model to concise high-confidence trajectories during exploration, synergized with a Step-Aware Relative Reward that dynamically penalizes redundancy based on group-level baselines. Extensive experiments across three model sizes on diverse benchmarks demonstrate that Step-GRPO achieves a superior accuracy-efficiency trade-off. On Qwen3-8B, our method reduces token consumption by 32.0\% compared to the vanilla model while avoiding the accuracy degradation observed in traditional length-penalty methods.

Existing feature-interpretation pipelines typically operate on uniformly sampled units, but only a small fraction of cross-layer transcoder (CLT) features matter for a target behavior, with the rest resulting in expensive feature explaining and evaluating costs. We introduce the first CLT-native end-to-end framework, PIE, connecting Pruning, automatic Interpretation, and interpretation Evaluation, enabling systematic measurement of behavioral fidelity and downstream interpretability under pruning. To achieve this, we propose Feature Attribution Patching (FAP), a patch-grounded attribution method that scores CLT features by aggregating gradient-weighted write contributions, and FAP-Synergy, a synergy-aware reranking procedure. We evaluate pruning using KL-divergence behavior retention and assess interpretation quality with FADE-style metrics. Across IOI and Doc-String, across budgets $K \in \{50, 100, 200, 400, 800\}$, and across FAP, FAP-Synergy, Activation-Magnitude, and ACDC-style pruning, the FAP family consistently achieves the best or near-best fidelity, with FAP-Synergy providing its clearest gains in strict-budget regimes. On IOI with CLTs for Llama-3.2-1B and Gemma-2-2B, pruning to $K=100$ features matches the KL fidelity that random selection from the active feature set requires $\approx 4$k features to achieve ($\approx 40\times$ compression), enabling $\approx 40\times$ fewer interpretation/evaluation calls while substantially reducing low-quality features.

Parameter-free stochastic optimization aims to design algorithms that are agnostic to the underlying problem parameters while still achieving convergence rates competitive with optimally tuned methods. While some parameter-free methods do not require the specific values of the problem parameters, they still rely on prior knowledge, such as the lower or upper bounds of them. We refer to such methods as ``partially parameter-free''. In this work, we target achieving ``fully parameter-free'' methods, i.e., the algorithmic inputs do not need to satisfy any unverifiable condition related to the true problem parameters. We propose a powerful and general grid search framework, named \textsc{Grasp}, with a novel self-bounding analysis technique that effectively determines the search ranges of parameters, in contrast to previous work. Our method demonstrates generality in: (i) the non-convex case, where we propose a fully parameter-free method that achieves near-optimal convergence rate, up to logarithmic factors; (ii) the convex case, where our parameter-free methods are competitive with strong performance in terms of acceleration and universality. Finally, we contribute a sharper guarantee for the model ensemble, a final step of the grid search framework, under interpolated variance characterization.

In long-context decoding for LLMs and LMMs, attention becomes increasingly memory-bound because each decoding step must load a large amount of KV-cache data from GPU memory. Existing acceleration strategies often trade efficiency for accuracy by relying on heuristic pruning that may discard useful information. At a deeper level, they also tend to indiscriminately preserve all high-scoring tokens, treat early tokens as indispensable anchors, or rely on heuristic head routing, reflecting an insufficient mechanistic understanding of the attention sink phenomenon. In this paper, we show that the attention sink phenomenon corresponds to a stable, reachable, and error-controllable fixed point constructed during training. Based on this insight, we propose SinkRouter, a training-free selective routing framework that detects the sink signal and skips computations that would otherwise produce near-zero output. To translate this mechanism into real-world acceleration, we develop a hardware-aware Triton kernel with block-level branching and Split-K parallelism. We conduct extensive evaluations on a diverse suite of long-context benchmarks, including LongBench, InfiniteBench, CVBench, MileBench, and MMVP, using both text-only and multimodal backbones such as Llama-3.1-8B, Llama-3.1-70B, Yi-9B-200K, LLaVA-1.5-7B, and LLaVA-1.5-13B. Across these settings, SinkRouter consistently improves decoding efficiency while maintaining competitive accuracy, and reaches 2.03x speedup with a 512K context.

Cross-cultural entity translation remains challenging for large language models (LLMs) as literal or phonetic renderings are usually yielded instead of culturally appropriate translations in context. However, relevant knowledge may already be encoded in model parameters during large-scale pre-training. To incentivize the effective use of parametric knowledge, we propose EA-RLVR (Entity-Anchored Reinforcement Learning with Verifiable Rewards), a training framework that optimizes cross-cultural entity translation without relying on external knowledge bases. EA-RLVR anchors supervision on a verifiable, entity-level reward signal and incorporates lightweight structural gates to stabilize optimization. This design steers the model toward learning a robust reasoning process rather than merely imitating reference translations. We evaluate EA-RLVR on XC-Translate and observe consistent improvements in both entity translation accuracy and out-of-domain generalization. Specifically, training on merely 7k samples boosts Qwen3-14B's entity translation accuracy from 23.66\% to 31.87\% on a 50k test set comprising entirely unseen entities. The learned entity translation ability also transfers to general translation, yielding +1.35 XCOMET on WMT24++, which scales to +1.59 with extended optimization. Extensive analyses of $pass@k$ dynamics and reward formulations attribute these gains to superior sampling efficiency and a stable optimization landscape.

Early prediction of severe clinical deterioration and remaining length of stay can enable timely intervention and better resource allocation in high-acuity settings such as the ICU. This has driven the development of machine learning models that leverage continuous streams of vital signs and other physiological signals for real-time risk prediction. Despite their promise, existing methods have important limitations. Contrastive pretraining treats all patients as equally strong negatives, failing to capture clinically meaningful similarity between patients with related diagnoses. Meanwhile, downstream fine-tuning typically ignores complementary modalities such as clinical notes, which provide rich contextual information unavailable in physiological signals alone. To address these challenges, we propose OC-Distill, a two-stage framework that leverages multimodal supervision during training while requiring only vital signs at inference. In the first stage, we introduce an ontology-aware contrastive objective that exploits the ICD hierarchy to quantify patient similarity and learn clinically grounded representations. In the second stage, we fine-tune the pretrained encoder via cross-modal knowledge distillation, transferring complementary information from clinical notes into the model. Across multiple ICU prediction tasks on MIMIC, OC-Distill demonstrates improved label efficiency and achieves state-of-the-art performance among methods that use only vital signs at inference.

A central question in computational neuroscience is whether the learning rule used to train a neural network determines how well its internal representations align with those of the human visual cortex. We present a systematic comparison of four learning rules -- backpropagation (BP), feedback alignment (FA), predictive coding (PC), and spike-timing-dependent plasticity (STDP) -- applied to identical convolutional architectures and evaluated against human fMRI data from the THINGS-fMRI dataset (720 stimuli, 3 subjects) using Representational Similarity Analysis (RSA). Crucially, we include an untrained random-weights baseline that reveals the dominant role of architecture. We find that early visual alignment (V1/V2) is primarily architecture-driven: an untrained CNN achieves rho = 0.071, statistically indistinguishable from BP (rho = 0.072, p = 0.43). Learning rules only differentiate at higher visual areas: BP dominates at LOC/IT, and PC with local Hebbian updates achieves IT alignment statistically indistinguishable from BP (p = 0.18). FA consistently impairs representations below the random baseline at V1. Partial RSA confirms all effects survive pixel-similarity control. These results demonstrate that the relationship between learning rules and cortical alignment is region-specific: architecture determines early alignment, while supervised objectives drive late alignment.

Neuro-symbolic Reinforcement Learning (NeSy-RL) combines symbolic reasoning with gradient-based optimization to achieve interpretable and generalizable policies. Relational concepts, such as "left of" or "close by", serve as foundational building blocks that structure how agents perceive and act. However, conventional approaches require human experts to manually define these concepts, limiting adaptability since concept semantics vary across environments. We propose GRAIL (Grounding Relational Agents through Interactive Learning), a framework that autonomously grounds relational concepts through environmental interaction. GRAIL leverages large language models (LLMs) to provide generic concept representations as weak supervision, then refines them to capture environment-specific semantics. This approach addresses both sparse reward signals and concept misalignment prevalent in underdetermined environments. Experiments on the Atari games Kangaroo, Seaquest, and Skiing demonstrate that GRAIL matches or outperforms agents with manually crafted concepts in simplified settings, and reveals informative trade-offs between reward maximization and high-level goal completion in the full environment.

AI agents increasingly call external tools (file system, network, APIs) through the Model Context Protocol (MCP). These tool calls are the agent's syscalls -- privileged operations with side effects on shared state -- yet today's safety enforcement lives entirely in userspace, where a 10-line script can bypass it. I propose Governed MCP, a kernel-resident tool governance gateway built on a logit-based safety primitive (ProbeLogits, companion paper: arXiv:2604.11943). The gateway interposes on every MCP tool call in a 6-layer pipeline: schema validation, trust tier check, rate limit, adversarial pre-filter, ProbeLogits gate (the load-bearing semantic check), and constitutional policy match, with a Blake3-hashed audit chain. I implement Governed MCP in Anima OS, a bare-metal x86_64 OS in approximately 86,000 lines of Rust. The five non-inference layers add 65.3 microseconds of overhead per call; ProbeLogits adds 65 ms (per-token-class semantic decision) on 7B Q4_0. A 4-config ablation on a 101-prompt MCP-domain benchmark shows that removing the ProbeLogits layer collapses F1 from 0.773 to 0.327 (Delta F1 = -0.446) -- hand-rule firewalling alone is insufficient. All 15 WASM-to-system host functions in the runtime route through the gateway (complete mediation of the WASM ABI surface; the scope and caveats of this claim are stated in Section 4.6); a 10-LoC userspace bypass that defeats existing guardrail libraries is structurally impossible against the kernel-resident gate.

In this paper, we consider the problem of extraction of most informative features from time series that are regarded as observed values of stochastic processes satisfying the It{ô} stochastic differential equations with unknown random drift and diffusion coefficients. We do not attract any additional information and use only the information contained in the time series as it is. Therefore, as additional features, we use the parameters of statistically adjusted mixture-type models of the observed regularities of the behavior of the time series. Several algorithms of construction of these parameters are discussed. These algorithms are based on statistical reconstruction of the coefficients which, in turn, is based on statistical separation of normal mixtures. We obtain two types of parameters by the techniques of the uniform and non-uniform statistical reconstruction of the coefficients of the underlying It{ô} process. The reconstructed coefficients obtained by uniform techniques do not depend on the current value of the process, while the non-uniform techniques reconstruct the coefficients with the account of their dependence on the value of the process. Actually, the non-uniform techniques used in this paper represent a stochastic analog of the Taylor expansion for the time series. The efficiency of the obtained additional features is compared by using them in the autoregressive algorithms of prediction of time series. In order to obtain pure conclusion that is not affected by unwanted factors, say, related to a special choice of the architecture of the neural network prediction methods, we used only simple autoregressive algorithms. We show that the use of additional statistical features improves the prediction.

The deployment of long-context Large Language Models (LLMs) poses significant challenges due to the intense computational cost of self-attention and the substantial memory overhead of the Key-Value Cache (KV Cache). In this paper, we introduce HieraSparse, a hierarchical KV Cache compression framework with acceleration kernels that leverage GPU sparse tensor cores to speed up semi-structured KV Cache attention for both the prefill and decode phases. With the hierarchical design, our method allows for a flexible quality-sparsity trade-off and successfully converts sparsity into efficiency. Compared to the state-of-the-art decode method that utilizes unstructured sparsity, HieraSparse achieves $\mathbf{1.2\times}$ KV compression ratio and $\mathbf{4.57\times}$ attention speedup at the same sparsity level. Furthermore, we extended the semi-structured KV Cache pruning to the prefill stage, which demonstrated up to $\mathbf{1.85\times}$ attention speedup at the highest sparsity. Lastly, we evaluate the generation quality of HieraSparse with a simple magnitude-based pruning method, and the results show that $\mathbf{1.37\times}$ prefill speedup and $\mathbf{1.77\times}$ decode speedup can be achieved without significant quality drop. The codebase can be found at https://github.com/psl-ntu/HieraSparse.

Recent deployments of large language models (LLMs) as autonomous trading agents raise questions about whether financial decision-making competence generalizes beyond specific market patterns and how it should be trained and evaluated in noisy markets lacking ground truth. We propose a structured framework for training and evaluating such models. Central to our approach is a curated, multiple-choice question (MCQ) dataset derived from classic textbooks and historical markets, verified by an AI committee, enriched with structured reasoning traces, and augmented to reduce shortcut learning. To evaluate whether performance on isolated MCQs generalizes to real-world trading, we introduce a two-stage protocol combining test-set evaluation with an MCQ-based chronological trading simulation. Extensive evaluations across market regimes provide statistically robust evidence that open models trained with our framework exhibit competitive, risk-aware behavior over time, outperform open-source baselines, and approach frontier-model performance at smaller scale. We release the dataset and evaluation framework to support further research.

Standard supervised learning optimizes for predictive accuracy but remains agnostic to the internal geometry of learned features, often yielding representations that are entangled and brittle. We propose Class-Conditional Activation Regularization (CCAR) to explicitly engineer the feature space, imposing a block-diagonal structure via a soft inductive bias. By shaping the latent representation to confine class energy to orthogonal subspaces, we create an intrinsic geometric scaffold that naturally filters noise and adversarial perturbations. We provide theoretical analysis linking this structural constraint to the maximization of the Fisher Discriminant Ratio, establishing a formal connection between geometric disentanglement and algorithmic stability. Empirically, this approach demonstrates that robustness is an emergent property of a well-engineered feature space, significantly outperforming baselines on label noise and input corruption benchmarks.

Accurate traffic forecasting is crucial for intelligent transportation systems, supporting effective traffic management, congestion reduction, and informed urban planning. However, traditional models often fail to adequately capture the intricate spatio-temporal dependencies present in traffic data. To overcome these limitations, we introduce GAMMA-Net, a novel approach that integrates Graph Attention Networks (GAT) with multi-axis Selective State Space Models (Mamba). The GAT component uses a self-attention mechanism to dynamically adjust the influence of nodes within the traffic network, enabling adaptive spatial dependency modeling based on real-time conditions. Simultaneously, the Mamba module efficiently models long-term temporal and spatial dynamics without the heavy computational cost of conventional recurrent architectures. Extensive experiments on several benchmark traffic datasets, including METR-LA, PEMS-BAY, PEMS03, PEMS04, PEMS07, and PEMS08, show that GAMMA-Net consistently outperforms existing state-of-the-art models across different prediction horizons, achieving up to a 16.25% reduction in Mean Absolute Error (MAE) compared to baseline models. Ablation studies highlight the critical contributions of both the spatial and temporal components, emphasizing their complementary role in improving prediction accuracy. In conclusion, the GAMMA-Net model sets a new standard in traffic forecasting, offering a powerful tool for next-generation traffic management and urban planning. The code for this study is available at https://github.com/hdy6438/GAMMA-Net

The proliferation of online debate platforms and social media has led to an unprecedented volume of argumentative content on controversial topics from multiple perspectives. While this wealth of perspectives offers opportunities for developing critical thinking and breaking filter bubbles (Pariser 2011), the sheer volume and complexity of arguments make it challenging for readers to synthesize and comprehend diverse viewpoints effectively. We present an unsupervised graph-based approach for community-based argument organization that helps users navigate and understand complex argumentative landscapes. Our system analyzes collections of topic-focused articles and constructs a rich interaction graph by capturing multiple relationship types between arguments: topic similarity, semantic coherence, shared keywords, and common entities. We then employ community detection to identify argument communities that reveal homogeneous and heterogeneous viewpoint distributions. The detected communities are simplified through strategic graph operations to present users with digestible, yet comprehensive summaries of key argumentative patterns. Our approach requires no training data and can effectively process hundreds of articles while preserving nuanced relationships between arguments. Experimental results demonstrate our system's ability to identify meaningful argument communities and present them in an interpretable manner, facilitating users' understanding of complex socio-political debates.

This dissertation explores how deep generative models can advance the analysis of challenging biological problems by integrating domain knowledge with deep learning. It focuses on two areas: DNA reaction kinetics and cryogenic electron microscopy (cryo-EM). In the first part, we present ViDa, a biophysics-informed framework leveraging variational autoencoders (VAEs) and geometric scattering transforms to generate biophysically-plausible embeddings of DNA reaction kinetics simulations. These embeddings are reduced to a two-dimensional space to visualize DNA hybridization and toehold-mediated strand displacement reactions. ViDa preserves structure and clusters trajectory ensembles into reaction pathways, making simulation results more interpretable and revealing new mechanistic insights. In the second part, we address key challenges in cryo-EM density map interpretation and protein structure modeling. We provide a comprehensive review and benchmarking of deep learning methods for atomic model building, with improved evaluation metrics and practical guidance. We then present Struc2mapGAN, a generative adversarial network that synthesizes high-fidelity experimental-like cryo-EM density maps from protein structures. Finally, we present CryoSAMU, a structure-aware multimodal U-Net that enhances intermediate-resolution cryo-EM maps by integrating density features with structural embeddings from protein language models via cross-attention. Overall, these contributions demonstrate the potential of deep generative models to interpret DNA reaction mechanisms and advance cryo-EM density map analysis and protein structure modeling.

Fast execution of contact-rich manipulation is critical for practical deployment, yet providing fast demonstrations for imitation learning (IL) remains challenging: humans cannot demonstrate at high speed, and naively accelerating demonstrations alters contact dynamics and induces large tracking errors. We present a method to autonomously refine time-accelerated demonstrations by repurposing Iterative Reference Learning Control (IRLC) to iteratively update the reference trajectory from observed tracking errors. However, applying IRLC directly at high speed tends to produce larger early-iteration errors and less stable transients. To address this issue, we propose Incremental Iterative Reference Learning Control (I2RLC), which gradually increases the speed while updating the reference, yielding high-fidelity trajectories. We validate on real-robot whiteboard erasing and peg-in-hole tasks using a teleoperation setup with a compliance-controlled follower and a 3D-printed haptic leader. Both IRLC and I2RLC achieve up to 10x faster demonstrations with reduced tracking error; moreover, I2RLC improves spatial similarity to the original trajectories by 22.5% on average over IRLC across three tasks and multiple speeds (3x-10x). We then use the refined trajectories to train IL policies; the resulting policies execute faster than the demonstrations and achieve 100% success rates in the peg-in-hole task at both seen and unseen positions, with I2RLC-trained policies exhibiting lower contact forces than those trained on IRLC-refined demonstrations. These results indicate that gradual speed scheduling coupled with reference adaptation provides a practical path to fast, contact-rich IL.

Fine-grained semantic segmentation of transmission-corridor point clouds is fundamental for intelligent power-line inspection. However, current progress is limited by realistic data scarcity and the difficulty of modeling global corridor structure and local geometric details in long, heterogeneous scenes. Existing public datasets usually provide only a few coarse categories or short cropped scenes which overlook long-range structural dependencies, severe long-tail distributions, and subtle distinctions among safety-critical components. As a result, current methods are difficult to evaluate under realistic inspection settings, and their ability to preserve and integrate complementary global and local cues remains unclear. To address the above challenges, we introduce TowerDataset, a heterogeneous benchmark for transmission-corridor segmentation. TowerDataset contains 661 real-world scenes and about 2.466 billion points. It preserves long corridor extents, defines a fine-grained 22-class taxonomy, and provides standardized splits and evaluation protocols. In addition, we present a global-local fusion framework which preserves and fuses whole-scene and local-detail information. A whole-scene branch with NoCrop training and prototypical contrastive learning captures long-range topology and contextual dependencies. A block-wise local branch retains fine geometric structures. Both predictions are then fused and refined by geometric validation. This design allows the model to exploit both global relationships and local shape details when recognizing rare and confusing components. Experiments on TowerDataset and two public benchmarks demonstrate the challenge of the proposed benchmark and the robustness of our framework in real, complex, and heterogeneous transmission-corridor scenes. The dataset will be released soon at https://huggingface.co/datasets/tccx18/Towerdataset/tree/main.

Large language models (LLMs) tuned for safety often avoid acknowledging demographic differences, even when such acknowledgment is factually correct (e.g., ancestry-based disease incidence) or contextually justified (e.g., religious hiring preferences). This identity-blindness yields incorrect responses, unnecessary refusals, or generic "equal-treatment" defaults. We study this via difference-awareness classification: given a question involving demographic groups, the task is not to answer directly, but to classify whether a correct answer requires recognizing group differences (yes) or whether groups should be treated identically (no). Crucially, fine-tuning for accuracy triggers harm drift: model-generated explanations become increasingly harmful as decision accuracy improves, whether by elaborating harmful content, introducing problematic assumptions, or failing to flag harms the baseline identified. To mitigate this, we introduce DART (Distill--Audit--Repair Training), which distills label-conditioned reasoning from a teacher, audits outputs for harm drift cases relative to baseline, and repairs problematic cases via severity-weighted fine-tuning. On eight benchmarks, DART improves Llama-3-8B-Instruct accuracy from 39.0% to 68.8%, with largest gains on equal-treatment prompts (11.3% -> 72.6%), while reducing harm drift cases by 72.6%. It also transfers to 280 open-ended real-world queries across medical, legal, policy, and educational domains, improving difference-appropriate responses from 39.8% to 77.5% while reducing refusals from 34.3% to 3.0%. Our results demonstrate that accuracy and safety need not conflict when explicit detection and repair mechanisms are in place.

This thesis develops numerical and theoretical approaches for understanding and analyzing singularity formation in Partial Differential Equations (PDEs). The singularity formation in the Navier-Stokes Equation (NSE) is famously challenging as one of the seven Clay Prize problems. Unlike simpler equations such as the Nonlinear Heat (NLH) or Keller-Segel (KS) equations, where formal asymptotics near blowup are better understood, the intrinsic complexity of NSE makes quantitative analytical treatment difficult, if not impossible, without numerical guidance. Building on numerical insights, we introduce a robust analytical framework to simplify and systematize pen-and-paper proofs for simpler singular PDEs. We present a novel approach based on enforcing vanishing modulation conditions for perturbations around approximate blowup profiles, complemented by singularly weighted energy estimates. We demonstrate the efficacy of our method on PDEs with complicated asymptotics, such as NLH and the Complex Ginzburg-Landau (CGL) equation, and address the open problem of singularity formation in the 3D KS equation with logistic damping. We develop and refine numerical approaches that facilitate deeper insights into singularity formation. We demonstrate that machine learning methods significantly enhance our capability to identify and characterize potential blowup solutions with high precision. We improve on existing Physics-Informed Neural Network (PINN) and Neural Operator (NO) frameworks. Moreover, we present a novel machine learning paradigm, the Kolmogorov-Arnold Network (KAN) architecture, whose interpretability and excellent scaling properties are achieved through learnable nonlinearities.

Land surface temperature (LST) super-resolution is important for environmental monitoring. However, it remains challenging as coarse thermal observations severely underdetermine fine-scale structure. In this paper, we propose Earth Foundation Model-guided Diffusion (EFDiff), a novel framework for super-resolution under extreme spatial degradation. EFDiff uses the Prithvi-EO-2.0 Earth foundation model to encode high-resolution multispectral reflectance into geospatial embeddings, which are injected into the denoising network via cross-attention to guide fine-scale reconstruction from highly degraded observations. We study two variants, EFDiff-$ε$ and EFDiff-$x_0$, which offer complementary trade-offs between perceptual realism and pixel-level fidelity. We evaluate EFDiff under an extreme $32\times$ scale gap using a globally diverse benchmark comprising 242,416 co-registered Landsat thermal-reflectance patches. Results show that EFDiff consistently outperforms baseline methods and that cross-attention conditioning by EFM is more effective than HLS channel concatenation. Although we present EFDiff in the context of LST super-resolution, the framework is broadly applicable to remote sensing problems in which pretrained geospatial representations can guide generative reconstruction.

Long-term memory is a critical challenge for Large Language Model agents, as fixed context windows cannot preserve coherence across extended interactions. Existing memory systems represent conversation history as unstructured embedding vectors, retrieving information through semantic similarity. This paradigm fails to capture the associative structure of human memory, wherein related experiences progressively strengthen interconnections through repeated co-activation. Inspired by cognitive neuroscience, we identify three mechanisms central to biological memory: association, consolidation, and spreading activation, which remain largely absent in current research. To bridge this gap, we propose HeLa-Mem, a bio-inspired memory architecture that models memory as a dynamic graph with Hebbian learning dynamics. HeLa-Mem employs a dual-level organization: (1) an episodic memory graph that evolves through co-activation patterns, and (2) a semantic memory store populated via Hebbian Distillation, wherein a Reflective Agent identifies densely connected memory hubs and distills them into structured, reusable semantic knowledge. This dual-path design leverages both semantic similarity and learned associations, mirroring the episodic-semantic distinction in human cognition. Experiments on LoCoMo demonstrate superior performance across four question categories while using significantly fewer context tokens. Code is available on GitHub: https://github.com/ReinerBRO/HeLa-Mem

We present enclawed, a hard-fork hardening framework built on top of the OpenClaw single-user personal artificial intelligence (AI) assistant gateway. enclawed targets deployments that need attestable peer trust, deny-by-default external connectivity, signed-module loading, and a tamper-evident audit trail typically regulated industries such as financial services, healthcare, defense contracting, regulated R&D, and government enclaves. The framework ships in two flavors: an open flavor that preserves OpenClaw compatibility while still emitting audit, classification, and data-loss-prevention (DLP) signals, and an enclaved flavor that activates strict allowlists, Federal Information Processing Standards (FIPS) cryptographic-module assertion, mandatory module-manifest signature verification, and high-assurance peer attestation for the Model Context Protocol (MCP). The classification ladder is fully data-driven: a deploying organization selects from five built-in presets (generic, US-government, healthcare, financial services, three-tier) or supplies its own JSON. We accompany the implementation with a security review, a 204-case test suite (146 unit tests, 58 adversarial pen-tests for tamper detection, signature forgery, egress bypass, trust-root mutation, DLP evasion, prompt injection, and code injection), real-time human-in-the-loop control (per-agent pause / resume / stop and approval queues), a memory-bounded secure transaction buffer with rollback (default cap 50% of system RAM, configurable), a strict-mode TypeScript typecheck of all 22 framework files, and a GitHub Actions workflow ready for continuous integration. enclawed is a hardening framework, not an accredited compliance certification. The deploying organization remains responsible for hardware, validated cryptographic modules, certified facilities, and assessor sign-off.

Semantic segmentation in hyperbolic space enables compact modeling of hierarchical structure while providing inherent uncertainty quantification. Prior approaches predominantly rely on the Poincaré ball model, which suffers from numerical instability, optimization, and computational challenges. We propose a novel, tractable, architecture-agnostic semantic segmentation framework (pixel-wise and mask classification) in the hyperbolic Lorentz model. We employ text embeddings with semantic and visual cues to guide hierarchical pixel-level representations in Lorentz space. This enables stable and efficient optimization without requiring a Riemannian optimizer, and easily integrates with existing Euclidean architectures. Beyond segmentation, our approach yields free uncertainty estimation, confidence map, boundary delineation, hierarchical and text-based retrieval, and zero-shot performance, reaching generalized flatter minima. We introduce a novel uncertainty and confidence indicator in Lorentz cone embeddings. Further, we provide analytical and empirical insights into Lorentz optimization via gradient analysis. Extensive experiments on ADE20K, COCO-Stuff-164k, Pascal-VOC, and Cityscapes, utilizing state-of-the-art per-pixel classification models (DeepLabV3 and SegFormer) and mask classification models (mask2former and maskformer), validate the effectiveness and generality of our approach. Our results demonstrate the potential of hyperbolic Lorentz embeddings for robust and uncertainty-aware semantic segmentation. Code is available at https://github.com/mxahan/Lorentz_semantic_segmentation.

Shanghai Composite Index prediction has become a hot issue for many investors and academic researchers. Deep learning models are widely applied in multivariate time series forecasting, including recurrent neural networks (RNN), convolutional neural networks (CNN), and transformers. Specifically, the Transformer encoder, with its unique attention mechanism and parallel processing capabilities, has become an important tool in time series prediction, and has an advantage in dealing with long sequence dependencies and multivariate data correlations. Drawing on the strengths of various models, we propose the CNN-Transformer-LSTM Networks (CTLNet). This paper explores the application of CTLNet for Shanghai Composite Index prediction and the comparative experiments show that the proposed model outperforms state-of-the-art baselines.

Privacy-preserving machine learning (PPML) has become increasingly important in applications where sensitive data must remain confidential. Homomorphic Encryption (HE) enables computation directly on encrypted data, allowing neural network inference without revealing raw inputs. While prior works have largely focused on inference over a single encrypted image, batch processing of encrypted inputs lags behind, despite being critical for high-throughput inference scenarios and training-oriented workloads. In this work, we address this gap by developing optimized algorithms for batched HE-friendly neural networks. We also introduced a pipeline architecture designed to maximize resource efficiency for different batch size execution. We implemented these algorithms and evaluated our work using HE-friendly ResNet-20 and ResNet-34 models on encrypted CIFAR-10 and CIFAR-100 datasets, respectively. For ResNet-20, our approach achieves an amortized inference time of 8.86 seconds per image when processing a batch of 512 encrypted images, with a peak memory usage of 98.96 GB. These results represent a 1.78x runtime improvement and a 3.74x reduction in memory usage compared to the state-of-the-art design. For the deeper ResNet-34 model, we achieve an amortized inference time of 28.14 on a batch of 256 encrypted images using 246.78GB of RAM

On-policy distillation (OPD) is an increasingly important paradigm for post-training language models. However, we identify a pervasive Scaling Law of Miscalibration: while OPD effectively improves task accuracy, it systematically traps models in severe overconfidence. We trace this failure to an information mismatch: teacher supervision is formed under privileged context available during training, whereas the deployed model must report confidence using only deployment-time information. We formalize this perspective theoretically, showing that teacher-conditioned success is generally not a valid target for deployment-time confidence and that helpful privileged context induces entropy collapse and a systematic optimism bias. To address this, we propose a calibration-aware OPD framework, CaOPD, that estimates empirical confidence from model rollouts, replaces self-reported confidence with this student-grounded target, and distills the revised response through the same self-distillation pipeline. Experiments across various models and domains show that CaOPD achieves Pareto-optimal calibration while maintaining competitive capability, generalizing robustly under out-of-distribution and continual learning. Our findings highlight that capability distillation does not imply calibrated confidence, and that confidence should be treated as an essential objective in post-training. Code: https://github.com/SalesforceAIResearch/CaOPD

Merging separately trained LoRA adapters is a practical alternative to joint multi-task training, but it often hurts performance. Existing methods usually treat the LoRA update $ΔW = BA$ as a single object and do not distinguish the two LoRA matrices. We show that the main source of LoRA merge interference comes from the output-side matrix $B$. Across tasks, $B$ repeatedly uses a small set of shared directions, while $A$ remains much more task-specific. As a result, the merged adapter overemphasizes these shared directions, and task-specific information is lost. We propose Pico (Pre-merge interference calibration in output-space), a data-free method that calibrates $B$ before merge by downscaling over-shared directions and then rescaling the merged update. Pico plugs directly into existing merging methods such as Task Arithmetic, TIES, and TSV-M. Across eight different benchmarks from math, coding, finance, and medical domains, Pico improves average accuracy by 3.4-8.3 points over the corresponding base method and achieves the best overall average performance. Pico also enables merged adapters to outperform the LoRA trained with all task data. These results show that LoRA merging works better when the two LoRA matrices are treated separately.

Multi-turn jailbreak attacks progressively erode LLM safety alignment across seemingly innocuous conversation turns, achieving success rates exceeding 90% against state-of-the-art models. Existing alignment-based and guardrail methods suffer from three key limitations: they require costly weight modification, evaluate each turn independently without modeling cumulative safety erosion, and detect attacks only after harmful content has been generated. To address these limitations, we first formulate the proactive early jailbreak detection problem with a new metric, detection lead, that measures how early an attack can be detected before the LLM complies. We then propose SAFEDREAM, a lightweight world-model-based framework that operates as an external module without modifying the LLM's weights. SAFEDREAM introduces three components: (1) a safety state world model that encodes LLM hidden states into a compact safety representation and predicts how it evolves across turns, (2) CUSUM detection that accumulates weak per-turn risk signals into reliable evidence, and (3) contrastive imagination that simultaneously rolls out attack and benign futures in latent space to issue early alarms before jailbreaks occur. On three multi-turn jailbreak benchmarks (XGuard-Train, SafeDialBench, SafeMTData) against 8 baselines, SAFEDREAM achieves the best detection timeliness across all benchmarks (1.06-1.20 turns before compliance) while maintaining competitive false positive rates and outperforming baselines in detection quality.

Vision Transformer (ViT) has brought new breakthroughs to the field of image classification by introducing the self-attention mechanism and Graph Convolutional Networks(GCN) have been proposed and successfully applied in data representation and analysis. However, there are key challenges which limit their further development: (1) The patch size selected by ViT is crucial for accurate predictions, which raises a natural question: How to select the size of patches properly or how to comprehensively combine small patches and larger patches; (2) While the spatial structure information is important in vision tasks, the 1D position embeddings fails to capture the spatial structure information of patches more accurately; (3) The GCN can capture the local connectivity relationships between image nodes, but it lacks the ability to capture global graph structural information. On the contrary, the self-attention mechanism of ViT can draw the global relation on image patches, but it is unable to model the local structure of image. To overcome such limitations, we propose the Hierarchical Vision Transformer Enhanced by Graph Convolutional Network (GCN-HViT) for image classification. Specifically, the Hierarchical ViT we designed can model patch-wise information interactions on a global scale within each level and model hierarchical relationships between small patches and large patches across multiple levels. In addition, the proposed GCN method functions as a local feature extractor to obtain the local representation of each image patch which serves as a 2D position embedding of each patch in the 2D space. Meanwhile, it models patch-wise information interactions on a local scale within each level. Extensive experiments on 3 real-world datasets demonstrate that GCN-HViT achieves state-of-the-art performance.

Reach and Frequency (R&F) contract advertising is an important form of widely used brand advertising. Unlike performance advertising, R&F contracts emphasize controllable delivery of UV and PV under given targeting, scheduling, and frequency control constraints. In practical systems, advertisers typically need to view the UV, PV change curves at different budget levels in real time when creating an R&F contract. However, most existing publicly available advertising datasets are based on independent samples, lacking a characterization of the core structure of the "budget-performance curve" (including UV and PV) in R&F contracts.This paper proposes and releases a large-scale R&F contract inventory estimation dataset. This dataset uses the R&F contract context consisting of "targeting-scheduling-frequency control" as the basic context, providing observations of UV and PV corresponding to multiple budget points within the same context, thus forming a complete budget-performance curve. The dataset explicitly includes a time-window-based frequency control mechanism (e.g.,"no more than 3 times within 5 days") and naturally satisfies the monotonicity and diminishing marginal returns characteristics in the budget and scheduling dimensions. We further derive the theoretical maximum exposure ceiling and use it as a consistency check to evaluate data quality and the feasibility of model predictions. Using this data set, this paper defines two standardized benchmark tasks: single-point performance prediction and reconstruction of budget-performance curves, and provides a set of reproducible baseline methods and evaluation protocols. This dataset can support systematic research on problems such as structural constraint learning, monotonic regression, curve consistency modeling, and R&F contract planning.The code for our experiments can be found at https://github.com/pengyunshan/RF-Inventory.

Imbalanced data is commonly present in real-world applications. While data synthesis can effectively mitigate the data scarcity problem of rare-classes, and LLMs have revolutionized text generation, the application of LLMs to relational/structured tabular data synthesis remains underexplored. Moreover, existing approaches lack an effective feedback mechanism that can guide LLMs towards continuously optimizing the quality of the generated data throughout the synthesis process. In this work, we propose RDDG, Relational Data generator with Dynamic Guidance, which is a unified in-context learning framework that employs progressive chain-of-thought (CoT) steps to generate tabular data for enhancing downstream imbalanced classification performance. RDDG first uses core set selection to identify representative samples from the original data, then utilizes in-context learning to discover the inherent patterns and correlations among attributes within the core set, and subsequently generates tabular data while preserving the aforementioned constraints. More importantly, it incorporates a self-reinforcing feedback mechanism that provides automatic assessments on the quality of the generated data, enabling continuous quality optimization throughout the generation process. Experimental results on multiple real and synthetic datasets demonstrate that RDDG outperforms existing approaches in both data fidelity and downstream imbalanced classification performance. We make our code available at https://github.com/cszhangLMU/RDDG.

Quantum error mitigation (QEM) provides a practical route for estimating reliable observables on noisy intermediate-scale quantum (NISQ) devices. Traditional QEM strategies, including zero-noise extrapolation (ZNE) and Clifford data regression (CDR), rely on noise scaling or global regression, and their performance is constrained by the exponential growth of the system degrees of freedom. We construct a graph-enhanced mitigation (GEM) framework, which incorporates physical information into the model representation. In this work, quantum circuits are encoded as attributed graphs. Hardware-level physical information is mapped to node and edge features: local noise parameters such as calibration parameters $T_1$, $T_2$, and readout errors are encoded at nodes, while coupling-related information such as two-qubit gate errors is encoded as edge features. Graph neural networks are used to model how errors propagate along the physical coupling structure and build up into non-local correlations. This allows the model to capture local interactions and part of the resulting non-local correlations across qubits. A dual-branch affine correction is applied to maintain consistency with physical constraints. Experiments on 10-qubit and 16-qubit random circuits executed on superconducting quantum processors show that GEM provides a level of accuracy comparable to CDR at small scales, while yielding lower mean absolute error and improved stability in zero-shot transfer to larger systems. Results of the traditional QEM strategy indicate that global regression methods remain effective in low-dimensional settings but become less reliable as system degrees of freedom grow. In contrast, GEM makes use of local physical structures to show better scalability and generalization, while preserving the overall error propagation patterns. This work provides a practical scalable approach to QEM for NISQ devices.

Agentic AI systems are rapidly advancing toward real-world applications, yet their readiness in complex and personalized environments remains insufficiently characterized. To address this gap, we introduce PersonalHomeBench, a benchmark for evaluating foundation models as agentic assistants in personalized smart home environments. The benchmark is constructed through an iterative process that progressively builds rich household states, which are then used to generate personalized, context-dependent tasks. To support realistic agent-environment interaction, we provide PersonalHomeTools, a comprehensive toolbox enabling household information retrieval, appliance control, and situational understanding. PersonalHomeBench evaluates both reactive and proactive agentic abilities under unimodal and multimodal observations. Thorough experimentation reveals a systematic performance reduction as task complexity increases, with pronounced failures in counterfactual reasoning and under partial observability, where effective tool-based information gathering is required. These results position PersonalHomeBench as a rigorous evaluation platform for analyzing the robustness and limitations of personalized agentic reasoning and planning.

When model developers or users fine-tune an LLM, this can induce behaviors that are unexpected, deliberately harmful, or hard to detect. It would be far easier to audit LLMs if they could simply describe their behaviors in natural language. Here, we study a scalable approach to rapidly identify learned behaviors of many LLMs derived from a shared base LLM. Given a model $M$, our method works by finetuning models $M_i$ from $M$ with implanted behaviors $b_i$; the $(M_i, b_i)$ pairs serve as labeled training data. We then train an \emph{introspection adapter} (IA): a single LoRA adapter jointly trained across the finetunes $M_i$ to cause them to verbalize their implanted behaviors. We find that this IA induces self-description of learned behaviors even in finetunes of $M$ that were trained in very different ways from the $M_i$. For example, IAs generalize to AuditBench, achieving state-of-the-art at identifying explicitly hidden concerning behaviors. IAs can also be used to detect encrypted finetuning API attacks. They scale favorably with model size and training data diversity. Overall, our results suggest that IAs are a scalable, effective, and practically useful approach to auditing fine-tuned LLMs.

Distributed tracing in microservices is critical for diagnostics but generates overwhelming data volumes, necessitating intelligent sampling. To maximize fidelity, state-of-the-art (SOTA) tail-based samplers analyze complete (or even log-enriched) traces by modeling them as graphs. However, this reliance on computationally expensive graph analysis creates a performance bottleneck that prohibits their use in online settings. To this end, we propose Gleaner, an online tail-sampling framework that breaks this trade-off. It is founded on the key insight that explicit graph structures are unnecessary for high-fidelity trace grouping. Instead, Gleaner represents each trace as a "bag-of-edges" augmented with log semantics, replacing slow graph algorithms with highly efficient set-based operations. It also employs an alarm-driven quota and a diversity-preserving strategy to prioritize anomalous and rare traces for downstream Root Cause Analysis (RCA). Experimentally, Gleaner processes traces at 0.74ms each, improving Trace Pattern Coverage by up to 128.7% and Shannon Entropy by up to 32.9% over baselines. At just a 1% sampling rate, Gleaner improves RCA accuracy by 42%-107% over the next-best sampler. Moreover, RCA on Gleaner's sampled data is more accurate than with the entire, unsampled dataset. This result reframes intelligent sampling from a data reduction technique to a powerful signal enhancement paradigm for automated operations.

Delayed loss spikes have been reported in neural-network training, but existing theory mainly explains earlier non-monotone behavior caused by overly large fixed learning rates. We study one stylized hypothesis: normalization can postpone instability by gradually increasing the effective learning rate during otherwise stable descent. To test this hypothesis at theorem level, we analyze batch-normalized linear models. Our flagship result concerns whitened square-loss linear regression, where we derive explicit no-rising-edge and delayed-onset conditions, bound the waiting time to directional onset, and show that the rising edge self-stabilizes within finitely many iterations. Combined with a square-loss decomposition, this yields a concrete delayed-spike mechanism in the whitened regime. For logistic regression, under highly restrictive active-margin assumptions, we prove only a supporting finite-horizon directional precursor in a knife-edge regime, with an optional appendix-only loss lower bound under an extra non-degeneracy condition. The paper should therefore be read as a stylized mechanism study rather than a general explanation of neural-network loss spikes. Within that scope, the results isolate one concrete delayed-instability pathway induced by batch normalization.

Optimization problems are central to decision-making in manufacturing, logistics, scheduling, and other industrial settings. Translating complicated descriptions of these problems into solver-ready formulations requires specialized operations research (OR) expertise, making it hard to scale. We present AutoOR, a scalable synthetic data generation and reinforcement learning pipeline that trains LLMs to autoformalize optimization problems specified in natural language across linear, mixed-integer, and non-linear categories. AutoOR generates verified training data from standard optimization forms and uses solver execution feedback as the reward signal for RL post-training. AutoOR applied to an 8B model achieves state-of-the-art or competitive results across six established OR benchmarks, matching significantly larger frontier models. For a non-linear problem class involving physical dynamics, where frontier models score near 0%, we introduce a curriculum RL strategy that bootstraps from limited initial training data to make this class tractable for post-training. We believe that methods such as AutoOR can significantly accelerate industrial decision-making with AI.

While modern representation learning relies heavily on global error signals, decentralized algorithms driven by local interactions offer a fundamental distributed alternative. However, the macroscopic convergence properties of these discrete dynamics on continuous data manifolds remain theoretically unresolved, notoriously suffering from parameter explosion. We bridge this gap by formalizing decentralized learning as a coupled slow-fast dynamical system on Riemannian manifolds. First, using measure-theoretic limits, we prove that the discrete spatial transitions converge uniformly to an overdamped Langevin stochastic differential equation. Second, via the Itô-Poisson resolvent and a stochastic extension of LaSalle's Invariance Principle, we establish that the representation weights unconditionally avoid divergence and align strictly with the principal eigenspace of the spatial measure. Finally, we construct a joint Lyapunov functional for the fully coupled spatial-parametric flow. This proves global dissipativity and demonstrates that orthogonally disentangled, linearly separable features emerge spontaneously at the stationary limit. Our framework bridges discrete algorithms with continuous stochastic analysis, providing a formal theoretical baseline for decentralized representation learning.

Large Language Models are increasingly used as judges to evaluate code artifacts when exhaustive human review or executable test coverage is unavailable. LLM-judge is increasingly relevant in agentic software engineering workflows, where it can help rank candidate solutions and guide patch selection. While attractive for scale, current practice lacks a principled account of reliability and bias: repeated evaluations of the same case can disagree; small prompt edits can swing outcomes; and seemingly semantics-preserving, human-equivalent perturbations may elicit divergent verdicts. This paper studies LLM-as-a-Judge for code through a measurement-first lens. We analyze two pointwise judging regimes across code generation, code repair task, and test generation, and we systematically probe prompt-induced biases. Our study considers difficulty levels for repeated runs and controlled prompt interventions that isolate one presentation cue at a time, and it evaluates judges using consistency and sensitivity to bias. We find that judge decisions are highly sensitive to prompt biases even when the underlying code snippet is unchanged. Across all three tasks, several biases systematically shift preferences toward the option favored by the prompt, improving accuracy when that option aligns with the gold answer but substantially reducing it otherwise. In some settings, these effects are large enough to change task-level conclusions and alter relative model rankings. These findings show that reported judge performance may reflect prompt artifacts rather than stable assessment ability, posing a direct threat to the validity and reproducibility of code evaluation. We therefore argue that LLM-as-a-Judge studies should report bias sensitivity alongside accuracy and incorporate explicit controls to support more trustworthy model comparison in software engineering.

We study how informal surface forms degrade NLI accuracy in ELECTRA-small (14M) and RoBERTa-large (355M) across four transforms applied to SNLI and MultiNLI: slang substitution, emoji replacement, Gen-Z filler tokens, and their combination. Slang substitution (replacing formal words with informal equivalents, e.g., "going to" -> "gonna", "friend" -> "homie") causes minimal degradation (at most 1.1pp): slang vocabulary falls largely within WordPiece coverage, so the tokenizer handles it without signal loss. Emoji replaces content words with Unicode characters that ELECTRA's WordPiece tokenizer maps to [UNK], destroying the input signal before any learned parameters see it (93.6% of emoji examples contain at least one [UNK], mean 2.91 per example). Noise tokens (no cap, deadass, tbh) are fully in-vocabulary but absent from NLI training data, consistent with the model assigning them inferential weight they do not carry. The two failure modes respond to different interventions: preprocessing recovers emoji accuracy by normalizing text before tokenization; augmentation handles noise by exposing the model to noise-bearing examples during training. A hybrid of both achieves 88.93% on the combined variant for ELECTRA on SNLI (up from 75.88%), with no statistically significant drop on clean text. Against GPT-4o-mini zero-shot, unmitigated ELECTRA is significantly worse on transformed variants (p < 0.0001); hybrid ELECTRA surpasses it across all SNLI variants and reaches statistical parity on MultiNLI.

Recent advances in Multimodal Large Language Models (MLLMs) have enabled open-ended object recognition, yet they struggle with fine-grained tasks. In contrast, CLIP-style models excel at fine-grained recognition but lack broad coverage of general object categories. To bridge this gap, we propose \textbf{HyMOR}, a \textbf{Hy}brid \textbf{M}ulti-granularity open-ended \textbf{O}bject \textbf{R}ecognition framework that integrates an MLLM with a CLIP model. In HyMOR, the MLLM performs open-ended and coarse-grained object recognition, while the CLIP model specializes in fine-grained identification of domain-specific objects such as animals and plants. This hybrid design enables accurate object understanding across multiple semantic granularities, serving as a robust perceptual foundation for downstream multi-modal content generation and interactive gameplay. To support evaluation in content-rich and educational scenarios, we introduce TBO (TextBook Objects), a dataset containing 20,942 images annotated with 8,816 object categories extracted from textbooks. Extensive experiments demonstrate that HyMOR narrows the fine-grained recognition gap with CLIP to 0.2\% while improving general object recognition by 2.5\% over a baseline MLLM, measured by average Sentence-BERT (SBert) similarity. Overall, HyMOR achieves a 23.2\% improvement in average SBert across all evaluated datasets, highlighting its effectiveness in enabling accurate perception for multi-modal game content generation and interactive learning applications.

This paper presents FairNVT, a lightweight debiasing framework for pretrained transformer-based encoders that improves both representation and prediction level fairness while preserving task accuracy. Unlike many existing debiasing approaches that address these notions separately, we argue they are inherently connected: suppressing sensitive information at the representation level can facilitate fairer predictions. Our approach learns task-relevant and sensitive embeddings via lightweight adapters, applies calibrated Gaussian noise to the sensitive embedding, and fuses it with the task representation. Together with orthogonality constraints and fairness regularization, these components jointly reduce sensitive-attribute leakage in the learned embeddings and encourage fairer downstream predictions. The framework is compatible with a wide range of pretrained transformer encoders. Across three datasets spanning vision and language, FairNVT reduces sensitive-attribute attacker accuracy, improves demographic-parity and equalized-odds metrics, and maintains high task performance.

Quantum feature maps offer expressive embeddings for classical learning tasks, and augmenting sparse identification of nonlinear dynamics (SINDy) with such features is a natural but unexplored direction. We introduce Q-SINDy, a quantum-kernel-augmented SINDy framework, and identify a specific failure mode that arises: coefficient cannibalization, in which quantum features absorb coefficient mass that rightfully belongs to the polynomial basis, corrupting equation recovery. We derive the exact cannibalization-bias formula Delta xi_P = (P^T P)^{-1} P^T Q xi_Q and prove that orthogonalizing quantum features against the polynomial column space at fit time eliminates this bias exactly. The claim is verified numerically to machine precision (<10^-12) on multiple systems. Empirically, across six canonical dynamical systems (Duffing, Van der Pol, Lorenz, Lotka-Volterra, cubic oscillator, Rossler) and three quantum feature map architectures (ZZ-angle encoding, IQP, data re-uploading), orthogonalized Q-SINDy consistently matches vanilla SINDy's structural recovery while uncorrected augmentation degrades true-positive rates by up to 100%. A refined dynamics-aware diagnostic, R^2_Q for X-dot, predicts cannibalization severity with statistical significance (Pearson r=0.70, p=0.023). An RBF classical-kernel control across 20 hyperparameter configurations fails more severely than any quantum variant, ruling out feature count as the cause. Orthogonalization remains robust under depolarizing hardware noise up to 2% per gate, and the framework extends without modification to Burgers' equation.

LLM-powered agents often reason from scratch when presented with a new problem instance and lack automatic mechanisms to transfer learned skills to other agents. We propose a federated learning-like framework, Federation over Text (FoT), that enables multiple agents solving different tasks to collectively generate a shared library of metacognitive insights by iteratively federating their local reasoning processes. Instead of federation over gradients (e.g., as in distributed training), FoT operates at the semantic level without any gradient optimization or supervision signal. Iteratively, each agent does local thinking and self-improvement on their specific tasks independently, and shares reasoning traces with a central server, which aggregates and distills them into a cross-task (and cross-domain) insight library that existing and future agents can leverage to improve performance on related tasks. Experiments show that FoT improves reasoning effectiveness and efficiency across a wide range of challenging applications, including mathematical problem solving, cross-domain collaboration, and machine learning research insight discovery. Specifically, it improves average accuracies of downstream tasks by 24% while reducing the reasoning tokens by 28% across the first two applications. In the research insight discovery application, FoT is able to generate insights that cover over 90% of the major contributions in the subsequent papers.

Modeling single-cell gene expression across diverse biological and technical conditions is crucial for characterizing cellular states and simulating unseen scenarios. Existing methods often treat genes as independent tokens, overlooking their high-level biological relationships and leading to poor performance. We introduce SAVE, a unified generative framework based on conditional Transformers for multi-condition single-cell modeling. SAVE leverages a coarse-grained representation by grouping semantically related genes into blocks, capturing higher-order dependencies among gene modules. A Flow Matching mechanism and condition-masking strategy further enhance flexible simulation and enable generalization to unseen condition combinations. We evaluate SAVE on a range of benchmarks, including conditional generation, batch effect correction, and perturbation prediction. SAVE consistently outperforms state-of-the-art methods in generation fidelity and extrapolative generalization, especially in low-resource or combinatorially held-out settings. Overall, SAVE offers a scalable and generalizable solution for modeling complex single-cell data, with broad utility in virtual cell synthesis and biological interpretation. Our code is publicly available at https://github.com/fdu-wangfeilab/sc-save

Every prediction from a generative medical event model is bounded by how clinical events are tokenized, yet input representation is rarely isolated from other system and architectural choices. We evaluate how representation decisions affect downstream prediction after a shared one-epoch pretraining budget. We train 28 matched transformers on MIMIC-IV and evaluate them on 30 clinical outcomes in three experiments: (1) quantization granularity, reference-range anchoring, and code-value fusion; (2) value encoding (hard bins, soft discretization, code-normalized xVal) crossed with temporal encoding (event order, time tokens, admission-relative RoPE); and (3) native MIMIC laboratory/vital codes versus the Common Longitudinal ICU Format (CLIF)-remapped laboratory/vital codes with compression-preserving perturbation arms. In Experiment 1, fused code-value tokenization improves mortality AUROC from 0.891 to 0.915 (BH-adjusted p < 0.001), hospital length-of-stay AUROC from 0.763 to 0.788 (BH-adjusted p < 0.001), and, for the decile fused-vs-unfused comparison, mean regression Spearman rho across the 13 regression outcomes from 0.414 to 0.494. Across the three temporal encodings, event order only and admission-relative RoPE match or exceed inserting time tokens on average while shortening sequences by 11%. CLIF remapping preserves downstream performance in our single-site setting while yielding a smaller, clinically interpretable token set compatible with multi-site use. Finer-than-decile quantization, reference-range anchoring, and soft discretization help in selective outcomes, while code-normalized xVal remains well below the discrete and soft families, consistent with near-median suppression that persists after the affine variant.

Long-horizon language model systems increasingly rely on persistent memory, yet many current designs still treat memory primarily as a static store: write an item, place it into memory, and retrieve it later if needed. We argue that this framing does not adequately capture the practical memory-control problem in deployed LLM systems. In realistic settings, the difficulty is often not merely forgetting useful information, but retaining too many uncertain items, forgetting important content in the wrong order, and giving users little trust in what will persist over time. We propose StageMem, a lifecycle-managed memory framework that treats memory as a stateful process rather than a passive repository. StageMem organizes memory into three stages -- transient, working, and durable memory -- and models each item with explicit confidence and strength. This separates shallow admission from long-term commitment: information may first be written at low cost and only later be promoted, retained, updated, or evicted as evidence and pressure evolve. Under controlled pressure regimes, this decomposition helps preserve late-important content while keeping memory burden and deeper-tier pollution more controlled. Adapted external tasks provide boundary evidence that the same schema remains compatible with stronger retrieval structure outside pure synthetic control. We present StageMem as a principled decomposition of the memory-control problem for language model systems.

Student engagement with large language models (LLMs) in academic writing is not a stable trait, an adoption decision, or a competency level; it is a continuously negotiated process that existing frameworks cannot adequately theorize. Typological models provide categories without mechanisms; technology acceptance models explain adoption but not post-adoption quality; AI literacy frameworks treat competency as a static predictor rather than a live input. None accounts for within-student variability across tasks, the developmental paradox whereby experience produces habituation rather than sophistication, or principled non-use as a form of ethical reasoning. This article introduces the Reliance Negotiation Framework (RNF), developed from a sequential explanatory mixed-methods study of 382 undergraduates at a public minority-serving institution in the United States (survey, N = 382; 14 semi-structured interviews; three qualitative survey strands; 1,435 coded instances). The RNF reconceptualizes LLM reliance as an ongoing negotiation among four concurrent inputs (perceived benefits, perceived risks, ethical commitments, and situational demands) with outputs that recursively modify subsequent decisions. A Two-Model Architecture accommodates the 13.0% of participants whose categorical ethical commitments foreclose negotiation entirely. The framework generates four falsifiable predictions with implications for AI literacy pedagogy, academic integrity policy, and equity-centered practice at minority-serving institutions.

Privacy, security, and accessibility, like ethical concerns in mobile applications (a.k.a. apps), commonly subsumed under non-functional requirements, are generally reported by users through app reviews available in app stores. However, these remain unidentified among other types of reviews, such as user experiences, problem reports, and new feature discussions. Over the past decade, extensive research has focused on extracting valuable information from app reviews, including feature requests and bug reports. However, there remains a lack of a synthesis of research related to app review analysis for exploring users' ethical concerns. This paper presents a comprehensive survey of this research area, covering 37 relevant studies published since 2012, identified from the initial 553 studies using specific inclusion and exclusion criteria. The studies examined vary in review counts, ranging from 500 to 626 million, and include between a single and 1.3 million apps. Our detailed analysis highlights diverse objectives, methodologies, and strategies, along with additional resources such as app privacy policies, which researchers generally utilize to analyze ethical concerns. Our findings also identify persistent barriers to privacy, security, accessibility, transparency, fairness, accountability, and safety, as reported by users in app reviews. Furthermore, we propose a research agenda that focuses on four key areas, including automated extraction and classification of ethical concerns-related app reviews. Our survey outcomes can assist developers and system architects in recognizing and prioritizing non-functional requirements at the initial stages of the development lifecycle, whereas researchers can expand upon this synthesis to create tools for the automated detection of ethical concerns.

Audio platforms have evolved beyond entertainment. They have become central to public discourse, from podcasts and radio to WhatsApp voice notes and live streams. With millions of shows and hundreds of millions of listeners, audio platforms are now a major channel for misinformation. Yet existing fact-checking pipelines are mostly designed for written claims, overlooking the unique properties of spoken media. We argue that audio misinformation is not merely textual content with transcripts: it is structurally different because it is both spoken - carrying persuasive force through prosody, pacing, and emotion - and conversational - unfolding across turns, speakers, and episodes. These dual properties introduce verification difficulties that traditional methods rarely face. This position paper synthesizes evidence across modalities and platforms, examines datasets and methods, and highlights why existing pipelines fail on audio. We argue that advancing fact-checking requires rethinking verification pipelines around the spoken and conversational realities of audio.

Online political hostility is pervasive, yet it remains unclear how toxicity varies across campaign issues and political ideology, and what psychosocial signals and framing accompany toxic expression online. In this work, we present a large-scale analysis of discourse on X (Twitter) during the five weeks surrounding the 2024 U.S. presidential election. We categorize posts into 10 major campaign issues, estimate the ideology of posts using a human-in-the-loop LLM-assisted annotation process, detect harmful content with an LLM-based toxicity detection model, and then examine the psychological drivers of toxic content. We use these annotated data to examine how harmful content varies across campaign issues and ideologies, as well as how emotional tone and moral framing shape toxicity in election discussions. Our results show issue heterogeneity in both the prevalence and intensity of toxicity. Identity-related issues displayed the highest toxicity intensity. As for specific harm categories, harassment was most prevalent and intense across most of the issues, while hate concentrated in identity-centered debates. Partisan posts contained more harmful content than neutral posts, and ideological asymmetries in toxicity varied by issue. In terms of psycholinguistic dimensions, we found that toxic discourse is dominated by high-arousal negative emotions. Left- and right-leaning posts often exhibit similar emotional profiles within the same issue domain, suggesting emotional mirroring. Partisan groups frequently rely on overlapping moral foundations, while issue context strongly shapes which moral foundations become most salient. These findings provide a fine-grained account of toxic political discourse on social media and highlight that online political toxicity is highly context-dependent, underscoring the need for issue-sensitive approaches to measuring and mitigating it.

Causal inference from electronic health records (EHR) is fundamentally limited by unmeasured confounding: critical clinical states such as frailty, goals of care, and mental status are documented in free-text notes but absent from structured data. Large language models can extract these latent confounders as interpretable, structured covariates, yet how to effectively integrate them into causal estimation pipelines has not been systematically studied. Using the MIMIC-IV database with 21,859 sepsis patients, we compare seven covariate-integration strategies for estimating the effect of early vasopressor initiation on 28-day mortality, spanning tabular-only baselines, traditional NLP representations, and three LLM-augmented approaches. A central finding is that not all integration strategies are equally effective: directly augmenting the propensity score model with LLM covariates achieves the best performance, while dual-caliper matching on text-derived categorical distances restricts the donor pool and degrades estimation. In semi-synthetic experiments with known ground-truth effects, LLM-augmented propensity scores reduce estimation bias from 0.0143 to 0.0003 relative to tabular-only methods, and this advantage persists under substantial simulated extraction error. On real data, incorporating LLM-extracted covariates reduces the estimated treatment effect from 0.055 to 0.027, directionally consistent with the CLOVERS randomized trial, and a doubly robust estimator yielding 0.019 confirms the robustness of this finding. Our results offer practical guidance on when and how text-derived covariates improve causal estimation in critical care.

Modern AI agents routinely depend on secrets such as API keys and SSH credentials, yet the dominant deployment model still exposes those secrets directly to the agent process through environment variables, local files, or forwarding sockets. This design fails against prompt injection, tool misuse, and model-controlled exfiltration because the agent can both use and reveal the same bearer credential. We present CapSeal, a capability-sealed secret mediation architecture that replaces direct secret access with constrained invocations through a local trusted broker. CapSeal combines capability issuance, schema-constrained HTTP execution, broker-executed SSH actions, anti-replay session binding, policy evaluation, and tamper-evident audit trails. We describe a Rust prototype integrated with an MCP-facing adapter, formulate conditional security goals for non-disclosure, constrained use, replay resistance, and auditability, and define an evaluation plan spanning prompt injection, tool misuse, and SSH abuse. The resulting system reframes secret handling for agentic systems from handing the model a key to granting the model a narrowly scoped, non-exportable action capability.

We present a system for automated detection, localization, and scoring of arrow punctures on 40\,cm indoor archery target faces, trained on only 48 annotated photographs (5{,}084 punctures). Our pipeline combines three components: a color-based canonical rectification stage that maps perspective-distorted photographs into a standardized coordinate system where pixel distances correspond to known physical measurements; a frozen self-supervised vision transformer (DINOv3 ViT-L/16) paired with AnyUp guided feature upsampling to recover sub-millimeter spatial precision from $32 \times 32$ patch tokens; and lightweight CenterNet-style detection heads for arrow-center heatmap prediction. Only 3.8\,M of 308\,M total parameters are trainable. Across three cross-validation folds, we achieve a mean F1 score of $0.893 \pm 0.011$ and a mean localization error of $1.41 \pm 0.06$\,mm, comparable to or better than prior fully-supervised approaches that require substantially more training data. An ablation study shows that the CenterNet offset regression head, typically essential for sub-pixel refinement, provides negligible detection improvement while degrading localization in our setting. This suggests that guided feature upsampling already resolves the spatial precision lost through patch tokenization. On downstream archery metrics, the system recovers per-image average arrow scores with a median error of 1.8\% and group centroid positions to within a median of 4.00\,mm. These results demonstrate that frozen foundation models with minimal task-specific adaptation offer a practical paradigm for dense prediction in small-data regimes.

The expression of emotions that serve social purposes, such as asserting independence or fostering interdependence, is central to human interactions and varies systematically across cultures. As LLMs are increasingly used to simulate human behavior in culturally nuanced interactions, it is important to understand whether they faithfully capture human patterns of social emotion expression. When LLM responses are not culturally aligned, their utility is compromised -- particularly when users assume they are interacting with a culturally attuned interlocutor, and may act on advice that proves inappropriate in their cultural context. We present a psychologically informed evaluation framework of cross-cultural social emotion expression in LLMs. Using a human study comparing European American and Latin American participants' expression of engaging and disengaging emotions, we evaluate six frontier LLMs on their ability to reflect culturally differentiated patterns for expressing social emotions. We find systematic misalignment between model and human behavior: all models express engaging emotions more than disengaging ones, with particularly stark differences observed for the generally well-represented European American persona. We further highlight that LLM responses are highly concentrated and deterministic, failing to capture the diversity of human responses in expressing social emotions. Our ablation analyses reveal that these patterns are robust to sampling temperatures, partially sensitive to prompt language, and dependent on the response elicitation format. Together, our findings highlight limitations in how current LLMs represent the interaction of cultural and emotional axes, particularly when expressing social emotions, with direct implications for their deployment in cross-cultural affective contexts.

Prompt-induced cognitive biases are changes in a general-purpose AI (GPAI) system's decisions caused solely by biased wording in the input (e.g., framing, anchors), not task logic. In software engineering (SE) decision support (where problem statements and requirements are natural language) small phrasing shifts (e.g., popularity hints or outcome reveals) can push GPAI models toward suboptimal decisions. We study this with PROBE-SWE, a dynamic benchmark for SE that pairs biased and unbiased versions of the same SE dilemmas, controls for logic and difficulty, and targets eight SE-relevant biases (anchoring, availability, bandwagon, confirmation, framing, hindsight, hyperbolic discounting, overconfidence). We ask whether prompt engineering mitigates bias sensitivity in practice, focusing on actionable techniques that practitioners can apply off-the-shelf in real environments. Testing common strategies (e.g., chain-of-thought, self-debiasing) on cost-effective GPAI systems, we find no statistically significant reductions in bias sensitivity on a per-bias basis. We then adopt a Prolog-style view of the reasoning process: solving SE dilemmas requires making explicit any background axioms and inference assumptions (i.e., SE best practices) that are usually implicit in the prompt. So, we hypothesize that bias-inducing features short-circuit assumptions elicitation, pushing GPAI models toward biased shortcuts. Building on this, we introduce an end-to-end method that elicits best practices and injects axiomatic reasoning cues into the prompt before answering, reducing overall bias sensitivity by 51% on average (p < .001). Finally, we report a thematic analysis that surfaces linguistic patterns associated with heightened bias sensitivity, clarifying when GPAI use is less advisable for SE decision support and where to focus future countermeasures.

As large language models (LLMs) are increasingly integrated into daily life, in roles ranging from high-stakes decision support to companionship, understanding their behavioral dispositions becomes critical. A growing literature uses psychometric inventories and cognitive paradigms to profile LLM dispositions. However, these approaches cannot determine whether behavioral differences reflect stable, stimulus-specific individuality or global response biases and stochastic noise. Here, we apply crossed random-effects models -- widely used in psychometrics to separate systematic effects -- to 74.9 million ratings provided by 10 open-weight LLMs for over 100,000 words across 14 psycholinguistic norms. On average, 16.9% of variance is attributable to stimulus-specific individuality, robustly exceeding a statistical null model. Cross-norm prediction analyses reveal this individuality as a coherent fingerprint, unique to each model. These results identify individual differences among LLMs that cannot be attributed to response biases or stochastic noise. We term these differences machine individuality.

In this article, we argue that AI slop in software is creating a tragedy of the commons. Individual productivity gains from AI-generated content externalize costs onto reviewer capacity, codebase integrity, public knowledge resources, collaborative trust, and the talent pipeline. AI slop is cheap to generate and expensive to review, and the review layer is already thin. Commons problems are not solved by individual restraint. We outline concrete next steps for tool developers, team leads, and educators, grounded in Ostrom's design principles for enduring commons institutions.

As large language models (LLMs) transition into autonomous agents integrated with extensive tool ecosystems, traditional routing heuristics increasingly succumb to context pollution and "overthinking". We argue that the bottleneck is not a deficit in algorithmic capability or skill diversity, but the absence of disciplined second-order metacognitive governance. In this paper, our scientific contribution focuses on the computational translation of human cognitive control - specifically, delayed appraisal, epistemic vigilance, and region-of-proximal offloading - into a single-agent architecture. We introduce MESA-S (Metacognitive Skills for Agents, Single-agent), a preliminary framework that shifts scalar confidence estimation into a vector separating self-confidence (parametric certainty) from source-confidence (trust in retrieved external procedures). By formalizing a delayed procedural probe mechanism and introducing Metacognitive Skill Cards, MESA-S decouples the awareness of a skill's utility from its token-intensive execution. Evaluated under an In-Context Static Benchmark Evaluation natively executed via Gemini 3.1 Pro, our early results suggest that explicitly programming trust provenance and delayed escalation mitigates supply-chain vulnerabilities, prunes unnecessary reasoning loops, and prevents offloading-induced confidence inflation. This architecture offers a scientifically cautious, behaviorally anchored step toward reliable, epistemically vigilant single-agent orchestration.

Current agent evaluations largely reward execution on fully specified tasks, while recent work studies clarification [11, 22, 2], capability awareness [9, 1], abstention [8, 14], and search termination [20, 5] mostly in isolation. This leaves open whether agents can diagnose why a task is blocked before acting. We introduce the Support-State Triage Audit (SSTA-32), a matched-item diagnostic framework in which minimal counterfactual edits flip the same base request across four support states: Complete (ANSWER), Clarifiable (CLARIFY), Support-Blocked (REQUEST SUPPORT), and Unsupported-Now (ABSTAIN). We evaluate a frontier model under four prompting conditions - Direct, Action-Only, Confidence-Only, and a typed Preflight Support Check (PSC) - using Dual-Persona Auto-Auditing (DPAA) with deterministic heuristic scoring. Default execution overcommits heavily on non-complete tasks (41.7% overcommitment rate). Scalar confidence mapping avoids overcommitment but collapses the three-way deferral space (58.3% typed deferral accuracy). Conversely, both Action-Only and PSC achieve 91.7% typed deferral accuracy by surfacing the categorical ontology in the prompt. Targeted ablations confirm that removing the support-sufficiency dimension selectively degrades REQUEST SUPPORT accuracy, while removing the evidence-sufficiency dimension triggers systematic overcommitment on unsupported items. Because DPAA operates within a single context window, these results represent upper-bound capability estimates; nonetheless, the structural findings indicate that frontier models possess strong latent triage capabilities that require explicit categorical decision paths to activate safely.

Audio deepfakes pose a significant security threat, yet current state-of-the-art (SOTA) detection systems do not generalize well to realistic in-the-wild deepfakes. We introduce a novel \textbf{I}n-\textbf{C}ontext \textbf{L}earning paradigm with comparison-guidance for \textbf{A}udio \textbf{D}eepfake detection (\textbf{ICLAD}). The framework enables the use of audio language models (ALMs) for training-free generalization to unseen deepfakes and provides textual rationales on the detection outcome. At the core of ICLAD is a pairwise comparative reasoning strategy that guides the ALM to discover and filter hallucinations and deepfake-irrelevant acoustic attributes. The ALM works alongside a specialized deepfake detector, whereby a routing mechanism feeds out-of-distribution samples to the ALM. On in-the-wild datasets, ICLAD improves macro F1 over the specialized detector, with up to $2\times$ relative improvement. Further analysis demonstrates the flexibility of ICLAD and its potential for deployment on recent open-source ALMs.

Time series forecasting plays a pivotal role in critical sectors such as finance, energy, transportation, and meteorology. However, Long-term Time Series Forecasting (LTSF) remains a significant challenge because real-world signals contain highly entangled temporal dynamics that are difficult to fully capture from a purely 1D perspective. To break this representation bottleneck, we propose TriTS, a novel cross-modal disentanglement framework that projects 1D time series into orthogonal time, frequency, and 2D-vision spaces.To seamlessly bridge the 1D-to-2D modality gap without the prohibitive $O(N^2)$ computational overhead of Vision Transformers (ViTs), we introduce a Period-Aware Reshaping strategy and incorporate Visual Mamba (Vim). This approach efficiently models cross-period dependencies as global visual textures while maintaining linear computational complexity. Complementing this, we design a Multi-Resolution Wavelet Mixing (MR-WM) module for the frequency modality, which explicitly decouples non-stationary signals into trend and noise components to achieve fine-grained time-frequency localization. Finally, a streaming linear branch is retained in the time domain to anchor numerical stability. By dynamically fusing these three complementary representations, TriTS effectively adapts to diverse data contexts. Extensive experiments across multiple benchmark datasets demonstrate that TriTS achieves state-of-the-art (SOTA) performance, fundamentally outperforming existing vision-based forecasters by drastically reducing both parameter count and inference latency.

Training-free token reduction methods for Vision Transformers (ToMe, ToFu, PiToMe, and MCTF) employ different scoring mechanisms, yet they share a closely matched cliff-like collapse at high compression. This paper explains \emph{why}. We develop a diagnostic framework with two tools, ranking consistency $ρ_s$ and off-diagonal correlation $ρ_\text{off}$, that decomposes the collapse into (1)a signal-agnostic error amplifier inherent to layer-wise reduction, predicting convex Pareto curves and $r_{\text{crit}} \propto 1/L$; and (2)shared reliance on \emph{pairwise} similarity signals whose ranking consistency degrades from $ρ_s{=}0.88$ to $0.27$ in deep layers. Pairwise rankings are inherently unstable ($O(N_p^2)$ joint perturbations) while unary signals enjoy greater stability ($O(N_p)$ perturbations, CLT). From three design principles derived from this diagnosis, we construct CATIS as a constructive validation: unary signals raise the trigger threshold, triage suppresses the gain. On ViT-Large at 63% FLOPs reduction, CATIS retains 96.9% of vanilla accuracy (81.0%) on ImageNet-1K where all baselines collapse to 43--65%.

We present a framework for evaluating adaptive personalization of educational reading materials with theory-grounded simulated learners. The system builds a learning-objective and knowledge-component ontology from open textbooks, curates it in a browser-based Ontology Atlas, labels textbook chunks with ontology entities, and generates aligned reading-assessment pairs. Simulated readers learn from passages through a Construction-Integration-inspired memory model with DIME-style reader factors, KREC-style misconception revision, and an open New Dale-Chall readability signal. Answers are produced by score-based option selection over the learner's explicit memory state, while BKT drives adaptation. Across three sampled subject ontologies and matched cohorts of 50 simulated learners per condition, adaptive reading significantly improved outcomes in computer science, yielded smaller positive but inconclusive gains in inorganic chemistry, and was neutral to slightly negative in general biology.

Scientists have long sought to accurately predict outcomes of real-world events before they happen. Can AI systems do so more reliably? We study this question through clinical trial outcome prediction, a high-stakes open challenge even for domain experts. We introduce CT Open, an open-access, live platform that will run four challenge every year. Anyone can submit predictions for each challenge. CT Open evaluates those submissions on trials whose outcomes were not yet public at the time of submission but were made public afterwards. Determining if a trial's outcome is public on the internet before a certain date is surprisingly difficult. Outcomes posted on official registries may lag behind by years, while the first mention may appear in obscure articles. To address this, we propose a novel, fully automated decontamination pipeline that uses iterative LLM-powered web search to identify the earliest mention of trial outcomes. We validate the pipeline's quality and accuracy by human expert's annotations. Since CT Open's pipeline ensures that every evaluated trial had no publicly reported outcome when the prediction was made, it allows participants to use any methodology and any data source. In this paper, we release a training set and two time-stamped test benchmarks, Winter 2025 and Summer 2025. We believe CT Open can serve as a central hub for advancing AI research on forecasting real-world outcomes before they occur, while also informing biomedical research and improving clinical trial design. CT Open Platform is hosted at $\href{https://ct-open.net/}{https://ct-open.net/}$

LLM-powered coding agents suffer from a poorly understood failure mode we term output stalling: the agent silently produces empty responses when attempting to generate large, format-heavy documents. We present a theoretical framework that explains and prevents this failure through three contributions. (1) We introduce Output Generation Capacity (OGC), a formal measure of an agent's effective ability to produce output given its current context state - distinct from and empirically smaller than the raw context window. (2) We prove a Format-Cost Separation Theorem showing that deferred template rendering is always at least as token-efficient as direct generation for any format with overhead multiplier $μ_f > 1$, and derive tight bounds on the savings. (3) We formalize Adaptive Strategy Selection, a decision framework that maps the ratio of estimated output cost to available OGC into an optimal generation strategy (direct, chunked, or deferred). We validate the theory through controlled experiments across three models (Claude 3.5 Sonnet, GPT-4o, Llama 3.1 70B), four document types, and an ablation study isolating each component's contribution. Deferred rendering reduces LLM generation tokens by 48-72% across all conditions and eliminates output stalling entirely. We instantiate the framework as GEN-PILOT, an open-source MCP server, demonstrating that the theory translates directly into a practical tool.

Multimodal large language models (MLLMs) have recently demonstrated strong capabilities in understanding and generating responses from diverse visual inputs, including high-resolution images and long video sequences. As these models scale to richer visual representations, inference increasingly relies on storing large numbers of vision tokens in the key-value (KV) cache, making memory consumption a central bottleneck. Existing methods address this issue by identifying redundancy in vision tokens and compressing the cache, but such compression is typically applied only after all inputs are processed, resulting in high peak memory usage during the prefill stage. In this work, we show that MLLMs exhibit inherent structural regularities and representational redundancy that can be exploited to control memory growth throughout inference. Based on this insight, we propose a sequential input-compression mechanism that enforces a fixed memory budget by performing structure-aware key-value cache compression during the prefill process. This approach substantially reduces peak memory usage while maintaining generative performance with only minimal degradation, enabling more practical and memory-efficient multimodal inference.

State-of-the-art large language models (LLMs) show high performance in general visual question answering. However, a fundamental limitation remains: current architectures lack the native 3D spatial reasoning required for direct analysis of volumetric medical imaging, such as CT or MRI. Emerging agentic AI offers a new solution, eliminating the need for intrinsic 3D processing by enabling LLMs to orchestrate and leverage specialized external tools. Yet, the feasibility of such agentic frameworks in complex, multi-step radiological workflows remains underexplored. In this work, we present a training-free agentic pipeline for automated brain MRI analysis. Validating our methodology on several LLMs (GPT-5.1, Gemini 3 Pro, Claude Sonnet 4.5) with off-the-shelf domain-specific tools, our system autonomously executes complex end-to-end workflows, including preprocessing (skull stripping, registration), pathology segmentation (glioma, meningioma, metastases), and volumetric analysis. We evaluate our framework across increasingly complex radiological tasks, from single-scan segmentation and volumetric reporting to longitudinal response assessment requiring multi-timepoint comparisons. We analyze the impact of architectural design by comparing single-agent models against multi-agent "domain-expert" collaborations. Finally, to support rigorous evaluation of future agentic systems, we introduce and release a benchmark dataset of image-prompt-answer tuples derived from public BraTS data. Our results demonstrate that agentic AI can solve highly neuro-radiological image analysis tasks through tool use without the need for training or fine-tuning.

GPU-based concurrent data structures (CDSs) achieve high throughput for read-only queries, but efficient support for dynamic updates on fully GPU-resident data remains challenging. Ordered CDSs (e.g., B-trees and LSM-trees) maintain an index layer that directs operations to a data layer (buckets or leaves), while hash tables avoid the cost of maintaining order but do not support range or successor queries. On GPUs, maintaining and traversing an index layer under frequent updates introduces contention and warp divergence. To tackle these problems, we flip the indexing paradigm on its head with FliX, a comparison-based, flipped indexing strategy for dynamic, fully GPU-resident CDSs. Traditional GPU CDSs typically take a batch of operations and assign each operation to a GPU thread or warp. FliX, however, assigns compute (e.g., a warp) to each bucket in the data layer, and each bucket then locates operations it is responsible for in the batch. FliX can replace many index layer traversals with a single binary search on the batch, reducing redundant work and warp divergence. Further, FliX simplifies updates as no index layer must be maintained. In our experiments, FliX achieves 6.5x reduced query latency compared to a leading GPU B-tree and 1.5x compared to a leading GPU LSM-tree, while delivering 4x higher throughput per memory footprint than ordered competitors. Despite maintaining order, FliX also surpasses state-of-the-art unordered GPU hash tables in query and deletion performance, and is highly competitive in insertion performance. In update-heavy workloads, it outperforms the closest fully dynamic ordered baseline by over 8x in insertion throughput while supporting dynamic memory reclamation. These results suggest that eliminating the index layer and adopting a compute-to-bucket mapping can enable practical, fully dynamic GPU indexing without sacrificing query performance.

Large Language Models (LLMs) have demonstrated potential in automating scientific ideation, yet current approaches relying on iterative prompting or complex multi-agent architectures often suffer from hallucination or computational inefficiency. A critical bottleneck in applying Reinforcement Learning (RL) to this open-ended domain is reward hacking -- where models exploit imperfect evaluation proxies to maximize scores without producing genuine scientific innovation. To address these limitations, we propose an RL framework explicitly tailored for high-quality scientific idea generation. We propose the first multi-agent reward function designed to serve as a judge, decoupling methodological validation from implementation details while providing strict binary rewards that are robust to reward hacking. To effectively optimize against this sparse signal, we utilize an unbiased variant of Group Relative Policy Optimization to mitigate artificial length bias. We grounded our training in ICLR-320, a curated dataset of problem-solution pairs extracted from ICLR 2024 proceedings. Experiments demonstrate that our framework significantly outperforms state-of-the-art baselines across expert-evaluated metrics of novelty, feasibility, and effectiveness.

Predicting full-field physics through the real-time virtual sensing of engineering systems can enhance limited physical sensors but often requires sparse-to-dense reconstruction, complex multiphysics, and highly irregular geometries as well as strict latency and energy constraints for edge-deployability. Neural operators have been presented as a potential candidate for such applications but few architectures exist that explicitly address power consumption. Spiking neuron integration can provide a potential solution when integrated on neuromorphic hardware but the current existing neuron models result in severe performance degradation towards regression-based virtual sensing. To address the performance concerns and edge-constraints, we present the Variable Spiking Graph Neural Operator (VS-GNO) which integrates a sophisticated spectral-spatial convolutional analysis and a previously developed Variable Spiking Neuron (VSN) and energy-error balance loss function. With a non-spiking $L_2$ error baseline of $0.4\%$, VS-GNO can provide a reconstruction error of $0.71\%$ with $15\%$ average spiking in its spectral-only form and $1.04\%$ with $24.5\%$ spiking in its entire form. These results position VS-GNO as a promising step towards energy-efficient, edge-deployable neural operators for real-time sparse-to-dense virtual sensing in complex, highly irregular engineering environments.

Developing neural operators that accurately predict the behavior of systems governed by partial differential equations (PDEs) across unseen parameter regimes is crucial for robust generalization in scientific and engineering applications. In practical applications, variations in physical parameters induce distribution shifts between training and prediction regimes, making extrapolation a central challenge. As a result, the way parameters are incorporated into neural operator models plays a key role in their ability to generalize, particularly when state and parameter representations are entangled. In this work, we introduce the Late Fusion Neural Operator, an architecture that disentangles learning state dynamics from parameter effects, improving predictive performance both within and beyond the training distribution. Our approach combines neural operators for learning latent state representations with sparse regression to incorporate parameter information in a structured manner. Across four benchmark PDEs including advection, Burgers, and both 1D and 2D reaction-diffusion equations, the proposed method consistently outperforms Fourier Neural Operator and CAPE-FNO. Late Fusion Neural Operators achieve consistently the best performance in all experiments, with an average RMSE reduction of 72.9% in-domain and 71.8% out-domain compared to the second-best method. These results demonstrate strong generalization across both in-domain and out-domain parameter regimes.

Time-series forecasting is central to many scientific and industrial domains, such as energy systems, climate modeling, finance, and retail. While forecasting methods have evolved from classical statistical models to automated, and neural approaches, the surrounding software ecosystem remains anchored to the traditional Python numerical stack. Existing libraries rely on interpreter-driven execution and object-oriented abstractions, limiting composability, large-scale parallelism, and integration with modern differentiable and accelerator-oriented workflows. Meanwhile, today's forecasting increasingly involves large collections of heterogeneous time series data, irregular covariates, and frequent retraining, placing new demands on scalability and execution efficiency. JAX offers an alternative paradigm to traditional stateful numerical computation frameworks based on pure functions and program transformations such as just-in-time compilation and automatic vectorization, enabling end-to-end optimization across CPUs, GPUs, and TPUs. However, this modern paradigm has not yet been fully incorporated into the design of forecasting systems. We introduce Chronax, a JAX-native time-series forecasting library that rethinks forecasting abstractions around functional purity, composable transformations, and accelerator-ready execution. By representing preprocessing, modeling, and multi-horizon prediction as pure JAX functions, Chronax enables scalable multi-series forecasting, model-agnostic conformal uncertainty quantification, and seamless integration with modern machine learning and scientific computing pipelines.

We investigate the potential of the Quantum Approximate Optimization Algorithm (QAOA) for reducing energy consumption in route planning, a key challenge in logistics due to the NP-hard nature of the Traveling Salesman and Vehicle Routing Problems. By encoding route optimization as a Quadratic Unconstrained Binary Optimization (QUBO) problem and implementing QAOA circuits at depth p = 3-5 alongside classical baselines of Simulated Annealing (SA) and Genetic Algorithms (GA), we perform systematic benchmarks on Euclidean graphs of sizes N = 5, 10, and 20. Our results demonstrate that QAOA attains higher solution quality with approximation ratios of 0.953 (N = 5), 0.921 (N = 10), and 0.903 (N = 20), outperforming SA and GA by 2.7-4.4%. Wall-clock runtimes for QAOA are 2-3x faster than SA across all tested sizes, and energy consumption measurements reveal a three-order-of-magnitude reduction, remaining in the picojoule range versus nanojoules for classical methods. Translating these gains to real-world logistics suggests an 8.2% improvement in routing efficiency could save approximately 2.62 EJ of fuel annually in the U.S., avoiding nearly 1.94 x 10^8 tonnes of CO2 emissions. These findings highlight QAOA's promise as a fast, energy-efficient optimizer for sustainable logistics applications and underscore its potential role in next-generation fleet-management systems.

This paper addresses a critical safety gap in the use Automated Verbal Response Scoring (AVRS). We present a novel hybrid framework for troubled student detection that combines a text classifier, trained to detect responses based on their content, and an audio classifier, trained to detect responses using prosodic markers. This approach overcomes key limitations of traditional AVRS systems by considering both content and prosody of responses, achieving enhanced performance in identifying potentially concerning responses. This system can expedite the review process by humans, which can be life-saving particularly when timely intervention may be crucial.

Graph foundation models have demonstrated remarkable adaptability across diverse downstream tasks through large-scale pretraining on graphs. However, existing implementations of the backbone model, graph transformers, are typically limited to single-GPU systems, leading to long training times or out-of-memory issues on large graphs. Moreover, parallelizing graph transformer training over the full graph is challenging, as efficiency depends heavily on both the graph structure and system characteristics, such as bandwidth and memory capacity. In this work, we introduce a distributed training framework for graph transformers, which automatically selects and optimizes parallelization strategies based on the graph structure and hardware configuration. With our implementation of distributed sparse operations, we accelerate sparse graph attention by up to 3.8x and reduce memory consumption by 78% compared to state-of-the-art frameworks. On large graph benchmarks, our proposed framework achieves up to 6x speedup with system scaling up to 8 GPUs. These results demonstrate that the proposed framework improves the scalability of graph transformers, bringing them closer to serving as practical graph foundation models.

Classical mixture models (MMs) are widely used tractable proposals for approximate inference settings such as variational inference (VI) and importance sampling (IS). Recently, mixture models with negative coefficients, called subtractive mixture models (SMMs), have been proposed as a potentially more expressive alternative. However, how to effectively use SMMs for VI and IS is still an open question as they do not provide latent variable semantics and therefore cannot use sampling schemes for classical MMs. In this work, we study how to circumvent this issue by designing several expectation estimators for IS and learning schemes for VI with SMMs, and we empirically evaluate them for distribution approximation. Finally, we discuss the additional challenges in estimation stability and learning efficiency that they carry and propose ways to overcome them. Code is available at: https://github.com/april-tools/delta-vi.

Automated evaluation of tool-using large language model (LLM) agents is widely assumed to be reliable, but this assumption has rarely been validated against human annotation. We introduce AgentProp-Bench, a 2,000-task benchmark with 2,300 traces across four domains, nine production LLMs, and a 100-label human-validated subset. We quantify judge reliability, characterize error propagation, and evaluate a runtime mitigation. Substring-based judging agrees with human annotation at kappa=0.049 (chance-level); a three-LLM ensemble reaches kappa=0.432 (moderate) with a conservative bias. Under validated evaluation, a parameter-level injection propagates to a wrong final answer with human-calibrated probability approximately 0.62 (range 0.46-0.73 across models). Rejection (catching bad parameters) and recovery (correcting after acceptance) are independent model capabilities (Spearman rho=0.126, p=0.747). A tuned runtime interceptor reduces hallucination on GPT-4o-mini by 23.0 percentage points under a concurrent n=600 control, but shows no significant effect on Gemini-2.0-Flash, whose aggressive parameter rejection eliminates the target failure mode. All code, data, traces, and human labels are released at https://github.com/bhaskargurram-ai/agenthallu-bench.

This study examines the role of artificial intelligence in translation, focusing on ChatGPT, specifically ChatGPT-4, and the extent to which human postediting is required in literary translation. A mixed-method approach was adopted, involving 30 professional translators who evaluated and postedited AI-generated translations of selected Arabic and English literary texts. The results show that although AI improves translation speed and accessibility, it remains limited in handling cultural, stylistic, and figurative aspects of language. Participants generally confirmed the necessity of human postediting, particularly in novels and drama. The findings indicate that emerging human-machine collaboration model rather than replacement of human translators. The study concludes that AI should be used as a supportive tool, while human expertise remains essential for ensuring translation quality and cultural appropriateness.

Large language models write production code, and yet they routinely introduce well-known vulnerabilities. We show that this is not a knowledge deficit: the same models that generate insecure code, correctly identify and explain the vulnerability when asked directly, this is a gap we call the Format-Reliability Gap. Mechanistic analysis reveals the cause: security representations are encoded from the earliest layers but remain computationally inert until the final layer, where format-compliance demands compete with them. Because the failure is localized to a single layer, per-vulnerability steering vectors reduce insecure generation by up to 74% with negligible overhead. The mechanism and the fix generalize across five models, three architecture families, and six vulnerability types, suggesting insecure code generation is an interpretability problem, not a training artifact.

3D object detection in point cloud data remains a challenging task due to the sparsity and lack of global structure inherent in the input. In this work, we propose a novel Multi-Scale Attention (MSA) mechanism integrated into the 3DETR architecture to better capture both local geometry and global context. Our method introduces an upsampling operation that generates high-resolution feature maps, enabling the network to better detect smaller and semantically related objects. Experiments conducted on the ScanNetv2 dataset demonstrate that our 3DETR + MSA model improves detection performance, achieving a gain of almost 1% in mAP@25 and 4.78% in mAP@50 over the baseline. While applying MSA to the 3DETR-m variant shows limited improvement, our analysis reveals the importance of adapting the upsampling strategy for lightweight models. These results highlight the effectiveness of combining hierarchical feature extraction with attention mechanisms in enhancing 3D scene understanding.

Large reasoning models (LRMs) enhance problem-solving capabilities by generating explicit multi-step chains of thought (CoT) reasoning; however, they incur substantial inference latency and computational overhead. To mitigate this issue, recent works have explored model collaboration paradigms, where small reasoning models (SRMs) generate intermediate reasoning steps to achieve a better accuracy--latency trade-off. Despite recent progress, effectively and efficiently detecting and mitigating SRM failures in collaborative systems remains a key challenge. To address this issue, we analyze SRM inference in both the generated text and hidden-state spaces, and identify three types of failure modes: \textit{overconfidence}, \textit{uncertainty}, and \textit{heavy revalidation}. Building on these insights, we propose \textbf{RankGuide}, a framework that improves the efficiency and effectiveness of SRM--LRM collaboration through tensor-rank-guided routing and steering. Specifically, RankGuide leverages a routing signal that incorporates tensor-rank signals derived from consecutive hidden states to detect when SRMs are likely to fail and selectively invoke LRMs. In addition, we introduce a tensor-rank-filtered steering vector extraction method to modulate the reasoning trajectory of SRMs, thereby improving their generation quality. By improving both routing and steering through tensor-rank signals, RankGuide enables SRM--LRM collaborative systems to achieve more efficient reasoning with fewer steps and improved accuracy. Experiments on multiple reasoning benchmarks demonstrate the efficacy of RankGuide in reducing latency by up to $1.75\times$ compared to LRM, while maintaining competitive accuracy relative to prior methods.

Masking-based post-hoc explanation methods, such as KernelSHAP and LIME, estimate local feature importance by querying a black-box model under randomized perturbations. This paper formulates this procedure as communication over a query channel, where the latent explanation acts as a message and each masked evaluation is a channel use. Within this framework, the complexity of the explanation is captured by the entropy of the hypothesis class, while the query interface supplies information at a rate determined by an identification capacity per query. We derive a strong converse showing that, if the explanation rate exceeds this capacity, the probability of exact recovery necessarily converges to one in error for any sequence of explainers and decoders. We also prove an achievability result establishing that a sparse maximum-likelihood decoder attains reliable recovery when the rate lies below capacity. A Monte Carlo estimator of mutual information yields a non-asymptotic query benchmark that we use to compare optimal decoding with Lasso- and OLS-based procedures that mirror LIME and KernelSHAP. Experiments reveal a range of query budgets where information theory permits reliable explanations but standard convex surrogates still fail. Finally, we interpret super-pixel resolution and tokenization for neural language models as a source-coding choice that sets the entropy of the explanation and show how Gaussian noise and nonlinear curvature degrade the query channel, induce waterfall and error-floor behavior, and render high-resolution explanations unattainable.

This paper introduces a multi-agent framework guided by Large Language Models (LLMs) to assist in the early stages of engineering design, a phase often characterized by vast parameter spaces and inherent uncertainty. Operating under a human-in-the-loop paradigm and demonstrated on the canonical problem of aerodynamic airfoil design, the framework employs a team of specialized agents: a Coding Assistant, a Design Agent, a Systems Engineering Agent, and an Analyst Agent - all coordinated by a human Manager. Integrated within a set-based design philosophy, the process begins with a collaborative phase where the Manager and Coding Assistant develop a suite of validated tools, after which the agents execute a structured workflow to systematically explore and prune a large set of initial design candidates. A key contribution of this work is the explicit integration of formal risk management, employing the Conditional Value-at-Risk (CVaR) as a quantitative metric to filter designs that exhibit a high probability of failing to meet performance requirements, specifically the target coefficient of lift. The framework automates labor-intensive initial exploration through a global sensitivity analysis conducted by the Analyst agent, which generates actionable heuristics to guide the other agents. The process culminates by presenting the human Manager with a curated final set of promising design candidates, augmented with high-fidelity Computational Fluid Dynamics (CFD) simulations. This approach effectively leverages AI to handle high-volume analytical tasks, thereby enhancing the decision-making capability of the human expert in selecting the final, risk-assessed design.

Large language models (LLMs) can answer questions and summarize documents when conditioned on external contexts (e.g., retrieved evidence), yet context use remains unreliable: models may overwrite an already-correct output (neutral regression) even when the context is non-informative. We formalize neutral regression as a do-no-harm requirement and quantify it by measuring accuracy drops on baseline-correct items under answer-consistent contexts. We propose No-Worse Context-Aware Decoding (NWCAD), a decode-time adapter built on a two-stream setup with a two-stage gate: it backs off to no-context decoding when the context is non-informative, and otherwise uses context-conditioned decoding with a CAD-style fallback under uncertainty. We evaluate NWCAD on benchmarks that separate do-no-harm reliability from context utilization (accuracy gains on genuinely helpful contexts). NWCAD prevents neutral regression on baseline-correct items while preserving strong context-driven accuracy on helpful contexts.

Accurate prediction of cancer progression remains a challenge due to the high heterogeneity of molecular omics data across patients. While biologically informed models have improved the interpretability of these predictions, a persistent limitation lies in how they encode individual genes to construct pathway representations. Existing hierarchical models typically derive gene features by directly mapping raw molecular inputs, whereas integration frameworks often rely on simple statistical aggregations of patient-level signals. These approaches often fail to explicitly learn a shared base representation for each gene, thereby limiting the expressiveness and biological accuracy of downstream pathway embeddings. To address this, we introduce PATH, a modulation-based, patient-conditioned gene embedding strategy. PATH represents a paradigm shift by starting from a shared base embedding for each gene, preserving a stable biological identity across the population, and then dynamically adapting it using patient-specific copy number variation (CNV) and mutation signals. This allows the model to capture subtle individual molecular variations while maintaining a consistent latent understanding of the gene itself. We integrate PATH into a graph transformer framework that models interactions among biologically connected pathways through pathway-guided attention. Across pancancer metastasis prediction, PATH achieves an F1 score of 0.8766, representing an 8.8 percent improvement over the current SOTA multi-omics benchmarks. Beyond superior predictive accuracy, our approach identifies biologically meaningful pathways and, crucially, reveals disease-state-specific pathway rewiring, offering new insights into the evolving pathway-pathway interactions that drive cancer progression.

We study model-free reinforcement learning (RL) in non-stationary finite-horizon episodic Markov decision processes (MDPs) without prior knowledge of the non-stationarity. We focus on the piecewise-stationary (PS) setting, where both the reward and transition dynamics can change an arbitrary number of times. We propose Detection Augmented Reinforcement Learning (DARLING), a modular wrapper for PS-RL that applies to both tabular and linear MDPs, without knowledge of the changes. Under certain change-point separation and reachability conditions, DARLING improves the best available dynamic regret bounds in both settings and yields strong empirical performance. We further establish the first minimax lower bounds for PS-RL in tabular and linear MDPs, showing that DARLING is the first nearly optimal algorithm. Experiments on standard benchmarks demonstrate that DARLING consistently surpasses the state-of-the-art methods across diverse non-stationary scenarios.

Imitation learning has enabled robots to acquire complex visuomotor manipulation skills from demonstrations, but deployment failures remain a major obstacle, especially for long-horizon action-chunked policies. Once execution drifts off the demonstration manifold, these policies often continue producing locally plausible actions without recovering from the failure. Existing runtime monitors either require failure data, over-trigger under benign feature drift, or stop at failure detection without providing a recovery mechanism. We present Rewind-IL, a training-free online safeguard framework for generative action-chunked imitation policies. Rewind-IL combines a zero-shot failure detector based on Temporal Inter-chunk Discrepancy Estimate (TIDE), calibrated with split conformal prediction, with a state-respawning mechanism that returns the robot to a semantically verified safe intermediate state. Offline, a vision-language model identifies recovery checkpoints in demonstrations, and the frozen policy encoder is used to construct a compact checkpoint feature database. Online, Rewind-IL monitors self-consistency in overlapping action chunks, tracks similarity to the checkpoint library, and, upon failure, rewinds execution to the latest verified safe state before restarting inference from a clean policy state. Experiments on real-world and simulated long-horizon manipulation tasks, including transfer to flow-matching action-chunked policies, demonstrate that policy-internal consistency coupled with semantically grounded respawning offers a practical route to improved reliability in imitation learning. Supplemental materials are available at https://sjay05.github.io/rewind-il

Power has become a central bottleneck for AI inference. This problem is becoming more urgent as agentic AI emerges as a major workload class, yet prior power-management techniques focus almost entirely on single-turn LLM serving. Our analysis shows that agentic serving behaves fundamentally differently: each request carries long-lived context that evolves across tool-interleaved turns, and lowering GPU frequency can push the system into a thrashing regime where memory pressure sharply worsens both performance and power efficiency. These observations show that power optimization for agentic serving requires rethinking. We present KAIROS, a context-aware power optimization system for agentic AI serving. KAIROS uses agent context as a first-class control signal to jointly manage GPU frequency, per-instance concurrency, and multi-instance request placement. This enables KAIROS to save power when memory headroom exists while avoiding thrashing and preserving performance targets. At a high level, KAIROS tracks requests at agent granularity, adapts local control to context growth and agent progress, and routes agents across instances to jointly improve power efficiency and memory stability. Evaluated across diverse software and data engineering agentic tasks, KAIROS achieves an average of 27% (up to 39.8%) power reduction while meeting the performance targets.

Contrastive objectives power state-of-the-art multimodal models, but their training remains slow, relying on long stochastic optimization. We propose a Unified Framework for Efficient Contrastive Alignment via Kernels (UniCon), which spans linear and nonlinear encoders as well as one-to-one and many-to-many alignments. At its core, UniCon introduces the contrastive similarity weight matrix $S(γ)$, which enables closed-form global solutions that provably replace minibatch back-propagation with exact updates. Through the lens of reproducing kernel Hilbert spaces (RKHS), UniCon provides a kernelized perspective that unifies contrastive alignment and reveals its connection to spectral methods. To validate the theory, we conduct experiments on synthetic, unimodal, multimodal, and zero-shot tasks, demonstrating that UniCon achieves substantial efficiency gains while preserving generality and strong empirical performance.

Vision-language-action (VLA) models have emerged as generalist robotic controllers capable of mapping visual observations and natural language instructions to continuous action sequences. However, VLAs provide no calibrated measure of confidence in their action predictions, thus limiting their reliability in real-world settings where uncertainty and failures must be anticipated. To address this problem we introduce ReconVLA, a reliable conformal model that produces uncertainty-guided and failure-aware control signals. Concretely, our approach applies conformal prediction directly to the action token outputs of pretrained VLA policies, yielding calibrated uncertainty estimates that correlate with execution quality and task success. Furthermore, we extend conformal prediction to the robot state space to detect outliers or unsafe states before failures occur, providing a simple yet effective failure detection mechanism that complements the action-level uncertainty. We evaluate ReconVLA in both simulation and real robot experiments across diverse manipulation tasks. Our results show that conformalized action predictions consistently improve failure anticipation, reduce catastrophic errors, and provide a calibrated measure of confidence without retraining or modifying the underlying VLA.

In active learning, membership queries (MQs) allow a learner to pose questions to a teacher, such as ''Is every apple a fruit?'', to which the teacher responds correctly with yes or no. These MQs can be viewed as subsumption tests with respect to the target ontology. Inspired by the standard reduction of subsumption to satisfiability in description logics, we reformulate each candidate axiom into its corresponding counter-concept and verbalise it in controlled natural language before presenting it to Large Language Models (LLMs). We introduce LLMs as a third component that provides real-world examples approximating an instance of the counter-concept. This design property ensures that only Type II errors may occur in ontology modelling; in the worst case, these errors merely delay the construction process without introducing inconsistencies. Experimental results on 13 commercial LLMs show that recall, corresponding to Type II errors in our framework, remains stable across several well-established ontologies.

Urgent blood donation seeking posts and messages on social media often go unnoticed due to the overwhelming volume of daily communications. Traditional app-based systems, reliant on manual input, struggle to reach users in low-resource settings, delaying critical responses. To address this, we introduce the Cognitive Blood Request System (CBRS), a multi-platform framework that efficiently filters and parses blood donation requests from social media streams using a cost-efficient dual-layered architecture. To do so, we curate a novel dataset of 11K parsed blood donation request messages in Bengali, English, and transliterated Bengali, capturing the linguistic diversity of real social media communications. The inclusion of adversarial negatives further enhances the robustness of our model. CBRS achieves an impressive 99% accuracy and precision in filtering, surpassing benchmark methods. In the parsing task, our LoRA finetuned Llama-3.2-3B model achieves 92% zero-shot accuracy, surpassing the base model by 41.54% and exceeding the few-shot performance of GPT-4o-mini, Gemini-2.0-Flash, and other LLMs, while resulting in a 35X reduction in input token usage. This work lays a robust foundation for scalable, inclusive information extraction in time-sensitive, object-focused tasks. Our code, dataset, and trained models are publicly available at [https://github.com/aaniksahaa/CBRS](https://github.com/aaniksahaa/CBRS).

Large pre-trained language models are increasingly adapted to downstream tasks using parameter-efficient fine-tuning (PEFT), but existing PEFT methods are typically deterministic and unimodal, making them poorly suited for low-resource multimodal settings where predictive uncertainty and cross-modal reliability both matter. We introduce CALIBER (Context-Aware Low-rank Inference with Bayesian Embedding Regularization), a multimodal uncertainty-aware PEFT framework for audio-text learning. CALIBER extends Bayesian low-rank adaptation by conditioning the variational posterior in the adapter space on per-layer, token-level text-audio cross-attention. Specifically, text-derived low-rank features attend to frame-level audio embeddings to produce localized acoustic context, which then modulates the mean and variance of a compact stochastic latent matrix within the rank-$r$ adapter space. This design treats audio not only as an additional feature source, but as a contextual reliability signal that shapes both adaptation and confidence. By confining stochasticity to a low-dimensional latent component, CALIBER retains the computational efficiency and scalability of PEFT while enabling heteroscedastic multimodal uncertainty estimation. Experimental results across diverse text and audio backbones show that CALIBER consistently matches or improves upon text-only Bayesian PEFT and conventional multimodal transfer-learning baselines, with token-level cross-attention yielding the most consistent gains. Our findings demonstrate that localized cross-modal conditioning is an effective and lightweight mechanism for uncertainty-aware multimodal adaptation.

All languages are equal; when it comes to tokenization, some are more equal than others. Tokens are the hidden currency that dictate the cost and latency of access to contemporary LLMs. However, many languages written in non-Latin scripts observe a poor exchange rate: LLMs take several multiples of tokens to encode the same information in many languages as they do for English. Our analysis reveals that this issue, known as 'token over-fragmentation', persists in modern open-weight LLMs. The standard remedy is vocabulary expansion that adds target language items missing from the model's vocabulary. In this work, we comprehensively study and advance interpretability-based vocabulary expansion, a new research direction. We focus on two core decisions in the vocabulary expansion process: What items should we add? and How should we initialize their corresponding input and output embeddings? First, we question the conventional use of frequency-based methods to choose candidate vocabulary items to add (a decision long treated as settled), and show that interpretability-based methods offer a superior performance-token efficiency trade-off. Next, we strengthen the case for interpretability-based embedding initialization by showing large gains (~20 pts) over baseline initialization methods for several languages written in non-Latin scripts. We identify the phenomenon of "subword detokenization" where models progressively merge fragmented subword tokens into larger subwords across layers. Grounded in our analysis of this phenomenon, we propose FragMend to further push the efficiency ceiling of interpretability-based expansion. We validate the effectiveness of FragMend through comparison against strong baselines and we present extensive analysis of its design choices.

The accurate quantification of brain age from MRI has emerged as an important biomarker of brain health. However, existing approaches are often restricted to narrow age ranges and single-modality MRI data, limiting their capacity to capture the coordinated macro- and microstructural changes that unfold across the human lifespan. To address these limitations, we developed a multi-modal brain age framework to characterize the integrated evolution of brain morphology and white matter organization. Our model adopts a two-stage architecture, where modalities are processed independently and integrated via late fusion in both stages: first to classify each subject into one of six developmental stages, and then to estimate age within the predicted stage. This design enables a unified and lifespan-spanning assessment of brain maturity across diverse developmental periods.

Reclaimed slur usage is a common and meaningful practice online for many marginalized communities. It serves as a source of solidarity, identity, and shared experience. However, contemporary automated and AI-based moderation tools for online content largely fail to distinguish between reclaimed and hateful uses of slurs, resulting in the suppression of marginalized voices. In this work, we use quantitative and qualitative methods to examine the attitudes of social media users in LGBTQIA+, Black, and women communities around reclaimed slurs targeting our focus groups including the f-word, n-word, and b-word. With social media users from these communities, we collect and analyze an annotated online slur usage corpus. The corpus includes annotators' perceptions of whether an online text containing a slur should be flagged as hate speech, as well as contextual features of the slur usage. Across all communities and annotation questions, we observe low inter-annotator agreement, indicating substantial disagreement among in-group annotators. This is compounded by the fact that, absent clear contextual signals of identity and intent, even in-group members may disagree on how to interpret reclaimed slur usage online. Semi-structured interviews with annotators suggest that differences in lived experience and personal history contribute to this variation as well. We find poor alignment between annotator judgments and automated hate speech assessments produced by Perspective API. We further observe that certain features of a text such as whether the slur usage was derogatory and if the slur was targeted at oneself are more associated with whether annotators report the text as hate speech. Together, these findings highlight the inherent subjectivity and contextual nature of how marginalized communities interpret slurs online.

Research shows news consumption differs across demographics, yet little is known about non-mainstream audiences, especially in relation to local media. Our study addresses this gap by examining how French-speaking migrants in a mid-size European city engage with local news, and whether their needs are reflected in coverage. Eight community members participated in focus groups, whose insights guided the selection of natural language processing methods (topic modeling, information retrieval, sentiment analysis, and readability) applied to over 2000 hyper-local news articles. Results showed that while articles frequently covered local events, gaps remained in topics important to participants. Sentiment analysis revealed a generally positive tone, and readability measures indicated an intermediate-advanced French level, raising questions about accessibility for integration. Our work contributes to bridging the gap between local news platforms' content and diverse readers' needs, and could inform local media organizations about opportunities to expand their current news story coverage to appeal to more diverse audiences.

Directed energy deposition (DED) produces complex thermo-mechanical responses that can lead to distortion and reduced dimensional accuracy of a manufactured part. Thermo-mechanical finite element simulations are widely used to estimate these effects, but their computational cost and the complexity of accurately capturing DED physics limit their use in design iteration and process optimization. This paper introduces FLARE (Field Prediction via Linear Affine Reconstruction in wEight-space), a data-efficient surrogate modeling framework for predicting post-cooling displacement fields in DED from geometric and process parameters. We develop a predefined-geometry DED simulation workflow using an open-source finite element framework and generate a dataset of simulations with varying geometry, laser power, and deposition velocity. Each simulation provides full-field displacement, stress, strain, and temperature data throughout the manufacturing process. FLARE encodes each simulation as an implicit neural field and regularizes the corresponding neural-network weights so that they follow the affine structure of the input parameter space. This enables prediction of unseen parameter combinations by reconstructing network weights through affine mixing of training examples. On this DED benchmark, the method shows improved accuracy compared to baseline methods in both in-distribution and extrapolation settings. Although the present study focuses on DED displacement prediction, the proposed affine weight-space reconstruction framework offers a promising approach for data-efficient surrogate modeling of physical fields.

In this work, we present FRIGID, a framework with a novel diffusion language model that generates molecular structures conditioned on mass spectra via intermediate fingerprint representations and determined chemical formulae, training at the scale of hundreds of millions of unlabeled structures. We then demonstrate how forward fragmentation models enable inference-time scaling by identifying spectrum-inconsistent fragments and refining them through targeted remasking and denoising. While FRIGID already achieves strong performance with its diffusion base, inference-time scaling significantly improves its accuracy, surpassing 18% Top-1 accuracy on the challenging MassSpecGym benchmark and tripling the Top-1 accuracy of the leading methods on NPLIB1. Further empirical analyses show that FRIGID exhibits log-linear performance scaling with increasing inference-time compute, opening a promising new direction for continued improvements in de novo structural elucidation. FRIGID code is publicly available at https://github.com/coleygroup/FRIGID

Recent advances in agentic frameworks have enabled AI agents to perform complex reasoning and decision-making. However, evidence comparing their reasoning performance, efficiency, and practical suitability remains limited. To address this gap, we empirically evaluate 22 widely used agentic frameworks across three reasoning benchmarks: BBH, GSM8K, and ARC. The frameworks were selected from 1,200 GitHub repositories collected between January 2023 and July 2025 and organized into a taxonomy based on architectural design. We evaluated them under a unified setting, measuring reasoning accuracy, execution time, computational cost, and cross-benchmark consistency. Our results show that 19 of the 22 frameworks completed all three benchmarks. Among these, 12 showed stable performance, with mean accuracy of 74.6-75.9%, execution time of 4-6 seconds per task, and cost of 0.14-0.18 cents per task. Poorer results were mainly caused by orchestration problems rather than reasoning limits. For example, Camel failed to complete BBH after 11 days because of uncontrolled context growth, while Upsonic consumed USD 1,434 in one day because repeated extraction failures triggered costly retries. AutoGen and Mastra also exhausted API quotas through iterative interactions that increased prompt length without improving results. We also found a sharp drop in mathematical reasoning. Mean accuracy on GSM8K was 44.35%, compared with 89.80% on BBH and 89.56% on ARC. Overall, this study provides the first large-scale empirical comparison of agentic frameworks for reasoning-intensive software engineering tasks and shows that framework selection should prioritize orchestration quality, especially memory control, failure handling, and cost management.

Recent large language model (LLM) agents have shown promise in using execution feedback for test-time adaptation. However, robust self-improvement remains far from solved: most approaches still treat each problem instance independently, without accumulating reusable knowledge. This limitation is particularly pronounced in domain-specific languages such as Triton, which are underrepresented in LLM pretraining data. Their strict constraints and non-linear optimization landscape further make naive generation and local refinement unreliable. We propose AdaExplore, an agent framework that enables self-improvement via accumulated execution feedback for performance-critical kernel code generation through two complementary stages: failure-driven adaptation and diversity-preserving search, jointly improving correctness and optimization performance without additional fine-tuning or external knowledge. In the adaptation stage, the agent synthesizes tasks and converts recurring failures into a reusable memory of validity rules, helping subsequent generations remain within the feasible set. In the search stage, the agent organizes candidate kernels as a tree and alternates between small local refinements and larger structural regeneration, allowing it to explore the optimization landscape beyond local optima. Experiments on kernel runtime optimization benchmarks validate these gains: AdaExplore achieves 3.12x and 1.72x speedups on KernelBench Level-2 and Level-3, respectively, within 100 steps, and continues to improve with additional computation.

Backchannels (e.g., `yeah', `mhm', and `right') are short, non-interruptive feedback signals whose lexical form and prosody jointly convey pragmatic meaning. While prior computational research has largely focused on predicting backchannel timing, the relationship between lexico-prosodic form and meaning remains underexplored. We propose a two-stage framework: first, fine-tuning large language models on dialogue transcripts to derive rich contextual representations; and second, learning a joint embedding space for dialogue contexts and backchannel realizations. We evaluate alignment with human perception via triadic similarity judgments (prosodic and cross-lexical) and a context-backchannel suitability task. Our results demonstrate that the learned projections substantially improve context-backchannel retrieval compared to previous methods. In addition, they reveal that backchannel form is highly sensitive to extended conversational context and that the learned embeddings align more closely with human judgments than raw WavLM features.

Analysis of Stochastic Gradient Descent (SGD) and its variants typically relies on the assumption of uniformly bounded variance, a condition that frequently fails in practical non-convex settings, such as neural network training, as well as in several elementary optimization settings. While several relaxations are explored in the literature, the Blum-Gladyshev (BG-0) condition, which permits the variance to grow quadratically with distance has recently been shown to be the weakest condition. However, the study of the oracle complexity of stochastic first-order non-convex optimization under BG-0 has remained underexplored. In this paper, we address this gap and establish information-theoretic lower bounds, proving that finding an $ε$-stationary point requires $Ω(ε^{-6})$ stochastic BG-0 oracle queries for smooth functions and $Ω(ε^{-4})$ queries under mean-square smoothness. These limits demonstrate an unavoidable degradation from classical bounded-variance complexities, i.e., $Ω(ε^{-4})$ and $Ω(ε^{-3})$ for smooth and mean-square smooth cases, respectively. To match these lower bounds, we consider Proximally Anchored STochastic Approximation (PASTA), a unified algorithmic framework that couples Halpern anchoring with Tikhonov regularization to dynamically mitigate the extra variance explosion term permitted by the BG-0 oracle. We prove that PASTA achieves minimax optimal complexities across numerous non-convex regimes, including standard smooth, mean-square smooth, weakly convex, star-convex, and Polyak-Lojasiewicz functions, entirely under an unbounded domain and unbounded stochastic gradients.

We introduce CoCo-LoRA, a multimodal, uncertainty-aware parameter-efficient fine-tuning method for text prediction tasks accompanied by audio context. Existing PEFT approaches such as LoRA are efficient but typically deterministic, while recent Bayesian low-rank adapters model uncertainty in a lightweight way yet remain largely unimodal and condition uncertainty primarily on internal text features. This leaves them poorly equipped to reflect uncertainty driven by external acoustic factors such as background noise, channel variability, or speaking style, which can materially affect reliability in speech-centered applications. CoCo-LoRA addresses this gap by conditioning a contextual variational posterior in the low-rank space on both local text-derived adapter features and an audio-derived context signal. A pooled audio embedding is projected once into a shared context space and then adapted through lightweight layer-wise heads, enabling global-to-local, depth-specific modulation of the adapter uncertainty and update without high-dimensional multimodal fusion. Stochasticity is confined to a compact latent component in the rank space, preserving PEFT scalability while producing audio-sensitive, heteroscedastic uncertainty. Based on our evaluations across diverse tasks and backbone combinations, CoCo-LoRA consistently matches or outperforms text-only PEFT and conventional feature-fusion transfer baselines, particularly on high-coverage labels where reliable adaptation is critical. The results indicate that using audio as a contextual uncertainty signal, rather than as a fused feature stream, provides a robust and parameter-efficient alternative for multimodal low-resource prediction.

Circuit-level decoders are essential for the realisation of low-overhead fault-tolerant quantum computing. However, they rely on complex hypergraphs that are traditionally compiled ahead-of-time. This static approach introduces a significant bottleneck for an emerging class of adaptive circuits, where the structure is modified during execution based on mid-circuit measurement outcomes. Pre-compiling hypergraphs for all possible circuit branches would incur an exponential memory cost, rendering current tools impractical for these workloads. Hence, we introduce GreenPeas, a C++/CUDA toolchain for the high-speed, just-in-time compilation of decoding hypergraphs. By lowering the circuit to a space-time error propagation graph, we show how Stim's backtracking algorithm can be mapped efficiently onto massively parallel GPU architectures, decomposing the O(nl) workload for a circuit with n qubits and l gate layers across thousands of concurrent threads. Our implementation achieves a greater than 10x average speedup over the Stim baseline across two of the leading fault-tolerant architectures: the surface and bivariate bicycle codes. As a key use case, we demonstrate that this speedup enables circuit-level decoding of adaptive syndrome measurement circuits, unlocking a regime previously restricted to less accurate phenomenological decoders. We aim to open-source GreenPeas to support the research of future adaptive circuit protocols.

Traffic prediction plays a central role in intelligent transportation systems (ITS) by supporting real-time decision-making, congestion management, and long-term planning. However, many existing approaches face practical limitations. Most spatio-temporal models are trained on centralized data, rely on numerical representations, and offer limited explainability. Recent Large Language Model (LLM) methods improve reasoning capabilities but typically assume centralized data availability and do not fully capture the distributed and heterogeneous nature of real-world traffic systems. To address these challenges, this study proposes FedLLM (Federated LLM), a privacy-preserving and distributed framework for explainable multi-horizon short-term traffic flow prediction (15-60 minutes). The framework introduces four key contributions: 1) a Composite Selection Score (CSS) for data-driven freeway selection that captures structural diversity across traffic regions 2) a domain-adapted LLM fine-tuned on structured traffic prompts encoding spatial, temporal, and statistical context 3) FedLLM framework, that enables collaborative training across heterogeneous clients while exchanging only lightweight LoRA adapter parameters, 4) a structured prompt representation that supports contextual reasoning and cross-region generalization. The FedLLM design allows each client to learn from local traffic patterns while contributing to a shared global model through efficient parameter exchange, reducing communication overhead and keeping data private. This setup supports learning under non-IID traffic distributions. Experimental results show that FedLLM achieves improved predictive performance over centralized baselines, while producing structured and explainable outputs. These findings highlight the potential of combining FL with domain-adapted LLMs for scalable, privacy-aware, and explainable traffic prediction.

Machine learning models often inherit biases from historical data, raising critical concerns about fairness and accountability. Conventional fairness interventions typically require access to sensitive attributes like gender or race, but privacy and legal restrictions frequently limit their use. To address this challenge, we propose a framework that infers sensitive attributes from auxiliary features and integrates fairness constraints into model training. Our approach mitigates bias while preserving predictive accuracy, offering a practical solution for fairness-aware learning. Empirical evaluations validate its effectiveness, contributing to the advancement of more equitable algorithmic decision-making.

With the rise of generative language models, machine-generated text detection has become a critical challenge. A wide variety of models is available, but inconsistent datasets, evaluation metrics, and assessment strategies obscure comparisons of model effectiveness. To address this, we evaluate 15 different detection models from six distinct systems, as well as seven trained models, across seven English-language textual test sets and three creative human-written datasets. We provide an empirical analysis of model performance, the influence of training and evaluation data, and the impact of key metrics. We find that no single system excels in all areas and nearly all are effective for certain tasks, and the representation of model performance is critically linked to dataset and metric choices. We find high variance in model ranks based on datasets and metrics, and overall poor performance on novel human-written texts in high-risk domains. Across datasets and metrics, we find that methodological choices that are often assumed or overlooked are essential for clearly and accurately reflecting model performance.

Large language models (LLMs) are increasingly deployed in high-stakes domains, yet a unified treatment of their overlapping safety challenges remains lacking. We present SafeLM, a framework that jointly addresses four pillars of LLM safety: privacy, security, misinformation, and adversarial robustness. SafeLM combines federated training with gradient smartification and Paillier encryption for privacy, integrates defenses against training and inference-time attacks, employs contrastive grounding with calibrated decoding to reduce hallucinations, and introduces alignment-aware binarized aggregation to enhance robustness while maintaining bounded reconstruction quality. Across benchmarks on factuality, toxicity, and membership inference, SafeLM achieves 98.0% harmful content detection accuracy, reduces communication by 96.9%, and lowers gradient inversion PSNR from 31.7 dB to 15.1 dB. Ablations show that each component contributes independently, whereas their integration yields a strong privacy utility efficiency trade-off for deploying trustworthy LLMs.

We present SemanticQA, an evaluation suite designed to assess language models (LMs) in semantic phrase processing tasks. The benchmark consolidates existing multiword expression (MwE) resources and reorganizes them into a unified testbed. It covers both general lexical phenomena, such as lexical collocations, and three fine-grained categories: idiomatic expressions, noun compounds, and verbal constructions. Through SemanticQA, we assess LMs of diverse architectures and scales in extraction, classification, and interpretation tasks, as well as sequential task compositions. We reveal substantial performance variation, particularly on tasks requiring semantic reasoning, highlighting differences in reasoning efficacy and semantic understanding of LMs, providing insights for pushing LMs with stronger comprehension on non-trivial semantic phrases. The evaluation harness and data of SemanticQA are available at https://github.com/jacklanda/SemanticQA.

This comprehensive report distinguishes prior works by the cognitive functions they innovate. Many works claim an almost "human-like" cognitive capability in their world models. To evaluate these claims requires a proper grounding in first principles in Cognitive Architecture Theory (CAT). We present a conceptual unified framework for world models that fully incorporates all the cognitive functions associated with CAT (i.e. memory, perception, language, reasoning, imagining, motivation, and meta-cognition) and identify gaps in the research as a guide for future states of the art. In particular, we find that motivation (especially intrinsic motivation) and meta-cognition remain drastically under-researched, and we propose concrete directions informed by active inference and global workspace theory to address them. We further introduce Epistemic World Models, a new category encompassing agent frameworks for scientific discovery that operate over structured knowledge. Our taxonomy, applied across video, embodied, and epistemic world models, suggests research directions where prior taxonomies have not.

As AI systems are increasingly used to conduct research autonomously, misaligned systems could introduce subtle flaws that produce misleading results while evading detection. We introduce ASMR-Bench (Auditing for Sabotage in ML Research), a benchmark for evaluating the ability of auditors to detect sabotage in ML research codebases. ASMR-Bench consists of 9 ML research codebases with sabotaged variants that produce qualitatively different experimental results. Each sabotage modifies implementation details, such as hyperparameters, training data, or evaluation code, while preserving the high-level methodology described in the paper. We evaluated frontier LLMs and LLM-assisted human auditors on ASMR-Bench and found that both struggled to reliably detect sabotage: the best performance was an AUROC of 0.77 and a top-1 fix rate of 42%, achieved by Gemini 3.1 Pro. We also tested LLMs as red teamers and found that LLM-generated sabotages were weaker than human-generated ones but still sometimes evaded same-capability LLM auditors. We release ASMR-Bench to support research on monitoring and auditing techniques for AI-conducted research.

Stochastic dynamical systems with slow or metastable behavior evolve, on long time scales, on an unknown low-dimensional manifold in high-dimensional ambient space. Building a reduced simulator from short-burst ambient ensembles is a long-standing problem: local-chart methods like ATLAS suffer from exponential landmark scaling and per-step reprojection, while autoencoder alternatives leave tangent-bundle geometry poorly constrained, and the errors propagate into the learned drift and diffusion. We observe that the ambient covariance~$Λ$ already encodes coordinate-invariant tangent-space information, its range spanning the tangent bundle. Using this, we construct a tangent-bundle penalty and an inverse-consistency penalty for a three-stage pipeline (chart learning, latent drift, latent diffusion) that learns a single nonlinear chart and the latent SDE. The penalties induce a function-space metric, the $ρ$-metric, strictly weaker than the Sobolev $H^1$ norm yet achieving the same chart-quality generalization rate up to logarithmic factors. For the drift, we derive an encoder-pullback target via Itô's formula on the learned encoder and prove a bias decomposition showing the standard decoder-side formula carries systematic error for any imperfect chart. Under a $W^{2,\infty}$ chart-convergence assumption, chart-level error propagates controllably to weak convergence of the ambient dynamics and to convergence of radial mean first-passage times. Experiments on four surfaces embedded in up to $201$ ambient dimensions reduce radial MFPT error by $50$--$70\%$ under rotation dynamics and achieve the lowest inter-well MFPT error on most surface--transition pairs under metastable Müller--Brown Langevin dynamics, while reducing end-to-end ambient coefficient errors by up to an order of magnitude relative to an unregularized autoencoder.

Large language models (LLMs) sometimes memorize undesirable knowledge, which must be removed after deployment. Prior work on machine unlearning has focused largely on optimization methods that adjust parameters to enforce forgetting while preserving retention. However, these approaches assume that the forget and retain sets are readily available, which rarely holds in practice. Unlearning is typically triggered by an undesired generation at inference time, making the retrieval of relevant data the central challenge. We introduce the notion of data Pareto improvement for LLM unlearning, which formalizes how retrieval can expand the achievable trade-off frontier between forgetting and retention. To realize this principle, we propose Randomized Antipodal Search on Linearized Influence Kernel (RASLIK), a retrieval algorithm that combines permutation-projection hashing with randomized antipodal search. RASLIK reduces selection variance, achieves sublinear complexity, and yields a double gain in both quality and efficiency. Across multiple models, datasets, and unlearning algorithms, RASLIK consistently outperforms deterministic baselines and even oracle sampling, establishing randomized search as a principled and scalable solution for data-centric unlearning.

Explaining Machine Learning (ML) results in a transparent and user-friendly manner remains a challenging task of Explainable Artificial Intelligence (XAI). In this paper, we present a method to enhance the interpretability of ML models by using a Knowledge Graph (KG). We store domain-specific data along with ML results and their corresponding explanations, establishing a structured connection between domain knowledge and ML insights. To make these insights accessible to users, we designed a selective retrieval method in which relevant triplets are extracted from the KG and processed by a Large Language Model (LLM) to generate user-friendly explanations of ML results. We evaluated our method in a manufacturing environment using the XAI Question Bank. Beyond standard questions, we introduce more complex, tailored questions that highlight the strengths of our approach. We evaluated 33 questions, analyzing responses using quantitative metrics such as accuracy and consistency, as well as qualitative ones such as clarity and usefulness. Our contribution is both theoretical and practical: from a theoretical perspective, we present a novel approach for effectively enabling LLMs to dynamically access a KG in order to improve the explainability of ML results. From a practical perspective, we provide empirical evidence showing that such explanations can be successfully applied in real-world manufacturing environments, supporting better decision-making in manufacturing processes.

Large Language Models (LLMs) have the potential to accelerate small molecule drug design due to their ability to reason about information from diverse sources and formats. However, their practical utility remains unclear due to the lack of benchmarks that reflect real-world scenarios. In this work, we introduce a suite of chemically-grounded tasks spanning molecular property prediction, molecular representation transformations, and molecular design. Importantly, we formulate these tasks as reinforcement learning (RL) environments, enabling a unified approach for evaluation and post-training. Across three model families, we find that frontier models are increasingly proficient at chemical tasks, but that there is significant room for improvement, especially in experimental settings with low data. Critically, we show that RL-based post-training can substantially improve performance. A smaller model post-trained on our environments becomes competitive with state-of-the-art frontier models, despite a significantly weaker base model. This suggests a practical route toward employing LLMs in drug discovery; by combining carefully-designed evaluation tasks with targeted post-training, we can both elucidate and close critical capability gaps.

Although most of the automated theorem-proving approaches depend on formal proof systems, informal theorem proving can align better with large language models' (LLMs) strength in natural language processing. In this work, we identify a primary bottleneck in informal theorem proving as a lack of insight, namely the difficulty of recognizing the core techniques required to solve complex problems. To address this, we propose a novel framework designed to cultivate this essential reasoning skill and enable LLMs to perform insightful reasoning. We propose $\mathtt{DeepInsightTheorem}$, a hierarchical dataset that structures informal proofs by explicitly extracting core techniques and proof sketches alongside the final proof. To fully exploit this dataset, we design a Progressive Multi-Stage SFT strategy that mimics the human learning process, guiding the model from basic proof writing to insightful thinking. Our experiments on challenging mathematical benchmarks demonstrate that this insight-aware generation strategy significantly outperforms baselines. These results demonstrate that teaching models to identify and apply core techniques can substantially improve their mathematical reasoning.

This paper explores the response of Large Language Models (LLMs) to user prompts with different degrees of politeness and impoliteness. The Politeness Theory by Brown and Levinson and the Impoliteness Framework by Culpeper form the basis of experiments conducted across three languages (English, Hindi, Spanish), five models (Gemini-Pro, GPT-4o Mini, Claude 3.7 Sonnet, DeepSeek-Chat, and Llama 3), and three interaction histories between users (raw, polite, and impolite). Our sample consists of 22,500 pairs of prompts and responses of various types, evaluated across five levels of politeness using an eight-factor assessment framework: coherence, clarity, depth, responsiveness, context retention, toxicity, conciseness, and readability. The findings show that model performance is highly influenced by tone, dialogue history, and language. While polite prompts enhance the average response quality by up to ~11% and impolite tones worsen it, these effects are neither consistent nor universal across languages and models. English is best served by courteous or direct tones, Hindi by deferential and indirect tones, and Spanish by assertive tones. Among the models, Llama is the most tone-sensitive (11.5% range), whereas GPT is more robust to adversarial tone. These results indicate that politeness is a quantifiable computational variable that affects LLM behaviour, though its impact is language- and model-dependent rather than universal. To support reproducibility and future work, we additionally release PLUM (Politeness Levels in Utterances, Multilingual), a publicly available corpus of 1,500 human-validated prompts across three languages and five politeness categories, and provide a formal supplementary analysis of six falsifiable hypotheses derived from politeness theory, empirically assessed against the dataset.

As AI-assisted video creation becomes increasingly practical, instruction-guided video editing has become essential for refining generated or captured footage to meet professional requirements. Yet the field still lacks both a large-scale human-annotated dataset with complete editing examples and a standardized evaluator for comparing editing systems. Existing resources are limited by small scale, missing edited outputs, or the absence of human quality labels, while current evaluation often relies on expensive manual inspection or generic vision-language model judges that are not specialized for editing quality. We introduce VEFX-Dataset, a human-annotated dataset containing 5,049 video editing examples across 9 major editing categories and 32 subcategories, each labeled along three decoupled dimensions: Instruction Following, Rendering Quality, and Edit Exclusivity. Building on VEFX-Dataset, we propose VEFX-Reward, a reward model designed specifically for video editing quality assessment. VEFX-Reward jointly processes the source video, the editing instruction, and the edited video, and predicts per-dimension quality scores via ordinal regression. We further release VEFX-Bench, a benchmark of 300 curated video-prompt pairs for standardized comparison of editing systems. Experiments show that VEFX-Reward aligns more strongly with human judgments than generic VLM judges and prior reward models on both standard IQA/VQA metrics and group-wise preference evaluation. Using VEFX-Reward as an evaluator, we benchmark representative commercial and open-source video editing systems, revealing a persistent gap between visual plausibility, instruction following, and edit locality in current models.

The complexity of Vietnam's legal texts presents a significant barrier to public access to justice. While Large Language Models offer a promising solution for legal text simplification, evaluating their true capabilities requires a multifaceted approach that goes beyond surface-level metrics. This paper introduces a comprehensive dual-aspect evaluation framework to address this need. First, we establish a performance benchmark for four state-of-the-art large language models (GPT-4o, Claude 3 Opus, Gemini 1.5 Pro, and Grok-1) across three key dimensions: Accuracy, Readability, and Consistency. Second, to understand the "why" behind these performance scores, we conduct a large-scale error analysis on a curated dataset of 60 complex Vietnamese legal articles, using a novel, expert-validated error typology. Our results reveal a crucial trade-off: models like Grok-1 excel in Readability and Consistency but compromise on fine-grained legal Accuracy, while models like Claude 3 Opus achieve high Accuracy scores that mask a significant number of subtle but critical reasoning errors. The error analysis pinpoints \textit{Incorrect Example} and \textit{Misinterpretation} as the most prevalent failures, confirming that the primary challenge for current LLMs is not summarization but controlled, accurate legal reasoning. By integrating a quantitative benchmark with a qualitative deep dive, our work provides a holistic and actionable assessment of LLMs for legal applications.

Existing multi-hazard susceptibility mapping (MHSM) studies often rely on spatially uniform models, treat hazards independently, and provide limited representation of cross-hazard dependence and uncertainty. To address these limitations, this study proposes a deep learning (DL) workflow for joint flood-landslide multi-hazard susceptibility mapping (FL-MHSM) that combines two-level spatial partitioning, probabilistic Early Fusion (EF), a tree-based Late Fusion (LF) baseline, and a soft-gating Mixture of Experts (MoE) model, with MoE serving as final predictive model. The proposed design preserves spatial heterogeneity through zonal partitions and enables data-parallel large-area prediction using overlapping lattice grids. In Kerala, EF remained competitive with LF, improving flood recall from 0.816 to 0.840 and reducing Brier score from 0.092 to 0.086, while MoE provided strongest performance for flood susceptibility, achieving an AUC-ROC of 0.905, recall of 0.930, and F1-score of 0.722. In Nepal, EF similarly improved flood recall from 0.820 to 0.858 and reduced Brier score from 0.057 to 0.049 relative to LF, while MoE outperformed both EF and LF for landslide susceptibility, achieving an AUC-ROC of 0.914, recall of 0.901, and F1-score of 0.559. GeoDetector analysis of MoE outputs further showed that dominant factors varied more across zones in Kerala, where susceptibility was shaped by different combinations of topographic, land-cover, and drainage-related controls, while Nepal showed a more consistent influence of topographic and glacier-related factors across zones. These findings show that EF and LF provide complementary predictive behavior, and that their spatially adaptive integration through MoE yields robust overall predictive performance for FL-MHSM while supporting interpretable characterization of multi-hazard susceptibility in spatially heterogeneous landscapes.

Vision Language models (VLMs) have demonstrated strong performance across a wide range of benchmarks, yet they often suffer from modality dominance, where predictions rely disproportionately on a single modality. Prior approaches primarily address this issue by steering model's attention allocation, implicitly assuming that all modalities provide sufficient information. However, attention only determines where the model focuses, and cannot enrich information that is missing or ambiguous. In the real world, input modalities often differ in information density and their signal-to-noise ratios. In such cases, simply adjusting model's attention does not resolve the underlying lack of information. In this paper, we propose \textsc{MoIR}: \textit{Multi-modal Information Router}, an information-level fusion method that explicitly reduces information disparity prior to fusion. \textsc{MoIR} identifies less informative tokens and routes complementary information from a stronger modality, constructing information-dense token representations before they are processed by a large language model. By modifying information availability, \textsc{MoIR} enables reliable shifts in modality dominance, even when one modality is degraded. We evaluate \textsc{MoIR} on three widely used multi-modal benchmarks across multiple model backbones. Experimental results show that \textsc{MoIR} consistently demonstrates more balanced modality contribution, and improves robustness and downstream performance, particularly even under modality degradation. These findings demonstrate that explicitly modifying cross-modal information is an effective and complementary strategy for mitigating modality dominance in multi-modal reasoning models.

Accurate weather and climate prediction relies on data assimilation (DA), which estimates the Earth system state by integrating observations with models. While exascale computing has significantly advanced earth simulation, scalable and accurate inference of the Earth system state remains a fundamental bottleneck, limiting uncertainty quantification and prediction of extreme events. We introduce a unified one-stage generative DA framework that reformulates assimilation as Bayesian posterior sampling, replacing the conventional forecast-update cycle with compute-dense, GPU-efficient inference. At the core is STORM, a novel spatiotemporal transformer with a global attention linear-complexity scaling algorithm that breaks the quadratic attention barrier. On 32,768 GPUs of the Frontier supercomputer, our method achieves 63% strong scaling efficiency and 1.6 ExaFLOP sustained performance. We further scale to 20 billion spatiotemporal tokens, enabling km-scale global modeling over 177k temporal frames, regimes previously unreachable, establishing a new paradigm for Earth system prediction.

Recent advances in language models have substantially improved Natural Language Understanding (NLU). Although widely used benchmarks suggest that Large Language Models (LLMs) can effectively disambiguate, their practical applicability in real-world narrative contexts remains underexplored. SemEval-2026 Task 5 addresses this gap by introducing a task that predicts the human-perceived plausibility of a word sense within a short story. In this work, we propose an LLM-based framework for plausibility scoring of homonymous word senses in narrative texts using a structured reasoning mechanism. We examine the impact of fine-tuning low-parameter LLMs with diverse reasoning strategies, alongside dynamic few-shot prompting for large-parameter models, on accurate sense identification and plausibility estimation. Our results show that commercial large-parameter LLMs with dynamic few-shot prompting closely replicate human-like plausibility judgments. Furthermore, model ensembling slightly improves performance, better simulating the agreement patterns of five human annotators compared to single-model predictions

Frontier models have demonstrated exceptional capabilities following the integration of task-reward-based reinforcement learning (RL) into their training pipelines, enabling systems to evolve from pure reasoning models into sophisticated agents. However, debate persists regarding whether RL genuinely instills new skills within a base model or merely sharpens its existing distribution to elicit latent capabilities. To address this dichotomy, we present an explicit comparison between distribution sharpening and task-reward-based learning, utilizing RL as a tool to implement both paradigms. Our analysis reveals the inherent limitations of distribution sharpening, demonstrating from first principles how and why the optima can be unfavorable and the approach fundamentally unstable. Furthermore, our experiments using Llama-3.2-3B-Instruct, Qwen2.5-3B-Instruct and Qwen3-4B-Instruct-2507 on math datasets confirm that sharpening yields limited gains, whereas incorporating task-based reward signal can greatly help achieve robust performance improvements and stable learning.

Competency Questions (CQs) are a cornerstone of requirement elicitation in ontology engineering. CQs represent requirements as a set of natural language questions that an ontology should satisfy; they are traditionally modelled by ontology engineers together with domain experts as part of a human-centred, manual elicitation process. The use of Generative AI automates CQ creation at scale, therefore democratising the process of generation, widening stakeholder engagement, and ultimately broadening access to ontology engineering. However, given the large and heterogeneous landscape of LLMs, varying in dimensions such as parameter scale, task and domain specialisation, and accessibility, it is crucial to characterise and understand the intrinsic, observable properties of the CQs they produce (e.g., readability, structural complexity) through a systematic, cross-domain analysis. This paper introduces a set of quantitative measures for the systematic comparison of CQs across multiple dimensions. Using CQs generated from well defined use cases and scenarios, we identify their salient properties, including readability, relevance with respect to the input text and structural complexity of the generated questions. We conduct our experiments over a set of use cases and requirements using a range of LLMs, including both open (KimiK2-1T, LLama3.1-8B, LLama3.2-3B) and closed models (Gemini 2.5 Pro, GPT 4.1). Our analysis demonstrates that LLM performance reflects distinct generation profiles shaped by the use case.

Reasoning in vision-language models (VLMs) has recently attracted significant attention due to its broad applicability across diverse downstream tasks. However, it remains unclear whether the superior performance of VLMs stems from genuine vision-grounded reasoning or relies predominantly on the reasoning capabilities of their textual backbones. To systematically measure this, we introduce CrossMath, a novel multimodal reasoning benchmark designed for controlled cross-modal comparisons. Specifically, we construct each problem in text-only, image-only, and image+text formats guaranteeing identical task-relevant information, verified by human annotators. This rigorous alignment effectively isolates modality-specific reasoning differences while eliminating confounding factors such as information mismatch. Extensive evaluation of state-of-the-art VLMs reveals a consistent phenomenon: a substantial performance gap between textual and visual reasoning. Notably, VLMs excel with text-only inputs, whereas incorporating visual data (image+text) frequently degrades performance compared to the text-only baseline. These findings indicate that current VLMs conduct reasoning primarily in the textual space, with limited genuine reliance on visual evidence. To mitigate this limitation, we curate a CrossMath training set for VLM fine-tuning. Empirical evaluations demonstrate that fine-tuning on this training set significantly boosts reasoning performance across all individual and joint modalities, while yielding robust gains on two general visual reasoning tasks. Source code is available at https://github.com/xuyige/CrossMath.

We propose HILBERT (HIerarchical Long-sequence Balanced Embedding with Reciprocal contrastive Training), a cross-attentive multimodal framework for learning document-level audio-text representations from long, segmented sequences in low-resource data settings. HILBERT leverages frozen pre-trained speech and language encoders to extract segment-level features, which are aggregated via cross-modal attention and self-attentive pooling to form modality-specific document representations and a joint cross-attentive embedding. To align modalities while preserving modality-specific structure under severe audio-text dimensional imbalance, we introduce a reciprocal dual contrastive objective that simultaneously aligns audio-to-joint and text-to-joint representations, rather than directly contrasting audio and text alone. Two auxiliary regularizers further stabilize long-sequence fusion: a Centered Kernel Alignment (CKA) loss that preserves structural consistency between each modality and the joint embedding, and a mutual information balancing loss that prevents dominance of a single modality by equalizing information flow from audio and text into the joint space. For downstream prediction, HILBERT employs a Mixture-of-Experts (MoE) classifier over concatenated audio, text, and joint representations to accommodate heterogeneous label regimes. Extensive evaluation across multiple audio-text backbone combinations demonstrates that HILBERT learns semantically meaningful long-sequence representations and achieves superior performance on highly imbalanced multi-class settings.

Reinforcement learning with verifiable rewards (RLVR) typically optimizes for outcome rewards without imposing constraints on intermediate reasoning. This leaves training susceptible to reward hacking, where models exploit loopholes (e.g., spurious patterns in training data) in the reward function to achieve high scores without solving the intended task. These reward-hacking behaviors are often implicit, as the intermediate chain-of-thought (CoT) may appear plausible on the surface, limiting the effectiveness of purely text-based monitoring. We propose Gradient Fingerprint (GRIFT), a method for detecting reward hacking using models' internal computations. Given a prompt and a model-generated CoT, GRIFT computes gradients of the CoT conditioned on the prompt and compresses them into a compact representation, which is then used to assess whether the CoT reflects reward hacking behavior. Across verifiable reasoning benchmarks spanning math, code, and logical reasoning, GRIFT substantially outperforms strong baselines, including CoT Monitor and TRACE, achieving over 25% relative improvement in detecting reward hacking behavior. Moreover, integrating GRIFT into the rejection fine-tuning pipeline for reasoning tasks reduces reward hacking and improves performance on the true task objective. Our results highlight a promising direction of leveraging gradient level representations for assessing the quality of CoT reasoning traces. Our code is available at: https://github.com/songtao-x/reward_hack.

Large language models have shown strong performance on broad-domain knowledge and reasoning benchmarks, but it remains unclear how well language models handle specialized animal-related knowledge under a unified closed-book evaluation protocol. We introduce BAGEL, a benchmark for evaluating animal knowledge expertise in language models. BAGEL is constructed from diverse scientific and reference sources, including bioRxiv, Global Biotic Interactions, Xeno-canto, and Wikipedia, using a combination of curated examples and automatically generated closed-book question-answer pairs. The benchmark covers multiple aspects of animal knowledge, including taxonomy, morphology, habitat, behavior, vocalization, geographic distribution, and species interactions. By focusing on closed-book evaluation, BAGEL measures animal-related knowledge of models without external retrieval at inference time. BAGEL further supports fine-grained analysis across source domains, taxonomic groups, and knowledge categories, enabling a more precise characterization of model strengths and systematic failure modes. Our benchmark provides a new testbed for studying domain-specific knowledge generalization in language models and for improving their reliability in biodiversity-related applications.

In multi-fidelity optimization, biased approximations of varying costs of the target function are available. This paper studies the problem of optimizing a locally smooth function with a limited budget, where the learner has to make a tradeoff between the cost and the bias of these approximations. We first prove lower bounds for the simple regret under different assumptions on the fidelities, based on a cost-to-bias function. We then present the Kometo algorithm which achieves, with additional logarithmic factors, the same rates without any knowledge of the function smoothness and fidelity assumptions, and improves previously proven guarantees. We finally empirically show that our algorithm outperforms previous multi-fidelity optimization methods without the knowledge of problem-dependent parameters.

Decision-makers rely on weather forecasts to plant crops, manage wildfires, allocate water and energy, and prepare for weather extremes. Today, such forecasts enjoy unprecedented accuracy out to two weeks thanks to steady advances in physics-based dynamical models and data-driven artificial intelligence (AI) models. However, model skill drops precipitously at subseasonal timescales (2 - 6 weeks ahead), due to compounding errors and persistent biases. To counter this degradation, we introduce probabilistic bias correction (PBC), a machine learning framework that substantially reduces systematic error by learning to correct historical probabilistic forecasts. When applied to the leading dynamical and AI models from the European Centre for Medium-Range Weather Forecasts (ECMWF), PBC doubles the subseasonal skill of the AI Forecasting System and improves the skill of the operationally-debiased dynamical model for 91% of pressure, 92% of temperature, and 98% of precipitation targets. We designed PBC for operational deployment, and, in ECMWF's 2025 real-time forecasting competition, its global forecasts placed first for all weather variables and lead times, outperforming the dynamical models from six operational forecasting centers, an international dynamical multi-model ensemble, ECMWF's AI Forecasting System, and the forecasting systems of 34 teams worldwide. These probabilistic skill gains translate into more accurate prediction of extreme events and have the potential to improve agricultural planning, energy management, and disaster preparedness in vulnerable communities.

This paper presents a systematic benchmark of state-of-the-art multilingual large language models (LLMs) adapted via token pruning - a compression technique that eliminates tokens and embedding parameters corresponding to languages irrelevant to the target application. Focusing on Korean-centric natural language processing (NLP) tasks, we evaluate architectures including Qwen3, Gemma-3, Llama-3, and Aya across three vocabulary configurations: Original, English-Korean (EnKo), and English-Korean-Chinese (EnKoZh). Performance is assessed using established benchmarks for general aptitude, cultural literacy, instruction following, and machine translation. Our findings indicate that token pruning significantly improves generation stability by eliminating language confusion, and in the case of machine translation, frequently enhances performance on Korean-specific tasks. While instruction-following capabilities display architecture-dependent variance linked to latent cross-lingual representations, the significant reduction in vocabulary size validates token pruning as a highly effective optimization strategy for memory-constrained, domain-specific deployments, despite modest gains in inference latency.

Academic integrity continues to face the persistent challenge of examination cheating. Traditional invigilation relies on human observation, which is inefficient, costly, and prone to errors at scale. Although some existing AI-powered monitoring systems have been deployed and trusted, many lack transparency or require multi-layered architectures to achieve the desired performance. To overcome these challenges, we propose an improvement over a simple two-stage framework for exam cheating detection that integrates object detection and behavioral analysis using well-known technologies. First, the state-of-the-art YOLOv8n model is used to localize students in exam-room images. Each detected region is cropped and preprocessed, then classified by a fine-tuned RexNet-150 model as either normal or cheating behavior. The system is trained on a dataset compiled from 10 independent sources with a total of 273,897 samples, achieving 0.95 accuracy, 0.94 recall, 0.96 precision, and 0.95 F1-score - a 13\% increase over a baseline accuracy of 0.82 in video-based cheating detection. In addition, with an average inference time of 13.9 ms per sample, the proposed approach demonstrates robustness and scalability for deployment in large-scale environments. Beyond the technical contribution, the AI-assisted monitoring system also addresses ethical concerns by ensuring that final outcomes are delivered privately to individual students after the examination, for example, via personal email. This prevents public exposure or shaming and offers students an opportunity to reflect on their behavior. For further improvement, it is possible to incorporate additional factors, such as audio data and consecutive frames, to achieve greater accuracy. This study provides a foundation for developing real-time, scalable, ethical, and open-source solutions.

Understanding natural and engineered systems often relies on symbolic formulations, such as differential equations, which provide interpretability and transferability beyond black-box models. We introduce Latent Grammar Flow (LGF), a neuro-symbolic generative framework for discovering ordinary differential equations from data. LGF embeds equations as grammar-based representations into a discrete latent space and forces semantically similar equations to be positioned closer together with a behavioural loss. Then, a discrete flow model guides the sampling process to recursively generate candidate equations that best fit the observed data. Domain knowledge and constraints, such as stability, can be either embedded into the rules or used as conditional predictors.

Structural health monitoring plays a critical role in ensuring structural safety by analyzing vibration responses from engineering systems. This paper proposes a Spectro-Temporal Alignment framework and a Hybrid Spectro-Temporal Fusion framework that integrate arrival-time interval descriptors with spectral features to capture both fine-scale and coarse-scale vibration dynamics. Experiments conducted on data collected from an LDS V406 electrodynamic shaker demonstrate that the proposed spectro-temporal representations significantly outperform conventional input formulations. The results indicate that a temporal resolution (Δτ) of 0.008 of 0.02 favors traditional machine learning models, whereas a finer resolution (Δτ) of 0.008 effectively unlocks the performance potential of deep learning architectures. Beyond classification accuracy, a comprehensive stability analysis based on condensed indices, including mean performance, standard deviation, coefficient of variation, and balanced score, shows that the proposed hybrid framework consistently achieves higher accuracy with substantially lower variability compared to baseline and alignment-only approaches. Overall, these results demonstrate that the proposed framework provides a robust, accurate, and reliable solution for vibration-based structural health monitoring.

Penalty kicks in soccer are decided under extreme time constraints, where goalkeepers benefit from anticipating shot direction from the kickers motion before or around ball contact. In this paper, MambaKick is presented as a learning-based framework for penalty direction prediction that leverages pretrained human action recognition (HAR) embeddings extracted from contact-centered short video segments and combines them with a lightweight temporal predictor. Rather than relying on explicit kinematic reconstruction or handcrafted biomechanical features, the approach reuses transferable spatiotemporal representations and utilizes selective state-spare models (Mamba) for efficient sequence aggregation. Simple contextual metadata (e.g., field side and footedness) are also considered as complementary cues that may reduce ambiguity in real-world footage. Across a range of HAR backbones, MambaKick consistently improves or matches strong embedding baselines, achieving up to 53.1% accuracy for three classes and 64.5% for two classes under the proposed methodology. Overall, the results indicate that combining pretrained HAR representations with efficient state-space temporal modeling is a practical direction for low-latency intention prediction in real-world sports video. The code will be available at GitHub: https://github.com/hvelesaca/MambaKick/

We present an amortized framework for real-time visual attribution streaming in multimodal thinking models. When these models generate code from a screenshot or solve math problems from images, their long reasoning traces should be grounded in visual evidence. However, verifying this reliance is challenging: faithful causal methods require costly repeated backward passes or perturbations, while raw attention maps offer instant access, they lack causal validity. To resolve this, we introduce an amortized approach that learns to estimate the causal effects of semantic regions directly from the rich signals encoded in attention features. Across five diverse benchmarks and four thinking models, our approach achieves faithfulness comparable to exhaustive causal methods while enabling visual attribution streaming, where users observe grounding evidence as the model reasons, not after. Our results demonstrate that real-time, faithful attribution in multimodal thinking models is achievable through lightweight learning, not brute-force computation.

Molecular property prediction integrates quantum chemistry, cheminformatics, and deep learning to connect molecular structure with physicochemical and biological behavior. This survey traces four complementary paradigms, including Quantum, Descriptor Machine Learning, Geometric Deep Learning, and Foundation Models, and outlines a unified taxonomy linking molecular representations, model architectures, and interdisciplinary applications. Benchmark analyses integrate evidence from both widely used datasets and datasets reflecting industry perspectives, encompassing quantum, physicochemical, physiological, and biophysical domains. The survey examines current standards in data curation, splitting strategies, and evaluation protocols, highlighting challenges including inconsistent stereochemistry, heterogeneous assay sources, and reproducibility limitations under random or poorly defined splits. These observations motivate the modernization of benchmark design toward more transparent, time- and scaffold-aware methodologies. We further propose three forward-looking directions: (i) physics-aware learning embedding quantum consistency, (ii) uncertainty-calibrated foundation models for trustworthy inference, and (iii) realistic multimodal benchmark ecosystems integrating computational and experimental data. Repository: https://github.com/Zongru-Li/Survey-and-Benchmarks-of-DL-for-Molecular-Property-Prediction-in-the-Foundation-Model-Era.

We present the Global Neural World Model (GNWM), a self-stabilizing framework that achieves topological quantization through balanced continuous entropy constraints. Operating as a continuous, action-conditioned Joint-Embedding Predictive Architecture (JEPA), the GNWM maps environments onto a discrete 2D grid, enforcing translational equivariance without pixel-level reconstruction. Our results show this architecture prevents manifold drift during autoregressive rollouts by using grid ``snapping'' as a native error-correction mechanism. Furthermore, by training via maximum entropy exploration (random walks), the model learns generalized transition dynamics rather than memorizing specific expert trajectories. We validate the GNWM across passive observation, active agent control, and abstract sequence regimes, demonstrating its capacity to act not just as a spatial physics simulator, but as a causal discovery model capable of organizing continuous, predictable concepts into structured topological maps.

Certified program synthesis (aka vericoding) is the process of automatically generating a program, its formal specification, and a machine-checkable proof of their alignment from a natural-language description. Two challenges make vericoding difficult. First, specifications synthesised from natural language are often either too weak to be meaningful or too strong to be implementable, yet existing approaches lack systematic means to detect such defects. Second, the landscape of program verifiers is fragmented: each tool supports a particular reasoning mode -- auto-active (e.g., Dafny, Verus) or interactive (e.g., Coq, Lean) -- with its own trade-off between automation and expressivity. This forces every synthesis methodology to be tailored to a single verification paradigm, limiting the class of tasks it can handle effectively. We overcome both challenges by structuring the certified synthesis workflow around a multi-modal verifier -- a single tool combining dynamic validation, automated proofs, and interactive proof scripting in one foundational framework. We realise this idea in LeetProof, an agentic pipeline built on Velvet, a multi-modal verifier embedded in Lean. Multi-modality enables LeetProof to validate generated specifications via randomised property-based testing before any code is synthesised, decompose the synthesis task into sub-problems guided by verification conditions, and delegate residual proof obligations to frontier AI provers specialised for Lean. We evaluate LeetProof on benchmarks derived from prior work on certified synthesis. Our specification validation uncovers defects in existing reference benchmarks, and LeetProof's staged pipeline achieves a significantly higher rate of fully certified solutions than a single-mode baseline at the same budget -- consistently across two frontier LLM backends.

Edge deployment of large language models (LLMs) increasingly relies on libraries of lightweight LoRA adapters, yet GPU/DRAM can keep only a small resident subset at a time. Serving a request through a non-resident adapter requires paging its weights from storage, incurring measurable latency. This creates a two-timescale online control problem: on a slow timescale, the system selects which adapters remain resident in fast memory, while on a fast timescale it routes each request to an adapter whose context-dependent utility is unknown a priori. The two decisions are tightly coupled: the cache determines the cost of exploration, and the router determines which adapters receive informative feedback. We formulate this joint caching-and-routing problem as a two-timescale contextual bandit and propose POLAR (Paging and Online Learning for Adapter Routing). POLAR pairs a cache-aware LinUCB router with an epoch-based cache controller. We study two variants. A fixed-epoch version provides a robust baseline with worst-case regret guarantees under arbitrary contexts. An epoch-doubling version, POLAR+, adds forced exploration and improved cache optimization to achieve $\widetilde{\mathcal{O}}(d\sqrt{NT}+\sqrt{KT})$ sublinear regret under stochastic regularity and cacheability conditions, where $N$ is the adapter count, $K$ the cache size, $d$ the context dimension, and $T$ the horizon. The routing term matches the standard contextual-bandit rate up to logarithmic factors, showing that the memory hierarchy does not fundamentally slow routing learning. Experiments using 15 real LoRA adapters for Qwen2.5-7B together with measured GPU paging latencies show that adaptive cache control substantially outperforms non-adaptive baselines and exhibits scaling trends consistent with the theory.

Pedestrian detection is fundamental to autonomous driving, robotics, and surveillance. Despite progress in deep learning, reliable identification remains challenging due to occlusions, cluttered backgrounds, and degraded visibility. While multispectral detection-combining visible and thermal sensors-mitigates poor visibility, the challenge of camouflaged pedestrians remains largely unexplored. Existing Camouflaged Object Detection (COD) benchmarks focus on biological species, leaving a gap in safety-critical human detection where targets blend into their surroundings. To address this, we introduce Camo-M3FD (derived from the M3FD dataset), a novel benchmark for cross-spectral camouflaged pedestrian detection, consisting of registered visible-thermal image pairs. The dataset is curated using quantitative metrics to ensure high foreground-background similarity. We provide high-quality pixel-level masks and establish a standardized evaluation framework using state-of-the-art COD models. Our results demonstrate that while thermal signals provide indispensable localization cues, multispectral fusion is essential for refining structural details. Camo-M3FD serves as a foundational resource for developing robust and safety-critical detection systems. The dataset is available on GitHub: https://cod-espol.github.io/Camo-M3FD/

Neural Combinatorial Optimization (NCO) has emerged as a powerful framework for solving combinatorial optimization problems by integrating deep learning-based models. This work focuses on improving existing inference techniques to enhance solution quality and generalization. Specifically, we modify the Random Re-Construct (RRC) approach of the Light Encoder Heavy Decoder (LEHD) model by incorporating Simulated Annealing (SA). Unlike the conventional RRC, which greedily replaces suboptimal segments, our SA-based modification introduces a probabilistic acceptance mechanism that allows the model to escape local optima and explore a more diverse solution space. Additionally, we enhance the Policy Optimization with Multiple Optima (POMO) approach by integrating Beam Search, enabling systematic exploration of multiple promising solutions while maintaining diversity in the search space. We further investigate different inference strategies, including Softmax Sampling, Greedy, Gumbel-Softmax, and Epsilon-Greedy, analyzing their impact on solution quality. Furthermore, we explore instance augmentation techniques, such as horizontal and vertical flipping and rotation-based augmentations, to improve model generalization across different CVRP instances. Our extensive experiments demonstrate that these modifications significantly reduce the optimality gap across various Capacitated Vehicle Routing Problem (CVRP) benchmarks, with Beam Search and SA-based RRC consistently yielding superior performance. By refining inference techniques and leveraging enhanced search strategies, our work contributes to the broader applicability of NCO models in real-world combinatorial optimization tasks.

Accurate assessment of lithium-ion battery ageing is challenged by cell-to-cell variability, heterogeneous cycling protocols, and limited transferability of data-driven models across datasets. In particular, robust identification of degradation transitions, such as the knee point, and reliable early-life prediction of remaining useful life (RUL) remain open problems. This study proposes a unified framework for battery ageing analysis based on continuous representations of voltage-capacity and capacity-cycle trajectories learned from heterogeneous public datasets (NASA, CALCE, ISU-ILCC). The continuous formulation enables consistent extraction of degradation descriptors, including curvature, plateau length and knee-related metrics, while reducing sensitivity to dataset-specific discretisation. Across more than 250 cells, statistically significant correlations between knee onset and end-of-life (Pearson 0.75-0.84) are observed. Additional early-life analysis confirms that knee-related features retain predictive value when estimated from partial trajectories. Early-life models provide increasingly stable RUL predictions as the number of observed cycles increases, with meaningful predictive performance emerging within the first 5-20 cycles and remain robust under cross-dataset domain shift. The framework integrates continuous modelling, feature extraction and uncertainty-aware prediction, providing an interpretable and dataset-consistent approach demonstrating robustness across heterogeneous dataset types. Compared with conventional discrete or feature-based methods, the proposed representation reduces sensitivity to sampling resolution and improves cross-dataset consistency. The study is limited to laboratory-scale datasets and capacity-based end-of-life definitions.

Automated depression estimation is highly vulnerable to signal corruption and ambient noise in real-world deployment. Prevailing deterministic methods produce uncalibrated point estimates, exposing safety-critical clinical systems to the severe risk of overconfident misdiagnoses. To establish a highly resilient and trustworthy assessment paradigm, we propose EviDep, an evidential learning framework that jointly quantifies depression severity alongside aleatoric and epistemic uncertainties via a Normal-Inverse-Gamma distribution. A fundamental vulnerability in multimodal evidential fusion is the uncontrolled accumulation of cross-modal redundancies. This structural flaw artificially inflates diagnostic confidence by double-counting overlapping evidence. To guarantee robust evidence synthesis, EviDep enforces strict information integrity. First, a Frequency-aware Feature Extraction module leverages a wavelet-based Mixture-of-Experts to dynamically isolate task-irrelevant noise, preserving the fidelity of diagnostic signals. Subsequently, a Disentangled Evidential Learning strategy separates the shared consensus from modality-specific nuances. By explicitly decoupling these representations before Bayesian fusion, EviDep systematically mitigates evidence redundancy. Extensive experiments on AVEC 2013, 2014, DAIC-WOZ, and E-DAIC confirm that EviDep achieves state-of-the-art predictive accuracy and superior uncertainty calibration, delivering a robust fail-safe mechanism for trustworthy clinical screening.

Human activity recognition (HAR) refers to the process of identifying human actions and activities using data collected from sensors. Neural networks, such as convolutional neural networks (CNNs), long short-term memory (LSTM) networks, convolutional LSTM, and their hybrid combinations, have demonstrated exceptional performance in various research domains. Developing a multilevel individual or hybrid model for HAR involves strategically integrating multiple networks to capitalize on their complementary strengths. The structural arrangement of these components is a critical factor influencing the overall performance. This study explores a novel framework of a two-level network architecture with dual-stage feature fusion: late fusion, which combines the outputs from the first network level, and intermediate fusion, which integrates the features from both the first and second levels. We evaluated $15$ different network architectures of CNNs, LSTMs, and convolutional LSTMs, incorporating late fusion with and without intermediate fusion, to identify the optimal configuration. Experimental evaluation on two public benchmark datasets demonstrates that architectures incorporating both late and intermediate fusion achieve higher accuracy than those relying on late fusion alone. Moreover, the optimal configuration outperforms baseline models, thereby validating its effectiveness for HAR.

Decoder-only large language models (LLMs) are increasingly replacing BERT-style architectures as the backbone for dense retrieval, achieving substantial performance gains and broad adoption. However, the robustness of these LLM-based retrievers remains underexplored. In this paper, we present the first systematic study of the robustness of state-of-the-art open-source LLM-based dense retrievers from two complementary perspectives: generalizability and stability. For generalizability, we evaluate retrieval effectiveness across four benchmarks spanning 30 datasets, using linear mixed-effects models to estimate marginal mean performance and disentangle intrinsic model capability from dataset heterogeneity. Our analysis reveals that while instruction-tuned models generally excel, those optimized for complex reasoning often suffer a ``specialization tax,'' exhibiting limited generalizability in broader contexts. For stability, we assess model resilience against both unintentional query variations~(e.g., paraphrasing, typos) and malicious adversarial attacks~(e.g., corpus poisoning). We find that LLM-based retrievers show improved robustness against typos and corpus poisoning compared to encoder-only baselines, yet remain vulnerable to semantic perturbations like synonymizing. Further analysis shows that embedding geometry (e.g., angular uniformity) provides predictive signals for lexical stability and suggests that scaling model size generally improves robustness. These findings inform future robustness-aware retriever design and principled benchmarking. Our code is publicly available at https://github.com/liyongkang123/Robust_LLM_Retriever_Eval.

Unsupervised anomaly detection is widely used to detect Distributed Denial-of-Service (DDoS) attacks in cloud-native 5G networks, yet most studies assume a fixed traffic representation, either temporal or structural, without validating which feature space best matches the data. We propose a lightweight decision framework that prioritizes temporal or structural features before training, using two diagnostics: lag-1 autocorrelation of an aggregated flow signal and PCA cumulative explained variance. When the probes are inconclusive, the framework reserves a hybrid option as a future fallback rather than an empirically validated branch. Experiments on two statistically distinct datasets with Isolation Forest, One-Class SVM, and KMeans show that structural features consistently match or outperform temporal ones, with the performance gap widening as temporal dependence weakens.

Federated Learning (FL) faces challenges from client data heterogeneity and resource-constrained mobile devices, which can degrade model accuracy. Personalized Federated Learning (PFL) addresses this issue by adapting shared global knowledge to local data distributions. A promising approach in PFL is model decoupling, which separates the model into global and personalized parameters, raising the key question of which parameters should be personalized to balance global knowledge sharing and local adaptation. In this paper, we propose a Federated Optimal Brain Personalization (FedOBP) algorithm with a quantile-based thresholding mechanism and introduce an element-wise importance score. This score extends Optimal Brain Damage (OBD) pruning theory by incorporating a federated approximation of the first-order derivative in the Taylor expansion to evaluate the importance of each parameter for personalization. Moreover, we move the metric computation originally performed on clients to the server side, to alleviate the burden on resource-constrained mobile devices. To the best of our knowledge, this is the first work to bridge classical saliency-based pruning theory with federated parameter decoupling, providing a rigorous theoretical justification for selecting personalized parameters based on their sensitivity to local loss landscapes. Extensive experiments demonstrate that FedOBP outperforms state-of-the-art methods across diverse datasets and heterogeneity scenarios, while requiring personalization of only a very small number of personalized parameters.

Wi-Fi Channel State Information (CSI) enables device-free human activity recognition, but existing multi-user approaches assume a fixed set of known users during both training and inference. This closed-set assumption limits deployment, as models trained on a specific user set degrade when applied to new individuals or environments. We reformulate multi-user activity recognition as activity counting, estimating how many users perform each activity type at a given time, without associating actions with specific individuals. We propose a pipeline that converts CSI measurements into spatial projections and extracts features using a pretrained convolutional backbone. Two formulations are evaluated on the WiMANS dataset: a conventional identity-dependent model that assigns activities to fixed user slots, and an identity-agnostic model that estimates scene-level activity composition through regression. Under standard evaluation, the identity-agnostic model achieves a mean absolute error of 0.1081 on a 0-5 count scale. Under unseen-user evaluation, the identity-dependent model's macro-F1 drops from 80.38 to 32.61, while the identity-agnostic model's counting error remains stable. Feature space analysis confirms that identity-agnostic representations are more user-invariant, which explains their stronger generalization. These results suggest that activity counting provides a more practical and generalizable alternative to identity-dependent formulations for multi-user WiFi sensing.

Ensuring functional consistency between high-level algorithmic models and low-level hardware implementations is a critical challenge, particularly as modern design flows increasingly span heterogeneous abstractions--from deep learning frameworks to hardware netlists. In this paper, we present EquivFusion, an end-to-end equivalence checking tool tailored for multi-modal circuit designs. Unlike traditional flows that rely on siloed tools or ad-hoc translation, EquivFusion leverages a verification-oriented MLIR lowering pipeline to unify diverse entry points, including PyTorch, C/C++, Chisel, Verilog, and gate-level netlists, into a common intermediate representation. This architecture enables automated, pairwise equivalence checking across diverse abstraction levels by rigorously translating designs into standard formal verification formats, i.e., SMT-LIB, BTOR2, AIGER. We demonstrate EquivFusion's feasibility to bridge the semantic gap between software specifications and hardware realizations, showcasing its effectiveness in facilitating "shift-left" formal verification for datapath-intensive hardware designs.

DNA sequence encoding is fundamental to gene function prediction, protein synthesis, and diverse downstream biological tasks. Despite the substantial progress achieved by large-scale DNA sequence pretraining, existing studies have overwhelmingly emphasized pretraining scale and custom downstream evaluation datasets, while neglecting some essential components of the pretraining paradigm. In this paper, we reveal three critical yet heretofore overlooked problems in DNA pretraining: inappropriate downstream datasets, inherent flaws in the neighbor-masking strategy, and the lack of detailed discussion on vocabulary. Therefore, we undertake comprehensive investigations and propose principled guidelines, including selection criteria for evaluation datasets, guiding task design, and in-depth vocabulary analysis. Extensive experiments validate the significance of our identified problems and support the rationale behind our recommendations. Finally, we introduce a standardized testbed that enables reproducible and rigorous benchmarking of DNA pretraining methods to advance the development of genomic foundation models.

Agentic Artificial Intelligence (AI) represents a paradigm shift from reactive systems to proactive, autonomous decision making frameworks. Existing AI-based educational systems remain fragmented and lack multi-level integration across stakeholders. This paper proposes the Agentic Unified Student Support System (AUSS), a novel multi-agent architecture integrating student-level personalization, educator-level automation, and institutional-level intelligence. The framework leverages Large Language Models (LLMs), reinforcement learning, predictive analytics, and rule-based reasoning. Experimental results demonstrate improvements in recommendation accuracy (92.4%), grading efficiency (94.1%), and dropout prediction (F1-score: 89.5%). The proposed system enables scalable, adaptive, and intelligent educational ecosystems.

While Diffusion Large Language Models (dLLMs) offer structural advantages for global planning, efficiently verifying that they arrive at correct answers via valid reasoning traces remains a critical challenge. In this work, we propose a geometric perspective: Reasoning on the Manifold. We hypothesize that valid generation trajectories reside as stable attractors on the high-density manifold of the learned distribution, whereas invalid paths exhibit off-manifold drift. To operationalize this, we introduce Bidirectional Manifold Consistency (BMC), a training-free, unsupervised metric that quantifies the stability of the generated sequence through a forward-masking and backward-reconstruction cycle. Empirically, we demonstrate BMC's versatility across the full reasoning lifecycle: (1) in Diagnosis, it serves as a robust discriminator of solution validity without ground truth answer; (2) in Inference, it enables rejection resampling to effectively concentrate computational resources on complex reasoning tasks; and (3) in Alignment, it functions as a dense geometric reward that transforms sparse outcome supervision into fine-grained guidance, empowering models to self-evolve beyond standard baselines. Our results establish intrinsic geometric stability as a robust indicator of correctness for dLLMs.

Systolic murmurs are extra heart sounds that occur during the contraction phase of the cardiac cycle, often indicating heart abnormalities caused by turbulent blood flow. Their intensity, pitch, and quality vary, requiring precise identification for the accurate diagnosis of cardiac disorders. This study presents an automatic classification system for systolic murmurs using a feature extraction module, followed by a classification model. The feature extraction module employs complex orthogonal matching pursuit to project single or multiple murmur segments onto a redundant dictionary composed of multiresolution complex Gabor basis functions (GBFs). The resulting projection weights are split and reshaped into variable-resolution time--frequency feature matrices. Processing multiple segments of a single recording using a shared dictionary mitigates murmur variability. This is achieved by learning the weights for each segment while enforcing that they correspond to the same set of basis functions in the dictionary, promoting consistent time--frequency feature matrices. The classification model is built based on a vision transformer to process multiple input matrices of different resolutions by passing each through a convolutional neural network for patch tokenization. All embedding tokens are then concatenated to form a matrix and forwarded to an encoder layer that includes multihead attention, residual connections, and a convolutional network with a kernel size of one. This integration of multiresolution feature extraction with transformer-based feature classification enhances the accuracy and reliability of heart murmur identification. An experimental analysis of four types of systolic murmurs from the CirCor DigiScope dataset demonstrates the effectiveness of the system, achieving a classification accuracy of $95.96\%$.

Generalizable gaze estimation methods have garnered increasing attention due to their critical importance in real-world applications and have achieved significant progress. However, they often overlook the effect of label noise, arising from the inherent difficulty of acquiring precise gaze annotations, on model generalization performance. In this paper, we are the first to comprehensively investigate the negative effects of label noise on generalization in gaze estimation. Further, we propose a novel solution, called See-Through-Noise (SeeTN) framework, which improves generalization from a novel perspective of mitigating label noise. Specifically, we propose to construct a semantic embedding space via a prototype-based transformation to preserve a consistent topological structure between gaze features and continuous labels. We then measure feature-label affinity consistency to distinguish noisy from clean samples, and introduce a novel affinity regularization in the semantic manifold to transfer gaze-related information from clean to noisy samples. Our proposed SeeTN promotes semantic structure alignment and enforces domain-invariant gaze relationships, thereby enhancing robustness against label noise. Extensive experiments demonstrate that our SeeTN effectively mitigates the adverse impact of source-domain noise, leading to superior cross-domain generalization without compromising the source-domain accuracy, and highlight the importance of explicitly handling noise in generalized gaze estimation.

Automatically generating formal specifications could reduce the effort needed to improve program correctness, but in practice, this is still challenging. Many developers avoid writing contracts by hand, which limits the use of automated verification tools. Recent large language models (LLMs) can generate specifications from code, but these specifications often fail in terms of verification. The reason is syntax errors, overly strict constraints, or mismatches with program behavior. We present SpecPylot, a Python tool that synthesizes executable specifications for Python programs as icontract annotations and checks them using crosshair's symbolic execution. The tool relies on LLMs to propose candidate contracts and uses crosshair to validate them. When crosshair finds a concrete counterexample, SpecPylot updates only the generated contracts and leaves the program itself untouched. In addition, the tool can produce coverage-driven pytest stubs and keep detailed execution artifacts that are useful during debugging. Overall, the evaluation indicates that SpecPylot is able to generate crosshair-compatible contracts for most programs, but it also highlights the practical limits introduced by bounded symbolic exploration and differences in LLM behavior.

AIGC has shown remarkable success in CV and NLP, and has recently demonstrated promising potential in the wireless domain. However, significant data imbalance exists across RF modalities, with abundant WiFi data but scarce mmWave and RFID data due to high acquisition cost. This makes it difficult to train high-quality generative models for these data-scarce modalities. In this work, we propose RF-CMG, a diffusion-based cross-modal generative method that leverages data-rich WiFi signals to synthesize high-fidelity RF data for scarce modalities including mmWave and RFID. The key insight of RF-CMG is to decouple cross-modal generation into high-frequency guidance and low-frequency constraint, which respectively learn high-frequency distribution from limited target modality data and preserve the underlying physical structure via low-frequency constraints during generation. On this basis, we introduce a Modality-Guided Embedding (MGE) module to steer the reverse diffusion trajectory toward the target high-frequency distribution, and a Low-Frequency Modality Consistency (LFMC) module to progressively enforce low-frequency constraints to suppress the accumulation of source-modality structural biases during inference, enabling high-quality target-modality generation. Performance comparison with several prevalent generative models demonstrates that RF-CMG achieves superior performance in synthesizing RFID and mmWave signals. We further showcase the effectiveness of the data generated by RF-CMG in gesture recognition tasks, and analyze the impact of the proportion of synthetic data on downstream performance.

Current post-training methodologies for adapting Large Vision-Language Models (LVLMs) generally fall into two paradigms: Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL). Despite their prevalence, both approaches suffer from inefficiencies when applied in isolation. SFT forces the model's generation along a single expert trajectory, often inducing catastrophic forgetting of general multimodal capabilities due to distributional shifts. Conversely, RL explores multiple generated trajectories but frequently encounters optimization collapse - a cold-start problem where an unaligned model fails to spontaneously sample any domain-valid trajectories in sparse-reward visual tasks. In this paper, we propose Supervised Group Relative Policy Optimization (S-GRPO), a unified post-training framework that integrates the guidance of imitation learning into the multi-trajectory exploration of preference optimization. Tailored for direct-generation visual tasks, S-GRPO introduces Conditional Ground-Truth Trajectory Injection (CGI). When a binary verifier detects a complete exploratory failure within a sampled group of trajectories, CGI injects the verified ground-truth trajectory into the candidate pool. By assigning a deterministic maximal reward to this injected anchor, S-GRPO enforces a positive signal within the group-relative advantage estimation. This mechanism reformulates the supervised learning objective as a high-advantage component of the policy gradient, compelling the model to dynamically balance between exploiting the expert trajectory and exploring novel visual concepts. Theoretical analysis and empirical results demonstrate that S-GRPO gracefully bridges the gap between SFT and RL, drastically accelerates convergence, and achieves superior domain adaptation while preserving the base model's general-purpose capabilities.

Neural Architecture Search (NAS) aims to automatically discover high-performing deep neural network (DNN) architectures. However, conventional algorithm-driven NAS relies on carefully hand-crafted search spaces to ensure executability, which restricts open-ended exploration. Recent coding-based agentic approaches using large language models (LLMs) reduce manual design, but current LLMs struggle to reliably generate complex, valid architectures, and their proposals are often biased toward a narrow set of patterns observed in their training data. To bridge reliable algorithmic search with powerful LLM assistance, we propose LLMasTool, a hierarchical tree-based NAS framework for stable and open-ended model evolution. Our method automatically extracts reusable modules from arbitrary source code and represents full architectures as hierarchical trees, enabling evolution through reliable tree transformations rather than code generation. At each evolution step, coarse-level planning is governed by a diversity-guided algorithm that leverages Bayesian modeling to improve exploration efficiency, while the LLM resolves the remaining degrees of freedom to ensure a meaningful evolutionary trajectory and an executable generated architecture. With this formulation, instead of fully agentic LLM approaches, our method explores diverse directions beyond the inherent biases in the LLM. Our method improves over existing NAS methods by 0.69, 1.83, and 2.68 points on CIFAR-10, CIFAR-100, and ImageNet16-120, demonstrating its effectiveness.

Stroke patient cross-subject electroencephalography (EEG) decoding of motor imagery (MI) brain-computer interface (BCI) is essential for motor rehabilitation, yet lesion-related abnormal temporal dynamics and pronounced inter-patient heterogeneity often undermine generalization. Existing adaptation methods are easily misled by pathological slow-wave activity and unstable target-domain pseudo-labels. To address this challenge, we propose PA-TCNet, a pathology-aware temporal calibration framework with physiology-guided target refinement for stroke motor imagery decoding. PA-TCNet integrates two coordinated components. The Pathology-aware Rhythmic State Mamba (PRSM) module decomposes EEG spatiotemporal features into slowly varying rhythmic context and fast transient perturbations, injecting the fused pathological context into selective state propagation to more effectively capture abnormal temporal dynamics. The Physiology-Guided Target Calibration (PGTC) module constructs source-domain sensorimotor region-of-interest templates, imposing physiological consistency constraints and dynamically refining target-domain pseudo-labels, thereby improving adaptation reliability. Leave-one-subject-out experiments on two independent stroke EEG datasets, XW-Stroke and 2019-Stroke, yielded mean accuracies of 66.56\% and 72.75\%, respectively, outperforming state-of-the-art baselines. These results indicate that jointly modeling pathological temporal dynamics and physiology-constrained pseudo-supervision can provide more robust cross-subject initialization for personalized post-stroke MI-BCI rehabilitation. The implemented code is available at https://github.com/wxk1224/PA-TCNet.

Recent text-to-scene generation approaches largely reduced the manual efforts required to create 3D scenes. However, their focus is either to generate a scene layout or to generate objects, and few generate both. The generated scene layout is often simple even with LLM's help. Moreover, the generated scene is often inconsistent with the text input that contains non-trivial descriptions of the shape, appearance, and spatial arrangement of the objects. We present a new paradigm of sequential text-to-scene generation and propose a novel generative model for interactive scene creation. At the core is a 3D Autoregressive Diffusion model 3D-ARD+, which unifies the autoregressive generation over a multimodal token sequence and diffusion generation of next-object 3D latents. To generate the next object, the model uses one autoregressive step to generate the coarse-grained 3D latents in the scene space, conditioned on both the current seen text instructions and already synthesized 3D scene. It then uses a second step to generate the 3D latents in the smaller object space, which can be decoded into fine-grained object geometry and appearance. We curate a large dataset of 230K indoor scenes with paired text instructions for training. We evaluate 7B 3D-ARD+, on challenging scenes, and showcase the model can generate and place objects following non-trivial spatial layout and semantics prescribed by the text instructions.

Despite the high accuracy of 'black box' deep learning models, drug discovery still relies on protein-ligand interaction principles and heuristics. To improve interpretability of protein-small molecule binding predictions, we developed the PWRules framework, which applies binding affinity data to identify privileged small molecule fragments and subsequently defines complementary pairing rules between these fragments and protein words (semantic sequence units) through an interpretability module. The resulting word-fragment rules are then ranked by the PWScore function to prioritize active compounds. Evaluations on benchmark datasets show that PWScore achieves competitive performance comparable to the physics-based model (Glide) and the deep learning model (PSICHIC) and shows broad applicability for protein targets outside the training dataset, e.g., SARS-CoV-2 main protease. Notably, PWScore captures complementary interaction information, yielding superior enrichment performance when integrated with these established methods. Structural analysis of protein-ligand complexes indicates that learned word-fragment rules are significantly enriched near ligand-binding pockets, despite training without explicit structural guidance. By extracting and applying complementary pairing rules, PWRules provides an interpretable framework for drug discovery.

Research on large language model (LLM) security is shifting from "will the model leak training data" to a more consequential question: can an agent with persistent, long-term memory be continuously shaped, cross-session poisoned, accessed without authorization, and propagated across shared organizational state? Recent surveys cover memory architectures and agent mechanisms, but fewer center the epistemic and governance properties of persistent, writable memory as the reason memory is an independent security problem. This survey addresses that gap. Drawing on cognitive neuroscience and the philosophy of memory, we characterize agent memory as malleable, rewritable, and socially propagating, and develop a memory-lifecycle framework organized around six phases -- Write, Store, Retrieve, Execute, Share, Forget/Rollback -- cross-tabulated against four security objectives: integrity, confidentiality, availability, governance. We organize the literature on memory poisoning, extraction, retrieval corruption, control-flow hijacking, cross-agent propagation, rollback, and governance, and situate representative architectures as determinants of which phases are explicitly governable. Three findings stand out: the literature concentrates on write- and retrieve-time integrity attacks, while confidentiality, availability, store/forget, and benign-persistence failures remain sparsely studied; no published architecture covers all nine governance primitives we identify; and using LLMs themselves for memory security remains sparse yet essential. We unify these under mnemonic sovereignty -- verifiable, recoverable governance over what may be written, who may read, when updates are authorized, and which states may be forgotten -- arguing future secure agents will be differentiated not only by recall capacity, but by memory governance quality.

Experimental evidence indicates that intrinsic temporal dynamics operating across multiple time scales are closely associated with the emergence of periodic spatial activity of increasing complexity. However, how information encoded in grid-like firing patterns for path integration is processed across these intrinsic time scales remains unclear. To address this question, we introduce adaptive time scales through a leak term in recurrent neural networks (RNNs), forming leaky RNNs discretized from the continuous attractors of firing rate models. Our results demonstrate that leaky RNNs substantially enhance the emergence of well-defined and highly regular hexagonal firing patterns. Compared with vanilla RNNs lacking a leak term, the trained leaky RNNs produce more accurate position estimates while generating reliable grid-cell-like representations. Furthermore, under identical noise conditions, leaky RNNs consistently exhibit more stable dynamics and better-defined grid structures. The learned dynamics also give rise to stable torus attractors with a clear central hole, supporting robust and regular grid-like activity. Overall, the dynamic leak acts as a low-pass filtering mechanism that protects recurrent neural circuitry from noise, stabilizes network dynamics, and improves path-integration accuracy in recurrent neural networks.

Most LLM safety work studies single-agent models, but many real applications rely on multiple interacting agents. In these systems, prompt segmentation and inter-agent routing create attack surfaces that single-agent evaluations miss. We study \emph{conjunctive prompt attacks}, where a trigger key in the user query and a hidden adversarial template in one compromised remote agent each appear benign alone but activate harmful behavior when routing brings them together. We consider an attacker who changes neither model weights nor the client agent and instead controls only trigger placement and template insertion. Across star, chain, and DAG topologies, routing-aware optimization substantially increases attack success over non-optimized baselines while keeping false activations low. Existing defenses, including PromptGuard, Llama-Guard variants, and system-level controls such as tool restrictions, do not reliably stop the attack because no single component appears malicious in isolation. These results expose a structural vulnerability in agentic LLM pipelines and motivate defenses that reason over routing and cross-agent composition. Code is available at https://github.com/UCF-ML-Research/ConjunctiveAgents.

Safety guardrails have become an active area of research in AI safety, aimed at ensuring the appropriate behavior of large language models (LLMs). However, existing research lacks consideration of nuances across linguistic and cultural contexts, resulting in a gap between reported performance and in-the-wild effectiveness. To address this issue, this paper proposes an approach to optimize guardrail models for a designated linguistic context by leveraging a curated dataset tailored to local linguistic characteristics, targeting the Taiwan linguistic context as a representative example of localized deployment challenges. The proposed approach yields TWGuard, a linguistic context-optimized guardrail model that achieves a huge gain (+0.289 in F1) compared to the foundation model and significantly outperforms the strongest baseline in practical use (-0.037 in false positive rate, a 94.9\% reduction). Together, this work lays a foundation for regional communities to establish AI safety standards grounded in their own linguistic contexts, rather than accepting boundaries imposed by dominant languages. The inadequacy of the latter is reconfirmed by our findings.

Poisoning input views of 3D reconstruction systems has been recently studied. However, we identify that existing studies simply backpropagate adversarial gradients through the 3D reconstruction pipeline as a whole, without uncovering the new vulnerability rooted in specific modules of the 3D reconstruction pipeline. In this paper, we argue that the structure-from-motion (SfM) initialization, as the geometric core of many widely used reconstruction systems, can be targeted to achieve transferable poisoning effects across diverse 3D reconstruction systems. To this end, we propose PoInit-of-View, which optimizes adversarial perturbations to intentionally introduce cross-view gradient inconsistencies at projections of corresponding 3D points. These inconsistencies disrupt keypoint detection and feature matching, thereby corrupting pose estimation and triangulation within SfM, eventually resulting in low-quality rendered views. We also provide a theoretical analysis that connects cross-view inconsistency to correspondence collapse. Experimental results demonstrate the effectiveness of our PoInit-of-View on diverse 3D reconstruction systems and datasets, surpassing the single-view baseline by 25.1% in PSNR and 16.5% in SSIM in black-box transfer settings, such as 3DGS to NeRF.

Automatic translation of natural language mathematics into faithful Lean 4 code is hindered by the fundamental dissonance between informal set-theoretic intuition and strict formal type theory. This gap often causes LLMs to hallucinate non-existent library definitions, resulting in code that fails to compile or lacks semantic fidelity. In this work, we investigate the effectiveness of tool-augmented agents for this task through a systematic factorial analysis of three distinct tool categories: Fine-tuned Model Querying (accessing expert drafts), Knowledge Search (retrieving symbol definitions), and Compiler Feedback (verifying code via a Lean REPL). We first benchmark the agent against one-shot baselines, demonstrating large gains in both compilation success and semantic equivalence. We then use the factorial decomposition to quantify the impact of each category, isolating the marginal contribution of each tool type to overall performance.

External validation is widely regarded as the gold standard for prognostic model evaluation. In this study, we challenge the assumption that successful external calibration guarantees model generalizability and propose two complementary strategies to improve transportability of prognostic models across cohorts. Using six real-world surgical cohorts from tertiary academic centers, we tested whether successful external calibration depends largely on similarity in covariates and outcomes between training and validation cohorts, quantified using Kullback-Leibler (KL) divergence, with calibration assessed by the Integrated Calibration Index (ICI). From the model-developer's perspective, we trained the "best-on-average" prognostic model by tuning toward a meta-analysis-derived covariate and outcome distribution as an approximation of the broader target population. From the end-user perspective, we proposed a simple measure for cohort outcome similarity to identify, among published models, the one most suitable for a given target cohort in terms of both calibration and clinical utility. External calibration worsened as distributional mismatch increased. Higher KL divergence was associated with higher ICI in both surgery-alone (Spearman $ρ=0.614$, $p=0.004$) and surgery + adjuvant chemotherapy cohorts (Spearman $ρ=0.738$, $p<0.001$). Meta-analysis-informed weighting improved calibration in most settings without materially affecting discrimination, with the clearest benefit when evaluated on the aggregated external population ($p=0.037$). Models developed in more similar cohorts achieved lower ICI in surgery-alone (Spearman $ρ=0.803$, $p<0.001$) and surgery + adjuvant chemotherapy cohorts (Spearman $ρ=0.737$, $p<0.001$), and provided greater clinical utility on DCA.

Machine learning components are now central to AI-infused software systems, from recommendations and code assistants to clinical decision support. As regulations and governance frameworks increasingly require deleting sensitive data from deployed models, machine unlearning is emerging as a practical alternative to full retraining. However, unlearning introduces a software quality-assurance challenge: under realistic deployment constraints and imperfect oracles, how can we test that a model no longer relies on targeted information? This paper frames unlearning testing as a first-class software engineering problem. We argue that practical unlearning tests must provide (i) thorough coverage over proxy and mediated influence pathways, (ii) debuggable diagnostics that localize where leakage persists, (iii) cost-effective regression-style execution under query budgets, and (iv) black-box applicability for API-deployed models. We outline a causal, pathway-centric perspective, causal fuzzing, that generates budgeted interventions to estimate residual direct and indirect effects and produce actionable "leakage reports". Proof-of-concept results illustrate that standard attribution checks can miss residual influence due to proxy pathways, cancellation effects, and subgroup masking, motivating causal testing as a promising direction for unlearning testing.

Test-time scaling (TTS) improves large language models (LLMs) by allocating additional compute at inference time. In practice, TTS is often achieved through parallel scaling: generating multiple candidate responses and selecting the best via a Best-of-N (BoN) strategy. Its effectiveness therefore hinges on the scoring function. Learned scorers such as process reward models (PRMs) can be strong, but they are expensive to train and run. Lightweight confidence heuristics based on token log-probabilities are much cheaper, yet we find that they often perform substantially worse. To improve on lightweight confidence heuristics without incurring the full cost of stronger learned scorers, we introduce SCATR, a simple and efficient BoN ranking method that learns a lightweight scorer from a small calibration set using hidden representations from the base model. Across coding and mathematical reasoning benchmarks, SCATR improves over prior confidence-based baselines by up to 9%. Relative to LoRA fine-tuning on the same calibration data, it achieves comparable accuracy with up to 8000x fewer trainable parameters and much lower compute, reducing training and inference latency by up to 150x and 1000x, respectively. SCATR is also competitive with strong PRM baselines, and in several settings improves accuracy by up to 7.8% on math and 4.2% on coding while enabling up to 1000x faster inference. Overall, SCATR offers a strong accuracy-efficiency trade-off for scalable test-time selection.

The communication protocols and data transfer mechanisms employed by IoT devices in smart buildings and corresponding digital twin systems predominantly rely on centralized architectures. Such centralized systems are vulnerable to single points of failure, where a malfunction can disrupt operational processes. This study introduces a blockchain-based decentralized protocol to enhance the cyber resilience of IoT data transfer for digital twins and enable decentralized automation of building operations. The framework incorporates public and private blockchain technologies alongside two case studies showcasing prototypes of each system. These prototypes were validated within a real-world building environment using smart home appliances and two digital twin platforms, with their performance evaluated based on cost, scalability, data security, and privacy. The findings reveal that the Hyperledger Fabric-based system excels in terms of scalability, speed, and cost-effectiveness, while both frameworks offer advantages over traditional centralized protocols in system cyber resilience, data security, and privacy.

Physics-aware recurrent convolutional networks (PARC) have demonstrated strong performance in predicting nonlinear spatiotemporal dynamics by embedding differential operators directly into the computational graph of a neural network. However, pixel-based convolutions are restricted to static, uniform Cartesian grids, making them ill-suited to following evolving localized structures in an efficient manner. Graph neural networks (GNNs) naturally handle irregular spatial discretizations, but existing graph-based physics-aware deep learning (PADL) methods have difficulty handling extreme nonlinear regimes. To address these limitations, we propose Graph PARC (G-PARC), which uses moving least squares (MLS) kernels to approximate spatial derivatives on unstructured graphs, and embeds the derivatives of governing partial differential equations into the network's computational graph. G-PARC achieves better accuracy with 2-3x fewer parameters than MeshGraphNet, MeshGraphKAN, and GraphSAGE, replacing the traditional encoder-processor-decoder framework with analytically computed differential operators. We demonstrate that G-PARC (1) generalizes across nonuniform spatial and temporal discretizations; (2) handles moving meshes required for structural deformation; and (3) outperforms existing graph-based PADL methods on nonlinear benchmarks including fluvial hydrology, planar shock waves, and elastoplastic dynamics. By embedding explicit physical operators within the flexibility of GNNs, G-PARC enables accurate modeling of extreme nonlinear phenomena on complex computational domains, moving PADLbeyond idealized Cartesian grids.

While deep learning systems are becoming increasingly prevalent in medical image analysis, their vulnerabilities to adversarial perturbations raise serious concerns for clinical deployment. These vulnerability evaluations largely rely on Attack Success Rate (ASR), a binary metric that indicates solely whether an attack is successful. However, the ASR metric does not account for other factors, such as perturbation strength, perceptual image quality, and cross-architecture attack transferability, and therefore, the interpretation is incomplete. This gap requires consideration, as complex, large-scale deep learning systems, including Vision Transformers (ViTs), are increasingly challenging the dominance of Convolutional Neural Networks (CNNs). These architectures learn differently, and it is unclear whether a single metric, e.g., ASR, can effectively capture adversarial behavior. To address this, we perform a systematic empirical study on four medical image datasets: PathMNIST, DermaMNIST, RetinaMNIST, and CheXpert. We evaluate seven models (VGG-16, ResNet-50, DenseNet-121, Inception-v3, DeiT, Swin Transformer, and ViT-B/16) against seven attack methods at five perturbation budgets, measuring ASR, Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), and $L_2$ perturbation magnitude. Our findings show a consistent pattern: perceptual and distortion metrics are strongly associated with one another and exhibit minimal correlation with ASR. This applies to both CNNs and ViTs. The results demonstrate that ASR alone is an inadequate indicator of adversarial robustness and transferability. Consequently, we argue that a thorough assessment of adversarial risk in medical AI necessitates multi-metric frameworks that encompass not only the attack efficacy but also its methodology and associated overheads.

MLP is a heavily used backbone in modern deep learning (DL) architectures for supervised learning on tabular data, and AdamW is the go-to optimizer used to train tabular DL models. Unlike architecture design, however, the choice of optimizer for tabular DL has not been examined systematically, despite new optimizers showing promise in other domains. To fill this gap, we benchmark 15 optimizers on 17 tabular datasets for training MLP-based models in the standard supervised learning setting under a shared experiment protocol. Our main finding is that the Muon optimizer consistently outperforms AdamW, and thus should be considered a strong and practical choice for practitioners and researchers, if the associated training efficiency overhead is affordable. Additionally, we find exponential moving average of model weights to be a simple yet effective technique that improves AdamW on vanilla MLPs, though its effect is less consistent across model variants.

Reliable uncertainty estimation is critical for medical image segmentation, where automated contours feed downstream quantification and clinical decision support. Many strong uncertainty methods require repeated inference, while efficient single-forward-pass alternatives often provide weaker failure ranking or rely on restrictive feature-space assumptions. We present $\textbf{SegWithU}$, a post-hoc framework that augments a frozen pretrained segmentation backbone with a lightweight uncertainty head. SegWithU taps intermediate backbone features and models uncertainty as perturbation energy in a compact probe space using rank-1 posterior probes. It produces two voxel-wise uncertainty maps: a calibration-oriented map for probability tempering and a ranking-oriented map for error detection and selective prediction. Across ACDC, BraTS2024, and LiTS, SegWithU is the strongest and most consistent single-forward-pass baseline, achieving AUROC/AURC of $0.9838/2.4885$, $0.9946/0.2660$, and $0.9925/0.8193$, respectively, while preserving segmentation quality. These results suggest that perturbation-based uncertainty modeling is an effective and practical route to reliability-aware medical segmentation. Source code is available at https://github.com/ProjectNeura/SegWithU.

Test-time scaling has become a powerful way to improve large language models. However, existing methods are best suited to short, bounded outputs that can be directly compared, ranked or refined. Long-horizon coding agents violate this premise: each attempt produces an extended trajectory of actions, observations, errors, and partial progress taken by the agent. In this setting, the main challenge is no longer generating more attempts, but representing prior experience in a form that can be effectively selected from and reused. We propose a test-time scaling framework for agentic coding based on compact representations of rollout trajectories. Our framework converts each rollout into a structured summary that preserves its salient hypotheses, progress, and failure modes while discarding low-signal trace details. This representation enables two complementary forms of inference-time scaling. For parallel scaling, we introduce Recursive Tournament Voting (RTV), which recursively narrows a population of rollout summaries through small-group comparisons. For sequential scaling, we adapt Parallel-Distill-Refine (PDR) to the agentic setting by conditioning new rollouts on summaries distilled from prior attempts. Our method consistently improves the performance of frontier coding agents across SWE-Bench Verified and Terminal-Bench v2.0. For example, by using our method Claude-4.5-Opus improves from 70.9% to 77.6% on SWE-Bench Verified (mini-SWE-agent) and 46.9% to 59.1% on Terminal-Bench v2.0 (Terminus 1). Our results suggest that test-time scaling for long-horizon agents is fundamentally a problem of representation, selection, and reuse.

Embryo selection is one of multiple crucial steps in in-vitro fertilization, commonly based on morphological assessment by clinical embryologists. Although artificial intelligence methods have demonstrated their potential to support embryo selection by automated embryo ranking or grading methods, the overall impact of AI-based solutions is still limited. This is mainly due to the required adaptation of automated solutions to custom clinical data, reliance on time lapse incubators and a lack of interpretability to understand AI reasoning. The modern, informed patient is questioning expert decisions, particularly if the treatment is not successful. Thus, evidence-based decision justification in tasks like embryo selection would support transparent decision making and respectful patient communication. To support this aim, we hereby present an expert-annotated dataset consisting of embryo images and corresponding morphological description using natural language. The description contains relevant information on embryonic cell cycle, developmental stage and morphological features. This dataset enables the finetuning of modern foundational vision-language models to learn and improve over time with high accuracy. Predicted embryo descriptions can then be leveraged to automatically extract scientific evidence from literature, facilitating well-informed, evidence-based decision-making and transparent communication with patients. Our proposed dataset supports research in language-based, interpretable, and transparent automated embryo assessment and has the potential to enhance the decision-making process and improve patient outcomes significantly over time.

Recent advancements in LLM agents are gradually shifting from reactive, text-based paradigms toward proactive, multimodal interaction. However, existing benchmarks primarily focus on reactive responses, overlooking the complexities of proactive intervention and monitoring. To bridge this gap, we introduce ProVoice-Bench, the first evaluation framework specifically designed for proactive voice agents, featuring four novel tasks. By leveraging a multi-stage data synthesis pipeline, we curate 1,182 high-quality samples for rigorous testing. Our evaluation of state-of-the-art Multimodal LLMs reveals a significant performance gap, particularly regarding over-triggering and reasoning capabilities. These findings highlight the limitations of current models and offer a roadmap for developing more natural, context-aware proactive agents.

In this paper, we proposed Bayesian Tucker decomposition (BTuD) in which residual is supposed to obey Gaussian distribution analogous to linear regression. Although we have proposed an algorithm to perform the proposed BTuD, the conventional higher-order orthogonal iteration can generate Tucker decomposition consistent with the present implementation. Using the proposed BTuD, we can perform unsupervised feature selection successfully applied to various synthetic datasets, global coupled maps with randomized coupling strength, and gene expression profiles. Thus we can conclude that our newly proposed unsupervised feature selection method is promising. In addition to this, BTuD based unsupervised FE is expected to coincide with TD based unsupervised FE that were previously proposed and successfully applied to a wide range of problems.

Many CAD learning pipelines discretize Boundary Representations (B-Reps) into triangle meshes, discarding analytic surface structure and topological adjacency and thereby weakening consistent instance-level analysis. We present STEP-Parts, a deterministic CAD-to-supervision toolchain that extracts geometric instance partitions directly from raw STEP B-Reps and transfers them to tessellated carriers through retained source-face correspondence, yielding instance labels and metadata for downstream learning and evaluation. The construction merges adjacent B-Rep faces only when they share the same analytic primitive type and satisfy a near-tangent continuity criterion. On ABC, same-primitive dihedral angles are strongly bimodal, yielding a threshold-insensitive low-angle regime for part extraction. Because the partition is defined on intrinsic B-Rep topology rather than on a particular triangulation, the resulting boundaries remain stable under changes in tessellation. Applied to the DeepCAD subset of ABC, the pipeline processes approximately 180{,}000 models in under six hours on a consumer CPU. We release code and precomputed labels, and show that STEP-Parts serves both as a tessellation-robust geometric reference and as a useful supervision source in two downstream probes: an implicit reconstruction--segmentation network and a dataset-level point-based backbone.

LLM-empowered agent simulations are increasingly used to study social emergence, yet the micro-to-macro causal mechanisms behind macro outcomes often remain unclear. This is challenging because emergence arises from intertwined agent interactions and meso-level feedback and nonlinearity, making generative mechanisms hard to disentangle. To this end, we introduce \textbf{\textsc{CAMO}}, an automated \textbf{Ca}usal discovery framework from \textbf{M}icr\textbf{o} behaviors to \textbf{M}acr\textbf{o} Emergence in LLM agent simulations. \textsc{CAMO} converts mechanistic hypotheses into computable factors grounded in simulation records and learns a compact causal representation centered on an emergent target $Y$. \textsc{CAMO} outputs a computable Markov boundary and a minimal upstream explanatory subgraph, yielding interpretable causal chains and actionable intervention levers. It also uses simulator-internal counterfactual probing to orient ambiguous edges and revise hypotheses when evidence contradicts the current view. Experiments across four emergent settings demonstrate the promise of \textsc{CAMO}.

Conformal prediction (CP) has attracted broad attention as a simple and flexible framework for uncertainty quantification through prediction sets. In this work, we study how to deploy CP under differential privacy (DP) in a statistically efficient manner. We first introduce differential CP, a non-splitting conformal procedure that avoids the efficiency loss caused by data splitting and serves as a bridge between oracle CP and private conformal inference. By exploiting the stability properties of DP mechanisms, differential CP establishes a direct connection to oracle CP and inherits corresponding validity behavior. Building on this idea, we develop Differentially Private Conformal Prediction (DPCP), a fully private procedure that combines DP model training with a private quantile mechanism for calibration. We establish the end-to-end privacy guarantee of DPCP and investigate its coverage properties under additional regularity conditions. We further study the efficiency of both differential CP and DPCP under empirical risk minimization and general regression models, showing that DPCP can produce tighter prediction sets than existing private split conformal approaches under the same privacy budget. Numerical experiments on synthetic and real datasets demonstrate the practical effectiveness of the proposed methods.

The deployment of Large Language Models in agentic, multi-turn conversational settings has introduced a class of privacy vulnerabilities that existing protection mechanisms are not designed to address. Current approaches to Personally Identifiable Information (PII) masking operate on a per-turn basis, scanning each user message in isolation and replacing detected entities with typed placeholders before forwarding sanitized text to the model. While effective against direct identifier leakage within a single message, these methods are fundamentally stateless and fail to account for the compounding privacy risk that emerges when PII fragments accumulate across conversation turns. A user who separately discloses their name, employer, location, and medical condition across several messages has revealed a fully re-identifiable profile - yet no individual message would trigger a per-turn masker. We formalize this phenomenon as Cumulative PII Exposure (CPE) and propose CAMP (Cumulative Agentic Masking and Pruning), a cross-turn privacy protection framework for multi-turn LLM conversations. CAMP maintains a session-level PII registry, constructs a co-occurrence graph to model combination risk between entity types, computes a CPE score after each turn, and triggers retroactive masking of conversation history when the score crosses a configurable threshold. We evaluate CAMP on four synthetic multi-turn scenarios spanning healthcare, hiring, finance, and general conversation, demonstrating that per-turn baselines expose re-identifiable profiles that CAMP successfully neutralizes while preserving full conversational utility.

The rapid expansion of gaming industry requires advanced recommender systems tailored to its dynamic landscape. Existing Graph Neural Network (GNN)-based methods primarily prioritize accuracy over diversity, overlooking their inherent trade-off. To address this, we previously proposed CPGRec, a balance-oriented gaming recommender system. However, CPGRec fails to account for critical disparities in player-game interactions, which carry varying significance in reflecting players' personal preferences and may exacerbate over-smoothness issues inherent in GNN-based models. Moreover, existing approaches underutilize the reasoning capabilities and extensive knowledge of large language models (LLMs) in addressing these limitations. To bridge this gap, we propose two new modules. First, Preference-informed Edge Reweighting (PER) module assigns signed edge weights to qualitatively distinguish significant player interests and disinterests while then quantitatively measuring preference strength to mitigate over-smoothing in graph convolutions. Second, Preference-informed Representation Generation (PRG) module leverages LLMs to generate contextualized descriptions of games and players by reasoning personal preferences from comparing global and personal interests, thereby refining representations of players and games. Experiments on \textcolor{black}{two Steam datasets} demonstrate CPGRec+'s superior accuracy and diversity over state-of-the-art models. The code is accessible at https://github.com/HsipingLi/CPGRec-Plus.

Magneto-active elastomers (MAEs) are polymer composites consisting of magnetic microparticles embedded in an elastomeric matrix. These materials exhibit strong magneto-mechanical coupling under external magnetic fields, resulting in tunable stiffness, reversible shape changes, and nonlinear magnetic responses. This study presents a multiscale theoretical framework to investigate the origin of magnetic hysteresis in MAEs, with emphasis on the evolution of the internal microstructure during magnetization and demagnetization. The total energy of the system is formulated as the sum of magnetic and micromechanical contributions, while macroscopic deformation of a cylindrical MAE sample is fully constrained. Particle interactions are modeled first via pure dipole-dipole interactions and then extended to include higher-order near-field effects at close particle separations. The results show that hysteresis in MAEs with magnetically soft particles primarily arises from trapped microstructural rearrangements, leading to distinct particle configurations under increasing and decreasing magnetic fields. Parametric studies demonstrate that particle volume fraction, sample aspect ratio, and matrix stiffness strongly influence the microstructure evolution and the width of resulting hysteresis loops. The proposed framework provides a solid foundation for modeling magnetic hysteresis, which is essential for the design and optimization of MAEs in practical applications.

The inverse Potts problem for estimating evolutionary single-site fields and pairwise couplings in homologous protein sequences from their single-site and pairwise amino acid frequencies observed in their multiple sequence alignment would be still one of useful methods in the studies of protein structure and evolution. Since the reproducibility of fields and couplings are the most important, the Boltzmann machine method is employed here, although it is computationally intensive. In order to reduce computational time required for the Boltzmann machine, parallel, persistent Markov chain Monte Carlo method is employed to estimate the single-site and pairwise marginal distributions in each learning step. Also, stochastic gradient descent methods are used to reduce computational time for each learning. Another problem is how to adjust the values of hyperparameters; there are two regularization parameters for evolutionary fields and couplings. The precision of contact residue pair prediction is often used to adjust the hyperparameters. However, it is not sensitive to these regularization parameters. Here, they are adjusted for the fields and couplings to satisfy a specific condition that is appropriate for protein conformations. This method has been applied to eight protein families.

Self-organizing memristive networks are physical circuits that dynamically reconfigure their circuitry in response to external input signals. Their adaptive behavior arises from intrinsic neuro-synaptic dynamics combined with a heterogeneous network topology. In this work, we demonstrate that such networks naturally generate neuronal population spiking dynamics similar to those observed in biological neuronal systems. This study investigates the intrinsic and emergent dynamics of memristive networks mathematically and numerically for both DC and AC input signals. Nonlinear spike-like features are maximized when the frequency of the input driving signal matches the network's intrinsic dynamical timescale, where nonlinear resonance is observed. Furthermore, the optimal frequency for computation is found to be the maximal frequency before the onset of resonance.

Magnetic solitons are nonlinear, local excitations in magnetic systems. In this study, we theoretically and numerically investigate the properties and generation of one-dimensional (1D) topologically trivial magnetic solitons in ferromagnetic nanowires. An approximate analytical soliton solution described by two free parameters is validated by comparing with the micromagnetic simulation. Across an interface between two media of different anisotropy, the reflection and refraction of a soliton are highly nonlinear that are different from the linear spin waves. A pair of magnetic solitons that propagate in opposite directions can be generated by alternately applying magnetic field or spin-polarized current pulses of opposite directions to at least two successive regions. Each soliton falls into a soliton solution that can be controlled by the generation process. These magnetic solitons can be used to drive domain wall motion over a certain distance determined by the soliton magnitude, allowing for discrete manipulation of domain walls compatible with the digital nature of information technology. Our findings pave the way for the application of topologically trivial solitons in spintronics.

We study the transport properties of non-Hermitian magnetic Josephson junctions, considering a superconductor-quantum dot-superconductor device coupled to a ferromagnetic metallic reservoir in the presence of an external magnetic field. We focus on the $0-π$ transitions that occur when the equilibrium phase difference between the superconductors shifts from $φ=0$ to $φ=π$ upon increasing the magnetic field amplitude. The coupling to the environment induces spin-dependent dissipation and leads to the broadening of the junction Andreev levels. By combining Green's function calculations with an effective non-Hermitian description restricted to the sub-gap Andreev quasi-bound states, we show that dissipation shifts the $0-π$ transition to higher magnetic fields. Remarkably, also the relative angle between the applied field and the reservoir magnetization can be used to drive the transition, at fixed field magnitude. We demonstrate that this effect can be entirely ascribed to the behavior of the complex eigenvalues of the non-Hermitian Hamiltonian. These findings highlight non-Hermiticity as a resource that can introduce new control knobs for engineering the current-phase relation in superconducting junctions.

We study the phase transition of the classical $XY$ model on a diamond lattice by Monte Carlo simulations using the Wolff cluster algorithm. Finite-size scaling (FSS) analysis of the Binder cumulant and the second-moment correlation length ratio $ξ_{2\rm nd}/L$ yields $T_c = 1.30036(1)$ and $ν= 0.671(6)$. Data collapse of both quantities confirms the three-dimensional $XY$ universality class.

Training in machine learning generally consists in finding one model, whose parameters minimize a data-dependent loss. Yet, empirical work shows that ensemble learning, an approach in which multiple models are sampled, can improve performance. Here, we provide an analytical framework to understand these observations in the case of Boltzmann machines, exploiting a duality between ensemble learning and large deviations of the free energy in spin-glass models. Replica calculations allow us to fully solve the case of spherical Boltzmann machine ensembles, and clarify when ensemble learning improves over standard loss minimization. Our findings are corroborated by numerical simulations on deep networks. Special care is brought to the case of nearly finite-dimensional data, for which we show that replica predictions are valid for arbitrarily large number of data points compared to the (large) embedding dimension.

Our understanding of vibrations in solids currently rests on concepts and techniques designed for crystals and explicitly relying on periodicity, hence inapplicable to amorphous materials. As a consequence, no established framework enables a systematic decomposition of the vibrational spectrum of amorphous solids into contributions associated with well-defined types of atomic motions. This methodological gap obscures the interpretation of various experimental probes of linear response, based on the measurements of acoustic, thermal, or optical properties. Here, we construct such a framework-Recursive Orthogonal Splitting Analysis (ROSA)-which decomposes the vibrational space by recursive applications of the projection formalism. Each step of the procedure exploits a dominant stiffness contrast to split a vibrational displacement subspace into two weakly coupled orthogonal complements. We illustrate ROSA by applying it to the archetypal covalent glass-vitreous silica. Successively separating bond-stretching, symmetric and antisymmetric oxygen motions, isotropic and deviatoric tetrahedral strains, and distinct classes of tetrahedral bending, reveals a hierarchical structure of the space of vibrational degrees of freedom, involving six mutually orthogonal subspaces. These subspaces selectively capture all salient spectral features, including the two-humped structure in the low-frequency range, the peak near 800 cm -1 , and the high-frequency doublet. In the low-frequency range, rigid-unit tetrahedral rotations do not constitute independent degrees of freedom but are kinematically enslaved to bending coordinates by no-stretch constraints. Because ROSA relies solely on the existence of contrasted stiffness scales associated with the point symmetry of local structural units, and on their separation by enforcing geometric constraints via the projection formalism, it is directly applicable to a broad class of covalent network glasses.

An analytical solution is obtained for the problem of the slow movement of a small drop of a fluid in another immiscible fluid in an infinitely large reservoir with the boundary condition of the normal slip and/or tangential partial slip at the interface. That generalizes the conventional Navier and Maxwellian boundary conditions of partial slip. Normal slip is accompanied by the density gradient in the fluid and is applicable only if one of the phases in contact at the interface is a gas. Although tangential partial slip and the associated generalization of the Hadamard-Rybczynski equation (HRE) have been considered previously, they were done using the friction coefficient formalism. Here, this issue is discussed within the more general formalism of slip lengths. It is proven that each of the two fluids separated by an interface has its own slip length. New equations describing the terminal velocity of gas bubble rise and aerosol falling have been obtained. The result is compared with experiment. It has been shown that the gas density within a rising bubble and around a falling droplet in the air is not uniform. The relative magnitude of the density increment increases with the size of the bubble or aerosol. Presumably, the best applicability of the generalized HRE should be expected for the interface of hydrophobic liquid and hydrophilic one (water and hydrocarbons, water and higher alcohols, in general: aqueous emulsions, water, lipophilic organic liquids and oils, etc.). These are quite important emulsions in practical terms, for example, for the oil industry and medicine. Experimental methods for determining the slip length are considered.

We investigate the effect of non-Hermitian Gamma interaction on the phase transitions and magnetic correlations for the transverse field Ising chain. We demonstrate that apart from the gapped ferromagnetic and paramagnetic phases, there is a gapless phase, where the system exhibits long-range spin-nematic order induced by parity-time symmetry breaking. Furthermore, we reveal that the parity-time symmetry breaking leads to the emergence of dynamical spin-nematic order, which also suggests a way of characterizing the spin-nematic phase diagram through non-equilibrium dynamics. Our findings show rich quantum phases stem from the competition among the Ising interaction, transverse field and non-Hermitian Gamma interaction, as well as providing a scheme for generating spin-nematic order in the spin chain.

Microscopic theories of magnetoresistance have traditionally focused on momentum relaxation and the plasma frequency of itinerant electrons. Here, we uncover a distinct mechanism in which magnetoresistance originates from quantum decoherence throughout the whole Fermi sea, specifically the decay of the off-diagonal components of the density matrix. The resulting conductivity, parameterized by two complex decoherence times, scales linearly with impurity density-markedly contrasting the conventional Drude picture, where conductivity is governed by momentum relaxation of Ferm-surface quasiparticles and is inversely proportional to impurity density. This unconventional scaling provides a direct electrical probe of quantum decoherence, a quantity central to both fundamental studies and emerging nanoscale technologies. Furthermore, the interplay between the external magnetic field and the exchange field gives rise to rich magnetotransport phenomena, including temperature-drive crossover from positive to negative magnetoresistance and a nonmonotonic temperature dependence with a conductivity maximum reminiscent of the Kondo effect. Our results establish quantum decoherence as a key ingredient in magnetoresistance and our findings should have an unprecedented impact on advancing research and applications involving magnetoresistance.

Charge-density-wave (CDW) order in correlated metals originates in a peaked electronic susceptibility at a finite wavevector $\mathbf Q_0$, set either by Fermi-surface features (nesting or saddle-point singularities) or by momentum-resolved electron-phonon coupling, or by a combination of the two. CDW precursor fluctuations can attenuate heat-carrying acoustic phonons even when long-range order is absent. We develop a Green's-function theory in which a damped-harmonic-oscillator propagator for a hybrid CDW--lattice soft mode at the ordering wavevector $\mathbf Q_0$ and a strain--intensity vertex obtained from an electron loop combine to give the acoustic phonon self-energy. The theory identifies two scattering channels: a local-intensity channel, controlled by a retarded composite CDW response and giving a narrow critical contribution when the CDW correlation length is large, and a texture (gradient) channel, which couples acoustic strain to spatial variations of the CDW envelope and, in a frozen-texture limit, reduces to a phenomenological form set by the measured diffraction peak weight and width. The same propagator fixes the lattice projection of a hybrid CDW--phonon soft pole measured by inelastic X-ray scattering, with an underdamped-to-overdamped crossover controlled by the distance to the CDW instability and a mass-tracking identity for the slow overdamped relaxation rate. The framework unifies diffraction, soft-mode spectroscopy, and thermal transport and applies broadly across CDW materials, including the transition-metal dichalcogenides, rare-earth tritellurides, kagome CDW compounds, and the cuprate fluctuating charge-order regime; we illustrate it by direct comparison with experimental IXS phonon softening and anomalous thermal transport in 2H-TaSe$_2$ at elevated temperatures.

Optimal transport and Wasserstein distance are prominent tools to quantify the space of probability distributions. From a novel viewpoint of manifold hypothesis in machine learning being a possible guide for the holographic principle, we study how holographic spacetime can emerge from quantum systems in general as a Wasserstein space through optimal transport. We employ the simplest example of a single quantum harmonic oscillator and demonstrate that, among various definitions of distance, the manifold hypothesis selects the 1-Wasserstein distance of optimal transport between Husimi Q-representations of states, and it gives rise to an emergent space. Furthermore, the Lindblad time evolution of the harmonic oscillator coupled to a bath, of the form of a Fokker-Planck equation, provides a time trajectory in the Wasserstein space, yielding an emergent Wasserstein spacetime that shares properties with black hole spacetimes and their event horizons. The methodology is applied to a Lindbladian subsystem of SYK model, revealing that the Wasserstein space is consistent with the AdS${}_2$ black hole geometry of the standard holographic dictionary. We remark that, in our examples, the 1-Wasserstein distance is identified as a generalized Krylov complexity, and argue that optimal transport with the manifold hypothesis can yield general emergent spacetimes, positioning the holographic principle on a broader basis.

The motivation behind the present work is to adopt methodology from field theory and high-energy physics to crystallography. In particular, we establish a relationship between the electromagnetic sector of the Standard-Model Extension (SME) for Lorentz invariance violation and optical media. At an effective level, electromagnetic properties associated with different crystal structures are demonstrated to be parametrized in the SME. Crystallographic and magnetic point groups provide the mathematical tools to show this correspondence. Birefringent and magnetoelectric media merit a dedicated study. Intriguing effects, which have not been described systematically in the modern literature, are rediscovered for the latter and expressed in SME language. With the setting developed at our disposal, materials with specific symmetries such as birefringent or multiferroic crystals serve as condensed-matter analogs for SME effects. It enables us to propose materials with unusual optical properties, which have not been thoroughly looked at in recent times.

We study an effective low-energy Kondo model on the triangular lattice in which itinerant electrons occupy a valence pocket at $Γ$ and three conduction pockets at the $M$ points of the Brillouin zone. This construction has a Fermi-surface nesting structure that favors triple-$Q$ magnetic order while only assuming the low-energy band-structure. Treating the local moments as classical spins on a four-sublattice magnetic unit cell, we find extended regions of non-coplanar order, including tetrahedral and related canted tetrahedral states, in addition to ferromagnetic and coplanar phases. The chiral phases remain stable over a broad range of inter-pocket Kondo couplings and persist in the presence of an external magnetic field. For certain chiral orders, the electronic bands can become gapped and host a quantum anomalous Hall state with $σ_{xy}=4\,e^2/h$. These results show that chiral magnetism and a quantized anomalous Hall effect on the triangular lattice do not rely on a specific tight-binding band structure, but can arise more generally from low-energy nested pockets at $Γ$ and $M$.

Advances in computational methods have made full-wave simulations in large disordered media increasingly feasible, but the resulting field data, scaling with the cube of the ratio of system size to wavelength, creates a severe storage and post-processing bottleneck. Generic compression methods are sample-specific and preclude operations on compressed data. We introduce OSCAR (On-Shell Compression And Reconstruction), a physics-based lossy compression scheme for weakly scattering media. OSCAR exploits the universal confinement of the Fourier representation of wave fields to a thin dispersion shell, a direct consequence of wave propagation when the scattering mean free path significantly exceeds the wavelength. The resulting compression ratio reflects two distinct scale separations: on-shell confinement due to weak scattering, and the excess Fourier-space volume introduced by sub-wavelength discretization of the scatterers. Crucially, second-order quantities such as intensity, correlations, and (optical) sensitivity can be computed via convolution entirely in compressed space and remain accurate even when individual field reconstruction incurs appreciable error, because coherent interference between independently compressed fields is preserved. Numerical simulations of electromagnetic waves in 2D and 3D confirm compression ratios up to ${\sim}380\times$ with sub-percent field error, enabling routine ensemble studies at scales relevant to biomedical optics, seismology, and underwater acoustics.

We investigate the creep dynamics of a driven elastic line at finite temperature, well below the depinning threshold. We show that creep is governed by two distinct length scales. The first, $\ell_{\mathrm{opt}}$, corresponds to the optimal activated rearrangements that control the dynamics' bottleneck and remains essentially temperature-independent. The second, $\ell_{\mathrm{av}}$, characterizes the spatial extent of thermally activated avalanches and grows as temperature decreases. By combining structural and dynamical observables, we show that $\ell_{\mathrm{av}}$ governs both the crossover in the structure factor and the growth of the four-point dynamical susceptibility, while the relaxation time remains controlled by activation over large barriers associated with $\ell_{\mathrm{opt}}$. We find that the avalanche scale follows $\ell_{\mathrm{av}}(T)\sim T^{-ν_{\mathrm{dep}}}$, thereby selecting a unique scenario among competing theoretical predictions. These results establish a unified picture of finite-temperature creep in which activation controls temporal scales while depinning criticality governs spatial correlations.

We theoretically investigate the magnetic and crystalline anisotropies of the superconducting diode effect (SDE) in proximitized planar Josephson junctions (JJs) with coexisting Rashba and Dresselhaus spin-orbit couplings (SOCs) under an in-plane magnetic field. A symmetry analysis identifies geometric constraints on magnetic-field and crystallographic orientations for which the SDE is suppressed independently of field strength, providing experimentally testable signatures of the interplay between SOC and Zeeman interaction. We develop a phenomenological model showing that the diode efficiency depends on the relative alignment between spin-orbit and magnetic fields, and corroborate this behavior in the narrow-junction, low-field regime using an analytical approach that links the anisotropy of the diode response to SOC-induced Fermi surface distortions and anisotropic Cooper pair momentum. These findings are supported by tight-binding simulations of the Bogoliubov-de Gennes equation, which reproduce recent experimental trends. The simulations reveal that electrostatic gating can induce polarity reversals of the SDE in the low-field regime even with only Rashba SOC, consistent with recent experiments, and predict additional reversals for specific field orientations, junction geometries, and SOC ratios. Our results elucidate the origin of anisotropic nonreciprocal superconducting transport and provide guidance for experimentally probing the mechanisms underlying the SDE in semiconductor-based planar JJs.

Many contemporary photodetectors operate beyond the readout of light intensity and enable the reconstruction of spectrum and polarization at the single-pixel level. However, the determination of light incidence direction with reconstructive detectors has not been realized so far. We show that photodetectors based on symmetric junctions of metals and 2d electron systems (2DES) enable (1) zero-bias photocurrent at oblique light incidence (2) reconstruction of incidence direction based on photocurrent measurements at variable carrier density. The former effect is based on peculiar electrodynamics of metal-contacted 2DES, where spatial variations of incident field phase translate into strong variations of local field amplitude. The local absorbances at two opposite metal-2DES junctions at oblique incidence are dissimilar, which results in finite photocurrent independent of microscopic rectification mechanism at these junctions. The direction of photocurrent uniquely determines the quadrant of light incidence. Quantitative determination of incidence angle becomes possible under conditions of 2d plasmon resonance at variable carrier density. In such a case, obliquely incident radiation excites the asymmetric plasmon modes, which amplitude carries unique information about angle of incidence.

Charged colloidal crystals exhibit a subtle interplay between electrostatic screening and elastic deformation. In an anisotropic elastic medium the coupling between dilation and the local ionic environment becomes direction dependent, leading to a preferential softening of the longitudinal acoustic response along specific crystallographic axes. This article provides a self-contained derivation of the long-wavelength static stability condition for cubic crystals subject to a generic electrostatic-elastic coupling. Starting from an effective static elastic tensor renormalized by a scalar coupling constant $λ_g$, we obtain an explicit condition for the onset of a homogeneous instability: the direction $\hat{\mathbf{k}}$ that first loses rigidity is determined by the inverse Christoffel matrix evaluated along that direction. Closed-form expressions for the critical coupling $λ_g^c$ are given for the $[100]$, $[110]$, and $[111]$ high-symmetry directions. We further provide a microscopic derivation of $λ_g$ from the Poisson-Boltzmann theory in a spherical Wigner-Seitz cell, linking the phenomenological constant to experimentally accessible parameters such as salt concentration, particle charge, and volume fraction. The analysis reveals that the most fragile direction can be identified without full lattice-dynamical calculations, and the associated unstable strain patterns are discussed. Numerical illustrations using experimentally measured elastic moduli of soft colloidal assemblies demonstrate the predictive power of the criterion. The present framework serves as a diagnostic tool for interpreting directional anomalies in static compressibility or low-frequency acoustic softening.

We present a systematic theory of chemically active emulsions in the hydrodynamic limit by constructing a thermodynamically consistent framework in which the equilibrium is broken by chemo-stating of fuel molecules. For ternary solutions with active chemical reactions, we obtain an effective dynamics of the conserved field dynamics at long length and time scales. The effective dynamics takes into account the broken time reversal symmetry that manifests itself by the emergence of gradient terms akin to those of Active Model B+, which is a generic theory of active phase separation. In addition to the active coefficients modifying the interfacial energy coefficient, the theory contains higher order terms in the gradient expansion that are necessary to correctly describe the dynamics of chemically active emulsions, extending thus Active Model B+. We study numerically a Flory-Huggins model with active chemical reactions. Our theory predicts the formation of microphases when the effective interfacial energy coefficient becomes negative. Moreover, including noise, we show the existence of bubbly phase separation. We also identify a new type of phase behavior, a dynamic active filament phase. Finally, we discuss the steady state entropy production rate in the system resulting from the active chemical reactions. We observe that the total entropy production rate increases with the driving chemical potential and exhibits a kink-like singularity at the transition to the dynamic active filament phase. Our work shows that the generic behaviors of active phase separation can emerge in chemically active emulsions.

We develop a theory of quantum spin Hall insulators with arbitrary spin $J$. Our analysis demonstrates that such systems support $J+\tfrac{1}{2}$ pairs of helical edge modes protected by nontrivial mirror Chern numbers. We establish that the corresponding edge theory is described by a generalized Dirac fermion with higher-order dispersion. These modes produce unique transport responses that are non-linear with voltage. An in-plane magnetic field opens a mass gap in the edge spectrum, and magnetic domain walls host $(J+\tfrac{1}{2})$-fold degenerate bound states characterized by nontrivial winding numbers. Our results extend quantum spin Hall physics to higher-spin systems and suggest possible realizations in ultracold atomic gases.

A Single-Photon Detector (SPD) can detect extremely low intensity of electromagnetic wave - down to a single photon. Driven by the rapid developments in quantum information science and an increasing demand for ultra-low-light sensing across various domains, there is a need for transformative advancements in the design and development of SPDs. In this context, low-dimensional platforms, including quantum dots, superconducting nanowires and layered materials have emerged as crucial frontiers of research. This review explores the state-of-the-art of different low-dimensional SPD platforms, focusing on the engineering physics across their device architectures, performance parameters and application potential. By critically comparing the performance and addressing current challenges inherent to each low-dimensional platform, the review aims to outline future research directions to advance next-generation SPD technologies.

The quantum metric is a fundamental ingredient of band quantum geometry and has recently at tracted intense interest, with most of its transport signatures appearing in the intrinsic second order nonlinear conductivity. In the clean limit, previous works argued that linear response conductivity is insensitive to the quantum metric, while the Berry curvature yields an intrinsic anomalous Hall con tribution. Here we combine analytic derivations with new numerics to show that disorder modifies the linear response conductivity dominated by geometric conductivity which is determined by the real space quantum metric. Focusing on a two dimensional multi-flatband stub-pyrochlore lattice, we identify a critical delocalized regime sandwiched between flat band localization and Anderson localization, characterized by finite geometric conductivity. Upon including spin orbit coupling, this regime evolves into a diffusive metallic phase, constituting a two dimensional inverse Anderson transition. Moreover, exploiting the connection between the real space quantum metric marker and the Wannier function spread, we construct a bond-percolation model on a square lattice. The resulting percolation region quantitatively coincides the critical delocalized regime, the exponent of which supports a classical percolation universality class. These findings suggest that flat band de localization can be understood as a classical percolation of quantum metric puddles. This advances our understanding of quantum geometric contributions to transport and establishes linear response measurements as a new avenue for accessing the quantum metric.

Quartz Crystal Microbalance (QCM) sensors are widely used to study biological and soft-matter interfaces due to their exceptional sensitivity to mass loading and interfacial mechanical properties. While classical QCM theory assumes predominantly shear-wave coupling into a semi-infinite Newtonian liquid, finite liquid thickness and acoustic reflections give rise to pronounced compressional (longitudinal) wave effects that strongly modulate both resonance frequency and motional resistance. Such compressional acoustic-wave responses should be properly accounted for when sensing in the liquid phase, for instance when working with cell suspensions. In this work, we systematically investigate compressional-wave responses in cell culture media including DMEM and RPMI-1640 across varying droplet volumes using a 5 MHz AT-cut QCM. Time-resolved measurements are analyzed using four parameters: the time period of compressional acoustic waves (Tca), the time associated with a phase shift between resonance frequency and resistance oscillations (Tp), the peak-to-peak shifts in frequency (Δfpp) and resistance (ΔRpp). DMEM and RPMI-1640 both exhibit strong volume-dependent periodic oscillations. At lower volumes, they exhibit low-frequency oscillations with a time period of approximately 40 minutes. However, as volume increases, the oscillations gradually evolve into high-frequency oscillations with a time period Tca of approximately 5 minutes. The peak-to-peak shifts (Δfpp) and (ΔRpp) are approximately 100-150 Hz and 40-60 Ω, respectively. The resonance frequency and resistance oscillations also exhibit a phase shift Tp of approximately 10 minutes. These results highlight that compressional-wave artifacts occur even in simple cell culture media, necessitating their explicit consideration when interpreting QCM data in the presence of cells.

During cell division active flows occur in the cortex, a thin layer of gel like network of acto myosin filaments, beneath the cell surface. The cortical flow and the associated stresses bring about change in the cell shape, in particular a sharp invagination at the mid cell. Using 3D phase field simulation of an active deformable shell, which captures coupled dynamics of cortical velocity and nematic order, we show how a nematic like actomyosin ring spontaneously emerge at the equator and drive sharp invagination. We further demonstrate how different cortical flow patterns, including counter rotating flows emerge near the division furrow. We show that these flow patterns, often attributed to intrinsic chirality of actomyosin filaments can instead arise from bias in the initial nematic alignment, revealing a memory effect in the system. By analyzing a simpler model of activity gradient driven compressive flow on a flat interface we decipher the main ingredients for surface instability leading to invagination and counter moving flows.

We present a systematic theoretical study of the Raman scattering effect for a two-level atomic system in near proximity of an ultrathin dielectric film with an embedded parallel array of periodically aligned single-wall semiconducting carbon nanotubes. More generally, our model provides a unified description of the quantum near-field medium-assisted enhancement effects for in-plane anisotropic metasurfaces, of which ultrathin periodically aligned carbon nanotube films are the representative example. Particular attention is given to incoming photon parameters of the external light radiation such as polarization and incidence plane orientation relative to the main anisotropy axis (nanotube alignment axis). By explicitly deriving the Raman scattering cross-section, we establish that for the two-level atomic system in the near-field zone of the carbon nanotube metasurface the effect can be enhanced by a factor of up to 10^4, not only for p-polarized but for s-polarized light as well.

A "mysterious" relation between the number variance and the variance of the $L$-th ordered eigenvalue, first suggested by French et al. [Ann. Phys. 113, 277 (1978)], is revisited and proven to be asymptotically exact for the $β=2$ Dyson symmetry class. Central to the proof is a previously unknown sum rule for the level spacing auto-covariances. Its derivation hinges on our previous work on the power spectrum description of eigenvalue fluctuations in random matrix theory. Analytical results for $β=2$ are complemented by conjectural extensions to the $β=1$ and $β=4$ symmetry classes. Our findings are corroborated by a comprehensive numerical analysis.

The Kramers-Kronig (KK) dispersion relations connect the real and imaginary parts of a causal response function through a Hilbert transform. While standard for linear-response theory, their validity for non-Markovian memory kernels -- the central object governing reduced dynamics in open quantum systems -- has not been rigorously established. We prove that the Nakajima-Zwanzig memory kernel belongs to the vector-valued Hardy space H+p(B) for every p>1, provided the initial system-bath state is factorized and the bath has a continuous spectral density. This membership yields rigorous KK reconstruction formulas. We establish three new results, including (i) a CP-Hardy obstruction theorem showing that upper-half-plane poles in approximate kernels imply non-physical (non-CPTP) dynamics; (ii) a passivity-analyticity theorem connecting dissipative kernels to the Herglotz-Nevanlinna class, realizing Gavassino's stability-causality chain for open quantum systems; and (iii) a moment-based Carleman criterion guaranteeing that Pade-reconstructed kernels from Memory Kernel Coupling Theory automatically satisfy KK. We validate the framework on solvable models and construct a counterexample demonstrating that correlated initial states break the Hardy property -- the quantum analogue of Gavassino's mechanism for how acausal macroscopic equations emerge from causal microscopic dynamics.

Among the statistical mechanical frameworks able to describe systems in non-equilibrium steady states such as collisionless plasmas, self-gravitating systems and other complex systems, superstatistics have gained recent attention. Superstatistics postulates a superposition of canonical systems with inverse temperatures $β$ described by a probability distribution depending on the external conditions. Unfortunately, the uncertainty about $β$ cannot be attributed to fluctuations of a phase space function, and this suggests that the distribution of $β$ is purely of statistical nature and must be inferred rather than measured. This lack of direct observability of the superstatistical temperature then becomes a conceptual issue in need of resolution. In this work we address this issue, showing that a mapping exists between functions of the superstatistical temperature and functions of the recently proposed fundamental temperature, a model-dependent function of the energy, in such a way that their expectation values coincide. We illustrate the use of this mapping by computing the conditional distribution of inverse temperature given energy for the $q$-canonical ensemble, as well as the full inverse temperature distribution, without the use of Laplace inversion.

Single photon emitters (SPEs) are integral parts of several quantum technology implementations. Over the past decade or so, monolayers of transition metal dichalcogenides (TMDCs) have emerged as one of the promising candidates for SPE platforms with attractive characteristics. To move ahead, it is necessary to understand the atomistic origin of SPEs in TMDCs - a topic which is highly debated with contradicting proposals. In this paper, we critically review these existing proposals to elucidate their origin. Further, we perform a critical trend analysis for different figures of merit of TMDC-based SPEs, and propose a characterization methodology to streamline the reporting process. Finally, we review several quantum technology implementations where solid state SPEs are being used, and identify the advancements required in TMDC-based SPEs for their successful adoption in these technologies.

Microorganisms navigating through narrow spaces encounter significant hydrodynamic challenges. To overcome these constraints and sustain efficient motion, they employ adaptive strategies, including adaptive oscillatory body deformations. While artificial microdroplets can traverse channels narrower than their diameter, studies of their locomotion have thus far been largely restricted to steady-shape regimes. In this work, we demonstrate a transition from steady shape to dynamic interfacial undulations in 5CB (4'-pentyl-4-cyanobiphenyl) droplets within aqueous trimethylammonium bromide (TTAB) solutions. We show that while droplets in dilute, additive-free solutions maintain a steady shape, the introduction of solutes or higher surfactant concentrations triggers pronounced interfacial undulations. Notably, both steady and undulating droplets exhibit a comparable velocity dependence on the confinement ratio, characterized by an initial deceleration followed by saturation, governed by the competition between hydrodynamic resistance and phoretic flow within the lubrication film. Furthermore, we find that increased surfactant concentration increases the capillary number, resulting in a thicker lubrication layer that facilitates a symmetry-breaking transition. Upon varying confinement, the droplet interface shifts from bilateral undulations to a mode localized on one side, forming a traveling-wave pattern strongly coupled to flow field fluctuations at the droplet's anterior. Linear stability analysis identifies the Yih-Marangoni instability as the underlying mechanism for these oscillations, revealing a previously unrecognized mode of adaptive locomotion in confined active matter.

The square lattice Ising model under the Brascamp-Kunz boundary conditions is a well-known exactly solvable lattice model. The exact solution of this system has been derived within the framework of Pfaffian-type method. In this paper we provide a derivation for the solution by the Schultz-Mattis-Lieb method in the transfer matrix formalism. We set special interactions on the boundaries and take certain limit of these interactions, so that the system under the Brascamp-Kunz boundary conditions is transformed into another system under the toroidal boundary conditions. The Schultz-Mattis-Lieb method is applied to the mapping system and the partition function is exactly solved in the fermionic representation. The Fisher zeros are analytically calculated and the physical critical point is identified. We also discuss the difference between the transfer matrix approaches to the Brascamp-Kunz and to the toroidal boundary conditions. Our work introduces a member to the family of transfer-matrix-based studies for Ising model under various boundary conditions.

Applications such as digital microfluidics and bio-diagnostics rely on droplet locomotion. A prominent example of such motion is durotaxis, a phenomenon that requires a stiffness gradient along a surface for the transport of liquids, cells, or other nano-objects. Using surfaces with varying properties in specific directions can be exploited as a universal concept for fluid transport with or without external energy supply. Changes in properties may refer to substrate patterns, Laplace pressure changes, wettability gradients, etc., leading to exciting phenomena, which can be employed in novel applications in various technologies. Here, we report on key results and progress in the area of directed droplet motion over the years, and we provide perspectives and implications for anticipated applications.

Low-temperature photoluminescence from MoSe2/WSe2 moiré superlattice often consists of a broad interlayer emission background with dense, narrow peaks, making microscopic line-by-line assignment difficult. Here, we use hyperspectral photoluminescence mapping and peak-decomposition-free spectral analyses to determine how this spectral complexity is organized in space. A 20 x 20 map acquired with a 400 nm pitch reveals three dominant spectral families that form contiguous real-space domains. Feature-wise spatial correlation analysis and whole-spectrum similarity yield a characteristic micron-scale length of 1.27-2.05 um, all exceeding the 0.85 um optical spot size. At the same time, individual pixels retain a dense, multi-peak structure, implying an unresolved local spectral manifold below optical resolution. Correlations among centroid, dominant energy, asymmetry, width, entropy, sharp fraction, and roughness indicate that the micron-scale energy landscape and local manifold complexity can be statistically separated, while remaining correlated across the map, consistent with a hierarchical organization of the emission spectrum. These results establish hierarchical inhomogeneity as an organizing principle of MoSe2/WSe2 moiré superlattice photoluminescence.

Reliable optical control of valley degrees of freedom in moire excitons requires that the emitted polarization faithfully reflect the underlying valley state. Here, we show that linearly polarized photoluminescence from WS2/WSe2 moiré excitons is largely insensitive to the excitation polarization and therefore is not primarily governed by valley-contrast selection rules. Automated polarization-resolved photoluminescence and Raman mapping at cryogenic temperatures reveal that the degree of linear polarization correlates strongly with local Raman shifts and moiré-exciton observables, identifying strain as the dominant experimental correlate. Exhaustive linear-regression analysis further shows that strain-related descriptors provide the best prediction of the observed polarization. Guided by theory, we attribute this behavior to strain-amplified breaking of C3 symmetry in the moire potential: weak uniaxial strain produces only partial cancellation of locally elliptical emission, yielding a finite far-field degree of linear polarization. These results establish strain as a key control parameter for reliable optical readout in TMD moire superlattices.

We explore the realization of distinct analog black hole horizons within tilted Weyl semimetals by comparing two models with contrasting spectral properties. We demonstrate that a spatially varying tilt in the Weyl cone structure creates an effect analogous to the tilting of light cones near a gravitational black hole horizon. By analyzing wave packet dynamics in both models, we reveal two fundamentally different types of analog horizons. The first model exhibits complete wave packet reflection, effectively mimicking an impenetrable barrier. In contrast, the second model permits wave packet transmission across the horizon. Critically, for both models, wave packets initialized with zero momentum ($k_0=0.0$) experience the strongest horizon effects, characterized by a dramatic slowing and significantly longer dwell times at the horizon region. Finally, we find that both systems exhibit substantial probability loss, which we demonstrate is directly correlated with the wave packet's dwell time near the horizon. Our findings establish tilted Weyl semimetals as a rich, tunable platform for investigating non-trivial quantum effects and information dynamics associated with analog black hole horizons.

Topological materials ranging from topological insulators to semimetals host many novel quantum phenomena including quantum spin Hall effect and topological Fermi arcs. Transitions between these topological phases have attracted much research interest. We performed angle-resolved photoemission spectroscopy (ARPES) on WTe2 ranging from a monolayer to the bulk and reveal the evolution of the electronic structure and the band gap. Notably, the gap observed in the monolayer system is suppressed in the three layers, where the film becomes metallic. Variations in the topological properties with thickness are demonstrated by the first-principles calculations. Topological Z2 invariant is shown to oscillate between 1 and 0 with the addition of layers, originating from the interlayer coupling-induced change in band crossing. The system evolves into a Weyl semimetal when the conduction and valence bands touch near the Fermi level and the topological nature is described by the Chern number. Our findings demonstrate the non-monotonic dependence of topological states on dimensionality and how layer-driven electronic band reconfiguration leads to phase transitions in solids.

The physics of a genuine one-dimensional system in which electrons are confined in one direction remains unclear. The actual electronic state of such a genuinely one-dimensional system has not been investigated in previous experiments, for they have all been conducted on quasi-one-dimensional specimens, namely in strongly anisotropic bulk crystals. Conventionally, charge-density waves (CDWs) driven by Fermi surface nesting have been considered to appear in one-dimensional electron-lattice systems. However, the CDW transitions actually observed to date have all occurred in quasi-one-dimensional systems and therefore do not directly indicate a genuine one-dimensional electronic state. We investigated, for the first time, isolated single and double-chain NbS$_{3}$ samples using the carbon-nanotube-sheath method and discovered CDWs in both systems. In the single-chain, surprisingly, a $(1/4)b^*$ CDW was observed, in contrast to the $(1/3)b^*$ CDW that has been observed in bulk samples. In the double-chain, the coexistence of a $(1/2)b^*$ dimer structure and a $(1/3)b^*$ CDW was confirmed. This discovery represents a significant advancement in the field of low-dimensional physics, surpassing the limitations of previous studies on bulk systems.

Phase-field models for anisotropic fracture employ two complementary mechanisms: (i) the anisotropic crack density function, controlling direction-dependent fracture resistance, and (ii) the anisotropic strain energy, governing the fracture driving force. Although the unified framework was presented in Pranavi et al.[Comput. Mech., 73 (2024)], the distinct roles of these mechanisms and their interaction remain uninvestigated. This work addresses this gap by first validating the formulation against mixed-mode fracture experiments on a soft elastomer (Lu et al. [Extreme Mech. Lett., 48 (2021)]), and then conducting systematic parametric studies on single-edge-notched (SEN) and open-hole tension (OHT) specimens to isolate each mechanism. The SEN studies show that the crack density anisotropy controls the crack path and toughness while leaving the elastic response unchanged, whereas the anisotropic strain energy deflects the crack but saturates rapidly. The OHT studies reveal a geometry-dependent role expansion: the anisotropic strain energy governs fiber-orientation-dependent stiffness, peak force, and fracture displacement. When both mechanisms act together, the combined response exhibits nonlinear synergistic interaction exceeding the linear sum of the individual contributions. These results establish that the crack density anisotropy governs the crack path (fracture resistance), while the anisotropic strain energy governs the driving force and, in stress-concentration geometries, additionally controls the elastic strain energy distribution around the stress concentrator.

In this work, we perform experiments with spherical polyacrylamide (PAAm) hydrogel drops/spheres, spanning a broad range of shear moduli and impact velocities on hydrophilic (plasma-treated glass) and hydrophobic (silane-coated) substrates, yielding an elastic number El variation of five orders of magnitude. Transient spreading morphology and impact force were simultaneously resolved using synchronized high-speed imaging and piezoelectric force sensing. At low elastic numbers ($El < 1$), impacting hydrogels exhibit a hybrid poroelastic response: a liquid-rich contact foot is expelled from the polymer network and spreads independently, while the bulk drop undergoes viscoelastic contact-line pinning into a pancake geometry at maximum deformation. At high elastic numbers ($El > 1$), contact foot spreading is suppressed, and deformation is accurately described by a neo-Hookean energy balance, yielding a maximum spreading factor independent of substrate wettability. Further, we show that the normalized peak impact force $F^*$ collapses to a constant value consistent with the Wagner limit for $El < 1$ and follows a power-law scaling $F^* \sim El^{0.38}$ for $El > 1$, in close agreement with both Hertzian and neo-Hookean predictions, and independent of substrate wettability. Furthermore, we highlight that post-impact retraction is suppressed across nearly the entire parameter space due to adsorbed polymer chains anchoring the receding gel network to the substrate, producing circumferential ridge instabilities; rebound occurs only when elastic restoring forces overcome the work of adhesion.

A simple framework is presented for unified applications in various fields of colloidal research, with minimal additional concepts & definitions. Several case studies concerning glass transition & crystallization are provided under the minimalist version, upon which adaptations can be made to suit more complicated topics. Major factors influencing accuracy are also discussed.

Dynamical properties of classical chaotic systems, for instance relaxation, can be understood as emerging from the time evolution of initially smooth long-wavelength densities to ever finer short-wavelength densities with fractal structure. Whether there is any analogous fractality by which one could characterize quantum many-body chaos is not known. By studying the spectral properties of the truncated operator propagator, we provide such structures. Namely, we show that the slowest-decaying operators, i.e., the leading Ruelle-Pollicott eigenvectors, have a nontrivial fractal dimension quantifying their non-locality, visible also in the divergence of their condition numbers. Furthermore, we find that unitarity imposes a constraint, i.e., an (approximate) equality, between the temporal decay rate of local correlations and this spatial operator fractal dimension. With this insight, a scenario for many-body quantum chaos becomes clear: over time, local operators evolve towards increasingly non-local ones with a quantifiable fractal structure, thereby naturally leading to effective non-unitary relaxation on the subspace of local operators - a kind of many-body Kolmogorov cascade in the space of operators. Our predictions are demonstrated in various quantum circuits: the kicked Ising model, brickwall circuits with a random 2-qubit gate, and dual-unitary circuits, where our results are exact.

While bulk silicon has long been understood to exhibit relatively weak spin-orbit coupling (SOC), confinement of electrons to quantum dots (QDs) at a silicon heterointerface results in significantly larger SOC. This is a concern for electron spin qubit performance, as intravalley and intervalley SOC can significantly perturb the operation of electron spin qubits. While these interactions can be harnessed to drive coherent rotations in a singlet-triplet qubit, coupling to low-lying excited valley states can lead to undesirable spin relaxation when valley splitting is on resonance with the Zeeman energy. In this work, we measure the angular dependence of the interfacial spin-orbit interaction as a function of the direction and magnitude of an applied external magnetic field in SiMOS and Si/SiGe heterostructures, two common material platforms for silicon spin qubits. We construct a physical model that accurately infers intra- and inter-valley SOC physics from fits to the data, allowing for a direct comparison between these two material systems. For the devices measured we find that, while the $g$-factor differences are comparable, the SiMOS QDs exhibit an order of magnitude larger spin-valley coupling than for Si/SiGe. Moreover, we find that the angular dependence of the spin-valley coupling is similar for both devices, with similar magnetic field orientations minimizing the spin-valley coupling. Our work points towards operational schemes for optimizing spin-valley coupling to avoid or exploit this mechanism for qubit operation.

We study wave propagation in 2D honeycomb structures with a non-commensurate or ``irrational'' line defect or edge. Our model is a Schrödinger operator which interpolates, across the edge, between two distinct bulk (asymptotic) Hamiltonians with a common spectral gap about the ``Dirac point'' of an unperturbed honeycomb operator. We seek edge states, eigenstates that are bounded and oscillatory parallel to the edge, and decaying in the transverse direction. For non-commensurate edges, the rigorous definition of these states is nontrivial due to the lack of translation invariance along the edge. To address this, we exploit quasiperiodicity along the edge by expressing the Hamiltonian as the restriction of a 3D (degenerate elliptic) Hamiltonian describing a 3D medium with a 2D interface within which there is periodicity. Via multiscale analysis, we construct approximate edge states in this 3D setting and obtain by restriction 2D edge states which are quasiperiodic along the irrational edge. These edge states are seeded by eigenfunctions of an effective Dirac operator, which has an infinite block-diagonal structure due to the non-commensurate geometry. A consequence is that infinitely many edge state eigenpairs arise, whose energies are dense in the perturbed bulk spectral gap. In a forthcoming paper, we rigorously construct these gap-filling edge states under a Diophantine condition. The main result here is a key tool in this construction: a resolvent expansion for the 3D Hamiltonian, whose leading term is the resolvent of the block-diagonal Dirac operator. The validity of this expansion requires an omnidirectional non-resonance (no-fold) condition on the dispersion functions of the unperturbed honeycomb Hamiltonian. This condition is satisfied in the strong binding regime. In contrast with earlier works on commensurate edges, the omnidirectional condition is independent of the edge.

We develop a geometric formulation of thermodynamic response in the classical Ising model by defining a curvature field over the control manifold spanned by inverse temperature $β$ and magnetic field $h$. We show that the existence of nontrivial curvature depends sensitively on the choice of control variables: while the $(J,h)$ manifold at fixed temperature is integrable and exhibits zero curvature, the $(β,h)$ manifold supports a finite curvature field arising from variations of the statistical ensemble. This curvature is given by a mixed derivative of the free energy and can be expressed directly as the covariance between energy and magnetization fluctuations. We evaluate the curvature field using Monte Carlo sampling and demonstrate that it develops a pronounced ridge structure extending from the critical point into the supercritical regime. This identifies the Widom line as a geometric feature of control space, corresponding to a locus of maximal thermodynamic response. More generally, the formulation provides a direct connection between geometric thermodynamics, critical phenomena, and experimentally accessible observables, and suggests that thermodynamic curvature may be probed through measurements of work performed under cyclic driving protocols.

Cooper quartets are aggregates of four electrons that generalize the concept of Cooper pairs, and their study can unfold unexplored perspectives in correlated matter and many-body physics. We propose a method to isolate them in a double-quantum-dot system coupled to conventional superconducting and normal leads. By driving the system out of equilibrium, we show that a resonance between the vacuum $|0\rangle$ and the four-electron state $|4e\rangle$ emerges in the high bias voltage regime, which involves a two-Cooper pair exchange process and is characterized by finite quartet correlations. We study the transport properties of the system and show that a peak in the Andreev current at high bias voltage has a width that scales with the magnitude of the quartet coupling $Γ_{4e}$, which can be tuned by the phase of additional superconducting leads, yielding distinctive signatures. By further studying the current-current correlations and the Fano factor, we establish a regime characterized by equal auto- and cross-correlations, which we interpret as a definitive signature of fast coherent two-Cooper-pair oscillations between the dots and the superconducting leads. The proposed platform, experimentally accessible in a quantum solid-state laboratory, enables exploration of quartet correlations and multifermion-correlated states of matter.

Flexible mechanical structures can undergo large deformations under small loads, enabling large, complex, and nonlinear wave responses under finite-frequency driving. Here, we study a dynamically driven canonical flexible mechanical metamaterial composed of rigid squares connected at their corners by flexible hinges. This metamaterial supports a uniform dilational mechanism and, in the limit of ideal joints, exhibits a Poisson ratio of -1. The presence of this dilational mode of deformation gives rise to a conformal symmetry, in which the dynamics are approximately invariant under a wide class of physical transformations -- conformal maps. We find that the low-frequency response of the system is dominated by conformal deformations consisting of spatially varying rotations and dilations concentrated at the boundary. Even at high frequencies, each conformal map implies a conserved spatially complex momentum. We explore how experimental parameters such as material stiffnesses and the geometry and number of unit cells allow experimental conformal momenta to approach this conservation, varying slowly compared to the non-conformal momenta of same order. These results constitute a new framework opening fundamental avenues for the study of conformal wave phenomena in dilational metamaterials as well as potential strategies for controlling nonlinear waves and vibrations.

We study phase transitions for repulsive-attractive mean-field free energies on the circle. For a $\frac{1}{n+1}$-periodic interaction whose Fourier coefficients satisfy a certain decay condition, we prove that the critical coupling strength $K_c$ coincides with the linear stability threshold $K_\#$ of the uniform distribution and that the phase transition is continuous, in the sense that the uniform distribution is the unique global minimizer at criticality. The proof is based on a sharp coercivity estimate for the free energy obtained from the constrained Lebedev--Milin inequality. We apply this result to three motivating models for which the exact value of the phase transition and its (dis)continuity in terms of the model parameters was not fully known. For the two-dimensional Doi--Onsager model $W(θ)=-|\sin(2πθ)|$, we prove that the phase transition is continuous at $K_c=K_\#=3π/4$. For the noisy transformer model $W_β(θ)=(e^{β\cos(2πθ)}-1)/β$, we identify the sharp threshold $β_*$ such that $K_c(β) = K_\#(β)$ and the phase transition is continuous for $β\leq β_*$, while $K_c(β)<K_\#(β)$ and the phase transition is discontinuous for $β> β_*$. We also obtain the corresponding sharp dichotomy for the noisy Hegselmann--Krause model $W_{R}(θ) = (R-2π|θ|)_{+}^2$ .

Accurate electrical amplification is essential in molecular electronics for measuring conductance through atomic and molecular junctions, where currents often span several orders of magnitude. In this work, we present a systematic design and comparative analysis of four current-to-voltage ($I\text{--}V$) amplifier architectures: single-stage linear, series-linear, logarithmic, and multi-stage cascaded, specifically optimized for break junction (BJ) techniques, including scanning tunneling microscopy (STM-BJ) and mechanically controllable break junctions (MCBJ). Each configuration is evaluated based on sensitivity, noise performance, and dynamic range. Our results characterize the trade-offs between circuit complexity and noise, providing a robust framework and practical guidelines for selecting amplification schemes in quantum transport experiments.

Freshwater scarcity demands desalination technologies that are efficient, scalable, and sustainable. Capacitive deionisation (CDI) is promising but remains limited by inefficient ion adsorption and poor charge utilisation. Here, we show that suitably chosen polyampholytic block copolymer grafting can substantially enhance CDI performance, via a combination of dipolar response and steric effects. Using mean-field classical density functional theory and grand-canonical Monte Carlo simulations, we demonstrate that such polymer grafted electrodes enable strongly improved desalination performance, without altering the pore architecture. Even an electrode grafting by simple neutral polymers can generate an improvement, although a suitably designed block polymer architecture offers an additional performance gain. These results establish interfacial block copolymer grafting as a powerful route toward high-performance, membrane-free desalination.

Casimir-Lifshitz forces generate an unavoidable, long-range attraction between protocells under prebiotically realistic conditions. This interaction stabilizes mesoscale clusters such as tetrahedra, octahedra, and 13-cell icosahedra. These highly symmetric assemblies act as persistent macrostates whose transitions remain reproducible despite microscopic noise. A physics-guided coarse-graining yields a well-defined mesodynamics that can be represented as an $ε$-machine: a small deterministic automaton whose causal states correspond to cluster attractors and whose transitions encode ordered reconfiguration pathways. The theory of Rosas et al. (Software in the natural world) shows that such systems can become informationally, causally, and computationally closed, thereby forming an autonomous proto-software layer. In this framework, prebiotic information does not arise from polymers but from attractor-based memory and structured transition dynamics in a purely physical cluster process.

Recent investigations of turbulent circulation fluctuations have uncovered substantial insights into the statistical organization of flow structures and revealed unexpected geometric features of turbulent intermittency. Of particular interest here is the observation that circulation probability distribution functions admit a superstatistical representation, namely a description based on "ensembles of Boltzmann-Gibbs ensembles". A fundamental phenomenological ingredient of this approach, which serves as a natural starting point for modeling, relies on the strong correlation between the dissipation field and the spatial distribution of elementary circulation-carrying structures, i.e., small-scale vortices. Within the language of superstatistics, this corresponds to characterizing circulation statistics through an appropriate choice of conditioned (Boltzmann-like) distributions and mixing distributions. We show that the superstatistical class of q-exponentials, known to have broad applicability in a wide range of multiscale and non-equilibrium systems, provides an accurate description of the observed circulation statistics in homogeneous and isotropic turbulence. This finding opens avenues for exploring the statistical structure of the turbulent cascade in the context of non-extensive statistical mechanics, rooted in the concept of non-additive entropies.

Establishing the fusion rules of anyonic quasiparticles in fractional quantum Hall fluids is essential for understanding their underlying topological order. Building on the conjecture that key topological properties are encoded in the "DNA" of candidate many-body wave functions - that is, the pattern of dominant orbital occupations restricted to a finite number of lowest Landau levels - we propose a combinatorial framework that derives these fusion rules directly from microscopic data. By extending Schrieffer's counting argument and introducing classes of topological excitations, our framework provides a unified route to the fusion rules for both Abelian and non-Abelian excitations. This approach elucidates the emergence of topological features from first principles in both fermionic and bosonic systems.

We study the properties and stability of networks with arbitrary Laplacian coupling. Classic approaches to studying networked systems require unrealistic assumptions, including homogeneous node dynamics, one-dimensional and undirected edges, or constant edge weights. We develop a unified formulation of Laplacian-style couplings that drops these assumptions, providing a unified notion for the edge weights of adaptive, directed, and multi-dimensional edges. We show that the recently developed theory of matrix phases can capture essential stability properties of the network and its edges. We quantify the impact of the asymmetry of the higher-dimensional edge dynamics on the system's phase properties by introducing the Asymmetry Rayleigh Ratio. These theoretical advances allow us to derive new sufficient stability conditions for AC power grids, directed diffusion, and the Kuramoto-Sakaguchi model. The resulting conditions are less conservative than the specific results known for these systems.

We present an experimental study of the Duffing--Holmes oscillator with a double-well potential, implemented as an analog electronic circuit under periodic external forcing. By systematically varying the forcing amplitude and frequency, we characterize the full dynamical landscape of the system through bifurcation diagrams, Poincaré maps, and maximum Lyapunov exponent calculations. The observed phenomenology includes period-doubling routes to chaos, periodic windows with multistability, dynamical intermittency, and antiperiodic orbits in which the trajectory recovers the global symmetry of the double-well potential. These results are synthesized into a high-resolution two-dimensional phase diagram in parameter space. The close agreement between all experimental diagnostics validates the fidelity of the analog implementation and demonstrates that continuous-time hardware provides a powerful platform for the quantitative study of nonlinear dynamics, free from the discretization artifacts inherent to numerical simulation.

The general relativistic magnetohydrodynamic (GRMHD) simulations are widely used to study accretion disk and jet dynamics around a black hole. Despite strong observational evidences for intrinsically nonlinear behavior, the interpretations of GRMHD simulation results, more precisely the underlying timeseries, have not been well-explored by nonlinear timeseries analysis. In this work, we characterize the jet and disk dynamics of different GRMHD simulated flows using the nonlinear timeseries analysis. As diagnostic tools, we consider Higuchi fractal dimension (HFD), Hurst Index (H) and spectral slope. We implement them for two model disk frameworks: magnetically arrested disk (MAD) and standard and normal evolution (SANE), across a range of black hole spins with the Kerr parameter spanning from -0.9375 to 0.9375. We simulate the disk/jet systems by two well-documented codes: HARMPI and BHAC, and obtain, respectively, low and high temporally resolved timeseries data. For both jet and disk dynamics, MADs are characterized by higher HFD, lower H and flatter spectral slopes than SANEs. High HFD in MAD could be due to its intermittent variability and indicates that it has lesser long-range temporal correlations than SANE. Moreover, HFD in MAD decreases with spin magnitude owing to increase in collimated, hence ordered, jets. However, in SANE, it increases with spin for positive ones due to interplay of winds and jets. Extending our analysis to observations, we attempt to segregate the classes of black hole: GRS 1915+105, into MAD- and SANE-like clusters based on their spectral properties extracted from X-ray data. The mean HFD of MAD-like cluster is higher than SANE-like cluster, thus, corroborating with the simulation results. Our work highlights the role of nonlinear timeseries analysis to understand the underlying dynamics of accretion flows and their connection to magnetic regulation.

Dissipative structures are open dynamical systems that sustain coherent macroscopic organization by continuously exchanging energy and matter with their environment and generating entropy. A recent thermodynamic analysis of the paradigmatic Malkus--Lorenz waterwheel interpreted the Lorenz system as an engine, deriving an exact formula for its thermodynamic efficiency, and showing that efficiency tends to increase as the system is driven far from equilibrium while displaying sharp drops near the Hopf subcritical bifurcation to chaos. Here, we extend that single-engine framework to coupled dissipative structures. We introduce two canonical couplings -- master-slave coupling (series) and symmetric diffusive coupling (parallel) -- and prove two fundamental association laws allowing us to reduce the composite systems to an equivalent engine with a specified efficiency. We then apply these abstract results to coupled Lorenz waterwheels, deriving efficiency formulas consistent with the underlying power balance. We perform numerical simulations confirming that (a) series coupling induces an increase in thermodynamic efficiency, (b) parallel coupling averages the efficiency of engines and increases total energy flow, (c) synchronization is typically neutral or beneficial for efficiency except in narrow parameter regions, and (d) coupling modifies the curvature of entropy-generation trends. Our theorems suggest a mathematically rigorous and transparent route to define and compute thermodynamic efficiency for generalized flow networks, with potential application to complex systems energetics.

Non-Hermitian dynamics in open systems can give rise to a variety of fascinating non-equilibrium phenomena, ranging from symmetry-breaking transitions to directional energy flow. Parity-time (PT) symmetry breaking determines the occurrence of dynamical instabilities, while non-reciprocal interactions enable asymmetric energy transfer between modes. Here, we present a versatile optomechanical platform based on a vacuum-levitated nanoparticle that allows full control over the coupling of its mechanical modes, including non-reciprocal and non-conservative interactions. By engineering the spatial ellipticity and polarization of the trapping beam, we continuously tune the system from a reciprocal to a strongly non-reciprocal regime. This allows us to observe PT-symmetry phase transitions and to isolate a unidirectional regime in which one mode remains effectively decoupled while driving the other. We demonstrate that elliptical polarisation of the trapping beam spanning unidirectional and reciprocal regimes induces asymmetric intermodal energy transfer. This results in the spontaneous cooling of one mechanical mode without external feedback. Both modes share identical mass, size, charge, and optical environment, providing a clean and robust setting for exploring non-Hermitian dynamics, exceptional-point physics, and energy redistribution in minimal systems. Combined with recent advances in ground-state cooling, our results provide a direct route to realising non-Hermitian phenomena in the quantum regime.

Do projects learn across space and time? The Olympics, among the largest publicly funded programmes in the world, offer a unique empirical setting. Theoretically, the Games seem ideal for generating "positive learning curves," driving down costs from one iteration to the next. In practice, they do not. Drawing on the concept of "myopia of learning," we argue that spatiotemporality (geographic distance, temporal gaps, and the temporary organisational form of each host committee) combines to block higher-level learning. Our analysis of cost overruns from 1960 to 2024 reveals no sustained improvement over 64 years. Tactical learning abounds, but none aggregates into strategic improvement. We propose four strategies for overcoming the spatiotemporal barrier (incremental, centralising, decentralising, and real options), arguing that radical reform is required.

International commerce has long been seen as a key way to keep the global food system stable, allowing agricultural surpluses in some areas to compensate for shortages in others. This strategy has led to the rise of highly specialised processing hubs that combine significant industrial capacity with agricultural inputs sourced from throughout the world. Türkiye's flour sector -- currently the largest wheat flour exporter in the world -- represents one of the most prominent examples of this model. However, increasing climate variability and geopolitical fragmentation raise important questions regarding the long-term resilience of food systems that rely heavily on imported biological inputs. Recent research shows the growing probability of synchronised crop failures across multiple agricultural regions due to atmospheric circulation anomalies and climate-induced extreme weather events. The assumption that global markets can consistently rebalance supply disruptions through trade is challenged by such events. Using the flour industry of Türkiye as a case study, this paper investigates the susceptibility of globally integrated grain processing centres. In order to assess the correlation between the scope of industrial processing and the capacity of domestic agricultural production, we introduce the Biophysical Autonomy Ratio~(BAR). The analysis demonstrates that Türkiye's BAR has declined consistently over time, suggesting that its processing sector has expanded beyond the domestic production base. The results suggest that in order to enhance the resilience of the food system in the future, it may be necessary to establish a more precise alignment between biological production systems and industrial food infrastructure. The paper concludes by addressing the policy implications for national food security governance in the context of escalating climate instability.

Continuous quantum phase transitions are characterized by an order parameter and correlation functions that are often challenging to access experimentally or in direct numerical simulations. The energy of an added impurity can on the other hand be probed by established polaron spectroscopy, or numerically with Monte Carlo methods. We provide evidence from ground-state quantum Monte Carlo calculations that the energy of a mobile impurity interacting weakly with a surrounding lattice Bose gas provides access to the critical behavior of the Mott insulator-superfluid phase transition. Finite-size scaling of the energy reveals that its value is scale invariant at the critical point of the quantum phase transition, and we extract a scaling exponent that is currently unexplained by theory. For a small lattice we further observe a flattening of the impurity-boson density-density correlations at the critical point, which hints at a divergence of a corresponding length scale in the thermodynamic limit. Our results suggest that impurity spectroscopy represents a useful way to probe the critical properties of quantum phase transitions in general.

Interference is conventionally attributed to path-accumulated phase differences, with measurement treated as a passive readout. Here we demonstrate that single-particle interference is governed by the relative phase between the prepared quantum state and the detector-defined measurement basis -- a joint quantity that is operationally inaccessible in any conventional interferometer. Using coherently seeded entangled nonlinear biphoton sources, we show that independently scanning the pump phase difference, the seed phase difference, or the signal interferometric phase each produces identical sinusoidal fringes ($V\approx0.99$) -- a three-scan equivalence impossible in any two-mode interferometer. The fringe visibility is continuously controlled through the idler-state overlap, directly encoding quantum distinguishability without idler detection. The same measurement-defined interference law persists from the single-photon to the high-flux regime. The bright-dark collective-state structure demonstrated here unifies coherent population trapping and electromagnetically induced transparency in atomic $Λ$-systems, discrete photonic interference, and single-slit diffraction within a common framework differing only in dark-subspace dimensionality, establishing measurement-defined photonic modes as a fundamental resource for quantum photonics.

Space-borne gravitational-wave telescopes are key to extend the observation band below $10\,\mathrm{Hz}$. The use of inter-satellite optical cavities linked by heterodyne interferometry is a promising approach to reach the sensitivity level of $10^{-22}/\sqrt{\mathrm{Hz}}$ in the decihertz band. While heterodyne interferometry is advantageous for relaxing arm-length control requirements, it introduces susceptibility to clock jitter, which can be a significant noise source. In the back-linked Fabry--Perot (BLFP) interferometer aiming at the decihertz band, the required clock stability exceeds that of current space-qualified oscillators by more than an order of magnitude. We propose a clock noise cancellation scheme that uses two heterodyne signals with positive and negative beat-note frequencies, naturally obtained using both incoming and outgoing laser beams of arm cavities without additional clock modulation schemes. By forming a weighted combination of these signals with time-dependent coefficients, clock jitter contributions are eliminated while preserving gravitational-wave information. We present the theoretical framework, analyze performance under realistic arm-length drifts, and validate the approach through time-domain simulations using parameters from the B-DECIGO concept. Results show that the synthesized signal recovers the original sensitivity and even improves the signal-to-noise ratio by a factor of $\sqrt{2}$ for shot noise.

State-of-the-art photon sources based on spontaneous parametric down-conversion (SPDC) currently rely on artificial structuring of the material nonlinearity to satisfy phase-matching conditions. This technique, known as periodic poling, is available only in a limited number of material platforms and introduces additional fabrication steps and errors, which are detrimental to up-scaling efforts. Here, we present a device architecture that enables SPDC of a wide range of frequencies without the need for periodic poling. We present explicit designs and calculations for 4H Silicon Carbide on-insulator, in which SPDC photon generation is so far unavailable, and thin-film Lithium Niobate on-insulator, a state-of-the-art quantum photonics platform. Our design, based on mode conversion and subsequent modal phase-matched SPDC, facilitates a CMOS compatible $χ^{(2)}$ platform, and simplifies photon sources by removing the requirement of periodic poling and the associated additional fabrication complexity.

We explore the role of intrinsic structural properties of hypergraphs in governing group-driven social dynamics with social reinforcement. First, we analyze simplicial contagion dynamics on random hypergraphs in which the level of hyperedge nestedness is systematically controlled. By developing the facet-based approximate master equation (FAME) method, we demonstrate that hyperedge nestedness induces a non-monotonic change in the outbreak threshold for simplicial contagion, displaying the lowest threshold at an intermediate level of hyperedge nestedness due to competition between simple and higher-order contagion processes. Next, we formulate the group-driven voter model (GVM) and investigate the consensus time for the GVM on hypergraphs with N nodes. Focusing on a representative case of the GVM, we show that the consensus time scales logarithmically with the system size as A ln N, where the prefactor A displays the fastest consensus formation at an intermediate level of social reinforcement due to competition between group-constraint and nonlinearity factors. Taken together, our results highlight the importance of competing effects arising from higher-order interactions in shaping optimality in group-driven social dynamical processes.

We demonstrate robust programmable state preparation in small VCSEL arrays with optical feedback using transient optical injection in the form of Gaussian pulses. In Lang--Kobayashi type models of delay-coupled 2- and 3-laser arrays, multistability gives rise to coexisting synchronized and symmetry-broken equilibrium branches. We show that short injection pulses applied to one or more lasers can steer the system from free-running operation to any stable equilibrium branch in the absence of noise by appropriate choice of pulse detuning and amplitude, after which the selected state persists without continued forcing. With correlated noise, injection steering remains effective, but branches with small basins of attraction are effectively destabilized by noise. These results validate pulsed injection as a practical mechanism for attractor selection in multistable VCSEL arrays and point to a feasible route toward experimental realization of programmable collective-state control.

The footprints of residential segregation have long been documented, yet the role of urban form as both medium and manifestation of segregation remains under-specified. We investigate whether the configuration of the built fabric may encode residential segregation in its spatial structure, hypothesising that built-form entropy (BFE) regimes are associated with the spatial distribution of income groups and their local clustering in non-linear ways. We examine this by quantifying BFE through a Shannon-based measure computed from building footprints, characterising income-based distributions using the Gini index and Moran's I, and placing both on a common spatial footing through a regular tessellation. Applying this framework to Sao Paulo, Latin America's largest city, we find non-linear relationships between BFE, income, and segregation: income levels and residential clustering increase toward both extremes of the entropy spectrum, with a stronger rise at the high-entropy end. This asymmetry suggests that high-entropy urban forms are associated with distinct spatial processes of segregation, including elite enclaving and incremental development in lower-income settlements, while low-entropy forms reflect more selective occupation shaped by planning and market filtering. Overall, the findings suggest that built form is more than a neutral backdrop, functioning as both affordance and signal of segregation.

Flat optics is now judged by more than a strong simulation or a single laboratory demonstration. To reach release, a device must survive a chain of handoffs: requirements, model selection, verification, layout release, fabrication, calibrated validation, packaging, and qualification. Diffractive optics brings mature routes for beam shaping and compact wavefront control, while meta-optics expands the design space through wavelength-scale control of phase, amplitude, and polarization. In both families, projects often slow down not because the optical function is impossible, but because the evidence required at each handoff is incomplete, poorly documented, or mismatched to the next decision. This tutorial organizes that problem into a stage-gate workflow, a set of compact technical checks, worked device examples, an artifact-based skills map, and an educational translation into workforce models, course deliverables, and assessment logic. The emphasis is practical: reduce avoidable redesign loops, make performance claims auditable, and clarify what students, instructors, and employers should be able to produce, review, and approve. The broader aim is to make the path from flat-optics concept to qualified hardware easier to understand, easier to teach, and easier to repeat.

The AI race amplifies security risks and international tensions. While the US restricts mobility and knowledge flows, challenges regulatory efforts to protect its advantage, China leads initiatives of global governance. Both strategies depend on cross-country relationships in AI innovation; yet, how this system evolves is unclear. Here, we measure the processes of polarization and integration in the global AI research over three decades by using large-scale data of scientific publications. Comparing cross-country collaboration and citation links to their random realizations, we find that the US and China have long diverged in both dimensions, forming two poles around which global AI research increasingly revolves. While the United Kingdom and Germany have integrated exclusively with the US, many European countries have converged with both poles. Developing and further developed countries, however, only integrate with China, signaling its expanding influence over the international AI research landscape. Our results inform national science policies and efforts toward global AI regulations.

This paper introduces two methods for analyzing periodic leaky-wave antennas (LWAs) within a new framework denoted as multi-modal network theory (MNT) with open periodic boundaries (OPBs). The approach is hybrid, combining analytical techniques with a commercial full-wave solver. The first method computes the dispersion diagram of periodic LWAs. It is iterative and relies on the full-wave simulation of a single unit-cell of a LWA, coupled with the analytical solution of an eigenvalue problem. This method effectively captures both the phase and attenuation constants of periodic LWAs while using fewer modes than previous methods with commercial frequency-domain solvers. The method is validated by computing the dispersion of classic LWA unit-cells and comparing them to those obtained through full-wave simulations of the full-length antenna and other state-of-the-art methods. The second, also based on OPB-MNT, focuses on LWA analysis in reception. Specifically, it determines the response of a unit-cell to an incident plane wave. To validate this method, we compute the response of LWA with different unit-cell designs. By comparing these results with the corresponding dispersion analysis, we show that the receiving case and the eigenvalue problem are related but not simply time-reversed versions of each other.

The recent detections of thioformyl cyanide (HCSCN) and propynethial (HCSCCH) in TMC-1 provide critical insights into the interstellar sulfur inventory, yet their sequestration and survivability on dust grain mantles remain poorly constrained. Here, we present a computational study of the site-specific adsorption of HCSCN and HCSCCH on amorphous solid water (ASW), modelled via water clusters (H2O)n, n = 6-16, at the wB97X-D/def2-TZVP level of theory, corroborated by QTAIM topological analyses and TD-DFT vertical excitations. Our results reveal a highly heterogeneous binding environment, with desorption energies spanning 1500 to 4900 K. Strongly bound cavity sites induce significant Stark shifts in the C=S stretching modes. Crucially, while the ice matrix exerts a negligible solvatochromic shift on UV transition wavelengths, deeply bound CN-cavity configurations exhibit a pronounced hyperchromic enhancement of the oscillator strength. Implementing these site-specific parameters into the UCLCHEM gas-grain code demonstrates that these species undergo a gradual thermal desorption profile rather than a singular sublimation event. Furthermore, the hyperchromic effect establishes a Survival Paradox: while deeply trapped populations are thermodynamically shielded against thermal desorption, they simultaneously possess enhanced UV absorption cross-sections, rendering them vulnerable to photodissociation by the interstellar radiation field prior to sublimation.

This work aims to address the bottleneck issues of hardware resource limitation and decoherence error in the Hamiltonian simulation of quantum fluids, which are caused by the standard quantum Fourier transform and the evolution of momentum operators, resulting in excessively deep circuits and excessive two-qubit gates. We propose an approximate operator optimization scheme aimed at reducing the circuit depth in Hamiltonian evolution. The proposed scheme successfully reduces the depth of analog circuits from $O(n^2)$ to $O(nlogn)$ or even $O(n)$ by eliminating $O(n^2)$ redundant two-qubit entangling gates. In this work, the numerical experiments are implemented on a supercomputing-oriented quantum simulator, simulating two-dimensional unsteady divergent flow. Experimental results demonstrate that although the truncation of high-frequency qubit coupling terms introduces deterministic theoretical errors, scaling at $O(n)$ for AQFT and $O(n^2)$ for momentum truncation, the optimized simulations successfully preserve the inherent macroscopic temporal evolution characteristics of the fluid in a 10-qubit simulation, achieving high correlation coefficients of $r$=0.933, $r$=0.941, and $r$=0.977 for density, X-momentum, and Y-momentum distributions respectively. Furthermore, we also analyzed the relationship between the algorithm truncation error and the hardware cumulative noise when the qubit number is extended to a higher level. This study proves that rationally adjusting truncation thresholds can establish an equilibrium point, preventing the hardware cumulative error from rapidly approaching 100% at the 20-30 qubit scale, providing a feasible engineering pathway for simulating complex fluid systems on real quantum devices in the future.

We study the hydrodynamic behaviour of a mesoscale numerical model for wetting dynamics based on the immersed boundary - lattice Boltzmann (IBLB) method. This IBLB model features a wetting potential to capture the interaction between a non-ideal droplet interface and a solid boundary; it is designed to prevent abrupt curvature changes near the contact line. As this approach prevents direct contact between the droplet and the solid, it forms a thin film beneath the droplet, which could compromise the hydrodynamic consistency in this region. This paper presents detailed comparisons against two other hydrodynamic solvers, respectively based on a boundary element method (BEM) and a volume of fluid (VoF) method, in order to examine the hydrodynamic behaviour of this IBLB scheme, elucidate its limits of validity in wetting applications, and explore the properties of its contact-line model.

Meta-optics promises compact, high-performance imaging and color routing. However, designing high-performance structures is a high-dimensional optimization problem: mapping a desired optical output back to a physical 3D structure requires solving computationally expensive Maxwell's equations iteratively. Even with adjoint optimization, broadband design can require thousands of Maxwell solves, making industrial-scale optimization slow and costly. To overcome this challenge, we propose the Neural Adjoint Method, a solver-supervised surrogate that predicts 3D adjoint gradient fields from a voxelized permittivity volume using a Fourier Neural Operator (FNO). By learning the dense, per-voxel sensitivity field that drives gradient-based updates, our method can replace per-iteration adjoint solves with fast predictions, greatly reducing the computational cost of full-wave simulations required during iterative refinement. To better preserve sensitivity peaks, we introduce a stage-wise FNO that progressively refines residual errors with increasing emphasis on higher-frequency components. We curate a meta-optics dataset from paired forward/adjoint FDTD simulations and evaluate it across three tasks: spectral sorting (color routers), achromatic focusing (metalenses), and waveguide mode conversion. Our method reduces design time from hours to seconds. These results suggest a practical route toward fast, large-scale volumetric meta-optical design enabled by AI-accelerated scientific computing.

The hallmark feature of photonic time crystals (PTCs) is the momentum bandgap, yet opening such a gap is extremely challenging, as it demands strong and rapid temporal modulation of the material properties. Recent theoretical advances have shown that resonance effects can substantially expand the momentum bandgap, and even give rise to a full (infinite) momentum bandgap spanning the entire momentum space. Despite these predictions, a full momentum bandgap has yet to be observed experimentally. Here, we report the first experimental observation of full momentum bandgaps in a microwave PTC. By enhancing the resonant effect, we demonstrate that the momentum bandgap can be drastically widened in a dynamically modulated microwave surface plasmon transmission-line metamaterial, leading to tighter spatiotemporal field confinement and greater robustness against temporal disorder. Remarkably, using a dynamically modulated microwave coupled resonator metamaterial characterized by coupled-resonator optical waveguide dispersion, we achieve a full momentum bandgap spanning the entire momentum space, thereby enabling arbitrary spatial localization and temporal amplification of microwave fields. Our findings establish a unified experimental framework for expanding momentum bandgaps up to an infinite extent with minimal requirements on modulation strength and speed, thus paving a viable route toward the first experimental realization of PTCs at optical frequencies.

Physical neural networks offer a transformative route to edge intelligence, providing superior inference speed and energy efficiency compared to conventional digital architectures. However, realizing scalable, end-to-end, fully analog recurrent neural networks for temporal information processing remains challenging due to the difficulty of faithfully mapping trained network models onto physical hardware. Here we present a fully analog resonant recurrent neural network (R$^2$NN) implemented via a metacircuit architecture composed of coupled electrical local resonators. A reformulated mechanical-electrical analogy establishes a direct mapping between the R$^2$NN model and metacircuit elements, enabling accurate physical implementation of trained neural network parameters. By integrating jointly trainable global resistive coupling and local resonances, which generate effective frequency-dependent negative resistances, the architecture shapes an impedance landscape that steers currents along frequency-selective pathways. This mechanism enables direct extraction of discriminative spectral features, facilitating real-time temporal classification of raw analog inputs while bypassing analog-to-digital conversion. We demonstrate the cross-domain versatility of this framework using integrated hardware for tactile perception, speech recognition, and condition monitoring. This work establishes a scalable, fully analog paradigm for intelligent temporal processing and paves the way for low-latency, resource-efficient physical neural hardware for edge intelligence.

Polycube structures provide parametric domains for all-hexahedral (all-hex) mesh generation and analysis-suitable volumetric spline construction in isogeometric analysis (IGA). Recent learning-based polycube pipelines have improved automation, yet several challenges remain when handling complex CAD geometries. These challenges include the limited diversity of primitive geometries, restricted grid configurations, and the increasing cost of genus-guided context search during inference as both the primitive set and the grid size grow. In this paper, we present {Scalable DDPM-Polycube}, an extended diffusion-based polycube construction method that addresses these limitations. First, we expand the primitive set from two primitive geometries to three by introducing a blind-hole cube primitive, thereby improving the representation of local hole-like features that do not change the global genus. Second, we extend the grid configuration from the previous $2\times 1$ setting to an enlarged three-dimensional grid configuration, which increases representational capacity and reduces mapping distortion for complex geometries. Third, we develop a genus-guided context generation strategy together with a hierarchical verification procedure, enabling robust context generation in both user-guided and automated modes. Once a valid polycube structure is generated, it is used for parametric mapping, all-hex control mesh generation, and volumetric spline construction. Experimental results demonstrate that scalable DDPM-Polycube improves the generality, scalability, and automation of diffusion-based polycube generation, and supports hex mesh generation and volumetric spline construction for IGA applications on complex geometries.

All-optical switching technology is a key solution to the future energy crisis in AI computing, where the performance of optical switches plays a critical role. Conventional integrated optical switches typically suffer from poor robustness to voltage fluctuations, fabrication variations, and temperature drifts. These limitations necessitate complex high-precision real-time calibration and control circuits, which greatly restrict their practical use. This paper presents a digital optical switch based on a thermally tuned multimode waveguide grating (MWG) filter. The switch maintains its on- and off-states across two voltage ranges: 0-0.7 V and 1.1-1.7 V, with a wide operating voltage margin of 0.6 V. It also exhibits excellent robustness to fabrication variations and temperature drifts. By introducing an innovative combination of positive dispersion and parabolic apodization design, the power consumption is reduced by two-thirds, reaching a maximum of only 6 mW. Owing to its low power consumption and wide voltage range, the device can be directly driven by digital signals, allowing for a simplified driver circuitry and a significant reduction in both energy use and overall cost. In addition, the switch offers low insertion loss (<0.5 dB), high extinction ratio (>20 dB), and fast switching (300 μs), demonstrating excellent overall performance and promising application prospects.

Confocal Brillouin microscopy enables high-resolution mechanical imaging but has low acquisition speed, partly due to its pixel-by-pixel mapping strategy. Line-scanning Brillouin microscopy (LSBM) significantly improves imaging speed by utilizing a multiplexing approach. However, current method is limited to a single-stage virtually imaged phased array (VIPA) spectrometer with insufficient capability of suppressing noise. Consequently, an absorptive gas chamber is often used to help reject excessive elastically scattered light. This approach requires specific tunable laser sources whose frequencies (e.g., around 780 nm) are locked to the absorption line of the gas chamber. Here, we developed a multiplexed Brillouin spectrometer for LSBM that increased the noise suppression to 57 dB without using any gas chamber. This is achieved by cascading two VIPA etalons with parallel dispersion axes in the spectrometer, where the first VIPA acts as a band-pass filter and the second as spectrum analyzer. We demonstrated its performance by acquiring Brillouin images of bio-printed phantoms with an inverted co-axial LSBM. This gas-chamber-free approach can expand the implementation of LSBM to other wavelengths where Brillouin scattering is more efficient and commercial laser sources are readily available.

Monolayer transition metal dichalcogenides (e.g., MoS2) exhibit exceptionally large optical nonlinearities for high-order nonlinear light generation (NLG), yet their inherent atomic thickness fundamentally limits light-matter interactions and thus conversion efficiency. Here, we overcome this intrinsic trade-off using a solution-processed bulk monolayer MoS2 (BM-MoS2) architecture composed of electronically decoupled MoS2 monolayers separated by organic interlayers. This layered superstructure preserves the exceptional intrinsic nonlinear susceptibility of monolayer MoS2 while enabling scalable interaction length. In the sub-wavelength regime, the NLG scales nearly quadratically with layer number (N^1.8), confirming the constructive buildup of nonlinear fields across stacked monolayers. As a result, a 100-nm-thick BM-MoS2 thin film exhibits colossal high-order NLG, including four-wave mixing and high harmonic generation nearly two orders of magnitude stronger than those from a 3-mm-thick ZnSe crystal. The generated nonlinear beam is directly visible to the naked eye and exhibits broad spectral tunability spanning more than 1000 nm in the mid-IR, enabling mid-IR-to-visible upconversion spectroscopy for resolving molecular vibrational fingerprints. By uniting monolayer-scale nonlinear susceptibility with bulk interaction length and coherent field buildup, BM-MoS2 establishes a thin-film platform for ultra-compact and substrate-agnostic nonlinear photonic systems beyond the constraints of conventional single crystals.

The solvation free energy (SFE) of molecules and ions is a fundamental property governing their solvation behavior and solubility. Molecular simulations offer a route to compute SFEs using alchemical free energy methods, such as thermodynamic integration or free energy perturbation. However, these methods suffer from the infamous end-point singularity, which leads to numerical instability when atoms approach closely, a challenge that becomes particularly acute in ab initio and machine learning molecular dynamics simulations. Here, we introduce the bubble method to calculate the SFEs of molecules and ions from first principles. Our approach avoids the end-state problem in both ab initio and machine learning molecular dynamics simulations and is applicable to molecules and ions of arbitrary shape. When calculating the SFEs of ions using periodic density functional theory, we incorporate corrections for the neutralizing background charge, spurious interactions between periodic images, and the vacuum-water interface potential. To validate our method, we successfully computed the SFEs of methane, methanol, water, and sodium ions using classical, ab initio, and machine learning molecular dynamics simulations. Importantly, our method requires no experimental inputs or empirical data. This makes it particularly well-suited for studying systems under extreme conditions, such as high pressure-temperature environments or under nanoconfinement, situations where experimental investigations are challenging and classical force fields, typically parameterized under ambient conditions, may be unreliable.

Biomolecular phase separation is typically attributed to the polymer physics of long, disordered chains. However, the underlying chemical grammar, i.e. the specific interactions between protein and RNA building blocks, remains poorly understood. We decouple those effects by screening the phase behavior of the complete dipeptide library in presence and absence of nucleic acids using full-atomistic molecular dynamics simulations. We demonstrate that (i) even these ultrashort units encode the instructions for spontaneous condensation, proving that phase separation is fundamentally rooted at a sub-polymeric level. (ii) Nucleic acids do not act as generic anionic glue but exert instead a base-specific regulatory logic. (iii) Individual nucleobases function as chemical tuners that dissolve, stabilize, or fluidize condensates based on their molecular identity. Overall, our minimal framework reveals that while polymer length enhances assembly, the core properties and regulatory control of condensates may be also governed by a fine-tuned chemical alphabet of peptides and nucleobases.

As the explosive growth of visual data increasingly strains the latency and energy limits of conventional electronic computing, optical analog computing has re-emerged as a disruptive paradigm for zero-power, speed-of-light information processing. Propelled by the unprecedented wave-manipulation capabilities of optical metasurfaces, this field is undergoing a rapid transition from macroscopic physical optics to ultra-compact, on-chip meta-processors. This Review examines the fundamental mechanisms of metasurface-empowered optical computing spanning Fourier-domain, nonlocal spatial-domain, and interferometric architectures that perform mathematical operations, with a particular focus on spatial differentiation and edge detection as representative computing tasks. By emphasizing recent breakthroughs, we highlight the evolution of meta-processors from static, linear regimes to dynamically reconfigurable, nonlinear, and quantum-assisted multidimensional platforms. We also envision how the synergy of AI-driven inverse design and the integration of analog meta-front-ends with optical neural networks will synergistically revolutionize next-generation intelligent machine vision.

We introduce the spatial disorder-generalized Langevin equation (SD-GLE), a data-driven method for constructing coarse-grained (CG) dynamics in heterogeneous systems. Unlike conventional CG approaches that rely on a mean-field potential, SD-GLE utilizes a variational Bayesian framework with a random field prior to explicitly disentangle static spatial disorder from viscoelastic friction. Numerical results demonstrate the limits of standard GLEs, whereas SD-GLE accurately extrapolates long-time dynamics to capture the anomalous diffusion crossover from short trajectories and recover the ensemble statistical properties inherent to the disordered nature of these systems.

We present a rigorous derivation and numerical validation of a universal projection-law factorization for quartic nonlinear coupling rates across physically distinct platforms. The central result is that observable Kerr-type interactions -- self-Kerr, cross-Kerr, and cross-phase modulation -- factorize into a dimensionless projection coefficient and an intrinsic quartic energy scale. This structure follows from canonical quantization of a quartic interaction projected onto a finite normal-mode basis. We state this result as a formal proposition with clearly enumerated assumptions, provide a proof sketch based on second quantization of the anharmonic potential, and enumerate the domain of validity. We then implement the scaling law as a lightweight computational toolkit (UEFT-Designer) with platform-specific kernels for superconducting circuits, photonic microcavities, and epsilon-near-zero (ENZ) structures. A complete, step-by-step worked example for a superconducting quarton device -- with full uncertainty propagation -- predicts a cross-Kerr rate $χ/2π= 361\pm 13\,\mathrm{MHz}$, agreeing with the independently measured value of $366\pm 0.5\,\mathrm{MHz}$ \cite{Ye2025} to within 1.4\%. A formal error-propagation analysis establishes that the dominant uncertainty is the Josephson-energy extraction uncertainty ($\sim\!2\%$), not the geometric projection factor. Cross-platform validation against five independent experiments spanning eight orders of magnitude in coupling strength confirms the universality of the factorization to within reported experimental uncertainties. The ghost-sector spectral correction introduced in earlier versions is reframed as an optional phenomenological self-energy model with no mandatory role in the validated scaling law.

Traditional melissopalynology is a time-consuming and subjective process, often taking 4-6 hours per sample. We present an automated, high-throughput microscopy system that integrates $H\infty$ robust mechanical control with advanced deep learning pipelines for the precise counting, classification, and morphological analysis of pollen grains from Bio Bio region in south central territory in Chile. Our system employs $U^{2}$-Net for salient object detection and a DINOv2 Vision Transformer backbone trained via Deep Metric Learning for classification. By integrating Gradient-Weighted Attention, the model provides human-interpretable texture and diagnostic feature annotations. The system achieves a 95.8$\%$ classification recall and a 6x processing speedup compared to manual expert analysis.

Time-bin encoded quantum states of light are crucial for quantum technology applications. The integration of manipulation functionalities into chip-scale devices is essential for deploying scalable, high-performance, and cost-effective quantum networks. Here we develop a fully integrated, high-throughput quantum receiver based on the thin-film lithium niobate (TFLN) platform, capable of high-speed electro-optic manipulation of time-bin encoded quantum states. The device's novel architecture enables active switching of time-bin quantum states with an electro-optic bandwidth exceeding 30 Ghz, while supporting real-time arbitrary projective measurements with a bandwidth of over 1 GHz. We showcase its versatility and performance through several applications, including the certification of entanglement with Bell's inequality violation by 38 standard deviations and with >95% visibility. We then apply it to a fiber-based quantum communication scenario, where we experimentally demonstrate an entanglement-based quantum key distribution (QKD) protocol, achieving stable finite-size secure key rates exceeding 25 kbit/s over 12 hours of continuous operation. By leveraging a high-speed active switching scheme, the system overcomes the need for temporal post-selection, eliminating a fundamental loophole that compromises the security of time-bin entanglement-based QKD protocols and relaxes the temporal resolution requirements of single-photon detectors. Moreover, it enables active selection of the projection basis, increasing the flexibility for communication parties. This approach establishes a versatile and scalable architecture for time-bin encoded quantum communication, enabling practical protocols on industry-grade photonic technology.

The performance of organic optoelectronic devices is critically dependent on how molecules orient within organic thin films. Yet, standard characterization techniques only reveal the first and second moments of the molecular orientation distribution. This limitation obscures the true molecular arrangement, as diverse distributions can yield identical low-order averages while exhibiting distinct functional properties. Here, we bridge this gap by combining multi-harmonic nonlinear polarimetry (second, third, and fourth harmonic) with the Maximum Entropy Method to reconstruct the probability distribution without any \textit{a priori} assumptions. This allows us to resolve features in the distribution such as asymmetry and bimodality, that remain invisible to conventional probes. Furthermore, we use this method to benchmark molecular dynamics simulations, revealing that these simulations often fail to capture the complex distribution despite correctly predicting the first and second moments. This work transforms molecular orientation from an inferred average into a precise observable, establishing essential validation standards for predictive material design.

Non-local thermodynamic equilibrium (NLTE) calculations remain a major computational bottleneck in radiation--hydrodynamics, while most existing machine-learning surrogates treat NLTE as a static input--output mapping rather than a kinetic evolution problem. Here, we present a physics-informed Latent Space Dynamics Identification (pLaSDI) framework specifically designed for NLTE atomic kinetics, which captures the time-dependent atomic kinetics of non-equilibrium plasmas through an explicit reduced governing equation. To ensure the physical reliability of the reduced model, we impose physics-informed loss terms that enforce macroscopic consistency, dynamical stability, and convergence to the correct steady state during long-time integration. Applied to tin NLTE population data generated along hydrodynamically modeled temperature--density trajectories relevant to extreme ultraviolet (EUV) lithography plasmas, the model accurately reproduces charge-state evolution and mean charge state with errors below 2\%, achieves speedups of approximately $5\times10^{4}$--$10^{5}$, and remains stable outside the training trajectories by converging toward physically admissible states and the correct steady-state solution under fixed plasma conditions. These results show that careful physics-informed design of the latent dynamics, rather than data fitting alone, is essential for constructing fast, stable, and physically reliable extrapolative surrogates for time-dependent NLTE kinetics.

Lensless cameras based on thin diffusers offer a compact alternative to conventional refractive imaging but rely on computational reconstruction, since the diffuser's point spread function (PSF) globally multiplexes every scene point across the sensor. Here, we report a Fourier Neural Operator (FNO) framework for this reconstruction task. Because a linear shift-invariant forward model reduces to a pointwise multiplication in Fourier space, the spectral-domain kernel of an FNO layer is structurally aligned with the DiffuserCam inverse problem. Using a compact DiffuserCam prototype and a 25,000-image natural-scene dataset, our FNO improves upon a U-Net baseline of comparable parameter count by $2.14$~dB in PSNR and $0.11$ in SSIM. The same FNO, trained exclusively at $128 \times 128$, reconstructs $256 \times 256$ and $512 \times 512$ measurements with less than $1$~dB loss in PSNR and no retraining, demonstrating resolution-agnostic inference. The framework is directly applicable to other lensless modalities with global PSFs, such as multimode-fiber endoscopy.

Holographic displays are widely regarded as the "ultimate" display technology, promising immersive 3D visuals with natural depth cues, continuous parallax, and perceptual realism. Realizing this potential, however, has remained elusive due to persistent image quality limitations -- most notably speckle noise, a byproduct of the random interference inherent to coherent light. This is typically further exacerbated by the hologram's phase randomness required for maintaining uniform energy distribution across the eyebox. While speckle suppression techniques like temporal multiplexing or smooth-phase heuristics exist, they often necessitate high-speed hardware and introduce visual artifacts, hindering their practical adoption. We introduce Ellipsography, a single-shot holography technique that achieves near-limit speckle suppression, reaching the image fidelity equivalent to averaging a million conventional scalar holograms -- in a single frame in simulation. By jointly modulating the phase and polarization of light, we structure optical interference and suppress speckle at its source. We present a full pipeline including a vectorial wave model, an end-to-end hologram synthesis algorithm, and a functional prototype display. Our experiments demonstrate substantial improvements in visual clarity, depth continuity, and focus cues over current state-of-the-art methods, achieving high-quality reconstructions approaching 30dB PSNR on a real holographic display for the first time -- a 10dB improvement over the best existing techniques. By pushing holographic reconstruction closer to the perceptual quality expected of modern displays, Ellipsography sets a new benchmark for practical, high-fidelity, speckle-free holography.

Deep learning has triggered explosive growth in the demand for specialized hardware processors, thus motivating the development of scalable and reconfigurable computing substrates. Optical processors offer a fundamentally different computing paradigm, combining massive parallelism and ultrahigh bandwidth with the potential for substantial energy savings. However, progress has been constrained by the absence of scalable and reconfigurable architectures that can implement a broad class of network architectures. Here, we introduce TRON, a scalable and trainable optoelectronic deep optical neural network that exploits a multi-scattering medium and a DMD as a learnable, high-dimensional dense optical matrix multiplier, processing with fixed and tunable optical operations. We perform in-situ optimization of the optical parameters involved in the scattering process, together with automated neural architecture search (NAS) and optimization directly on optics. The experimental results demonstrate that in-situ NAS is essential to discover architectures that adapt to both the task and hardware constraints, establishing a viable path towards large-scale optical processors for next-generation machine learning and data-intensive computing.

Call detail records (CDR) from mobile phone networks are widely used to study human mobility however CDR data from a single mobile operator are inherently biased because the observed users do not mirror the population distribution. Using data from a major Chilean carrier in Santiago, we observe the user base is skewed by socioeconomic group, so aggregate metrics like radius of gyration are distorted by the population that is actually observed. To correct this sampling bias, we apply multilevel regression and poststratification (MRP), a method that is not yet standard for CDR-based mobility studies. We fit a Bayesian multilevel model for individual mobility using socioeconomic status, gender, and geography, with partial pooling across comunas, and then poststratify the predictions to match census demographics. This approach reduces the naive CDR estimate of average radius of gyration by about 17%. Importantly, a version of the model that uses only geographic information still captures much of the bias, showing that MRP can be useful even when the socioeconomic composition of users is not fully known, as long as spatial patterns of socioeconomic groups exist. This example demonstrates how MRP can provide a principled correction for non-representative CDR-derived mobility estimates, rather than treating the carrier sample as if it were a random population sample.

We present a high-yield method for fabricating concave micromirror templates in silica using feedback-controlled CO2 laser ablation with precise in situ positioning. Real-time monitoring of the white-light emission generated during ablation is used to terminate laser exposure, thereby reducing shot-to-shot variability in mirror depth and radius of curvature. To ensure reproducible single-shot processing across different substrates, the sample position relative to the laser focus is calibrated using an in situ phase-scanning interferometric microscope integrated into the fabrication workflow. The method enables reliable fabrication of shallow mirror templates with tunable radii of curvature spanning from approximately 20$\mathrm{μm}$ to several hundred micrometers, with relative geometric variances as low as 3%. The suitability of the fabricated mirrors for optical resonators is verified by realizing a compact plano-concave Fabry--Perot microcavity with a finesse of 37000 at telecom wavelengths. The setup provides a simple and automated route to reproducible micromirror fabrication for applications in cavity quantum electrodynamics and cavity optomechanics.

Stable dissemination of terahertz (THz) signals over long distances is important for next-generation synchronization networks, radio astronomy, and high-capacity wireless systems. Optical fiber provides a low-loss platform for coherent frequency transfer; however, when a THz carrier is encoded as the difference between two optical wavelengths, chromatic dispersion introduces differential phase noise that degrades spectral purity. Here, we demonstrate phase-coherent distribution of opto-THz carriers over 38 km of standard single-mode fiber using a dual-wavelength Brillouin laser (DWBL) combined with a dual-channel round-trip noise-cancellation architecture. By extracting the differential phase noise between the two optical lines via a dual-channel round-trip measurement, dispersion-mediated phase fluctuations are compensated, and the intrinsic stability of the source is effectively preserved at the remote end within the measurement sensitivity. Opto-THz carriers at 150, 300, and 600 GHz exhibit sub-femtosecond timing stability and fractional frequency instabilities below 1e-17 at 10,000 seconds of averaging.

Recent advances in the use of High-Resolution Cross-Correlation Spectroscopy (HRCCS) to detect molecular species in exoplanet atmospheres, presents a new challenge for the accuracy of reference spectroscopic line lists. While parent isotopologues of key atmospheric tracers are often well-characterized, minor isotopologues, crucial for diagnosing planetary formation histories and evolution, suffer from a scarcity of experimental data, often leading to reliance on less accurate theoretical predictions. In this work, a comprehensive machine learning framework is designed to mitigate these inaccuracies by modelling the residual errors of the isotopologue extrapolation (IE) method used within the ExoMol project. A fully connected neural network architecture for carbon dioxide (CO$_2$) is shown to predict energy corrections with high fidelity, reducing the mean absolute error (MAE) relative to the original IE approach for more than 87\% of the levels when benchmarked against empirical (\Marvel) energies. Furthermore, development of a novel hybrid, molecule-aware transfer learning architecture is presented that successfully propagates correction patterns from the data-rich CO$_2$ system to the data-poor carbon monoxide (CO) system. This transfer learning approach yields MAE improvements in over 93\% of CO samples, demonstrating that physical correction factors related to isotopic substitution can be generalized across chemically related molecular systems. Updated and improved line lists are presented for 11 CO$_2$ isotopologues and energy levels for excited states of CO isotopologues are predicted. The methodology establishes a scalable, data-driven paradigm for refining molecular line lists, helping to bridge the gap between theoretical calculations and experimental precision.

Laser-plasma wakefield acceleration (LWFA) offers ultrahigh accelerating gradients in compact setups, but the complex non-linear nature of the process makes it challenging to generate high-quality beams. Injection of electron bunches from an external source into a plasma accelerator provides a promising route to improved performance; however, electron bunches from conventional radio-frequency (RF)-based injectors suffer from non-linear compression and laser-beam asynchrony, leading to energy jitter and emittance growth. We present a fundamental concept of terahertz-controlled electron bunches for external injection into LWFA. This terahertz-frequency approach provides temporal locking between the electron beam and the drive laser, and enables the compression of high-quality beams to sub-10-fs durations before injection into the LWFA. Numerical simulations demonstrate that GeV-scale acceleration with excellent beam quality and stability -- energy jitter and energy spread around 0.2% -- can be achieved using this method. This concept opens new opportunities for stable, multi-stage laser-driven accelerators and supports the development of next-generation applications such as free-electron lasers (FELs).

Scale-resolving simulations of high Reynolds number incompressible flows are often limited by the Courant-Friedrichs-Lewy (CFL) stability restriction imposed by explicit time-stepping schemes, resulting in small time step sizes and long time-to-solution. In this work, we systematically compare two implicit formulations of the velocity correction scheme -- a linear-implicit approach and a sub-stepping (or semi-Lagrangian) method -- against a standard semi-implicit formulation within a high-order spectral/hp element framework. The schemes are assessed in terms of stability limits, temporal accuracy, and computational performance for implicit large-eddy simulation of the Imperial Front Wing benchmark, a complex high Reynolds number geometry with curved surfaces that imposes strict CFL constraints. Both implicit schemes extend the stability limit by up to two orders of magnitude in time step size. While increasing the cost per time step, they reduce the overall time-to-solution by up to a factor of eleven. Accuracy analysis shows that time step sizes up to twenty times larger than the explicit limit have only minor impact on resolving laminar-turbulent transition and key flow statistics. The results quantify the trade-off between stability, accuracy, and computational cost for implicit velocity correction schemes on complex geometries and provide guidance for selecting time integration strategies in large-scale scale-resolving simulations.

For head and neck cancer (HNC) patients, prognostic outcome prediction can support personalized treatment strategy selection. Improving prediction performance of HNC outcomes has been extensively explored by using advanced artificial intelligence (AI) techniques on PET/CT data. However, the interpretability of AI remains a critical obstacle for its clinical adoption. Unlike previous HNC studies that empirically selected explainable AI (XAI) techniques, we are the first to comprehensively evaluate and rank 13 XAI methods across 24 metrics, covering faithfulness, robustness, complexity and plausibility. Experimental results on the multi-center HECKTOR challenge dataset show large variations across evaluation aspects among different XAI methods, with Integrated Gradients (IG) and DeepLIFT (DL) consistently obtained high rankings for faithfulness, complexity and plausibility. This work highlights the importance of comprehensive XAI method evaluation and can be extended to other medical imaging tasks.

We theoretically investigate an annularly confined Bose-Einstein Condensate (BEC) coupled to a four-mirror ring cavity supporting traveling-wave optical modes. Under symmetric driving by counter-propagating Laguerre-Gaussian beams carrying equal and opposite orbital angular momenta, the system realizes supersolid phases coexisting with persistent superfluid circulation. Specifically, we obtain a supersolid state if we start with a BEC of winding number $L_p$ as well as supersolid packets with coherent superpositions of two different BEC $L_p$ values. Under asymmetric pumping, realized with Laguerre-Gaussian beams of different orbital angular momenta, chiral symmetry is broken in the system, resulting in asymmetric cavity field amplitudes, directional density modulations, and tunable rotational dynamics of the resulting supersolid lattice. This leads to rotating supersolid density structures for a single winding-number state, and rotating wave packets for an initial superposition of rotational eigenstates. Finally, we probe the presence of Goldstone and Higgs modes which can be observed using minimally destructive measurements of the cavity output spectrum. Our mean-field theory reveals interference-driven rotation without physical stirring, and distinguishes our work from prior static cavity supersolids. Our results establish the ring cavity annular BEC as a versatile platform for generating chiral quantum matter, implementing rotation-sensing devices and generating atomtronic circuits with supersolids.

The serial nature of two-photon polymerization (2PP) limits fabrication throughput. While diffractive optical elements (DOEs) can be used to generate multiple write spots in a single plane, three-dimensional structures still require sequential layer-by-layer fabrication. We demonstrate a dual-plane multi-spot 2PP approach using a single static DOE capable of generating two independent focal-spot arrays in distinct planes. This configuration enables simultaneous fabrication of two layers during continuous scanning while maintaining a simple scanning strategy compatible with woodpile structures. Using 29 write spots distributed across two planes separated by 1.8 $μ$m, we fabricate four-layer woodpile structures with an effective writing speed of 1 mm 2 in 90 s. The results demonstrate that combining multi-write-spot parallelization with simultaneous multi-plane writing provides a powerful route to significantly increase 2PP fabrication throughput.

We present density response estimators for Monte Carlo simulations that are based on a reweighting procedure, where the samples of an unperturbed system are used to estimate the properties of a system perturbed by an external harmonic potential. This allows the linear and nonlinear static density response to be estimated purely from simulations of the unperturbed system. The method is demonstrated for the uniform electron gas under warm dense matter and strongly coupled conditions using ab initio path integral Monte Carlo simulations. The performance of the method with respect to the number of particles and the number of imaginary time slices is investigated. The scheme is generalised to consider multiple external perturbations, acting on different species and with different wavenumbers, giving one access to additional cross-species density response functions and the complete quadratic response function resolved for both wave number arguments through mode coupling. The flexibility of the methodology opens the possibility to investigate numerous new density response properties to further advance our understanding of interacting quantum many-body systems across a broad range of applications.

We present a phase-only time-reversal framework for steering photonic nanojets without mechanical motion or amplitude modulation. Time-reversed radiation by a synthetic source placed at the target PNJ location helps define a phase-only modulation on a control line, compatible with a spatial light modulator, that produces the desired PNJ. Full-wave finite-difference frequency-domain (FDFD) simulations demonstrate robust lateral and axial steering with subwavelength confinement and low sidelobes. A parametric study of microelement geometries shows that nanojet formation is largely insensitive to moderate boundary variations, with simple shapes providing competitive performance. Robustness to fabrication and alignment errors is confirmed via uncertainty analysis.

Rain introduces broadband and frequency-selective attenuation in wideband terahertz (THz) links, making it necessary to identify a compact spectral descriptor that captures how the dominant loss region evolves with rainfall conditions. This article investigates the peak-frequency behavior of rain attenuation by combining Mie-theory calculations with one separable laboratory Gaussian drop-size distribution (DSD) and seven outdoor empirical DSD models whose spectral shapes vary with rainfall rate. The analysis compares total-loss, absorption, and scattering components, examines the roles of characteristic DSD scale and representative drop-size statistics, and evaluates the effect of temperature on the peak location. The results show that, unlike the fixed-shape laboratory case where the peak frequency remains unchanged with rainfall rate, all outdoor empirical DSD models exhibit a monotonic migration of the attenuation peak toward lower frequencies as rainfall rate increases; this behavior is well described by an asymptotic power-law relation and is governed primarily by the rainfall-dependent DSD characteristic scale rather than by total drop concentration or fixed-temperature dielectric dispersion.

Direct orientation contrast imaging of zinc-blende III-V materials is studied using scanning electron microscopy. A quantitative approach is taken using a 3 μm thick orientation-patterned GaP grown on GaAs sample, studying the anti-phase domain contrast with respect to the electron beam energy and the tilt angle. A qualitative approach is taken for III-V grown on non-polar materials with and without chemical mechanical polishing. Finally, a processing of the acquired image for GaP on Si reveals in plane preferential anti-phase boundaries.

Flow and transport in fractured geological media are strongly controlled by aperture heterogeneity and uncertainty in subsurface characterisation, yet most upscaling approaches rely on deterministic representations of fracture permeability. This study presents a scalable probabilistic workflow that bridges image-based fracture geometry and uncertainty-aware hydraulic predictions across scales. The approach integrates Bayesian correction of aperture-permeability model misspecification, a deep learning surrogate for predicting spatially distributed permeability statistics, and Darcy-scale flow upscaling to propagate uncertainty to effective transmissivity. The workflow is applied to natural shear fractures from core material in the Little Grand Wash Fault damage zone (Utah) and to simplified geometries derived from the same datasets. The Bayesian component quantifies uncertainty due to measurement errors and imperfect constitutive relations, while a Residual U-Net learns the effects of local heterogeneity and spatial correlation on predicted permeability uncertainty. Together, these components generate ensembles of permeability fields that are subsequently upscaled to probabilistic macroscopic flow responses. Results show that common empirical aperture-permeability relations are systematically biased for natural fractures, whereas the proposed probabilistic workflow yields uncertainty-aware permeability estimates consistent with physics-based behaviour. The method captures the impact of channelisation, connectivity, and complex 3D void geometries on transmissivity while quantifying the resulting uncertainty bounds. Computational efficiency arises from the proposed hybrid strategy for probabilistic upscaling, which combines physics-informed and data-driven approaches, preserves Stokes-flow consistency and supports uncertainty propagation without repeated high-fidelity simulations.

The strategic placement of bike-sharing infrastructure shapes urban accessibility and mobility outcomes. However, station-allocation approaches vary in their assumptions and decision logic. This study examines how alternative modelling paradigms prioritise urban space when applied to the same planning problem in Trondheim, Norway. We developed a unified analytical framework to compare three location-allocation approaches: weighted linear combination (WLC), maximal covering location problem (MCLP), and a data-driven suitability score based on exogenous spatial features (SSE). Each model designs a 68-station bike-sharing network from scratch using the same 24 spatial features and hierarchical weighting scheme. The resulting configurations are compared with the existing network, and consensus-based synthesis identifies 12 priority locations for expansion. The findings reveal systematic differences in spatial prioritisation across modelling approaches. WLC achieves the strongest coverage of population and transit demand, MCLP produces the widest spatial distribution prioritising geographic reach, and SSE balances demand intensity with accessibility. All model-derived configurations diverge from the existing network, highlighting the influence of historical and institutional factors on real-world deployment. Consensus synthesis identifies 12 expansion sites characterised by multimodal integration potential, underserved residential clusters, and high latent demand. This analysis demonstrates that methodological choices fundamentally shape spatial decision-support outcomes. By systematically evaluating classical optimisation and data-driven approaches under controlled conditions, the study provides evidence-based recommendations for bike-sharing network expansion and clarifies the strengths and limitations of alternative analytical frameworks for location-allocation planning.

We comment on Ref.[D. Karlovets, D. Grosman, and I. Pavlov, Phys. Rev. Lett. 136, 085002 (2026)], which proposes a BMT-like equation for the mean kinetic orbital angular momentum (OAM) of vortex particles in accelerator fields and draws spin-like conclusions about depolarization, resonances, and control. We show that the proposed closure is not generally valid even at the mean-value level. In the authors' own homogeneous-field model, Eq.(8) already makes $\langle L_z\rangle$ depend on the packet second moment $\langle ρ^2\rangle(τ)$; for an exact family of breathing Landau/LG packets this yields an explicit oscillation incompatible with Eq.(9) except in the nongeneric matched case. Moreover, the Appendix A assumption that mixed correlators are negligible suppresses the transverse kinetic-OAM components themselves, since those correlators are precisely the building blocks of $L_x$ and $L_y$. We also stress that, even if a closed equation for $\langle \hat{\mathbf L}\rangle$ were available, it would still not constitute a transport equation for a vortex quantum state. Mean-OAM transport does not determine OAM spectra, inter-mode coherences, or fidelity. State-level claims therefore require a mode-resolved density-matrix treatment rather than an Ehrenfest equation for a low-order moment.

Orbital degrees of freedom offer a largely untapped route to emergent dynamical phenomena in correlated quantum materials. However, it remains unclear whether collective orbital states can intrinsically generate both reactive and memory functionalities in a bulk system. Here we show that in the ferrimagnet Mn3Si2Te6, nonequilibrium reconfiguration of chiral orbital currents produces both emergent inductance and nonvolatile memristance as intrinsic properties of a single crystal. At low frequency and under a magnetic field along the c axis, coherent orbital-current domains generate robust clockwise inductive I-V loops. At higher frequency and low field, current-driven first-order reconfiguration leads to incomplete reversal and metastable trapping, producing an intrinsic electromotive force and a finite remanent voltage at zero current. These results establish orbital currents as a class of quantum state variables that encode both reactive and memory functionalities, opening routes toward intrinsically reconfigurable and energy-efficient electronic systems.

Time-varying media, characterized by dynamic or spacetime-modulated constitutive parameters such as permittivity and permeability, have recently emerged as a transformative paradigm for advanced wave control, transcending the constraints imposed by temporal translation symmetry and energy conservation in static systems. By incorporating time as an active degree of freedom, such media unlock unique phenomena including broadband frequency conversion, temporal refraction, significant field enhancement, and magnet-free nonreciprocity. These capabilities are reshaping the landscape of photonic technologies, enabling groundbreaking applications such as broadband nonreciprocal amplifiers, non-resonant lasers, and highly efficient particle accelerators. This review systematically classifies time-varying media based on their modulation schemes and elucidates the underlying physical principles and distinctive wave-matter interactions. We comprehensively survey the latest advances in this rapidly evolving field, highlighting exotic wave behaviors and practical implementations across electromagnetic and photonic systems. Furthermore, we summarize experimental platforms that realize time-varying responses across different frequency regimes. Finally, we assess the current state of progress, identify key challenges, and offer a forward-looking perspective on future research directions in this dynamic and promising area.

Nonlinear frequency conversion unlocks technologies ranging from telecommunications to quantum computation; however, weak nonlinearities and architectures that resist miniaturization currently limit devices. Here, we combine a bandstructure-engineered GaAs/AlGaAs heterostructure with a high quality factor dielectric metasurface to simultaneously tailor the intrinsic nonlinear susceptibility and optimize the electromagnetic field within the heterostructure. By engineering a resonant interband transition, we realize a large second-order nonlinear tensor element, 1.6 nm/V at 1.57 um wavelength. We then make it free-space-accessible and boost the effective nonlinearity to ~ 14 nm/V using a metasurface patterned on the material. Our proof-of-concept experiment establishes that interband transition engineering and metasurfaces accessing otherwise unusable nonlinear tensor elements enable giant effective nonlinearities in the near-infrared to visible spectrum. This addresses material and device-level constraints in nonlinear photonics, providing a scalable route to compact, efficient devices.

Porous media containing cracks, fractures, or internal discontinuities arise throughout subsurface geomechanics, biomechanics, and materials science. Numerical simulation of the coupled hydromechanical response is inherently challenging because the pressure and displacement fields are tightly coupled through the Biot equations, requiring stable mixed formulations. These difficulties are compounded when cracks are present, because standard mesh-conforming approaches require costly, labor-intensive, body-fitted meshing, while unfitted methods often require cut-cell integration, enrichment functions, or additional stabilization. In this work, we use an alternative approach, we adapt the shifted interface method to coupled transient poroelasticity with embedded interfaces. The method replaces the true crack by a surrogate approximation where interface conditions are transferred through local expansions. A unified derivation yields shifted forms for both hydraulic transmission and mechanical traction coupling. Two enforcement strategies are extensively compared: a weak (integral) enforcement and a strong (pointwise) enforcement. Four test cases of increasing geometric complexity (offset mesh-aligned, boundary-intersecting angled, embedded angled, and multi-crack configurations) validate the formulation. Away from crack tips, interface residuals converge as O(h); near tips, localized post-processing artifacts degrade the global rate, but first-order convergence is recovered when a small tip region is excluded. A multi-crack demonstration with four simultaneously embedded cracks of distinct geometry and interface properties confirms the practical applicability of the framework. These results support the shifted interface method as a practical framework for poroelastic crack modeling on non-body-fitted meshes with geometrically complex embedded interfaces.

Thermal boundary conductance (TBC) at dissimilar interfaces imposes a fundamental limit on electronic device performance, yet predicting and understanding heat transport across realistic, disordered boundaries remains elusive. Here, we develop a computational framework that combines machine-learned interatomic potentials with lattice dynamics to address the long-standing problem of how interfacial structure, from disordered to atomically sharp, affects thermal transport in the technologically important $β$-Ga$_2$O$_3$/4H-SiC heterostructure. By explicitly accounting for phonon wave-particle duality, we show that interfacial disorder introduces additional interfacial phonon modes that facilitate vibrational impedance matching between the two highly dissimilar crystals, yet it simultaneously disrupts interfacial phonon coherence and limits the potential heat-transport benefit. Our atomistic simulations further indicate that restoring atomic-scale order preserves coherence and yields markedly higher conductance. These insights motivate the controlled epitaxial growth of $β$-Ga$_2$O$_3$/4H-SiC heterostructures with systematically tuned interfacial order. Experimental measurements validate our predictions, achieving a record-high TBC of 231 MW m$^{-2}$ K$^{-1}$ at atomically sharp junctions. Beyond the immediate implications for $β$-Ga$_2$O$_3$-based power electronics, our results establish the preservation of interfacial phonon coherence as an effective strategy for mitigating thermal bottlenecks in mismatched systems.

Recent studies reveal striking representational alignment between artificial neural networks (ANNs) and biological brains, leading to proposals that all sufficiently capable systems converge on universal representations of reality. Here, we argue that this claim of Universality is premature. We introduce the Umwelt Representation Hypothesis (URH), proposing that alignment arises not from convergence toward a single global optimum, but from overlap in ecological constraints under which systems develop. We review empirical evidence showing that representational differences between species, individuals, and ANNs are systematic and adaptive, which is difficult to reconcile with Universality. Finally, we reframe ANN model comparison as a method for mapping clusters of alignment in ecological constraint space rather than searching for a single optimal world model.

The parameters of many classes of birth-death processes cannot be inferred uniquely from phylogenetic trees: infinitely many parameter combinations yield the same distribution of phylogenetic trees. Here, we show that parameter identifiability can be recovered even for the most general cases of time-dependent rates when additional information on hidden birth events along branches of the reconstructed tree is available. This holds both for models in which individuals are sampled at a single point in time or through time at a time-dependent rate. Moreover, we prove that when mutations occur at birth - assuming two different models for the accumulation of mutations at a birth event - then information about hidden birth events is available in the sequences and thus all parameters of time-dependent birth-death models become identifiable. Thus, phylodynamic inference is identifiable whenever evolutionary models with mutation accumulation at birth (such as at speciation, transmission, or cell division) are plausible.

How much data is enough to make a scientific discovery? As biomedical datasets scale to millions of samples and AI models grow in capacity, progress increasingly depends on predicting when additional data will substantially improve performance. In practice, model development often relies on empirical scaling curves measured across architectures, modalities, and dataset sizes, with limited theoretical guidance on when performance should improve, saturate, or exhibit cross-over behavior. We propose a scaling-law framework for cross-modal discoverability based on spectral structure of data covariance operators, task-aligned signal projections, and learned representations. Many performance metrics, including AUC, can be expressed in terms of cumulative signal-to-noise energy accumulated across identifiable spectral modes of an encoder and cross-modal operator. Under mild assumptions, this accumulation follows a zeta-like scaling law governed by power-law decay of covariance spectra and aligned signal energy, leading naturally to the appearance of the Riemann zeta function. Representation learning methods such as sparse models, low-rank embeddings, and multimodal contrastive objectives improve sample efficiency by concentrating useful signal into earlier stable modes, effectively steepening spectral decay and shifting scaling curves. The framework predicts cross-over regimes in which simpler models perform best at small sample sizes, while higher-capacity or multimodal encoders outperform them once sufficient data stabilizes additional degrees of freedom. Applications include multimodal disease classification, imaging genetics, functional MRI, and topological data analysis. The resulting zeta law provides a principled way to anticipate when scaling data, improving representations, or adding modalities is most likely to accelerate discovery.

Background: Three-dimensional dual-energy X-ray absorptiometry reconstructs three-dimensional maps of the proximal femur's density distribution from standard hip scans, enabling the estimation of trabecular and cortical bone parameters. The aim of this study was to assess the agreement of these three-dimensional cortical and trabecular femur parameters across different series and models of Hologic densitometers. Methodology: The study cohort was composed of 103 women and men recruited from four clinical centers in Spain and France. Subjects had duplicated hip scans using different Hologic scanners from the Horizon, Discovery, and QDR4500 series. Analyses were performed using 3D-Shaper software. Inter-scanner agreement was evaluated using Deming regression and Bland-Altman analysis. Results: The parameters demonstrated strong inter-device agreement across all clinical centers and scanner models, with coefficients of determination greater than 0.91. Absolute biases were less than 2.5 mg$/$cm$^3$ for integral volumetric bone mineral density, less than 2.9 mg$/$cm$^3$ for trabecular volumetric bone mineral density, and less than 1.7 mg$/$cm$^2$ for cortical surface bone mineral density. No statistically significant bias was found between parameters obtained from different scanners. Furthermore, the observed bias was lower than the expected least significant change, indicating that inter-scanner variability across these devices is not clinically significant. Conclusions: This study demonstrated excellent agreement for standard and three-dimensional derived bone parameters at the hip across Hologic densitometers. These findings support their suitability for clinical use.

Cortical folding reflects coordinated neurodevelopmental processes and provides a sensitive marker of neurological disease. In juvenile myoclonic epilepsy (JME), structural abnormalities are subtle and spatially distributed, limiting the sensitivity of conventional morphometric measures such as cortical thickness. We introduce a Poisson flow model derived from gradients of the mean curvature field on the cortical surface. The method yields a smooth scalar field obtained from a Poisson equation, whose surface gradient defines a flow representation of folding organization. This representation enables spatially coherent characterization of sulcal--gyral patterns and provides a principled geometric framework for studying distributed cortical alterations in JME.

Classical causal models, such as Granger causality and structural equation modeling, are largely restricted to acyclic interactions and struggle to represent cyclic and higher-order dynamics in complex networks. We introduce a causal framework grounded in a variational principle, interpreting causality as directional energy flow from high- to low-energy states along network connections. Using Hodge theory, network flows are decomposed into dissipative components and a persistent harmonic component that captures stable cyclic interactions. Applied to resting-state fMRI connectivity, our variational framework reveals robust cyclic causal patterns that are not detected by conventional causal models, highlighting the value of variational principles for causality.

Fitness landscapes are mappings between genotypes, phenotypes, and fitness that shape evolution. In recent years, empirical work and theoretical models have greatly advanced our understanding of how populations navigate rugged fitness landscapes. Here, we provide a timely review of this field. Its rapidly growing literature employs a wide range of terms, which are sometimes used ambiguously or inconsistently. We therefore begin by defining the major concepts and the field's vocabulary, highlighting our own terminology choices wherever needed. We then review key results on the relationships between epistasis, ruggedness, accessibility, and navigability for genotype-fitness maps, highlighting several complex and sometimes counterintuitive connections that have emerged. Further, we review how the conserved structural properties of the underlying genotype-phenotype map -- that leads to the formation of large connected neutral networks of genotypes -- influence dynamics on fitness landscapes. We then compare the two levels to study landscape navigation -- the level of the genotype-phenotype maps and the level of genotype-fitness maps. Our review leads us to propose a new measure of navigability, based on evolutionary outcomes, that is broadly applicable and overcomes limitations of existing measures. Finally, we review the smaller body of work that relaxes the common assumption of fitness-monotonic paths on static landscapes, and discuss how this can fundamentally change the nature of fitness landscape navigation. Throughout the review, we identify directions for future work to fill existing gaps and to synthesize the disparate strands of research within the field.

Linear-threshold networks (LTNs) capture the mesoscale behavior of interacting populations of neurons and are of particular interest to control theorists due to their dynamical richness and relative ease of analysis. The aim of this paper is to advance the study of global asymptotic stability in LTNs with asymmetric neural interactions and heterogeneous dissipation under the structural Lyapunov diagonal stability (LDS) condition. To this end, we introduce a one-parameter family of LTNs that preserves the LDS condition and has a parameter-independent equilibrium set. In the fast limit, this family converges to a projected dynamical system (PDS), while in the slow limit, it converges to a discontinuous hard-selector system (HSS). Under LDS, we prove that the fast PDS limit is globally exponentially stable and that the HSS limit is globally asymptotically stable. This alignment suggests that the limiting systems capture essential mechanisms governing stability across the entire LTN family. Together with numerical evidence, these findings indicate that resolving stability at the fast and slow endpoints provides a promising and structurally grounded path toward establishing global stability for LTNs with biologically plausible recurrence and diagonal dissipation.

Genome engineering has achieved remarkable sequence-level precision, yet predicting the transcriptomic state that a cell will occupy after perturbation remains an open problem. Single-cell CRISPR screens measure how far cells move from their unperturbed state, but this effect magnitude ignores a fundamental question: do the cells move together? Two perturbations with identical magnitude can produce qualitatively different outcomes if one drives cells coherently along a shared trajectory while the other scatters them across expression space. We introduce a geometric stability metric, Shesha, that quantifies the directional coherence of single-cell perturbation responses as the mean cosine similarity between individual cell shift vectors and the mean perturbation direction. Across five CRISPR datasets (2,200+ perturbations spanning CRISPRa, CRISPRi, and pooled screens), stability correlates strongly with effect magnitude (Spearman $ρ=0.75-0.97$), with a calibrated cross-dataset correlation of 0.97. Crucially, discordant cases where the two metrics decouple expose regulatory architecture: pleiotropic master regulators such as CEBPA and GATA1 pay a "geometric tax," producing large but incoherent shifts, while lineage-specific factors such as KLF1 produce tightly coordinated responses. After controlling for magnitude, geometric instability is independently associated with elevated chaperone activation (HSPA5/BiP; $ρ_{partial}=-0.34$ and $-0.21$ across datasets), and the high-stability/high-stress quadrant is systematically depleted. The magnitude-stability relationship persists in scGPT foundation model embeddings, confirming it is a property of biological state space rather than linear projection. Perturbation stability provides a complementary axis for hit prioritization in screens, phenotypic quality control in cell manufacturing, and evaluation of in silico perturbation predictions.

Tone injection (TI) is a promising distortionless PAPR reduction technique that incurs no spectral efficiency loss. However, state-of-the-art TI schemes based on random candidate generation or clipping noise spectrum suffer from fundamental limitations in PAPR performance. In this paper, we propose novel TI schemes compatible with both OFDM and AFDM systems. The proposed schemes iteratively update the TI sequence via a candidate ranking procedure guided by time-domain local peaks. This accurately selects effective candidates while achieving a complexity comparable to that of the fast Fourier transform. Depth-first search is further integrated to enhance PAPR performance by exploiting the tree structure of the process. Simulations demonstrate that the proposed schemes achieve over 1 dB PAPR gain over baseline TI schemes at comparable complexity. The gain is consistent across various numbers of subcarriers under controlled per-iteration complexities, confirming a superior performance-complexity trade-off for both OFDM and AFDM.

This paper develops a Doppler-aware sensing framework for cell-free massive MIMO (CF-mMIMO) networks operating under OFDM-based integrated sensing and communication (ISAC). The framework explicitly incorporates the 3D-bistatic Doppler geometry across distributed access points (APs) into a generalized likelihood ratio test (GLRT) detector. To address the scalability, a user-target-centric AP association approach is utilized. The 3D tangential components of the target's velocity vector are estimated, and several search and optimization strategies, including coarse grid search, gradient-based refinement, and particle swarm optimization (PSO), are developed and evaluated. The Doppler-aware GLRT statistic and receive sensing signal-to-noise ratio (SNR) are derived. Simulation results demonstrate that the proposed PSO-aided detector achieves the most favorable accuracy-complexity trade-off, while Doppler mismatch can cause substantial sensing-SNR degradation in high-mobility scenarios. Additionally, leveraging more OFDM subcarriers enhances frequency-domain diversity and yields further sensing-SNR gains.

Near-field propagation is often unavoidable at terahertz (THz) frequencies due to the large apertures needed for sufficient array gain, yet near-field operation complicates practical system design, especially under user mobility. This paper asks whether a mobile THz link can remain broadband, achieve the desired high rates and coverage, while operating exclusively in the radiative far field. To answer this question, we develop a proof-by-contradiction feasibility framework that jointly enforces (i) a far-field requirement based on the Fraunhofer distance and (ii) a reliability requirement specified by a target SNR at the worst-case link distance. We derive closed-form upper bounds on the far-field-feasible bandwidth for stationary and mobile links. We further incorporate practical misalignment through several UE rotation and mobility scenarios. Numerical results show that stationary THz links can remain far-field-only with physically realizable apertures while supporting extremely large bandwidths, whereas practical mobile THz systems cannot. In practically relevant mobile THz access settings, the far-field-feasible bandwidth becomes a severe limiting factor: achieving tens-of-GHz targets would require unrealistically high UE transmit power. A cross-band comparison further shows that far-field-only operation is largely attainable at sub-6~GHz and, to a significant extent, at mmWave for moderate bandwidths, while near-field-aware designs become essential for mobile THz access.

As 6G advances, ubiquitous connectivity and higher capacity requirements of the air interface pose substantial challenges for accurate and real-time wireless channel acquisition in diverse environments. Conventional statistical channel modeling relies on offline measurement data from limited environments, struggling to support online applications facing diverse environments. To this end, the digital twin channel (DTC) has emerged as a novel paradigm that constructs a digital replica of the physical environment through high-fidelity sensing and predicts corresponding channel in real time utilizing artificial intelligence (AI) models. As the engine of DTC, existing AI models struggle to simultaneously achieve strong environmental generalization in real-world and end-to-end channel prediction for real time tasks. Therefore, this paper proposes a channel large model (ChannelLM)-driven DTC architecture comprising three modules: low-complexity and high-accuracy environment reconstruction based on dynamic object detection and multimodal alignment of image and point cloud data, physically interpretable environment feature extraction, and a ChannelLM core to mapping these features into generalized environment representations for multi-task channel prediction. Simulation results demonstrate that, in unseen test environments, compared with small-scale AI models, ChannelLM reduces prediction errors by 4.23 dB in channel state information prediction while achieving an end-to-end inference latency of 70 milliseconds in the real world.

Reconfigurable intelligent surface (RIS) technology is a promising enabler for next-generation (NextG) wireless systems, capable of dynamically shaping the propagation environment. Integrating RIS within the open radio access network (O-RAN) architecture enables flexible and intelligent control of wireless links. However, practical RIS-assisted operation requires efficient acquisition and reporting of channel state information (CSI) to support real-time control from the base station side. This paper proposes a CSI reference signal (CSI-RS)-based reporting scheme for downlink complex channel information (CCI) to facilitate RIS optimization in an O-RAN-compliant environment. The proposed framework establishing CCI extraction and CSI-RS reporting procedures is experimentally validated on a real-world testbed integrating an open-source O-RAN system with an RIS prototype operating in the n78 frequency band. Existing channel estimation-based RIS optimization algorithms, including Hadamard and orthogonal matching pursuit (OMP), are tailored for integration into the O-RAN architecture. Experimental results demonstrate notable improvements in received signal power for both near and far users, highlighting the effectiveness and practical viability of the proposed scheme.

Determining the optimal phase configurations of reconfigurable intelligent surface (RIS) elements typically requires complex channel estimation procedures with high pilot overhead, creating a bottleneck for real-time deployment in time-varying wireless environments. In this paper, we propose a digital twin (DT)-driven framework for RIS phase shift optimization that eliminates extensive signaling overhead associated with estimating high-dimensional RIS channels. Leveraging the NVIDIA Sionna ray-tracing library, we construct a DT of the physical environment based on a three-dimensional map. The proposed system utilizes the location information of the transceivers to compute the optimal RIS phase shift configurations within the DT. These computationally generated configurations are then transferred to a physical RIS prototype. Experimental results demonstrate that the phase configurations obtained from the DT significantly enhance the received signal power in the physical environment, validating the fidelity of the ray-tracing model and the feasibility of the proposed optimization strategy.

In practical affine frequency division multiplexing (AFDM) systems, the intricate coupling of oscillator phase noise (PN) and off-grid fractional shifts traps conventional estimators in a severe high-SNR error floor. To address these challenges, we propose a joint PN and channel estimation method based on sparse Bayesian learning (JPNCE-SBL). Specifically, a reduced-rank subspace projection is first introduced to capture the dominant eigen-energy of the Wiener PN process. Concurrently, a dynamic grid evolution strategy is designed to iteratively eliminate off-grid errors without requiring computationally prohibitive global grid densification. Both components are integrated into a unified Expectation-Maximization (EM) framework, where the channel and PN estimates are jointly updated at each iteration to prevent error propagation. Simulation results demonstrate that JPNCE-SBL significantly outperforms existing benchmarks in both NMSE and BER, closely approaching the perfect channel state information case under practical PN conditions.

Millimeter-wave (mmWave) technology is a crucial enabler for next-generation networks because it offers substantially greater available bandwidth. mmWave multiple-input multiple-output (MIMO) systems cannot rely solely on fully digital precoding due to hardware costs. As a result, hybrid precoding, which combines digital baseband processing with RF precoding, has emerged as a practical solution that balances performance and implementation complexity. As mmWave links typically operate over wideband, frequency-selective channels, orthogonal frequency-division multiplexing (OFDM) is commonly used to mitigate dispersive effects, yet OFDM introduces practical drawbacks, including out-of-band (OOB) emissions from abrupt spectral transitions among subcarriers and additional spectral leakage induced by windowing. Moreover, nonideal phase shifters (PS) in the RF transmit precoder and the user combiner impose inherent implementation limits that result in phase errors. We investigate robust joint digital--RF precoder design for minimizing the downlink sum mean-squared error (MSE) in hybrid multi-user (MU) MIMO--OFDM systems subject to maximum transmit-power, clipping, and OOB spectral-mask constraints. The resulting optimization is nonconvex and challenging to solve. To address this, we develop a minimum mean-squared error (MMSE) based block coordinate descent (BCD) algorithm that alternates between updating the transmitter-side digital--RF precoders and the user-side digital--RF combiners. For each BCD subproblem, we propose computationally efficient and scalable, closed-form solution strategies suitable for practical implementation. Extensive simulations validate the proposed methods and show clear performance improvements over established benchmark schemes.

Terahertz (THz) communications have emerged as a key technology for escalating data rates in future generation wireless networks. However, severe propagation losses at THz frequencies pose significant challenges, which can be mitigated via ultra-massive multiple-input multiple-output (UM-MIMO) systems employing highly directional transmissions. To this end, codebook-based analog beamforming constitutes a realistic solution, eliminating the need for explicit channel estimation. However, in UM-MIMO systems, the use of extremely narrow beams makes beam training and alignment increasingly challenging, leading to a substantial increase in the number of codewords to be tested and, thus, to high computational complexity. In this paper, a novel artificial neural network architecture for low-complexity beam training in UM-MIMO THz systems is presented, which does not require a constant feedback link between transmitter and receiver to obtain the best beamformer and combiner pair. An inception and residual network, which is trained based on the received signal powers using the transmit and receive codewords generated from predefined hierarchical codebooks, is designed. Our numerical investigations demonstrate that the proposed machine learning approach significantly reduces the complexity of UM-MIMO transmit and receive beamforming design, as compared to the standard exhaustive and hierarchical beam searching methods.

The accelerating advancement of intelligent transportation systems has established consumer-oriented vehicular networks (CVNs) as a critical infrastructure for next-generation connected mobility. However, the high mobility of vehicular users (VUs) introduces significant channel state information (CSI) uncertainty, which severely undermines the performance of conventional fixed-position antenna systems. To address this, this paper explores the deployment of movable-antennas (MAs) to enhance communication robustness in CVNs under imperfect CSI conditions. We develop a joint optimization framework that dynamically coordinates the spatial positioning of MAs and transmit beamforming at the base station, with the objective of maximizing the worst-case sum rate across all VUs. The problem is formulated as a non-convex max-min optimization problem, subject to bounded CSI estimation errors, transmit power limits, and physical constraints on antenna displacement. By adopting an alternating optimization strategy, the original problem is decomposed into tractable subproblems, solved via techniques including the S-Procedure, Schur complement, and successive convex approximation. Numerical evaluations confirm that the proposed approach achieves substantial gains over existing benchmarks in terms of worst-case throughput.

Low-altitude communication networks (LACNs) serve as the critical infrastructure of the emerging low-altitude economy (LAE), supporting services such as drone delivery and infrastructure inspection. However, LACNs operate in highly dynamic three-dimensional (3D) environments characterized by high mobility and predominantly line-of-sight (LoS) propagation, creating strong coupling among key performance objectives including coverage, interference mitigation, handover management, and sensing capability. Isolated tuning of individual objectives cannot capture these cross-objective interactions, rendering conventional approaches based on experience-driven tuning and repeated field trials inefficient and costly. To address these challenges, we propose DT-MOO, a Digital Twin-based Multi-Objective Optimization framework for LACNs. By constructing a high-fidelity virtual replica that integrates realistic environmental models, electromagnetic (EM) propagation, and traffic dynamics within a unified environment, DT-MOO enables joint evaluation and systematic optimization of interdependent network parameters, scoring candidate configurations by their combined effect on multiple objectives. As the foundational validation of the framework, we report real-world experiments in a 5G-enabled LACN focusing on coverage-interference co-optimization, where DT-MOO increases the high-quality coverage rate from 14.0% to 52.9% across all evaluated altitudes compared to an operator-provisioned, experience-based baseline, while achieving a net SINR gain under stringent criteria despite local spatial trade-offs, confirming its ability to handle coupled objectives in practical LACN deployment.

This paper explores the integration of reconfigurable intelligent surfaces (RISs) into cell-free massive multiple-input-multiple-output (CF-mMIMO) networks operating in FR1 and FR3 frequency bands. We present a comprehensive framework for analyzing RIS-assisted CF-mMIMO systems under realistic propagation conditions, accounting for frequency-dependent characteristics and RIS configurations. A novel RIS-user association algorithm is proposed to optimize phase-shift settings by assigning each RIS to a single user based on line of sight (LoS) connectivity. The system model incorporates spatially correlated Ricean fading channels and employs scalable partial-minimum mean square error (P-MMSE) combining. The numerical results demonstrate that the proposed RIS-user selection strategy significantly improves the spectral efficiency compared to random or exhaustive RIS configurations, particularly when the number of RISs is moderate. We also analyze the trade-off between training overhead and performance gains, showing that excessive pilot requirements can offset benefits when RIS density or element count increases. The results highlight the potential of the FR3 bands for RIS-assisted CF-mMIMO, provided advanced channel estimation techniques are adopted to mitigate overhead. These findings emphasize the importance of intelligent RIS-user pairing and scalable estimation methods for future 6G deployments.

This paper studies scalable conjugate beamforming (CB) variants for physical-layer multicasting in cell-free massive multiple-input multiple-output (CF-mMIMO) systems. Focusing on fully distributed precoding, we analyze classical CB, normalized CB (NCB), and enhanced CB (ECB) within a subgroup-centric multicast framework. Multicast users are partitioned into subgroups based on large-scale fading similarity, which enables composite channel estimation, pilot reuse, and distributed precoding with low complexity. The performance of the different CB variants is evaluated in terms of aggregated spectral efficiency (ASE) under representative user geometries, including uniformly distributed users, spatially clustered deployments, and heterogeneous scenarios combining hotspots with more dispersed users. Monte Carlo simulations reveal a strong spatial geometry-dependent behavior: unicast transmission is preferable in uniform deployments, while subgroup-based multicasting becomes essential in clustered and heterogeneous scenarios. Among the CB-based precoders, NCB offers a robust performance-complexity trade-off across most scenarios, whereas ECB provides additional gains only when sufficient channel hardening is present. These results provide practical insights into the selection of low-complexity distributed precoders and multicast transmission modes in CF-mMIMO systems supporting broadband and multimedia services.

This paper develops an active sensing framework for designing the transmit and receive beamformers of a multiple-input multiple-output (MIMO) radar system. In the proposed technique, the beamformers are adaptively designed in each sensing stage based on the measurements made in the previous sensing stages. The beamformers are determined by minimizing the Bayesian Cram{é}r-Rao bound (BCRB) for the estimation of the unknown sensing parameters at each stage via Lagrangian dual optimization. To address the exploration-exploitation tradeoff that is inherent to such an adaptive design, this paper proposes two variants of the BCRB optimization problem: an exploration-centric variant, that ensures that multiple orthogonal beamforming directions are probed in each sensing stage, and an exploitation-centric variant, that does not restrict the number of optimal beamformers. Each variant of the optimization problem is solved via an alternating optimization algorithm that alternates between solving for the transmit beamformers and solving for the receive beamformers. The algorithm is shown to converge to a stationary point provided that each optimization problem is solved to global optimality. Moreover, this paper studies each of the two BCRB optimization sub-problems in the Lagrangian dual domain and shows that despite the non-convexity, global optimality is guaranteed provided that certain sufficient conditions hold. The conditions pertain to the multiplicity of the eigenvalues of a specific direction matrix that can be analytically written in terms of the optimal dual variables. These conditions further imply the tightness of the semidefinite relaxation of the optimization problems. Simulation results demonstrate the benefits of the proposed BCRB-based design compared to state-of-the-art adaptive beamforming strategies.

Spatial correlation poses a significant challenge in massive multiple-input multiple-output (MIMO) high-altitude platform station (HAPS) systems. The inherent spatial correlation among antenna elements on the HAPS induces high correlation and interference among users' channel gains. To mitigate this issue, we propose an integrated approach that combines spatial resource allocation and user clustering. In our proposed solution, we assign the same resource blocks to users with orthogonal channel gains, while users with non-orthogonal channel gains receive different resource blocks. Additionally, we propose a sectorized antenna architecture for the massive MIMO HAPS base station, specifically designed to directly transmit three-dimensional beams to users and reduce spatial correlation among antenna elements. This work addresses the joint optimization problem of power allocation and resource allocation to maximize the overall data rate of the massive MIMO HAPS system. Simulation results revealed the role of spatial resource allocation in managing spatial correlation and interference among users.

This paper addresses the problem of adaptive reconfigurable intelligent surfaces (RIS) configuration design for user localization in rich-scattering environment (RSE), where electromagnetic waves undergo multiple interactions with dynamic scatterers and RIS elements. We propose an adaptive learning-based localization approach for a distributed RIS-assisted network in a RSE using a bidirectional long-short term memory (biLSTM) model that captures temporal correlations between observations. The proposed approach actively senses the environment using sequential pilot transmissions from the base station (BS), accounting for scattering effects, and adaptively updates the RIS configuration based on prior measurements to eventually accurately estimate and minimize the user localization error. The proposed model comprises two neural sub-networks: Scattering Estimation Network (Bi-SEN), for estimation of scattering in the environment, and Adaptive RIS-Assisted User Localization Network (Bi-ARULN), for RIS configuration and localization. Bayesian optimization is used for hyperparameter tuning of the model. The simulation results demonstrate the effectiveness of the proposed approach, achieving significantly lower localization root mean squared error (RMSE compared to random configuration, prestored codebook look-ups, and adaptive baselines in both single-input-single-output (SISO) and multiple-input-multiple output(MIMO) RIS-assisted networks in RSE. The design is generalized across configurations and scales with RIS size and network dimensions. The results highlight the strong potential of RIS deployment and of the proposed approach to enable reliable location services in RSE.

This paper investigates an uplink user equipment (UE) location and orientation estimation problem in an indoor rich-scattering environment (RSE) for a multiple-input-multiple-output (MIMO) narrowband reconfigurable intelligent surfaces (RIS)-assisted communication system. The localization problem in RSE is challenging as the uplink pilot signal undergoes multiple interactions with the RIS and dynamic scattering objects (SOs). This paper proposes an approach where base station (BS) adaptively senses the environment with the help of RIS. Based on this sensing, it sequentially designs RIS configuration, BS beamforming and UE beamforming vectors, using the sequence of pilot transmissions from the UE to the BS, with an objective of progressively focusing them onto the UE. Towards this end, we train a bidirectional long-short term memory (biLSTM) network based controller to capture the temporal dependencies between measurements to first adaptively sense the RSE and then design RIS, BS and UE beamforming vectors to localize the UE. We evaluate the proposed approach under various RSE conditions such as various distributed RIS installations, varying number of randomly moving SOs and sensing RIS elements. Simulation results illustrate that it effectively enables adaptive sensing to achieve low localization error with robustness in various RSEs.

Radio map estimation from sparse measurements is fundamental to wireless network planning, optimization, and localized map updating. Most recent learning-based approaches formulate the problem as dense map completion over a predefined grid, whereas many practical deployments require estimating transmitter-specific received signal strength only at queried locations or refining an existing map after local changes. This paper proposes a physics-aware query-conditioned hierarchical graph attention network for transmitter-resolved point-wise radio map estimation. For each queried target--transmitter pair, the proposed encoder constructs a bounded local graph over sampled reference observations and aggregates reference-to-query evidence through transmitter-referenced geometric descriptors. A global graph then exchanges representation-level context among nearby target locations to improve neighborhood consistency without revisiting a large number of reference measurements. On top of this shared architecture, we instantiate three operating regimes: direct RSS estimation, prior-conditioned residual correction, and post-hoc gated attenuation of the learned correction. The framework uses only measurement-side quantities and does not rely on environment-side inputs. Simulations on the DeepMIMO scenario show that, in the direct regime, the proposed HGAT achieves the lowest RMSE and MAE among the evaluated learning-based baselines on all reported sites. When conventional prior estimate is available, the residual and gated regimes further reduce the prior error.

This paper investigates communication-efficient neural network transmission by exploiting structured symmetry constraints in convolutional kernels. Instead of transmitting all model parameters, we propose a degrees-of-freedom (DoF) based codec that sends only the unique coefficients implied by a chosen symmetry group, enabling deterministic reconstruction of the full weight tensor at the receiver. The proposed framework is evaluated under quantization and noisy channel conditions across multiple symmetry patterns, signal-to-noise ratios, and bit-widths. To improve robustness against transmission impairments, a projection step is further applied at the receiver to enforce consistency with the symmetry-invariant subspace, effectively denoising corrupted parameters. Experimental results on MNIST and CIFAR-10 using a DeepCNN architecture demonstrate that DoF-based transmission achieves substantial bandwidth reduction while preserving significantly higher accuracy than pruning-based baselines, which often suffer catastrophic degradation. Among the tested symmetries, \textit{central-skew symmetry} consistently provides the best accuracy-compression tradeoff, confirming that structured redundancy can be leveraged for reliable and efficient neural model delivery over constrained links.

Precise aerial radio environment characterization is vital for low-altitude planning. However, existing datasets and estimation methods lack the high-resolution granularity required for complex aerial spaces. Additionally, current schemes suffer from poor generalization and heavy reliance on environmental priors. To address these gaps, this paper introduces FARM, a pioneering foundation model for unified aerial radio map estimation. This model is supported by a newly curated, high-resolution dataset featuring multi-band and multi-antenna configurations specifically for low-altitude environments. FARM utilizes a masked autoencoder to extract deep latent representations of the aerial radio environment, which then guide a diffusion-based decoder to generate high-fidelity signal distributions through iterative refinement. Extensive experiments demonstrate that FARM significantly outperforms state-of-the-art benchmarks and exhibits superior generalization capabilities across unseen scenarios. Ultimately, FARM serves as a critical infrastructure for low-altitude economy by enabling autonomous aerial logistics and intelligent urban networking.

Time series forecasting is traditionally dominated by sequence-based architectures such as recurrent neural networks and attention mechanisms, which process all time steps uniformly and often incur substantial computational cost. However, real-world temporal signals typically exhibit heterogeneous structure, where informative patterns are sparsely distributed and interspersed with redundant observations. This work introduces \textbf{SPaRSe-TIME}, a structured and computationally efficient framework that models time series through a decomposition into three complementary components: saliency, memory, and trend. The proposed approach reformulates temporal modeling as a projection onto informative subspaces, where saliency acts as a data-dependent sparsification operator, memory captures dominant low-rank temporal patterns, and trend encodes low-frequency dynamics. These components are integrated through a lightweight, adaptive mapping that enables simplified, selective, and interpretable temporal reasoning. Extensive experiments on diverse real-world datasets demonstrate that SPaRSe-TIME achieves competitive predictive performance compared to recurrent and attention-based architectures, while significantly reducing computational complexity. The model is particularly effective in structured time series with clear temporal components and provides explicit interpretability through component-wise contributions. Furthermore, analysis reveals both the strengths and limitations of decomposition-based modeling, highlighting challenges in highly stochastic and complex multivariate settings. Overall, SPaRSe-TIME offers a principled alternative to monolithic sequence models, bridging efficiency, interpretability, and performance, and providing a scalable framework for time series learning.

This paper addresses the critical issue of spectrum scarcity and the need to support diverse services, including communication and learning tasks, by presenting a reconfigurable intelligent surface (RIS)-aided wireless network framework that integrates non-orthogonal multiple access (NOMA) with over-the-air federated learning (AirFL). The proposed system leverages the ability of RIS to adaptively shape wireless channels, aiming to enhance overall network performance for both communication and learning through concurrent uplink transmissions. To tackle critical challenges such as co-channel interference, imperfect channel state information (CSI), and successive interference cancellation (SIC), we develop an optimization framework that focuses on minimizing the optimality gap. This joint optimization is formulated as a non-convex problem, complicated by the intricate interactions between NOMA and AirFL users as well as the impact of imperfect CSI and SIC. To overcome these challenges and reduce the optimality gap, we reformulate the optimization problem as a Markov decision process and solve it using a long short-term memory deep deterministic policy gradient (LSTM-DDPG) algorithm, a memory-based approach within deep reinforcement learning (DRL). Simulation results demonstrate that the proposed approach achieves faster convergence, lower variance, and improved robustness under channel uncertainty, outperforming baseline DRL algorithms such as DDPG, soft actor-critic (SAC), and advantage actor-critic (A2C).

Near Field Communication (NFC) cards are widely used for identification, but their passive nature often limits the ability to incorporate additional security mechanisms. As a result, anyone holding the card may be incorrectly recognized as an authenticated user. To overcome this limitation, this paper presents a secure manual password input framework using a virtual PIN pad for passive NFC cards. Users input passwords by pressing designated regions on the card, which induces measurable impedance variations in the NFC antenna. These variations change the RF signals subtly, and a deep learning model is used to infer the intended password from the resulting signal patterns. A key challenge is that identical press interactions can produce significantly different responses across NFC cards, which yields unreliable recognition. To address this, we introduce a lightweight recognition approach that operates directly within the RF feature space at the penultimate layer of a temporal neural encoder. An adversarial domain-alignment module reshapes virtual PIN pad press-response embeddings into compact, card-invariant clusters, which enables stable and consistent recognition across heterogeneous cards. To support model training and evaluation, a reconfigurable software-defined radio (SDR) testbed is developed, and PIN pad press-response data are collected from commercially available ISO/IEC 15693 cards. Recognition is performed using a Mahalanobis distance metric derived from a calibration-based covariance model that captures feature correlations. Experimental results show that the proposed system achieves a 98.20\% recognition acceptance rate and remains robust under substantial noise degradation. The framework is fully card-agnostic and can be seamlessly integrated into existing NFC infrastructures.

In sixth-generation (6G) cellular networks and beyond, aerial platforms, such as uncrewed aerial vehicles (UAVs) and high-altitude platform stations (HAPS), are anticipated to play a crucial role in enhancing connectivity, expanding network coverage, and supporting advanced communication services. However, the deployment of energy-efficient onboard communication systems is essential for their widespread adoption and effectiveness. The integration of energy harvesting (EH) into aerial platforms is envisioned to be pivotal in promoting both energy and cost efficiency. In this paper, we propose a new paradigm for aerial platforms in which they can collect energy from the transmitted signals of nearby aerial platforms. The paper employs a two-tier architecture with HAPS super-macro base stations (HAPS-SMBS) system: regular HAPS-SMBS nodes serve as base stations, while a "mother" HAPS-SMBS node acts as a manager to coordinate communications between regular HAPS-SMBS and the ground station, thus enabling wireless energy transfer. Specifically, we analyze the characteristics of EH-enabled HAPS-SMBS and compare their performance with those without EH. Additionally, we derive the optimal regular HAPS-SMBS positioning to mitigate signal attenuation and power loss. Subsequently, we formulate a joint optimization problem for regular HAPS-SMBS positioning and the EH factor. We solve the problem using the iterative distance and EH factor algorithm (IDFA); however, we employ $Q$-learning to verify its effectiveness. Our findings indicate that, compared to conventional EH systems, IDFA and $Q$-learning exhibit higher data rate performance. In contrast, $Q$-learning outperforms IDFA systems in linear modelswith intensive training in approximating optimal values. Furthermore, maximizing transmit power achieves higher gains than systems without EH.

To meet the growing data traffic demand in future wireless systems, novel transmission architectures capable of adapting to complex propagation environments are required. Movable antenna (MA) systems have recently emerged as a promising approach, enabling the physical repositioning of antenna elements to exploit spatial degrees of freedom. However, existing studies largely rely on idealized or simplistic channel models, leaving open the question of whether the performance gains of MA systems persist under realistic propagation conditions. This paper investigates the performance of downlink multi-user MIMO systems with movable antennas using deterministic ray-traced channel models. A simulation framework combining three-dimensional ray tracing and field-response channel modeling is developed, and antenna positions are optimized using particle swarm optimization and genetic algorithms. Simulation results reveal that while simplified distance-based channel models predict large performance disparities between competing array configurations, realistic ray-traced channels significantly compress these differences, indicating that propagation effects dominate over pure array geometry optimization. Nevertheless, movable antenna systems retain strong effectiveness over conventional fixed arrays across different user distributions, array sizes, and multipath conditions, even in geometry-constrained propagation environments.

This study presents a physically informed hybrid time-frequency and machine learning (STFT-ML) framework for arc stability monitoring in electric arc welding systems. The primary current signal is modeled as a stochastic representation of plasma dynamics and transformed into a structured feature space using localized spectral energy distributions. Within this framework, the Arc Stability Index (ASI), spectral entropy (Hs), and harmonic distortion (THDarc) are defined as energy-based descriptors and integrated with complementary time-domain features to capture both spectral redistribution and temporal variability. Experimental evaluation demonstrates that the SVM-RBF classifier achieves a hold-out accuracy of 94.4%. However, cross-validation results (85.6% for Leave-One-Out and 87.5% +/- 9.4 for 10-fold) and a 95% confidence interval of [81.65%, 92.50%] provide a more realistic assessment of generalization performance. Receiver Operating Characteristic (ROC) and Precision-Recall (PR) analyses further confirm strong class separability, particularly for stable and extinction regimes, while transient states remain more challenging due to their non-stationary nature. Compared to high-dimensional deep learning approaches, the proposed framework significantly reduces computational complexity and inference latency, enabling real-time deployment in resource-constrained environments. The results indicate that spectral energy redistribution around the fundamental frequency serves as a reliable precursor to arc instability. The main contribution of this work lies in the development of a computationally efficient and physically interpretable feature representation framework that bridges time-frequency analysis and machine learning-based classification for industrial diagnostic applications.

Ensuring strict safety guarantees is the paramount challenge for emerging 5G/6G wireless systems, particularly as they increasingly govern mission-critical applications ranging from autonomous UAV swarms to industrial automation. While deep reinforcement learning (DRL) offers a promising solution for complex resource allocation, standard algorithms frequently violate essential constraints, such as QoS mandates and power limits, posing unacceptable risks of system failure and regulatory non-compliance. We propose Safe-Deep Q-Learning, a novel algorithm that simultaneously addresses all three challenges: it handles mixed-integer nonconvex problems by approximating the Q-function, adapts to stochastic dynamics, and enforces dual-timescale constraints using integrated Lagrangian methods. Our framework features adaptive penalty scaling and constraint violation tracking, specifically tailored for wireless environments, and is designed to operate in both distributed and centralized architectural modes. We prove convergence to optimal constraint-satisfying policies under mild conditions and demonstrate robustness through dual variable stabilization. Validation on unmanned aerial vehicle (UAV) swarm control network and post-disaster emergency communications applications shows that Safe-Deep Q-Learning achieves stringent adherence to safety bounds with near-zero violation rates, significantly outperforming existing constrained RL baselines, establishing its effectiveness for safety-critical wireless deployments.

Large Language Models have revolutionized natural language processing, achieving unprecedented success across a vast range of tasks. However, their practical application in long-context scenarios is severely hampered by the formidable memory footprint of the Key-Value cache. While channel pruning has emerged as a promising compression strategy, existing methods evaluate channel importance in isolation, fundamentally ignoring the inter-channel interactions that collectively dictate model performance. This oversight leads to suboptimal pruning decisions. To address this, we introduce \textbf{GRACE} (\textbf{GR}aph-guided \textbf{A}daptive \textbf{C}hannel \textbf{E}limination), a novel framework that reframes KV cache compression as a graph-based optimization problem. GRACE models channels as nodes and their interactions as weighted edges, enabling the identification of a near-optimal channel subset for pruning by minimizing the reconstruction error of the attention weight matrix. Furthermore, GRACE incorporates an adaptive protection mechanism that shields salient key channels from removal, ensuring a robust autoregressive decoding process. Extensive experiments show that GRACE can reduce KV cache size by 60\% with negligible performance degradation, consistently outperforming the state-of-the-art method.

Generative semantic communication (SemCom) harnesses pretrained generative priors to improve the perceptual quality of wireless image transmission. Existing generative SemCom receivers, however, rely on maximum a posteriori (MAP) estimation, which fundamentally cannot preserve the data distribution and thus limits achievable perceptual quality. Moreover, current diffusion-based approaches using single-domain guidance face significant limitations: latent-domain guidance is sensitive to channel noise, while image-domain guidance inherits decoder bias. Simply combining both domains simultaneously yields an overconfident pseudo-posterior. In this paper, we formulate semantic decoding as a Bayesian inverse problem and prove that posterior sampling achieves optimal perceptual quality by preserving the data distribution. Building on this insight, we propose alternating dual-domain posterior sampling (ADDPS), a diffusion-based SemCom receiver that alternately enforces latent-domain and image-domain consistency during the sampling process. This alternating strategy decomposes joint posterior sampling into simpler subproblems, avoiding gradient conflicts while retaining the complementary strengths of both domains. Experiments on FFHQ demonstrate that the proposed ADDPS achieves superior perceptual quality compared with existing methods.

This tutorial derives the mathematical foundations of what it means for a carrier waveform to be predictable and non-selective. We focus on Zak-OTFS, where each carrier waveform is a pulse in the delay-Doppler (DD) domain, formally a quasi-periodic localized function with specific periods along delay and Doppler. Viewed in the time domain, the Zak-OTFS carrier is realized as a pulse train modulated by a tone (termed a pulsone). We start by providing physical intuition, describing what it means for the Zak-OTFS carrier waveforms to be geometric modes of the Heisenberg-Weyl (HW) group of discrete delay and Doppler shifts that define the discrete-time communication model. In fact, we show that these geometric modes are common eigenvectors of a maximal commutative subgroup of our discrete HW group. When the channel delay spread is less than the delay period, and the channel Doppler spread is less than the Doppler period, we show that the Zak-OTFS input-output (I/O) relation is predictable and non-selective. Given the I/O response at one DD point in a frame, it is possible to predict the I/O response at all other points, without recourse to some mathematical model of the channel. While it may be intuitive that geometric modes of the HW group are predictable and non-selective wireless carriers, this is not a requirement. We provide a necessary and sufficient condition that depends on the ambiguity properties of the basis of carrier waveforms. In fact, we show that the structure of a pulse train modulated by a Hadamard matrix is common to several families of waveforms proposed for 6G, including Zak-OTFS, AFDM, OTSM and ODDM.

The high computational cost of approaching the performance of Maximum-likelihood (ML) decoding has limited its practical use for decades. Because the complexity grows exponentially with the message length, researchers have spent years developing algorithms like Ordered Statistics Decoding (OSD), Partial Ordered Statistics Decoding (POSD) and Guessing Random Additive Noise decoding (GRAND) which try to approach ML performance. OSD, POSD and GRAND work by trying to guess the error patterns affecting the received signals. However, there does not exist a systematic method to extend the error pattern guesses to novel channels. This work introduces a systematic method that uses the Probability Density Function (PDF) of a memoryless channel to generate a set of error patterns that can be applied on any future received signal on this channel. Simulation results show that our proposed method applied on GRAND, OSD and POSD generally matches or outperforms current pre-generated error patterns on additive white Gaussian noise (AWGN) channel, mixture of Gaussian distribution channels, Rayleigh fading channel with perfect knowledge of Channel State Information (CSI) and Rayleigh fading channel with no perfect knowledge of Channel State Information (NCSI).

Beam training and prediction in real-world millimeter-wave (mmWave) communications systems are challenging due to rapidly time-varying channels and strong interference from surrounding objects. In this context, widely available sensors, such as cameras and radars, can capture rich environmental information, enabling efficient beam management. This paper proposes a knowledge-distillation (KD)-enabled learning framework for developing lightweight and low-complexity models for beam prediction and tracking using real-world camera and radar data from the DeepSense 6G dataset. Specifically, a powerful teacher network based on convolutional neural networks (CNNs) and gated recurrent units (GRUs) is first designed to predict current and future beams from historical sensor observations. Then, a compact student model is constructed and trained via KD to transfer the predictive capability of the teacher model to a lightweight architecture. Simulation results demonstrate that jointly leveraging radar and image modalities significantly outperforms single-modality approaches. Moreover, the proposed student model achieves over 96% Top-5 beam prediction accuracy while reducing computational complexity by more than 4 times and the number of parameters by over 27 times compared with the teacher model.

In this paper, we consider resource allocation for a collaborative integrated sensing and communication (ISAC) scenario, in which distributed smart devices can be scheduled to perform sensing and transmit their sensing features to a fusion center. The fusion center aims to perform classification tasks on the environment based on received features. A scalable networksensing framework is proposed to balance the performance of the sensing service with that of the classical enhanced Mobile Broadband (eMBB) service. We adopt a tractable theoretical metric, the discriminant gain, as a proxy for the classification goal. We formulate cross-layer optimization problems to maximize discriminant gain under constraints on energy consumption and eMBB communication quality for the independent and joint scheduling policies. The joint scheduling policy has considerably higher complexity than the independent scheduling policy, in exchange for better collaborative sensing performance. A simplified gain model is proposed to reduce the complexity and practicality of the joint scheduling policy. Both policies are obtained via successive convex approximation and parametric convex optimization. Extensive experiments are conducted to verify the goal-oriented framework and the two policies. It is demonstrated that the two policies outperform the baseline policies with both synthetic and realistic radar simulation datasets. The joint scheduling policy can exploit device correlations and thus performs better than the independent scheduling policy under strong correlations and strict communication constraints.

We propose a cosmological scenario in which, beyond matter and radiation, an additional barotropic fluid with positive equation of state $ω_s$ contributes to the cosmic energy budget, in contrast to Early Dark Energy (EDE). We investigate the theoretical implications of this framework, here dubbed the $Λ_{ω_s}$CDM model, at both the background and perturbative levels, exploring its impact on the expansion history and structure formation. We show that, while remaining subdominant at late times and therefore consistent with current observational bounds, the additional fluid modifies the early-time expansion rate, leading to a higher inferred value of the Hubble constant. Thus, we perform a full Bayesian analysis using a modified version of the \texttt{CLASS} Boltzmann code interfaced with \texttt{MontePython}, considering combinations of \textit{Planck} 2018 Cosmic Microwave Background (CMB) data, DESI DR2 Baryon Acoustic Oscillations (BAO) measurements, Pantheon Type Ia supernovae (SNe Ia), and SH0ES determinations of $H_0$. We find that the inclusion of the SH0ES prior, $H_0 = 73.04 \pm 1.04\,\mathrm{km/s/Mpc}$, leads to a preference for a nonvanishing barotropic fluid. In particular, we obtain $ω_s = 0.290^{+0.017(0.021)}_{-0.007(0.028)}$ and density $10^5Ω_s = 1.47^{+0.35(1.14)}_{-0.62(0.94)}$ for the dataset combination CMB + BAO + Pantheon + SH0ES, and $ω_s = 0.302^{+0.024(0.034)}_{-0.013(0.038)}$ and $10^5Ω_s = 1.21^{+0.31(1.10)}_{-0.65(0.86)}$ when BAO data are excluded. We further compare our scenario with the EDE framework and show that, statistically, no strong evidence is found against the $Λ_{ω_s}$CDM model. Finally, we provide a physical interpretation of our fluid in terms of matter with pressure, indicating that the standard cosmological model may be incomplete in its current minimal formulation.

We present a matrix product state (MPS) method for including decoherence processes in calculations involving waveguide quantum electrodynamics (waveguide QED) using density matrices. The approach is based on collision quantum optics, where the many-body state of the waveguide is represented as discrete time bins, which are efficiently represented using an MPS chain. Our method is a generalization of previous MPS methods, and we demonstrate how one can efficiently expand to density matrices, allowing for the inclusion of various loss processes in the form of Lindblad terms in the Liouvillian superoperator responsible for the relevant dissipation dynamics. As an application of the theory, we study various waveguide QED systems and the influence of emitter pure dephasing (which is one of the most important processes in real systems) on the light-matter interactions, including a two-level system (TLS) in a semi-infinite waveguide with time-delayed feedback, two spatially separated TLSs with finite delays, and finally the scattering of few-photon Fock pulses on a TLS. In addition to emitter pure dephasing, we also show how to include off-chip radiative decay, and show how it differs qualitatively from pure dephasing.

We have considered the Yukawa theory at two loop in the inflationary de Sitter spacetime, for a massless minimally coupled scalar and a massless fermion. The one loop computation for the same has been investigated in detail in the earlier literatures. The chief motivation behind this study is the fact that at one loop, the scalar self energy contains only fermions, which are conformally invariant. At two loop, there are two diagrams, each containing one internal scalar line, thereby breaking the conformal invariance. This should result in the appearance of infrared secular logarithms in the scalar self energy. The renormalisation of this two loop self energy has been performed. We next compute the loop corrected coincident two point correlation function, $\langle φ^2\rangle$, due to the self energies. The expectation value has been taken with respect to the initial Bunch-Davies vacuum. We argue that the late time secular contribution from the local or UV self energy must dominate the non-local or IR ones in the present case, from the point of view of the powers of these large logarithms of the scale factor. This corresponds to the fact that fermion lines do not show any IR secular effect. The leading behaviour of $\langle φ^2\rangle$ at one and two loop are respectively found to be $\ln^3 a$ and $\ln^4 a$. These are hybrids of UV and IR logarithms, where the latter originate from the massless and minimal two external scalar lines. A resummed expression for $\langle φ^2\rangle$ has also been computed. The same is found to be bounded and decreasing monotonically with the increasing magnitude of the Yukawa coupling. Accordingly, the dynamically generated scalar mass increases with the increasing coupling.

Near-term quantum devices with limited qubits motivate the study of space-bounded quantum computation in the data stream model. We show that Shannon entropy estimation exhibits an exponential separation between quantum and classical space complexity in this setting. Technically, we develop a two-stage quantum streaming algorithm based on a quantum procedure with an explicitly constructed oracle derived from the streaming input. This algorithm achieves logarithmic space complexity in the accuracy parameter, whereas any classical streaming algorithm requires polynomial space. In sharp contrast, existing results for Shannon entropy estimation in the quantum query model achieve only a quadratic speedup. Our work establishes a natural problem with practical applications in computer networking that admits an exponential quantum space advantage, revealing a fundamental gap between quantum query complexity and streaming space complexity.

We investigate the leading gravitational eikonal in nonlocal $D$ dimensional theories of gravity. We analyze the simplest cases of $2\rightarrow2$ massless and massive scalar scattering at tree level, studying the effects of nonlocal form factors in the gravitational sector. We give an interpretation of our results in terms of geodesic motion in effective generalized Aichelburg-Sexl geometries for the massless case, and in smeared linearized Schwarzschild metrics for the massive case in the probe limit. Combining our results for the geometries at linearized level with general requirements about the behaviour of the solutions in the core, we propose a nonlinear completion of the geometries. The resulting spacetimes describe singularity-free, asymptotically flat deformations of the Schwarzschild solution with a de Sitter core. We also analyze the main geometric and thermodynamic features of these solutions.

We show numerically that a repulsive Bose-Einstein condensate can be driven into implosive dynamics by a direct topological quench. We first realize giant vortices by quasi-adiabatic phase imprinting, and then perform a sudden anti-imprint that cancels the accumulated winding in a single step, abruptly switching the condensate from a highly charged vortex state to the trivial sector. The resulting phase-density mismatch launches a rapid inward radial flow and produces a strong central density buildup, despite the repulsive interactions. We find a clear threshold in the initial winding for the onset of this focusing. After the first implosion, the dynamics evolves into circular nonlinear wave fronts that subsequently undergo breaking of azimuthal symmetry (axisymmetry) down to a polygonal one, whose shape is determined by the way the giant vortex is built. These results establish topological engineering as a new tool for studying implosive dynamics and symmetry-breaking instabilities in quantum fluids.

Accurate modeling of surface catalytic processes often requires methods capable of describing strong correlation, charge transfer, and multiple closely lying electronic states. While density functional theory remains widely used, its limitations for localized electronic states motivate the use of wavefunction-based approaches and, more recently, quantum computing algorithms. However, the performance of quantum ansätze in chemically motivated, multistate settings remains largely unexplored. Here, we benchmark state-averaged factorized unitary coupled cluster with singles and doubles (SA-fUCCSD) and the adaptive, problem-tailored ansatz (SA-ADAPT) using an embedded cluster model of NO adsorption on Rh-doped TiO2(110). The system exhibits pronounced multiconfigurational character and multiple state crossings, providing a stringent test. State-averaged CASSCF serves as a reference, and the quantum ansätze are evaluated as solvers for the corresponding CASCI problem within a fixed orbital basis. We find that SA-fUCCSD improves with increasing circuit depth but requires many parameters and shows sensitivity to initialization. In contrast, SA-ADAPT achieves near-CASSCF accuracy with significantly fewer operators. A modified operator selection scheme, incorporating multiple operators per iteration, substantially accelerates convergence. Our results demonstrate the efficiency of adaptive ansätze for multistate problems and establish a controlled benchmark for quantum algorithms in chemically motivated systems beyond minimal models.

We introduce a traceable operator-algebraic framework for incompressible transport on M= T3 (and, more generally, compact Riemannian manifolds endowed with a smooth invariant probability measure). Given an autonomous divergence-free velocity field u, the time-1 map $Φ$ induces the Koopman unitary U on L2(M) and the crossed-product finite von Neumann algebra Mu\,:= L$\infty$(M) ___$α$ Z= W$\star$(L$\infty$(M),U), equipped with its canonical faithful normal trace $τ$u. Within Mu we define tracial complexity functionals from commutators [U,Mf] (with Mf the multiplication operators) and associated positive elements, and we connect these quantities to Fuglede--Kadison determinants and entropy-like tracial functionals. In parallel, we introduce bounded regularized advection operators T(s) u\,:= KsTuKs as differential-level probes of transport oncommutativity, and we recall the Lie-bracket commutator identity at the formal generator level. This provides a natural algebraic setting in which tracial invariants are well posed and, in principle, computable on discretizations (e.g. cavity flow and vortex benchmarks).

We construct a family of positive but not completely positive linear maps acting on four dimensional space. We employ these maps to detect bound entanglement in high dimensional quantum systems.

Microwave photon detection is a key technology for low-temperature superconducting electronics and quantum information processing. A promising possibility is to use switching processes in parametric superconducting devices at criticality, which can be triggered by small perturbations. Here we demonstrate the unique sensing properties of the superconducting Kerr parametric resonator when operated in the proximity of the phase transition boundary. We utilize a semiclassical approximation to provide numerical and analytical results for the Heisenberg-Langevin and Fokker-Planck equations that describe the switching mechanism. We show that the probability of switching events is enhanced by probe input states with energies down to single quanta levels.

Molecular optomechanical cavities have recently emerged as a promising platform for frequency upconversion, enabling the quantum coherent conversion of infrared signal into the visible range. In a recent work [F. Zou et al., Phys. Rev. Lett. 132, 153602 (2024)], we proposed an amplification mechanism that can enhance the intensity of the upconverted infrared signals by a factor of 1000 or more within such a cavity under the ideal case without any noise. In this work, we employ the power spectrum method to investigate the noise added to the upconverted signal in a molecular optomechanical cavity along with the conversion efficiency from infrared signal into visible range. In the red-detuned regime, the anti-Stokes sideband achieves superior conversion efficiency relative to the Stokes sideband. Conversely, the Stokes sideband dominates under the blue-detuned condition, which amplifies the infrared signal. We further demonstrate the dependence of the added noise on the coupling strength and decay rates of the system. In particular, we find that when the infrared signal is amplified, the added noise approaches the quantum limit of one quantum.

In curved spacetime, deformable bodies can undergo a displacement of their center of mass through cyclic internal motions without the use of propellant, as shown by Wisdom. In this paper, we demonstrate a different mechanism: in Schwarzschild spacetime, static structures with internal stress can generate a buoyancy-like force without non-gravitational external forces or cyclic internal motions. The resulting effect is extremely small for realistically sized structures and does not lead to an actual ascent against gravity. Nevertheless, it reveals a new aspect of extended-body dynamics in spacetimes with curvature gradients, in particular regarding the influence on motion of effects arising from the coupling between internal stress and spacetime curvature.

In this paper, we apply the deterministic quantum imaginary time evolution (QITE) algorithm to obtain the ground state of a two-dimensional pure $\mathbb{Z}_2$ lattice gauge theory. We first construct the set of Pauli operators commuting with Gauss's law constraints, generalizing a previous result. This makes the deterministic QITE gauge-invariant and reduces both the measurement and gate costs significantly without adding extra algorithm errors in the QITE. Then, the classical numerical simulation of the deterministic QITE using tensor networks is performed, and the results are compared with the density matrix renormalization group (DMRG) to evaluate the accuracy of the algorithm. Specifically, we investigate the coupling and system size dependence, and find that the deterministic QITE can achieve a relative error of less than $0.1\%$ up to a twelve-plaquette system and coupling values in a regime that we study. Furthermore, the error dependence on the number of time steps is studied and discussed.

\texttt{gwNRHME} is a framework that converts multi-modal (i.e., containing several spherical harmonic modes) quasi-circular waveforms into their eccentric counterparts, provided the quadrupolar eccentric mode is known, by exploiting universal eccentric modulation functions. Leveraging this framework, we combine the quasi-circular NR surrogate model \texttt{NRHybSur3dq8} with the quadrupolar, non-spinning, eccentric surrogate \texttt{NRSurE\_q4NoSpin\_22} to construct a multi-modal, non-spinning, eccentric model, denoted as \model{}, which includes nine modes: $(2,\{1,2\})$, $(3,\{1,2,3\})$, $(4,\{2,3,4\})$, and $(5,5)$. When compared against 156 eccentric SXS NR waveforms, \model{} achieves median frequency-domain mismatches (computed using the Advanced LIGO design sensitivity) of $\sim 9\times 10^{-5}$, with a standard deviation of $\sim 2 \times 10^{-4}$. To demonstrate the modularity of the framework, we further combine \texttt{NRSurE\_q4NoSpin\_22} with effective-one-body (EOB) models \texttt{SEOBNRv5HM} and \texttt{TEOBResumS-Dali} in their non-spinning limits, yielding eccentric waveforms with median mismatches of $\sim 2\times10^{-4}$ and $\sim 10^{-3}$, respectively, with standard deviation of $\sim 2 \times 10^{-3}$ and $\sim 2 \times 10^{-2}$ respectively. Finally, we provide both a surrogate model, \texttt{gwEccEvolve\_q4NoSpin\_Sur}, and an analytical model, \texttt{gwEccEvNSv2}, for the eccentricity evolution up to $2M$ before merger, based on eccentricity definitions derived from the universal modulation functions. The \texttt{gwNRHME} framework is publicly available through the \texttt{gwModels} package, and the resulting waveform models will be released via the \texttt{gwsurrogate} package.

We derive a tight and saturable monogamy relation for three-qubit pure states that bounds the sum of concurrence and concurrence of assistance by the entanglement with an external qubit. The bound decreases strictly with increasing external entanglement, establishing a precise trade-off between internal and environment-induced entanglement. Equivalent formulations in terms of negativity and its convex-roof extensions follow. Our result provides a unified and quantitative constraint on entanglement distribution in open multipartite quantum systems.

We present a systematic prolongation procedure and its implementation for Killing two-tensors, especially in the locally symmetric case. We use the resulting machinery to elucidate the natural quadratic mapping from Killing fields to Killing two-tensors on irreducible locally symmetric spaces of compact type.

We study resonances for the semiclassical magnetic Laplacian in the full plane with a compactly supported magnetic field in the framework of semiclassical complex scaling and black box scattering theory. Assuming that the magnetic field is locally constant, we prove the existence of semiclassical resonances near the Landau levels with exponentially small imaginary parts. We also prove that resonances emerge from a magnetic step discontinuity along a curved interface or a non-degenerate magnetic well, and in the vicinity of anharmonic Landau levels if the field has an isolated zero.

Recently, Hausmann and Renner have pointed out that several famous paradoxes relating to black holes have a similar character to various Extended Wigner's Friend paradoxes. In this paper I consider what the connection between these things could teach us about the Wigner's Friend scenarios. I argue that if we take the analogy between these cases seriously, the black hole paradoxes appear to favour a certain class of response to the Wigner's Friend scenario - specifically, those which posit intrinsic relationality, rather than effective and emergent relationality, and also those which posit some kind of retrocausality.

Classical opacity theory for discrete-event systems relies strictly on observable event sequences, fundamentally failing to capture security breaches in hybrid architectures where an attacker exploits both classical traces and localized quantum correlations. To address this gap, we formalize current-state opacity within the framework of safe partially observed quantum Petri nets by introducing a true-concurrency semantics that represents classical observations as partially ordered multisets via unfolding configurations. Building upon this, we define quantitative posterior-state leakage as the trace distance between the attacker's localized quantum states, evaluated conditionally on whether the underlying system has reached a secret or non-secret marking. This formulation strictly preserves classical opacity definitions. To achieve computational tractability, we apply the stabilizer formalism and develop an exact symbolic verification algorithm. By combining targeted unfolding exploration, state aggregation exclusively at maximal unobservable reach, and stabilizer-tableau propagation, this procedure circumvents both concurrent interleaving explosions and exponential density-matrix overhead. Finally, an entanglement-swapping case study validates the exact leakage evaluation, demonstrates substantial computational gains, and establishes a rigorous interface for counterexample-guided leakage enforcement.

Employing mutually-commuting von Neumann algebras to represent the algebra of observables on quantum systems provides a framework for studying quantum information theory in systems with infinite degrees of freedom and quantum field theory, yielding many profound results that differ from non-relativistic quantum systems. In this paper, we establish a mutually-commuting von Neumann algebra model of quantum networks with arbitrary structures. We derive Bell-type inequalities on this model, and determine various bounds for Bell-type inequalities based on the structure of underline von Neumann algebras, and identify the algebraic structural conditions required for their violation. The conditions on the algebraic structure of observables for maximal violation of Bell-type inequalities, which we discovered in the context of von Neumann algebra models, can in turn guide the search for measurements in the non-relativistic setting.

We establish a quantitative relation between local spin polarization and quantum entanglement in two-qubit systems by deriving an upper bound on the concurrence at fixed local polarization, showing that increasing polarization constrains the maximum achievable entanglement. We further demonstrate that this bound is saturated by pure states in certain cases. As a concrete physical application, we consider the parity-violating process $e^+e^- \to Z^0 \to q\bar{q}$, which generates final-state spin polarization. We show that the maximal concurrence is attained in specific kinematic regions and is significantly reduced relative to the unpolarized case. These results establish a general, process-independent framework connecting local polarization, maximal entanglement, and the role of pure states.

The cosmological principle asserts that the Universe is homogeneous and isotropic on large enough scales. However, alternative cosmological models can bring about anisotropies through local inhomogeneities, anisotropic evolution, or exotic physics. In addition, select studies have also hinted at mild evidence of anisotropies in SNe Ia, CMB, and GRB data, though these remain unconfirmed. In this work, we test for cosmological anisotropies using gravitational waves and gamma-ray bursts, adopting the latest O4a release from the LIGO-Virgo-KAGRA collaboration and GRBWeb (including all known GRBs since 1991). If the cosmological principle holds, the sky localisation and the characteristics of the GRBs and GWs (masses, luminosities, redshifts) should be statistically isotropic when corrected for selection biases. We employ a couple statistical methods, including angular power spectra and two-point correlation functions, and compare the results against synthetic data. The work extends previous analyses by including the most recent datasets, and the use of multiple complementary statistical tests. We find no significant evidence for anisotropy in the current GW and GRB datasets, consistent with the cosmological principle.

The long-range transmission of quantum information relies on multiple interfaces between photons, acting as flying qubits, and localized memories, serving as repeaters, to mitigate transmission losses. Efficient, long-range transmission necessitates the use of short, picosecond-scale photons, which are markedly different from the narrowband, nanosecond-scale photons optimal for absorption by memory elements, typically operating at wavelengths far from telecom. In this article, we point toward designs capable of bridging these regimes, leveraging the interplay between sum-frequency generation-based quantum frequency conversion and resonant confinement in an integrated ring resonator.

Experimental access to many-body quantum systems is often limited by measurement backaction, and key dynamical properties are typically obtained by perturbing a system and measuring its response. Here we replace this active paradigm with a minimally invasive protocol based on a pair of weak quantum measurements that leverages measurement backaction as a strength. By correlating time-separated measurements with the first detecting fluctuations -- of any sort -- and the second tracking their time evolution, our method directly measures dynamical correlation functions without external perturbation. We demonstrate this technique in an atomic Bose-Einstein condensate using phase-contrast imaging to obtain the two-time density-density correlation function known as the Van Hove function and, through its Fourier transform, the dynamical structure factor. Due to the role of spatial correlations in scattering, these quantities underpin neutron and X-ray scattering and atomic Bragg spectroscopy. This approach is broadly applicable, providing access to correlation functions between any pair of observables amenable to weak measurement, thereby going beyond the capabilities of conventional strong measurements. We further isolate the role of quantum backaction through Aharonov's post-selection-based quantum weak values.

We investigate quantum spectroscopy with undetected photons for protein detection in the mid-infrared spectral region. Classical Fourier-transform infrared spectroscopy of protein samples (bovine serum albumin and N-terminal pro-brain natriuretic peptide) is used as reference to define the sample's mid-infrared absorption, which is then embedded in a numerical model of a double-pass quantum interferometer. We analyse parameters that influence visibility of the interference pattern formed by the signal beams, including the length of nonlinear crystal, sample length and mirror-sample distance. This leads us to a practical quantum spectrometer design with optimal image contrast at the specific amide I-II spectral bands. The simulated visibility spectra reproduce nearly identically the protein absorption features in the mid-IR and reveal temperature-induced changes to the protein secondary structure. Overall, this provides practical design rules for future quantum bio-spectroscopy applications that use only visible wavelength sources and detectors.

We investigate the thermodynamic and phenomenological implications of a cosmological model governed by fractional entropy applied to the apparent horizon of a flat Friedmann-Lemaître-Robertson-Walker (FLRW) universe. By utilizing the unified first law of thermodynamics alongside the Kodama-Hayward temperature, we derive a generalized set of Friedmann equations characterized by a fractional parameter $α\in (1,2]$. The thermodynamic analysis reveals that the specific heats $C_V$ and $C_p$ share the same sign and depend solely on the deceleration parameter, demonstrating that the fractional model is thermodynamically stable during the late-time accelerated expansion and does not exhibit phase transitions. To constrain the background dynamics, we confront the truncated fractional model with a joint sample of late-time observational data, including Cosmic Chronometers, Pantheon+SH0ES supernovae, and the latest DESI DR2 Baryon Acoustic Oscillations. Exploring the physically motivated range $ 1 < α\leq 2 $, we find that the fit quality degrades monotonically as $α$ decreases from the General Relativity limit, with the data favoring $α$ close to $2$ while yielding $H_0 = 69.50 \pm 0.42$ km/s/Mpc and $Ω_{m0} = 0.292 \pm 0.008 $ at $α= 2$. Decreasing $α$ coherently shifts $H_0$ upward and $Ω_{m0} $ downward, revealing that the fractional parameter modulates the background expansion in a physically nontrivial and observationally distinguishable way.

We propose a versatile and efficient algorithmic framework for optimizing fermion-to-qubit mappings by generalizing the idea of randomized block coordinate descent. Our greedy approach, termed Randomized Subsystem Descent, iteratively samples a tractable subsystem from the full Hamiltonian, performs optimization within the subsystem under a given metric, and then reintegrates the updated subsystem into the global operator. Restricting the optimization to a subsystem at each iteration ensures computational efficiency, bypassing the dimensional bottlenecks that usually hinder global search heuristics. We benchmark our algorithm on one- and two-dimensional lattice hopping models, the Hubbard model with up to $16 \times 16$ sites, alongside a collection of molecular electronic-structure Hamiltonians with up to 54 modes and more than 180,000 Pauli strings. Across all benchmarks, our method consistently provides appreciable reduction in (weighted) Pauli weight, suggesting that Randomized Subsystem Descent is a practical and scalable framework for lowering the resource overhead of finding hardware-efficient Hamiltonian encodings.

We consider a search for the anisotropic stochastic gravitational-wave background (SGWB) that decomposes the sky map into its spherical harmonics components in order to obtain estimators of the angular power spectrum. Such a search often requires the inversion of a Fisher information matrix which contains small singular values. Rather than dealing with biases induced by regularization methods used to facilitate this matrix inversion, we opt to avoid this inversion step entirely by working in the so-called ``dirty map" space, and we introduce methodology for statistical model inference in this space. We apply our methodology to simulated model signals added to detector noise characterized by Advanced LIGO's third observing run and consider angular power spectra for both the SGWB auto-correlation search as well as a cross-correlation search between the SGWB and electromagnetic tracers of matter structure in the universe. In both cases we are able to reliably recover simulated model parameters for sufficiently strong signals up to maximum order spherical harmonic modes of $\ell_{max}=10$. We find the limitations of our methodology to arise from the computational cost of testing complex models, the assumption of Gaussianity of the angular power spectrum, and a cosmic variance-like source of uncertainty which scales with the strength of the underlying signal.

We present a novel approach to deriving the Einstein Lagrangian density based on energy-momentum conservation. In this framework, starting from a field-theoretic setting with background Minkowski spacetime and Poincaré invariance, we investigate whether conservation of the total energy-momentum tensor, in which the field contribution is given by the symmetrized Belinfante tensor, determines the form of the field Lagrangian density for a symmetric rank-2 tensor field. We find that the requirement of total energy-momentum conservation is satisfied only if the field Lagrangian density is proportional to the Einstein Lagrangian density.

We investigate the combined effects of a brane intersecting the AdS boundary and background gravitational field on the local characteristics of the electromagnetic vacuum. Two types of boundary conditions on the brane are considered, which are higher-dimensional generalizations of the perfect electric (PEC) and perfect magnetic (PMC) boundary conditions in Maxwell's electrodynamics. The brane-induced contributions to the Wightman functions of the vector potential and field tensor are explicitly extracted. Simple expressions in terms of elementary functions are provided. The behavior of the vacuum expectation values (VEVs) is mimicked by a scalar field with a negative effective mass squared determined by the radius of the AdS spacetime. The expectation values of the electric and magnetic fields squares and of the energy-momentum tensor are investigated as local characteristics of the vacuum state. The brane-induced contributions to these VEVs have opposite signs for the PEC and PMC conditions. For the PMC condition, this contribution is negative for the electric field squared and positive for the magnetic field squared. The VEV of the energy-momentum tensor has a nonzero off-diagonal component. The brane-induced vacuum energy density is positive for PMC condition, whereas the normal and parallel stresses change sign as functions of the distance from the brane. Unlike the problem involving a planar boundary in the Minkowski bulk, the vacuum energy-momentum tensor does not vanish in (3+1)-dimensional AdS spacetime.

We study finite-window black-hole spectroscopy in the loud-event regime and ask when a multimode ringdown fit supports a stable common-remnant Kerr interpretation. Starting from whitened, tapered detector-frame data, we prove a deterministic frequency-extraction theorem for a projected sampled Prony--matrix-pencil pipeline with explicit statistical, algorithmic, omitted-tail, and mismatch terms. We then construct a local inverse atlas for the Kerr $(\ell,m,n)=(2,2,0)$ map on an event-local detector-frame remnant box for GW250114 and propagate the resulting primary uncertainty into $(2,2,1)$ and $(4,4,0)$ consistency tests. These ingredients yield a detector-frame trust criterion for individual windows. We calibrate mismatch and colored-noise radii on a GW250114-like synthetic waveform bank built from public surrogate, CCE, and numerical-relativity information, and we apply the resulting bounds to the public H1/L1 strain and public parameter-estimation products for GW250114. The accepted windows form an intermediate post-peak band: earlier windows remain sensitive to start-time drift and structured nuisance fits, whereas later windows become variance dominated. Within that band, the recovered remnant remains consistent with the public inspiral--merger--ringdown estimates and supports a common-remnant Kerr interpretation that survives the full preprocessing and robustness checks. For loud events, the relevant question is therefore which finite detector-frame windows sustain spectroscopy, not whether some multimode fit can be made in isolation.

Conventional quantum thermometry assumes completely positive (CP) encoding maps, where the probe is initially uncorrelated with the environment. We consider realistic scenarios with initial probe-environment correlations leading to physically realizable non-completely positive (NCP) encoding, and show how such encodings can significantly impact temperature estimation of the environment. We first consider pure entangled probe-environment initial states (Type-I NCP encoding) and analytically show that for probes and environments of equal but arbitrary dimension, the maximum achievable precision matches the thermal-state bound, as in the CP case. However, upon relaxing the constraint of pure probe-environment states and considering general correlated initial states (Type-II NCP encoding), we demonstrate that the estimation precision can surpass the thermal-state limit. This establishes a clear advantage of NCP encoding in enhancing thermometric performance. We illustrate the results using qubit probes interacting with qubit environments via XY interactions.

We prove that the Penrose limit of a Lorentzian metric along an affinely parametrized null geodesic is intrinsic, but intrinsic on a weighted associated-graded model determined by the null filtration rather than on a canonically identified spacetime neighborhood. Under the standard dilation scaling $(u,v,x)\mapsto (u,λ^2 v,λx)$, admissible adapted coordinate changes collapse to their weighted homogeneous principal parts, so the large coordinate freedom of the classical construction degenerates to a small residual weighted gauge group, namely the group attached to the splittings of the null filtration. On the manifold of unparametrized null geodesics, the same weighted dilation is the grading derivation of a Heisenberg tangent model, and a $1$-jet of contact scale determines a realized degree-two direction without changing the underlying graded limit. These residual data assemble into an intrinsic unpolarized Penrose gauge bundle over the $1$-jet bundle of contact scales, with a polarized parabolic reduction after choosing a Lagrangian. Pulling the resulting model to the incidence space of spacetime points lying on null geodesics yields a canonical tautological soldering to the ambient weighted normal geometry, and fiberwise identifies the corresponding homogeneous plane-wave germ with the weighted associated graded of the ambient metric germ along the null geodesic. On the pullback of the incidence correspondence to the first jet bundle of contact scales, the tautological soldering canonically identifies the Penrose limit with an actual metric on the corresponding soldered spacetime neighborhood of the null geodesic.

We show that the diffusion-based models proposed in Refs.~\cite{Perez2021,Landau2022} within the framework of Unimodular Gravity (UG) to alleviate the $H_0$ tension are incompatible with the second law of thermodynamics. Starting from the Gibbs equation for a pressureless matter fluid, we derive a general thermodynamic admissibility condition for the $Λ$CDM$+$diffusion class in UG, demonstrating that the second law requires $\dot{Q} < 0$, independently of the specific form of the diffusion function $Q(t)$. This condition implies that energy must flow from the effective cosmological term $Λ_{\rm eff}$ into the matter sector, rather than in the opposite direction. We then establish a no-go theorem: no choice of $Q(t)$ can simultaneously satisfy the second law and generate the growing effective cosmological term $\dotΛ_{\rm eff} > 0$ required to alleviate the $H_0$ tension. We confirm this result explicitly for the models of Perez, Sudarsky, and Wilson-Ewing~\cite{Perez2021} and Landau et al.~\cite{Landau2022}, noting that the former explicitly identify $\dotΛ_{\rm eff} > 0$ as a necessary feature of their proposal without addressing its thermodynamic implications. The incompatibility is therefore structural and applies to the entire class of $Λ$CDM$+$diffusion models in UG with a pressureless matter component.

Quantum batteries harness non-classical resources, such as quantum coherence and entanglement, to surpass the performance limits of classical energy-storage devices. Here we realize a room-temperature quantum battery based on a collective atomic spin ensemble in a thermal alkali-metal vapor, containing approximately $10^{12}$ $^{87}$Rb atoms with coherence times exceeding 110 ms. We operationally determine the battery capacity by directly measuring the extremal internal energies accessible under unitary evolution. This tomography-free protocol agrees closely with the conventional state-based definition and verifies the decomposition of capacity into coherent and incoherent contributions. We further show that quantum coherence can substantially enhance the storage capability independently of level populations, and experimentally establish quantitative relations linking battery capacity to von Neumann, Tsallis and linear entropies. By introducing a controlled dephasing channel with a magnetic-field gradient, we observe a monotonic reduction of capacity with coherence loss and track the corresponding evolution of the entropy-capacity relations. Our results identify thermal atomic spin ensembles as a scalable platform for quantum batteries and connect macroscopic quantum energy storage with operational quantum thermodynamics.

Quantum Neural Networks (QNNs) represent a promising direction within Quantum Machine Learning (QML), yet their realization on noisy intermediate-scale quantum (NISQ) devices remains constrained by decoherence, gate imperfections, crosstalk, and readout errors. This study provides a systematic evaluation of noise effects and mitigation strategies in hybrid quantum neural networks (HQNNs). Zero-Noise Extrapolation (ZNE), Digital Dynamical Decoupling (DDD), and Layerwise Richardson Extrapolation (LRE) are integrated into end-to-end QNN training pipelines developed with PennyLane, simulated under Qiskit Aer noise models, and integrated with the Mitiq framework, while Probabilistic Error Cancellation (PEC) is evaluated separately under depolarizing noise due to its computational cost. Experiments conducted on the Iris dataset with five representative noise channels show that the impact of noise and the effect of mitigation are strongly dependent on the noise model and its strength. The model maintains comparatively strong performance under phase-flip and phase-damping noise, while substantial degradation is observed under high depolarizing and amplitude-damping noise. Across the evaluated mitigation methods, the observed benefits remain limited and noise-dependent: ZNE, DDD, and LRE generally follow the same degradation trends as the unmitigated baseline, while PEC shows limited gains only in the low-noise depolarizing regime. These findings highlight the need for context-specific mitigation strategies to improve the robustness of QNNs in practical NISQ settings.

Relativistic extensions of de Broglie-Bohm theory postulate a preferred foliation of space-time, an additional structure essential for defining simultaneous configurations on Minkowski space-time, but conventionally believed to be empirically undetectable at quantum equilibrium. In this paper, we outline an experimental protocol for empirically detecting the preferred foliation, which is assumed to be flat for simplicity. Building on the arrival-time distributions for spin-1/2 particles predicted by Das and Dürr, we show that in an EPRB-type experiment with spacelike-separated spin and arrival-time measurements, the observed arrival-time statistics will depend crucially on the temporal order of these measurements relative to the preferred foliation of space-time. This dependence offers a potential experimental signature of the preferred foliation postulated by relativistic Bohmian models. Moreover, it implies the possibility of superluminal signaling.

We present a quantum-dynamical framework for identifying structurally important residues in proteins based on continuous time quantum walks (CTQWs) on weighted residue interaction networks constructed from experimentally resolved structures. By mapping the weighted adjacency matrix to a Hamiltonian, residue importance emerges from the long-time averaged occupation probability, confirmed analytically through its spectral decomposition. Across a dataset of approximately 150 proteins spanning diverse structural and functional classes, CTQW centrality exhibits consistently strong agreement with classical eigenvector centrality in identifying central residues, while extending beyond it through incorporating signatures of quantum interference. Analyzing the time-averaged quantum transition matrix reveals consistently larger spectral gaps than the classical random-walk operator. Furthermore, biological relevance is confirmed through recovery of experimentally established functional residues in proteins kinase A and oxytocin. CTQW-derived centrality rankings are accessible on near-term intermediate-scale quantum hardware, as we demonstrate through a proof-of-principle implementation on IBM superconducting quantum hardware. These results establish continuous-time quantum walks as a computationally tractable framework for protein network analysis, that connects network theoretical treatments of protein structural biology to continuous-time quantum walk dynamics.

Small modular nuclear reactors (SMRs) are redefining the energy generation landscape by enabling the deployment of modular, scalable, and pre-built power units that can be used to build distributed autonomous microgrids for critical infrastructure and burgeoning AI factories. Often, these microgrids are linked together to provide a resilient, decentralized power generation infrastructure. Consequently, the cybersecurity of microgrids is of critical importance. In this work, we propose a quantum augmented network framework for resilient microgrids. We integrate the ideas of secure quantum networking, quantum anonymous notification, and quantum random number generation to strengthen the integrity, confidentiality, and privacy of microgrid networks. To substantiate the possible benefits of using quantum augmented microgrids, we simulate a practical high-impact classical attack: a traffic analysis and priority-action spoofing campaign that can (1) deanonymize the anonymous notification for a high-priority action, (2) force excessive key usage, and (3) induce harmful allow/block operations at the control level. We quantify how these attacks affect information leakage, spoof acceptance, key sufficiency, and operational outcomes such as latency, deadline misses, unserved energy, etc. This quantum augmented microgrid (QuAM) framework lets us evaluate trade-offs between privacy, availability, and the operational cost of mitigation (cover traffic, verification delays, and key-rotation policies), further paving the path for the study of more nuanced attacks that arise due to the use of quantum-classical integrated frameworks.

We present an effective numerical method that can be used to straightforwardly calculate the full spectrum of primordial gravitational waves produced during inflation and reheating. Our method is based on the Bogoliubov approach with several key improvements to overcome its shortcomings such as numerical instabilities at high frequencies and issues with tachyonic modes. We also present a few useful analytical examples from which one can gain crucial insights into the numerical instabilities. The improved method allows us to demonstrate that anharmonicity of inflaton oscillations can leave interesting fingerprints on the high-frequency part of the GW spectrum. Our numerical code is publicly available on [GitHub](We present an effective numerical method that can be used to straightforwardly calculate the full spectrum of primordial gravitational waves produced during inflation and reheating. Our method is based on the Bogoliubov approach with several key improvements to overcome its shortcomings such as numerical instabilities at high frequencies and issues with tachyonic modes. We also present a few useful analytical examples from which one can gain crucial insights into the numerical instabilities. The improved method allows us to demonstrate that anharmonicity of inflaton oscillations can leave interesting fingerprints on the high-frequency part of the GW spectrum. Our numerical code is publicly available on GitHub https://github.com/xunjiexu/Unified-Bogoliubov.git.

These notes give a concise introduction to General Relativity at the advanced undergraduate level, starting from the weak field limit and gravitational waves, then introducing curved manifolds and Riemannian geometry. The nonlinear gravitational action is used to derive the nonlinear field equations, with applications to black holes and cosmology. It is assumed that special relativity and electromagnetic waves have been previously studied. Some advanced topics such as Rindler and Hawking radiation are derived, and recent developments in gravitational wave detection are briefly covered. Problems are included, both those suitable for homework, and simpler ones that could be worked out by students during class sessions.

Quantum imaginary-time evolution (QITE) is a fundamental framework for preparing ground and thermal states, yet its computational cost scales significantly with the evolution duration $τ$. Reducing this duration is critical for practical quantum advantage. Here, we establish a unified theoretical framework for the Mpemba effect in QITE -- a counterintuitive phenomenon where a state initially farther from the ground state relaxes to it faster than one initially closer. We derive a remarkably simple necessary and sufficient condition for the occurrence of this effect, showing it is uniquely determined by the population ratios of excited states to the ground state. For practical state preparation, we introduce a rigorous sufficient condition for the finite-time Mpemba effect, ensuring the crossing occurs before reaching a prescribed proximity threshold. Furthermore, we unveil unique dynamical features, including a multiple-crossing phenomenon in multi-level systems and simultaneous intersections for collinear initial states. Our results provide criteria for identifying favorable initial states in QITE and offer deep insights into the speed limit of quantum state preparation.

How many black-box queries to a quantum channel are needed to learn its full classical description? This question lies at the heart of quantum channel tomography (also known as quantum process tomography), a fundamental task in the characterization and validation of quantum hardware. Despite extensive prior work, the optimal query complexity for quantum channel tomography is far from fully understood. In this paper, we study tomography of an unknown quantum channel with input dimension $d_1$, output dimension $d_2$, and Kraus rank at most $r$, to within error $\varepsilon$. We identify the dilation rate $τ= r d_2 / d_1$ (which always satisfies $τ\geq 1$ due to the trace preservation of quantum channels) as a key parameter, and establish that the optimal query complexity of channel tomography exhibits distinct scaling laws across three regimes of $τ$. - In the boundary regime ($τ= 1$): we show that the query complexity is $Θ(r d_1 d_2/\varepsilon)$ for Choi trace norm error $\varepsilon$, and is upper bounded by $O(\min\{r d_1^{1.5} d_2/\varepsilon, r d_1 d_2/\varepsilon^2\})$ and lower bounded by $Ω(r d_1 d_2/\varepsilon)$ for diamond norm error $\varepsilon$. - In the away-from-boundary regime ($τ\geq 1+Ω(1)$): we show that the query complexity is $Θ(r d_1 d_2/\varepsilon^2)$ for both Choi trace norm and diamond norm errors $\varepsilon$. Our results uncover a sharp Heisenberg-to-classical phase transition in the query complexity of quantum channel tomography: at $τ=1$, the optimal query complexity exhibits Heisenberg scaling $1/\varepsilon$, whereas for $τ\geq 1+Ω(1)$, it exhibits classical scaling $1/\varepsilon^2$. In addition, we show that in the near-boundary regime ($1< τ< 1+o(1)$), the query complexity exhibits a mixture of Heisenberg and classical scaling behaviors.

Altermagnetism, known for its non-relativistic spin-split band structures with yet compensated moments, is being intensively investigated. Discovering new altermagnetic materials with characteristics suitable for practical use remains an important ongoing task. Recently a metallic room-temperature altermagnet candidate Rb1-xV2Te2O with a layered structure and d-wave spin symmetry has been reported based on experimental results from the spin-resolved photoemission spectroscopy and scanning tunnelling microscopy/spectroscopy (STM/STS) measurements. Here we report neutron powder diffraction (NPD) investigations on the magnetic structure of Rb1-xV2Te2O, which shows a G-type antiferromagnetic structure below the transition temperature of 337 K. The result is different from the original theoretical expectation, which might lead to new insights on the physics of this altermagnet candidate.

Gravitational lensing constitutes one of the most direct observational manifestations of spacetime curvature and provides a powerful probe of compact astrophysical objects. In this work, we present a comprehensive analysis of the bending of light in curved spacetime, beginning with the fundamental aspects of gravitational lensing and the Newtonian approximation to light deflection. The relativistic formulation of lensing is then developed through the lens equation, lensing potential, and a geometrical interpretation of Fermat's principle using an effective refractive index in curved spacetime. Photon trajectories and light deflection are subsequently investigated in static, spherically symmetric geometries, followed by a detailed study of photon motion in the equatorial plane of the Kerr spacetime. Analytical expressions for the closest approach distance and critical parameters governing photon orbits are derived. Furthermore, the bending angle is examined using the Rindler-Ishak method and the Gauss-Bonnet theorem within the optical geometry. Finally, the analysis is extended to axisymmetric spacetimes using the OIA and GW-OIA formalisms, providing a unified geometrical framework for computing light deflection in both static and rotating gravitational fields.

Reliable quantum computation requires fault-tolerant protocols to prevent errors from propagating during syndrome extraction in quantum error correction. We present a novel fault-tolerant syndrome extraction technique for CSS codes, which we refer to as the cut-cat state scheme. While each ancilla qubit interacts non-fault-tolerantly with a pair of data qubits, we introduce additional cat stabilizer measurements to identify and correct the resulting hook errors. Our approach maintains the key benefit of cat-based extraction, i.e., parallelized data qubit interactions, while reducing the number of simultaneous qubits required by more than half. Compared to flag-based state-of-the-art protocols, the cut-cat scheme offers a notable advantage in terms of two-qubit gate count as the code distance increases.

Quantum Noise Characterization (QNC) is indispensable for benchmarking and mitigating errors in Noisy Intermediate-Scale Quantum (NISQ) devices. However, traditional Quantum Process Tomography (QPT) suffers from an exponential parameter explosion $O(4^N)$, severely hindering its scalability. In this paper, we propose a Hierarchical Progressive Optimization (HPO) framework to efficiently extract high-order spatial crosstalk in multi-qubit systems. By introducing a mathematically rigorous combinatorial projection mask, the HPO framework strategically freezes foundational low-weight topologies and exclusively isolates high-weight Pauli correlations. This progressive masking mechanism effectively reduces the optimization complexity from $O(4^N)$ to a scalable $O(N \cdot 4^w)$, successfully mitigating the barren plateau phenomenon. Simulations show that our method achieves a remarkable parameter compression rate of 96.3% on a 5-qubit system while maintaining machine precision convergence. Furthermore, to validate its practical utility, we apply the extracted spatial crosstalk model to perform Quantum Error Mitigation (QEM) on a deep-circuit 10-qubit Harrow-Hassidim-Lloyd (HHL) algorithm. Compared to the traditional global depolarizing baseline, the HPO-guided mitigation scheme breaks the unmitigated crosstalk bottleneck, achieving an unprecedented state fidelity recovery from 0.7431 to 0.9381 ($ΔF \approx 19.5\%$). Our work provides a scalable, highly accurate, and indispensable blueprint for modeling and mitigating complex multi-body errors in large-scale quantum algorithms.

Within the present noisy intermediate-scale quantum-computing era, hybrid quantum-classical-processor algorithms have emerged as promising avenues for tackling electronic-structure eigenproblems. Among them, the so-called ensemble-variational quantum eigensolver has been designed to treat ground and excited states on an equal footing and proven effective in capturing features such as conical intersections and avoided crossings between two electronic states, as we recently demonstrated for formaldimine. We also showed on that example how the underlying ensemble-variational principle was prone to provide a quasi-diabatic representation "for free". To date, this method has been limited to computing only two eigenstates of a Hamiltonian. The aim of the present paper is to show how and under what conditions this can be generalized to models that involve three coupled electronic states or more. Our approach relies on designing a parameterized basis transformation that can directly be implemented on a quantum computer for further post-treatment. This nontrivial step is accompanied by the development of quantum circuits specifically adapted to the several states of interest. An algebraic optimization strategy for the parameters of the basis transformation is formulated to obtain the target eigenstates as well as the optimally diabatic states under various objective flavors of the ensemble-variational principle. Our approach was tested for addressing the first three coupled electronic states of the H$_4^+$ molecular ion as a proof of principle, with three electrons in four spatial orbitals, along various geometries.

Practical utilization of Gottesman-Kitaev-Preskill (GKP) qubits requires not only the preparation of logical basis states, but also the ability to prepare and evaluate arbitrary logical qubit superpositions. Currently, this is typically done via quantum state tomography, which is resource-intensive. We introduce a family of positive semidefinite Hermitian operators, one for each point on the logical Bloch sphere, whose unique zero-eigenvalue ground states are the corresponding ideal GKP qubit states. We show that the expectation value of each operator serves as a witness of non-Gaussianity, and corresponds to twice the logical infidelity for states in the ideal logical GKP subspace. Furthermore, the truncated finite-dimensional counterparts of these operators yield physical approximations of arbitrary logical GKP states as their ground states. The evaluation of the proposed operators requires only three quadrature measurements, making this framework practical for both the experimental characterization and numerical optimization of GKP state preparation circuits.

We establish a deformation framework for highly symmetric solutions to the Einstein equations. In this framework, four-dimensional metrics are constructed from three-dimensional η-Einstein metrics admitting a deformation determined by a single function. Under this deformation, the resulting spacetime solves the Einstein equations with a string-cloud source. Within this framework , a wide range of symmetric spacetimes can be treated in a unified manner. These include FLRW, Kantowski-Sachs, and LRS Bianchi cosmological models (including Taub-NUT-(A)dS solutions), as well as Reissner-Nordström-(A)dS black holes admitting spherical, planar, or hyperbolic symmetry. In the cosmological setting, the deformation leaves the evolution equations for the scale factors unchanged, and hence the expansion history coincides with that of the corresponding undeformed models. For the deformed Reissner-Nordström-(A)dS black holes, the structure of Killing horizons is insensitive to the deformation.

In this paper we introduce a new general framework for the study of phenomenological quantum gravity theories (PQG). The key idea is the introduction of two different types of spacetime, an observer-independent spacetime (modeled by a smooth orientable manifold) and an observer-dependent one (which has an inherently discrete causal set structure). The interaction between the two allows us to prove the main result of the paper: relative locality can be obtained in general PQG models, regardless of momentum-space curvature. {We also discuss the treatment of spacetime symmetries in our model, and introduce a direct link between spacetime symmetries and relative locality effects.} Our construction is presented in a coordinate-independent way and is based on fibre bundles where spacetime, rather than momentum-space, is the base manifold. This makes causality manifest even in general models with relative locality. Furthermore, it allows for the application of this formalism to cosmology on general backgrounds, something which is not clearly possible in the canonical approach to relative locality, where momentum-space serves as the base manifold.

In this paper, we study the quasinormal modes of the generalized Joshi-Malafarina-Narayan (JMN) naked singularity spacetime using the exact Wentzel-Kramers-Brillouin (WKB) method. Working in the complex radial plane, we construct the exact WKB momentum function, determine its turning points, and compute the associated Stokes geometry for representative quasinormal mode (QNM) frequencies. We obtained a bow-shaped deformation of Stokes curves on the side of the complex plane containing the central singularity in JMN spacetime. We show analytically that this structure originates from the logarithmic branch-point singularity of the WKB phase at $(r = 0)$, which is absent in Schwarzschild spacetime. This establishes the bow-shaped Stokes topology as a direct signature of the naked singularity in the global analytic structure of the perturbation equation. Our results demonstrate that exact WKB analysis provides a powerful framework for probing the analytic structure of compact objects, and suggest that topological features of Stokes geometry may offer a new avenue for distinguishing black holes from horizonless alternatives.

Stark-localized quantum probes have recently been shown to enable quantum-enhanced weak-field sensing with polynomial or super-polynomial scaling. In this paper, we show that the spatial geography of the encoded field can elevate this advantage to a genuine exponential scaling. We study a one-dimensional Stark probe subject to an exponential gradient profile, \(V_j=e^{aj}\), and analyze its metrological performance in both equilibrium and non-equilibrium regimes, for single-particle and interacting many-body settings. In the equilibrium single-particle case, we derive an analytical lower bound showing that the quantum Fisher information grows exponentially with system size, and confirm numerically that this enhancement persists throughout the extended phase and at the localization transition. We further show that the same exponential scaling survives for mid-spectrum eigenstates and in the interacting many-body regime. This advantage remains intact under a fair resource analysis because the relevant preparation gap closes only algebraically, so the polynomial preparation overhead cannot offset the exponential gain in sensitivity. In the non-equilibrium regime, a simple product-state initialization followed by free evolution already retains exponential enhancement, eliminating the need for cooling, adiabatic preparation, or operation within a narrowly tuned sensing window. Finally, we outline a superconducting implementation based on flux-tunable transmon qubits with graded mutual inductive coupling to a common sensing bus. Our results identify exponentially graded Stark potentials as a distinct and experimentally plausible route to weak-field sensing with exponentially improving precision.

Advancement in quantum information and quantum technologies has ushered in a new era of technological revolution in large scale atomistic simulation and efficient system on a chip device fabrication. This has led to innovative ways of harnessing rigorous search algorithms for functional quantum materials and steered scientists to dig deeper into the world of quantum phenomenon and applications. In this work, we delineate the advanced electronic structure and vibrational properties utilizing the popular meta-GGA functionals, spectral signatures of the Raman active phonon modes, explored their average mean free paths, and whether they conserve helicity, by leveraging first principles density functional theory and density functional perturbation theory. A systematic analysis of the role of phonon lifetimes, consequences of phonon-phonon and three phonon scattering rates and phonon linewidths have been presented. Further, a study of the the frequency and temperature dependent Gruneisen parameter has been employed in conjecture with the temperature dependent thermal expansion and thermal conductivity to portray the effect of anharmonicity in the phonon spectra of these two materials. Finally, we provide strategies for tuning the properties of these materials in an effort to improve their efficacy for advanced thermoelectric, photovoltaic and optoelectronic device applications.

Various classical machine learning models, including linear regression, kernel methods, and deep neural networks, exhibit double descent, in which the test risk peaks near the interpolation threshold and then decreases in the overparameterized regime. However, this phenomenon has received less attention in the quantum setting. In this work, we investigate the double descent phenomenon in quantum kernel ridge regression (QKRR). By applying deterministic equivalents from random matrix theory (RMT), we derive an asymptotic expression for the test risk of QKRR in the high-dimensional limit. Our analysis rigorously characterizes the interpolation peak and reveals how explicit regularization can effectively suppress it. We corroborate our theoretical results with numerical simulations, demonstrating close agreement even for finite-size quantum systems.

Detecting coherence transfer in complex quantum networks can be challenging due to uncharacterized experimental conditions and limited system access. Here, we use static and dynamic coherence features to introduce a nonlinear criterion for identifying coherence transfer. The criterion requires only two measurement settings for network-state populations in an experimental state basis, regardless of the network's size. It remains valid even when the verification capabilities of checkpoint nodes are uncharacterized. The principle and method are general, encompassing networks with different access levels and scenarios, from those requiring no input changes to those involving coherence dynamics in the time domain. Experimentally, using remote state preparation and entanglement swapping, we transfer single polarization qubits and polarization-entangled pairs in four- and six-photon entanglement networks. The criterion provides experimental evidence of coherence transfer in multi-photon entanglement networks. Our findings offer a practical tool for coherence transfer in quantum information and open quantum systems in networks.

We introduce map-dependent quantum characteristic functions constructed from the normalized Choi operator of quantum dynamical maps. We prove a Bochner--Choi positivity theorem establishing that the positive-type condition of the associated Gram matrix is equivalent to complete positivity of the underlying quantum channel. Applying the construction to intermediate dynamical maps, we obtain a characterization of CP-divisibility in terms of positivity of two-time characteristic functions. Numerical examples for amplitude damping and pure dephasing models demonstrate that negativity of the Gram matrix coincides with the breakdown of CP-divisibility and the emergence of information backflow. The proposed framework provides a new bridge between characteristic-function methods in quantum statistics and structural properties of quantum dynamical maps.

Quantum dots are nanometer-scale semiconductor particles that exhibit size-dependent quantum mechanical properties. In this work, we investigate the dynamics of quantum correlations, quantified by the concurrence and the quantum coherence, in a bipartite system of coupled double quantum dots. The analysis is carried out within both Markovian and non-Markovian regimes, and further extended to different noisy quantum channels, including amplitude damping, phase flip, and phase damping. Our results show that environmental memory plays a crucial role in the preservation of quantum correlations, leading to oscillatory behavior and partial revivals in the non-Markovian regime, in contrast to the monotonic decay observed under Markovian dynamics. Moreover, distinct decoherence mechanisms induce qualitatively different effects: dissipative channels rapidly suppress correlations, while phase-based channels lead to either redistribution or gradual degradation. A key finding is that quantum coherence exhibits a higher robustness compared to entanglement under all considered conditions, highlighting its relevance as a reliable quantum resource in noisy environments. These results provide valuable insights into the control and protection of quantum correlations in realistic solid-state systems.

A CaTherine wheel is a space-filling curve $f : S^1\to S^2$ such that for every closed interval $J\subset S^1$, $f(J)$ is homeomorphic to a closed disk and $f(\partial J)$ is contained in $\partial f(J)$. A CaTherine wheel gives rise to a pair of disjoint, dense topological trees in $S^2$ which roughly speaking lie to the left and right of $f$. We give necessary and sufficient conditions for a topological tree in $S^2$ to arise as one of these trees for some CaTherine wheel $f$. We apply this result to show that there is a unique CaTherine wheel corresponding to the geodesic tree rooted at $\infty$ for the $γ$-Liouville quantum gravity (LQG) metric, for $γ\in (0,2)$. In other words, we construct the space-filling curve which is the contour exploration of the LQG geodesic tree.

Periodic driving plays a central role in quantum control, but its application in interacting spin systems is often restricted to near-resonant conditions, where standard averaging techniques remain valid. Here we investigate how detuning from resonance can be used to dynamically spin-lock a dipolar-coupled ensemble. We show that the combination of offset and pulse structure generates an effective Rabi field with sharply structured amplitude and tilt. This behavior - supported by a semi-analytical framework, numerical simulations and experiment - enables new approaches to many-body system control, here exemplified via offset-induced reversible interconversion of Zeeman and dipolar order, and heterospin polarization transfer away from rf-field matching conditions. Further, we leverage artificial-intelligence-assisted sequence design to explore regimes with long control cycles - where average Hamiltonian theory breaks down, but effective locking persists - opening pathways to rich offset-dependent responses. These findings position offset-enabled dynamic locking as a promising tool for quantum sensing, energy transfer, and spin-order manipulation beyond traditional approaches.

Gravitational spectroscopy tests the coupling of gravity to matter by measuring gravitationally induced frequency shifts of quantum transitions. While modern optical clocks probe the gravitational response of electronic transitions with extraordinary precision, tests in the nuclear sector have not progressed since the Mössbauer measurements of the gravitational redshift by Pound and Rebka. Here we introduce a new approach to nuclear gravitational spectroscopy based on phase-sensitive heterodyne interferometry of time-resolved nuclear resonant scattering of synchrotron radiation. In this scheme the gravitational redshift appears as a slowly accumulating phase drift of a delayed heterodyne beat signal, converting nuclear gravitational spectroscopy from energy-domain detection to time-domain interferometry. A Fisher-information analysis supported by Monte Carlo simulations and experimentally confirmed photon count rates shows that the nuclear gravitational redshift of $^{57}$Fe can be detected within hours on a few-meter-scale vertical baseline, with percent-level precision on deviations from general relativity becoming accessible on day-scale timescales. The method thus establishes an experimentally realistic and scalable platform for precision tests of gravitational coupling to nuclear structure.

We develop an exact theory of coherence-selective stroboscopic resetting for quadratic open quantum systems within the single-particle density-matrix formalism. We focus on the survival of coherences and the associated thermodynamic cost at the stroboscopic fixed point. To this end, we introduce a one-parameter family of reset channels that continuously interpolates between complete coherence erasure and complete coherence preservation. This unifies the reset-map description, the repeated-interaction and evolving-correlation endpoint channels, and the thermodynamic cost of environmental reinitialization. For a single fermionic level coupled to a structured semi-infinite tight-binding bath, we derive the exact affine stroboscopic map, solve for its unique fixed point, and compute the retained coherence spectrum, the post-reset occupation, and the reset heat current. We find that retained coherence increases monotonically with the retention parameter, whereas the reset heat current is generically nonmonotonic and is maximized at an intermediate operating point. Thus the protocol that stores the most coherence is not the one that dissipates the most heat. Exact operating diagrams further show that coherence-optimal and coherence-per-cost-optimal protocols are both driven toward the coherence-preserving endpoint, while the heat-optimal protocol depends strongly on the reset interval. We also show that this coherence-cost geometry survives at nonzero chemical potential as a filling-biased deformation of the same fixed-point tradeoff, rather than as an independent particle-current optimization problem. These results establish coherence-selective resetting as a distinct control principle for structured-bath open quantum systems and provide an exactly solvable benchmark for memory engineering and thermodynamic optimization under repeated environmental reinitialization.

As physical systems, qubits must evolve from input to output state. We describe a simple scheme in which the effect of a quantum gate is described by the action of an effective Hamiltonian acting for some characteristic time. This model shows that the action of common unary gates is to induce Bloch sphere trajectories along lines of latitude relative to an eigenvector of the gate. Such trajectories would immediately move a `rebit', initially confined to a line of latitude, off this line and acquire a complex phase. The role of the complex phase in bringing about the entanglement of two qubits is also highlighted. It is then asked whether such dynamics could be modelled using real QM. It is shown that the continuous evolution required for such dynamics can only be provided by members of the special orthogonal group of the vector space. Since the matrices representing many quantum gates of interest have determinant -1, no real special orthogonal operators can model their evolution if the dimension of the real vector space is the same as that of the complex space. Next we look at the mapping from a complex vector space of dimension $N$ to a real space of dimension $2N$ that is often used to construct `real' QM. It is shown that this is just an isomorphic mapping from the scalar representation of complex numbers to their $2\times2$ matrix equivalents, so that the resulting matrices are actually represent complex matrices. Finally, we investigate the endomorphism of real vector spaces of dimension $N = 2^{n}$, where $n \in \mathbb{Z}^{+}$, suitable for the modelling of $n-1$ qubits. We confirm that the mapping from $\mathbb{C}^{2^{n-1}} \to \mathbb{R}^{2^{n}}$ only maps elements of $\mathrm{End}(\mathbb{C}^{2^{n-1}})$ to a restricted subspace of $\mathrm{End}(\mathbb{R}^{2^{n}})$ that reproduces the `real' representation of complex matrices.

We present the cosmic expansion scenario in the framework of $f(T)$ gravity by employing a dark energy equation of state (EoS) parameter. Specifically, we proceed with the power-law form of the function $f(T) = α$$(-T)^{n}$, in conjunction with the Gong-Zhang parametrization of the dark energy EoS. We derive the expansion rate in terms of the redshift for the considered model, providing deeper insights into the underlying cosmic dynamics. The model is further utilized to explore the expansion history of the universe and the evolution of several cosmological parameters. By using the Bayesian methods based on the $χ^{2}$-minimization technique, the median values of the model parameters are determined for the cosmic chronometer (CC) and joint (CC+Pantheon) datasets. The evolution of the deceleration parameter, energy density, pressure, EoS parameter and the energy conditions for dark energy is analyzed in detail. The model captures the observed acceleration as a transient phenomenon, followed by future deceleration. Additionally, the nature of both geometrical and dynamical diagnostics robustly indicates a quintessence-like behavior at the present epoch. Finally, the thermodynamic viability of the model is confirmed through the generalized second law of thermodynamics and the estimated age of the universe further supports the model's compatibility with astronomical observations.

In this work, we develop a theoretical description of the collective behavior of interacting dipolar planar rotors by using time independent perturbation theory and a small angle quadratic approximation. The ground state properties for both the ordered and disordered quantum phases of the system are directly calculated and analyzed. Time-independent perturbation theory is shown to be appropriate for the disordered phase. For the ordered phase, we construct a quadratic approximation based on the stable equilibrium configurations of the dipolar ordering; we show that the inclusion of the quartic terms from the expansion of the potential energy are essential to correct the shift in the energy spectrum due to quantization ambiguities. Numerical techniques such as Exact Diagonalization and Density Matrix Renormalization Group are used for the benchmark the quality of both approximations.

The admissibility of a gauge-fixing is governed by the invertibility of $Δ=\{σ^a,γ_b\}$ where $σ^a$ are gauge-fixing conditions and $γ_b$ are independent first-class constraints. We prove, via the Schur complement, that the determinant of the combined constraint matrix $\mathcal{M}=\{Φ_A, Φ_B\}$ built from all constraints and gauge-fixing conditions factorises as $\det\mathcal{M}\approx\pm(\detΔ)^2\det C$, where $C$ is the second-class constraint matrix, providing an alternative criterion for admissibility. Since $\det C\neq0$ by definition, the second-class sector decouples entirely from the gauge-fixing sector. In the algebraic case, this factorisation identifies the Hamiltonian admissibility criterion of Henneaux and Teitelboim with the Lagrangian completeness criterion of Motohashi, Suyama, and Takahashi. We identify a metric ansatz as gauge-fixing at the action level and analyse completeness in the context of spherically symmetric spacetime. The factorisation ensures that completeness is robust to the second-class sector that arises in modified theories of gravity.

Recently, Pang et al reported a novel polarization transfer scheme applicable to three-spin systems, whereby a rotating-frame NMR analogue of the cross effect could transfer polarization between; e.g., two 13Cs and an 15N in a single crystal. The present work furthers this scheme to the case of powder NMR under magic angle spinning (MAS) conditions, as well as to solution NMR. It is found that in all such cases a second-order average Hamiltonian can transfer polarization between non-equivalent, coupled abundant spins (e.g., two 1Hs) prepared in anti-longitudinal magnetization states, and the spin-locked magnetization of a rare spins (e.g., one 13C). The average Hamiltonian for such three-spin (S1-S2) to I transfer was derived for both liquids and solids, and found in good quantitative agreement with numerical simulations and experiments. At an optimal transfer condition whereby an I-spin RF irradiation field matches the S1-S2 chemical-shift-difference, a maximum polarization enhancement equal to the ratio of gyromagnetic ratios is achieved; as explained and demonstrated in the study, ca. half of this can be effectively obtained for I = 13C in powdered solids and in multi-spin systems in solutions. All such processes display an oscillatory nature, meaning that the transverse spin-locked polarization of a rare spin can become anti-longitudinal magnetization of abundant spins -without ever pulsing on the latter. The roles played by many-body interactions, RF inhomogeneities, and interferences of other coherences during the execution of these novel forms of cross-polarization were investigated, and are exemplified with experiments and simulations.

Recent work investigated the use of Stimulated Brillouin Scattering (SBS) to measure changes in fiber parameters, thereby enhancing the security of a Quantum Key Distribution(QKD) system. In this work, we cast the problem into a binary hypothesis testing scenario, with the goal of comparing the state of the art with potential future quantum technology-based detection methods. We derive an effective input-output model for the Stimulated Brillouin Scattering (SBS) interaction, and utilize it to compare three detection methods: First, the established state of the art. Second, a photon-counting based method which will likely be available in the near future. Finally, we compare to the ultimate quantum limit, which is given by the quantum error exponent of asymmetric hypothesis testing.

Although information, strictly speaking, is not a physical entity, it generally requires physical entities as its carriers, e.g., writing it down on paper, encoding it with quantum particles, or transmitting it using electro-magnetic fields. And it seems natural that these carriers cannot travel faster than light. Here we reveal that if we use quantum correlations as the carrier of information (either quantum or classical), then it can display a kind of nonlocality, which bears both similarities to and distinctions from the nonlocality of physical particles. Notably, though superluminal signaling is still not allowed so that the special relativity is not violated, it is possible to select at our will whether to decode the information at one location, or to dispatch it to another location far away (i.e., to give up the chance of decoding the information and let it be decodable in somewhere else only) without needing the assistance of classical information, so that it occurs instantaneously without being limited by the speed of light. This phenomenon differs sharply from the nonlocality of physical particles that we once knew, where whether a particle can be detected in one location or another is governed by quantum uncertainty, which cannot be chosen freely.

Density functional theory with plane-wave basis sets is widely employed in computational materials science, including applications to isolated molecular systems. However, the inadequate description of electron correlation remains a fundamental limitation. Accurate correlation treatments based on many-body Hamiltonians require reliable representations of both occupied and virtual orbitals, yet virtual orbitals are often poorly described in conventional computational schemes, resulting in reduced accuracy. In this work, we introduce localized correlation-converged virtual orbitals (LCCVOs) as an efficient basis for constructing accurate many-body Hamiltonians in molecular systems. Using a substantially reduced number of orbitals, the LCCVO framework yields dissociation energies for singlet, doublet, and triplet molecules that are comparable to, and in many cases exceed, those obtained with high-level correlation-consistent basis sets such as cc-pVXZ (X = D, T, Q, 5). These results demonstrate the efficiency, scalability, and robustness of the LCCVO approach for high-accuracy quantum chemical calculations.

We formulate a first-order thermodynamic description of Jordan-frame tensor--multi-scalar gravity. From the Einstein-like field equations we obtain the exact covariant $1+3$ decomposition of the geometric sector and interpret it as an effective imperfect fluid. In a generic frame, the heat flux can be written exactly as $q_a^{(g)}=-χ(a_a+W_a)$, with $χ=-\dot{\FF}/(8π\FF)$, where $\FF=\FF(φ^C)$ is the nonminimal coupling function, and with $W_a$ the residual temperature-gradient sector. In the $\FF$-comoving frame this yields the inertial variable $χ_{\FF}\equiv K_{\FF}T_{\FF}$ together with a generally nonvanishing spatial contribution $W_a^{(\FF)}$ sourced by scalar directions not aligned with the coupling, showing that the multi-field thermodynamic description is not generically reducible to a single $KT$-type quantity. We derive transport equations for $χ_{\FF}$, for the field-space thermal vector $χ^A$ and covector $χ_A$, and for the residual gradient sector. We further introduce the scalar diagnostics $\mathfrak D_χ=χ_Aχ^A$ and $\mathfrak D_{\rm grad}=\mathcal{B}_{AB}\D_a^{(\FF)}φ^A\D^a_{(\FF)}φ^B$, where $\mathcal{B}_{AB}$ is the field-space kinetic matrix of the multi-scalar theory and $\D_a^{(\FF)}$ is the covariant derivative projected orthogonally to the $\FF$-comoving 4-velocity. These diagnostics characterize the full time-like and spatial multi-scalar sectors and lead to a precise GR-attractor criterion: freezing the effective coupling is, in general, weaker than full relaxation to the GR sector. Finally, we construct the entropy current and entropy production in the coupling frame and show that homogeneous cosmology suppresses the spatial sector while retaining nontrivial time-like multi-scalar thermal dynamics.

Tetraphenylpyrazine (TPP) and 2,3,4,5-tetraphenyl-1H-pyrrole (TePP) are closely related heterocycles bearing four phenyl substituents, whose structural similarity makes them a useful pair for comparing how intramolecular flexibility influences excited-state relaxation and emission in the gas phase and in the solid state. TPP is a prototypical solid-state luminescence enhancement (SLE) emitter, exhibiting a markedly increased quantum yield upon molecular aggregation. In contrast, TePP displays similar quantum yields in solution and solid state, characteristic of dual-state emission (DSE). This behaviour indicates that intramolecular rotations are already significantly hindered in the isolated-molecule regime, consistent with our previous observations for TPP and other solid-state emitters (Hernández-Rodríguez et al., ChemPhysChem, 2024, 25, e202400563). To unravel the excited-state dynamics underlying this contrasting behaviour, we performed mixed quantum-classical trajectory simulations on a single molecule of TPP and TePP employing the surface-hopping method. Twelve singlet states were included at the TD-B3LYP-D3/def2-SVP level, which were previously benchmarked against coupled cluster methods. Simulated observables such as gas phase ultrafast electron diffraction (GUED) and time-resolved fluorescence (TR-FL) signals allow us to dissect the distinct deactivation pathways operating in both systems in the gas phase, while also providing mechanistic insight into how these pathways are expected to evolve in solution and solid-state environments.

Quantum classification faces two key challenges. First, the difficulty of distinguishing between different classes varies: some class pairs are easy to separate, while others are more challenging. Second, practical execution is affected by noise, finite sampling, and measurement overhead. To address these issues, we propose PAPUS, a framework for pair-adaptive quantum classification in Pauli space. The method evaluates candidate upload circuits using low-weight Pauli features and formulates upload design as a structured model selection problem based on discriminative representations. By dynamically adjusting circuit complexity according to class-pair difficulty, the framework achieves a better balance between classification accuracy and resource efficiency. Experiments on 9 data sets with 474 tasks show that PAPUS achieves a favorable balance between predictive performance and execution cost. Specifically, PAPUS attains classification accuracies above 90% in both local noiseless simulation and the IonQ noisy simulator, while requiring substantially lower measurement and circuit cost (fewer total measurement shots and fewer quantum gates for data upload). Compared with the two conventional baselines, template_cv and kta_exact, PAPUS also shows much stronger robustness under noise: accuracy decreases by only 1.67% in the noisy setting, whereas both baselines degrade by 9.44%.

Understanding the structure and stability of open quantum dynamics is increasingly important for both fundamental studies of nonequilibrium quantum systems and the development of quantum simulation algorithms. In this work, we introduce a structural framework for Markovian open quantum systems that characterizes Lindbladian generators in terms of two scalar quantities: the dissipative strength and the nonnormality. We show that normal generators admit an exact decoupling between dissipative and norm-preserving dynamics, leading to purely exponential behavior governed by the dissipative scale. In contrast, nonnormality is an intrinsically dissipative feature: it vanishes in the absence of dissipation but is not implied by it. Moreover, it is structurally constrained by the interplay between the Hermitian and anti-Hermitian components of the generator. For generic Markovian open quantum systems, we identify parametric regimes controlled by a dimensionless ratio between nonnormality and dissipative strength, governing the onset of transient amplification. These structural features have direct implications for quantum simulation. While Hamiltonian and normal dissipative dynamics exhibit stable evolution with standard scaling behavior, nonnormal generators can induce transient growth that amplifies numerical errors and increases simulation cost. Our results provide a unified generator-level perspective on irreversibility, stability, and quantum simulation of open quantum systems.

We propose a scheme to simultaneously cool multiple degenerate mechanical modes in optomechanical systems beyond the resolved sideband regime. In general, one of the main obstacles for cooling degenerate mechanical modes is the so-called dark-mode effect. The Duffing nonlinearities (mechanical nonlinearities) can be used to overcome the dark-mode effect of degenerate mechanical modes. A second-order nonlinear medium (optical nonlinearity) is introduced to accomplish the ground-state cooling of degenerate mechanical modes beyond the resolved sideband regime. We find the dark mode of degenerate mechanical modes can be broken when the mechanical nonlinearities of different mechanical modes are not very close. Our scheme paves the way toward the implementation of simultaneous ground-state cooling of degenerate mechanical modes of optomechanical systems beyond the resolved sideband regime in experiments.

Although non-Hermitian dynamics near exceptional points (EPs) provide a route to accelerated entanglement generation, entanglement can also be generated far from EPs at comparable or even higher rates. However, the behavior of such entanglement in open systems remains largely unexplored, rendering it highly susceptible to environmental noise. Here, we study entanglement dynamics in two coupled qubits described by a non-Hermitian Hamiltonian, where the dissipative rates are subject to stochastic noise. We focus on the regime away from EPs. Our approach demonstrates that, even in the presence of classical noise, rich dynamical control can be achieved, enabling highly efficient entanglement generation with a timescale that is significantly shorter than that of both Hermitian systems and EP-based non-Hermitian protocols. Additionally, the timescale is independent of the number of qubits, highlighting favorable scalability for multipartite entanglement generation and facilitating integration into future photonic quantum processors.

We develop a reference-renormalized (photon-sphere-free) normalization scheme for Gauss-Bonnet gravitational lensing at finite distance in static, spherically symmetric spacetimes. The method treats the curvature primitive used to reduce the Gauss-Bonnet curvature-area integral as a quantity defined only modulo an additive constant (an additive gauge freedom). We fix this gauge by matching to a physically chosen reference optical geometry in an outer regime where the physical geometry approaches that reference, thereby defining a unique renormalized discrepancy primitive $\mathcal{P}_e(r)$ by reference subtraction. The resulting master formula yields the Ishihara-Li finite-distance deflection angle without invoking any circular null orbit, while remaining fully compatible with orbit-normalized prescriptions whenever a suitable photon sphere exists (the two gauges differ only by a constant shift and give identical $α$). In asymptotically flat settings the canonical reference is Minkowski, while in Kottler-type backgrounds the canonical reference is de Sitter within the static patch, making the operational fiducial explicit. We validate the method by reproducing Ishihara's finite-distance weak-deflection formulas for Schwarzschild, Reissner-Nordström, and Kottler spacetimes, including the mixed $r_gΛ$ term in the Kottler case within the static-patch fiducial. We also present a demonstrative example in which orbit normalization is genuinely inapplicable because no circular null orbit exists in the physical optical region (the Janis-Newman-Winicour spacetime for $γ\le \tfrac12$). The result is a unified, geometrically transparent route to finite-distance lensing that preserves compatibility with orbit-normalized prescriptions whenever those apply.

Ion trap quantum computers often store qubits on field-sensitive S_1/2 ground state Zeeman levels of the valence electron, such as in 40Ca+, 88Sr+, and 138Ba+ atomic systems. We experimentally synthesize magnetically insensitive qubit states in multiple metastable electronic D_3/2 Zeeman levels in such an atomic system. We demonstrate coherent operations within the D_3/2 manifold of 138Ba+, including coherent flopping between the synthesized qubit states, and our results agree with theory. Such an encoding may allow for more flexible use of atomic levels for photonic interfaces, and with a measured improvement in the qubit coherence time T2* by a factor of 3, this lays the foundation for further improvement for quantum computing and network applications.

We study the estimation of parameters pertaining to non-Markovian quantum open systems, such as the dissipation rate and environmental memory time. A key challenge is identifying the optimal measurement time, which must allow sufficient time to acquire information about the environment, yet be short enough to avoid dissipation that erases the information. Using machine learning approaches, we develop an optimized control scheme trained over a representative ensemble to fix the optimal measurement time at a prescribed runtime. The protocol is robust to errors in the training process, enhances precision by exploiting non-Markovian memory effects, and achieves measurement uncertainties approaching the quantum limits set by the Cramér-Rao bound.

A simple translation between a standard representation of $\mathfrak{sl}_2\mathbb{C}$ and the complex-quaternions ($\mathbb{H}\otimes_\mathbb{R}\mathbb{C}$) is established and exploited to construct a novel hyper-complex description of the Dirac theory, electrodynamics, and ultimately the electroweak sector of the standard model. We find that coupling the constructed Dirac spinors to electromagnetism yields the correct magnetic moment for charged spin-1/2 particles. Extending electrodynamics to electroweak theory necessitates an algebraic distinction between the structures of the leptonic and Higgs fields not present in the standard model. The conditions of spontaneous symmetry breaking are explored using an alternative representation of weak isospin and hypercharge equivalent to an irreducible representation of $\mathfrak{su}(2)\oplus\mathfrak{u}(1)$ on $\mathbb{C}^4$. This alternative representation disagrees with the standard model on the overall signs of weak neutral currents.

Quantum teleportation is a cornerstone of quantum information processing, enabling the nonlocal transmission of quantum states across arbitrary distances using shared entanglement and classical communication. While the standard protocol typically employs Z-basis Bell-state measurements, this fixed-basis approach limits flexibility in practical quantum networks, where dynamic operations, hardware variability, and advanced communications demand alternative measurement bases. In this work, we introduce a multi-axis quantum teleportation protocol that generalizes the measurement process by allowing arbitrary basis choices. We provide a formal derivation and self-contained mathematical proof demonstrating that the input quantum state can be faithfully reconstructed under basis-adaptive restoration operations. By establishing a rigorous algorithmic and analytical foundation, this work validates the generalized teleportation protocol and paves the way toward advanced strategies for quantum communication. The demonstrations of the proposed protocol are available at: https://github.com/JJJayyyy/Multi-Axis-Quantum-Teleportation.

Claims of successful quantum teleportation are backed up by showing that fidelity exceeds some specified threshold, but whether fidelity is the performance metric and what the threshold should be has been a subject of vigorous debate. We construct adversarial models for quantum teleportation, i.e., involving cheating parties, and show that fidelity thresholds can be justified in the context of the type of adversary trying to prove unsuccessful quantum teleportation has been successful. In particular we show how previously established average-fidelity thresholds of 1/2 and 2/3 arise from our adversarial approach. Mathematically, we describe adversarial quantum teleportation as a multi-partite protocol with explicit quantum-logic circuits in both honest and cheating settings, and our methods are relevant beyond quantum teleportation to other quantum-information gadgets.

The memory effect for Robinson-Trautman waves is explicitly worked out. In a first step, we construct the combined frame rotation and coordinate transformation in which Robinson-Trautman waves are manifestly locally asymptotically flat at future null infinity. This allows us to apply well-established results on how to derive the memory effect in this context. In a second step, we construct a suitably improved generalized mass aspect that provides a local Lyapunov function for the flow in the sense that it is manifestly positive. News-free solutions are studied in detail and shown to coincide with the vacuum sector of Euclidean Liouville theory. They correspond to a boosted and rescaled Schwarzschild black hole. As a by-product, we show that the displacement and non-linear memory effects in locally asymptotically flat spacetimes at future null infinity are invariant under supertranslations and covariant under $\mathrm{BMS}_4$ Lorentz transformations and constant rescalings. A novel interpretation of modified flows that control the low harmonics in terms of keeping the system in its instantaneous rest frame is provided.

Efficient classical simulation has matured to a critical component of the quantum computing stack, driving hardware validation, algorithm design, and the study of structured quantum dynamics. Lie-algebraic simulation ($\mathfrak{g}$-sim) is a compelling approach: it replaces exponentially large Hilbert-space evolution by dynamics in a reduced adjoint space whose dimension is set by the dynamical Lie algebra (DLA) of the circuit, enabling efficient simulation whenever the DLA grows only polynomially with system size. In practice, however, existing applications of $\mathfrak{g}$-sim have been confined to free-fermionic (matchgate) regimes, and it has been unclear how to extend the paradigm to other structured circuits whose generators may have large Pauli expansions. Here we enable Lie-algebraic classical simulation beyond free fermions by identifying additional non-trivial families of polynomial-dimensional DLAs and introducing symmetry-adapted basis representations that make the adjoint space mapping tractable. In particular, we develop an explicit Pauli orbit basis for permutation-equivariant dynamics, supporting cubic-dimensional algebras despite exponential Pauli support, and a subspace-adapted (modified) generalized Gell-Mann basis for bounded Hamming-weight ($U(1)$-equivariant) dynamics, yielding polynomial costs on fixed excitation sectors. Together with streamlined routines for free-fermionic Pauli algebras and translation-invariant variants thereof, these constructions significantly broaden the practical scope of $\mathfrak{g}$-sim as a unifying simulation tool for structured quantum dynamics. Numerical benchmarks confirm favorable preprocessing scaling and validate large-scale proof-of-concept simulations.

The interaction of quantum fields with fractal and self-similar geometries encompasses multiple distinct physical regimes, including spectral geometry on intrinsic fractals, macroscopic self-similar Casimir configurations, and bounded Euclidean cavities with fractal boundaries. While the thermal equations of state and spectral asymptotics for these systems are well established, a cohesive treatment of the vacuum trace frequently conflates rigorous mathematical bounds with phenomenological models. In this manuscript, we systematically decouple these regimes and advance a unified effective framework combining the rigorous thermal trace of fractal radiation with a zero-temperature integrated vacuum trace for plate-like self-similar geometries. We demonstrate that for systems governed by a scale-dependent Casimir coefficient $C(d_s, \ln(d/\ell_*))$, the anisotropic stress-energy tensor produces an integrated vacuum trace proportional to its logarithmic running, $\partial_{\ln d}C$. We strictly differentiate this effective macroscopic backreaction from first-principles local trace anomalies on genuine fractal boundaries. Finally, we analyze finite-level ($n$) prefractal realizations, establishing the analytical prerequisites necessary to transition this effective formalism into a quantitatively predictive electromagnetic theory amenable to experimental verification.

Quantum sensing can enhance imaging performance by reducing measurement noise below the classical limit, thereby improving the signal-to-noise ratio (SNR) of acquired data. In conventional quantum imaging schemes, squeezing is applied independently to each pixel or spatial mode, leading to a quantum resource cost that scales linearly with image dimension. This approach implicitly separates quantum enhancement from classical post-processing, treating them as independent layers. In this work, we demonstrate that integrating quantum resource allocation with the guidance from classical compressive imaging, via co-design between the quantum hardware layer and the classical software layer, substantially reduces the required quantum resources. We employ principal component analysis (PCA) to identify a low-dimensional principal component subspace for measurement and apply squeezing selectively to the most informative spatial modes corresponding to these principal components. Our numerical experiments show that high-accuracy image classification and high-fidelity image reconstruction can be achieved with significantly fewer squeezed modes compared to pixel-wise squeezing. Our results establish a joint quantum classical co-design framework for resource-efficient quantum-enhanced imaging.

Here we present models of hot neutron star (NS) atmospheres consisting of thermonuclear ashes of various chemical compositions. These models are essential for studying thermonuclear flashes in X-ray bursting NSs in which nuclear-burning ashes are transported to the stellar surface. We consider four different mixtures, each dominated by helium, chromium, iron, or nickel. In addition to the opacity sources previously used in NS atmosphere modeling, we include photoionization from excited ionic states as well as approximately 5000 spectral lines. We also develop a method that enables the simultaneous treatment of Compton scattering and a large number of spectral lines. A key feature of the modeled NS atmospheres is the presence of a layer in the transition region between the optically thin and optically thick parts of the atmosphere where the radiation-pressure force increases significantly. This enhanced force sets an upper limit on the maximum attainable bolometric flux for a given surface gravity and chemical composition. The emergent spectra from the computed atmospheres display pronounced absorption edges, whose energies are determined by the dominant chemical species. We fit the model spectra using a diluted blackbody modified by a single absorption edge, and we investigate how the fit parameters depend on both the relative bolometric flux and the chemical composition of the atmosphere. Finally, we discuss constraints on these models imposed by the properties of X-ray bursts that exhibit absorption edges in their spectra, as observed in the systems HETE~J1900.1$-$2455 and GRS~1747$-$312.

For the timelike geodesic equations in Schwarzschild spacetime, three hidden conserved quantities were found recently, which are analogues of dynamical quantities related to the well-known Laplace-Runge-Lenz (LRL) vector in Newtonian gravity. In particular, the geodesic equations possess an LRL angle, an LRL Killing-vector time and an LRL proper-time, each of which is a conserved quantity for all timelike geodesics. The present work provides a natural symmetry interpretation for these three quantities by applying Noether's theorem in reverse to the geodesic Lagrangian. This yields three hidden symmetry transformations. They are shown to commute with the Killing isometries and act on the equatorial geodesics by separate shifts and scaling of the geodesic energy and angular momentum. Together with the Killing symmetries, these transformations comprise the complete Noether symmetry group of the timelike equatorial geodesic equations.

A low-energy nuclear transition in the isotope thorium-229 has been excited in thorium-doped crystals with laser light. This opens the perspective towards a highly stable and robust solid-state optical nuclear clock. The required laser radiation at 148 nm wavelength has so far been produced using pulsed laser systems where only a small fraction of the incident photons has been resonant with the narrow nuclear transition. Here we show that the nuclear resonance can be excited with a continuous-wave laser source with a power of less than 1 nW, and that the resonance signal can be detected in absorption rather than in fluorescence. This eliminates the slow nuclear fluorescence decay from the detection process and offers a considerable advantage for clock operation through fast signal acquisition. The laser is based on three sequential frequency doublings, starting from a diode laser at 1187 nm that is well suited for linewidth narrowing and for frequency comparisons with optical atomic clocks. We use absorption spectroscopy for the quantitative characterization of two different Th-centers in calcium fluoride and measure the isomeric shift between them. One of the centers shows a very small static electric crystal field gradient < 0.1V/$Å^2$, to be compared to gradients in the range of 100 V/$Å^2$ observed earlier. This indicates a center with high symmetry of the ions surrounding the Th nucleus, promising nuclear resonance lines that are less sensitive to the lattice spacing.

The aim of this paper is to provide a largely self-contained, compact and comprehensible introduction to the basic ideas behind correlation geometry, which underlies the theory of causal fermion system (CFS). A key focus here is on the manner in which the framework deals with gauge transformations, including diffeomorphisms via the principle of unitary equivalence. We will argue that, conceptually, the fundamental description of a physical system in terms of its correlation geometry is much closer to thermodynamics than quantum theory.

In this work, we investigate the optical, dynamical, and radiative properties of an Euler--Heisenberg black hole immersed in a perfect fluid dark matter (PFDM) background. We analyze the photon sphere and shadow, the scalar quasinormal-mode spectrum in the eikonal regime, the grey-body factor through the eikonal QNM correspondence, the sparsity of Hawking radiation, and the corresponding energy emission rate. Our results show that both the black-hole charge and the PFDM parameter significantly affect the photon sphere, shadow size, quasinormal frequencies, Hawking temperature, and emission profile, whereas the Euler--Heisenberg correction is typically subleading in the parameter range explored, although it may become more visible in strong-charge regimes for selected observables. Overall, the dark-matter environment provides the dominant imprint on the phenomenology of the system, indicating that shadow and ringdown-related quantities may serve as useful probes of PFDM effects within the approximations considered.

Nonassociative deformations of phase-space structures arise naturally in the presence of magnetic charge, where the Jacobi identity for momentum components fails and the corresponding Moyal product becomes nonassociative. While such structures are well understood at the level of single-particle kinematics, their implications for open-system quantum dynamics remain largely unexplored. Here we derive a Born-Markov master equation for a system coupled to a bath when the underlying operator product is weakly nonassociative. The deformation enters through associators appearing in the second-order kernel, while pairwise operator products and dissipators retain their standard form. The resulting correction is dispersive and modifies the Liouville-von Neumann part of the generator without introducing additional dissipative channels. We then embed this structure into a two-qubit transverse-field Ising model using a Stratonovich-Weyl representation and an Ising-aligned twisted Poisson structure. In the zero-temperature limit, the nonassociative terms produce a nonlinear correction in which the instantaneous population imbalance of each qubit feeds back into the dynamics as a state-dependent longitudinal field. Numerical simulations in the weak-coupling regime, where the Born-Markov derivation is quantitatively controlled, show that increasing the nonassociativity parameter suppresses steady-state entanglement by up to 59%, reduces purity, and increases entropy, while leaving the relaxation timescale set by the dissipative rates unchanged. These results demonstrate that weak nonassociativity manifests as a coherent, population-dependent deformation of open-system dynamics rather than an additional dissipative mechanism.

We study entropy-based certification for non-commutative Markov semigroups with rank-deficient stationary states. We introduce a boundary-activation functional that quantifies deviation from the support of the stationary state and prove a logarithmic lower bound on relative entropy in terms of this quantity, based on a BKM coercivity argument and an exact block-decomposition identity. Combining this estimate with dynamical bounds for Davies semigroups under a modified logarithmic Sobolev inequality, we show that, under explicit local conditions, entropy-based certification of convergence can exhibit a slowdown near the boundary regime, with a rate of order e^{-alpha t}/sqrt(alpha t). The mechanism is identified as a coherence-population effect and disappears for incoherent initial states, where classical Pinsker-type bounds are recovered.

Matrix product states (MPS) are a central language for one-dimensional quantum matter and a practical target for near-term quantum simulators and variational algorithms. Yet, while substantial effort has focused on preparing MPS with shallow circuits, scalable methods to \emph{verify} that a many-body device has actually produced the intended state remain underdeveloped. Direct fidelity estimation (DFE) relies only on local Pauli measurements, but in many-body settings it suffers an exponential classical overhead from the preprocessing needed to sample Pauli strings. We eliminate this obstacle by introducing an \emph{autoregressive} importance sampler that draws Pauli strings sequentially from efficiently computable conditional distributions, reducing the per-shot classical overhead to linear scaling in the number of qubits. We further develop a grouped extension that constructs qubit-wise commuting measurement settings via a \emph{sorting string} and simultaneously estimates the entire commuting group from a single setting, significantly reducing estimator variance while preserving efficient postprocessing. Our approach extends naturally to matrix product operators (MPO), enabling scalable verification of tensor-network states and observables in long one-dimensional quantum systems. We utilize random MPS as a natural benchmark for generic 1D entangled states.

We investigate momentum reconstruction for particle processes observed by Unruh-deWitt detector setups. In particular, we derive the probability distributions for particle momenta conditioned on detector clicks in three spatial dimensions. We investigate the statistical properties of such detector setups and discuss their use as models of measurement devices in particle physics.

Accurate sky localization is essential for gravitational-wave (GW) astronomy, particularly for multimessenger follow-up and host galaxy identification. For strongly lensed GW events, achieving localization at the level of $\sim 10~\mathrm{deg}^2$ is critical for associating signals with their lensing structures and enabling targeted searches for additional faint images. We investigate how sky localization improves when combining multiple lensed images of the same source. Using simulated lensed compact binary coalescences and \textsc{BAYESTAR} sky localization, we evaluate localization performance as a function of image multiplicity. We find that combining multiple images leads to a systematic improvement in localization, with the largest gain occurring when combining two images, typically reducing the 90\% credible region area by an order of magnitude. Additional images provide further improvements, with four-image systems achieving localization areas of $\sim 10$--$100~\mathrm{deg}^2$. We also show that subthreshold images contribute modest but non-degrading improvements, enabling their safe inclusion in localization analyses. These results demonstrate that strongly lensed GW events provide a natural pathway to improved localization and motivate hierarchical search strategies for detecting faint lensed images.

We investigate black holes with non-minimal couplings between the electromagnetic field and spacetime curvature, focusing on their event horizons, shadows, and photon rings. Such couplings can naturally arise from both classical effective field theories of gravity and quantum effects in curved spacetime. Starting from a general action with three independent coupling terms, we derive static and spherically symmetric black hole solutions using a series expansion method. We find that all couplings enlarge the event horizon and photon sphere, while their observational consequences differ. The coupling $F^μ_{\ ν}F_{μρ}R^{νρ}$ slightly increases the shadow size and the separation between the zeroth- and first-order photon rings, leaving higher-order spacings nearly unchanged. The coupling $F_{μν}F_{σρ}R^{μνσρ}$ significantly enlarges the shadow and the zeroth-first ring separation, but rapidly suppresses the spacing between higher-order rings. In contrast, the $F^2R$ coupling reduces the shadow size and causes the zeroth- and first-order rings to nearly coincide, leading to an enhanced brightness, while increasing the separation of higher-order rings and leaving them easier to resolve observationally. We further generate black hole images via backward ray tracing and confirm these features within the observationally resolvable regime. These findings can make observational constraints on the non-minimal couplings or might provide new evidence for the modifications to gravity caused by classical or quantum effects.

Classical energy conditions are investigated in generic static and spherically symmetric spacetimes. In setups with nonconstant $g_{tt} g_{rr}$, the appearance of horizons can signal the violation of the null energy condition and the breakdown of some standard near-horizon properties. For configurations satisfying $g_{tt}g_{rr}=-1$, we devise a systematic algorithm to generate solutions of the Einstein field equations that automatically obey the null energy condition. Within this family, we select a particularly significant metric that incorporates a logarithmic correction to the Schwarzschild model and fulfills all standard energy criteria. We examine its main features, including the horizon structure, geodesic behavior, and junction conditions. Our analysis shows that this geometry can be interpreted as an effective exterior description for both horizon-bearing and horizonless compact objects, and suggests that it can potentially act, in certain regimes, as a black hole mimicker.

Higher derivative quantum corrections are essential components of scalar tensor effective field theories (EFTs), yet they typically reintroduce the Ostrogradsky ghost instability that the classical theory was designed to evade. This paper resolves this fundamental tension by establishing a rigorous equivalence between two distinct criteria for theoretical consistency. We analyze a general DHOST theory augmented by Gauss Bonnet and Weyl squared operators with coefficients that are arbitrary functions of the scalar field and its kinetic term. We then pursue two independent paths: first, we derive a set of differential equations for these coefficients by demanding that the full action remains invariant under the protective gauge symmetry of the classical theory. Second, we perform a first principles Hamiltonian analysis using the ADM formalism, deriving a separate set of conditions by imposing the primary and secondary constraints required to eliminate the ghost. Our central result is the proof that these two sets of conditions, one algebraic and one dynamical, are mathematically identical. This equivalence demonstrates that the gauge symmetry is the fundamental origin of Hamiltonian stability in the quantum corrected theory and establishes the symmetry principle as a powerful and practical tool for constructing consistent, ghost free gravitational EFTs without resorting to a full Hamiltonian analysis

We propose a coherent-control scheme for engineering quantum correlations in a cavity optomechanical (COM) system consisting of a driven optical cavity with an embedded nonlinear medium and a membrane, assisted by a coherent feedback loop. The nonlinear medium and the membrane are pumped to implement optical and mechanical parametric amplifications with controllable modulation frequencies and pump amplitudes. Through the combined modulation of the two parametric amplifications and the coherent feedback loop, we engineer the effective cavity decay rate and the distribution of quantum fluctuations, thereby strengthening quantum correlations and improving their robustness against thermal noise. Our scheme provides an efficient route to realizing highly tunable, strong, thermally robust quantum correlations in COM systems, which is promising for the protection of fragile quantum resources.

Variational quantum circuits (VQCs) are typically evaluated at the logical design level when analyzing trainability. However, execution on real quantum devices requires hardware-aware compilation (transpilation) to satisfy qubit connectivity and native gate-set constraints. In this paper, we examine how transpilation can alter the gradient statistics. Using parameter-shift differentiation and gradient variance estimation, we compare logical and transpiled circuits across three representative ansatz families: EfficientSU2 (dense entanglement), TTN (tree tensor network), and RealAmplitudes (linear entanglement). We observe architecture-dependent trainability shifts where densely entangling circuits exhibit pronounced gradient reshaping in shallow regimes, structured tensor-network circuits remain comparatively robust, and linear architectures show mixed behavior. Deep circuits across all families display minimal sensitivity to hardware-aware compilation. These findings demonstrate that transpilation acts as an implicit structural transformation of the optimization landscape, motivating compilation-aware analysis and co-design for VQCs.

We develop integration-by-parts (IBP) reduction and differential equations for massive loop integrals of cosmological correlators in de Sitter (dS) spacetime, demonstrating the feasibility of this approach. We identify a structural property of the dS IBP system: for an $n$-propagator family, it splits into $2^n$ closed subsystems classified by the parity of the propagator indices. We further formulate a Baikov representation for loop integrals in dS space and derive the corresponding dimensional recurrence relations. In flat spacetime, intersection theory shows that $\mathrm{d}\log$-form master integrands lead to $\mathrm{d}\log$-form differential equations. Motivated by fibration intersection theory, we conjecture that this construction extends to dS integrands involving Hankel functions. We verify this conjecture in the one-loop bubble family and determine the associated alphabet.

We report enhanced evidence for the $X(7200)$ state and significantly improved measurements of the $X(6900)$ resonance parameters through a combined analysis of the di-$J/ψ$ mass spectrum using published data from LHCb, ATLAS, and CMS. By performing simultaneous fits to all three experiments, we observe the $X(6900)$ with overwhelming significance ($>12σ$) and determine its mass and width with improved precision. For the $X(7200)$, we find consistent signals across multiple interference models, with significances ranging from $3.7σ$ to $6.6σ$; the best-fit model (adopting the CMS three-resonance scheme) yields $6.6σ$ significance, providing substantially strengthened evidence for this state. Our results underscore the essential role of interference effects in fully-charmed tetraquark spectroscopy and offer new constraints on their production mechanisms at the LHC.

China's first proton test beam facility, named the High-energy Proton-beam Experiment Station (HPES), is currently under construction in campus of CSNS, as part of the CSNS-II project. Utilizing protons slowly extracted from the Rapid Cycling Synchrotron of CSNS, HPES will deliver 1.6 GeV proton beam with an adjustable flux ranging from 1E3 to 1E8 protons per second. The station is composed of two dedicated test terminals designed to support comprehensive beam tests, serving as an advanced platform for particle detector development, irradiation hardness studies of aerospace chips, and GeV-proton-induced nuclear data measurements.To characterize the beam, HPES incorporates dedicated flux and profile monitors. For user experiments, the facility is equipped with a high-precision proton telescope offering a positioning resolution of 10 $μ$m, and a Time-of-Flight (TOF) spectrometer achieving an energy resolution of 1%. Furthermore, a compatible trigger logic unit have been designed to provide precise event tagging, which is essential for data alignment. This paper presents an overview of the detector systems within HPES, discusses their design considerations, and outlines the future prospects of the facility.

We review the salient features of next-to-leading-order QCD and electroweak corrections to the scattering of two and the production of three weak gauge bosons at the Large Hadron Collider. Results for the tower of $O(α_s^mα^n)$ corrections are shown for the exemplary processes of like-sign WW scattering and triple-W production, emphasizing the large impact of purely electroweak corrections which generically grow to $\sim-16\%$ and $\sim-7\%$ for these process types, respectively, even for integrated cross sections. Moreover, we discuss the possibility to reproduce the results of full off-shell calculations by the "vector-boson scattering approximation", "leading-pole approximations", and the "effective vector-boson approximation".

We present perturbative quantum chromodynamics (pQCD) predictions for high-momentum particle yield modification in very light ion collisions - ${}^{10}\mathrm{B}+{}^{10}\mathrm{B}$, ${}^{6}\mathrm{Li}+{}^{6}\mathrm{Li}$, ${}^{4}\mathrm{He}+{}^{4}\mathrm{He}$, and ${}^{3}\mathrm{He}+{}^{3}\mathrm{He}$ - with and without medium-induced energy loss. We find non-trivial suppression in symmetric systems from ${}^{208}\mathrm{Pb}+{}^{208}\mathrm{Pb}$ to ${}^{3}\mathrm{He}+{}^{3}\mathrm{He}$ and in asymmetric $A+B$ systems, with the suppression scaling approximately as $R_{AB} \simeq (\sqrt{AB})^{1/3}$. Further, we find that ${}^{3}\mathrm{He}$ and ${}^{6}\mathrm{Li}$ offer particularly clean environments for observing final-state partonic energy loss from quark-gluon plasma (QGP) formation in extremely small systems. Finally, we show that energy loss models generically predict $v_2\{\mathrm{SP}\} \approx 0$ in small systems, indicating that the large measured $v_2 > 0$ in $p+{}^{208}\mathrm{Pb}$ is not due to energy loss.

In a mixed dark matter scenario in which primordial black holes (PBHs) would co-exist with thermally produced self-annihilating particles, one expects the former to be surrounded by extremely dense halos made of the latter, built up during radiation domination. Here, as a continuation of previous work, we derive observational limits on such a scenario from a full statistical analysis of cosmic microwave background (CMB) data. We quantify how a tiny fraction $\fbh$ of PBHs could restrict the parameter space available to thermal particle dark matter, limiting the $s$-wave annihilation cross section to values $\lesssim 10^{-30}\,{\rm cm^3/s}\,(\mchi/100\,{\rm GeV})\,(\fbh/10^{-6})^{-3}$ if PBHs are typically heavier than $\sim 10^{-10}\,\Msun$, which can also be turned into constraints on PBHs in this mass range. In contrast, asteroid mass or lighter PBHs could live in perfect peace with these particles. Finally, we shortly discuss the implications of the recent tentative interpretation of Subaru-HSC microlensing events as PBHs.

We consider two-dimensional $\mathcal{N}=(2,2)$ supersymmetric field theories living on a spindle $\mathbb{WCP}_{[n_1,n_2]}^1$. Starting from the spindle solutions of five-dimensional STU gauged supergravity, we construct theories on a spindle which preserve supersymmetry via either the twist or anti-twist mechanism and admit two Killing spinors of opposite R-charge. While the study of field theories on anti-twisted spindles has already been undertaken in some detail, the advantage of our approach allows for the derivation of analogous results in the twist case. We apply the technique of supersymmetric localisation to compute the exact partition function for a theory consisting of an abelian vector multiplet and a charged chiral multiplet in the presence of a Fayet-Iliopoulos term. We compare and contrast the results for the twisted and anti-twisted spindle and find a general formula which encompasses the partition function for both cases simultaneously.

The LHCb detector has demonstrated a proven competitiveness across a wide range of physics analyses thanks to its forward coverage. These proceedings describe: i) complementary measurements using heavy flavour jets, ii) Electroweak (EW) measurements with the top and W boson, and iii) searches for New Physics states such as axion-like particles (ALPs), heavy-neutral lepton (HNLs) and B-meson decays to multi-muon final states.

Motivated by the first observation of CP violation in $b$-baryon decays, the search for baryonic decays exhibiting large CP violation will be a primary focus in the coming years. We propose that significant CP-violating effects exist in the decay $Λ_b \to ΛD$, where $D$ denotes a CP eigenstate of the $D^0 - \bar{D}^0$ system. The predicted CP asymmetries for both the CP-even and CP-odd modes can reach magnitudes as large as $50\%$, making these decays promising targets for measurement at the LHCb experiment. Additionally, we predict for the first time several nonzero CP-violating observables associated with angular distribution parameters, providing valuable complementary information in the search for CP violation in baryon decays. Furthermore, we propose a novel strategy to extract the CKM angle $γ$ by combining data on angular distribution parameters and decay rates from the relevant channels. We emphasize that $Λ_b \rightarrow ΛD$ decays are among the most promising candidates for determining $γ$ in the baryon sector. Our findings may offer new insights for future theoretical and experimental investigations.

Astrophysical dark matter particles with masses well below GeV-scale can be difficult to detect using conventional nuclear recoil experiments due to their low velocities in our Milky Way halo. Elastic scattering with high-energy cosmic rays or thermal production inside core-collapse supernovae can accelerate sub-GeV DM to (semi-)relativistic velocities, producing nuclear recoil energies above the keV threshold that paleo-detectors can record over geological timescales. Using olivine as the target with 100$\,$g$\cdot$Gyr exposure, we compute track length distributions from such (semi-)relativistic dark matter fluxes, incorporating all major backgrounds (neutrinos, uranium-chain neutrons, thorium recoils) with a statistical analysis on an Asimov dataset. We derive 95 C.L. projected sensitivity of paleo-detectors to the DM-nucleon cross section for dark matter masses between a few MeV and hundreds of MeV. Our results show that paleo-detectors are able to probe large parameter regions that are not covered by current and near-future experiments designed to detect dark matter and neutrinos. In particular, paleo-detectors offer a unique ability to record the dark matter flux from Galactic supernova events over geological times. Such cumulative exposure enables sensitivity gains of a few orders of magnitude compared to conventional experiments.

We numerically investigate whether the center-symmetric confined phase of large-$N$ $SU(N)$ gauge theory with one adjoint Dirac fermion persists under spatial compactification on $\mathbb{R}^3 \times S^1$. To this end, we employ a partially reduced twisted Eguchi-Kawai (TEK) model on a $1^3 \times L_4$ lattice with an adjoint Wilson fermion, and measure both the Polyakov loop around $S^1$ and order parameters for volume independence in the reduced directions. For $N=36$, $L_4=2$, $b=0.30\text{-}0.46$, and $κ=0.03\text{-}0.16$, we find that, with periodic boundary conditions, the Polyakov loop remains near zero in the light-fermion regime as the circle size is reduced. For the modified twist, the volume-independence order parameters are also consistent with zero in the explored region, supporting the validity of the partially reduced description. These results provide numerical evidence, within the reduced-model setup and parameter range studied, for an adiabatic-continuity scenario in which the confined phase is smoothly connected between large and small circles. By contrast, with antiperiodic boundary conditions, the Polyakov loop exhibits a clear deconfinement transition. We also discuss how this scenario is compatible with the anomaly constraints of the underlying four-dimensional theory. The symmetric twist is examined as a useful comparison, although its volume-independence properties appear less robust at the present value of $N$.

We study how rotation modifies the constraints on MeV-scale axion-like particles (ALPs) coupled to photons derived from SN 1987A. We constrain the ALP parameter space based on both the energy-loss argument and the gamma-ray limits, and examine how these constraints are affected by stellar rotation. Adopting initial angular velocities of $Ω_{0} = 0.0 and 1.0 rad s^{-1}$ in the iron core, we carry out two-dimensional core-collapse supernova simulations for three progenitor models - a $14 + 9M_{\odot}$ binary and $13M_{\odot}$ and $18M_{\odot}$ single stars with solar metallicity - and estimate ALP emission rates through post-processing. We find that rotation suppresses ALP emission by reducing the core temperature via centrifugal support. Rotation also reduces the neutrino luminosity, but the suppression of ALP emission is more effective, leading to relaxed constraints within a simplified criterion based on the energy-loss argument. This relaxation is particularly pronounced in the rotating $18M_{\odot}$ model, where a substantial decrease in the central temperature occurs at $t_{pb} = 0.8 - 1 s$. In this simplified criterion, such rapid temporal variations in temperature indicate that the resulting constraints depend sensitively on both the evaluation time and the underlying supernova model. For a gamma-ray limit from the SN 1987A observation, rotation has a negligible impact on the constraint. This is because the ALP-induced gamma-ray fluence observed at Earth is proportional to the fourth power of the ALP-photon coupling constant, making the constraint relatively insensitive to the rotational suppression of ALP emission.

We investigate the internal structure of near-threshold $s$-wave eigenstates in a two-body system with Coulomb plus short-range interactions. Using a nonrelativistic effective field theory, we derive the expression for the compositeness in terms of the energy derivative of the self-energy, which is applicable to the present system with the non-separable Coulomb interaction. For near-threshold states, the compositeness can be written solely in terms of the Coulomb scattering length, the Coulomb effective range, and the Bohr radius, providing the weak-binding relation in the presence of the Coulomb interaction. We numerically study the pole trajectories and the compositeness and find that the Coulomb interaction qualitatively modifies the threshold behavior of the poles and the internal structure of the eigenstates. We show that when the Coulomb interaction is relatively strong, the enhancement of the compositeness near the threshold is absent, in contrast to purely short-range interactions. On the other hand, for a weak Coulomb interaction, a remnant of short-range universality survives, and near-threshold bound states tend to be composite dominant. Furthermore, even resonances are dominated by the composite component in the presence of the Coulomb interaction, owing to their continuous connection to the bound-state regime. We apply the formalism to realistic systems with near-threshold eigenstates, including exotic hadrons and nuclei.

We investigate the decay $φ\toπ^+π^-π^0$ in an effective-Lagrangian framework that keeps the dominant $ρπ$ mechanism and the direct three-pion term explicitly separated at the amplitude level. The resonant contribution is described with the Gounaris-Sakurai propagator, the neutral channel includes $ρ^0$-$ω$ mixing, and the leading elastic $P$-wave $ππ$ final-state interaction is incorporated through a constant on-shell Omnès factor $C_Ω\equiv|Ω_1(m_ρ^2)|$. The purpose of this approximation is not to provide a full dispersive reconstruction of $φ\to3π$, but to isolate the leading rescattering effect in a transparent phenomenological setting. With this setup, we obtain $Γ_{\rm th}=0.6950$ MeV, about $5\%$ above the empirical estimate $Γ_{\rm exp}\approx0.660\pm0.020$ MeV, while the direct integrated weight is reproduced at a realistic level, $I_{\mathrm{dir}}=8.457\times10^{-3}$. The computed on-shell Omnès factor, $C_Ω=4.794$, is quantitatively sizable, indicating that $ππ$ rescattering provides a nontrivial enhancement in the $ρ$-dominated channel. At the same time, the $x$ and especially the $y$ Dalitz projections still exhibit visible discrepancies near the phase-space boundary, showing that the present treatment should be viewed as an intermediate phenomenological step rather than a precision amplitude analysis. These residual tensions motivate the next stage: a fully $s$-dependent Omnès implementation and a direct fit to the efficiency-corrected KLOE Dalitz-bin data.

We give a systematic account of the soft mode dynamics of QCD critical point(QCD-CP) and the two-flavor color-superconductivity(2SC-CP) based on the 2-flavor Nambu-Jona-Lasinio model, and investigate their effects on electromagnetic observables in relativistic heavy-ion collisions (HIC). We first demonstrate that the collective excitations coupled to the fluctuations of the respective order parameters are the soft modes associated to the respective phase transitions, in the sense that they acquire a prominent spectral strength in the low-energy and low-momentum region above the respective critical temperatures, and the peak energy of the respective spectral functions goes down, i.e., gets softened, and eventually vanishes at the the critical point. It is shown that the diquark soft mode of the 2SC gives rise to the pseudogap, i.e., a depression in the density of states of the quark spectra around the Fermi surface above but in the vicinity of the critical temperature. Then, exploiting the ideas that were developed in condensed matter physics for describing the `para-conductivity' in the normal phase of metal superconductors, we show that the soft modes cause an anomalous enhancement of electric conductivity and the dilepton production rate, and discuss their relevance to HIC.

We extend the Hodge atoms framework of Katzarkov--Kontsevich--Pantev--Yu to one-parameter conifold degenerations of Calabi--Yau threefolds. For a degeneration $π\colon X \to Δ$ whose central fiber $X_0$ has $r$ ordinary double points, we construct a canonical rigid-flexible decomposition of the Hodge atoms of the nearby smooth fiber attached to the corrected degeneration object. The rigid atom $A(\IC^H_{X_0})$ is preserved across the degeneration, while the flexible atoms $A(i_{k*}\QQ^H_{\{p_k\}}(-1))$ are rank-one contributions, one for each vanishing cycle. The total degeneration atom $A(P^H)$ is the atom of the corrected mixed Hodge module $P^H\in\MHM(X_0)$ and fits into an exact sequence of atoms whose non-split structure is controlled by the intersection matrix $(\langleδ_i,δ_j\rangle)$. The technical core is the Stokes--Extension Identification, which identifies the Stokes matrix of the Dubrovin connection at the conifold locus with the matrix of the variation morphism $\varF\colon\varphi_π(F) \to ψ_π(F)$ under mixed Hodge module realization.

We investigate the holographic Schwinger effect in a confining background with a step dilaton profile, which induces a sharp transition between ultraviolet and infrared regimes and provides a qualitatively distinct realization of confinement. Within this framework, the quark--antiquark potential is extracted from the classical configuration of a fundamental string, allowing for a direct analysis of vacuum instability and pair production. In the absence of a magnetic field, the step dilaton leads to a significantly sharper suppression of the potential barrier as the electric field increases, implying an enhanced sensitivity of the critical electric field compared to smooth soft-wall models and demonstrating that the abrupt geometric transition qualitatively enhances the onset of vacuum decay. Incorporating an external magnetic field through the Dirac--Born--Infeld action, we find a nontrivial and amplified deformation of the potential barrier, resulting in a pronounced shift of the critical electric field that depends on both the magnitude and orientation of the magnetic field. Overall, the step dilaton background exhibits a substantially stronger response of the Schwinger effect to external electromagnetic fields than conventional soft-wall models, providing a novel mechanism for controlling pair production and highlighting the crucial role of dilaton structure in non-perturbative dynamics of holographic QCD.

A dark photon is an Abelian gauge boson from a new $U(1)_D$ gauge symmetry, coupled to the Standard Model via kinetic mixing, with $\varepsilon$ inducing an effective coupling to the electromagnetic current and $g_χ$ to a stable dark matter particle $χ$. We study $J/ψ$ two-body and four-body decays via a light-mass dark photon ($m_U < 3.0$ GeV) in the framework of non-relativistic QCD (NRQCD), considering both visible and invisible decays of the dark photon into SM fermions or dark sector particles. Numerical results for the decay ratios $Γ/Γ_{J/ψ}$ and expected event numbers at BESIII are presented, along with the significance $S/\sqrt{B}$ and $p_T$ distributions where applicable. Our results show that, for $m_U < 2m_χ$, the two-body final state decay channels of the $J/ψ$ mediated by a dark photon have event yields $0\sim 37$ with significances of $10^{-5}\sim10^{-6}$, while the four-body final state channel yields about $94\sim172$ events in the very low mass region $m_U < 0.2\ \text{GeV}$. For $m_U \ge 2m_χ$, the invisible two-body final state decay channel yields $0\sim12$ events with better significance $10^{-1}\sim10^{-3}$, and the invisible four-body final state decay channel yields $0\sim129$ events, whereas visible decays are all severely suppressed.

Precision Higgs physics offers a sensitive window into physics beyond the Standard Model. In parallel, neutrino-oscillation experiments have established the existence of nonzero neutrino masses, thus implying the presence of new physics. Motivated by these facts, we investigate the one-loop contributions of light and heavy Majorana neutrinos to the $ZZh$ vertex within a variant of the type-I seesaw mechanism in which light-neutrino masses vanish at tree level and are then generated radiatively. We analyze the $CP$-conserving and $CP$-violating anomalous couplings which characterize the $ZZh$ vertex and study their phenomenological implications in two relevant kinematic scenarios at future lepton colliders: Higgsstrahlung production and Higgs production via neutral-current vector-boson fusion. We find that $CP$-conserving effects can reach magnitudes of order $10^{-3}$, potentially within projected future experimental sensitivities, whereas $CP$-violating contributions are strongly suppressed, lying well beyond such projections.

The multiplicity-dependent suppression of $Υ(n\mathrm{S})$ states measured by CMS in $pp$ at $\sqrt s=7\,$TeV \cite{CMS2020}, and of $ψ(2S)\big/J/ψ$ measured by LHCb at $\sqrt s=13\,$TeV \cite{LHCb2024}, is subjected to four multi-differential tests: \emph{cone isolation}, \emph{azimuthal sectors}, \emph{transverse sphericity}, and \emph{prompt vs. non-prompt}. Cone and sphericity close a \emph{scissors constraint}: the local reading of the Comover Interaction Model is in tension with the cone data, its global reading with the sphericity data. The non-prompt flatness forces the mechanism to act at early proper times. None of the considered hadronic or string-based frameworks -- CIM local or global, PYTHIA 8 MPI \cite{Sjostrand2015}, rope hadronisation \cite{Bierlich2015}, CGC \cite{Ma2015}, Trainor TCM \cite{Trainor2008} -- naturally satisfies the four constraints simultaneously in its published form. The surviving class is an early, globally correlated medium consistent with partonic degrees of freedom, co-occurring with the ALICE strangeness enhancement \cite{ALICE_SE}, the long-range ridge \cite{CMS_ridge}, and below the threshold of the partonic baryon-meson $v_2$ \cite{ALICE_v2}, in a density window compatible with the Campanini \& Ferri equation of state \cite{Campanini2011}.

The Belle and Belle II experiments have collected a 1.3 ab$^{-1}$ sample of $e^+e^-\to B\bar B$ collisions at $Υ(4S)$ centre-of-mass energy. This is ideal environment to search for rare electroweak penguin $B$ decays and notably those involving $B$ decays to final states with missing energy. Results on these datasets of $b\to s \ell^+\ell^-$ $(\ell=e,μ)$, $b\to sτ^+τ^-$, and $b\to sν\bar ν$ transitions are presented.

A search is performed for the single production of a heavy vector-like quark (VLQ), decaying into a W boson and a b quark. The analysis uses proton-proton collision data collected by the CMS experiment at the CERN LHC at a center-of-mass energy of 13 TeV and corresponding to an integrated luminosity of 138 fb$^{-1}$. The search targets events with leptonic W boson decays. The event signature consists of one electron or muon, large transverse momentum imbalance, at least one jet consistent with coming from the fragmentation of a b quark and having large transverse momentum, and at least one jet in the forward region of the detector. No significant excess over the standard model prediction is observed. Upper limits are set at the 95% confidence level on the production cross section of a VLQ and its coupling $κ_\mathrm{W}$ to the standard model sector. For a VLQ decaying exclusively into Wb, the upper limit on $κ_\mathrm{W}$ depends on the VLQ mass and reaches values as low as 0.086 for masses around 1.4 TeV. For $κ_\mathrm{W}$ = 0.2 the lower limit on the VLQ mass is 2.4 TeV. These are the most stringent limits to date on the single production of VLQs decaying into Wb.

Two-particle two-hole (2p2h) excitations driven by meson-exchange currents (MEC) are among the leading nuclear uncertainties in long-baseline neutrino oscillation experiments. Three models currently implemented in neutrino event generators disagree by 20--40% on the $ω$-integrated 2p2h cross section in the dip region on carbon (differential disagreements can reach factors of 2--3), and the axial two-body current has no direct experimental constraint beyond tritium $β$-decay at $Q^2 = 0$. We propose a measurement program at the Electron-Ion Collider (EIC) using polarized electron scattering on deuteron and $^3$He. Electromagnetic (EM) scattering ($γ^*$ exchange) measures the vector MEC. Charged-current (CC) scattering ($W^-$ exchange) on the same targets measures the vector$+$axial MEC. Subtracting the two provides the first direct sensitivity to the axial two-body current, including the $V$--$A$ interference, as a function of momentum transfer. Using $^3$He (2~$pn$ $+$ 1~$pp$ pair) extends the decomposition to $pp$ pairs. Polarized beams and targets give access to six EM response functions on deuteron, four of which have not been previously measured. The tensor analyzing power provides a sign-flip test for $Δ$-excitation MEC. We present projected sensitivities at $50 fb^{-1}$ on deuteron ($\sim$5 years at $10^{33}$~cm$^{-2}$s$^{-1}$). The EM program can deliver $\sim\!5\!\times\!10^4$ events per $Q^2$ bin constraining the MEC transverse response to $\sim$2% per bin, the beam--target double-spin asymmetry reaches $6$--$13σ$ per bin, and the vector MEC $V_{pn}$ is measured to $\sim$6% per bin. The CC channel is statistics-limited, with $\sim$6--38 events per $Q^2$ bin at $50 fb^{-1}$, requiring a luminosity upgrade beyond the current EIC baseline.

Recently arxiv:2604.12897 urged that the terminology "neutrinoless double beta decay" should be changed to "Majorana double beta decay" to properly give credit to Majorana, and to focus on the positive aspects of the phenomenon -- supposed creation of matter in the laboratory -- rather than the negative: absence of something, embarrassment over false claims of detection, and a "sociology of suspicion." I argue that the current terminology is more accurate and descriptive, and that the claimed reasons for its adoption are lacking in credibility.

This contribution discusses the physics potential of a future muon collider operating at a center-of-mass energy of $\sqrt{s} = 10$ TeV for precision studies in the Higgs sector. Using a detailed detector simulation that incorporates the dominant sources of machine-induced background, the expected sensitivity to key Higgs processes is evaluated. These include the measurement of production cross sections for $H\to b\bar{b}$, $H\to WW^*$, and double-Higgs production $H\!H\to b\bar{b}b\bar{b}$. A central focus of the study is the determination of the Higgs boson trilinear self-coupling, a critical parameter for understanding the structure of the Higgs potential and electroweak symmetry breaking. The analysis is based on the MUSIC multi-purpose detector concept, specifically optimized for the muon collider environment, and assumes an integrated luminosity of $10$ ab$^{-1}$ collected over five years. The results presented highlight the exceptional prospects of a multi-TeV muon collider for exploring the Higgs potential with a level of precision unattainable by any other proposed future collider within a comparable timeframe.

The $B \to πK$ system provides a rich laboratory for testing the Standard Model and studying CP violation. A particularly important channel is $B^0_d\toπ^0 K_{\rm S}$, the only mode exhibiting both direct and mixing-induced CP violation. Recent Belle II measurements of the CP asymmetries in this decay provide valuable new input. An updated analysis incorporating these new data provides new insight on the long-standing $B \to πK$ puzzle. Looking ahead to the high-precision era of flavour physics, the $B \to πK$ system can be further exploited to potentially reveal new sources of CP violation.

The coadjoint orbit action for a multifermion system, as an exact description of its dynamics, is considered. A parametrization of the variables involved is given which facilitates the approximation of this by another coadjoint orbit action suitable for expansions around the Fermi surface, recovering various actions which have been used in previous literature. The presentation of this in terms of functions on phase space with star-products as well as further truncations are briefly considered.

We introduce a refined version of the 3D index for 3-manifolds, building on the construction of the 3D $\mathcal{N}=2$ gauge theory $T[M]$ by Dimofte-Gaiotto-Gukov and Gang-Yonekura. The refined index is a superconformal index of $T[M]$ equipped with additional gradings that capture enhanced flavor symmetries of the effective theory. Our construction is based on a Dehn surgery presentation of $M$ in terms of an ideally triangulated link complement $N$. We derive an explicit infinite-sum formula for the refined index and provide nontrivial checks in representative examples, supporting its invariance under changes of triangulation, Dehn surgery presentation, and other auxiliary data. As a strictly stronger invariant, the refined index enables finer distinctions among 3-manifolds and among distinct IR phases of the associated gauge theories. We also introduce a computational tool, \textsc{Refined Index Calculator}, for its explicit evaluation.

In a recently proposed approach to testing models of inflation by Cosmic Microwave Background (CMB) radiation the reheating temperature is directly expressed in terms of the CMB observables. Its model independent bounds translate in a given model into narrow ranges of those observables. In that approach we analyse the polynomial class of the $α$-attractor inflaton potential models (P-models), in a broad range of polynomials and with the inflaton decays and fragmentation in the reheating period taken into account. The predictions for the CMB observables, the scalar spectral index $n_s$ and tensor-to-scalar ratio $r$, are compared with the Planck and Planck combined with ACT data. Both can be accommodated by that class of the $α$ attractor models. The sensitivity of the results of that comparison to the reheating temperature and to the upper bound on the ratio $r$ is clearly demonstrated.

The 1D Mueller dipole model, its high energy limit, and its generalization were investigated. To address the ambiguity stemming from different definitions of the pseudorapidity ranges in experimental measurements, we propose the entropy as the function of the logarithm of the average multiplicity, $S(\ln\langle n\rangle$, as a universal observable. From the solutions of the models, we calculate both the entropy and the average charged particle multiplicity and compare to data measured in proton-proton collisions. We obtained these quantities directly from the measured charged particle multiplicity distributions and determine the model parameters via fits. We find that the generalized dipole model provides a significantly better description of the data than the 1D Mueller model.

We analyze the effect of using the Fermi-Dirac statistics, rather than its Boltzmann approximation, in numerical simulations of perfect spin hydrodynamics of particles with spin 1/2. The system considered is boost invariant, transversely homogeneous, with corrections to the baryon current and the energy-momentum tensor that are second order in the spin polarization tensor $ω$, and the spin tensor considered is first order in $ω$. The study shows the feasibility of this approach, as the special functions defined by integrals that appear in the coefficients in the Fermi-Dirac case can be conveniently parametrized. For sets of initial conditions used in previous works, the differences in parameter evolution between the two underlying particle statistics are about one order of magnitude smaller than corrections coming from spin feedback. We also discuss when and why the numerical solutions of the equations of perfect spin hydrodynamics break down for very large values of spin polarization in one of the geometric configurations considered.

We study the hydrodynamics of the Filtered Dark Matter (Filtered DM) scenario during a first-order phase transition (FOPT). In this scenario, the bubble wall is highly reflective of the dark matter (DM) fluid but transparent to radiation, making the hydrodynamic problem fundamentally different from that of the electroweak FOPT. Motivated by this property, we formulate the hydrodynamics of this system as a two-component fluid composed of DM and radiation, and find that the solutions can be classified into detonation-like and deflagration-like branches in the ballistic regime and in the local thermal equilibrium (LTE) regime. In the ballistic regime, the energy--momentum of DM that cannot enter the wall appears as a reflected mode, while in the LTE regime, it relaxes into the energy--momentum of radiation. We find that this difference in the fate of the DM fluid that cannot enter the interior of the wall leads to different hydrodynamic behaviors in the DM and radiation fluids independently and, in particular, results in different existence conditions for solutions in the deflagration-like branch. Based on these results, we further revisit the impact of hydrodynamic effects on the relic abundance of Filtered DM and demonstrate the change in the abundance induced by hydrodynamic effects. In addition, we also discuss the non-conservation of the entropy current from the viewpoint of the two-fluid system, and briefly comment on the similarity between the Filtered DM system and information-thermodynamic systems.

CP violation offers powerful probes to explore the quark-flavour sector, where decays of B mesons have been key players since decades. I discuss a variety of probes ranging from non-leptonic to rare B decays, offering exciting opportunities at the FCC in the era after the HL-LHC and Belle II.

We systematically investigate the scaling properties of the transverse momentum spectra for pions, kaons, and protons in Au+Au collisions at $\sqrt{s_{NN}}$ = 7.7, 11.5, 14.5, 19.6, 27, 39, 62.4, and 200 GeV, as well as in U+U collisions at $\sqrt{s_{NN}}$ = 193 GeV, across different centrality classes, using experimental data from the collaborations at the Relativistic Heavy Ion Collider (RHIC). Universal scaling emerges when the particle transverse momentum spectra are scaled by global physical quantities, i.e., the average total particle multiplicity and mean transverse momentum, confirming recent scaling findings from the data at the Large Hadron Collider (LHC) by the ExTrEMe collaboration. The scaling behavior breaks down in the high $p_{T}$ region and in peripheral collisions. We provide a natural explanation for these observations by invoking the Cooper-Frye formula, which is used for hadronization in hydrodynamics. Furthermore, we demonstrate the equivalence between the scaling found by the ExTrEMe collaboration and the Hwa-Yang scaling which was proposed two decades ago.

Recently, a mechanism for generating astrophysically relevant magnetic fields via ultralight pseudoscalar dark matter, through the coupling term $g_{φγ} φF_{μν}\tilde{F}^{μν}$ in the Lagrangian density, was proposed in Brandenberger et al (2026) (see Ref. 1). In this scenario, the electromagnetic fields are amplified through the phenomena of parametric resonance due to the oscillatory behaviour of the pseudoscalar field. However, the analysis presented in that work does not account for the effects of a conducting medium. In this paper, we incorporate the finite conductivity of the plasma into the dynamics of the pseudoscalar and electromagnetic fields. We show that, due to the large conductivity relative to the Hubble parameter, the amplification of the electromagnetic fields due to parametric resonance is significantly suppressed. Consequently, we find that, for observationally viable values of the coupling between the electromagnetic field and the ultralight pseudoscalar field, it is not possible to generate magnetic fields of sufficient strength to explain their presence in cosmic voids.

The upcoming PandaX-xT experiment will deploy over 3,000 readout channels operating at a 500 MSa/s sampling rate, generating a sustained data bandwidth up to 1.6 GB/s. To meet this demanding requirement, we present AURORA, a high-performance, distributed data acquisition (DAQ) framework designed for scalability, low latency, and efficient resource utilization. Built on a modular architecture and leveraging modern I/O and networking technologies, including multi-level buffering, deferred and asynchronous processing, AURORA achieves a projected throughput of over 3 GB/s on the aggregation node in benchmark tests. While developed to support PandaX-xT, the framework is experiment-agnostic and readily adaptable to other large-scale particle and nuclear physics experiments.

We investigate bubble dynamics in a holographic superfluid undergoing a first-order phase transition with spontaneous $U(1)$ symmetry breaking. Near the nucleation threshold, the system exhibits universal critical behavior governed by a single unstable mode, leading to logarithmic scaling of the time spent near the critical solution. The terminal bubble wall velocity increases with charge density but remains small due to strong dissipation. In multi-bubble collisions, vortex formation depends sensitively on the initial phases and deviates significantly from the geodesic rule. Notably, we identify a regime where three-bubble collisions produce a vortex-antivortex pair that subsequently annihilates, a phenomenon not predicted by the geodesic rule. The lifetime of this pair scales logarithmically with the distance to the critical collision radius. Our results underscore the crucial role of non-equilibrium dynamics in strongly coupled superfluids and provide new insights into topological defect formation during first-order phase transitions.

We investigate $2P$-$1F$ mixing in charmonium, focusing on the close-in-mass $χ_{c2}(2P)$ and $χ_{c2}(1F)$ states. The conventional tensor force yields negligible mixing, motivating the inclusion of coupled-channel effects. Our unquenched calculation reveals sizable mixing angles of $7.5^\circ$ and $15.4^\circ$. We predict the corresponding two-photon and two-gluon decay widths as key observables for experimental verification. Additionally, we discuss the production of these two $2P$-$1F$ mixed states of charmonium via $γγ$ fusion. Current data are insufficient to determine the mixing, highlighting the need for precise future measurements to resolve this aspect of charmonium spectroscopy.

A significant excess of $t\bar{t}$ events near the production threshold was observed in LHC Run-2 data by the ATLAS Collaboration. It is consistent with the formation of $t\bar{t}$ quasi-bound states, which were first hypothesised almost 40 years ago. This contribution summarises the experimental results and outlines a path toward further characterisation of the excess.

We derive a complementary cosmological upper bound on the fraction of dark matter residing inside massive compact halo objects (MACHOs) using the cosmic dawn and dark ages global 21-cm signal $(T_{21})$. MACHOs of masses $M\gtrsim 10^3~M_\odot$ moving through the post-recombination baryonic fluid transfer kinetic energy to the intergalactic medium via dynamical friction, raising the gas temperature and distorting the 21-cm signal from the $Λ$CDM prediction. We consider both a monochromatic and two extended MACHO mass distributions: log normal and critical collapse. Imposing the conditions that the deviation in the global 21-cm signal $ΔT_{21}$ does not exceed $50~\rm mK$ at $z\sim 17$ or $15~\rm mK$ at $z\sim 89$, and that no emission signal appears at $z \gtrsim 300$, we derive upper bounds on the MACHO fraction $f_M$ across the mass range $10^3 \lesssim M_c/M_\odot \lesssim 10^7$. The dark ages criterion yields constraints that are both tighter and free from astrophysical uncertainties associated with star formation, providing a complementary cosmological window. Extended distributions produce bounds that are generally more stringent than their monochromatic counterpart, with the critical collapse models yielding the strongest constraints at intermediate masses.

We compute the rate of growth of operator size in matrix models by probing the Lin-Maldacena class of geometries with classical probes. We consider massive point particle probes whose proper momentum equals the size of the gauge invariant operator in the matrix model. We work out the example of the BMN Plane Wave Matrix Model using the electrostatic approach and the method of background fluxes. We also work out complexities in the D2 brane as well as NS5 brane limits of the BMN matrix model along with an example of the irrelevant deformation namely the non-Abelian T-dual of $AdS_5 \times S^5$. Finally, we carry out a possible calculation of the Krylov complexity on the matrix model counterpart by using a simple reduction ansatz known as the pulsating fuzzy sphere model. We outline an algorithm to define Krylov basis elements for the matrix model and compute a few Lanczos coefficients. Our analysis reveals that both the Krylov basis states as well as Lanczos coefficients are uniquely fixed in terms of the mass parameter of the matrix model.

Semileptonic $b$-hadron decays proceed via charged-current interactions and provide powerful probes for testing the Standard Model and searching for New Physics effects. The advantages of studying such decays include the large branching fractions and reliable calculations of the hadronic matrix elements. Several features can be studied, such as the ratios of branching fractions, CKM parameters, properties of $b$-hadron production, form factor parameters and New Physics Wilson coefficients. In this contribution, LHCb measurements of branching fraction in $Λ\to p μ^{-} \barν_μ$ and form factor parameters with $B^0 \to D^{*+} μ^{-} ν_μ$ decays are presented.

Axion dark matter searches use a microwave cavity for the resonant conversion of axions into photons to enhance experimental sensitivity, with the cavity generally configured as a two-port system for both signal pickup and cavity characterization measurements. In this study, we investigated cavity-mode couplings in such a two-port system and examined their impact on axion dark matter search experiments, which typically use one strongly coupled port and one weakly coupled port. We found that, in such a two-port cavity system, the measured coupling strength of one port depends on that of the other; hence, the coupling coefficients appearing in the relation for the unloaded quality factor of the cavity mode can vary substantially with the measured coupling strengths. Meanwhile, the scanning rate, the figure-of-merit for axion dark matter searches, cancels the systematic contribution from the strongly coupled port; hence, the remaining systematic uncertainty arises only from the weakly coupled port and may be negligible, depending on its coupling strength. Nevertheless, we recommend measuring the coupling strength of the weakly coupled port to eliminate this systematic uncertainty and thereby recover any experimental sensitivity that may have been lost, for example by approximately 10\% when the coupling strength of the weakly coupled port is 0.05.

Many recent efforts at the LHC have been made to search for new particles that do not decay promptly but are instead long-lived. This has been done via many different exotic signatures, including searches performed at ATLAS for displaced vertices (DV), where the new long-lived particle decays into multiple visible tracks after having traveled a certain distance into the detector. This talk covers two such searches: a Run 2 search for DVs in events triggered by missing transverse energy, and a Run 3 search for DVs in events triggered by muons. The former search is the first to use a new "fuzzy" displaced vertex reconstruction algorithm, alongside the standard one, to effectively reconstruct cases where the long-lived particle decays into heavy quarks that are themselves slightly long-lived, hence causing the final visible decay products to not point back exactly to the same vertex, setting limits on Higgs Portal, SUSY, and DFSZ axino models. The latter search is the first to use a new displaced muon trigger, setting limits on RPV SUSY models.

We calculate the anomalous dimension of the $|ΔS| = 1$ current-current operators of the weak effective Lagrangian at next-to-next-to-next-to-leading order (NNNLO) in QCD. This constitutes the first step towards a full four-loop calculation of the QCD correction to $ε_K$, the measure for indirect CP violation in the neutral kaon system. We present fully analytic results, together with the expressions necessary to transform our results to a basis with an arbitrary different definition of evanescent operators. As an application, we calculate the corresponding results in the ``standard'' operator basis used in $B$ physics.

We present a comparative collider study of three flavor-violating Higgs signatures in the Type-III Two-Higgs-Doublet Model (\ddHmIII) at \(\sqrt{s}=14\)~TeV: \(pp \to H \to t\bar{c} \ (\bar{t}c)\), \(pp \to H^\pm \to c\bar{b} \ (\bar{c}b)\), and \(pp \to H^\pm \to t\bar{b} \ (\bar{t}b)\). Using a common cut-based analysis and realistic detector simulation, we identify a clear phenomenological hierarchy. The neutral mode and the heavy charged mode emerge as the most robust signatures, containing parameter regions that already exceed the \(5σ\) benchmark at an integrated luminosity of \(300~\text{fb}^{-1}\). In contrast, the light charged channel \(H^{\pm}\to c\bar{b}\) is considerably more sensitive to the event-selection strategy because of large QCD backgrounds, although one competitive benchmark survives. Our results single out the neutral and heavy charged flavor-violating channels as the most promising targets for discovery-oriented searches and for constraining the \ddHmIII\ parameter space at the LHC and HL-LHC. All quoted significances are purely statistical and do not include systematic uncertainties.

Fermion bound states in the background of the electroweak SU(2) monopole are investigated for various values of gauge coupling constant $g$, the Higgs self-coupling constant $λ$, and the Yukawa coupling constant $y_q$. Numerical solutions to the set of coupled differential equations for various selected points in the parameter space reveal only a zero mode, for which we also present an analytic argument. We show that the monopole profile functions and the zero mode wave function become more localized with increasing $g$ and $λ$, while the right-handed component of the latter decreases with $g$. However, as expected, this component increases with $y_q$. We find that the zero mode in the limit $g\to 0$ differs from the zero mode held by the Higgs alone, highlighting the nonlinear and nonperturbative character of the system. Finally, we prove the spectral mirror symmetry of the fermion, whence, together with the existence of the zero mode, we infer the fermion number $1/2$ of the electroweak monopole.

We explore what may be deduced about the neutrino mass ordering problem from the observation of core-collapse supernova burst neutrinos in modern terrestrial detectors. We employ ternary plots in a novel way to visualize the time evolution of the flavor composition of various supernova neutrino flux models from the SNEWPY software package. Through our analysis of several models using a simplified unfolding process, we have explored potential robust discriminants between the normal and inverted mass orderings. We find that the normal and inverted mass orderings tend to occupy different regions in ternary space across different models.

Searches for QCD axions with masses in the neV/$c^2$ mass range are strongly motivated by new physics at the GUT scale and by well-motivated pre-inflationary axion symmetry breaking scales. This parameter space is challenging to probe due to the small axion-photon couplings, which typically require large, high-field magnets with substantial stored energy. In this paper, we propose a new experimental geometry based on a narrow-bore, segmented solenoid that optimizes the collection of the axion-induced signal using LC resonators outside the high-field region of the magnet bore. This alternative optimization significantly reduces the required stored magnetic energy while preserving sensitivity, enabling a near-term experiment in the 30-200 MHz (120-830 neV/$c^2$) range, with a cost-effective, staged scaling to a GUT-scale experiment in the 100 kHz-30 MHz (0.4-120 neV/$c^2$) range.

We investigate electroweak baryogenesis from domain walls with electroweak-symmetric cores moving through the electroweak-broken plasma. In the thick-wall regime, CP-violating semiclassical forces generate chiral asymmetries that source baryon number through transport and weak sphaleron processes. We show that the baryon yield is governed by the hierarchy between the wall width, the CP-violating source width, and the diffusion length, and we identify the corresponding scaling behavior in the relevant parametric limits. A distinctive feature of this mechanism is the interference between the two faces of the domain wall, which leads to qualitatively different behavior for CP-violating sources that are even or odd under wall-orientation reversal. We construct a simplified description that captures these effects and reproduces the predictions of the full transport system in a broad range of parameter space. Applying our framework to a singlet-extended Standard Model, we delineate the region in which electroweak-symmetric domain walls can generate the observed baryon asymmetry.

We predict the primordial deuterium abundance using a novel, fully data-driven approach, where we use Gaussian process regression to fit experimental nuclear reaction data for $d$,($d$,$n$)$^3$He, $d$,($d$,$p$)$t$, and $d$($p$,$γ$)$^3$He, three reactions to which the primordial deuterium abundance is most sensitive. Using the Planck determination of the baryon density, we predict $10^5\times\mathrm{D/H} = 2.442\pm0.040$ in standard Big Bang Nucleosynthesis, $1.70σ$ below the Cooke et al. measurement. Our result is consistent with predictions relying on first principles calculations of the deuterium burning cross sections. With the inferred baryon density from a combined fit to Planck, ACT DR6, and SPT-3G D1, this discrepancy worsens to $1.98σ$. We validate our approach and confirm that Gaussian processes make unbiased D/H predictions with appropriately-sized uncertainties. We repeat our validation tests for low-degree polynomial fits, a technique used in previous analyses, and find that they systematically over-predict D/H. Our results highlight the need for improved measurements of the $d$,($d$,$n$)$^3$He and $d$,($d$,$p$)$t$ S-factors at energies between 0.1 and 0.6 MeV.

Asymmetry measurements in $e^+e^-\toτ^+τ^-$ constitute a promising avenue to obtain competitive constraints on the $τ$ dipole moments, the anomalous magnetic moment $a_τ$ and the electric dipole moment $d_τ$, especially, once a polarized electron beam becomes available, as possible at a future polarization upgrade of the SuperKEKB collider. While the main challenges concern the measurement of these asymmetries and the calculation of radiative corrections at the relevant level of precision, at subleading orders also electroweak effects and the potential impact of four-fermion operators parameterizing other beyond-the-Standard-Model scenarios besides those described by dipole operators need to be taken into consideration. Here, we show that $Z$-boson contributions arise at the level of $\simeq 3\times 10^{-6}$, while we estimate the largest possible effect from four-fermion operators as $\simeq 10^{-5} C \, v^2/Λ^2$. In addition, we observe that four-fermion-operator insertions at the loop level can probe Wilson coefficients that are otherwise not constrained directly, and that the imaginary part generated by insertions of the dipole operator at loop level opens another potential avenue towards a determination of $a_τ$ without the need for a polarized electron beam. Despite the inherent loop suppression, a measurement of the required normal asymmetry $A_N^\pm$ with a precision of $\lesssim 10^{-5}$ would allow one to probe the Schwinger term, which could define an intermediate goal to be realized in the current setting at Belle II.

The search for New Physics requires ever increasing precision from experimental and theoretical efforts. Within the Standard Model Effective Field Theory (SMEFT) framework, the latest achievement in this quest has been the complete computation of the two-loop Renormalisation Group Equations (RGEs) for the Wilson Coefficients of dimension-six operators. In this work, we solve the two-loop SMEFT RGEs with full numerical integration and compare the evolution matrix obtained at one and two loops to analyze how two-loop contributions alter mixing patterns and break zeroes present at one-loop order. Then, we perform a first comprehensive analysis of the impact of the two-loop RGEs in phenomenological studies at HL-LHC and FCC-ee. From a bottom-up perspective, we carry out individual and global fits at linear and quadratic level for a set of 61 Wilson coefficients and compare against the results obtained by including only one-loop RGE effects. We find non-negligible two-loop induced effects in some cases, in particular for four-quark, top Yukawa and Higgs-gluon operators. From a top-down perspective, we perform fits to all the scalar and fermion extensions of the Granada dictionary matched onto SMEFT at one-loop level, including for the first time the couplings that enter only at one loop, and find percent-level effects in the sensitivity to the couplings of some models.

We present projections for the sensitivity of future high-energy colliders to new physics through precision measurements of the Standard Model (SM) interactions, focusing on near-term electron-positron facilities: FCC-ee, LEP3, and the Linear Collider Facility. We interpret these projections in three complementary frameworks: Higgs coupling modifiers, effective Higgs and electroweak couplings, and global SMEFT fits. The SMEFT analysis includes renormalisation-group evolution, linear/quadratic contributions, and NLO corrections to EFT cross sections where available. By matching the EFT to UV-complete models, we also quantify the sensitivity of future colliders to representative benchmark scenarios, including composite Higgs models and single-particle SM extensions. In parallel, we release an updated version of the open-source SMEFiT framework, enabling the results presented here to be fully reproduced, extended, and customised.

We report the first measurement of one- and two-point energy correlators within jets in transversely polarized proton-proton collisions at $\sqrt{s}=200$ GeV, using the STAR detector at RHIC. These observables quantify the energy-weighted angular distribution of single hadrons and hadron pairs within jets, respectively. Sizable spin-dependent asymmetries are observed for $π^+$, $π^-$, and $π^+π^-$ pairs, revealing the onset of nonperturbative dynamics at specific angular scales. By projecting the fragmentation dynamics onto Mellin moments, these measurements provide sensitivity to the nucleon's transversity while minimizing uncertainties from nonperturbative fragmentation functions. These results establish energy correlators as a novel and precise probe of nucleon structure and open a promising avenue for three-dimensional nucleon tomography at the future Electron-Ion Collider.

In this paper, we construct eight pairs of hexaquark currents to search the $Λ_cΣ_c$ and $\barΛ_cΣ_c$ dibaryon states via QCD sum rules. We show that the two currents of each pair are equivalent and we choose one of them to calculate the masses and pole residues of ground states. For either $Λ_cΣ_c$ or $\barΛ_cΣ_c$, the $J^P$ of the considered hexaquark currents are $0^-$, $0^+$, $1^+$ and $1^-$, respectively. We found three possible molecular states, they are $Λ_cΣ_c$ dibaryon with the $J^P=1^+$ and $\barΛ_cΣ_c$ dibaryons with the $J^P=0^-$ and $1^-$. The other five are unlikely to form the bound dibaryon states, and we assign them as the resonance states.