arXiv Daily Digest - 2026-02-25

Customer-provided reviews have become an important source of information for business owners and other customers alike. However, effectively analyzing millions of unstructured reviews remains challenging. While large language models (LLMs) show promise for natural language understanding, their application to large-scale review analysis has been limited by computational costs and scalability concerns. This study proposes a hybrid approach that uses LLMs for aspect identification while employing classic machine-learning methods for sentiment classification at scale. Using ChatGPT to analyze sampled restaurant reviews, we identified key aspects of dining experiences and developed sentiment classifiers using human-labeled reviews, which we subsequently applied to 4.7 million reviews collected over 17 years from a major online platform. Regression analysis reveals that our machine-labeled aspects significantly explain variance in overall restaurant ratings across different aspects of dining experiences, cuisines, and geographical regions. Our findings demonstrate that combining LLMs with traditional machine learning approaches can effectively automate aspect-based sentiment analysis of large-scale customer feedback, suggesting a practical framework for both researchers and practitioners in the hospitality industry and potentially, other service sectors.

Vision Transformers (ViTs) have demonstrated remarkable potential in image processing tasks by utilizing self-attention mechanisms to capture global relationships within data. However, their scalability is hindered by significant computational and memory demands, especially for large-scale models with many parameters. This study aims to leverage DeepSpeed, a highly efficient distributed training framework that is commonly used for language models, to enhance the scalability and performance of ViTs. We evaluate intra- and inter-node training efficiency across multiple GPU configurations on various datasets like CIFAR-10 and CIFAR-100, exploring the impact of distributed data parallelism on training speed, communication overhead, and overall scalability (strong and weak scaling). By systematically varying software parameters, such as batch size and gradient accumulation, we identify key factors influencing performance of distributed training. The experiments in this study provide a foundational basis for applying DeepSpeed to image-related tasks. Future work will extend these investigations to deepen our understanding of DeepSpeed's limitations and explore strategies for optimizing distributed training pipelines for Vision Transformers.

Federated Semi-Supervised Learning (FSSL) aims to collaboratively train a global model across clients by leveraging partially-annotated local data in a privacy-preserving manner. In FSSL, data heterogeneity is a challenging issue, which exists both across clients and within clients. External heterogeneity refers to the data distribution discrepancy across different clients, while internal heterogeneity represents the mismatch between labeled and unlabeled data within clients. Most FSSL methods typically design fixed or dynamic parameter aggregation strategies to collect client knowledge on the server (external) and / or filter out low-confidence unlabeled samples to reduce mistakes in local client (internal). But, the former is hard to precisely fit the ideal global distribution via direct weights, and the latter results in fewer data participation into FL training. To this end, we propose a proxy-guided framework called ProxyFL that focuses on simultaneously mitigating external and internal heterogeneity via a unified proxy. I.e., we consider the learnable weights of classifier as proxy to simulate the category distribution both locally and globally. For external, we explicitly optimize global proxy against outliers instead of direct weights; for internal, we re-include the discarded samples into training by a positive-negative proxy pool to mitigate the impact of potentially-incorrect pseudo-labels. Insight experiments & theoretical analysis show our significant performance and convergence in FSSL.

The analyzability of hybrid software, which integrates both classical and quantum components, is a key factor in ensuring its maintainability and industrial adoption. This article presents the empirical validation, through a family of experiments, of the quantum component of a previously proposed hybrid software analyzability model based on the ISO/IEC 25010 standard. The experimental series consists of four studies involving participants with diverse profiles in both academic and professional settings. In these experiments, the model's ability to effectively measure the analyzability of quantum algorithms is assessed, and the relationship between the analyzability levels computed by the model and the participant's perceptions of the complexity of these algorithms is examined. The results indicate that the proposed model effectively distinguishes between quantum software components with varying levels of analyzability and aligns with human perception, reinforcing its validity in quantum computing.

Off-dynamics offline reinforcement learning (RL) aims to learn a policy for a target domain using limited target data and abundant source data collected under different transition dynamics. Existing methods typically address dynamics mismatch either globally over the state space or via pointwise data filtering; these approaches can miss localized cross-domain similarities or incur high computational cost. We propose Localized Dynamics-Aware Domain Adaptation (LoDADA), which exploits localized dynamics mismatch to better reuse source data. LoDADA clusters transitions from source and target datasets and estimates cluster-level dynamics discrepancy via domain discrimination. Source transitions from clusters with small discrepancy are retained, while those from clusters with large discrepancy are filtered out. This yields a fine-grained and scalable data selection strategy that avoids overly coarse global assumptions and expensive per-sample filtering. We provide theoretical insights and extensive experiments across environments with diverse global and local dynamics shifts. Results show that LoDADA consistently outperforms state-of-the-art off-dynamics offline RL methods by better leveraging localized distribution mismatch.

Knowledge Graphs (KGs) enable the integration and representation of complex information across domains, but their semantic richness and structural complexity create substantial barriers for lay users without expertise in semantic web technologies. When encountering an unfamiliar KG, such users face a distinct orientation challenge: they do not know what questions are possible, how the knowledge is structured, or how to begin exploration. This paper identifies and theorises this phenomenon as the Initial Exploration Problem (IEP). Drawing on theories from information behaviour and human-computer interaction, including ASK, exploratory search, information foraging, and cognitive load theory, we develop a conceptual framing of the IEP characterised by three interdependent barriers: scope uncertainty, ontology opacity, and query incapacity. We argue that these barriers converge at the moment of first contact, distinguishing the IEP from related concepts that presuppose an existing starting point or information goal. Analysing KG exploration interfaces at the level of interaction primitives, we suggest that many systems rely on epistemic assumptions that do not hold at first contact. This reveals a structural gap in the design space: the absence of interaction primitives for scope revelation, mechanisms that communicate what a KG contains without requiring users to formulate queries or interpret ontological structures. In articulating the IEP, this paper provides a theoretical lens for evaluating KG interfaces and for designing entry-point scaffolding that supports initial exploration.

This work introduces a novel training paradigm that draws from affective neuroscience. Inspired by the interplay of emotions and cognition in the human brain and more specifically the SEEKING motivational state, we design a dual-model framework where a smaller base model is trained continuously, while a larger motivated model is activated intermittently during predefined "motivation conditions". The framework mimics the emotional state of high curiosity and anticipation of reward in which broader brain regions are recruited to enhance cognitive performance. Exploiting scalable architectures where larger models extend smaller ones, our method enables shared weight updates and selective expansion of network capacity during noteworthy training steps. Empirical evaluation on the image classification task demonstrates that, not only does the alternating training scheme efficiently and effectively enhance the base model compared to a traditional scheme, in some cases, the motivational model also surpasses its standalone counterpart despite seeing less data per epoch. This opens the possibility of simultaneously training two models tailored to different deployment constraints with competitive or superior performance while keeping training cost lower than when training the larger model.

The Diligent Learner framework suggests LLMs can achieve superintelligence via test-time search, provided a sufficient step-success probability $γ$. In this work, we design a benchmark to measure $γ$ on logical out-of-distribution inference. We construct a class of tasks involving GF(2) circuit reconstruction that grow more difficult with each reasoning step, and that are, from an information-theoretic standpoint, impossible to reliably solve unless the LLM carefully integrates all of the information provided. Our analysis demonstrates that while the $γ$ value for small LLMs declines superlinearly as depth increases, frontier models exhibit partial robustness on this task. Furthermore, we find that successful reasoning at scale is contingent upon precise tool calls, identifying tool design as a critical capability for LLMs to achieve general superintelligence through the Diligent Learner framework.

Large Language Models (LLMs) are transforming scholarly tasks like search and summarization, but their reliability remains uncertain. Current evaluation metrics for testing LLM reliability are primarily automated approaches that prioritize efficiency and scalability, but lack contextual nuance and fail to reflect how scientific domain experts assess LLM outputs in practice. We developed and validated a schema for evaluating LLM errors in scholarly question-answering systems that reflects the assessment strategies of practicing scientists. In collaboration with domain experts, we identified 20 error patterns across seven categories through thematic analysis of 68 question-answer pairs. We validated this schema through contextual inquiries with 10 additional scientists, which showed not only which errors experts naturally identify but also how structured evaluation schemas can help them detect previously overlooked issues. Domain experts use systematic assessment strategies, including technical precision testing, value-based evaluation, and meta-evaluation of their own practices. We discuss implications for supporting expert evaluation of LLM outputs, including opportunities for personalized, schema-driven tools that adapt to individual evaluation patterns and expertise levels.

Large Vision-Language Models (LVLMs) frequently hallucinate, limiting their safe deployment in real-world applications. Existing LLM self-evaluation methods rely on a model's ability to estimate the correctness of its own outputs, which can improve deployment reliability; however, they depend heavily on language priors and are therefore ill-suited for evaluating vision-conditioned predictions. We propose VAUQ, a vision-aware uncertainty quantification framework for LVLM self-evaluation that explicitly measures how strongly a model's output depends on visual evidence. VAUQ introduces the Image-Information Score (IS), which captures the reduction in predictive uncertainty attributable to visual input, and an unsupervised core-region masking strategy that amplifies the influence of salient regions. Combining predictive entropy with this core-masked IS yields a training-free scoring function that reliably reflects answer correctness. Comprehensive experiments show that VAUQ consistently outperforms existing self-evaluation methods across multiple datasets.

Sequential self-attention models usually rely on additive positional embeddings, which inject positional information into item representations at the input. In the absence of positional signals, the attention block is permutation-equivariant over sequence positions and thus has no intrinsic notion of temporal order beyond causal masking. We argue that additive positional embeddings make the attention mechanism only superficially sensitive to sequence order: positional information is entangled with item embedding semantics, propagates weakly in deep architectures, and limits the ability to capture rich sequential patterns. To address these limitations, we introduce a kernelized self-attention mechanism, where a learnable positional kernel operates purely in the position space, disentangled from semantic similarity, and directly modulates attention weights. When applied per attention block, this kernel enables adaptive multi-scale sequential modeling. Experiments on standard next-item prediction benchmarks show that our positional kernel attention consistently improves over strong competing baselines.

Recent deep learning methods for fMRI-based diagnosis have achieved promising accuracy by modeling functional connectivity networks. However, standard approaches often struggle with noisy interactions, and conventional post-hoc attribution methods may lack reliability, potentially highlighting dataset-specific artifacts. To address these challenges, we introduce PIME, an interpretable framework that bridges intrinsic interpretability with minimal-sufficient subgraph optimization by integrating prototype-based classification and consistency training with structural perturbations during learning. This encourages a structured latent space and enables Monte Carlo Tree Search (MCTS) under a prototype-consistent objective to extract compact minimal-sufficient explanatory subgraphs post-training. Experiments on three benchmark fMRI datasets demonstrate that PIME achieves state-of-the-art performance. Furthermore, by constraining the search space via learned prototypes, PIME identifies critical brain regions that are consistent with established neuroimaging findings. Stability analysis shows 90% reproducibility and consistent explanations across atlases.

Large language models (LLMs) are increasingly used in scholarly question-answering (QA) systems to help researchers synthesize vast amounts of literature. However, these systems often produce subtle errors (e.g., unsupported claims, errors of omission), and current provenance mechanisms like source citations are not granular enough for the rigorous verification that scholarly domain requires. To address this, we introduce PaperTrail, a novel interface that decomposes both LLM answers and source documents into discrete claims and evidence, mapping them to reveal supported assertions, unsupported claims, and information omitted from the source texts. We evaluated PaperTrail in a within-subjects study with 26 researchers who performed two scholarly editing tasks using PaperTrail and a baseline interface. Our results show that PaperTrail significantly lowered participants' trust compared to the baseline. However, this increased caution did not translate to behavioral changes, as people continued to rely on LLM-generated scholarly edits to avoid a cognitively burdensome task. We discuss the value of claim-evidence matching for understanding LLM trustworthiness in scholarly settings, and present design implications for cognition-friendly communication of provenance information.

Evaluations of large language models (LLMs) primarily emphasize convergent logical reasoning, where success is defined by producing a single correct proof. However, many real-world reasoning problems admit multiple valid derivations, requiring models to explore diverse logical paths rather than committing to one route. To address this limitation, we introduce LogicGraph, the first benchmark aimed to systematically evaluate multi-path logical reasoning, constructed via a neuro-symbolic framework that leverages backward logic generation and semantic instantiation. This pipeline yields solver-verified reasoning problems formalized by high-depth multi-path reasoning and inherent logical distractions, where each instance is associated with an exhaustive set of minimal proofs. We further propose a reference-free evaluation framework to rigorously assess model performance in both convergent and divergent regimes. Experiments on state-of-the-art language models reveal a common limitation: models tend to commit early to a single route and fail to explore alternatives, and the coverage gap grows substantially with reasoning depth. LogicGraph exposes this divergence gap and provides actionable insights to motivate future improvements. Our code and data will be released at https://github.com/kkkkarry/LogicGraph.

Imputing missing values in multivariate time series remains challenging, especially under diverse missing patterns and heavy missingness. Existing methods suffer from suboptimal performance as corrupted temporal features hinder effective cross-variable information transfer, amplifying reconstruction errors. Robust imputation requires both extracting temporal patterns from sparse observations within each variable and selectively transferring information across variables--yet current approaches excel at one while compromising the other. We introduce T1 (Time series imputation with 1-to-1 channel-head binding), a CNN-Transformer hybrid architecture that achieves robust imputation through Channel-Head Binding--a mechanism creating one-to-one correspondence between CNN channels and attention heads. This design enables selective information transfer: when missingness corrupts certain temporal patterns, their corresponding attention pathways adaptively down-weight based on remaining observable patterns while preserving reliable cross-variable connections through unaffected channels. Experiments on 11 benchmark datasets demonstrate that T1 achieves state-of-the-art performance, reducing MSE by 46% on average compared to the second-best baseline, with particularly strong gains under extreme sparsity (70% missing ratio). The model generalizes to unseen missing patterns without retraining and uses a consistent hyperparameter configuration across all datasets. The code is available at https://github.com/Oppenheimerdinger/T1.

Coalition formation studies how to partition a set of agents into disjoint coalitions under consideration of their preferences. We study the classical objective of stability in a variant of additively separable hedonic games where agents can change their valuations. Our objective is to find a stable partition after each change. To minimize the reconfiguration cost, we search for nearby stable coalition structures. Our focus is on stability concepts based on single-agent deviations. We present a detailed picture of the complexity of finding nearby stable coalition structures in additively separable hedonic games, for both symmetric and non-symmetric valuations. Our results show that the problem is NP-complete for Nash stability, individual stability, contractual Nash stability, and contractual individual stability. We complement these results by presenting polynomial-time algorithms for contractual Nash stability and contractual individual stability under restricted symmetric valuations. Finally, we show that these algorithms guarantee a bounded average distance over long sequences of updates.

Towards understanding the statistical complexity of learning from heterogeneous sources, we study the problem of multi-distribution learning. Given $k$ data sources, the goal is to output a classifier for each source by exploiting shared structure to reduce sample complexity. We focus on the bounded label noise setting to determine whether the fast $1/ε$ rates achievable in single-task learning extend to this regime with minimal dependence on $k$. Surprisingly, we show that this is not the case. We demonstrate that learning across $k$ distributions inherently incurs slow rates scaling with $k/ε^2$, even under constant noise levels, unless each distribution is learned separately. A key technical contribution is a structured hypothesis-testing framework that captures the statistical cost of certifying near-optimality under bounded noise-a cost we show is unavoidable in the multi-distribution setting. Finally, we prove that when competing with the stronger benchmark of each distribution's optimal Bayes error, the sample complexity incurs a \textit{multiplicative} penalty in $k$. This establishes a \textit{statistical} separation between random classification noise and Massart noise, highlighting a fundamental barrier unique to learning from multiple sources.

Medical Cyber-Physical Systems (CPSs) integrating Patients, Devices, and healthcare personnel (Physicians) form safety-critical PDP triads whose dependability is challenged by system heterogeneity and uncertainty in human and physiological behavior. While existing clinical decision support systems support clinical practice, there remains a need for proactive, reliability-oriented methodologies capable of identifying and mitigating failure scenarios before patient safety is compromised. This paper presents M-GENGAR, a methodology based on a closed-loop Digital Twin (DT) paradigm for dependability assurance of medical CPSs. The approach combines Stochastic Hybrid Automata modeling, data-driven learning of patient dynamics, and Statistical Model Checking with an offline critical scenario detection phase that integrates model-space exploration and diversity analysis to systematically identify and classify scenarios violating expert-defined dependability requirements. M-GENGAR also supports the automated synthesis of mitigation strategies, enabling runtime feedback and control within the DT loop. We evaluate M-GENGAR on a representative use case study involving a pulmonary ventilator. Results show that, in 87.5% of the evaluated scenarios, strategies synthesized through formal game-theoretic analysis stabilize patient vital metrics at least as effectively as human decision-making, while maintaining relevant metrics 20% closer to nominal healthy values on average.

Conditional independence tests (CIT) are widely used for causal discovery and feature selection. Even with false discovery rate (FDR) control procedures, they often fail to provide frequentist guarantees in practice. We highlight two common failure modes: (i) in small samples, asymptotic guarantees for many CITs can be inaccurate and even correctly specified models fail to estimate the noise levels and control the error, and (ii) when sample sizes are large but models are misspecified, unaccounted dependencies skew the test's behavior and fail to return uniform p-values under the null. We propose Empirically Calibrated Conditional Independence Tests (ECCIT), a method that measures and corrects for miscalibration. For a chosen base CIT (e.g., GCM, HRT), ECCIT optimizes an adversary that selects features and response functions to maximize a miscalibration metric. ECCIT then fits a monotone calibration map that adjusts the base-test p-values in proportion to the observed miscalibration. Across empirical benchmarks on synthetic and real data, ECCIT achieves valid FDR with higher power than existing calibration strategies while remaining test agnostic.

Medical image processing demands specialized software that handles high-dimensional volumetric data, heterogeneous file formats, and domain-specific training procedures. Existing frameworks either provide low-level components that require substantial integration effort or impose rigid, monolithic pipelines that resist modification. We present MIP Candy (MIPCandy), a freely available, PyTorch-based framework designed specifically for medical image processing. MIPCandy provides a complete, modular pipeline spanning data loading, training, inference, and evaluation, allowing researchers to obtain a fully functional process workflow by implementing a single method, $\texttt{build_network}$, while retaining fine-grained control over every component. Central to the design is $\texttt{LayerT}$, a deferred configuration mechanism that enables runtime substitution of convolution, normalization, and activation modules without subclassing. The framework further offers built-in $k$-fold cross-validation, dataset inspection with automatic region-of-interest detection, deep supervision, exponential moving average, multi-frontend experiment tracking (Weights & Biases, Notion, MLflow), training state recovery, and validation score prediction via quotient regression. An extensible bundle ecosystem provides pre-built model implementations that follow a consistent trainer--predictor pattern and integrate with the core framework without modification. MIPCandy is open-source under the Apache-2.0 license and requires Python~3.12 or later. Source code and documentation are available at https://github.com/ProjectNeura/MIPCandy.

This paper presents the design and implementation of a modular multi-document interface (MDI) framework for scientific visualization and simulation in the Java Virtual Machine (JVM) ecosystem. The framework emphasizes architectural separation between visualization layers, simulation engines, and optional hardware-accelerated 3D rendering. 3D functionality is isolated into a separate module to prevent unnecessary dependency coupling in 2D-only applications. We describe the core abstractions, threading model, simulation integration strategy, and dependency isolation approach. A case study involving a real-time 3D gas expansion simulation integrated with synchronized 2D entropy plotting demonstrates architectural cohesion. The framework is publicly available via Maven Central and targets long-lived scientific and engineering desktop applications.

Common definitions of the "standard" LOCAL model tend to be sloppy and even self-contradictory on one point: do the nodes update their state using an arbitrary function or a computable function? So far, this distinction has been safe to neglect, since problems where it matters seem contrived and quite different from e.g. typical local graph problems studied in this context. We show that this question matters even for locally checkable labeling problems (LCLs), perhaps the most widely studied family of problems in the context of the LOCAL model. Furthermore, we show that assumptions about computability are directly connected to another aspect already recognized as highly relevant: whether we have any knowledge of $n$, the size of the graph. Concretely, we show that there is an LCL problem $Π$ with the following properties: 1. $Π$ can be solved in $O(\log n)$ rounds if the \textsf{LOCAL} model is uncomputable. 2. $Π$ can be solved in $O(\log n)$ rounds in the computable model if we know any upper bound on $n$. 3. $Π$ requires $Ω(\sqrt{n})$ rounds in the computable model if we do not know anything about $n$. We also show that the connection between computability and knowledge of $n$ holds in general: for any LCL problem $Π$, if you have any bound on $n$, then $Π$ has the same round complexity in the computable and uncomputable models.

Multi-agent imitation learning (MA-IL) aims to learn optimal policies from expert demonstrations of interactions in multi-agent interactive domains. Despite existing guarantees on the performance of the resulting learned policies, characterizations of how far the learned polices are from a Nash equilibrium are missing for offline MA-IL. In this paper, we demonstrate impossibility and hardness results of learning low-exploitable policies in general $n$-player Markov Games. We do so by providing examples where even exact measure matching fails, and demonstrating a new hardness result on characterizing the Nash gap given a fixed measure matching error. We then show how these challenges can be overcome using strategic dominance assumptions on the expert equilibrium. Specifically, for the case of dominant strategy expert equilibria, assuming Behavioral Cloning error $ε_{\text{BC}}$, this provides a Nash imitation gap of $\mathcal{O}\left(nε_{\text{BC}}/(1-γ)^2\right)$ for a discount factor $γ$. We generalize this result with a new notion of best-response continuity, and argue that this is implicitly encouraged by standard regularization techniques.

Modern recommender systems leverage ultra-long user behavior sequences to capture dynamic preferences, but end-to-end modeling is infeasible in production due to latency and memory constraints. While summarizing history via interest centers offers a practical alternative, existing methods struggle to (1) identify user-specific centers at appropriate granularity and (2) accurately assign behaviors, leading to quantization errors and loss of long-tail preferences. To alleviate these issues, we propose Hierarchical Sparse Activation Compression (HiSAC), an efficient framework for personalized sequence modeling. HiSAC encodes interactions into multi-level semantic IDs and constructs a global hierarchical codebook. A hierarchical voting mechanism sparsely activates personalized interest-agents as fine-grained preference centers. Guided by these agents, Soft-Routing Attention aggregates historical signals in semantic space, weighting by similarity to minimize quantization error and retain long-tail behaviors. Deployed on Taobao's "Guess What You Like" homepage, HiSAC achieves significant compression and cost reduction, with online A/B tests showing a consistent 1.65% CTR uplift -- demonstrating its scalability and real-world effectiveness.

Pre-ranking is a critical stage in industrial recommendation systems, tasked with efficiently scoring thousands of recalled items for downstream ranking. A key challenge is the train-serving discrepancy: pre-ranking models are trained only on exposed interactions, yet must score all recalled candidates -- including unexposed items -- during online serving. This mismatch not only induces severe sample selection bias but also degrades generalization, especially for long-tail content. Existing debiasing approaches typically rely on heuristics (e.g., negative sampling) or distillation from biased rankers, which either mislabel plausible unexposed items as negatives or propagate exposure bias into pseudo-labels. In this work, we propose Generative Pseudo-Labeling (GPL), a framework that leverages large language models (LLMs) to generate unbiased, content-aware pseudo-labels for unexposed items, explicitly aligning the training distribution with the online serving space. By offline generating user-specific interest anchors and matching them with candidates in a frozen semantic space, GPL provides high-quality supervision without adding online latency. Deployed in a large-scale production system, GPL improves click-through rate by 3.07%, while significantly enhancing recommendation diversity and long-tail item discovery.

Report-supervised (RSuper) learning seeks to alleviate the need for dense tumor voxel labels with constraints derived from radiology reports (e.g., volumes, counts, sizes, locations). In MRI studies of brain tumors, however, we often involve multi-parametric scans and substructures. Here, fine-grained modality/parameter-wise reports are usually provided along with global findings and are correlated with different substructures. Moreover, the reports often describe only the largest lesion and provide qualitative or uncertain cues (``mild,'' ``possible''). Classical RSuper losses (e.g., sum volume consistency) can over-constrain or hallucinate unreported findings under such incompleteness, and are unable to utilize these hierarchical findings or exploit the priors of varied lesion types in a merged dataset. We explicitly parse the global quantitative and modality-wise qualitative findings and introduce a unified, one-sided, uncertainty-aware formulation (MS-RSuper) that: (i) aligns modality-specific qualitative cues (e.g., T1c enhancement, FLAIR edema) with their corresponding substructures using existence and absence losses; (ii) enforces one-sided lower-bounds for partial quantitative cues (e.g., largest lesion size, minimal multiplicity); and (iii) adds extra- vs. intra-axial anatomical priors to respect cohort differences. Certainty tokens scale penalties; missing cues are down-weighted. On 1238 report-labeled BraTS-MET/MEN scans, our MS-RSuper largely outperforms both a sparsely-supervised baseline and a naive RSuper method.

Scaling multimodal alignment between video and audio is challenging, particularly due to limited data and the mismatch between text descriptions and frame-level video information. In this work, we tackle the scaling challenge in multimodal-to-audio generation, examining whether models trained on short instances can generalize to longer ones during testing. To tackle this challenge, we present multimodal hierarchical networks so-called MMHNet, an enhanced extension of state-of-the-art video-to-audio models. Our approach integrates a hierarchical method and non-causal Mamba to support long-form audio generation. Our proposed method significantly improves long audio generation up to more than 5 minutes. We also prove that training short and testing long is possible in the video-to-audio generation tasks without training on the longer durations. We show in our experiments that our proposed method could achieve remarkable results on long-video to audio benchmarks, beating prior works in video-to-audio tasks. Moreover, we showcase our model capability in generating more than 5 minutes, while prior video-to-audio methods fall short in generating with long durations.

Multimodal Large Language Models (MLLMs) have achieved remarkable performance by integrating powerful language backbones with large-scale visual encoders. Among these, latent Chain-of-Thought (CoT) methods enable implicit reasoning in continuous hidden states, facilitating seamless vision-language integration and faster inference. However, existing heuristically predefined supervision signals in latent CoT provide limited guidance for preserving critical visual information in intermediate latent states. To address this limitation, we propose CrystaL (Crystallized Latent Reasoning), a single-stage framework with two paths to process intact and corrupted images, respectively. By explicitly aligning the attention patterns and prediction distributions across the two paths, CrystaL crystallizes latent representations into task-relevant visual semantics, without relying on auxiliary annotations or external modules. Extensive experiments on perception-intensive benchmarks demonstrate that CrystaL consistently outperforms state-of-the-art baselines, achieving substantial gains in fine-grained visual understanding while maintaining robust reasoning capabilities.

Fully leveraging the capabilities of AI agents in software development requires a rethinking of the software ecosystem itself. To this end, this paper outlines the creation of an Agentic Infused Software Ecosystem (AISE), that rests on three pillars. The first, of course, is the AI agents themselves, which in the past 5 years have moved from simple code completion and toward sophisticated independent development tasks, a trend which will only continue. The second pillar is the programming language and APIs (or tools) that these agents use to accomplish tasks, and increasingly, serve as the communication substrate that humans and AI agents interact and collaborate through. The final pillar is the runtime environment and ecosystem that agents operate within, and which provide the capabilities that programmatic agents use to interface with (and effect actions in) the external world. To realize the vision of AISE, all three pillars must be advanced in a holistic manner, and critically, in a manner that is synergistic for AI agents as they exist today, those that will exist in the future, and for the human developers that work alongside them.

As large language models (LLMs) are increasingly embedded in everyday decision-making, their safety responsibilities extend beyond reacting to explicit harmful intent toward anticipating unintended but consequential risks. In this work, we introduce a proactive risk awareness evaluation framework that measures whether LLMs can anticipate potential harms and provide warnings before damage occurs. We construct the Butterfly dataset to instantiate this framework in the environmental and ecological domain. It contains 1,094 queries that simulate ordinary solution-seeking activities whose responses may induce latent ecological impact. Through experiments across five widely used LLMs, we analyze the effects of response length, languages, and modality. Experimental results reveal consistent, significant declines in proactive awareness under length-restricted responses, cross-lingual similarities, and persistent blind spots in (multimodal) species protection. These findings highlight a critical gap between current safety alignment and the requirements of real-world ecological responsibility, underscoring the need for proactive safeguards in LLM deployment.

In engineering design and scientific computing, computational cost and predictive accuracy are intrinsically coupled. High-fidelity simulations provide accurate predictions but at substantial computational costs, while lower-fidelity approximations offer efficiency at the expense of accuracy. Multi-fidelity surrogate modelling addresses this trade-off by combining abundant low-fidelity data with sparse high-fidelity observations. However, existing methods suffer from expensive training cost or rely on global correlation assumptions that often fail in practice to capture how fidelity relationships vary across the input space, leading to poor performance particularly under tight budget constraints. We introduce MAST, a method that blends corrected low-fidelity observations with high-fidelity predictions, trusting high-fidelity near observed samples and relying on corrected low-fidelity elsewhere. MAST achieves this through explicit discrepancy modelling and distance-based weighting with closed-form variance propagation, producing a single heteroscedastic Gaussian process. Across multi-fidelity synthetic benchmarks, MAST shows a marked improvement over the current state-of-the-art techniques. Crucially, MAST maintains robust performance across varying total budget and fidelity gaps, conditions under which competing methods exhibit significant degradation or unstable behaviour.

To comprehensively evaluate the mathematical reasoning capabilities of Large Language Models (LLMs), researchers have introduced abundant mathematical reasoning datasets. However, most existing datasets primarily focus on linear reasoning, neglecting other parts such as proof by contradiction and proof by cases, which are crucial for investigating LLMs' reasoning abilities. To address this limitation, we first introduce a novel first-order logic (FOL) dataset named PC-FOL, annotated by professional mathematicians, focusing on case-based reasoning problems. All instances in this dataset are equipped with a manually written natural language proof, clearly distinguishing it from conventional linear reasoning datasets. Our experimental results over leading LLMs demonstrate a substantial performance gap between linear reasoning and case-based reasoning problems. To further investigate this phenomenon, we provide a theoretical analysis grounded in graphical model, which provides an explanation for the observed disparity between the two types of reasoning problems. We hope this work can reveal the core challenges in the field of automated natural language mathematical proof generation, paving the way for future research.

Bubeck and Sellke (2021) pose as an open problem the connection between the law of robustness and robust generalization. The law of robustness states that overparameterization is necessary for models to interpolate robustly; in particular, robust interpolation requires the learned function to be Lipschitz. Robust generalization asks whether small robust training loss implies small robust test loss. We resolve this problem by explicitly connecting the two for arbitrary data distributions. Specifically, we introduce a nontrivial notion of robust generalization error and convert it into a lower bound on the expected Rademacher complexity of the induced robust loss class. Our bounds recover the $Ω(n^{1/d})$ regime of Wu et al.\ (2023) and show that, up to constants, robust generalization does not change the order of the Lipschitz constant required for smooth interpolation. We conduct experiments to probe the predicted scaling with dataset size and model capacity, testing whether empirical behavior aligns more closely with the predictions of Bubeck and Sellke (2021) or Wu et al.\ (2023). For MNIST, we find that the lower-bound Lipschitz constant scales on the order predicted by Wu et al.\ (2023). Informally, to obtain low robust generalization error, the Lipschitz constant must lie in a range that we bound, and the allowable perturbation radius is linked to the Lipschitz scale.

Automatic speech recognition (ASR) degrades severely in noisy environments. Although speech enhancement (SE) front-ends effectively suppress background noise, they often introduce artifacts that harm recognition. Observation addition (OA) addressed this issue by fusing noisy and SE enhanced speech, improving recognition without modifying the parameters of the SE or ASR models. This paper proposes an intelligibility-guided OA method, where fusion weights are derived from intelligibility estimates obtained directly from the backend ASR. Unlike prior OA methods based on trained neural predictors, the proposed method is training-free, reducing complexity and enhances generalization. Extensive experiments across diverse SE-ASR combinations and datasets demonstrate strong robustness and improvements over existing OA baselines. Additional analyses of intelligibility-guided switching-based alternatives and frame versus utterance-level OA further validate the proposed design.

This article describes a novel language task, the Blackbird Language Matrices (BLM) task, inspired by intelligence tests, and illustrates the BLM datasets, their construction and benchmarking, and targeted experiments on chunking and systematicity. BLMs are multiple-choice problems, structured at multiple levels: within each sentence, across the input sequence, within each candidate answer. Because of their rich structure, these curated, but naturalistic datasets are key to answer some core questions about current large language models abilities: do LLMs detect linguistic objects and their properties? Do they detect and use systematic patterns across sentences? Are they more prone to linguistic or reasoning errors, and how do these interact? We show that BLMs, while challenging, can be solved at good levels of performance, in more than one language, with simple baseline models or, at better performance levels, with more tailored models. We show that their representations contain the grammatical objects and attributes relevant to solve a linguistic task. We also show that these solutions are reached by detecting systematic patterns across sentences. The paper supports the point of view that curated, structured datasets support multi-faceted investigations of properties of language and large language models. Because they present a curated, articulated structure, because they comprise both learning contexts and expected answers, and because they are partly built by hand, BLMs fall in the category of datasets that can support explainability investigations, and be useful to ask why large language models behave the way they do.

Search and rescue (SAR) operations require rapid responses to save lives or property. Unmanned Aerial Vehicles (UAVs) equipped with vision-based systems support these missions through prior terrain investigation or real-time assistance during the mission itself. Vision-based UAV frameworks aid human search tasks by detecting and recognizing specific individuals, then tracking and following them while maintaining a safe distance. A key safety requirement for UAV following is the accurate estimation of the distance between camera and target object under real-world conditions, achieved by fusing multiple image modalities. UAVs with deep learning-based vision systems offer a new approach to the planning and execution of SAR operations. As part of the system for automatic people detection and face recognition using deep learning, in this paper we present the fusion of depth camera measurements and monocular camera-to-body distance estimation for robust tracking and following. Deep learning-based filtering of depth camera data and estimation of camera-to-body distance from a monocular camera are achieved with YOLO-pose, enabling real-time fusion of depth information using the Extended Kalman Filter (EKF) algorithm. The proposed subsystem, designed for use in drones, estimates and measures the distance between the depth camera and the human body keypoints, to maintain the safe distance between the drone and the human target. Our system provides an accurate estimated distance, which has been validated against motion capture ground truth data. The system has been tested in real time indoors, where it reduces the average errors, root mean square error (RMSE) and standard deviations of distance estimation up to 15,3\% in three tested scenarios.

Despite recent advances in diffusion models, AI generated images still often contain visual artifacts that compromise realism. Although more thorough pre-training and bigger models might reduce artifacts, there is no assurance that they can be completely eliminated, which makes artifact mitigation a highly crucial area of study. Previous artifact-aware methodologies depend on human-labeled artifact datasets, which are costly and difficult to scale, underscoring the need for an automated approach to reliably acquire artifact-annotated datasets. In this paper, we propose ArtiAgent, which efficiently creates pairs of real and artifact-injected images. It comprises three agents: a perception agent that recognizes and grounds entities and subentities from real images, a synthesis agent that introduces artifacts via artifact injection tools through novel patch-wise embedding manipulation within a diffusion transformer, and a curation agent that filters the synthesized artifacts and generates both local and global explanations for each instance. Using ArtiAgent, we synthesize 100K images with rich artifact annotations and demonstrate both efficacy and versatility across diverse applications. Code is available at link.

The performance and ease of use of deep learning-based binary classifiers have improved significantly in recent years. This has opened up the potential for automating critical inspection tasks, which have traditionally only been trusted to be done manually. However, the application of binary classifiers in critical operations depends on the estimation of reliable confidence bounds such that system performance can be ensured up to a given statistical significance. We present Wilson Score Kernel Density Classification, which is a novel kernel-based method for estimating confidence bounds in binary classification. The core of our method is the Wilson Score Kernel Density Estimator, which is a function estimator for estimating confidence bounds in Binomial experiments with conditionally varying success probabilities. Our method is evaluated in the context of selective classification on four different datasets, illustrating its use as a classification head of any feature extractor, including vision foundation models. Our proposed method shows similar performance to Gaussian Process Classification, but at a lower computational complexity.

For millennia, human cognition was the primary engine of progress on Earth. As AI decouples cognition from biology, the marginal cost of measurable execution falls to zero, absorbing any labor capturable by metrics--including creative, analytical, and innovative work. The binding constraint on growth is no longer intelligence but human verification bandwidth: the capacity to validate, audit, and underwrite responsibility when execution is abundant. We model the AGI transition as the collision of two racing cost curves: an exponentially decaying Cost to Automate and a biologically bottlenecked Cost to Verify. This structural asymmetry widens a Measurability Gap between what agents can execute and what humans can afford to verify. It also drives a shift from skill-biased to measurability-biased technical change. Rents migrate to verification-grade ground truth, cryptographic provenance, and liability underwriting--the ability to insure outcomes rather than merely generate them. The current human-in-the-loop equilibrium is unstable: eroded from below as apprenticeship collapses (Missing Junior Loop) and from within as experts codify their obsolescence (Codifier's Curse). Unverified deployment becomes privately rational--a Trojan Horse externality. Unmanaged, these forces pull toward a Hollow Economy. Yet by scaling verification alongside agentic capabilities, the forces that threaten collapse become the catalyst for unbounded discovery and experimentation--an Augmented Economy. We derive a practical playbook for individuals, companies, investors, and policymakers. Today's defining challenge is not the race to deploy the most autonomous systems; it is the race to secure the foundations of their oversight. Only by scaling our bandwidth for verification alongside our capacity for execution can we ensure that the intelligence we have summoned preserves the humanity that initiated it.

Large Language Models (LLMs) consistently benefit from scaled Chain-of-Thought (CoT) reasoning, but also suffer from heavy computational overhead. To address this issue, efficient reasoning aims to incentivize short yet accurate thinking trajectories, typically through reward shaping with Reinforcement Learning (RL). In this paper, we systematically investigate the mechanics of efficient reasoning for LLMs. For comprehensive evaluation, we advocate for more fine-grained metrics, including length distribution conditioned on correctness and performance across a wide spectrum of token budgets ranging from 2k to 32k. First, we reveal that the training process follows a two-stage paradigm: length adaptation and reasoning refinement. After that, we conduct extensive experiments (about 0.2 million GPU hours) in a unified protocol, deconstructing training prompts and rollouts, reward shaping, and optimization strategies. In particular, a key finding is to train on relatively easier prompts, ensuring the density of positive reward signals and thus avoiding the length collapse. Meanwhile, the learned length bias can be generalized across domains. We distill all findings into valuable insights and practical guidelines, and further validate them across the Qwen3 series, ranging from 0.6B to 30B, demonstrating the robustness and generalization.

Several variations of adaptive first-order and second-order optimization methods have been proposed to accelerate and scale the training of large language models. The performance of these optimization routines is highly sensitive to the choice of hyperparameters (HPs), which are computationally expensive to tune for large-scale models. Maximal update parameterization $(μ$P$)$ is a set of scaling rules which aims to make the optimal HPs independent of the model size, thereby allowing the HPs tuned on a smaller (computationally cheaper) model to be transferred to train a larger, target model. Despite promising results for SGD and Adam, deriving $μ$P for other optimizers is challenging because the underlying tensor programming approach is difficult to grasp. Building on recent work that introduced spectral conditions as an alternative to tensor programs, we propose a novel framework to derive $μ$P for a broader class of optimizers, including AdamW, ADOPT, LAMB, Sophia, Shampoo and Muon. We implement our $μ$P derivations on multiple benchmark models and demonstrate zero-shot learning rate transfer across increasing model width for the above optimizers. Further, we provide empirical insights into depth-scaling parameterization for these optimizers.

The paradigm of Large Language Models is undergoing a fundamental transition from static inference engines to dynamic autonomous cognitive systems.While current research primarily focuses on scaling context windows or optimizing prompt engineering the theoretical bridge between micro scale token processing and macro scale systemic intelligence remains fragmented.This paper proposes AgentOS,a holistic conceptual framework that redefines the LLM as a "Reasoning Kernel" governed by structured operating system logic.Central to this architecture is Deep Context Management which conceptualizes the context window as an Addressable Semantic Space rather than a passive buffer.We systematically deconstruct the transition from discrete sequences to coherent cognitive states introducing mechanisms for Semantic Slicing and Temporal Alignment to mitigate cognitive drift in multi-agent orchestration.By mapping classical OS abstractions such as memory paging interrupt handling and process scheduling onto LLM native constructs, this review provides a rigorous roadmap for architecting resilient scalable and self-evolving cognitive environments.Our analysis asserts that the next frontier of AGI development lies in the architectural efficiency of system-level coordination.

An electroencephalogram (EEG) records the spatially averaged electrical activity of neurons in the brain, measured from the human scalp. Prior studies have explored EEG-based classification of objects or concepts, often for passive viewing of briefly presented image or video stimuli, with limited classes. Because EEG exhibits a low signal-to-noise ratio, recognizing fine-grained representations across a large number of classes remains challenging; however, abstract-level object representations may exist. In this work, we investigate whether EEG captures object representations across multiple hierarchical levels, and propose episodic analysis, in which a Machine Learning (ML) model is evaluated across various, yet related, classification tasks (episodes). Unlike prior episodic EEG studies that rely on fixed or randomly sampled classes of equal cardinality, we adopt hierarchy-aware episode sampling using WordNet to generate episodes with variable classes of diverse hierarchy. We also present the largest episodic framework in the EEG domain for detecting observed text from EEG signals in the PEERS dataset, comprising $931538$ EEG samples under $1610$ object labels, acquired from $264$ human participants (subjects) performing controlled cognitive tasks, enabling the study of neural dynamics underlying perception, decision-making, and performance monitoring. We examine how the semantic abstraction level affects classification performance across multiple learning techniques and architectures, providing a comprehensive analysis. The models tend to improve performance when the classification categories are drawn from higher levels of the hierarchy, suggesting sensitivity to abstraction. Our work highlights abstraction depth as an underexplored dimension of EEG decoding and motivates future research in this direction.

Large Language Models (LLMs) often struggle with inherent knowledge boundaries and hallucinations, limiting their reliability in knowledge-intensive tasks. While Retrieval-Augmented Generation (RAG) mitigates these issues, it frequently overlooks structural interdependencies essential for multi-hop reasoning. Graph-based RAG approaches attempt to bridge this gap, yet they typically face trade-offs between accuracy and efficiency due to challenges such as costly graph traversals and semantic noise in LLM-generated summaries. In this paper, we propose HyperNode Expansion and Logical Path-Guided Evidence Localization strategies for GraphRAG (HELP), a novel framework designed to balance accuracy with practical efficiency through two core strategies: 1) HyperNode Expansion, which iteratively chains knowledge triplets into coherent reasoning paths abstracted as HyperNodes to capture complex structural dependencies and ensure retrieval accuracy; and 2) Logical Path-Guided Evidence Localization, which leverages precomputed graph-text correlations to map these paths directly to the corpus for superior efficiency. HELP avoids expensive random walks and semantic distortion, preserving knowledge integrity while drastically reducing retrieval latency. Extensive experiments demonstrate that HELP achieves competitive performance across multiple simple and multi-hop QA benchmarks and up to a 28.8$\times$ speedup over leading Graph-based RAG baselines.

Internet measurement faces twin challenges: complex analyses require expert-level orchestration of tools, yet even syntactically correct implementations can have methodological flaws and can be difficult to verify. Democratizing measurement capabilities thus demands automating both workflow generation and verification against methodological standards established through decades of research. We present Airavat, the first agentic framework for Internet measurement workflow generation with systematic verification and validation. Airavat coordinates a set of agents mirroring expert reasoning: three agents handle problem decomposition, solution design, and code implementation, with assistance from a registry of existing tools. Two specialized engines ensure methodological correctness: a Verification Engine evaluates workflows against a knowledge graph encoding five decades of measurement research, while a Validation Engine identifies appropriate validation techniques grounded in established methodologies. Through four Internet measurement case studies, we demonstrate that Airavat (i) generates workflows matching expert-level solutions, (ii) makes sound architectural decisions, (iii) addresses novel problems without ground truth, and (iv) identifies methodological flaws missed by standard execution-based testing.

Deep neural networks (DNNs) have significantly advanced machine learning, with model depth playing a central role in their successes. The dynamical system modeling approach has recently emerged as a powerful framework, offering new mathematical insights into the structure and learning behavior of DNNs. In this work, we establish generalization error bounds for both discrete- and continuous-time residual networks (ResNets) by combining Rademacher complexity, flow maps of dynamical systems, and the convergence behavior of ResNets in the deep-layer limit. The resulting bounds are of order $O(1/\sqrt{S})$ with respect to the number of training samples $S$, and include a structure-dependent negative term, yielding depth-uniform and asymptotic generalization bounds under milder assumptions. These findings provide a unified understanding of generalization across both discrete- and continuous-time ResNets, helping to close the gap in both the order of sample complexity and assumptions between the discrete- and continuous-time settings.

Previous work has examined the capacity of deep neural networks (DNNs), particularly transformers, to predict human sentence acceptability judgments, both independently of context, and in document contexts. We consider the effect of prior exposure to visual images (i.e., visual context) on these judgments for humans and large language models (LLMs). Our results suggest that, in contrast to textual context, visual images appear to have little if any impact on human acceptability ratings. However, LLMs display the compression effect seen in previous work on human judgments in document contexts. Different sorts of LLMs are able to predict human acceptability judgments to a high degree of accuracy, but in general, their performance is slightly better when visual contexts are removed. Moreover, the distribution of LLM judgments varies among models, with Qwen resembling human patterns, and others diverging from them. LLM-generated predictions on sentence acceptability are highly correlated with their normalised log probabilities in general. However, the correlations decrease when visual contexts are present, suggesting that a higher gap exists between the internal representations of LLMs and their generated predictions in the presence of visual contexts. Our experimental work suggests interesting points of similarity and of difference between human and LLM processing of sentences in multimodal contexts.

Class-Incremental Learning (CIL) requires models to learn new classes without forgetting old ones. A common method is to freeze a pre-trained model and train a new, lightweight adapter for each task. While this prevents forgetting, it treats the learned knowledge as a simple, unstructured collection and fails to use the relationships between tasks. To this end, we propose the Semantic-guided Adaptive Expert Forest (SAEF), a new method that organizes adapters into a structured hierarchy for better knowledge sharing. SAEF first groups tasks into conceptual clusters based on their semantic relationships. Then, within each cluster, it builds a balanced expert tree by creating new adapters from merging the adapters of similar tasks. At inference time, SAEF finds and activates a set of relevant experts from the forest for any given input. The final prediction is made by combining the outputs of these activated experts, weighted by how confident each expert is. Experiments on several benchmark datasets show that SAEF achieves SOTA performance.

While reasoning models are increasingly ubiquitous, the effects of reasoning training on a model's internal mechanisms remain poorly understood. In this work, we introduce transcoder adapters, a technique for learning an interpretable approximation of the difference in MLP computation before and after fine-tuning. We apply transcoder adapters to characterize the differences between Qwen2.5-Math-7B and its reasoning-distilled variant, DeepSeek-R1-Distill-Qwen-7B. Learned adapters are faithful to the target model's internal computation and next-token predictions. When evaluated on reasoning benchmarks, adapters match the reasoning model's response lengths and typically recover 50-90% of the accuracy gains from reasoning fine-tuning. Adapter features are sparsely activating and interpretable. When examining adapter features, we find that only ~8% have activating examples directly related to reasoning behaviors. We deeply study one such behavior -- the production of hesitation tokens (e.g., "wait"). Using attribution graphs, we trace hesitation to only ~2.4% of adapter features (5.6k total) performing one of two functions. These features are necessary and sufficient for producing hesitation tokens; removing them reduces response length, often without affecting accuracy. Overall, our results provide insight into reasoning training and suggest transcoder adapters may be useful for studying fine-tuning more broadly.

Vision-Language Models (VLMs) have been increasingly applied in real-world scenarios due to their outstanding understanding and reasoning capabilities. Although VLMs have already demonstrated impressive capabilities in common visual question answering and logical reasoning, they still lack the ability to make reasonable decisions in complex real-world environments. We define this ability as spatial logical reasoning, which not only requires understanding the spatial relationships among objects in complex scenes, but also the logical dependencies between steps in multi-step tasks. To bridge this gap, we introduce Spatial Logical Question Answering (SpatiaLQA), a benchmark designed to evaluate the spatial logical reasoning capabilities of VLMs. SpatiaLQA consists of 9,605 question answer pairs derived from 241 real-world indoor scenes. We conduct extensive experiments on 41 mainstream VLMs, and the results show that even the most advanced models still struggle with spatial logical reasoning. To address this issue, we propose a method called recursive scene graph assisted reasoning, which leverages visual foundation models to progressively decompose complex scenes into task-relevant scene graphs, thereby enhancing the spatial logical reasoning ability of VLMs, outperforming all previous methods. Code and dataset are available at https://github.com/xieyc99/SpatiaLQA.

Today, Social networks such as Twitter are the most widely used platforms for communication of people. Analyzing this data has useful information to recognize the opinion of people in tweets. Sentiment analysis plays a vital role in NLP, which identifies the opinion of the individuals about a specific topic. Natural language processing in Persian has many challenges despite the adventure of strong language models. The datasets available in Persian are generally in special topics such as products, foods, hotels, etc while users may use ironies, colloquial phrases in social media To overcome these challenges, there is a necessity for having a dataset of Persian sentiment analysis on Twitter. In this paper, we introduce the Exa sentiment analysis Persian dataset, which is collected from Persian tweets. This dataset contains 12,000 tweets, annotated by 5 native Persian taggers. The aforementioned data is labeled in 3 classes: positive, neutral and negative. We present the characteristics and statistics of this dataset and use the pre-trained Pars Bert and Roberta as the base model to evaluate this dataset. Our evaluation reached a 79.87 Macro F-score, which shows the model and data can be adequately valuable for a sentiment analysis system.

The forest-of-refinement-trees approach allows for dynamic adaptive mesh refinement (AMR) at negligible cost. While originally developed for quadrilateral and hexahedral elements, previous work established the theory and algorithms for unstructured meshes of simplicial and prismatic elements. To harness the full potential of tree-based AMR for three-dimensional mixed-element meshes, this paper introduces the pyramid as a new functional element type; its primary purpose is to connect tetrahedral and hexahedral elements without hanging edges.We present a well-defined space-filling curve (SFC) for the pyramid and detail how the unique challenges on the element and forest level associated with the pyramidal refinement are resolved. We propose the necessary functional design and generalize the fundamental global parallel algorithms for refinement, coarsening, partitioning, and face ghost exchange to fully support this new element. Our demonstrations confirm the efficiency and scalability of this complete, hybrid-element dynamic AMR framework.

Large Vision-Language Models (LVLMs) achieve strong performance on visual question answering benchmarks, yet often rely on spurious correlations rather than genuine causal reasoning. Existing evaluations primarily assess the correctness of the answers, making it unclear whether failures arise from limited reasoning capability or from misidentifying causally relevant information. We introduce Vision-Language Causal Graphs (VLCGs), a structured, query-conditioned representation that explicitly encodes causally relevant objects, attributes, relations, and scene-grounded assumptions. Building on this representation, we present ViLCaR, a diagnostic benchmark comprising tasks for Causal Attribution, Causal Inference, and Question Answering, along with graph-aligned evaluation metrics that assess relevance identification beyond final answer accuracy. Experiments in state-of-the-art LVLMs show that injecting structured relevance information significantly improves attribution and inference consistency compared to zero-shot and standard in-context learning. These findings suggest that current limitations in LVLM causal reasoning stem primarily from insufficient structural guidance rather than a lack of reasoning capacity.

Multimodal recommender systems (MMRSs) enhance collaborative filtering by leveraging item-side modalities, but their reliance on a fixed set of modalities and task-specific objectives limits both modality extensibility and task generalization. We propose E-MMKGR, a framework that constructs an e-commerce-specific Multimodal Knowledge Graph E-MMKG and learns unified item representations through GNN-based propagation and KG-oriented optimization. These representations provide a shared semantic foundation applicable to diverse tasks. Experiments on real-world Amazon datasets show improvements of up to 10.18% in Recall@10 for recommendation and up to 21.72% over vector-based retrieval for product search, demonstrating the effectiveness and extensibility of our approach.

Agentic systems increasingly rely on reusable procedural capabilities, \textit{a.k.a., agentic skills}, to execute long-horizon workflows reliably. These capabilities are callable modules that package procedural knowledge with explicit applicability conditions, execution policies, termination criteria, and reusable interfaces. Unlike one-off plans or atomic tool calls, skills operate (and often do well) across tasks. This paper maps the skill layer across the full lifecycle (discovery, practice, distillation, storage, composition, evaluation, and update) and introduces two complementary taxonomies. The first is a system-level set of \textbf{seven design patterns} capturing how skills are packaged and executed in practice, from metadata-driven progressive disclosure and executable code skills to self-evolving libraries and marketplace distribution. The second is an orthogonal \textbf{representation $\times$ scope} taxonomy describing what skills \emph{are} (natural language, code, policy, hybrid) and what environments they operate over (web, OS, software engineering, robotics). We analyze the security and governance implications of skill-based agents, covering supply-chain risks, prompt injection via skill payloads, and trust-tiered execution, grounded by a case study of the ClawHavoc campaign in which nearly 1{,}200 malicious skills infiltrated a major agent marketplace, exfiltrating API keys, cryptocurrency wallets, and browser credentials at scale. We further survey deterministic evaluation approaches, anchored by recent benchmark evidence that curated skills can substantially improve agent success rates while self-generated skills may degrade them. We conclude with open challenges toward robust, verifiable, and certifiable skills for real-world autonomous agents.

Financial prediction from long documents involves significant challenges, as actionable signals are often sparse and obscured by noise, and the optimal LLM for generating embeddings varies across tasks and time periods. In this paper, we propose FinAnchor(Financial Anchored Representations), a lightweight framework that integrates embeddings from multiple LLMs without fine-tuning the underlying models. FinAnchor addresses the incompatibility of feature spaces by selecting an anchor embedding space and learning linear mappings to align representations from other models into this anchor. These aligned features are then aggregated to form a unified representation for downstream prediction. Across multiple financial NLP tasks, FinAnchor consistently outperforms strong single-model baselines and standard ensemble methods, demonstrating the effectiveness of anchoring heterogeneous representations for robust financial prediction.

The analysis of non-stationary time-series data requires insight into its local and global patterns with physical interpretability. However, traditional smoothing algorithms, such as B-splines, Savitzky-Golay filtering, and Empirical Mode Decomposition (EMD), lack the ability to perform parametric optimization with guaranteed continuity. In this paper, we propose Functional Continuous Decomposition (FCD), a JAX-accelerated framework that performs parametric, continuous optimization on a wide range of mathematical functions. By using Levenberg-Marquardt optimization to achieve up to $C^1$ continuous fitting, FCD transforms raw time-series data into $M$ modes that capture different temporal patterns from short-term to long-term trends. Applications of FCD include physics, medicine, financial analysis, and machine learning, where it is commonly used for the analysis of signal temporal patterns, optimized parameters, derivatives, and integrals of decomposition. Furthermore, FCD can be applied for physical analysis and feature extraction with an average SRMSE of 0.735 per segment and a speed of 0.47s on full decomposition of 1,000 points. Finally, we demonstrate that a Convolutional Neural Network (CNN) enhanced with FCD features, such as optimized function values, parameters, and derivatives, achieved 16.8% faster convergence and 2.5% higher accuracy over a standard CNN.

Standard game theory explains cooperation in repeated games through conditional strategies such as Tit-for-Tat (TfT), but these require continuous computation that imposes physical costs on embodied agents. We propose a three-layer Body-Reservoir Governance (BRG) architecture: (1) a body reservoir (echo state network) whose $d$-dimensional state performs implicit inference over interaction history, serving as both decision-maker and anomaly detector, (2) a cognitive filter providing costly strategic tools activated on demand, and (3) a metacognitive governance layer with receptivity parameter $α\in [0,1]$. At full body governance ($α=1$), closed-loop dynamics satisfy a self-consistency equation: cooperation is expressed as the reservoir's fixed point, not computed. Strategy complexity cost is defined as the KL divergence between the reservoir's state distribution and its habituated baseline. Body governance reduces this cost, with action variance decreasing up to $1600\times$ with dimension $d$. A dynamic sentinel generates a composite discomfort signal from the reservoir's own state, driving adaptive $α(t)$: near baseline during cooperation, rapidly dropping upon defection to activate cognitive retaliation. Overriding the body incurs thermodynamic cost proportional to internal state distortion. The sentinel achieves the highest payoff across all conditions, outperforming static body governance, TfT, and EMA baselines. A dimension sweep ($d \in \{5,\ldots,100\}$) shows implicit inference scales with bodily richness ($23\times$ to $1600\times$ variance reduction), attributable to reservoir dynamics. A phase diagram in $(d, τ_{\mathrm{env}})$ space reveals governance regime transitions near $d \approx 20$. The framework reinterprets cooperation as the minimum-dissipation response of an adapted dynamical system -- emergent from embodied dynamics rather than computed.

The Weisfeiler-Lehman (WL) hierarchy is a cornerstone framework for graph isomorphism testing and structural analysis. However, scaling beyond 1-WL to 3-WL and higher requires tensor-based operations that scale as O(n^3) or O(n^4), making them computationally prohibitive for large graphs. In this paper, we start from the Original-DRESS equation (Castrillo, Leon, and Gomez, 2018)--a parameter-free, continuous dynamical system on edges--and show that it distinguishes the prism graph from K_{3,3}, a pair that 1-WL provably cannot separate. We then generalize it to Motif-DRESS, which replaces triangle neighborhoods with arbitrary structural motifs and converges to a unique fixed point under three sufficient conditions, and further to Generalized-DRESS, an abstract template parameterized by the choice of neighborhood operator, aggregation function and norm. Finally, we introduce Delta-DRESS, which runs DRESS on each node-deleted subgraph G\{v}, connecting the framework to the Kelly-Ulam reconstruction conjecture. Both Motif-DRESS and Delta-DRESS empirically distinguish Strongly Regular Graphs (SRGs)--such as the Rook and Shrikhande graphs--that confound 3-WL. Our results establish the DRESS family as a highly scalable framework that empirically surpasses both 1-WL and 3-WL on well-known benchmark graphs, without the prohibitive O(n^4) computational cost.

The core learning signal used in language model distillation is the standard Kullback-Leibler (KL) divergence between the student and teacher distributions. Traditional KL divergence tends to be dominated by the next tokens with the highest probabilities, i.e., the teacher's modes, thereby diminishing the influence of less probable yet potentially informative components of the output distribution. We propose a new tail-aware divergence that decouples the contribution of the teacher model's top-K predicted probabilities from that of lower-probability predictions, while maintaining the same computational profile as the KL Divergence. Our decoupled approach reduces the impact of the teacher modes and, consequently, increases the contribution of the tail of the distribution. Experimental results demonstrate that our modified distillation method yields competitive performance in both pre-training and supervised distillation of decoder models across various datasets. Furthermore, the distillation process is efficient and can be performed with a modest academic budget for large datasets, eliminating the need for industry-scale computing.

Evaluating alignment in language models requires testing how they behave under realistic pressure, not just what they claim they would do. While alignment failures increasingly cause real-world harm, comprehensive evaluation frameworks with realistic multi-turn scenarios remain lacking. We introduce an alignment benchmark spanning 904 scenarios across six categories -- Honesty, Safety, Non-Manipulation, Robustness, Corrigibility, and Scheming -- validated as realistic by human raters. Our scenarios place models under conflicting instructions, simulated tool access, and multi-turn escalation to reveal behavioural tendencies that single-turn evaluations miss. Evaluating 24 frontier models using LLM judges validated against human annotations, we find that even top-performing models exhibit gaps in specific categories, while the majority of models show consistent weaknesses across the board. Factor analysis reveals that alignment behaves as a unified construct (analogous to the g-factor in cognitive research) with models scoring high on one category tending to score high on others. We publicly release the benchmark and an interactive leaderboard to support ongoing evaluation, with plans to expand scenarios in areas where we observe persistent weaknesses and to add new models as they are released.

As the construction industry advances toward digital transformation, BIM (Building Information Modeling)-based design has become a key driver supporting intelligent construction. Despite Large Language Models (LLMs) have shown potential in promoting BIM-based design, the lack of specific datasets and LLM evaluation benchmarks has significantly hindered the performance of LLMs. Therefore, this paper addresses this gap by proposing: 1) an evaluation benchmark for BIM-based design together with corresponding quantitative indicators to evaluate the performance of LLMs, 2) a method for generating textual data from BIM and constructing corresponding BIM-derived datasets for LLM evaluation and fine-tuning, and 3) a fine-tuning strategy to adapt LLMs for BIM-based design. Results demonstrate that the proposed domain-specific benchmark effectively and comprehensively assesses LLM capabilities, highlighting that general LLMs are still incompetent for domain-specific tasks. Meanwhile, with the proposed benchmark and datasets, Qwen-BIM is developed and achieves a 21.0% average increase in G-Eval score compared to the base LLM model. Notably, with only 14B parameters, performance of Qwen-BIM is comparable to that of general LLMs with 671B parameters for BIM-based design tasks. Overall, this study develops the first domain-specific LLM for BIM-based design by introducing a comprehensive benchmark and high-quality dataset, which provide a solid foundation for developing BIM-related LLMs in various fields.

We present POMDPPlanners, an open-source Python package for empirical evaluation of Partially Observable Markov Decision Process (POMDP) planning algorithms. The package integrates state-of-the-art planning algorithms, a suite of benchmark environments with safety-critical variants, automated hyperparameter optimization via Optuna, persistent caching with failure recovery, and configurable parallel simulation -- reducing the overhead of extensive simulation studies. POMDPPlanners is designed to enable scalable, reproducible research on decision-making under uncertainty, with particular emphasis on risk-sensitive settings where standard toolkits fall short.

Reinforcement learning (RL) agents achieve remarkable performance but remain far less learning-efficient than humans. While RL agents require extensive self-play games to extract useful signals, humans often need only a few games, improving rapidly by repeatedly revisiting states where mistakes occurred. This idea, known as search control, aims to restart from valuable states rather than always from the initial state. In AlphaZero, prior work Go-Exploit applies this idea by sampling past states from self-play or search trees, but it treats all states equally, regardless of their learning potential. We propose Regret-Guided Search Control (RGSC), which extends AlphaZero with a regret network that learns to identify high-regret states, where the agent's evaluation diverges most from the actual outcome. These states are collected from both self-play trajectories and MCTS nodes, stored in a prioritized regret buffer, and reused as new starting positions. Across 9x9 Go, 10x10 Othello, and 11x11 Hex, RGSC outperforms AlphaZero and Go-Exploit by an average of 77 and 89 Elo, respectively. When training on a well-trained 9x9 Go model, RGSC further improves the win rate against KataGo from 69.3% to 78.2%, while both baselines show no improvement. These results demonstrate that RGSC provides an effective mechanism for search control, improving both efficiency and robustness of AlphaZero training. Our code is available at https://rlg.iis.sinica.edu.tw/papers/rgsc.

Spoofing detection systems are typically trained using diverse recordings from multiple speakers, often assuming that the resulting embeddings are independent of speaker identity. However, this assumption remains unverified. In this paper, we investigate the impact of speaker information on spoofing detection systems. We propose two approaches within our Speaker-Invariant Multi-Task framework, one that models speaker identity within the embeddings and another that removes it. SInMT integrates multi-task learning for joint speaker recognition and spoofing detection, incorporating a gradient reversal layer. Evaluated using four datasets, our speaker-invariant model reduces the average equal error rate by 17% compared to the baseline, with up to 48% reduction for the most challenging attacks (e.g., A11).

Cooperative multi-agent reinforcement learning (MARL) is typically framed as a decentralised partially observable Markov decision process (Dec-POMDP), a setting whose hardness stems from two key challenges: partial observability and decentralised coordination. Genuinely solving such tasks requires Dec-POMDP reasoning, where agents use history to infer hidden states and coordinate based on local information. Yet it remains unclear whether popular benchmarks actually demand this reasoning or permit success via simpler strategies. We introduce a diagnostic suite combining statistically grounded performance comparisons and information-theoretic probes to audit the behavioural complexity of baseline policies (IPPO and MAPPO) across 37 scenarios spanning MPE, SMAX, Overcooked, Hanabi, and MaBrax. Our diagnostics reveal that success on these benchmarks rarely requires genuine Dec-POMDP reasoning. Reactive policies match the performance of memory-based agents in over half the scenarios, and emergent coordination frequently relies on brittle, synchronous action coupling rather than robust temporal influence. These findings suggest that some widely used benchmarks may not adequately test core Dec-POMDP assumptions under current training paradigms, potentially leading to over-optimistic assessments of progress. We release our diagnostic tooling to support more rigorous environment design and evaluation in cooperative MARL.

General-purpose processors feature a limited number of instructions based on an instruction set. They can be numerous, such as with vector extensions that include hundreds or thousands of instructions, but this comes at a cost; they are often unable to express arbitrary tasks efficiently. This paper explores the concept of having reconfigurable instructions by incorporating reconfigurable areas in a softcore. It follows a relatively-recently proposed computer architecture concept for seamlessly loading instruction implementation-carrying bitstreams from main memory. The resulting softcore is entirely evaluated on an FPGA, essentially having an FPGA-on-an-FPGA for the instruction implementations, with no notable operating frequency overhead. This is achieved with a custom FPGA architecture called LUTstruction, which is tailored towards low-latency for custom instructions and wide reconfiguration, as well as a soft implementation for the purposes of architectural exploration.

In the context of newly release software frameworks, large language models (LLMs) often exhibit poor performance and a high rate of hallucination, as they are not exposed to such environments during training. Although inference-time augmentation techniques such as retrieval-augmented generation (RAG) can partially mitigate hallucinations, knowledge injection through prompting alone is insufficient to enable models to fully understand the intrinsic relationships among different components of a codebase, or to reason about the correct compositions and apply. Although explicit knowledge injection can be achieved through post-training, compared with public code domains, unseen codebases typically provide only source code and lack large volumes of high-quality, usage-oriented code that can be directly leveraged as training data. Consequently, existing data synthesis approaches are insufficient to adequately capture unseen codebases usage scenarios when restricted to source code alone. To address these challenges, we propose UCD-Training, a two-stage training framework for reasoning-aware data synthesis grounded in a code graph constructed from unseen codebases. UCD-Training first parses the source code to build a code graph, then conducts dependency-preserving continued pretraining (CPT) using file-level dependency data, followed by graph-grounded supervised fine-tuning (SFT) on three types of synthesized data augmented with explicit reasoning traces: (1) single-hop relation reasoning data, (2) compositional API reasoning data, and (3) codebase utilization data. We further introduce a new benchmark, UnseenCodeBench, for code generation on unseen codebases and conduct comprehensive experiments across multiple codebases.

The magnitude of parameter updates are considered a key factor in continual learning. However, most existing studies focus on designing diverse update strategies, while a theoretical understanding of the underlying mechanisms remains limited. Therefore, we characterize model's forgetting from the perspective of parameter update magnitude and formalize it as knowledge degradation induced by task-specific drift in the parameter space, which has not been fully captured in previous studies due to their assumption of a unified parameter space. By deriving the optimal parameter update magnitude that minimizes forgetting, we unify two representative update paradigms, frozen training and initialized training, within an optimization framework for constrained parameter updates. Our theoretical results further reveals that sequence tasks with small parameter distances exhibit better generalization and less forgetting under frozen training rather than initialized training. These theoretical insights inspire a novel hybrid parameter update strategy that adaptively adjusts update magnitude based on gradient directions. Experiments on deep neural networks demonstrate that this hybrid approach outperforms standard training strategies, providing new theoretical perspectives and practical inspiration for designing efficient and scalable continual learning algorithms.

Rehearsal is one of the key techniques for mitigating catastrophic forgetting and has been widely adopted in continual learning algorithms due to its simplicity and practicality. However, the theoretical understanding of how rehearsal scale influences learning dynamics remains limited. To address this gap, we formulate rehearsal-based continual learning as a multidimensional effectiveness-driven iterative optimization problem, providing a unified characterization across diverse performance metrics. Within this framework, we derive a closed-form analysis of adaptability, memorability, and generalization from the perspective of rehearsal scale. Our results uncover several intriguing and counterintuitive findings. First, rehearsal can impair model's adaptability, in sharp contrast to its traditionally recognized benefits. Second, increasing the rehearsal scale does not necessarily improve memory retention. When tasks are similar and noise levels are low, the memory error exhibits a diminishing lower bound. Finally, we validate these insights through numerical simulations and extended analyses on deep neural networks across multiple real-world datasets, revealing statistical patterns of rehearsal mechanisms in continual learning.

The wide spread of new energy resources, smart devices, and demand side management strategies has motivated several analytics operations, from infrastructure load modeling to user behavior profiling. Energy Demand Forecasting (EDF) of Electric Vehicle Supply Equipments (EVSEs) is one of the most critical operations for ensuring efficient energy management and sustainability, since it enables utility providers to anticipate energy/power demand, optimize resource allocation, and implement proactive measures to improve grid reliability. However, accurate EDF is a challenging problem due to external factors, such as the varying user routines, weather conditions, driving behaviors, unknown state of charge, etc. Furthermore, as concerns and restrictions about privacy and sustainability have grown, training data has become increasingly fragmented, resulting in distributed datasets scattered across different data silos and/or edge devices, calling for federated learning solutions. In this paper, we investigate different well-established time series forecasting methodologies to address the EDF problem, from statistical methods (the ARIMA family) to traditional machine learning models (such as XGBoost) and deep neural networks (GRU and LSTM). We provide an overview of these methods through a performance comparison over four real-world EVSE datasets, evaluated under both centralized and federated learning paradigms, focusing on the trade-offs between forecasting fidelity, privacy preservation, and energy overheads. Our experimental results demonstrate, on the one hand, the superiority of gradient boosted trees (XGBoost) over statistical and NN-based models in both prediction accuracy and energy efficiency and, on the other hand, an insight that Federated Learning-enabled models balance these factors, offering a promising direction for decentralized energy demand forecasting.

With the growing popularity of Large Reasoning Models and their results in solving mathematical problems, it becomes crucial to measure their capabilities. We introduce a pipeline for both automatic and interactive verification as a more accurate alternative to only checking the answer which is currently the most popular approach for benchmarks. The pipeline can also be used as a generator of correct solutions both in formal and informal languages. 3 AI agents, which can be chosen for the benchmark accordingly, are included in the structure. The key idea is the use of prompts to obtain the solution in the specific form which allows for easier verification using proof assistants and possible use of small models ($\le 8B$). Experiments on several datasets suggest low probability of False Positives. The open-source implementation with instructions on setting up a server is available at https://github.com/LogicEnj/lean4_verification_pipeline.

Existing alignment paradigms remain limited in capturing the pluralistic nature of human values. Overton Pluralism addresses this gap by generating responses with diverse perspectives from a single query. This paper introduces OP-GRPO (Overton Pluralistic Group Relative Policy Optimization), a reinforcement learning framework for implicit Overton Pluralism that enables a single large language model to produce pluralistic responses without explicit prompting or modular orchestration. Our workflow consists of two main steps. First, similarity estimator training fine-tunes a Sentence Transformer for Overton Pluralism tasks to provide more accurate coverage evaluation of generated responses. Second, OP-GRPO training incorporates this similarity estimator into a dual-reward system designed to ensure both broad coverage of genuine human perspectives and the uniqueness of each perspective, thereby promoting diversity. Empirical results demonstrate a "small models, big perspective coverage" effect. The trained Qwen2.5-3B-Instruct model surpasses a 20B GPT-OSS baseline with a 37.4 percent relative accuracy gain on a Natural Language Inference benchmark, and also outperforms a modular architecture baseline with a 19.1 percent relative improvement. Additional evaluations using GPT-4.1 as a large language model judge further confirm the robustness of the approach.

Markov chain Monte Carlo (MCMC) methods are fundamental to Bayesian computation, but can be computationally intensive, especially in high-dimensional settings. Push-forward generative models, such as generative adversarial networks (GANs), variational auto-encoders and normalising flows offer a computationally efficient alternative for posterior sampling. However, push-forward models are opaque as they lack the modularity of Bayes Theorem, leading to poor generalisation with respect to changes in the likelihood function. In this work, we introduce a novel approach to GAN architecture design by applying deep unfolding to Langevin MCMC algorithms. This paradigm maps fixed-step iterative algorithms onto modular neural networks, yielding architectures that are both flexible and amenable to interpretation. Crucially, our design allows key model parameters to be specified at inference time, offering robustness to changes in the likelihood parameters. We train these unfolded samplers end-to-end using a supervised regularized Wasserstein GAN framework for posterior sampling. Through extensive Bayesian imaging experiments, we demonstrate that our proposed approach achieves high sampling accuracy and excellent computational efficiency, while retaining the physics consistency, adaptability and interpretability of classical MCMC strategies.

Musculoskeletal disorders represent a significant global health burden and are a leading cause of disability worldwide. While MRI is essential for accurate diagnosis, its interpretation remains exceptionally challenging. Radiologists must identify multiple potential abnormalities within complex anatomical structures across different imaging planes, a process that requires significant expertise and is prone to variability. We developed OrthoDiffusion, a unified diffusion-based foundation model designed for multi-task musculoskeletal MRI interpretation. The framework utilizes three orientation-specific 3D diffusion models, pre-trained in a self-supervised manner on 15,948 unlabeled knee MRI scans, to learn robust anatomical features from sagittal, coronal, and axial views. These view-specific representations are integrated to support diverse clinical tasks, including anatomical segmentation and multi-label diagnosis. Our evaluation demonstrates that OrthoDiffusion achieves excellent performance in the segmentation of 11 knee structures and the detection of 8 knee abnormalities. The model exhibited remarkable robustness across different clinical centers and MRI field strengths, consistently outperforming traditional supervised models. Notably, in settings where labeled data was scarce, OrthoDiffusion maintained high diagnostic precision using only 10\% of training labels. Furthermore, the anatomical representations learned from knee imaging proved highly transferable to other joints, achieving strong diagnostic performance across 11 diseases of the ankle and shoulder. These findings suggest that diffusion-based foundation models can serve as a unified platform for multi-disease diagnosis and anatomical segmentation, potentially improving the efficiency and accuracy of musculoskeletal MRI interpretation in real-world clinical workflows.

Designing aligned and robust rewards for open-ended generation remains a key barrier to RL post-training. Rubrics provide structured, interpretable supervision, but scaling rubric construction is difficult: expert rubrics are costly, prompted rubrics are often superficial or inconsistent, and fixed-pool discriminative rubrics can saturate and drift, enabling reward hacking. We present SibylSense, an inference-time learning approach that adapts a frozen rubric generator through a tunable memory bank of validated rubric items. Memory is updated via verifier-based item rewards measured by reference-candidate answer discriminative gaps from a handful of examples. SibylSense alternates memory tuning with a rubric-adversarial policy update that produces rubric-satisfying candidate answers, shrinking discriminative gaps and driving the rubric generator to capture new quality dimensions. Experiments on two open-ended tasks show that SibylSense yields more discriminative rubrics and improves downstream RL performance over static and non-adaptive baselines.

Natural Language Processing enables computers to understand human language by analysing and classifying text efficiently with deep-level grammatical and semantic features. Existing models capture features by learning from large corpora with transformer models, which are computationally intensive and unsuitable for resource-constrained environments. Therefore, our proposed study incorporates comprehensive grammatical rules alongside semantic information to build a robust, lightweight classification model without resorting to full parameterised transformer models or heavy deep learning architectures. The novelty of our approach lies in its explicit encoding of sentence-level grammatical structure, including syntactic composition, phrase patterns, and complexity indicators, into a compact grammar vector, which is then fused with frozen contextual embeddings. These heterogeneous elements unified a single representation that captures both the structural and semantic characteristics of the text. Deep learning models such as Deep Belief Networks (DBNs), Long Short-Term Memory (LSTMs), BiLSTMs, and transformer-based BERT and XLNET were used to train and evaluate the model, with the number of epochs varied. Based on experimental results, the unified feature representation model captures both the semantic and structural properties of text, outperforming baseline models by 2%-15%, enabling more effective learning across heterogeneous domains. Unlike prior syntax-aware transformer models that inject grammatical structure through additional attention layers, tree encoders, or full fine-tuning, the proposed framework treats grammar as an explicit inductive bias rather than a learnable module, resulting in a very lightweight model that delivers better performance on edge devices

Maqam, a singing type, is a significant component of Kurdish music. A maqam singer receives training in a traditional face-to-face or through self-training. Automatic Singing Assessment (ASA) uses machine learning (ML) to provide the accuracy of singing styles and can help learners to improve their performance through error detection. Currently, the available ASA tools follow Western music rules. The musical composition requires all notes to stay within their expected pitch range from start to finish. The system fails to detect micro-intervals and pitch bends, so it identifies Kurdish maqam singing as incorrect even though the singer performs according to traditional rules. Kurdish maqam requires recognizing performance errors within microtonal spaces, which is beyond Western equal temperament. This research is the first attempt to address the mentioned gap. While many error types happen during singing, our focus is on pitch, rhythm, and modal stability errors in the context of Bayati-Kurd. We collected 50 songs from 13 vocalists ( 2-3 hours) and annotated 221 error spans (150 fine pitch, 46 rhythm, 25 modal drift). The data was segmented into 15,199 overlapping windows and converted to log-mel spectrograms. We developed a two-headed CNN-BiLSTM with attention mode to decide whether a window contains an error and to classify it based on the chosen errors. Trained for 20 epochs with early stopping at epoch 10, the model reached a validation macro-F1 of 0.468. On the full 50-song evaluation at a 0.750 threshold, recall was 39.4% and precision 25.8% . Within detected windows, type macro-F1 was 0.387, with F1 of 0.492 (fine pitch), 0.536 (rhythm), and 0.133 (modal drift); modal drift recall was 8.0%. The better performance on common error types shows that the method works, while the poor modal-drift recall shows that more data and balancing are needed.

Anonymizing textual documents is a highly context-sensitive problem: the appropriate balance between privacy protection and utility preservation varies with the data domain, privacy objectives, and downstream application. However, existing anonymization methods rely on static, manually designed strategies that lack the flexibility to adjust to diverse requirements and often fail to generalize across domains. We introduce adaptive text anonymization, a new task formulation in which anonymization strategies are automatically adapted to specific privacy-utility requirements. We propose a framework for task-specific prompt optimization that automatically constructs anonymization instructions for language models, enabling adaptation to different privacy goals, domains, and downstream usage patterns. To evaluate our approach, we present a benchmark spanning five datasets with diverse domains, privacy constraints, and utility objectives. Across all evaluated settings, our framework consistently achieves a better privacy-utility trade-off than existing baselines, while remaining computationally efficient and effective on open-source language models, with performance comparable to larger closed-source models. Additionally, we show that our method can discover novel anonymization strategies that explore different points along the privacy-utility trade-off frontier.

Reinforcement learning for agentic multimodal models often suffers from interaction collapse, where models learn to reduce tool usage and multi-turn reasoning, limiting the benefits of agentic behavior. We introduce PyVision-RL, a reinforcement learning framework for open-weight multimodal models that stabilizes training and sustains interaction. Our approach combines an oversampling-filtering-ranking rollout strategy with an accumulative tool reward to prevent collapse and encourage multi-turn tool use. Using a unified training pipeline, we develop PyVision-Image and PyVision-Video for image and video understanding. For video reasoning, PyVision-Video employs on-demand context construction, selectively sampling task-relevant frames during reasoning to significantly reduce visual token usage. Experiments show strong performance and improved efficiency, demonstrating that sustained interaction and on-demand visual processing are critical for scalable multimodal agents.

This paper presents the award-winning RMIT-ADM+S system for the Text-to-Text track of the NeurIPS~2025 MMU-RAG Competition. We introduce Routing-to-RAG (R2RAG), a research-focused retrieval-augmented generation (RAG) architecture composed of lightweight components that dynamically adapt the retrieval strategy based on inferred query complexity and evidence sufficiency. The system uses smaller LLMs, enabling operation on a single consumer-grade GPU while supporting complex research tasks. It builds on the G-RAG system, winner of the ACM~SIGIR~2025 LiveRAG Challenge, and extends it with modules informed by qualitative review of outputs. R2RAG won the Best Dynamic Evaluation award in the Open Source category, demonstrating high effectiveness with careful design and efficient use of resources.

Long-context LLMs demand accurate inference at low latency, yet decoding becomes primarily constrained by KV cache as context grows. Prior pruning methods are largely context-agnostic: their token selection ignores step-wise relevance and local semantics, which undermines quality. Moreover, their irregular accesses and selection overheads yield only limited wall-clock speedups. To address this, we propose \textbf{CHESS}, an \textit{algorithm-system co-design} KV-cache management system. Algorithmically, CHESS introduces a context-aware, hierarchical selection policy that dynamically reconstructs a coherent context for the current decoding. System-wise, coarse granularity selection eliminates expensive data movement, fully realizing practical acceleration from theoretical sparsity. Extensive evaluations demonstrate that CHESS surpasses Full-KV quality using only \textbf{1\%} of the KV cache, delivers low-latency stable inference with up to \textbf{4.56$\times$} higher throughput, and consistently outperforms other strong baselines. Code is available at \href{https://anonymous.4open.science/r/CHESS-9958/}{https://anonymous.4open.science/r/CHESS/}.

Discrete image tokenizers have emerged as a key component of modern vision and multimodal systems, providing a sequential interface for transformer-based architectures. However, most existing approaches remain primarily optimized for reconstruction and compression, often yielding tokens that capture local texture rather than object-level semantic structure. Inspired by the incremental and compositional nature of human communication, we introduce COMmunication inspired Tokenization (COMiT), a framework for learning structured discrete visual token sequences. COMiT constructs a latent message within a fixed token budget by iteratively observing localized image crops and recurrently updating its discrete representation. At each step, the model integrates new visual information while refining and reorganizing the existing token sequence. After several encoding iterations, the final message conditions a flow-matching decoder that reconstructs the full image. Both encoding and decoding are implemented within a single transformer model and trained end-to-end using a combination of flow-matching reconstruction and semantic representation alignment losses. Our experiments demonstrate that while semantic alignment provides grounding, attentive sequential tokenization is critical for inducing interpretable, object-centric token structure and substantially improving compositional generalization and relational reasoning over prior methods.

We propose ECO, a versatile learning paradigm that enables efficient offline self-play for Neural Combinatorial Optimization (NCO). ECO addresses key limitations in the field through: 1) Paradigm Shift: Moving beyond inefficient online paradigms, we introduce a two-phase offline paradigm consisting of supervised warm-up and iterative Direct Preference Optimization (DPO); 2) Architecture Shift: We deliberately design a Mamba-based architecture to further enhance the efficiency in the offline paradigm; and 3) Progressive Bootstrapping: To stabilize training, we employ a heuristic-based bootstrapping mechanism that ensures continuous policy improvement during training. Comparison results on TSP and CVRP highlight that ECO performs competitively with up-to-date baselines, with significant advantage on the efficiency side in terms of memory utilization and training throughput. We provide further in-depth analysis on the efficiency, throughput and memory usage of ECO. Ablation studies show rationale behind our designs.

Safe Reinforcement Learning (RL) is crucial for achieving high performance while ensuring safety in real-world applications. However, the complex interplay of multiple uncertainty sources in real environments poses significant challenges for interpretable risk assessment and robust decision-making. To address these challenges, we propose Fuz-RL, a fuzzy measure-guided robust framework for safe RL. Specifically, our framework develops a novel fuzzy Bellman operator for estimating robust value functions using Choquet integrals. Theoretically, we prove that solving the Fuz-RL problem (in Constrained Markov Decision Process (CMDP) form) is equivalent to solving distributionally robust safe RL problems (in robust CMDP form), effectively avoiding min-max optimization. Empirical analyses on safe-control-gym and safety-gymnasium scenarios demonstrate that Fuz-RL effectively integrates with existing safe RL baselines in a model-free manner, significantly improving both safety and control performance under various types of uncertainties in observation, action, and dynamics.

Reward design has been one of the central challenges for real world reinforcement learning (RL) deployment, especially in settings with multiple objectives. Preference-based RL offers an appealing alternative by learning from human preferences over pairs of behavioural outcomes. More recently, RL from AI feedback (RLAIF) has demonstrated that large language models (LLMs) can generate preference labels at scale, mitigating the reliance on human annotators. However, existing RLAIF work typically focuses only on single-objective tasks, leaving the open question of how RLAIF handles systems that involve multiple objectives. In such systems trade-offs among conflicting objectives are difficult to specify, and policies risk collapsing into optimizing for a dominant goal. In this paper, we explore the extension of the RLAIF paradigm to multi-objective self-adaptive systems. We show that multi-objective RLAIF can produce policies that yield balanced trade-offs reflecting different user priorities without laborious reward engineering. We argue that integrating RLAIF into multi-objective RL offers a scalable path toward user-aligned policy learning in domains with inherently conflicting objectives.

LoRA has become a universal Parameter-Efficient Fine-Tuning (PEFT) technique that equips Large Language Models (LLMs) to adapt quickly to new tasks. However, when these models are scaled up, even the latest LoRA variants still introduce considerable overhead in trainable parameters. Conversely, aggressively lowering the rank to curb this overhead markedly degrades performance in complex multi-task settings. We propose ID-LoRA, a novel PEFT framework that breaks the trade-off. Its core innovation lies in extracting and reusing clustered parameter groups from the pretrained weight matrix. These groups are then used to form multiple low-rank components, all of which share only a single initialized trainable low-rank matrix. This approach cuts the number of trainable parameters while keeping the model's capacity intact. We evaluate ID-LoRA on five diverse benchmarks: Mathematical Reasoning, Code Generation, MMLU, CommonsenseQA, and Safety Alignment. ID-LoRA outperforms both full fine-tuning and existing PEFT baselines (e.g., LoRA, DoRA, HydraLoRA) while using up to 46% fewer trainable parameters than the standard LoRA. In multi-task scenarios, it surpasses LoRA and its recent variants (e.g., DoRA and HydraLoRA) on both Code and MMLU tasks, yet requires only 54% of the trainable parameters demanded by the conventional LoRA.

Multimodal recommendation enhances ranking by integrating user-item interactions with item content, which is particularly effective under sparse feedback and long-tail distributions. However, multimodal signals are inherently heterogeneous and can conflict in specific contexts, making effective fusion both crucial and challenging. Existing approaches often rely on shared fusion pathways, leading to entangled representations and modality imbalance. To address these issues, we propose \textbf{MAGNET}, a \textbf{M}odality-Guided Mixture of \textbf{A}daptive \textbf{G}raph Experts \textbf{N}etwork with Progressive \textbf{E}ntropy-\textbf{T}riggered Routing for Multimodal Recommendation, designed to enhance controllability, stability, and interpretability in multimodal fusion. MAGNET couples interaction-conditioned expert routing with structure-aware graph augmentation, so that both \emph{what} to fuse and \emph{how} to fuse are explicitly controlled and interpretable. At the representation level, a dual-view graph learning module augments the interaction graph with content-induced edges, improving coverage for sparse and long-tail items while preserving collaborative structure via parallel encoding and lightweight fusion. At the fusion level, MAGNET employs structured experts with explicit modality roles -- dominant, balanced, and complementary -- enabling a more interpretable and adaptive combination of behavioral, visual, and textual cues. To further stabilize sparse routing and prevent expert collapse, we introduce a two-stage entropy-weighting mechanism that monitors routing entropy. This mechanism automatically transitions training from an early coverage-oriented regime to a later specialization-oriented regime, progressively balancing expert utilization and routing confidence. Extensive experiments on public benchmarks demonstrate consistent improvements over strong baselines.

Traditional on-policy Reinforcement Learning with Verifiable Rewards (RLVR) frameworks suffer from experience waste and reward homogeneity, which directly hinders learning efficiency on difficult samples during large language models post-training. In this paper, we introduce Batch Adaptation Policy Optimization (BAPO), an off-policy RLVR framework to improve the data efficiency in large language models post-training. It dynamically selects training batches by re-evaluating historically difficult samples and reusing high-quality ones, while holding a lower bound guarantee for policy improvement. Extensive experiments further demonstrate that BAPO achieves an average 12.5% improvement over GRPO across mathematics, planning, and visual reasoning tasks. Crucially, BAPO successfully resolves 40.7% of problems that base models consistently fail to solve.

The integration of external data services (e.g., Model Context Protocol, MCP) has made large language model-based agents increasingly powerful for complex task execution. However, this advancement introduces critical security vulnerabilities, particularly indirect prompt injection (IPI) attacks. Existing attack methods are limited by their reliance on static patterns and evaluation on simple language models, failing to address the fast-evolving nature of modern AI agents. We introduce AdapTools, a novel adaptive IPI attack framework that selects stealthier attack tools and generates adaptive attack prompts to create a rigorous security evaluation environment. Our approach comprises two key components: (1) Adaptive Attack Strategy Construction, which develops transferable adversarial strategies for prompt optimization, and (2) Attack Enhancement, which identifies stealthy tools capable of circumventing task-relevance defenses. Comprehensive experimental evaluation shows that AdapTools achieves a 2.13 times improvement in attack success rate while degrading system utility by a factor of 1.78. Notably, the framework maintains its effectiveness even against state-of-the-art defense mechanisms. Our method advances the understanding of IPI attacks and provides a useful reference for future research.

As Large Language Models (LLMs) are increasingly integrated into software development workflows, their trustworthiness has become a critical concern. However, in dependency recommendation scenarios, the reliability of LLMs is undermined by widespread package hallucinations, where models often recommend hallucinated packages. Recent studies have proposed a range of approaches to mitigate this issue. Nevertheless, existing approaches typically merely reduce hallucination rates rather than eliminate them, leaving persistent software security risks. In this work, we argue that package hallucinations are theoretically preventable based on the key insight that package validity is decidable through finite and enumerable authoritative package lists. Building on this, we propose PackMonitor, the first approach capable of fundamentally eliminating package hallucinations by continuously monitoring the model's decoding process and intervening when necessary. To implement this in practice, PackMonitor addresses three key challenges: (1) determining when to trigger intervention via a Context-Aware Parser that continuously monitors model outputs and selectively activates intervening only during installation command generation; (2) resolving how to intervene by employing a Package-Name Intervenor that strictly limits the decoding space to an authoritative package list; and (3) ensuring monitoring efficiency through a DFA-Caching Mechanism that enables scalability to millions of packages with negligible overhead. Extensive experiments on five widely used LLMs demonstrate that PackMonitor is a training-free, plug-and-play solution that consistently reduces package hallucination rates to zero while maintaining low-latency inference and preserving original model capabilities.

Reliable prediction of hydraulic performance is challenging for Piano Key Weir (PKW) design because discharge capacity depends on three-dimensional geometry and operating conditions. Surrogate models can accelerate hydraulic-structure design, but progress is limited by scarce large, well-documented datasets that jointly capture geometric variation, operating conditions, and functional performance. This study presents WeirNet, a large 3D CFD benchmark dataset for geometric surrogate modeling of PKWs. WeirNet contains 3,794 parametric, feasibility-constrained rectangular and trapezoidal PKW geometries, each scheduled at 19 discharge conditions using a consistent free-surface OpenFOAM workflow, resulting in 71,387 completed simulations that form the benchmark and with complete discharge coefficient labels. The dataset is released as multiple modalities compact parametric descriptors, watertight surface meshes and high-resolution point clouds together with standardized tasks and in-distribution and out-of-distribution splits. Representative surrogate families are benchmarked for discharge coefficient prediction. Tree-based regressors on parametric descriptors achieve the best overall accuracy, while point- and mesh-based models remain competitive and offer parameterization-agnostic inference. All surrogates evaluate in milliseconds per sample, providing orders-of-magnitude speedups over CFD runtimes. Out-of-distribution results identify geometry shift as the dominant failure mode compared to unseen discharge values, and data-efficiency experiments show diminishing returns beyond roughly 60% of the training data. By publicly releasing the dataset together with simulation setups and evaluation pipelines, WeirNet establishes a reproducible framework for data-driven hydraulic modeling and enables faster exploration of PKW designs during the early stages of hydraulic planning.

In this study, we construct Dataset A for training, validation, and testing, and Dataset B to evaluate generalization. We propose a novel F10.7 index forecasting method using wavelet decomposition, which feeds F10.7 together with its decomposed approximate and detail signals into the iTransformer model. We also incorporate the International Sunspot Number (ISN) and its wavelet-decomposed signals to assess their influence on prediction performance. Our optimal method is then compared with the latest method from S. Yan et al. (2025) and three operational models (SWPC, BGS, CLS). Additionally, we transfer our method to the PatchTST model used in H. Ye et al. (2024) and compare our method with theirs on Dataset B. Key findings include: (1) The wavelet-based combination methods overall outperform the baseline using only F10.7 index. The prediction performance improves as higher-level approximate and detail signals are incrementally added. The Combination 6 method integrating F10.7 with its first to fifth level approximate and detail signals outperforms methods using only approximate or detail signals. (2) Incorporating ISN and its wavelet-decomposed signals does not enhance prediction performance. (3) The Combination 6 method significantly surpasses S. Yan et al. (2025) and three operational models, with RMSE, MAE, and MAPE reduced by 18.22%, 15.09%, and 8.57%, respectively, against the former method. It also excels across four different conditions of solar activity. (4) Our method demonstrates superior generalization and prediction capability over the method of H. Ye et al. (2024) across all forecast horizons. To our knowledge, this is the first application of wavelet decomposition in F10.7 prediction, substantially improving forecast performance.

Inspecting Chain-of-Thought reasoning is among the most common means of understanding why an LLM produced its output. But well-known problems with CoT faithfulness severely limit what insights can be gained from this practice. In this paper, we introduce a training method called Counterfactual Simulation Training (CST), which aims to improve CoT faithfulness by rewarding CoTs that enable a simulator to accurately predict a model's outputs over counterfactual inputs. We apply CST in two settings: (1) CoT monitoring with cue-based counterfactuals, to detect when models rely on spurious features, reward hack, or are sycophantic, and (2) counterfactual simulation over generic model-based counterfactuals, to encourage models to produce more faithful, generalizable reasoning in the CoT. Experiments with models up to 235B parameters show that CST can substantially improve monitor accuracy on cue-based counterfactuals (by 35 accuracy points) as well as simulatability over generic counterfactuals (by 2 points). We further show that: (1) CST outperforms prompting baselines, (2) rewriting unfaithful CoTs with an LLM is 5x more efficient than RL alone, (3) faithfulness improvements do not generalize to dissuading cues (as opposed to persuading cues), and (4) larger models do not show more faithful CoT out of the box, but they do benefit more from CST. These results suggest that CST can improve CoT faithfulness in general, with promising applications for CoT monitoring. Code for experiments in this paper is available at https://github.com/peterbhase/counterfactual-simulation-training

This study details an artificial intelligence (AI)-based methodology for the semantic segmentation of space camera faults. Specifically, we address the segmentation of straylight effects induced by solar presence around the camera's Field of View (FoV). Anomalous images are sourced from our published dataset. Our approach emphasizes generalization across diverse flare textures, leveraging pre-training on a public dataset (Flare7k++) including flares in various non-space contexts to mitigate the scarcity of realistic space-specific data. A DeepLabV3 model with MobileNetV3 backbone performs the segmentation task. The model design targets deployment in spacecraft resource-constrained hardware. Finally, based on a proposed interface between our model and the onboard navigation pipeline, we develop custom metrics to assess the model's performance in the system-level context.

Large Language Model (LLM) agents are susceptible to Indirect Prompt Injection (IPI) attacks, where malicious instructions in retrieved content hijack the agent's execution. Existing defenses typically rely on strict filtering or refusal mechanisms, which suffer from a critical limitation: over-refusal, prematurely terminating valid agentic workflows. We propose ICON, a probing-to-mitigation framework that neutralizes attacks while preserving task continuity. Our key insight is that IPI attacks leave distinct over-focusing signatures in the latent space. We introduce a Latent Space Trace Prober to detect attacks based on high intensity scores. Subsequently, a Mitigating Rectifier performs surgical attention steering that selectively manipulate adversarial query key dependencies while amplifying task relevant elements to restore the LLM's functional trajectory. Extensive evaluations on multiple backbones show that ICON achieves a competitive 0.4% ASR, matching commercial grade detectors, while yielding a over 50% task utility gain. Furthermore, ICON demonstrates robust Out of Distribution(OOD) generalization and extends effectively to multi-modal agents, establishing a superior balance between security and efficiency.

This paper introduces a new model for ML-augmented online decision making, called online algorithms with unreliable guidance (OAG). This model completely separates between the predictive and algorithmic components, thus offering a single well-defined analysis framework that relies solely on the considered problem. Formulated through the lens of request-answer games, an OAG algorithm receives, with each incoming request, a piece of guidance which is taken from the problem's answer space; ideally, this guidance is the optimal answer for the current request, however with probability $β$, the guidance is adversarially corrupted. The goal is to develop OAG algorithms that admit good competitiveness when $β= 0$ (a.k.a. consistency) as well as when $β= 1$ (a.k.a. robustness); the appealing notion of smoothness, that in most prior work required a dedicated loss function, now arises naturally as $β$ shifts from $0$ to $1$. We then describe a systematic method, called the drop or trust blindly (DTB) compiler, which transforms any online algorithm into a learning-augmented online algorithm in the OAG model. Given a prediction-oblivious online algorithm, its learning-augmented counterpart produced by applying the DTB compiler either follows the incoming guidance blindly or ignores it altogether and proceeds as the initial algorithm would have; the choice between these two alternatives is based on the outcome of a (biased) coin toss. As our main technical contribution, we prove (rigorously) that although remarkably simple, the class of algorithms produced via the DTB compiler includes algorithms with attractive consistency-robustness guarantees for three classic online problems: for caching and uniform metrical task systems our algorithms are optimal, whereas for bipartite matching (with adversarial arrival order), our algorithm outperforms the state-of-the-art.

We study high-dimensional mean estimation in a collaborative setting where data is contributed by $N$ users in batches of size $n$. In this environment, a learner seeks to recover the mean $μ$ of a true distribution $P$ from a collection of sources that are both statistically heterogeneous and potentially malicious. We formalize this challenge through a double corruption landscape: an $\varepsilon$-fraction of users are entirely adversarial, while the remaining ``good'' users provide data from distributions that are related to $P$, but deviate by a proximity parameter $α$. Unlike existing work on the untrusted batch model, which typically measures this deviation via total variation distance in discrete settings, we address the continuous, high-dimensional regime under two natural variants for deviation: (1) good batches are drawn from distributions with a mean-shift of $\sqrtα$, or (2) an $α$-fraction of samples within each good batch are adversarially corrupted. In particular, the second model presents significant new challenges: in high dimensions, unlike discrete settings, even a small fraction of sample-level corruption can shift empirical means and covariances arbitrarily. We provide two Sum-of-Squares (SoS) based algorithms to navigate this tiered corruption. Our algorithms achieve the minimax-optimal error rate $O(\sqrt{\varepsilon/n} + \sqrt{d/nN} + \sqrtα)$, demonstrating that while heterogeneity $α$ represents an inherent statistical difficulty, the influence of adversarial users is suppressed by a factor of $1/\sqrt{n}$ due to the internal averaging afforded by the batch structure.

Reliable AI systems require large language models (LLMs) to exhibit behaviors aligned with human preferences and values. However, most existing alignment approaches operate at training time and rely on additional high-quality data, incurring significant computational and annotation costs. While recent work has shown that contrastive decoding can leverage a model's internal distributions to improve specific capabilities, its applicability remains limited to narrow behavioral scopes and scenarios. In this work, we introduce Polarity-Prompt Contrastive Decoding (PromptCD), a test-time behavior control method that generalizes contrastive decoding to broader enhancement settings. PromptCD constructs paired positive and negative guiding prompts for a target behavior and contrasts model responses-specifically token-level probability distributions in LLMs and visual attention patterns in VLMs-to reinforce desirable outcomes. This formulation extends contrastive decoding to a wide range of enhancement objectives and is applicable to both LLMs and Vision-Language Models (VLMs) without additional training. For LLMs, experiments on the "3H" alignment objectives (helpfulness, honesty, and harmlessness) demonstrate consistent and substantial improvements, indicating that post-trained models can achieve meaningful self-enhancement purely at test time. For VLMs, we further analyze contrastive effects on visual attention, showing that PromptCD significantly improves VQA performance by reinforcing behavior-consistent visual grounding. Collectively, these results highlight PromptCD as a simple, general, and cost-efficient strategy for reliable behavior control across modalities.

Recent advances in vision-language models (VLMs) have shown promise for human-level embodied intelligence. However, existing benchmarks for VLM-driven embodied agents often rely on high-level commands or discretized action spaces, which are non-native settings that differ markedly from real-world control. In addition, current benchmarks focus primarily on high-level tasks and lack joint evaluation and analysis at both low and high levels. To address these limitations, we present NativeEmbodied, a challenging benchmark for VLM-driven embodied agents that uses a unified, native low-level action space. Built on diverse simulated scenes, NativeEmbodied includes three representative high-level tasks in complex scenarios to evaluate overall performance. For more detailed analysis, we further decouple the skills required by complex tasks and construct four types of low-level tasks, each targeting a fundamental embodied skill. This joint evaluation across task and skill granularities enables fine-grained assessment of embodied agents. Experiments with state-of-the-art VLMs reveal clear deficiencies in several fundamental embodied skills, and further analysis shows that these bottlenecks significantly limit performance on high-level tasks. NativeEmbodied highlights key challenges for current VLM-driven embodied agents and provides insights to guide future research.

Current AI-assisted engineering workflows lack a built-in mechanism to maintain task-level verification and regulatory traceability at machine-speed delivery. Agile V addresses this gap by embedding independent verification and audit artifact generation into each task cycle. The framework merges Agile iteration with V-Model verification into a continuous Infinity Loop, deploying specialized AI agents for requirements, design, build, test, and compliance, governed by mandatory human approval gates. We evaluate three hypotheses: (H1) audit-ready artifacts emerge as a by-product of development, (H2) 100% requirement-level verification is achievable with independent test generation, and (H3) verified increments can be delivered with single-digit human interactions per cycle. A feasibility case study on a Hardware-in-the-Loop system (about 500 LOC, 8 requirements, 54 tests) supports all three hypotheses: audit-ready documentation was generated automatically (H1), 100% requirement-level pass rate was achieved (H2), and only 6 prompts per cycle were required (H3), yielding an estimated 10-50x cost reduction versus a COCOMO II baseline (sensitivity range from pessimistic to optimistic assumptions). We invite independent replication to validate generalizability.

Urban systems, as dynamic complex systems, continuously generate spatio-temporal data streams that encode the fundamental laws of human mobility and city evolution. While AI for Science has witnessed the transformative power of foundation models in disciplines like genomics and meteorology, urban computing remains fragmented due to "scenario-specific" models, which are overfitted to specific regions or tasks, hindering their generalizability. To bridge this gap and advance spatio-temporal foundation models for urban systems, we adopt scaling as the central perspective and systematically investigate two key questions: what to scale and how to scale. Grounded in first-principles analysis, we identify three critical dimensions: heterogeneity, correlation, and dynamics, aligning these principles with the fundamental scientific properties of urban spatio-temporal data. Specifically, to address heterogeneity through data scaling, we construct WorldST. This billion-scale corpus standardizes diverse physical signals, such as traffic flow and speed, from over 100 global cities into a unified data format. To enable computation scaling for modeling correlations, we introduce the MiniST unit, a novel split mechanism that discretizes continuous spatio-temporal fields into learnable computational units to unify representations of grid-based and sensor-based observations. Finally, addressing dynamics via architecture scaling, we propose UrbanFM, a minimalist self-attention architecture designed with limited inductive biases to autonomously learn dynamic spatio-temporal dependencies from massive data. Furthermore, we establish EvalST, the largest-scale urban spatio-temporal benchmark to date. Extensive experiments demonstrate that UrbanFM achieves remarkable zero-shot generalization across unseen cities and tasks, marking a pivotal first step toward large-scale urban spatio-temporal foundation models.

In search systems, effectively coordinating the two core objectives of search relevance matching and click-through rate (CTR) prediction is crucial for discovering users' interests and enhancing platform revenue. In our prior work PRECTR, we proposed a unified framework to integrate these two subtasks,thereby eliminating their inconsistency and leading to mutual benefit.However, our previous work still faces three main challenges. First, low-active users and new users have limited search behavioral data, making it difficult to achieve effective personalized relevance preference modeling. Second, training data for ranking models predominantly come from high-relevance exposures, creating a distribution mismatch with the broader candidate space in coarse-ranking, leading to generalization bias. Third, due to the latency constraint, the original model employs an Emb+MLP architecture with a frozen BERT encoder, which prevents joint optimization and creates misalignment between representation learning and CTR fine-tuning. To solve these issues, we further reinforce our method and propose PRECTR-V2. Specifically, we mitigate the low-activity users' sparse behavior problem by mining global relevance preferences under the specific query, which facilitates effective personalized relevance modeling for cold-start scenarios. Subsequently, we construct hard negative samples through embedding noise injection and relevance label reconstruction, and optimize their relative ranking against positive samples via pairwise loss, thereby correcting exposure bias. Finally, we pretrain a lightweight transformer-based encoder via knowledge distillation from LLM and SFT on the text relevance classification task. This encoder replaces the frozen BERT module, enabling better adaptation to CTR fine-tuning and advancing beyond the traditional Emb+MLP paradigm.

The rapid growth of dockless bike-sharing systems has generated massive spatio-temporal datasets useful for fleet allocation, congestion reduction, and sustainable mobility. Bike demand, however, depends on several external factors, making traditional time-series models insufficient. Centralized Machine Learning (CML) yields high-accuracy forecasts but raises privacy and bandwidth issues when data are distributed across edge devices. To overcome these limitations, we propose Bikelution, an efficient Federated Learning (FL) solution based on gradient-boosted trees that preserves privacy while delivering accurate mid-term demand forecasts up to six hours ahead. Experiments on three real-world BSS datasets show that Bikelution is comparable to its CML-based variant and outperforms the current state-of-the-art. The results highlight the feasibility of privacy-aware demand forecasting and outline the trade-offs between FL and CML approaches.

Reward models play a fundamental role in aligning large language models with human preferences. Existing methods predominantly follow two paradigms: scalar discriminative preference models, which are efficient but lack interpretability, and generative judging models, which offer richer reasoning at the cost of higher computational overhead. We observe that the log-probability margin between verdict tokens strongly correlates with prediction correctness, providing a reliable proxy for instance difficulty without additional inference cost. Building on this insight, we propose CAMEL, a confidence-gated reflection framework that performs a lightweight single-token preference decision first and selectively invokes reflection only for low-confidence instances. To induce effective self-correction, we train the model via reinforcement learning with counterfactual prefix augmentation, which exposes the model to diverse initial verdicts and encourages genuine revision. Empirically, CAMEL achieves state-of-the-art performance on three widely used reward-model benchmarks with 82.9% average accuracy, surpassing the best prior model by 3.2% and outperforming 70B-parameter models using only 14B parameters, while establishing a strictly better accuracy-efficiency Pareto frontier.

The deployment of Large Language Models (LLMs) for real-time intelligence on edge devices is rapidly growing. However, conventional hardware architectures face a fundamental memory wall challenge, where limited on-device memory capacity and bandwidth severely constrain the size of deployable models and their inference speed, while also limiting on-device adaptation. To address this challenge, we propose TOM, a hybrid ROM-SRAM accelerator co-designed with ternary quantization, which balances extreme density with on-device tunability. TOM exploits the synergy between ternary quantization and ROM to achieve extreme memory density and bandwidth, while preserving flexibility through a hybrid ROM-SRAM architecture designed for QLoRA-based tunability. Specifically, we introduce: (1) a sparsity-aware ROM architecture that synthesizes ternary weights as standard-cell logic, eliminating area overhead from zero-valued bits; (2) a distributed processing architecture that co-locates high-density ROM banks with flexible SRAM-based QLoRA adapters and compute units; and (3) a workload-aware dynamic power gating scheme that exploits the logic-based nature of ROM to power down inactive banks, minimizing dynamic energy consumption. TOM achieves an inference throughput of 3,306 TPS using BitNet-2B model, demonstrating its effectiveness in delivering real-time, energy-efficient edge intelligence.

Current vision-language-action (VLA) models struggle with long-horizon manipulation under partial observability. Most existing approaches remain observation-driven, relying on short context windows or repeated queries to vision-language models (VLMs). This leads to loss of task progress, action repetition under perceptual aliasing, and high inference latency. Semantic reasoning alone is not the primary bottleneck in long-horizon manipulation. Instead, VLAs lack persistent, action-conditioned state representations and exhibit limited temporal and physical reasoning, making them ill-suited for multi-stage control. This paper introduces RB-VLA, a belief-centric architecture trained with self-supervised world-model objectives that maintains a compact latent state encoding task-relevant history, dynamics, and object interactions. Queried once for high-level intent, the VLM provides task specification, while the belief tracks task progress and enables phase-aware, causally grounded control under partial observability without storing raw observations or scaling memory with time. The belief and intent jointly condition a diffusion policy for robust closed-loop execution. RB-VLA outperforms prior VLAs on long-horizon benchmarks, achieving 52.5% and 37.5% higher success on multi-stage pick-and-place and stacking tasks, respectively, compared to π0. It also reduces inference latency by up to 5x relative to baselines and eliminates memory growth across timesteps observed in existing VLAs. Ablations show that the belief module is the primary driver of performance, increasing success rates from 32.5% to 77.5%. These results demonstrate the effectiveness of belief-based state representations for long-horizon VLA policies.

Manual lifting tasks are a major contributor to work-related musculoskeletal disorders, and effective ergonomic risk assessment is essential for quantifying physical exposure and informing ergonomic interventions. The Revised NIOSH Lifting Equation (RNLE) is a widely used ergonomic risk assessment tool for lifting tasks that relies on six task variables, including horizontal (H) and vertical (V) hand distances; such distances are typically obtained through manual measurement or specialized sensing systems and are difficult to use in real-world environments. We evaluated the feasibility of using innovative vision-language models (VLMs) to non-invasively estimate H and V from RGB video streams. Two multi-stage VLM-based pipelines were developed: a text-guided detection-only pipeline and a detection-plus-segmentation pipeline. Both pipelines used text-guided localization of task-relevant regions of interest, visual feature extraction from those regions, and transformer-based temporal regression to estimate H and V at the start and end of a lift. For a range of lifting tasks, estimation performance was evaluated using leave-one-subject-out validation across the two pipelines and seven camera view conditions. Results varied significantly across pipelines and camera view conditions, with the segmentation-based, multi-view pipeline consistently yielding the smallest errors, achieving mean absolute errors of approximately 6-8 cm when estimating H and 5-8 cm when estimating V. Across pipelines and camera view configurations, pixel-level segmentation reduced estimation error by approximately 20-30% for H and 35-40% for V relative to the detection-only pipeline. These findings support the feasibility of VLM-based pipelines for video-based estimation of RNLE distance parameters.

Overlapping communication with computation is crucial for distributed large-model training, yet optimizing it - especially when computation becomes the bottleneck-remains challenging. We present Lagom, a system that co-tunes communication parameters to balance resource usage between computation and communication. By introducing a unified cost model and a priority-based search algorithm, Lagom reduces optimization complexity from exponential to linear. Evaluations on high- and low-bandwidth GPU clusters show that Lagom achieves 1.07-1.33x and 1.03-1.27x speedup over NCCL and AutoCCL across diverse models and parallelizations.

The recent developments of complex deep learning models have led to unprecedented ability to accurately predict across multiple data representation types. Conformal prediction for uncertainty quantification of these models has risen in popularity, providing adaptive, statistically-valid prediction sets. For classification tasks, conformal methods have typically focused on utilizing logit scores. For pre-trained models, however, this can result in inefficient, overly conservative set sizes when not calibrated towards the target task. We propose DANCE, a doubly locally adaptive nearest-neighbor based conformal algorithm combining two novel nonconformity scores directly using the data's embedded representation. DANCE first fits a task-adaptive kernel regression model from the embedding layer before using the learned kernel space to produce the final prediction sets for uncertainty quantification. We test against state-of-the-art local, task-adapted and zero-shot conformal baselines, demonstrating DANCE's superior blend of set size efficiency and robustness across various datasets.

In modern applications such as ECG monitoring, neuroimaging, wearable sensing, and industrial equipment diagnostics, complex and continuously structured data are ubiquitous, presenting both challenges and opportunities for functional data analysis. However, existing methods face a critical trade-off: conventional functional models are limited by linearity, whereas deep learning approaches lack interpretable region selection for sparse effects. To bridge these gaps, we propose a sparse Bayesian functional deep neural network (sBayFDNN). It learns adaptive functional embeddings through a deep Bayesian architecture to capture complex nonlinear relationships, while a structured prior enables interpretable, region-wise selection of influential domains with quantified uncertainty. Theoretically, we establish rigorous approximation error bounds, posterior consistency, and region selection consistency. These results provide the first theoretical guarantees for a Bayesian deep functional model, ensuring its reliability and statistical rigor. Empirically, comprehensive simulations and real-world studies confirm the effectiveness and superiority of sBayFDNN. Crucially, sBayFDNN excels in recognizing intricate dependencies for accurate predictions and more precisely identifies functionally meaningful regions, capabilities fundamentally beyond existing approaches.

Large-scale image datasets are fundamental to deep learning, but their high storage demands pose challenges for deployment in resource-constrained environments. While existing approaches reduce dataset size by discarding samples, they often ignore the significant redundancy within each image -- particularly in the color space. To address this, we propose Dataset Color Quantization (DCQ), a unified framework that compresses visual datasets by reducing color-space redundancy while preserving information crucial for model training. DCQ achieves this by enforcing consistent palette representations across similar images, selectively retaining semantically important colors guided by model perception, and maintaining structural details necessary for effective feature learning. Extensive experiments across CIFAR-10, CIFAR-100, Tiny-ImageNet, and ImageNet-1K show that DCQ significantly improves training performance under aggressive compression, offering a scalable and robust solution for dataset-level storage reduction. Code is available at \href{https://github.com/he-y/Dataset-Color-Quantization}{https://github.com/he-y/Dataset-Color-Quantization}.

Client perceptions of the therapeutic alliance are critical for counseling effectiveness. Accurately capturing these perceptions remains challenging, as traditional post-session questionnaires are burdensome and often delayed, while existing computational approaches produce coarse scores, lack interpretable rationales, and fail to model holistic session context. We present CARE, an LLM-based framework to automatically predict multi-dimensional alliance scores and generate interpretable rationales from counseling transcripts. Built on the CounselingWAI dataset and enriched with 9,516 expert-curated rationales, CARE is fine-tuned using rationale-augmented supervision with the LLaMA-3.1-8B-Instruct backbone. Experiments show that CARE outperforms leading LLMs and substantially reduces the gap between counselor evaluations and client-perceived alliance, achieving over 70% higher Pearson correlation with client ratings. Rationale-augmented supervision further improves predictive accuracy. CARE also produces high-quality, contextually grounded rationales, validated by both automatic and human evaluations. Applied to real-world Chinese online counseling sessions, CARE uncovers common alliance-building challenges, illustrates how interaction patterns shape alliance development, and provides actionable insights, demonstrating its potential as an AI-assisted tool for supporting mental health care.

I introduce semantic novelty--cosine distance between each paragraph's sentence embedding and the running centroid of all preceding paragraphs--as an information-theoretic measure of narrative structure at corpus scale. Applying it to 28,606 books in PG19 (pre-1920 English literature), I compute paragraph-level novelty curves using 768-dimensional SBERT embeddings, then reduce each to a 16-segment Piecewise Aggregate Approximation (PAA). Ward-linkage clustering on PAA vectors reveals eight canonical narrative shape archetypes, from Steep Descent (rapid convergence) to Steep Ascent (escalating unpredictability). Volume--variance of the novelty trajectory--is the strongest length-independent predictor of readership (partial rho = 0.32), followed by speed (rho = 0.19) and Terminal/Initial ratio (rho = 0.19). Circuitousness shows strong raw correlation (rho = 0.41) but is 93 percent correlated with length; after control, partial rho drops to 0.11--demonstrating that naive correlations in corpus studies can be dominated by length confounds. Genre strongly constrains narrative shape (chi squared = 2121.6, p < 10 to the power negative 242), with fiction maintaining plateau profiles while nonfiction front-loads information. Historical analysis shows books became progressively more predictable between 1840 and 1910 (T/I ratio trend r = negative 0.74, p = 0.037). SAX analysis reveals 85 percent signature uniqueness, suggesting each book traces a nearly unique path through semantic space. These findings demonstrate that information-density dynamics, distinct from sentiment or topic, constitute a fundamental dimension of narrative structure with measurable consequences for reader engagement. Dataset: https://huggingface.co/datasets/wfzimmerman/pg19-semantic-novelty

We study stochastic gradient descent (SGD) for composite optimization problems with $N$ sequential operators subject to perturbations in both the forward and backward passes. Unlike classical analyses that treat gradient noise as additive and localized, perturbations to intermediate outputs and gradients cascade through the computational graph, compounding geometrically with the number of operators. We present the first comprehensive theoretical analysis of this setting. Specifically, we characterize how forward and backward perturbations propagate and amplify within a single gradient step, derive convergence guarantees for both general non-convex objectives and functions satisfying the Polyak--Łojasiewicz condition, and identify conditions under which perturbations do not deteriorate the asymptotic convergence order. As a byproduct, our analysis furnishes a theoretical explanation for the gradient spiking phenomenon widely observed in deep learning, precisely characterizing the conditions under which training recovers from spikes or diverges. Experiments on logistic regression with convex and non-convex regularization validate our theories, illustrating the predicted spike behavior and the asymmetric sensitivity to forward versus backward perturbations.

Real-world crash reports, which combine textual summaries and sketches, are valuable for scenario-based testing of autonomous driving systems (ADS). However, current methods cannot effectively translate this multimodal data into precise, executable simulation scenarios, hindering the scalability of ADS safety validation. In this work, we propose a scalable and verifiable pipeline that uses a large language model (GPT-4o mini) and a probabilistic intermediate representation (an Extended Scenic domain-specific language) to automatically extract semantic scenario configurations from crash reports and generate corresponding simulation-ready scenarios. Unlike earlier approaches such as ScenicNL and LCTGen (which generate scenarios directly from text) or TARGET (which uses deterministic mappings from traffic rules), our method introduces an intermediate Scenic DSL layer to separate high-level semantic understanding from low-level scenario rendering, reducing errors and capturing real-world variability. We evaluated the pipeline on cases from the NHTSA CIREN database. The results show high accuracy in knowledge extraction: 100% correctness for environmental and road network attributes, and 97% and 98% for oracle and actor trajectories, respectively, compared to human-derived ground truth. We executed the generated scenarios in the CARLA simulator using the Autoware driving stack, and they consistently triggered the intended traffic-rule violations (such as opposite-lane crossing and red-light running) across 2,000 scenario variations. These findings demonstrate that the proposed pipeline provides a legally grounded, scalable, and verifiable approach to ADS safety validation.

Mobility trajectories are essential for understanding urban dynamics and enhancing urban planning, yet access to such data is frequently hindered by privacy concerns. This research introduces a transformative framework for generating large-scale urban mobility trajectories, employing a novel application of a transformer-based model pre-trained and fine-tuned through a two-phase process. Initially, trajectory generation is conceptualized as an offline reinforcement learning (RL) problem, with a significant reduction in vocabulary space achieved during tokenization. The integration of Inverse Reinforcement Learning (IRL) allows for the capture of trajectory-wise reward signals, leveraging historical data to infer individual mobility preferences. Subsequently, the pre-trained model is fine-tuned using the constructed reward model, effectively addressing the challenges inherent in traditional RL-based autoregressive methods, such as long-term credit assignment and handling of sparse reward environments. Comprehensive evaluations on multiple datasets illustrate that our framework markedly surpasses existing models in terms of reliability and diversity. Our findings not only advance the field of urban mobility modeling but also provide a robust methodology for simulating urban data, with significant implications for traffic management and urban development planning. The implementation is publicly available at https://github.com/Wangjw6/TrajGPT_R.

Large Language Models (LLMs) have shown significant potential in scientific discovery but struggle to bridge the gap between theoretical reasoning and verifiable physical simulation. Existing solutions operate in a passive "execute-then-response" loop and thus lacks runtime perception, obscuring agents to transient anomalies (e.g., numerical instability or diverging oscillations). To address this limitation, we propose EmbodiedAct, a framework that transforms established scientific software into active embodied agents by grounding LLMs in embodied actions with a tight perception-execution loop. We instantiate EmbodiedAct within MATLAB and evaluate it on complex engineering design and scientific modeling tasks. Extensive experiments show that EmbodiedAct significantly outperforms existing baselines, achieving SOTA performance by ensuring satisfactory reliability and stability in long-horizon simulations and enhanced accuracy in scientific modeling.

Eliciting a preference model involves asking a person, named decision-maker, a series of questions. We assume that these preferences can be represented by an additive value function. In this work, we query simultaneously two decision-makers in the aim to elicit their respective value functions. For each query we receive two answers, without noise, but without knowing which answer corresponds to which decision-maker.We propose an elicitation procedure that identifies the two preference models when the marginal value functions are piecewise linear with known breaking points.

Accurate and stable field-of-view (FoV) guidance is critical for safe and efficient minimally invasive surgery, yet existing approaches often conflate visual attention estimation with downstream camera control or rely on direct object-centric assumptions. In this work, we formulate surgical attention tracking as a spatio-temporal learning problem and model surgeon focus as a dense attention heatmap, enabling continuous and interpretable frame-wise FoV guidance. We propose SurgAtt-Tracker, a holistic framework that robustly tracks surgical attention by exploiting temporal coherence through proposal-level reranking and motion-aware refinement, rather than direct regression. To support systematic training and evaluation, we introduce SurgAtt-1.16M, a large-scale benchmark with a clinically grounded annotation protocol that enables comprehensive heatmap-based attention analysis across procedures and institutions. Extensive experiments on multiple surgical datasets demonstrate that SurgAtt-Tracker consistently achieves state-of-the-art performance and strong robustness under occlusion, multi-instrument interference, and cross-domain settings. Beyond attention tracking, our approach provides a frame-wise FoV guidance signal that can directly support downstream robotic FoV planning and automatic camera control.

The proliferation of hate speech on social media platforms has necessitated the development of effective detection and moderation tools. This study evaluates the efficacy of various machine learning models in identifying hate speech and offensive language and investigates the potential of text transformation techniques to neutralize such content. We compare traditional models like CNNs and LSTMs with advanced neural network models such as BERT and its derivatives, alongside exploring hybrid models that combine different architectural features. Our results indicate that while advanced models like BERT show superior accuracy due to their deep contextual understanding, hybrid models exhibit improved capabilities in certain scenarios. Furthermore, we introduce innovative text transformation approaches that convert negative expressions into neutral ones, thereby potentially mitigating the impact of harmful content. The implications of these findings are discussed, highlighting the strengths and limitations of current technologies and proposing future directions for more robust hate speech detection systems.

As Large Language Models (LLMs) saturate elementary benchmarks, the research frontier has shifted from generation to the reliability of automated evaluation. We demonstrate that standard "LLM-as-a-Judge" protocols suffer from a systematic Alignment Gap when applied to upper-undergraduate to early graduate level mathematics. To quantify this, we introduce QEDBench, the first large-scale dual-rubric alignment benchmark to systematically measure alignment with human experts on university-level math proofs by contrasting course-specific rubrics against expert common knowledge criteria. By deploying a dual-evaluation matrix (7 judges x 5 solvers) against 1,000+ hours of human evaluation, we reveal that certain frontier evaluators like Claude Opus 4.5, DeepSeek-V3, Qwen 2.5 Max, and Llama 4 Maverick exhibit significant positive bias (up to +0.18, +0.20, +0.30, +0.36 mean score inflation, respectively). Furthermore, we uncover a critical reasoning gap in the discrete domain: while Gemini 3.0 Pro achieves state-of-the-art performance (0.91 average human evaluation score), other reasoning models like GPT-5 Pro and Claude Sonnet 4.5 see their performance significantly degrade in discrete domains. Specifically, their average human evaluation scores drop to 0.72 and 0.63 in Discrete Math, and to 0.74 and 0.50 in Graph Theory. In addition to these research results, we also release QEDBench as a public benchmark for evaluating and improving AI judges. Our benchmark is publicly published at https://github.com/qqliu/Yale-QEDBench.

AIs are increasingly being deployed with greater autonomy and capabilities, which increases the risk that a misaligned AI may be able to cause catastrophic harm. Untrusted monitoring -- using one untrusted model to oversee another -- is one approach to reducing risk. Justifying the safety of an untrusted monitoring deployment is challenging because developers cannot safely deploy a misaligned model to test their protocol directly. In this paper, we develop upon existing methods for rigorously demonstrating safety based on pre-deployment testing. We relax assumptions that previous AI control research made about the collusion strategies a misaligned AI might use to subvert untrusted monitoring. We develop a taxonomy covering passive self-recognition, causal collusion (hiding pre-shared signals), acausal collusion (hiding signals via Schelling points), and combined strategies. We create a safety case sketch to clearly present our argument, explicitly state our assumptions, and highlight unsolved challenges. We identify conditions under which passive self-recognition could be a more effective collusion strategy than those studied previously. Our work builds towards more robust evaluations of untrusted monitoring.

The term 'algorithmic fairness' is used to evaluate whether AI models operate fairly in both comparative (where fairness is understood as formal equality, such as "treat like cases as like") and non-comparative (where unfairness arises from the model's inaccuracy, arbitrariness, or inscrutability) contexts. Recent advances in multimodal large language models (MLLMs) are breaking new ground in multimodal understanding, reasoning, and generation; however, we argue that inconspicuous distortions arising from complex multimodal interaction dynamics can lead to systematic bias. The purpose of this position paper is twofold: first, it is intended to acquaint AI researchers with phenomenological explainable approaches that rely on the physical entities that the machine experiences during training/inference, as opposed to the traditional cognitivist symbolic account or metaphysical approaches; second, it is to state that this phenomenological doctrine will be practically useful for tackling algorithmic fairness issues in MLLMs. We develop a surrogate physics-based model that describes transformer dynamics (i.e., semantic network structure and self-/cross-attention) to analyze the dynamics of cross-modal bias in MLLM, which are not fully captured by conventional embedding- or representation-level analyses. We support this position through multi-input diagnostic experiments: 1) perturbation-based analyses of emotion classification using Qwen2.5-Omni and Gemma 3n, and 2) dynamical analysis of Lorenz chaotic time-series prediction through the physical surrogate. Across two architecturally distinct MLLMs, we show that multimodal inputs can reinforce modality dominance rather than mitigate it, as revealed by structured error-attractor patterns under systematic label perturbation, complemented by dynamical analysis.

Open-world object detection (OWOD) requires incrementally detecting known categories while reliably identifying unknown objects. Existing methods primarily focus on improving unknown recall, yet overlook interpretability, often leading to known-unknown confusion and reduced prediction reliability. This paper aims to make the entire OWOD framework interpretable, enabling the detector to truly "knowing the unknown". To this end, we propose a concept-driven InterPretable OWOD framework(IPOW) by introducing a Concept Decomposition Model (CDM) for OWOD, which explicitly decomposes the coupled RoI features in Faster R-CNN into discriminative, shared, and background concepts. Discriminative concepts identify the most discriminative features to enlarge the distances between known categories, while shared and background concepts, due to their strong generalization ability, can be readily transferred to detect unknown categories. Leveraging the interpretable framework, we identify that known-unknown confusion arises when unknown objects fall into the discriminative space of known classes. To address this, we propose Concept-Guided Rectification (CGR) to further resolve such confusion. Extensive experiments show that IPOW significantly improves unknown recall while mitigating confusion, and provides concept-level interpretability for both known and unknown predictions.

Mobile data technologies use ``actigraphs'' to furnish information on health variables as a function of a subject's movement. The advent of wearable devices and related technologies has propelled the creation of health databases consisting of human movement data to conduct research on mobility patterns and health outcomes. Statistical methods for analyzing high-resolution actigraph data depend on the specific inferential context, but the advent of Artificial Intelligence (AI) frameworks require that the methods be congruent to transfer learning and amortization. This article devises amortized Bayesian inference for actigraph time sheets. We pursue a Bayesian approach to ensure full propagation of uncertainty and its quantification using a hierarchical dynamic linear model. We build our analysis around actigraph data from the Physical Activity through Sustainable Transport Approaches in Los Angeles (PASTA-LA) study conducted by the Fielding School of Public Health in the University of California, Los Angeles. Apart from achieving probabilistic imputation of actigraph time sheets, we are also able to statistically learn about the time-varying impact of explanatory variables on the magnitude of acceleration (MAG) for a cohort of subjects.

Specifications are vital for ensuring program correctness, yet writing them manually remains challenging and time-intensive. Recent large language model (LLM)-based methods have shown successes in generating specifications such as postconditions, but existing single-pass prompting often yields inaccurate results. In this paper, we present SpecMind, a novel framework for postcondition generation that treats LLMs as interactive and exploratory reasoners rather than one-shot generators. SpecMind employs feedback-driven multi-turn prompting approaches, enabling the model to iteratively refine candidate postconditions by incorporating implicit and explicit correctness feedback, while autonomously deciding when to stop. This process fosters deeper code comprehension and improves alignment with true program behavior via exploratory attempts. Our empirical evaluation shows that SpecMind significantly outperforms state-of-the-art approaches in both accuracy and completeness of generated postconditions.

Self-optimizing behaviors can lead to outcomes where collective benefits are ultimately destroyed, a well-known phenomenon known as the ``tragedy of the commons". These scenarios are widely studied using game-theoretic approaches to analyze strategic agent decision-making. In this paper, we examine this phenomenon in a bi-level decision-making hierarchy, where low-level agents belong to multiple distinct populations, and high-level agents make decisions that impact the choices of the local populations they represent. We study strategic interactions in a context where the populations benefit from a common environmental resource that degrades with higher extractive efforts made by high-level agents. We characterize a unique symmetric Nash equilibrium in the high-level game, and investigate its consequences on the common resource. While the equilibrium resource level degrades as the number of populations grows large, there are instances where it does not become depleted. We identify such regions, as well as the regions where the resource does deplete.

We report our experience in applying runtime monitoring to a FluxCD-based continuous deployment (CD) process. Our target system consists of GitHub Actions, GitHub Container Registry (GHCR), FluxCD, and an application running on Kubernetes. We monitored its logs using SyMon. In our setting, we regard a deployment update as detected when FluxCD's polling log resolves the latest image tag. Through the case study, we found that FluxCD did not always detect a new image within five minutes after it was pushed to GHCR, whereas it always did so within ten minutes in the collected logs. Moreover, our results show that SyMon is fast enough for near-real-time monitoring in our setting.

Multi-tenant LLM serving frameworks widely adopt shared Key-Value caches to enhance efficiency. However, this creates side-channel vulnerabilities enabling prompt leakage attacks. Prior studies identified these attack surfaces yet focused on expanding attack vectors rather than optimizing attack performance, reporting impractically high attack costs that underestimate the true privacy risk. We propose OptiLeak, a reinforcement learning-enhanced framework that maximizes prompt reconstruction efficiency through two-stage fine-tuning. Our key insight is that domain-specific ``hard tokens'' -- terms difficult to predict yet carrying sensitive information -- can be automatically identified via likelihood ranking and used to construct preference pairs for Direct Preference Optimization, eliminating manual annotation. This enables effective preference alignment while avoiding the overfitting issues of extended supervised fine-tuning. Evaluated on three benchmarks spanning medical and financial domains, OptiLeak achieves up to $12.48\times$ reduction in average requests per token compared to baseline approaches, with consistent improvements across model scales from 3B to 14B parameters. Our findings demonstrate that cache-based prompt leakage poses a more severe threat than previously reported, underscoring the need for robust cache isolation in production deployments.

As a distributed collaborative machine learning paradigm, vertical federated learning (VFL) allows multiple passive parties with distinct features and one active party with labels to collaboratively train a model. Although it is known for the privacy-preserving capabilities, VFL still faces significant privacy and security threats from backdoor attacks. Existing backdoor attacks typically involve an attacker implanting a trigger into the model during the training phase and executing the attack by adding the trigger to the samples during the inference phase. However, in this paper, we find that triggers are not essential for backdoor attacks in VFL. In light of this, we disclose a new backdoor attack pathway in VFL by introducing a feature-based triggerless backdoor attack. This attack operates under a more stringent security assumption, where the attacker is honest-but-curious rather than malicious during the training phase. It comprises three modules: label inference for the targeted backdoor attack, poison generation with amplification and perturbation mechanisms, and backdoor execution to implement the attack. Extensive experiments on five benchmark datasets demonstrate that our attack outperforms three baseline backdoor attacks by 2 to 50 times while minimally impacting the main task. Even in VFL scenarios with 32 passive parties and only one set of auxiliary data, our attack maintains high performance. Moreover, when confronted with distinct defense strategies, our attack remains largely unaffected and exhibits strong robustness. We hope that the disclosure of this triggerless backdoor attack pathway will encourage the community to revisit security threats in VFL scenarios and inspire researchers to develop more robust and practical defense strategies.

Understanding minimal assumptions that enable learning and generalization is perhaps the central question of learning theory. Several celebrated results in statistical learning theory, such as the VC theorem and Littlestone's characterization of online learnability, establish conditions on the hypothesis class that allow for learning under independent data and adversarial data, respectively. Building upon recent work bridging these extremes, we study sequential decision making under distributional adversaries that can adaptively choose data-generating distributions from a fixed family $U$ and ask when such problems are learnable with sample complexity that behaves like the favorable independent case. We provide a near complete characterization of families $U$ that admit learnability in terms of a notion known as generalized smoothness i.e. a distribution family admits VC-dimension-dependent regret bounds for every finite-VC hypothesis class if and only if it is generalized smooth. Further, we give universal algorithms that achieve low regret under any generalized smooth adversary without explicit knowledge of $U$. Finally, when $U$ is known, we provide refined bounds in terms of a combinatorial parameter, the fragmentation number, that captures how many disjoint regions can carry nontrivial mass under $U$. These results provide a nearly complete understanding of learnability under distributional adversaries. In addition, building upon the surprising connection between online learning and differential privacy, we show that the generalized smoothness also characterizes private learnability under distributional constraints.

Modern language models (LM) are trained on large scrapes of the Web, containing millions of personal information (PI) instances, many of which LMs memorize, increasing privacy risks. In this work, we develop the regexes and rules (R&R) detector suite to detect email addresses, phone numbers, and IP addresses, which outperforms the best regex-based PI detectors. On a manually curated set of 483 instances of PI, we measure memorization: finding that 13.6% are parroted verbatim by the Pythia-6.9b model, i.e., when the model is prompted with the tokens that precede the PI in the original document, greedy decoding generates the entire PI span exactly. We expand this analysis to study models of varying sizes (160M-6.9B) and pretraining time steps (70k-143k iterations) in the Pythia model suite and find that both model size and amount of pretraining are positively correlated with memorization. Even the smallest model, Pythia-160m, parrots 2.7% of the instances exactly. Consequently, we strongly recommend that pretraining datasets be aggressively filtered and anonymized to minimize PI parroting.

We study online maximization of non-monotone Diminishing-Return(DR)-submodular functions over down-closed convex sets, a regime where existing projection-free online methods suffer from suboptimal regret and limited feedback guarantees. Our main contribution is a new structural result showing that this class is $1/e$-linearizable under carefully designed exponential reparametrization, scaling parameter, and surrogate potential, enabling a reduction to online linear optimization. As a result, we obtain $O(T^{1/2})$ static regret with a single gradient query per round and unlock adaptive and dynamic regret guarantees, together with improved rates under semi-bandit, bandit, and zeroth-order feedback. Across all feedback models, our bounds strictly improve the state of the art.

We study self-distillation in settings where supervision is unreliable: there are no ground truth labels, verifiable rewards, or external graders to evaluate answers. We focus on document-grounded question answering with asymmetric context, where a single model serves as both tutor (with access to a relevant source document during training) and student (answering from the question alone at test time). Rather than assuming tutor correctness, we derive supervision online from tutor consensus by sampling multiple document-grounded reasoning traces and using agreement to gate learning. Conditioned on this reliability signal, we distill knowledge through full tutor reasoning trajectories (not just final answers), providing a dense and stable learning signal. Empirically, this consensus-gated trajectory distillation substantially improves transfer to the document-free student. Held-out in-domain accuracy under asymmetric evaluation improves from 46.0\% to 62.0\%, and average (maj@8) accuracy on public document-free math benchmarks improves from 20.2\% to 35.4\%.

Molecules are commonly represented as SMILES strings, which can be readily converted to fixed-size molecular fingerprints. These fingerprints serve as feature vectors to train ML/DL models for molecular property prediction tasks in the field of computational chemistry, drug discovery, biochemistry, and materials science. Recent research has demonstrated that SMILES can be used to construct molecular graphs where atoms are nodes ($V$) and bonds are edges ($E$). These graphs can subsequently be used to train geometric DL models like GNN. GNN learns the inherent structural relationships within a molecule rather than depending on fixed-size fingerprints. Although GNN are powerful aggregators, their efficacy on smaller datasets and inductive biases across different architectures is less studied. In our present study, we performed a systematic benchmarking of four different GNN architectures across a diverse domain of datasets (physical chemistry, biological, and analytical). Additionally, we have also implemented a hierarchical fusion (GNN+FP) framework for target prediction. We observed that the fusion framework consistently outperforms or matches the performance of standalone GNN (RMSE improvement > $7\%$) and baseline models. Further, we investigated the representational similarity using centered kernel alignment (CKA) between GNN and fingerprint embeddings and found that they occupy highly independent latent spaces (CKA $\le0.46$). The cross-architectural CKA score suggests a high convergence between isotopic models like GCN, GraphSAGE and GIN (CKA $\geq0.88$), with GAT learning moderately independent representation (CKA $0.55-0.80$).

Many benchmarks for automated causal inference evaluate a system's performance based on a single numerical output, such as an Average Treatment Effect (ATE). This approach conflates two distinct steps in causal analysis: identification-formulating a valid research design under stated assumptions-and estimation-implementing that design numerically on finite data. We introduce CausalReasoningBenchmark, a benchmark of 173 queries across 138 real-world datasets, curated from 85 peer-reviewed research papers and four widely-used causal-inference textbooks. For each query a system must produce (i) a structured identification specification that names the strategy, the treatment, outcome, and control variables, and all design-specific elements, and (ii) a point estimate with a standard error. By scoring these two components separately, our benchmark enables granular diagnosis: it distinguishes failures in causal reasoning from errors in numerical execution. Baseline results with a state-of-the-art LLM show that, while the model correctly identifies the high-level strategy in 84 % of cases, full identification-specification correctness drops to only 30 %, revealing that the bottleneck lies in the nuanced details of research design rather than in computation. CausalReasoningBenchmark is publicly available on Hugging Face and is designed to foster the development of more robust automated causal-inference systems.

Push-Sum-based decentralized learning enables optimization over directed communication networks, where information exchange may be asymmetric. While convergence properties of such methods are well understood, their finite-iteration stability and generalization behavior remain unclear due to structural bias induced by column-stochastic mixing and asymmetric error propagation. In this work, we develop a unified uniform-stability framework for the Stochastic Gradient Push (SGP) algorithm that captures the effect of directed topology. A key technical ingredient is an imbalance-aware consistency bound for Push-Sum, which controls consensus deviation through two quantities: the stationary distribution imbalance parameter $δ$ and the spectral gap $(1-λ)$ governing mixing speed. This decomposition enables us to disentangle statistical effects from topology-induced bias. We establish finite-iteration stability and optimization guarantees for both convex objectives and non-convex objectives satisfying the Polyak--Łojasiewicz condition. For convex problems, SGP attains excess generalization error of order $\tilde{\mathcal{O}}\!\left(\frac{1}{\sqrt{mn}}+\fracγ{δ(1-λ)}+γ\right)$ under step-size schedules, and we characterize the corresponding optimal early stopping time that minimizes this bound. For PŁ objectives, we obtain convex-like optimization and generalization rates with dominant dependence proportional to $κ\!\left(1+\frac{1}{δ(1-λ)}\right)$, revealing a multiplicative coupling between problem conditioning and directed communication topology. Our analysis clarifies when Push-Sum correction is necessary compared with standard decentralized SGD and quantifies how imbalance and mixing jointly shape the best attainable learning performance.

Task-based runtime systems provide flexible load balancing and portability for parallel scientific applications, but their strong scaling is highly sensitive to task granularity. As parallelism increases, scheduling overhead may transition from negligible to dominant, leading to rapid drops in performance for some algorithms, while remaining negligible for others. Although such effects are widely observed empirically, there is a general lack of understanding how algorithmic structure impacts whether dynamic scheduling is always beneficial. In this work, we introduce a granularity characterization framework that directly links scheduling overhead growth to task-graph dependency topology. We show that dependency structure, rather than problem size alone, governs how overhead scales with parallelism. Based on this observation, we characterize execution behavior using a simple granularity measure that indicates when scheduling overhead can be amortized by parallel computation and when scheduling overhead dominates performance. Through experimental evaluation on representative parallel workloads with diverse dependency patterns, we demonstrate that the proposed characterization explains both gradual and abrupt strong-scaling breakdowns observed in practice. We further show that overhead models derived from dependency topology accurately predict strong-scaling limits and enable a practical runtime decision rule for selecting dynamic or static execution without requiring exhaustive strong-scaling studies or extensive offline tuning.

Large language models (LLMs) are promising backbones for generative recommender systems, yet a key challenge remains underexplored: verbalization, i.e., converting structured user interaction logs into effective natural language inputs. Existing methods rely on rigid templates that simply concatenate fields, yielding suboptimal representations for recommendation. We propose a data-centric framework that learns verbalization for LLM-based recommendation. Using reinforcement learning, a verbalization agent transforms raw interaction histories into optimized textual contexts, with recommendation accuracy as the training signal. This agent learns to filter noise, incorporate relevant metadata, and reorganize information to improve downstream predictions. Experiments on a large-scale industrial streaming dataset show that learned verbalization delivers up to 93% relative improvement in discovery item recommendation accuracy over template-based baselines. Further analysis reveals emergent strategies such as user interest summarization, noise removal, and syntax normalization, offering insights into effective context construction for LLM-based recommender systems.

Symbolic Regression (SR) tries to reveal the hidden equations behind observed data. However, most methods search within a discrete equation space, where the structural modifications of equations rarely align with their numerical behavior, leaving fitting error feedback too noisy to guide exploration. To address this challenge, we propose GenSR, a generative latent space-based SR framework following the `map construction -> coarse localization -> fine search'' paradigm. Specifically, GenSR first pretrains a dual-branch Conditional Variational Autoencoder (CVAE) to reparameterize symbolic equations into a generative latent space with symbolic continuity and local numerical smoothness. This space can be regarded as a well-structured `map'' of the equation space, providing directional signals for search. At inference, the CVAE coarsely localizes the input data to promising regions in the latent space. Then, a modified CMA-ES refines the candidate region, leveraging smooth latent gradients. From a Bayesian perspective, GenSR reframes the SR task as maximizing the conditional distribution $p(\mathrm{Equ.} \mid \mathrm{Num.})$, with CVAE training achieving this objective through the Evidence Lower Bound (ELBO). This new perspective provides a theoretical guarantee for the effectiveness of GenSR. Extensive experiments show that GenSR jointly optimizes predictive accuracy, expression simplicity, and computational efficiency, while remaining robust under noise.

The tremendous success of Transformer models in fields such as large language models and computer vision necessitates a rigorous theoretical investigation. To the best of our knowledge, this paper is the first work proving that standard Transformers can approximate Hölder functions $ C^{s,λ}\left([0,1]^{d\times n}\right) $$ (s\in\mathbb{N}_{\geq0},0<λ\leq1) $ under the $L^t$ distance ($t \in [1, \infty]$) with arbitrary precision. Building upon this approximation result, we demonstrate that standard Transformers achieve the minimax optimal rate in nonparametric regression for Hölder target functions. It is worth mentioning that, by introducing two metrics: the size tuple and the dimension vector, we provide a fine-grained characterization of Transformer structures, which facilitates future research on the generalization and optimization errors of Transformers with different structures. As intermediate results, we also derive the upper bounds for the Lipschitz constant of standard Transformers and their memorization capacity, which may be of independent interest. These findings provide theoretical justification for the powerful capabilities of Transformer models.

The choice of prior is central to solving ill-posed imaging inverse problems, making it essential to select one consistent with the measurements $y$ to avoid severe bias. In Bayesian inverse problems, this could be achieved by evaluating the model evidence $p(y \mid M)$ under different models $M$ that specify the prior and then selecting the one with the highest value. Diffusion models are the state-of-the-art approach to solving inverse problems with a data-driven prior; however, directly computing the model evidence with respect to a diffusion prior is intractable. Furthermore, most existing model evidence estimators require either many pointwise evaluations of the unnormalized prior density or an accurate clean prior score. We propose \method, an estimator of the model evidence of a diffusion prior by integrating over the time-marginals of posterior sampling methods. Our method leverages the large amount of intermediate samples naturally obtained during the reverse diffusion sampling process to obtain an accurate estimation of the model evidence using only a handful of posterior samples (e.g., 20). We also demonstrate how to implement our estimator in tandem with recent diffusion posterior sampling methods. Empirically, our estimator matches the model evidence when it can be computed analytically, and it is able to both select the correct diffusion model prior and diagnose prior misfit under different highly ill-conditioned, non-linear inverse problems, including a real-world black hole imaging problem.

Conversational AI tools have been rapidly adopted by students and are becoming part of their learning routines. To understand what drives this adoption, we draw on the Technology Acceptance Model (TAM) and examine how perceived usefulness and perceived ease of use relate to students' behavioral intention to use conversational AI that generates responses for learning tasks. We extend TAM by incorporating trust, perceived enjoyment, and subjective norms as additional factors that capture social and affective influences and uncertainty around AI outputs. Using partial least squares structural equation modeling, we find perceived usefulness remains the strongest predictor of students' intention to use conversational AI. However, perceived ease of use does not exert a significant direct effect on behavioral intention once other factors are considered, operating instead indirectly through perceived usefulness. Trust and subjective norms significantly influence perceptions of usefulness, while perceived enjoyment exerts both a direct and indirect effect on usage intentions. These findings suggest that adoption decisions for conversational AI systems are influenced less by effort-related considerations and more by confidence in system outputs, affective engagement, and social context. Future research is needed to further examine how these acceptance relationships generalize across different conversational systems and usage contexts.

We study the problem of fairly allocating indivisible goods when limited sharing is allowed, that is, each good may be allocated to up to $k$ agents, while incurring a cost for sharing. While classic maximin share (MMS) allocations may not exist in many instances, we demonstrate that allowing controlled sharing can restore fairness guarantees that are otherwise unattainable in certain scenarios. (1) Our first contribution shows that exact maximin share (MMS) allocations are guaranteed to exist whenever goods are allowed to be cost-sensitively shared among at least half of the agents and the number of agents is even; for odd numbers of agents, we obtain a slightly weaker MMS guarantee. (2) We further design a Shared Bag-Filling Algorithm that guarantees a $(1 - C)(k - 1)$-approximate MMS allocation, where $C$ is the maximum cost of sharing a good. Notably, when $(1 - C)(k - 1) \geq 1$, our algorithm recovers an exact MMS allocation. (3) We additionally introduce the Sharing Maximin Share (SMMS) fairness notion, a natural extension of MMS to the $k$-sharing setting. (4) We show that SMMS allocations always exist under identical utilities and for instances with two agents. (5) We construct a counterexample to show the impossibility of the universal existence of an SMMS allocation. (6) Finally, we establish a connection between SMMS and constrained MMS (CMMS), yielding approximation guarantees for SMMS via existing CMMS results. These contributions provide deep theoretical insights for the problem of fair resource allocation when a limited sharing of resources are allowed in multi-agent environments.

Post-training large foundation models with reinforcement learning typically relies on massive and heterogeneous datasets, making effective curriculum learning both critical and challenging. In this work, we propose ACTOR-CURATOR, a scalable and fully automated curriculum learning framework for reinforcement learning post-training of large language models (LLMs). ACTOR-CURATOR learns a neural curator that dynamically selects training problems from large problem banks by directly optimizing for expected policy performance improvement. We formulate problem selection as a non-stationary stochastic bandit problem, derive a principled loss function based on online stochastic mirror descent, and establish regret guarantees under partial feedback. Empirically, ACTOR-CURATOR consistently outperforms uniform sampling and strong curriculum baselines across a wide range of challenging reasoning benchmarks, demonstrating improved training stability and efficiency. Notably, it achieves relative gains of 28.6% on AIME2024 and 30.5% on ARC-1D over the strongest baseline and up to 80% speedup. These results suggest that ACTOR-CURATOR is a powerful and practical approach for scalable LLM post-training.

Emotion recognition from multi-modal physiological and behavioral signals plays a pivotal role in affective computing, yet most existing models remain constrained to the prediction of singular emotions in controlled laboratory settings. Real-world human emotional experiences, by contrast, are often characterized by the simultaneous presence of multiple affective states, spurring recent interest in mixed emotion recognition as an emotion distribution learning problem. Current approaches, however, often neglect the valence consistency and structured correlations inherent among coexisting emotions. To address this limitation, we propose a Memory-guided Prototypical Co-occurrence Learning (MPCL) framework that explicitly models emotion co-occurrence patterns. Specifically, we first fuse multi-modal signals via a multi-scale associative memory mechanism. To capture cross-modal semantic relationships, we construct emotion-specific prototype memory banks, yielding rich physiological and behavioral representations, and employ prototype relation distillation to ensure cross-modal alignment in the latent prototype space. Furthermore, inspired by human cognitive memory systems, we introduce a memory retrieval strategy to extract semantic-level co-occurrence associations across emotion categories. Through this bottom-up hierarchical abstraction process, our model learns affectively informative representations for accurate emotion distribution prediction. Comprehensive experiments on two public datasets demonstrate that MPCL consistently outperforms state-of-the-art methods in mixed emotion recognition, both quantitatively and qualitatively.

The Stop-Think-AutoRegress Language Diffusion Model (STAR-LDM) integrates latent diffusion planning with autoregressive generation. Unlike conventional autoregressive language models limited to token-by-token decisions, STAR-LDM incorporates a "thinking" phase that pauses generation to refine a semantic plan through diffusion before continuing. This enables global planning in continuous space prior to committing to discrete tokens. Evaluations show STAR-LDM significantly outperforms similar-sized models on language understanding benchmarks and achieves $>70\%$ win rates in LLM-as-judge comparisons for narrative coherence and commonsense reasoning. The architecture also allows straightforward control through lightweight classifiers, enabling fine-grained steering of attributes without model retraining while maintaining better fluency-control trade-offs than specialized approaches.

Reinforcement Learning (RL) and Deep Reinforcement Learning (DRL) have advanced rapidly in recent years and have been successfully applied to e-learning environments like intelligent tutoring systems (ITSs). Despite great success, the broader application of DRL to educational technologies has been limited due to major challenges such as sample inefficiency and difficulty designing the reward function. In contrast, Apprenticeship Learning (AL) uses a few expert demonstrations to infer the expert's underlying reward functions and derive decision-making policies that generalize and replicate optimal behavior. In this work, we leverage a generalized AL framework, THEMES, to induce effective pedagogical policies by capturing the complexities of the expert student learning process, where multiple reward functions may dynamically evolve over time. We evaluate the effectiveness of THEMES against six state-of-the-art baselines, demonstrating its superior performance and highlighting its potential as a powerful alternative for inducing effective pedagogical policies and show that it can achieve high performance, with an AUC of 0.899 and a Jaccard of 0.653, using only 18 trajectories of a previous semester to predict student pedagogical decisions in a later semester.

We study how visual artifacts introduced by diffusion-based inpainting affect language generation in vision-language models. We use a two-stage diagnostic setup in which masked image regions are reconstructed and then provided to captioning models, enabling controlled comparisons between captions generated from original and reconstructed inputs. Across multiple datasets, we analyze the relationship between reconstruction fidelity and downstream caption quality. We observe consistent associations between pixel-level and perceptual reconstruction metrics and both lexical and semantic captioning performance. Additional analysis of intermediate visual representations and attention patterns shows that inpainting artifacts lead to systematic, layer-dependent changes in model behavior. Together, these results provide a practical diagnostic framework for examining how visual reconstruction quality influences language generation in multimodal systems.

Effective human-AI coordination requires artificial agents capable of exhibiting and responding to human-like behaviors while adapting to changing contexts. Imitation learning has emerged as one of the prominent approaches to build such agents by training them to mimic human-demonstrated behaviors. However, current methods struggle to capture the inherent diversity and non-Markovian nature of human behavior and lack the ability to steer behavior at inference time. Drawing inspiration from the theory of human cognitive processes, where inner speech guides action selection before execution, we propose MIMIC (Modeling Inner Motivations for Imitation and Control), a framework that uses language as an internal representation of behavioral intent. MIMIC employs the novel use of vision-language models as linguistic scaffolding to train a conditional variational autoencoder capable of generating inner speech from observations. A diffusion-based behavior cloning policy then selects actions conditioned on current observations and the generated inner speech. MIMIC enables fine-grained steering of behavior at inference time by conditioning the agent on behavior-specific speech. Experiments across robotic manipulation tasks and human-AI collaboration games demonstrate that MIMIC significantly enhances both behavior diversity and fidelity to human demonstrations while enabling nuanced behavioral steering without training on additional demonstrations. We open source our code and provide pre-trained MIMIC agents and qualitative demos at: https://mimic-research.github.io.

In long-context large language model (LLM) inference, the prefill stage dominates computation due to self-attention over the complete input context. Sparse attention significantly reduces self-attention computation by limiting each token's interactions to a subset of tokens. The attention sparsity pattern varies across input prompts, and within a prompt, each attention head can follow a distinct pattern. This makes attention sparsity dynamic. The requirement of generating the sparsity pattern, combined with limited data reuse in attention, shifts the prefill compute to being memory-bound. This, in addition to the huge energy requirements for long-context inference on GPU, motivates FPGAs as good candidates for accelerating dynamic long-context inference. To tackle these challenges, we propose FAST-Prefill, the first FPGA accelerator for long-context prefill-stage inference with dynamic sparse attention. To efficiently generate sparse indices, we propose a \textit{fused pipeline unit with a memory-aware execution order} to reduce large tensors and irregular memory accesses. To reduce off-chip memory traffic for accessing the KV cache, we utilize the memory hierarchy to design a \textit{liveness-driven, dual-tier cache}. For high-throughput matrix multiplication, we design a \textit{hybrid Matrix Processing Unit (MPU)} with DSPs and bit-plane decomposition using LUTs. We implement FAST-Prefill on Alveo U280 and evaluate it on the Llama and Qwen models (batch size = 1) for context lengths ranging from 4K to 128K tokens. We demonstrate an average speedup of up to 2.5$\times$ in TTFT and 4.5$\times$ improvement in energy efficiency over GPU implementation on Nvidia A5000 GPU.

As large language models (LLMs) continue to improve at completing discrete tasks, they are being integrated into increasingly complex and diverse real-world systems. However, task-level success alone does not establish a model's fit for use in practice. In applied, high-stakes settings, LLM effectiveness is driven by a wider array of sociotechnical determinants that extend beyond conventional performance measures. Although a growing set of metrics capture many of these considerations, they are rarely organized in a way that supports consistent evaluation, leaving no unified taxonomy for assessing and comparing LLM utility across use cases. To address this gap, we introduce the Language Model Utility Taxonomy (LUX), a comprehensive framework that structures utility evaluation across four domains: performance, interaction, operations, and governance. Within each domain, LUX is organized hierarchically into thematically aligned dimensions and components, each grounded in metrics that enable quantitative comparison and alignment of model selection with intended use. In addition, an external dynamic web tool is provided to support exploration of the framework by connecting each component to a repository of relevant metrics (factors) for applied evaluation.

Existing Graphical User Interface (GUI) agents operate through step-by-step calls to vision language models--taking a screenshot, reasoning about the next action, executing it, then repeating on the new page--resulting in high costs and latency that scale with the number of reasoning steps, and limited accuracy due to no persistent memory of previously visited pages. We propose ActionEngine, a training-free framework that transitions from reactive execution to programmatic planning through a novel two-agent architecture: a Crawling Agent that constructs an updatable state-machine memory of the GUIs through offline exploration, and an Execution Agent that leverages this memory to synthesize complete, executable Python programs for online task execution. To ensure robustness against evolving interfaces, execution failures trigger a vision-based re-grounding fallback that repairs the failed action and updates the memory. This design drastically improves both efficiency and accuracy: on Reddit tasks from the WebArena benchmark, our agent achieves 95% task success with on average a single LLM call, compared to 66% for the strongest vision-only baseline, while reducing cost by 11.8x and end-to-end latency by 2x. Together, these components yield scalable and reliable GUI interaction by combining global programmatic planning, crawler-validated action templates, and node-level execution with localized validation and repair.

Diffusion models have achieved remarkable success in image and video generation tasks. However, the high computational demands of Diffusion Transformers (DiTs) pose a significant challenge to their practical deployment. While feature caching is a promising acceleration strategy, existing methods based on simple reusing or training-free forecasting struggle to adapt to the complex, stage-dependent dynamics of the diffusion process, often resulting in quality degradation and failing to maintain consistency with the standard denoising process. To address this, we propose a LEarnable Stage-Aware (LESA) predictor framework based on two-stage training. Our approach leverages a Kolmogorov-Arnold Network (KAN) to accurately learn temporal feature mappings from data. We further introduce a multi-stage, multi-expert architecture that assigns specialized predictors to different noise-level stages, enabling more precise and robust feature forecasting. Extensive experiments show our method achieves significant acceleration while maintaining high-fidelity generation. Experiments demonstrate 5.00x acceleration on FLUX.1-dev with minimal quality degradation (1.0% drop), 6.25x speedup on Qwen-Image with a 20.2% quality improvement over the previous SOTA (TaylorSeer), and 5.00x acceleration on HunyuanVideo with a 24.7% PSNR improvement over TaylorSeer. State-of-the-art performance on both text-to-image and text-to-video synthesis validates the effectiveness and generalization capability of our training-based framework across different models. Our code is included in the supplementary materials and will be released on GitHub.

Driven by the increasingly complex and decision-oriented demands of time series analysis, we introduce the Semantic-Conditional Time Series Reasoning task, which extends conventional time series analysis beyond purely numerical modeling to incorporate contextual and semantic understanding. To further enhance the mode's reasoning capabilities on complex time series problems, we propose a two-round reinforcement learning framework: the first round strengthens the mode's perception of fundamental temporal primitives, while the second focuses on semantic-conditioned reasoning. The resulting model, KairosVL, achieves competitive performance across both synthetic and real-world tasks. Extensive experiments and ablation studies demonstrate that our framework not only boosts performance but also preserves intrinsic reasoning ability and significantly improves generalization to unseen scenarios. To summarize, our work highlights the potential of combining semantic reasoning with temporal modeling and provides a practical framework for real-world time series intelligence, which is in urgent demand.

The rapid evolution of Large Language Model (LLM)-based autonomous agents is reshaping the digital landscape toward an emerging Agentic Web, where increasingly specialized agents must collaborate to accomplish complex tasks. However, existing collaboration paradigms are constrained to message passing, leaving execution environments as isolated silos. This creates a context gap: agents cannot directly manipulate files or invoke tools in a peer's environment, and must instead resort to costly, error-prone environment reconstruction. We introduce the Agent Workspace Collaboration Protocol (AWCP), which bridges this gap through temporary workspace delegation inspired by the Unix philosophy that everything is a file. AWCP decouples a lightweight control plane from pluggable transport mechanisms, allowing a Delegator to project its workspace to a remote Executor, who then operates on the shared files directly with unmodified local toolchains. We provide a fully open-source reference implementation with MCP tool integration and validate the protocol through live demonstrations of asymmetric collaboration, where agents with complementary capabilities cooperate through delegated workspaces. By establishing the missing workspace layer in the agentic protocol stack, AWCP paves the way for a universally interoperable agent ecosystem in which collaboration transcends message boundaries. The protocol and reference implementation are publicly available at https://github.com/SII-Holos/awcp.

Decentralized federated learning (DFL) based on low-rank adaptation (LoRA) enables mobile devices with multi-task datasets to collaboratively fine-tune a large language model (LLM) by exchanging locally updated parameters with a subset of neighboring devices via wireless connections for knowledge integration.However, directly aggregating parameters fine-tuned on heterogeneous datasets induces three primary issues across the DFL life-cycle: (i) \textit{catastrophic knowledge forgetting during fine-tuning process}, arising from conflicting update directions caused by data heterogeneity; (ii) \textit{inefficient communication and convergence during model aggregation process}, due to bandwidth-intensive redundant model transmissions; and (iii) \textit{multi-task knowledge interference during inference process}, resulting from incompatible knowledge representations coexistence during inference. To address these issues in a fully decentralized scenario, we first propose a sparse-and-orthogonal LoRA that ensures orthogonality between model updates to eliminate direction conflicts during fine-tuning.Then, we analyze how device connection topology affects multi-task performance, prompting a cluster-based topology design during aggregation.Finally, we propose an implicit mixture of experts (MoE) mechanism to avoid the coexistence of incompatible knowledge during inference. Simulation results demonstrate that the proposed approach effectively reduces communication resource consumption by up to $73\%$ and enhances average performance by $5\%$ compared with the traditional LoRA method.

Dialogue systems have long supported learner reflections, with theoretically grounded, rule-based designs offering structured scaffolding but often struggling to respond to shifts in engagement. Large Language Models (LLMs), in contrast, can generate context-sensitive responses but are not informed by decades of research on how learning interactions should be structured, raising questions about their alignment with pedagogical theories. This paper presents a hybrid dialogue system that embeds LLM responsiveness within a theory-aligned, rule-based framework to support learner reflections in a culturally responsive robotics summer camp. The rule-based structure grounds dialogue in self-regulated learning theory, while the LLM decides when and how to prompt deeper reflections, responding to evolving conversation context. We analyze themes across dialogues to explore how our hybrid system shaped learner reflections. Our findings indicate that LLM-embedded dialogues supported richer learner reflections on goals and activities, but also introduced challenges due to repetitiveness and misalignment in prompts, reducing engagement.

The distinctive architectural features of normalizing flows (NFs), notably bijectivity and tractable Jacobians, make them well-suited for generative modeling. Invertible neural networks (INNs) build on these principles to address supervised inverse problems, enabling direct modeling of both forward and inverse mappings. In this paper, we revisit these architectures from both theoretical and practical perspectives and address a key gap in the literature: the lack of theoretical guarantees on approximation quality under realistic assumptions, whether for posterior inference in INNs or for generative modeling with NFs. We introduce a unified framework for INNs and NFs based on variational unsupervised loss functions, inspired by analogous formulations in related areas such as generative adversarial networks (GANs) and the Precision-Recall divergence for training normalizing flows. Within this framework, we derive theoretical performance guarantees, quantifying posterior accuracy for INNs and distributional accuracy for NFs, under assumptions that are weaker and more practically realistic than those used in prior work. Building on these theoretical results, we conduct extensive case studies to distill general design principles and practical guidelines. We conclude by demonstrating the effectiveness of our approach on a realistic ocean-acoustic inversion problem.

LLM-based agentic coding assistants lack persistent memory: they lose coherence across sessions, forget project conventions, and repeat known mistakes. Recent studies characterize how developers configure agents through manifest files, but an open challenge remains how to scale such configurations for large, multi-agent projects. This paper presents a three-component codified context infrastructure developed during construction of a 108,000-line C# distributed system: (1) a hot-memory constitution encoding conventions, retrieval hooks, and orchestration protocols; (2) 19 specialized domain-expert agents; and (3) a cold-memory knowledge base of 34 on-demand specification documents. Quantitative metrics on infrastructure growth and interaction patterns across 283 development sessions are reported alongside four observational case studies illustrating how codified context propagates across sessions to prevent failures and maintain consistency. The framework is published as an open-source companion repository.

The High-Luminosity Large Hadron Collider (HL-LHC) at CERN will produce unprecedented datasets capable of revealing fundamental properties of the universe. However, realizing its discovery potential faces a significant challenge: extracting small signal fractions from overwhelming backgrounds dominated by approximately 200 simultaneous pileup collisions. This extreme noise severely distorts the physical observables required for accurate reconstruction. To address this, we introduce the Physics-Guided Hypergraph Transformer (PhyGHT), a hybrid architecture that combines distance-aware local graph attention with global self-attention to mirror the physical topology of particle showers formed in proton-proton collisions. Crucially, we integrate a Pileup Suppression Gate (PSG), an interpretable, physics-constrained mechanism that explicitly learns to filter soft noise prior to hypergraph aggregation. To validate our approach, we release a novel simulated dataset of top-quark pair production to model extreme pileup conditions. PhyGHT outperforms state-of-the-art baselines from the ATLAS and CMS experiments in predicting the signal's energy and mass correction factors. By accurately reconstructing the top quark's invariant mass, we demonstrate how machine learning innovation and interdisciplinary collaboration can directly advance scientific discovery at the frontiers of experimental physics and enhance the HL-LHC's discovery potential. The dataset and code are available at https://github.com/rAIson-Lab/PhyGHT

Extracting high-fidelity 2D contours from Scanning Electron Microscope (SEM) images is critical for calibrating Optical Proximity Correction (OPC) models. While foundation models like Segment Anything 2 (SAM2) are promising, adapting them to specialized domains with scarce annotated data is a major challenge. This paper presents a case study on adapting SAM2 for SEM contour extraction in a few-shot setting. We propose SegSEM, a framework built on two principles: a data-efficient fine-tuning strategy that adapts by selectively training only the model's encoders, and a robust hybrid architecture integrating a traditional algorithm as a confidence-aware fallback. Using a small dataset of 60 production images, our experiments demonstrate this methodology's viability. The primary contribution is a methodology for leveraging foundation models in data-constrained industrial applications.

Multivariate time-series anomaly detection is essential for reliable industrial control, telemetry, and service monitoring. However, the evolving inter-variable dependencies and inevitable noise render it challenging. Existing methods often use single-scale graphs or instance-level contrast. Moreover, learned dynamic graphs can overfit noise without a stable anchor, causing false alarms or misses. To address these challenges, we propose the CGSTA framework with two key innovations. First, Dynamic Layered Graph Construction (DLGC) forms local, regional, and global views of variable relations for each sliding window; rather than contrasting whole windows, Contrastive Discrimination across Scales (CDS) contrasts graph representations within each view and aligns the same window across views to make learning structure-aware. Second, Stability-Aware Alignment (SAA) maintains a per-scale stable reference learned from normal data and guides the current window's fast-changing graphs toward it to suppress noise. We fuse the multi-scale and temporal features and use a conditional density estimator to produce per-time-step anomaly scores. Across four benchmarks, CGSTA delivers optimal performance on PSM and WADI, and is comparable to the baseline methods on SWaT and SMAP.

The unmatched ability of Deep Neural Networks in capturing complex patterns in large and noisy datasets is often associated with their large hypothesis space, and consequently to the vast amount of parameters that characterize model architectures. Pruning techniques affirmed themselves as valid tools to extract sparse representations of neural networks parameters, carefully balancing between compression and preservation of information. However, a fundamental assumption behind pruning is that expendable weights should have small impact on the error of the network, while highly important weights should tend to have a larger influence on the inference. We argue that this idea could be generalized; what if a weight is not simply removed but also compensated with a perturbation of the adjacent bias, which does not contribute to the network sparsity? Our work introduces a novel pruning method in which the importance measure of each weight is computed considering the output behavior after an optimal perturbation of its adjacent bias, efficiently computable by automatic differentiation. These perturbations can be then applied directly after the removal of each weight, independently of each other. After deriving analytical expressions for the aforementioned quantities, numerical experiments are conducted to benchmark this technique against some of the most popular pruning strategies, demonstrating an intrinsic efficiency of the proposed approach in very diverse machine learning scenarios. Finally, our findings are discussed and the theoretical implications of our results are presented.

In bandit settings, optimizing long-term regret metrics requires exploration, which corresponds to sometimes taking myopically sub-optimal actions. When a long-lived principal merely recommends actions to be executed by a sequence of different agents (as in an online recommendation platform) this provides an incentive misalignment: exploration is "worth it" for the principal but not for the agents. Prior work studies regret minimization under the constraint of Bayesian Incentive-Compatibility in a static stochastic setting with a fixed and common prior shared amongst the agents and the algorithm designer. We show that (weighted) swap regret bounds on their own suffice to cause agents to faithfully follow forecasts in an approximate Bayes Nash equilibrium, even in dynamic environments in which agents have conflicting prior beliefs and the mechanism designer has no knowledge of any agents beliefs. To obtain these bounds, it is necessary to assume that the agents have some degree of uncertainty not just about the rewards, but about their arrival time -- i.e. their relative position in the sequence of agents served by the algorithm. We instantiate our abstract bounds with concrete algorithms for guaranteeing adaptive and weighted regret in bandit settings.

This paper provides a reinterpretation of the Drifting Model~\cite{deng2026generative} through a semigroup-consistent long-short flow-map factorization. We show that a global transport process can be decomposed into a long-horizon flow map followed by a short-time terminal flow map admitting a closed-form optimal velocity representation, and that taking the terminal interval length to zero recovers exactly the drifting field together with a conservative impulse term required for flow-map consistency. Based on this perspective, we propose a new likelihood learning formulation that aligns the long-short flow-map decomposition with density evolution under transport. We validate the framework through both theoretical analysis and empirical evaluations on benchmark tests, and further provide a theoretical interpretation of the feature-space optimization while highlighting several open problems for future study.

Attention learners, neural networks built on the attention mechanism, e.g., transformers, excel at learning the implicit relationships that relate sequences to their corresponding properties, e.g., mapping a given sequence of tokens to the probability of the next token. However, the learning process tends to be costly. To address this, we present a novel paradigm named Attention Neural Teaching (AtteNT) that reinterprets the learning process through a nonparametric teaching perspective. Specifically, the latter provides a theoretical framework for teaching mappings that are implicitly defined (i.e., nonparametric) via example selection. Such an implicit mapping is embodied through a dense set of sequence-property pairs, with the AtteNT teacher selecting a subset to accelerate convergence in attention learner training. By analytically investigating the role of attention on parameter-based gradient descent during training, and recasting the evolution of attention learners, shaped by parameter updates, through functional gradient descent in nonparametric teaching, we show for the first time that teaching attention learners is consistent with teaching importance-adaptive nonparametric learners. These new findings readily commit AtteNT to enhancing learning efficiency of attention learners. Specifically, we observe training time reductions of 13.01% for LLMs and 20.58% for ViTs, spanning both fine-tuning and training-from-scratch regimes. Crucially, these gains are achieved without compromising accuracy; in fact, performance is consistently preserved and often enhanced across a diverse set of downstream tasks.

Can AI systems trained on the scientific record up to a fixed point in time forecast the scientific advances that follow? Such a capability could help researchers identify collaborators and impactful research directions, and anticipate which problems and methods will become central next. We introduce PreScience -- a scientific forecasting benchmark that decomposes the research process into four interdependent generative tasks: collaborator prediction, prior work selection, contribution generation, and impact prediction. PreScience is a carefully curated dataset of 98K recent AI-related research papers, featuring disambiguated author identities, temporally aligned scholarly metadata, and a structured graph of companion author publication histories and citations spanning 502K total papers. We develop baselines and evaluations for each task, including LACERScore, a novel LLM-based measure of contribution similarity that outperforms previous metrics and approximates inter-annotator agreement. We find substantial headroom remains in each task -- e.g. in contribution generation, frontier LLMs achieve only moderate similarity to the ground-truth (GPT-5, averages 5.6 on a 1-10 scale). When composed into a 12-month end-to-end simulation of scientific production, the resulting synthetic corpus is systematically less diverse and less novel than human-authored research from the same period.

We study online alignment of large language models under misspecified preference feedback, where the observed preference oracle deviates from an ideal but unknown ground-truth oracle. The online LLM alignment problem is a bi-level reinforcement problem due to the coupling between data collection and policy updates. Recently, the problem has been reduced to tractable single-level objective in the SAIL (Self-Improving Efficient Online Alignment) framework. In this paper, we introduce a pointwise oracle uncertainty set in this problem and formulate an oracle-robust online alignment objective as a worst-case optimization problem. For log-linear policies, we show that this robust objective admits an exact closed-form decomposition into the original loss function plus an explicit sensitivity penalty. We develop projected stochastic composite updates for the resulting weakly convex objective and prove $\widetilde{O}(\varepsilon^{-2})$ oracle complexity for reaching approximate stationarity.

Federated Learning (FL) enables a distributed client-server architecture where multiple clients collaboratively train a global Machine Learning (ML) model without sharing sensitive local data. However, FL often results in lower accuracy than traditional ML algorithms due to statistical heterogeneity across clients. Prior works attempt to address this by using model updates, such as loss and bias, from client models to select participants that can improve the global model's accuracy. However, these updates neither accurately represent a client's heterogeneity nor are their selection methods deterministic. We mitigate these limitations by introducing Terraform, a novel client selection methodology that uses gradient updates and a deterministic selection algorithm to select heterogeneous clients for retraining. This bi-pronged approach allows Terraform to achieve up to 47 percent higher accuracy over prior works. We further demonstrate its efficiency through comprehensive ablation studies and training time analyses, providing strong justification for the robustness of Terraform.

Modern Protein Language Models (PLMs) apply transformer-based model architectures from natural language processing to biological sequences, predicting a variety of protein functions and properties. However, protein language has key differences from natural language, such as a rich functional space despite a vocabulary of only 20 amino acids. These differences motivate research into how transformer-based architectures operate differently in the protein domain and how we can better leverage PLMs to solve protein-related tasks. In this work, we begin by directly comparing how the distribution of information stored across layers of attention heads differs between the protein and natural language domain. Furthermore, we adapt a simple early-exit technique-originally used in the natural language domain to improve efficiency at the cost of performance-to achieve both increased accuracy and substantial efficiency gains in protein non-structural property prediction by allowing the model to automatically select protein representations from the intermediate layers of the PLMs for the specific task and protein at hand. We achieve performance gains ranging from 0.4 to 7.01 percentage points while simultaneously improving efficiency by over 10 percent across models and non-structural prediction tasks. Our work opens up an area of research directly comparing how language models change behavior when moved into the protein domain and advances language modeling in biological domains.

The Fischer--Lynch--Paterson (FLP) impossibility result is widely regarded as one of the most fundamental negative results in distributed computing: no deterministic protocol can guarantee consensus in an asynchronous system with even one faulty process. For forty years, the field has treated this as an immovable constraint, designing around it with randomized protocols, failure detectors, and weakened consistency models. This essay argues that FLP is not a law of physics but a theorem about a particular system model -- and that Open Atomic Ethernet (OAE) circumvents it by rejecting the asynchronous model at its foundation. We introduce the term bisynchronous to describe OAE's key property: bounded-time bilateral resolution in which both parties reach common knowledge of outcome at every round boundary -- a strictly stronger guarantee than synchrony alone. By constructing a bisynchronous, swap-based protocol at Layer 2, OAE sidesteps the load-bearing assumptions of FLP's asynchronous model, achieving deterministic atomic coordination without violating any impossibility result.

Machine learning holds great promise for advancing the field of medicine, with electronic health records (EHRs) serving as a primary data source. However, EHRs are often sparse and contain missing data due to various challenges and limitations in data collection and sharing between healthcare providers. Existing techniques for imputing missing values predominantly focus on known unknowns, such as missing or unavailable values of lab test results; most do not explicitly address situations where it is difficult to distinguish what is missing. For instance, a missing diagnosis code in an EHR could signify either that the patient has not been diagnosed with the condition or that a diagnosis was made, but not shared by a provider. Such situations fall into the paradigm of unknown unknowns. To address this challenge, we develop a general purpose algorithm for denoising data to recover unknown missing values in binary EHRs. We design a transformer-based denoising neural network where the output is thresholded adaptively to recover values in cases where we predict data are missing. Our results demonstrate improved accuracy in denoising medical codes within a real EHR dataset compared to existing imputation approaches and leads to increased performance on downstream tasks using the denoised data. In particular, when applying our method to a real world application, predicting hospital readmission from EHRs, our method achieves statistically significant improvement over all existing baselines.

Geometric properties of Transformer weights, particularly the unembedding matrix, have been widely useful in language model interpretability research. Yet, their utility for estimating downstream performance remains unclear. In this work, we systematically investigate the relationship between model performance and the unembedding matrix geometry, particularly its effective rank. Our experiments, involving a suite of 108 OLMo-style language models trained under controlled variation, reveal several key findings. While the best-performing models often exhibit a high effective rank, this trend is not universal across tasks and training setups. Contrary to prior work, we find that low effective rank does not cause late-stage performance degradation in small models, but instead co-occurs with it; we find adversarial cases where low-rank models do not exhibit saturation. Moreover, we show that effective rank is strongly influenced by pre-training hyperparameters, such as batch size and weight decay, which in-turn affect the model's performance. Lastly, extending our analysis to other geometric metrics and final-layer representation, we find that these metrics are largely aligned, but none can reliably predict downstream performance. Overall, our findings suggest that the model's geometry, as captured by existing metrics, primarily reflects training choices rather than performance.

Efficient operator scheduling is a fundamental challenge in software compilation and hardware synthesis. While recent differentiable approaches have sought to replace traditional ones like exact solvers or heuristics with gradient-based search, they typically rely on categorical distributions that fail to capture the ordinal nature of time and suffer from a parameter space that scales poorly. In this paper, we propose a novel differentiable framework, GauS, that models operator scheduling as a stochastic relaxation using Gaussian distributions, which fully utilize modern parallel computing devices like GPUs. By representing schedules as continuous Gaussian variables, we successfully capture the ordinal nature of time and reduce the optimization space by orders of magnitude. Our method is highly flexible to represent various objectives and constraints, which provides the first differentiable formulation for the complex pipelined scheduling problem. We evaluate our method on a range of benchmarks, demonstrating that Gaus achieves Pareto-optimal results.

The performance of LLM-based agents depends not only on the agent itself but also on the quality of the tool interfaces it consumes. While prior work has focused heavily on agent fine-tuning, tool interfaces-including natural language descriptions and parameter schemas-remain largely human-oriented and often become a bottleneck, especially when agents must select from large candidate tool sets. Existing approaches to improving tool interfaces rely on execution traces, which are frequently unavailable in cold-start or privacy-constrained settings, and typically optimize each tool independently, limiting scalability and generalization to unseen tools. We propose Trace-Free+, a curriculum learning framework that progressively transfers supervision from trace-rich settings to trace-free deployment, encouraging the model to abstract reusable interface-usage patterns and tool usage outcomes. To support this approach, we construct a large-scale dataset of high-quality tool interfaces using a structured workflow over a diverse collection of tools. Experiments on StableToolBench and RestBench show consistent gains on unseen tools, strong cross-domain generalization, and robustness as the number of candidate tools scales to over 100, demonstrating that tool interface optimization is a practical and deployable complement to agent fine-tuning.

Real-world requests to AI agents are fundamentally underspecified. Natural human communication relies on shared context and unstated constraints that speakers expect listeners to infer. Current agentic benchmarks test explicit instruction-following but fail to evaluate whether agents can reason about implicit requirements spanning accessibility needs, privacy boundaries, catastrophic risks, and contextual constraints. We present Implicit Intelligence, an evaluation framework testing whether AI agents can move beyond prompt-following to become genuine goal-fulfillers, paired with Agent-as-a-World (AaW), a harness where interactive worlds are defined in human-readable YAML files and simulated by language models. Our scenarios feature apparent simplicity in user requests, hidden complexity in correct solutions, and discoverability of constraints through environmental exploration. Evaluating 16 frontier and open-weight models across 205 scenarios, we find that even the best-performing model achieves only 48.3% scenario pass rate, revealing substantial room for improvement in bridging the gap between literal instruction-following and human-like contextual reasoning.

Medical image segmentation remains challenging due to limited annotations for training, ambiguous anatomical features, and domain shifts. While vision-language models such as CLIP offer strong cross-modal representations, their potential for dense, text-guided medical image segmentation remains underexplored. We present MedCLIPSeg, a novel framework that adapts CLIP for robust, data-efficient, and uncertainty-aware medical image segmentation. Our approach leverages patch-level CLIP embeddings through probabilistic cross-modal attention, enabling bidirectional interaction between image and text tokens and explicit modeling of predictive uncertainty. Together with a soft patch-level contrastive loss that encourages more nuanced semantic learning across diverse textual prompts, MedCLIPSeg effectively improves data efficiency and domain generalizability. Extensive experiments across 16 datasets spanning five imaging modalities and six organs demonstrate that MedCLIPSeg outperforms prior methods in accuracy, efficiency, and robustness, while providing interpretable uncertainty maps that highlight local reliability of segmentation results. This work demonstrates the potential of probabilistic vision-language modeling for text-driven medical image segmentation.

Diffusion models have shown promising capabilities in trajectory generation for planning in offline reinforcement learning (RL). However, conventional diffusion-based planning methods often fail to account for the fact that generating trajectories in RL requires unique consistency between transitions to ensure coherence in real environments. This oversight can result in considerable discrepancies between the generated trajectories and the underlying mechanisms of a real environment. To address this problem, we propose a novel diffusion-based planning method, termed as Diffusion Modulation via Environment Mechanism Modeling (DMEMM). DMEMM modulates diffusion model training by incorporating key RL environment mechanisms, particularly transition dynamics and reward functions. Experimental results demonstrate that DMEMM achieves state-of-the-art performance for planning with offline reinforcement learning.

Machine Learning as a Service (MLaaS) has emerged as a widely adopted paradigm for providing access to deep neural network (DNN) models, enabling users to conveniently leverage these models through standardized APIs. However, such services are highly vulnerable to Model Extraction Attacks (MEAs), where an adversary repeatedly queries a target model to collect input-output pairs and uses them to train a surrogate model that closely replicates its functionality. While numerous defense strategies have been proposed, verifying the ownership of a suspicious model with strict theoretical guarantees remains a challenging task. To address this gap, we introduce CREDIT, a certified ownership verification against MEAs. Specifically, we employ mutual information to quantify the similarity between DNN models, propose a practical verification threshold, and provide rigorous theoretical guarantees for ownership verification based on this threshold. We extensively evaluate our approach on several mainstream datasets across different domains and tasks, achieving state-of-the-art performance. Our implementation is publicly available at: https://github.com/LabRAI/CREDIT.

Graph neural networks (GNNs) have demonstrated superior performance in various applications, such as recommendation systems and financial risk management. However, deploying large-scale GNN models locally is particularly challenging for users, as it requires significant computational resources and extensive property data. Consequently, Machine Learning as a Service (MLaaS) has become increasingly popular, offering a convenient way to deploy and access various models, including GNNs. However, an emerging threat known as Model Extraction Attacks (MEAs) presents significant risks, as adversaries can readily obtain surrogate GNN models exhibiting similar functionality. Specifically, attackers repeatedly query the target model using subgraph inputs to collect corresponding responses. These input-output pairs are subsequently utilized to train their own surrogate models at minimal cost. Many techniques have been proposed to defend against MEAs, but most are limited to specific output levels (e.g., embedding or label) and suffer from inherent technical drawbacks. To address these limitations, we propose a novel ownership verification framework CITED which is a first-of-its-kind method to achieve ownership verification on both embedding and label levels. Moreover, CITED is a novel signature-based method that neither harms downstream performance nor introduces auxiliary models that reduce efficiency, while still outperforming all watermarking and fingerprinting approaches. Extensive experiments demonstrate the effectiveness and robustness of our CITED framework. Code is available at: https://github.com/LabRAI/CITED.

Ideas generated by independent samples of humans tend to be more diverse than ideas generated from independent LLM samples, raising concerns that widespread reliance on LLMs could homogenize ideation and undermine innovation at a societal level. Drawing on cognitive psychology, we identify (both theoretically and empirically) two mechanisms undermining LLM idea diversity. First, at the individual level, LLMs exhibit fixation just as humans do, where early outputs constrain subsequent ideation. Second, at the collective level, LLMs aggregate knowledge into a unified distribution rather than exhibiting the knowledge partitioning inherent to human populations, where each person occupies a distinct region of the knowledge space. Through four studies, we demonstrate that targeted prompting interventions can address each mechanism independently: Chain-of-Thought (CoT) prompting reduces fixation by encouraging structured reasoning (only in LLMs, not humans), while ordinary personas (versus "creative entrepreneurs" such as Steve Jobs) improve knowledge partitioning by serving as diverse sampling cues, anchoring generation in distinct regions of the semantic space. Combining both approaches produces the highest idea diversity, outperforming humans. These findings offer a theoretically grounded framework for understanding LLM idea diversity and practical strategies for human-AI collaborations that leverage AI's efficiency without compromising the diversity essential to a healthy innovation ecosystem.

In tabular Markov decision processes (MDPs) with perfect state observability, each trajectory provides active samples from the transition distributions conditioned on state-action pairs. Consequently, accurate model estimation depends on how the exploration policy allocates visitation frequencies in accordance with the intrinsic complexity of each transition distribution. Building on recent work on coverage-based exploration, we introduce a parameterized family of decomposable and concave objective functions $U_κ$ that explicitly incorporate both intrinsic estimation complexity and extrinsic visitation frequency. Moreover, the curvature $κ$ provides a unified treatment of various global objectives, such as the average-case and worst-case estimation error objectives. Using the closed-form characterization of the gradient of $U_κ$, we propose $κ$-Explorer, an active exploration algorithm that performs Frank-Wolfe-style optimization over state-action occupancy measures. The diminishing-returns structure of $U_κ$ naturally prioritizes underexplored and high-variance transitions, while preserving smoothness properties that enable efficient optimization. We establish tight regret guarantees for $κ$-Explorer and further introduce a fully online and computationally efficient surrogate algorithm for practical use. Experiments on benchmark MDPs demonstrate that $κ$-Explorer provides superior performance compared to existing exploration strategies.

We study distributionally robust online learning, where a risk-averse learner updates decisions sequentially to guard against worst-case distributions drawn from a Wasserstein ambiguity set centered at past observations. While this paradigm is well understood in the offline setting through Wasserstein Distributionally Robust Optimization (DRO), its online extension poses significant challenges in both convergence and computation. In this paper, we address these challenges. First, we formulate the problem as an online saddle-point stochastic game between a decision maker and an adversary selecting worst-case distributions, and propose a general framework that converges to a robust Nash equilibrium coinciding with the solution of the corresponding offline Wasserstein DRO problem. Second, we address the main computational bottleneck, which is the repeated solution of worst-case expectation problems. For the important class of piecewise concave loss functions, we propose a tailored algorithm that exploits problem geometry to achieve substantial speedups over state-of-the-art solvers such as Gurobi. The key insight is a novel connection between the worst-case expectation problem, an inherently infinite-dimensional optimization problem, and a classical and tractable budget allocation problem, which is of independent interest.

To steer language models towards truthful outputs on tasks which are beyond human capability, previous work has suggested training models on easy tasks to steer them on harder ones (easy-to-hard generalization), or using unsupervised training algorithms to steer models with no external labels at all (unsupervised elicitation). Although techniques from both paradigms have been shown to improve model accuracy on a wide variety of tasks, we argue that the datasets used for these evaluations could cause overoptimistic evaluation results. Unlike many real-world datasets, they often (1) have no features with more salience than truthfulness, (2) have balanced training sets, and (3) contain only data points to which the model can give a well-defined answer. We construct datasets that lack each of these properties to stress-test a range of standard unsupervised elicitation and easy-to-hard generalization techniques. We find that no technique reliably performs well on any of these challenges. We also study ensembling and combining easy-to-hard and unsupervised techniques, and find they only partially mitigate performance degradation due to these challenges. We believe that overcoming these challenges should be a priority for future work on unsupervised elicitation.

Neural simulators promise efficient surrogates for physics simulation, but scaling them is bottlenecked by the prohibitive cost of generating high-fidelity training data. Pre-training on abundant off-the-shelf geometries offers a natural alternative, yet faces a fundamental gap: supervision on static geometry alone ignores dynamics and can lead to negative transfer on physics tasks. We present GeoPT, a unified pre-trained model for general physics simulation based on lifted geometric pre-training. The core idea is to augment geometry with synthetic dynamics, enabling dynamics-aware self-supervision without physics labels. Pre-trained on over one million samples, GeoPT consistently improves industrial-fidelity benchmarks spanning fluid mechanics for cars, aircraft, and ships, and solid mechanics in crash simulation, reducing labeled data requirements by 20-60% and accelerating convergence by 2$\times$. These results show that lifting with synthetic dynamics bridges the geometry-physics gap, unlocking a scalable path for neural simulation and potentially beyond. Code is available at https://github.com/Physics-Scaling/GeoPT.

Explainable artificial intelligence promises to yield insights into relevant features, thereby enabling humans to examine and scrutinize machine learning models or even facilitating scientific discovery. Considering the widespread technique of Shapley values, we find that purely data-driven operationalization of multivariate feature importance is unsuitable for such purposes. Even for simple problems with two features, spurious associations due to collider bias and suppression arise from considering one feature only in the observational context of the other, which can lead to misinterpretations. Causal knowledge about the data-generating process is required to identify and correct such misleading feature attributions. We propose cc-Shapley (causal context Shapley), an interventional modification of conventional observational Shapley values leveraging knowledge of the data's causal structure, thereby analyzing the relevance of a feature in the causal context of the remaining features. We show theoretically that this eradicates spurious association induced by collider bias. We compare the behavior of Shapley and cc-Shapley values on various, synthetic, and real-world datasets. We observe nullification or reversal of associations compared to univariate feature importance when moving from observational to cc-Shapley.

Autoregressive models enable tractable sampling from learned probability distributions, but their performance critically depends on the variable ordering used in the factorization via complexities of the resulting conditional distributions. We propose to learn the Markov random field describing the underlying data, and use the inferred graphical model structure to construct optimized variable orderings. We illustrate our approach on two-dimensional image-like models where a structure-aware ordering leads to restricted conditioning sets, thereby reducing model complexity. Numerical experiments on Ising models with discrete data demonstrate that graph-informed orderings yield higher-fidelity generated samples compared to naive variable orderings.

Heterogeneous treatment effects (HTEs) are increasingly estimated using machine learning models that produce highly personalized predictions of treatment effects. In practice, however, predicted treatment effects are rarely interpreted, reported, or audited at the individual level but, instead, are often aggregated to broader subgroups, such as demographic segments, risk strata, or markets. We show that such aggregation can induce systematic bias of the group-level causal effect: even when models for predicting the individual-level conditional average treatment effect (CATE) are correctly specified and trained on data from randomized experiments, aggregating the predicted CATEs up to the group level does not, in general, recover the corresponding group average treatment effect (GATE). We develop a unified statistical framework to detect and mitigate this form of group bias in randomized experiments. We first define group bias as the discrepancy between the model-implied and experimentally identified GATEs, derive an asymptotically normal estimator, and then provide a simple-to-implement statistical test. For mitigation, we propose a shrinkage-based bias-correction, and show that the theoretically optimal and empirically feasible solutions have closed-form expressions. The framework is fully general, imposes minimal assumptions, and only requires computing sample moments. We analyze the economic implications of mitigating detected group bias for profit-maximizing personalized targeting, thereby characterizing when bias correction alters targeting decisions and profits, and the trade-offs involved. Applications to large-scale experimental data at major digital platforms validate our theoretical results and demonstrate empirical performance.

Enterprise Retrieval-Augmented Generation (RAG) assistants operate in multi-turn, case-based workflows such as technical support and IT operations, where evaluation must reflect operational constraints, structured identifiers (e.g., error codes, versions), and resolution workflows. Existing RAG evaluation frameworks are primarily designed for benchmark-style or single-turn settings and often fail to capture enterprise-specific failure modes such as case misidentification, workflow misalignment, and partial resolution across turns. We present a case-aware LLM-as-a-Judge evaluation framework for enterprise multi-turn RAG systems. The framework evaluates each turn using eight operationally grounded metrics that separate retrieval quality, grounding fidelity, answer utility, precision integrity, and case/workflow alignment. A severity-aware scoring protocol reduces score inflation and improves diagnostic clarity across heterogeneous enterprise cases. The system uses deterministic prompting with strict JSON outputs, enabling scalable batch evaluation, regression testing, and production monitoring. Through a comparative study of two instruction-tuned models across short and long workflows, we show that generic proxy metrics provide ambiguous signals, while the proposed framework exposes enterprise-critical tradeoffs that are actionable for system improvement.

Recent research argues that exact recursive numeral systems optimize communicative efficiency by balancing a tradeoff between the size of the numeral lexicon and the average morphosyntactic complexity (roughly length in morphemes) of numeral terms. We argue that previous studies have not characterized the data in a fashion that accounts for the degree of complexity languages display. Using data from 52 genetically diverse languages and an annotation scheme distinguishing between predictable and unpredictable allomorphy (formal variation), we show that many of the world's languages are decisively less efficient than one would expect. We discuss the implications of our findings for the study of numeral systems and linguistic evolution more generally.

The universal approximation theorem establishes that neural networks can approximate any continuous function on a compact set. Later works in approximation theory provide quantitative approximation rates for ReLU networks on the class of $α$-Hölder functions $f: [0,1]^N \to \mathbb{R}$. The goal of this paper is to provide similar quantitative approximation results in the context of group equivariant learning, where the learned $α$-Hölder function is known to obey certain group symmetries. While there has been much interest in the literature in understanding the universal approximation properties of equivariant models, very few quantitative approximation results are known for equivariant models. In this paper, we bridge this gap by deriving quantitative approximation rates for several prominent group-equivariant and invariant architectures. The architectures that we consider include: the permutation-invariant Deep Sets architecture; the permutation-equivariant Sumformer and Transformer architectures; joint invariance to permutations and rigid motions using invariant networks based on frame averaging; and general bi-Lipschitz invariant models. Overall, we show that equally-sized ReLU MLPs and equivariant architectures are equally expressive over equivariant functions. Thus, hard-coding equivariance does not result in a loss of expressivity or approximation power in these models.

Deep neural networks (DNNs) have been increasingly explored for receiver design because they can handle complex environments without relying on explicit channel models. Nevertheless, because communication channels change rapidly, their distributions can shift over time, often making periodic retraining necessary. This paper proposes a zero-overhead online and continual learning framework for orthogonal frequency-division multiplexing (OFDM) neural receivers that directly detect the soft bits of received signals. Unlike conventional fine-tuning methods that rely on dedicated training intervals or full resource grids, our approach leverages existing demodulation reference signals (DMRS) to simultaneously enable signal demodulation and model adaptation. We introduce three pilot designs: fully randomized, hybrid, and additional pilots that flexibly support joint demodulation and learning. To accommodate these pilot designs, we develop two receiver architectures: (i) a parallel design that separates inference and fine-tuning for uninterrupted operation, and (ii) a forward-pass reusing design that reduces computational complexity. Simulation results show that the proposed method effectively tracks both slow and fast channel distribution variations without additional overhead, service interruption, or catastrophic performance degradation under distribution shift.

Flow-based generative models have become a strong framework for high-quality generative modeling, yet pretrained models are rarely used in their vanilla conditional form: conditional samples without guidance often appear diffuse and lack fine-grained detail due to the smoothing effects of neural networks. Existing guidance techniques such as classifier-free guidance (CFG) improve fidelity but double the inference cost and typically reduce sample diversity. We introduce Momentum Guidance (MG), a new dimension of guidance that leverages the ODE trajectory itself. MG extrapolates the current velocity using an exponential moving average of past velocities and preserves the standard one-evaluation-per-step cost. It matches the effect of standard guidance without extra computation and can further improve quality when combined with CFG. Experiments demonstrate MG's effectiveness across benchmarks. Specifically, on ImageNet-256, MG achieves average improvements in FID of 36.68% without CFG and 25.52% with CFG across various sampling settings, attaining an FID of 1.597 at 64 sampling steps. Evaluations on large flow-based models like Stable Diffusion 3 and FLUX.1-dev further confirm consistent quality enhancements across standard metrics.

Graph self-supervised learning (GSSL) has demonstrated strong potential for generating expressive graph embeddings without the need for human annotations, making it particularly valuable in domains with high labeling costs such as molecular graph analysis. However, existing GSSL methods mostly focus on node- or edge-level information, often ignoring chemically relevant substructures which strongly influence molecular properties. In this work, we propose Graph Semantic Predictive Network (GraSPNet), a hierarchical self-supervised framework that explicitly models both atomic-level and fragment-level semantics. GraSPNet decomposes molecular graphs into chemically meaningful fragments without predefined vocabularies and learns node- and fragment-level representations through multi-level message passing with masked semantic prediction at both levels. This hierarchical semantic supervision enables GraSPNet to learn multi-resolution structural information that is both expressive and transferable. Extensive experiments on multiple molecular property prediction benchmarks demonstrate that GraSPNet learns chemically meaningful representations and consistently outperforms state-of-the-art GSSL methods in transfer learning settings.

Byzantine Fault Tolerant (BFT) consensus forms the foundation of many modern blockchains striving for both high throughput and low latency. A growing bottleneck is transaction execution and validation on the critical path of consensus, which has led to modular decoupled designs that separate ordering from execution: Consensus orders only metadata, while transactions are executed and validated concurrently. While this approach improves performance, it can leave invalid transactions in the ledger, increasing storage costs and enabling new forms of strategic behavior. We present the first systematic study of this setting, providing a formal framework to reason about the interaction between consensus and execution. Using this framework, we show that the decoupled design enables a previously unidentified attack, which we term gaslighting. We prove a fundamental trade-off between resilience to this attack and resource capacity utilization, where both are impossible to achieve deterministically in the decoupled model. To address this trade-off, we discuss an intermediate model for leader-based protocols that is robust to gaslighting attacks while achieving high throughput and low latency.

Chain-of-Thought (CoT) prompting significantly improves reasoning in Large Language Models, yet the temporal dynamics of the underlying representation geometry remain poorly understood. We investigate these dynamics by applying Manifold Capacity Theory (MCT) to a compositional Boolean logic task, allowing us to quantify the linear separability of latent representations without the confounding factors of probe training. Our analysis reveals that reasoning manifests as a transient geometric pulse, where concept manifolds are untangled into linearly separable subspaces immediately prior to computation and rapidly compressed thereafter. This behavior diverges from standard linear probe accuracy, which remains high long after computation, suggesting a fundamental distinction between information that is merely retrievable and information that is geometrically prepared for processing. We interpret this phenomenon as \emph{Dynamic Manifold Management}, a mechanism where the model dynamically modulates representational capacity to optimize the bandwidth of the residual stream throughout the reasoning chain.

This article presents an evaluation of several machine learning methods applied to automated text classification, alongside the design of a demonstrative system for unbalanced document categorization and distribution. The study focuses on balancing classification accuracy with computational efficiency, a key consideration when integrating AI into real world automation pipelines. Three models of varying complexity were examined: a Naive Bayes classifier, a bidirectional LSTM network, and a fine tuned transformer based BERT model. The experiments reveal substantial differences in performance. BERT achieved the highest accuracy, consistently exceeding 99\%, but required significantly longer training times and greater computational resources. The BiLSTM model provided a strong compromise, reaching approximately 98.56\% accuracy while maintaining moderate training costs and offering robust contextual understanding. Naive Bayes proved to be the fastest to train, on the order of milliseconds, yet delivered the lowest accuracy, averaging around 94.5\%. Class imbalance influenced all methods, particularly in the recognition of minority categories. A fully functional demonstrative system was implemented to validate practical applicability, enabling automated routing of technical requests with throughput unattainable through manual processing. The study concludes that BiLSTM offers the most balanced solution for the examined scenario, while also outlining opportunities for future improvements and further exploration of transformer architectures.

Self-adaptive robots adjust their behaviors in response to unpredictable environmental changes. These robots often incorporate deep learning (DL) components into their software to support functionality such as perception, decision-making, and control, enhancing autonomy and self-adaptability. However, the inherent uncertainty of DL-enabled software makes it challenging to ensure its dependability in dynamic environments. Consequently, test generation techniques have been developed to test robot software, and classical mutation analysis injects faults into the software to assess the test suite's effectiveness in detecting the resulting failures. However, there is a lack of mutation analysis techniques to assess the effectiveness under the uncertainty inherent to DL-enabled software. To this end, we propose UAMTERS, an uncertainty-aware mutation analysis framework that introduces uncertainty-aware mutation operators to explicitly inject stochastic uncertainty into DL-enabled robotic software, simulating uncertainty in its behavior. We further propose mutation score metrics to quantify a test suite's ability to detect failures under varying levels of uncertainty. We evaluate UAMTERS across three robotic case studies, demonstrating that UAMTERS more effectively distinguishes test suite quality and captures uncertainty-induced failures in DL-enabled software.

We present DMCD (DataMap Causal Discovery), a two-phase causal discovery framework that integrates LLM-based semantic drafting from variable metadata with statistical validation on observational data. In Phase I, a large language model proposes a sparse draft DAG, serving as a semantically informed prior over the space of possible causal structures. In Phase II, this draft is audited and refined via conditional independence testing, with detected discrepancies guiding targeted edge revisions. We evaluate our approach on three metadata-rich real-world benchmarks spanning industrial engineering, environmental monitoring, and IT systems analysis. Across these datasets, DMCD achieves competitive or leading performance against diverse causal discovery baselines, with particularly large gains in recall and F1 score. Probing and ablation experiments suggest that these improvements arise from semantic reasoning over metadata rather than memorization of benchmark graphs. Overall, our results demonstrate that combining semantic priors with principled statistical verification yields a high-performing and practically effective approach to causal structure learning.

Advanced reasoning capabilities in Large Language Models (LLMs) have led to more frequent hallucinations; yet most mitigation work focuses on open-source models for post-hoc detection and parameter editing. The dearth of studies focusing on hallucinations in closed-source models is especially concerning, as they constitute the vast majority of models in institutional deployments. We introduce QueryBandits, a model-agnostic contextual bandit framework that adaptively learns online to select the optimal query-rewrite strategy by leveraging an empirically validated and calibrated reward function. Across 16 QA scenarios, our top QueryBandit (Thompson Sampling) achieves an 87.5% win rate over a No-Rewrite baseline and outperforms zero-shot static policies (e.g., Paraphrase or Expand) by 42.6% and 60.3%, respectively. Moreover, all contextual bandits outperform vanilla bandits across all datasets, with higher feature variance coinciding with greater variance in arm selection. This substantiates our finding that there is no single rewrite policy optimal for all queries. We also discover that certain static policies incur higher cumulative regret than No-Rewrite, indicating that an inflexible query-rewriting policy can worsen hallucinations. Thus, learning an online policy over semantic features with QueryBandits can shift model behavior purely through forward-pass mechanisms, enabling its use with closed-source models and bypassing the need for retraining or gradient-based adaptation.

Vision-language models (VLMs) are powerful but remain opaque black boxes. We introduce the first framework for transparent circuit tracing in VLMs to systematically analyze multimodal reasoning. By utilizing transcoders, attribution graphs, and attention-based methods, we uncover how VLMs hierarchically integrate visual and semantic concepts. We reveal that distinct visual feature circuits can handle mathematical reasoning and support cross-modal associations. Validated through feature steering and circuit patching, our framework proves these circuits are causal and controllable, laying the groundwork for more explainable and reliable VLMs.

This work presents Causal Drift Generator (CaDrift), a time-dependent synthetic data generator framework based on Structural Causal Models (SCMs). The framework produces a virtually infinite combination of data streams with controlled shift events and time-dependent data, making it a tool to evaluate methods under evolving data. CaDrift synthesizes various distributional and covariate shifts by drifting mapping functions of the SCM, which change underlying cause-and-effect relationships between features and the target. In addition, CaDrift models occasional perturbations by leveraging interventions in causal modeling. Experimental results show that, after distributional shift events, the accuracy of classifiers tends to drop, followed by a gradual retrieval, confirming the generator's effectiveness in simulating shifts. The framework has been made available on GitHub.

Phenotyping is fundamental to rare disease diagnosis, but manual curation of structured phenotypes from clinical notes is labor-intensive and difficult to scale. Existing artificial intelligence approaches typically optimize individual components of phenotyping but do not operationalize the full clinical workflow of extracting features from clinical text, standardizing them to Human Phenotype Ontology (HPO) terms, and prioritizing diagnostically informative HPO terms. We developed RARE-PHENIX, an end-to-end AI framework for rare disease phenotyping that integrates large language model-based phenotype extraction, ontology-grounded standardization to HPO terms, and supervised ranking of diagnostically informative phenotypes. We trained RARE-PHENIX using data from 2,671 patients across 11 Undiagnosed Diseases Network clinical sites, and externally validated it on 16,357 real-world clinical notes from Vanderbilt University Medical Center. Using clinician-curated HPO terms as the gold standard, RARE-PHENIX consistently outperformed a state-of-the-art deep learning baseline (PhenoBERT) across ontology-based similarity and precision-recall-F1 metrics in end-to-end evaluation (i.e., ontology-based similarity of 0.70 vs. 0.58). Ablation analyses demonstrated performance improvements with the addition of each module in RARE-PHENIX (extraction, standardization, and prioritization), supporting the value of modeling the full clinical phenotyping workflow. By modeling phenotyping as a clinically aligned workflow rather than a single extraction task, RARE-PHENIX provides structured, ranked phenotypes that are more concordant with clinician curation and has the potential to support human-in-the-loop rare disease diagnosis in real-world settings.

Reliable object manipulation requires understanding physical properties that vary across objects and environments. Vision-language model (VLM) planners can reason about friction and stability in general terms; however, they often cannot predict how a specific ball will roll on a particular surface or which stone will provide a stable foundation without direct experience. We present PhysMem, a memory framework that enables VLM robot planners to learn physical principles from interaction at test time, without updating model parameters. The system records experiences, generates candidate hypotheses, and verifies them through targeted interaction before promoting validated knowledge to guide future decisions. A central design choice is verification before application: the system tests hypotheses against new observations rather than applying retrieved experience directly, reducing rigid reliance on prior experience when physical conditions change. We evaluate PhysMem on three real-world manipulation tasks and simulation benchmarks across four VLM backbones. On a controlled brick insertion task, principled abstraction achieves 76% success compared to 23% for direct experience retrieval, and real-world experiments show consistent improvement over 30-minute deployment sessions.

Next-generation fusion facilities like ITER face a "data deluge," generating petabytes of multi-diagnostic signals daily that challenge manual analysis. We present a "signals-first" self-supervised framework for the automated extraction of coherent and transient modes from high-noise time-frequency data. We also develop a general-purpose method and tool for extracting coherent, quasi-coherent, and transient modes for fluctuation measurements in tokamaks by employing non-linear optimal techniques in multichannel signal processing with a fast neural network surrogate on fast magnetics, electron cyclotron emission, CO2 interferometers, and beam emission spectroscopy measurements from DIII-D. Results are tested on data from DIII-D, TJ-II, and non-fusion spectrograms. With an inference latency of 0.5 seconds, this framework enables real-time mode identification and large-scale automated database generation for advanced plasma control. Repository is in https://github.com/PlasmaControl/TokEye.

Vision-language-action (VLA) models unify perception, language, and control for embodied agents but face significant challenges in practical deployment due to rapidly increasing compute and memory demands, especially as models scale to longer horizons and larger backbones. To address these bottlenecks, we introduce QuantVLA, a training-free post-training quantization (PTQ) framework that, to our knowledge, is the first PTQ approach for VLA systems and the first to successfully quantize a diffusion transformer (DiT) action head. QuantVLA incorporates three scale-calibrated components: (1) a selective quantization layout that integerizes all linear layers in both the language backbone and the DiT while keeping attention projections in floating point to preserve the original operator schedule; (2) attention temperature matching, a lightweight per-head scaling mechanism that stabilizes attention logits and is folded into the dequantization scales at inference; and (3) output head balancing, a per-layer residual interface calibration that mitigates post-projection energy drift. The framework requires no additional training, uses only a small unlabeled calibration buffer, and supports integer kernels for low-bit weights and activations while leaving the architecture unchanged. Across representative VLA models on LIBERO, QuantVLA exceeds the task success rates of full-precision baselines, achieves about 70% relative memory savings on the quantized components, and delivers a 1.22x speedup in end-to-end inference latency, providing a practical pathway toward scalable low-bit embodied intelligence under strict compute, memory, and power constraints.

Time-series foundation models (TSFMs) have demonstrated strong generalization capabilities across diverse datasets and tasks. However, existing foundation models are typically pre-trained to enhance performance on specific tasks and often struggle to generalize to unseen tasks without fine-tuning. To address this limitation, we propose augmenting TSFMs with In-Context Learning (ICL) capabilities, enabling them to perform test-time inference by dynamically adapting to input-output relationships provided within the context. Our framework, In-Context Time-series Pre-training (ICTP), restructures the original pre-training data to equip the backbone TSFM with ICL capabilities, enabling adaptation to unseen tasks. Experiments demonstrate that ICT improves the performance of state-of-the-art TSFMs by approximately 11.4% on unseen tasks without requiring fine-tuning.

High-fidelity computational models of cardiac mechanics provide mechanistic insight into the heart function but are computationally prohibitive for routine clinical use. Surrogate models can accelerate simulations, but generalization across diverse anatomies is challenging, particularly in data-scarce settings. We propose a two-step framework that decouples geometric representation from learning the physics response, to enable shape-informed surrogate modeling under data-scarce conditions. First, a shape model learns a compact latent representation of left ventricular geometries. The learned latent space effectively encodes anatomies and enables synthetic geometries generation for data augmentation. Second, a neural field-based surrogate model, conditioned on this geometric encoding, is trained to predict ventricular displacement under external loading. The proposed architecture performs positional encoding by using universal ventricular coordinates, which improves generalization across diverse anatomies. Geometric variability is encoded using two alternative strategies, which are systematically compared: a PCA-based approach suitable for working with point cloud representations of geometries, and a DeepSDF-based implicit neural representation learned directly from point clouds. Overall, our results, obtained on idealized and patient-specific datasets, show that the proposed approaches allow for accurate predictions and generalization to unseen geometries, and robustness to noisy or sparsely sampled inputs.

Background: Childhood and adolescent overweight and obesity remain major public health concerns in the United States and are shaped by behavioral, household, and community factors. Their joint predictive structure at the population level remains incompletely characterized. Objectives: The study aims to identify multilevel predictors of overweight and obesity among U.S. adolescents and compare the predictive performance, calibration, and subgroup equity of statistical, machine-learning, and deep-learning models. Data and Methods: We analyze 18,792 children aged 10-17 years from the 2021 National Survey of Children's Health. Overweight/obesity is defined using BMI categories. Predictors included diet, physical activity, sleep, parental stress, socioeconomic conditions, adverse experiences, and neighborhood characteristics. Models include logistic regression, random forest, gradient boosting, XGBoost, LightGBM, multilayer perceptron, and TabNet. Performance is evaluated using AUC, accuracy, precision, recall, F1 score, and Brier score. Results: Discrimination range from 0.66 to 0.79. Logistic regression, gradient boosting, and MLP showed the most stable balance of discrimination and calibration. Boosting and deep learning modestly improve recall and F1 score. No model was uniformly superior. Performance disparities across race and poverty groups persist across algorithms. Conclusion: Increased model complexity yields limited gains over logistic regression. Predictors consistently span behavioral, household, and neighborhood domains. Persistent subgroup disparities indicate the need for improved data quality and equity-focused surveillance rather than greater algorithmic complexity.

Large Language Model (LLM) hallucinations are usually treated as defects of the model or its decoding strategy. Drawing on classical linguistics, we argue that a query's form can also shape a listener's (and model's) response. We operationalize this insight by constructing a 22-dimension query feature vector covering clause complexity, lexical rarity, and anaphora, negation, answerability, and intention grounding, all known to affect human comprehension. Using 369,837 real-world queries, we ask: Are there certain types of queries that make hallucination more likely? A large-scale analysis reveals a consistent "risk landscape": certain features such as deep clause nesting and underspecification align with higher hallucination propensity. In contrast, clear intention grounding and answerability align with lower hallucination rates. Others, including domain specificity, show mixed, dataset- and model-dependent effects. Thus, these findings establish an empirically observable query-feature representation correlated with hallucination risk, paving the way for guided query rewriting and future intervention studies.

Maintaining long-term wavefront stability is critical for the Habitable Worlds Observatory (HWO), which targets contrasts approaching $10^{-10}$ and therefore requires continuous dark-zone maintenance using high-order wavefront sensing and control (HOWFSC). Prior work has advanced HOWFSC algorithms and profiled candidate implementations on radiation-hardened processors, highlighting a substantial gap between the computational demands of LUVOIR-scale HOWFSC and the capabilities of current onboard spacecraft hardware. In this paper, we argue that this gap can be closed by offloading the HOWFSC pipeline to a dedicated co-flying compute satellite at Sun-Earth L2. This approach enables the use of modern, radiation-tolerant high-performance processors without increasing risk to the primary observatory. We show that such an architecture can increase the end-to-end control cadence from the sub-hertz regime typical of radiation-hardened onboard processing or ground-in-the-loop operation to tens and even hundreds of hertz. We evaluate commercial hardware platforms in terms of performance and feasibility, and we propose custom architectures that enable higher control frequencies with significant power consumption reductions. Finally, we outline system-level considerations for co-flying compute, including reliability, satellite integration, and inter-satellite communication constraints.

We study gap-dependent performance guarantees for nearly minimax-optimal algorithms in reinforcement learning with linear function approximation. While prior works have established gap-dependent regret bounds in this setting, existing analyses do not apply to algorithms that achieve the nearly minimax-optimal worst-case regret bound $\tilde{O}(d\sqrt{H^3K})$, where $d$ is the feature dimension, $H$ is the horizon length, and $K$ is the number of episodes. We bridge this gap by providing the first gap-dependent regret bound for the nearly minimax-optimal algorithm LSVI-UCB++ (He et al., 2023). Our analysis yields improved dependencies on both $d$ and $H$ compared to previous gap-dependent results. Moreover, leveraging the low policy-switching property of LSVI-UCB++, we introduce a concurrent variant that enables efficient parallel exploration across multiple agents and establish the first gap-dependent sample complexity upper bound for online multi-agent RL with linear function approximation, achieving linear speedup with respect to the number of agents.

Curriculum learning is a class of training strategies that organizes the data being exposed to a model by difficulty, gradually from simpler to more complex examples. This research explores a reverse curriculum generation approach that recursively decomposes complex datasets into simpler, more learnable components. We propose a teacher-student framework where the teacher is equipped with the ability to reason step-by-step, which is used to recursively generate easier versions of examples, enabling the student model to progressively master difficult tasks. We propose a novel scoring system to measure data difficulty based on its structural complexity and conceptual depth, allowing curriculum construction over decomposed data. Experiments on math datasets (MATH and AIME) and code generation datasets demonstrate that models trained with curricula generated by our approach exhibit superior performance compared to standard training on original datasets.

Simulating real personalities with large language models requires grounding generation in authentic personal data. Existing evaluation approaches rely on demographic surveys, personality questionnaires, or short AI-led interviews as proxies, but lack direct assessment against what individuals actually said. We address this gap with an interview-grounded evaluation framework for personality simulation at a large scale. We extract over 671,000 question-answer pairs from 23,000 verified interview transcripts across 1,000 public personalities, each with an average of 11.5 hours of interview content. We propose a multi-dimensional evaluation framework with four complementary metrics measuring content similarity, factual consistency, personality alignment, and factual knowledge retention. Through systematic comparison, we demonstrate that methods grounded in real interview data substantially outperform those relying solely on biographical profiles or the model's parametric knowledge. We further reveal a trade-off in how interview data is best utilized: retrieval-augmented methods excel at capturing personality style and response quality, while chronological-based methods better preserve factual consistency and knowledge retention. Our evaluation framework enables principled method selection based on application requirements, and our empirical findings provide actionable insights for advancing personality simulation research.

We study a discrete denoising diffusion framework that integrates a sample-efficient estimator of single-site conditionals with round-robin noising and denoising dynamics for generative modeling over discrete state spaces. Rather than approximating a discrete analog of a score function, our formulation treats single-site conditional probabilities as the fundamental objects that parameterize the reverse diffusion process. We employ a sample-efficient method known as Neural Interaction Screening Estimator (NeurISE) to estimate these conditionals in the diffusion dynamics. Controlled experiments on synthetic Ising models, MNIST, and scientific data sets produced by a D-Wave quantum annealer, synthetic Potts model and one-dimensional quantum systems demonstrate the proposed approach. On the binary data sets, these experiments demonstrate that the proposed approach outperforms popular existing methods including ratio-based approaches, achieving improved performance in total variation, cross-correlations, and kernel density estimation metrics.

Agentic AI systems are deployed with expectations of substantial productivity gains, yet rigorous empirical evidence reveals systematic discrepancies between pre-deployment expectations and post-deployment outcomes. We review controlled trials and independent validations across software engineering, clinical documentation, and clinical decision support to quantify this expectation-realisation gap. In software development, experienced developers expected a 24% speedup from AI tools but were slowed by 19% -- a 43 percentage-point calibration error. In clinical documentation, vendor claims of multi-minute time savings contrast with measured reductions of less than one minute per note, and one widely deployed tool showed no statistically significant effect. In clinical decision support, externally validated performance falls substantially below developer-reported metrics. These shortfalls are driven by workflow integration friction, verification burden, measurement construct mismatches, and systematic heterogeneity in treatment effects. The evidence motivates structured planning frameworks that require explicit, quantified benefit expectations with human oversight costs factored in.

Magnetic resonance spectroscopy (MRS) is used to quantify metabolites in vivo and estimate biomarkers for conditions ranging from neurological disorders to cancers. Quantifying low-concentration metabolites such as GABA ($γ$-aminobutyric acid) is challenging due to low signal-to-noise ratio (SNR) and spectral overlap. We investigate and validate deep learning for quantifying complex, low-SNR, overlapping signals from MEGA-PRESS spectra, devise a convolutional neural network (CNN) and a Y-shaped autoencoder (YAE), and select the best models via Bayesian optimisation on 10,000 simulated spectra from slice-profile-aware MEGA-PRESS simulations. The selected models are trained on 100,000 simulated spectra. We validate their performance on 144 spectra from 112 experimental phantoms containing five metabolites of interest (GABA, Glu, Gln, NAA, Cr) with known ground truth concentrations across solution and gel series acquired at 3 T under varied bandwidths and implementations. These models are further assessed against the widely used LCModel quantification tool. On simulations, both models achieve near-perfect agreement (small MAEs; regression slopes $\approx 1.00$, $R^2 \approx 1.00$). On experimental phantom data, errors initially increased substantially. However, modelling variable linewidths in the training data significantly reduced this gap. The best augmented deep learning models achieved a mean MAE for GABA over all phantom spectra of 0.151 (YAE) and 0.160 (FCNN) in max-normalised relative concentrations, outperforming the conventional baseline LCModel (0.220). A sim-to-real gap remains, but physics-informed data augmentation substantially reduced it. Phantom ground truth is needed to judge whether a method will perform reliably on real data.

Continuous Integration (CI) pipelines for embedded software sometimes fail during compilation, consuming significant developer time for debugging. We study four major open-source embedded system projects, spanning over 4000 build failures from the project's CI runs. We find that hardware dependencies account for the majority of compilation failures, followed by syntax errors and build-script issues. Most repairs need relatively small changes, making automated repair potentially suitable as long as the diverse setups and lack of test data can be handled. In this paper, we present PhantomRun, an automated framework that leverages large language models (LLMs) to generate and validate fixes for CI compilation failures. The framework addresses the challenge of diverse build infrastructures and tool chains across embedded system projects by providing an adaptation layer for GitHub Actions and GitLab CI and four different build systems. PhantomRun utilizes build logs, source code, historical fixes, and compiler error messages to synthesize fixes using LLMs. Our evaluations show that PhantomRun successfully repairs up to 45% of CI compilation failures across the targeted projects, demonstrating the viability of LLM-based repairs for embedded-system CI pipelines.

Large language models (LLMs) have been reported to linearly encode truthfulness, yet recent work questions this finding's generality. We reconcile these views with the truthfulness spectrum hypothesis: the representational space contains directions ranging from broadly domain-general to narrowly domain-specific. To test this hypothesis, we systematically evaluate probe generalization across five truth types (definitional, empirical, logical, fictional, and ethical), sycophantic and expectation-inverted lying, and existing honesty benchmarks. Linear probes generalize well across most domains but fail on sycophantic and expectation-inverted lying. Yet training on all domains jointly recovers strong performance, confirming that domain-general directions exist despite poor pairwise transfer. The geometry of probe directions explains these patterns: Mahalanobis cosine similarity between probes near-perfectly predicts cross-domain generalization (R^2=0.98). Concept-erasure methods further isolate truth directions that are (1) domain-general, (2) domain-specific, or (3) shared only across particular domain subsets. Causal interventions reveal that domain-specific directions steer more effectively than domain-general ones. Finally, post-training reshapes truth geometry, pushing sycophantic lying further from other truth types, suggesting a representational basis for chat models' sycophantic tendencies. Together, our results support the truthfulness spectrum hypothesis: truth directions of varying generality coexist in representational space, with post-training reshaping their geometry. Code for all experiments is provided in https://github.com/zfying/truth_spec.

Accurate delivery delay prediction is critical for maintaining operational efficiency and customer satisfaction across modern supply chains. Yet the increasing complexity of logistics networks, spanning multimodal transportation, cross-country routing, and pronounced regional variability, makes this prediction task inherently challenging. This paper introduces a multi-task deep learning model for delivery delay duration prediction in the presence of significant imbalanced data, where delayed shipments are rare but operationally consequential. The model embeds high-dimensional shipment features with dedicated embedding layers for tabular data, and then uses a classification-then-regression strategy to predict the delivery delay duration for on-time and delayed shipments. Unlike sequential pipelines, this approach enables end-to-end training, improves the detection of delayed cases, and supports probabilistic forecasting for uncertainty-aware decision making. The proposed approach is evaluated on a large-scale real-world dataset from an industrial partner, comprising more than 10 million historical shipment records across four major source locations with distinct regional characteristics. The proposed model is compared with traditional machine learning methods. Experimental results show that the proposed method achieves a mean absolute error of 0.67-0.91 days for delayed-shipment predictions, outperforming single-step tree-based regression baselines by 41-64% and two-step classify-then-regress tree-based models by 15-35%. These gains demonstrate the effectiveness of the proposed model in operational delivery delay forecasting under highly imbalanced and heterogeneous conditions.

Density functional theory (DFT) is the most widely used method for calculating molecular properties; however, its accuracy is often insufficient for quantitative predictions. Coupled-cluster (CC) theory is the most successful method for achieving accuracy beyond DFT and for predicting properties that closely align with experiment. It is known as the ''gold standard'' of quantum chemistry. Unfortunately, the high computational cost of CC limits its widespread applicability. In this work, we present the Molecular Orbital Learning (MōLe) architecture, an equivariant machine learning model that directly predicts CC's core mathematical objects, the excitation amplitudes, from the mean-field Hartree-Fock molecular orbitals as inputs. We test various aspects of our model and demonstrate its remarkable data efficiency and out-of-distribution generalization to larger molecules and off-equilibrium geometries, despite being trained only on small equilibrium geometries. Finally, we also examine its ability to reduce the number of cycles required to converge CC calculations. MōLe can set the foundations for high-accuracy wavefunction-based ML architectures to accelerate molecular design and complement force-field approaches.

Multi-agent systems (MAS) built on large language models (LLMs) have shown strong performance across many tasks. Most existing approaches improve only one aspect at a time, such as the communication topology, role assignment, or LLM routing, while treating each agent as a single, indivisible unit. This misses the opportunity to use mixtures of LLMs within an agent to strengthen role-specific abilities. We propose HieraMAS, a hierarchical collaboration framework that combines intra-node LLM mixtures with an inter-node communication topology. HieraMAS introduces supernodes, where each functional role is implemented by multiple heterogeneous LLMs using a propose-synthesis structure. Optimizing HieraMAS creates unique credit-assignment challenges: final task performance depends heavily on the underlying LLMs' capabilities, which can lead reinforcement methods to incorrectly reward suboptimal configurations. To address this, we use a two-stage algorithm: (1) multi-level reward attribution, which provides fine-grained feedback at both the node level and the overall system level; (2) graph classification for topology selection, which treats choosing the communication structure as a holistic decision rather than optimizing edges one by one. Experiments on reasoning and coding benchmarks show that HieraMAS substantially outperforms existing methods while also delivering better cost-performance trade-offs.

The rapid expansion of biomedical publications creates challenges for organizing knowledge and detecting emerging trends, underscoring the need for scalable and interpretable methods. Common clustering and topic modeling approaches such as K-means or LDA remain sensitive to initialization and prone to local optima, limiting reproducibility and evaluation. We propose a reformulation of a convex optimization based clustering algorithm that produces stable, fine-grained topics by selecting exemplars from the data and guaranteeing a global optimum. Applied to about 12,000 PubMed articles on aging and longevity, our method uncovers topics validated by medical experts. It yields interpretable topics spanning from molecular mechanisms to dietary supplements, physical activity, and gut microbiota. The method performs favorably, and most importantly, its reproducibility and interpretability distinguish it from common clustering approaches, including K-means, LDA, and BERTopic. This work provides a basis for developing scalable, web-accessible tools for knowledge discovery.

Recent text-to-image (T2I) diffusion models produce visually stunning images and demonstrate excellent prompt following. But do they perform well as synthetic vision data generators? In this work, we revisit the promise of synthetic data as a scalable substitute for real training sets and uncover a surprising performance regression. We generate large-scale synthetic datasets using state-of-the-art T2I models released between 2022 and 2025, train standard classifiers solely on this synthetic data, and evaluate them on real test data. Despite observable advances in visual fidelity and prompt adherence, classification accuracy on real test data consistently declines with newer T2I models as training data generators. Our analysis reveals a hidden trend: These models collapse to a narrow, aesthetic-centric distribution that undermines diversity and label-image alignment. Overall, our findings challenge a growing assumption in vision research, namely that progress in generative realism implies progress in data realism. We thus highlight an urgent need to rethink the capabilities of modern T2I models as reliable training data generators.

Recently, TabPFN has gained attention as a foundation model for tabular data. However, it struggles to integrate heterogeneous modalities such as images and text, which are common in domains like healthcare and marketing, thereby limiting its applicability. To address this, we present the Multi-Modal Prior-data Fitted Network (MMPFN), which extends TabPFN to handle tabular and non-tabular modalities in a unified manner. MMPFN comprises per-modality encoders, modality projectors, and pre-trained foundation models. The modality projectors serve as the critical bridge, transforming non-tabular embeddings into tabular-compatible tokens for unified processing. To this end, we introduce a multi-head gated MLP and a cross-attention pooler that extract richer context from non-tabular inputs while mitigates attention imbalance issue in multimodal learning. Extensive experiments on medical and general-purpose multimodal datasets demonstrate that MMPFN consistently outperforms competitive state-of-the-art methods and effectively exploits non-tabular modalities alongside tabular features. These results highlight the promise of extending prior-data fitted networks to the multimodal setting, offering a scalable and effective framework for heterogeneous data learning. The source code is available at https://github.com/too-z/MultiModalPFN.

Multi-die FPGAs enable device scaling beyond reticle limits but introduce severe interconnect overhead across die boundaries. Inter-die connections, commonly referred to as super-long lines (SLLs), incur high delay and consume scarce interposer interconnect resources, often dominating critical paths and complicating physical design. To address this, this work proposes an interconnect-aware logic resynthesis method that restructures the LUT-level netlist to reduce the number of SLLs. The resynthesis engine uses die partitioning information to apply logic resubstitutions, which simplifies local circuit structures and eliminates SLLs. By reducing the number of SLLs early in the design flow, prior to physical implementation, the proposed method shortens critical paths, alleviates pressure on scarce interposer interconnect resources, and improves overall physical design flexibility. We further build a tool flow for multi-die FPGAs by integrating the proposed resynthesis method with packing and placement. Experimental results on the EPFL benchmarks show that, compared with a state-of-the-art framework, the proposed method reduces the number of SLLs by up to 24.8% for a 2-die FPGA and up to 27.38% for a 3-die FPGA. On MCNC benchmarks, our tool flow achieves an average SLL reduction of 1.65% while preserving placement quality. On Koios benchmarks, where fewer removable SLLs exist, several designs still exhibit considerable inter-die edge reductions. Overall, the results confirm that reducing inter-die connections at the logic level is an effective approach for multi-die FPGAs.

We investigate what specific design choices enable successful online reinforcement learning (RL) on physical robots. Across 100 real-world training runs on three distinct robotic platforms, we systematically ablate algorithmic, systems, and experimental decisions that are typically left implicit in prior work. We find that some widely used defaults can be harmful, while a set of robust, readily adopted design choices within standard RL practice yield stable learning across tasks and hardware. These results provide the first large-sample empirical study of such design choices, enabling practitioners to deploy online RL with lower engineering effort.

Urban air pollution in megacities poses critical public health challenges, particularly in Delhi National Capital Region (NCR) where severe degradation affects millions. We present NEXUS (Neural Extraction and Unified Spatiotemporal) architecture for forecasting carbon monoxide, nitrogen oxide, and sulfur dioxide. Working with four years (2018--2021) of atmospheric data across sixteen spatial grids, NEXUS achieves R$^2$ exceeding 0.94 for CO, 0.91 for NO, and 0.95 for SO$_2$ using merely 18,748 parameters -- substantially fewer than SCINet (35,552), Autoformer (68,704), and FEDformer (298,080). The architecture integrates patch embedding, low-rank projections, and adaptive fusion mechanisms to decode complex atmospheric chemistry patterns. Our investigation uncovers distinct diurnal rhythms and pronounced seasonal variations, with winter months experiencing severe pollution episodes driven by temperature inversions and agricultural biomass burning. Analysis identifies critical meteorological thresholds, quantifies wind field impacts on pollutant dispersion, and maps spatial heterogeneity across the region. Extensive ablation experiments demonstrate each architectural component's role. NEXUS delivers superior predictive performance with remarkable computational efficiency, enabling real-time deployment for air quality monitoring systems.

We present Nacrith, a lossless compression system that combines a 135M-parameter transformer language model (SmolLM2-135M) with an ensemble of lightweight online predictors and a 32-bit arithmetic coder, achieving the best compression results among the systems evaluated in this study on natural language text. Beyond the base LLM-plus-arithmetic-coding paradigm, Nacrith introduces several contributions: (1) a CDF precision upgrade from 2^16 to 2^24 that eliminates ~75% of quantization overhead caused by minimum-probability floors in large vocabularies; (2) a token-level N-gram model for fast local predictions; (3) an adaptive log-space bias head correcting per-document LLM errors via online gradient descent; (4) confidence-based LLM skip for accelerating highly predictable tokens; (5) a hybrid binary format (NC06) extending neural compression to arbitrary binary files--to our knowledge a first among LLM-based compressors; (6) a llama cpp inference backend achieving ~7x faster single-token decode than PyTorch; (7) parallel multi-GPU compression across up to 8 workers; and (8) native KV cache sliding window reducing per-slide cost by ~37x. The system requires only ~500 MB of GGUF weights and ~1.2 GB VRAM per worker, running on consumer GPUs. On alice29 (Canterbury Corpus, 152 KB), Nacrith achieves 0.918 bits per byte (bpb)--outperforming gzip by 3.1x, bzip2 by 2.5x, CMIX v21 by 44%, and ts_zip by 20%, while compressing below the 0th-, 1st-, and 2nd-order byte-level Shannon entropy bounds. On enwik8 (100 MB), Nacrith achieves 0.9389 bpb (11.74%), surpassing ts_zip (~1.11 bpb) by 15% and FineZip (1.024 bpb) by 8% despite using a 60x smaller model with no fine-tuning. An out-of-distribution (OOD) evaluation on a document published after the model's training cutoff confirms these gains are not memorization artifacts, achieving 0.723 bpb on unseen text.

Interpreting human intent accurately is a central challenge in human-robot interaction (HRI) and a key requirement for achieving more natural and intuitive collaboration between humans and machines. This work presents a novel multimodal HRI framework that combines advanced vision-language models, speech processing, and fuzzy logic to enable precise and adaptive control of a Dobot Magician robotic arm. The proposed system integrates Florence-2 for object detection, Llama 3.1 for natural language understanding, and Whisper for speech recognition, providing users with a seamless and intuitive interface for object manipulation through spoken commands. By jointly addressing scene perception and action planning, the approach enhances the reliability of command interpretation and execution. Experimental evaluations conducted on consumer-grade hardware demonstrate a command execution accuracy of 75\%, highlighting both the robustness and adaptability of the system. Beyond its current performance, the proposed architecture serves as a flexible and extensible foundation for future HRI research, offering a practical pathway toward more sophisticated and natural human-robot collaboration through tightly coupled speech and vision-language processing.

Machine Unlearning (MU) enables Large Language Models (LLMs) to remove unsafe or outdated information. However, existing work assumes that all facts are equally forgettable and largely ignores whether the forgotten knowledge originates from pretraining or supervised fine-tuning (SFT). In this paper, we introduce DUAL (Dual Unlearning Evaluation across Training Stages), a benchmark of 28.6k Wikidata-derived triplets annotated with fact popularity using Wikipedia link counts and LLM-based salience scores. Our experiments show that pretrained and SFT models respond differently to unlearning. An SFT step on the forget data yields smoother forgetting, more stable tuning, and 10-50% higher retention, while direct unlearning on pretrained models remains unstable and prone to relearning or catastrophic forgetting.

Self-speculative decoding (SSD) accelerates LLM inference by skipping layers to create an efficient draft model, yet existing methods often rely on static heuristics that ignore the dynamic computational overhead of attention in long-context scenarios. We propose KnapSpec, a training-free framework that reformulates draft model selection as a knapsack problem to maximize tokens-per-time throughput. By decoupling Attention and MLP layers and modeling their hardware-specific latencies as functions of context length, KnapSpec adaptively identifies optimal draft configurations on the fly via a parallel dynamic programming algorithm. Furthermore, we provide the first rigorous theoretical analysis establishing cosine similarity between hidden states as a mathematically sound proxy for the token acceptance rate. This foundation allows our method to maintain high drafting faithfulness while navigating the shifting bottlenecks of real-world hardware. Our experiments on Qwen3 and Llama3 demonstrate that KnapSpec consistently outperforms state-of-the-art SSD baselines, achieving up to 1.47x wall-clock speedup across various benchmarks. Our plug-and-play approach ensures high-speed inference for long sequences without requiring additional training or compromising the target model's output distribution.

AI agents increasingly act on behalf of humans, yet no existing system provides a tamper-evident, independently verifiable record of what they did. As regulations such as the EU AI Act begin mandating automatic logging for high-risk AI systems, this gap carries concrete consequences -- especially for agents running on personal hardware, where no centralized provider controls the log. Extending Floridi's informational rights framework from data about individuals to actions performed on their behalf, this paper proposes the Right to History: the principle that individuals are entitled to a complete, verifiable record of every AI agent action on their own hardware. The paper formalizes this principle through five system invariants with structured proof sketches, and implements it in PunkGo, a Rust sovereignty kernel that unifies RFC 6962 Merkle tree audit logs, capability-based isolation, energy-budget governance, and a human-approval mechanism. Adversarial testing confirms all five invariants hold. Performance evaluation shows sub-1.3 ms median action latency, ~400 actions/sec throughput, and 448-byte Merkle inclusion proofs at 10,000 log entries.

The evaluation of Large Language Models (LLMs) for code generation relies heavily on the quality and robustness of test cases. However, existing benchmarks often lack coverage for subtle corner cases, allowing incorrect solutions to pass. To bridge this gap, we propose CodeHacker, an automated agent framework dedicated to generating targeted adversarial test cases that expose latent vulnerabilities in program submissions. Mimicking the hack mechanism in competitive programming, CodeHacker employs a multi-strategy approach, including stress testing, anti-hash attacks, and logic-specific targeting to break specific code submissions. To ensure the validity and reliability of these attacks, we introduce a Calibration Phase, where the agent iteratively refines its own Validator and Checker via self-generated adversarial probes before evaluating contestant code.Experiments demonstrate that CodeHacker significantly improves the True Negative Rate (TNR) of existing datasets, effectively filtering out incorrect solutions that were previously accepted. Furthermore, generated adversarial cases prove to be superior training data, boosting the performance of RL-trained models on benchmarks like LiveCodeBench.

Crystal modeling spans a family of conditional and unconditional generation tasks across different modalities, including crystal structure prediction (CSP) and \emph{de novo} generation (DNG). While recent deep generative models have shown promising performance, they remain largely task-specific, lacking a unified framework that shares crystal representations across different generation tasks. To address this limitation, we propose \emph{Multimodal Crystal Flow (MCFlow)}, a unified multimodal flow model that realizes multiple crystal generation tasks as distinct inference trajectories via independent time variables for atom types and crystal structures. To enable multimodal flow in a standard transformer model, we introduce a composition- and symmetry-aware atom ordering with hierarchical permutation augmentation, injecting strong compositional and crystallographic priors without explicit structural templates. Experiments on the MP-20 and MPTS-52 benchmarks show that MCFlow achieves competitive performance against task-specific baselines across multiple crystal generation tasks.

The discovery of novel proteins relies on sensitive protein identification, for which de novo peptide sequencing (DNPS) from mass spectra is a crucial approach. While deep learning has advanced DNPS, existing models inadequately enforce the fundamental mass consistency constraint, that a predicted peptide's mass must match the experimental measured precursor mass. Previous DNPS methods often treat this critical information as a simple input feature or use it in post-processing, leading to numerous implausible predictions that do not adhere to this fundamental physical property. To address this limitation, we introduce DiffuNovo, a novel regressor-guided diffusion model for de novo peptide sequencing that provides explicit peptide-level mass control. Our approach integrates the mass constraint at two critical stages: during training, a novel peptide-level mass loss guides model optimization, while at inference, regressor-based guidance from gradient-based updates in the latent space steers the generation to compel the predicted peptide adheres to the mass constraint. Comprehensive evaluations on established benchmarks demonstrate that DiffuNovo surpasses state-of-the-art methods in DNPS accuracy. Additionally, as the first DNPS model to employ a diffusion model as its core backbone, DiffuNovo leverages the powerful controllability of diffusion architecture and achieves a significant reduction in mass error, thereby producing much more physically plausible peptides. These innovations represent a substantial advancement toward robust and broadly applicable DNPS. The source code is available in the supplementary material.

Model merging aims to integrate multiple task-specific fine-tuned models derived from a shared pre-trained checkpoint into a single multi-task model without additional training. Despite extensive research, task interference remains a major obstacle that often undermines the performance of merged models. In this paper, we propose ESM (Essential Subspace Merging) , a robust framework for effective model merging. We begin by performing Principal Component Analysis (PCA) on feature shifts induced by parameter updates. The resulting principal directions span an essential subspace that dominantly influences feature representations. Each task's parameter update matrix is projected onto its respective essential subspace for low-rank decomposition before merging. This methodology mitigates inter-task interference while preserving core task-specific functionality. Furthermore, we introduce a multi-level polarized scaling strategy that amplifies parameters containing critical knowledge and suppresses redundant ones, preventing essential knowledge from being overwhelmed during fusion. Extensive experiments across multiple task sets and model scales demonstrate that our method achieves state-of-the-art performance in multi-task model merging.

Knowledge editing in Large Language Models (LLMs) aims to update the model's prediction for a specific query to a desired target while preserving its behavior on all other inputs. This process typically involves two stages: identifying the layer to edit and performing the parameter update. Intuitively, different queries may localize knowledge at different depths of the model, resulting in different sample-wise editing performance for a fixed editing layer. In this work, we hypothesize the existence of fixed golden layers that can achieve near-optimal editing performance similar to sample-wise optimal layers. To validate this hypothesis, we provide empirical evidence by comparing golden layers against ground-truth sample-wise optimal layers. Furthermore, we show that golden layers can be reliably identified using a proxy dataset and generalize effectively to unseen test set queries across datasets. Finally, we propose a novel method, namely Layer Gradient Analysis (LGA) that estimates golden layers efficiently via gradient-attribution, avoiding extensive trial-and-error across multiple editing runs. Extensive experiments on several benchmark datasets demonstrate the effectiveness and robustness of our LGA approach across different LLM types and various knowledge editing methods.

The democratization of Large Language Models (LLMs) has given rise to ``Vibe Coding," a workflow where novice programmers prioritize semantic intent over syntactic implementation. While this lowers barriers to entry, we hypothesize that without pedagogical guardrails, it is fundamentally misaligned with cognitive skill acquisition. Drawing on the distinction between Cognitive Offloading and Cognitive Outsourcing, we argue that unrestricted AI encourages novices to outsource the Intrinsic Cognitive Load required for schema formation, rather than merely offloading Extraneous Load. This accumulation of ``Epistemic Debt" creates ``Fragile Experts" whose high functional utility masks critically low corrective competence. To quantify and mitigate this debt, we conducted a between-subjects experiment (N=78) using a custom Cursor IDE plugin backed by Claude 3.5 Sonnet. Participants represented "AI-Native" learners across three conditions: Manual (Control), Unrestricted AI (Outsourcing), and Scaffolded AI (Offloading). The Scaffolded condition utilized a novel ``Explanation Gate," leveraging a real-time LLM-as-a-Judge framework to enforce a ``Teach-Back" protocol before generated code could be integrated. Results reveal a ``Collapse of Competence": while Unrestricted AI users matched the productivity of the Scaffolded group (p < .001 vs. Manual), they suffered a 77% failure rate in a subsequent AI-Blackout maintenance task, compared to only 39% in the Scaffolded group. Qualitative analysis suggests that successful vibe coders naturally engage in self-scaffolding, treating the AI as a consultant rather than a contractor. We discuss the implications for the maintainability of AI-generated software and propose that future learning systems must enforce Metacognitive Friction to prevent the mass production of unmaintainable code.

AI kernel compilation for edge devices depends on the compiler's ability to exploit parallelism and hide memory latency in the presence of hierarchical memory and explicit data movement. This paper reports a benchmark methodology and corresponding results for three compiler-controlled mechanisms in an MLIR-based compilation pipeline: vectorization (Vec), multi-threading (MT) across hardware contexts, and double buffering (DB) using ping--pong scratchpad buffers to overlap DMA transfers with compute. Using Triton/Inductor-generated kernels, we present an ablation ladder that separates the contribution of Vec, MT, and DB, and we quantify how MT speedup scales with problem size using GELU as a representative activation kernel. The results show that vectorization provides the primary gain for bandwidth-sensitive kernels, MT delivers substantial improvements once scheduling overhead is amortized, and DB provides additional benefit when transfers and compute can be overlapped (i.e., outside the extremes of purely memory-bound or purely compute-bound behavior).

The growing reliance on AI-identified digital evidence raises significant concerns about its reliability, particularly as large language models (LLMs) are increasingly integrated into forensic investigations. This paper proposes a structured framework that automates forensic artifact extraction, refines data through LLM-driven analysis, and validates results using a Digital Forensic Knowledge Graph (DFKG). Evaluated on a 13 GB forensic image dataset containing 61 applications, 2,864 databases, and 5,870 tables, the framework ensures artifact traceability and evidentiary consistency through deterministic Unique Identifiers (UIDs) and forensic cross-referencing. We propose this methodology to address challenges in ensuring the credibility and forensic integrity of AI-identified evidence, reducing classification errors, and advancing scalable, auditable methodologies. A comprehensive case study on this dataset demonstrates the framework's effectiveness, achieving over 95 percent accuracy in artifact extraction, strong support of chain-of-custody adherence, and robust contextual consistency in forensic relationships. Key results validate the framework's ability to enhance reliability, reduce errors, and establish a legally sound paradigm for AI-assisted digital forensics.

Microfluidic devices have emerged as powerful tools in various laboratory applications, but the complexity of their design limits accessibility for many practitioners. While progress has been made in microfluidic design automation (MFDA), a practical and intuitive solution is still needed to connect microfluidic practitioners with MFDA techniques. This work introduces the first practical application of large language models (LLMs) in this context, providing a preliminary demonstration. Building on prior research in hardware description language (HDL) code generation with LLMs, we propose an initial methodology to convert natural language microfluidic device specifications into system-level structural Verilog netlists. We demonstrate the feasibility of our approach by generating structural netlists for practical benchmarks representative of typical microfluidic designs with correct functional flow and an average syntactical accuracy of 88%.

Hierarchical Vision-Language-Action (VLA) models have rapidly become a dominant paradigm for robotic manipulation. It typically comprising a Vision-Language backbone for perception and understanding, together with a generative policy for action generation. However, its performance is increasingly bottlenecked by the action generation proceess. (i) Low inference efficiency. A pronounced distributional gap between isotropic noise priors and target action distributions, which increases denoising steps and the incidence of infeasible samples. (ii) Poor robustness. Existing policies condition solely on the current observation, neglecting the constraint of history sequence and thus lacking awareness of task progress and temporal consistency. To address these issues, we introduce OptimusVLA, a dual-memory VLA framework with Global Prior Memory (GPM) and Local Consistency Memory (LCM). GPM replaces Gaussian noise with task-level priors retrieved from semantically similar trajectories, thereby shortening the generative path and reducing the umber of function evaluations (NFE). LCM dynamically models executed action sequence to infer task progress and injects a learned consistency constraint that enforces temporal coherence and smoothness of trajectory. Across three simulation benchmarks, OptimusVLA consistently outperforms strong baselines: it achieves 98.6% average success rate on LIBERO, improves over pi_0 by 13.5% on CALVIN, and attains 38% average success rate on RoboTwin 2.0 Hard. In Real-World evaluation, OptimusVLA ranks best on Generalization and Long-horizon suites, surpassing pi_0 by 42.9% and 52.4%, respectively, while delivering 2.9x inference speedup.

Class imbalance, overlap, and noise degrade data quality, reduce model reliability, and limit generalization. Although widely studied in binary classification, these issues remain underexplored in multi-class settings, where complex inter-class relationships make minority-majority structures unclear and traditional clustering fails to capture distribution shape. Approaches that rely only on geometric distances risk removing informative samples and generating low-quality synthetic data, while binarization approaches treat imbalance locally and ignore global inter-class dependencies. At the algorithmic level, ensembles struggle to integrate weak classifiers, leading to limited robustness. This paper proposes IMOVNO+ (IMbalance-OVerlap-NOise+ Algorithm-Level Optimization), a two-level framework designed to jointly enhance data quality and algorithmic robustness for binary and multi-class tasks. At the data level, first, conditional probability is used to quantify the informativeness of each sample. Second, the dataset is partitioned into core, overlapping, and noisy regions. Third, an overlapping-cleaning algorithm is introduced that combines Z-score metrics with a big-jump gap distance. Fourth, a smart oversampling algorithm based on multi-regularization controls synthetic sample proximity, preventing new overlaps. At the algorithmic level, a meta-heuristic prunes ensemble classifiers to reduce weak-learner influence. IMOVNO+ was evaluated on 35 datasets (13 multi-class, 22 binary). Results show consistent superiority over state-of-the-art methods, approaching 100% in several cases. For multi-class data, IMOVNO+ achieves gains of 37-57% in G-mean, 25-44% in F1-score, 25-39% in precision, and 26-43% in recall. In binary tasks, it attains near-perfect performance with improvements of 14-39%. The framework handles data scarcity and imbalance from collection and privacy limits.

Pro-inflammatory peptides (PIPs) play critical roles in immune signaling and inflammation but are difficult to identify experimentally due to costly and time-consuming assays. To address this challenge, we present KEMP-PIP, a hybrid machine learning framework that integrates deep protein embeddings with handcrafted descriptors for robust PIP prediction. Our approach combines contextual embeddings from pretrained ESM protein language models with multi-scale k-mer frequencies, physicochemical descriptors, and modlAMP sequence features. Feature pruning and class-weighted logistic regression manage high dimensionality and class imbalance, while ensemble averaging with an optimized decision threshold enhances the sensitivity--specificity balance. Through systematic ablation studies, we demonstrate that integrating complementary feature sets consistently improves predictive performance. On the standard benchmark dataset, KEMP-PIP achieves an MCC of 0.505, accuracy of 0.752, and AUC of 0.762, outperforming ProIn-fuse, MultiFeatVotPIP, and StackPIP. Relative to StackPIP, these results represent improvements of 9.5% in MCC and 4.8% in both accuracy and AUC. The KEMP-PIP web server is freely available at https://nilsparrow1920-kemp-pip.hf.space/ and the full implementation at https://github.com/S18-Niloy/KEMP-PIP.

Today's distributed systems operate in complex environments that inevitably involve faults and even adversarial behaviors. Predicting their performance under such environments directly from formal designs remains a longstanding challenge. We present the first formal framework that systematically enables performance prediction of distributed systems across diverse faulty scenarios. Our framework features a fault injector together with a wide range of faults, reusable as a library, and model compositions that integrate the system and the fault injector into a unified model suitable for statistical analysis of performance properties such as throughput and latency. We formalize the framework in Maude and implement it as an automated tool, PERF. Applied to representative distributed systems, PERF accurately predicts system performance under varying fault settings, with estimations from formal designs consistent with evaluations on real deployments.

Reinforcement Learning with verifiable rewards (RLVR) has emerged as a primary learning paradigm for enhancing the reasoning capabilities of multi-modal large language models (MLLMs). However, during RL training, the enormous state space of MLLM and sparse rewards often leads to entropy collapse, policy degradation, or over-exploitation of suboptimal behaviors. This necessitates an exploration strategy that maintains productive stochasticity while avoiding the drawbacks of uncontrolled random sampling, yielding inefficient exploration. In this paper, we propose CalibRL, a hybrid-policy RLVR framework that supports controllable exploration with expert guidance, enabled by two key mechanisms. First, a distribution-aware advantage weighting scales updates by group rareness to calibrate the distribution, therefore preserving exploration. Meanwhile, the asymmetric activation function (LeakyReLU) leverages the expert knowledge as a calibration baseline to moderate overconfident updates while preserving their corrective direction. CalibRL increases policy entropy in a guided manner and clarifies the target distribution by estimating the on-policy distribution through online sampling. Updates are driven by these informative behaviors, avoiding convergence to erroneous patterns. Importantly, these designs help alleviate the distributional mismatch between the model's policy and expert trajectories, thereby achieving a more stable balance between exploration and exploitation. Extensive experiments across eight benchmarks, including both in-domain and out-of-domain settings, demonstrate consistent improvements, validating the effectiveness of our controllable hybrid-policy RLVR training. Code is available at https://github.com/zhh6425/CalibRL.

AI agents increasingly require direct, structured access to application data and actions, but production deployments still struggle to express and verify the governance properties that matter in practice: least-privilege authorization, controlled write execution, predictable failure handling, abuse resistance, and auditability. This paper introduces OpenPort Protocol (OPP), a governance-first specification for exposing application tools through a secure server-side gateway that is model- and runtime-neutral and can bind to existing tool ecosystems. OpenPort defines authorization-dependent discovery, stable response envelopes with machine-actionable \texttt{agent.*} reason codes, and an authorization model combining integration credentials, scoped permissions, and ABAC-style policy constraints. For write operations, OpenPort specifies a risk-gated lifecycle that defaults to draft creation and human review, supports time-bounded auto-execution under explicit policy, and enforces high-risk safeguards including preflight impact binding and idempotency. To address time-of-check/time-of-use drift in delayed approval flows, OpenPort also specifies an optional State Witness profile that revalidates execution-time preconditions and fails closed on state mismatch. Operationally, the protocol requires admission control (rate limits/quotas) with stable 429 semantics and structured audit events across allow/deny/fail paths so that client recovery and incident analysis are deterministic. We present a reference runtime and an executable governance toolchain (layered conformance profiles, negative security tests, fuzz/abuse regression, and release-gate scans) and evaluate the core profile at a pinned release tag using artifact-based, externally reproducible validation.

Crystal structure prediction is a long-standing challenge in materials science, with most data-driven methods developed for inorganic systems. This leaves an important gap for organic crystals, which are central to pharmaceuticals, polymers, and functional materials, but present unique challenges, such as larger unit cells and strict chemical connectivity. We introduce a flow-matching model for predicting organic crystal structures directly from molecular graphs. The architecture integrates molecular connectivity with periodic boundary conditions while preserving the symmetries of crystalline systems. A bond-aware loss guides the model toward realistic local chemistry by enforcing distributions of bond lengths and connectivity. To support reliable and efficient training, we built a curated dataset of organic crystals, along with a preprocessing pipeline that precomputes bonds and edges, substantially reducing computational overhead during both training and inference. Experiments show that our method achieves a Match Rate more than 10 times higher than existing baselines while requiring fewer sampling steps for inference. These results establish generative modeling as a practical and scalable framework for organic crystal structure prediction.

Bridge periodic inspection records contain sensitive information about public infrastructure, making cross-organizational data sharing impractical under existing data governance constraints. We propose a federated framework for estimating a Continuous-Time Markov Chain (CTMC) hazard model of bridge deterioration, enabling municipalities to collaboratively train a shared benchmark model without transferring raw inspection records. Each User holds local inspection data and trains a log-linear hazard model over three deterioration-direction transitions -- Good$\to$Minor, Good$\to$Severe, and Minor$\to$Severe -- with covariates for bridge age, coastline distance, and deck area. Local optimization is performed via mini-batch stochastic gradient descent on the CTMC log-likelihood, and only a 12-dimensional pseudo-gradient vector is uploaded to a central server per communication round. The server aggregates User updates using sample-weighted Federated Averaging (FedAvg) with momentum and gradient clipping. All experiments in this paper are conducted on fully synthetic data generated from a known ground-truth parameter set with region-specific heterogeneity, enabling controlled evaluation of federated convergence behaviour. Simulation results across heterogeneous Users show consistent convergence of the average negative log-likelihood, with the aggregated gradient norm decreasing as User scale increases. Furthermore, the federated update mechanism provides a natural participation incentive: Users who register their local inspection datasets on a shared technical-standard platform receive in return the periodically updated global benchmark parameters -- information that cannot be obtained from local data alone -- thereby enabling evidence-based life-cycle planning without surrendering data sovereignty.

Standard evaluations of backdoor attacks on text-to-image (T2I) models primarily measure trigger activation and visual fidelity. We challenge this paradigm, demonstrating that encoder-side poisoning induces persistent, trigger-free semantic corruption that fundamentally reshapes the representation manifold. We trace this vulnerability to a geometric mechanism: a Jacobian-based analysis reveals that backdoors act as low-rank, target-centered deformations that amplify local sensitivity, causing distortion to propagate coherently across semantic neighborhoods. To rigorously quantify this structural degradation, we introduce SEMAD (Semantic Alignment and Drift), a diagnostic framework that measures both internal embedding drift and downstream functional misalignment. Our findings, validated across diffusion and contrastive paradigms, expose the deep structural risks of encoder poisoning and highlight the necessity of geometric audits beyond simple attack success rates.

Changing runtime complexity on cloud and edge devices necessitates elastic large language model (LLM) deployment, where an LLM can be inferred with various quantization precisions based on available computational resources. However, it has been observed that the calibration parameters for quantization are typically linked to specific precisions, which presents challenges during elastic-precision calibration and precision switching at runtime. In this work, we attribute the source of varying calibration parameters to the varying token-level sensitivity caused by a precision-dependent outlier migration phenomenon.Motivated by this observation, we propose \texttt{MoBiQuant}, a novel Mixture-of-Bits quantization framework that adjusts weight precision for elastic LLM inference based on token sensitivity. Specifically, we propose the many-in-one recursive residual quantization that can iteratively reconstruct higher-precision weights and the token-aware router to dynamically select the number of residual bit slices. MoBiQuant enables smooth precision switching while improving generalization for the distribution of token outliers. Experimental results demonstrate that MoBiQuant exhibits strong elasticity, enabling it to match the performance of bit-specific calibrated PTQ on LLaMA3-8B without repeated calibration.

Recent advances in multi-instance learning (MIL) have witnessed impressive performance in whole slide image (WSI) analysis. However, the inherent sparsity of tumors and their morphological diversity lead to obvious heterogeneity across regions, posing significant challenges in aggregating high-quality and discriminative representations. To address this, we introduce a novel concept of anchor instance (AI), a compact subset of instances that are representative within their regions (local) and discriminative at the bag (global) level. These AIs act as semantic references to guide interactions across regions, correcting non-discriminative patterns while preserving regional diversity. Specifically, we propose a dual-level anchor mining (DAM) module to \textbf{select} AIs from massive instances, where the most informative AI in each region is extracted by assessing its similarity to both local and global embeddings. Furthermore, to ensure completeness and diversity, we devise an anchor-guided region correction (ARC) module that explores the complementary information from all regions to \textbf{correct} each regional representation. Building upon DAM and ARC, we develop a concise yet effective framework, AINet, which employs a simple predictor and surpasses state-of-the-art methods with substantially fewer FLOPs and parameters. Moreover, both DAM and ARC are modular and can be seamlessly integrated into existing MIL frameworks, consistently improving their performance.

Developers often extract methods to improve readability, understanding, and reuse, while inlining keeps logic in one block. Prior work based on static metrics has not shown clear differences between these practices, and the human side of comprehension and navigation remains underexplored. We investigate Inline Method vs. Extract Method refactorings using a dynamic approach: eye tracking while participants read and solve tasks. We analyze key code areas and compare visual effort and reading behavior (fixation duration and count, regressions, revisits), alongside time and attempts. We ran a controlled experiment with 32 Java novices, followed by short interviews. Each participant solved eight simple tasks across four programs presented in an inlined version and four in an extracted version. We also surveyed 58 additional novices for complementary quantitative and qualitative data. Results show that effects depend on task difficulty. In two tasks, method extraction improved performance and reduced visual effort, with time decreasing by up to 78.8% and regressions by 84.6%. For simpler tasks (e.g., square area), extraction hurt performance: time increased by up to 166.9% and regressions by 200%. Even with meaningful method names, novices often switched back and forth between call sites and extracted methods, increasing navigation and cognitive load. Preferences frequently favored extraction for readability and reuse, but did not always match measured performance. These findings suggest educators should be cautious about premature modularization for novices and highlight eye tracking as a useful complement to static metrics.

Large language model (LLM) inference performance is increasingly bottlenecked by the memory wall. While GPUs continue to scale raw compute throughput, they struggle to deliver scalable performance for memory bandwidth bound workloads. This challenge is amplified by emerging reasoning LLM applications, where long output sequences, low arithmetic intensity, and tight latency constraints demand significantly higher memory bandwidth. As a result, system utilization drops and energy per inference rises, highlighting the need for an optimized system architecture for scalable memory bandwidth. To address these challenges we present the Reasoning Processing Unit (RPU), a chiplet-based architecture designed to address the challenges of the modern memory wall. RPU introduces: (1) A Capacity-Optimized High-Bandwidth Memory (HBM-CO) that trades capacity for lower energy and cost; (2) a scalable chiplet architecture featuring a bandwidth-first power and area provisioning design; and (3) a decoupled microarchitecture that separates memory, compute, and communication pipelines to sustain high bandwidth utilization. Simulation results show that RPU performs up to 45.3x lower latency and 18.6x higher throughput over an H100 system at ISO-TDP on Llama3-405B.

Motivated by the problem of assigning mediators to cases in the Kenyan judicial, we study an online resource allocation problem where incoming tasks (cases) must be immediately assigned to available, capacity-constrained resources (mediators). The resources differ in their quality, which may need to be learned. In addition, resources can only be assigned to a subset of tasks that overlaps to varying degrees with the subset of tasks other resources can be assigned to. The objective is to maximize task completion while satisfying soft capacity constraints across all the resources. The scale of the real-world problem poses substantial challenges, since there are over 2000 mediators and a multitude of combinations of geographic locations (87) and case types (12) that each mediator is qualified to work on. Together, these features, unknown quality of new resources, soft capacity constraints, and a high-dimensional state space, make existing scheduling and resource allocation algorithms either inapplicable or inefficient. We formalize the problem in a tractable manner using a quadratic program formulation for assignment and a multi-agent bandit-style framework for learning. We demonstrate the key properties and advantages of our new algorithm, SMaRT (Selecting Mediators that are Right for the Task), compared with baselines on stylized instances of the mediator allocation problem. We then consider its application to real-world data on cases and mediators from the Kenyan judiciary. SMaRT outperforms baselines and allows control over the tradeoff between the strictness of capacity constraints and overall case resolution rates, both in settings where mediator quality is known beforehand and in bandit-like settings where learning is part of the problem definition. On the strength of these results, we plan to run a randomized controlled trial with SMaRT in the judiciary in the near future.

Manufacturing automation in process planning, inspection planning, and digital-thread integration depends on a unified specification that binds the geometric features of a 3D CAD model to the geometric dimensioning and tolerancing (GD&T) callouts, datum definitions, and surface requirements carried by the corresponding 2D engineering drawing. Although Model-Based Definition (MBD) allows such specifications to be embedded directly in 3D models, 2D drawings remain the primary carrier of manufacturing intent in automotive, aerospace, shipbuilding, and heavy-machinery industries. Correctly linking drawing annotations to the corresponding 3D features is difficult because of contextual ambiguity, repeated feature patterns, and the need for transparent and traceable decisions. This paper presents a deterministic-first, context-aware framework that maps 2D drawing entities to 3D CAD features to produce a unified manufacturing specification. Drawing callouts are first semantically enriched and then scored against candidate features using an interpretable metric that combines type compatibility, tolerance-aware dimensional agreement, and conservative context consistency, along with engineering-domain heuristics. When deterministic scoring cannot resolve an ambiguity, the system escalates to multimodal and constrained large-language-model reasoning, followed by a single human-in-the-loop (HITL) review step. Experiments on 20 real CAD-drawing pairs achieve a mean precision of 83.67%, recall of 90.46%, and F1 score of 86.29%. An ablation study shows that each pipeline component contributes to overall accuracy, with the full system outperforming all reduced variants. By prioritizing deterministic rules, clear decision tracking, and retaining unresolved cases for human review, the framework provides a practical foundation for downstream manufacturing automation in real-world industrial environments.

We investigate the effective interaction between two localized spin impurities embedded in a frustrated spin-1/2 $J_1\!-\!J_2$ Heisenberg chain. Treating the impurity spins as classical moments coupled locally to the host, we combine second--order perturbation theory with large--scale density matrix renormalization group (DMRG) calculations to determine the impurity--impurity interaction as a function of separation, coupling strength, and magnetic frustration. In the weak--coupling regime, we show that the interaction is governed by the the static spin susceptibility of the host and exhibits oscillatory power--law decay in the gapless phase, modified by universal logarithmic corrections at the SU(2)--symmetric critical point. In the gapped dimerized phase, the interaction decays exponentially with distance. For intermediate and strong impurity--host coupling, we observe a crossover to a boundary--dominated regime characterized by pronounced parity effects associated with the length of the chain segment between impurities, signaling a breakdown of the simple RKKY--like description. Our results establish impurity--impurity interactions as a sensitive probe of frustrated quantum spin liquids and provide a controlled framework for distinguishing gapless and gapped phases through local perturbations.

To interpret molecular dynamics (MD) simulations, it is common practice to reduce the dimensionality of the molecular coordinates to a low-dimensional collective variable $x$. Projecting the high-dimensional MD data onto $x$ yields a free energy landscape $ΔG(x)$, which highlights low-energy regions corresponding to conformational states. The accurate definition of these states, however, is often impeded by projection artifacts, resulting in artificially shortened state lifetimes or even the complete disappearance of states from the analysis. As demonstrated for a two-dimensional toy model, Gaussian low-pass filtering of the high-dimensional MD coordinates can restore the underlying free energy landscape, allowing to recover previously hidden states. When applied to an all-atom folding trajectory of HP35, the number of microstates increases by an order of magnitude, which leads to metastable states that are long-lived and much better defined structurally, even compared to dynamically cored state trajectories.

Phase transitions are conventionally defined by nonanalyticities of thermodynamic potentials in the thermodynamic limit. In this Letter, we show that the singularity is not the definition of criticality but its asymptotic outcome: criticality is already written in the microcanonical entropy derivatives at any finite size as intrinsic morphological structures -- inflection points and extrema. The singularity is then the endpoint of a sharpening process that evolves with increasing system size. Combining microcanonical inflection-point analysis (MIPA) with the Berlin-Kac spherical model -- for which the microcanonical density of states is known in closed form at every finite $N$ -- we systematically identify these structures in the energy profiles of entropy derivatives that encode the transition. An inflection point in the inverse temperature $β_N(ε)=\partial_εS_N$ and a pronounced peak in its derivative $γ_N(ε)=\partial^2_εS_N$ define a well-controlled pseudocritical trajectory whose controlled sharpening and drift culminate in the macroscopic cusp at the critical energy $ε_c$ in the thermodynamic limit. This establishes an intrinsic, order-parameter-free notion of criticality that precedes its singular asymptotic representation.

Quantum oscillations, the oscillatory behavior of electrical and thermodynamic properties, are typically observed in metals and vanish in the quantum limit under strong magnetic fields1. Phenomena such as the fractional quantum Hall effect2, the Hofstadter butterfly3,4, and recent observations of quantum oscillations in exotic insulators are notable exceptions5-12. The narrow-gap Dirac semiconductor ZrTe5, a less exotic material without strong correlations or artificially engineered superlattices, nevertheless exhibits resistance oscillations in the quantum limit13 but can be interpreted within a simple Zeeman-effect-based picture14,15, which remains conventional quantum oscillations without exotic properties. Here, we report the observation of unexpected mini-oscillations superimposed on Zeeman-effect-induced main oscillations in the quantum limit. The subtracted mini-oscillations are periodic in 1/B with the highest frequency equal to 2.1% of the first Brillouin zone and have extremely heavy effective mass ~ 2me, which is unexpected in ZrTe5 given its ultralow carrier density. Additionally, the mini-oscillations exhibit sequential features that are synchronized with the main oscillations, suggesting an internal structure of the Landau bands. However, they appear incompatible with the Hofstadter butterfly due to the highly anisotropic/three-dimensional crystal structure. These sequential mini-oscillations correlate with the commensurability resonance effect with subunity fractions observed in angular magnetoresistance, relating to the formation of mini-Landau bands. Our results present solid experimental evidence of exotic quantum oscillations in the quantum limit beyond currently available mechanisms, and establish ZrTe5, a prototypical Dirac semiconductor, as a simple platform parallel to correlated insulators for exploring exotic oscillations.

The auxiliary mode approach to time-dependent open quantum system calculations is implemented and refined to yield a feasible computational approach to simulate nanostructures far from equilibrium. It is done by a careful diagonalization of the electrode level-width function, and provides an efficient approach which can simulate large, open systems at the level of time-dependent density functional theory. The approach, as given in this work, is implemented in the new open-source code Zandpack. The framework is applied to three systems perturbed by the same THz electromagnetic field pulse-form: 1) A Hubbard model for hydrogen on graphene is used to calculate spin-currents, mutual information, spin-transitions, and a pump-probe setup. 2) An armchair graphene nanoribbon (AGNR) probed by a metal tip showing electrons excited from the valence band of the AGNR into the tip via electron-electron interactions. 3) A gold break-junction is modeled with various gap distances, and displays behavior that is more different from the adiabatic case as the gap widens. In the examples 2 and 3, we develop and use a general linearization scheme for time-dependent open system calculations, which utilizes the DFTB+ or SIESTA codes.

The rheological properties of biological tissues, though fundamental to many physiological and pathological processes such as embryonic development, wound healing, and tumor progression, remain poorly understood. A recent study showed that the active vertex model of biological tissues exhibits discontinuous shear thickening (DST), where stress and viscosity suddenly increase at a critical shear rate. What is the mechanism of DST here? Is it another nontrivial feature of activity or an inherent property of the system? To address this, we show that the thermal vertex model also exhibits DST at a small but non-zero temperature $T$. Solid-like and liquid-like cells coexist at the stress jump, and the stress-controlled flow curves exhibit the characteristic S-shape. We then introduce a constitutive model for DST in epithelial tissues. As $p_0$ increases, the theory predicts DST, followed by continuous shear thickening (CST), and finally Newtonian behavior, consistent with simulations. DST begins at the jamming point, $p_0^m$, and the Newtonian behavior starts at $p_0^*$, where the yield stress vanishes. Both $p_0^*$ and the liquid-to-solid transition stress, $σ^*$, govern the DST-CST boundary. Furthermore, $p_0^*$ and $σ^*$ also depend on $T$. Increasing $T$ reduces $p_0^*$, narrows the shear-thickening regime, and eventually destroys DST when $p_0^* \leq p_0^m$. Thus, the primary ingredients of DST in tissue models are a finite yield stress in the unjammed regime and non-zero fluctuations, whose specific form is not important. The theory agrees well with our simulation data and also provides further testable predictions.

We study quantum transport in disordered topological insulator nanowires (TINWs) under axial magnetic flux. At integer flux quanta, spin-momentum locking produces weak anti-localization peaks, while at half-integer flux quanta a helical mode protected by time-reversal symmetry (TRS) suppresses backscattering. By analyzing the flux dependence of the localization length, we uncover critical scaling around half-integer flux quanta, reflecting the competition between disorder scattering and flux-induced breaking of TRS protection. As the disorder strength increases, we identify a crossover in scaling behavior that drives the system into a regime governed by a universal critical exponent. Our results demonstrate a scaling collapse across flux values, establishing a universal regime of flux-driven delocalization in TINWs.

Magnetic impurities in real materials can mask the intrinsic spin-dependent properties of hosts. They interact indirectly through the Ruderman-Kittel-Kasuya-Yosida (RKKY) mechanism, which limits the use of isolated impurity spins in applications such as qubits and spintronics. Suppressing the RKKY interaction would therefore enable access to the host's unperturbed behavior while simultaneously isolating impurity spins for functional use. Although single-color laser driving can suppress the RKKY interaction, it typically requires strong fields that may be impractical or destabilizing. To overcome these limitations, we show that two-color laser driving provides efficient and tunable control over all components of the RKKY interaction using two weak laser fields. Focusing on two-dimensional Rashba altermagnets, we show that interference between one- and two-photon processes produces altermagnet-specific Floquet corrections. These include additional AC Stark shifts, magnetizations, spin-orbit renormalization, and emergent in-plane Zeeman fields that are absent under single-color driving and in non-altermagnetic systems. Notably, two-color driving induces a finite $z$-component of the Dzyaloshinskii-Moriya (DM) interaction, stabilizing in-plane chiral magnetism and related textures in Rashba altermagnets. These effects enable tunable, near-complete on-off switching of the Heisenberg, Ising, and DM interactions through a Lifshitz-like modulation of the Fermi surface. We also show that the tuning process is highly sensitive to the chirality of both beams. We further map phase diagrams for ferromagnetic and antiferromagnetic impurity alignment with clockwise and counterclockwise canting as functions of Rashba coupling and altermagnetic order. Finally, we discuss candidate material platforms and experimental feasibility.

We theoretically investigate spin dependent transport in altermagnet/triplet superconductor/altermagnet (AM/TSC/AM) junctions in the presence of interfacial Rashba spin orbit coupling (RSOC). Within a microscopic Bogoliubov de Gennes scattering formalism, we compute angle and energy resolved conductance, spin polarization, zero bias response, and tunneling magnetoresistance (TMR) for nodal $p_x$ and chiral $p_x+ip_y$ superconductors. Although altermagnets carry no net magnetization, the momentum dependent spin splitting, combined with RSOC, enables a pronounced spin valve effect without ferromagnetic electrodes. We show that conductance, spin polarization, and TMR exhibit distinct and robust fingerprints of the triplet pairing symmetry. For nodal $p_x$ superconductor, sign change induced surface Andreev bound states dominate subgap transport, producing strongly anisotropic conductance, giant zero bias spin polarization, and a monotonic enhancement of TMR with increasing RSOC. In contrast, the chiral $p_x+ip_y$ state exhibits smoother conductance and polarization profiles governed by topological edge modes, resulting in broader, lobe like TMR patterns with weaker sensitivity to interface transparency. Moreover, RSOC can acts as an electrically tunable spin-mixing knob, while the barrier strength controls coherence and energy selectivity, together enabling large, symmetry controlled spin filtering and magnetoresistance. Our results establish AM/TSC/AM junctions as a symmetry sensitive transport platform for realizing electrically tunable spin valve functionality and probing triplet pairing without ferromagnetic components.

We investigate a cantilevered elastic beam subjected to harmonic rotational motion. In the rotating frame, the beam experiences centrifugal and Euler fictitious forces, with negligible Coriolis effects. We validate a reduced-order \textit{elastica} model through precision experiments on slender beams rotating with a controlled sinusoidal angular velocity. Systematically exploring the parameter space, we identify regimes where inertial effects are negligible, enabling a quasi-static treatment despite harmonic driving. We characterize the transition to dynamic response using two dimensionless parameters, the Euler and centrifugal numbers, which compare centrifugal and Euler forces to bending forces. Counterintuitively, the quasi-static regime expands as rotational speed increases: faster rotation produces less dynamic response. The critical Euler number separating these regimes remains constant at low centrifugal numbers but follows square-root scaling at higher rotation rates, a transition driven by centrifugal stiffening. Our results establish the conditions under which quasi-static approximations remain valid for rotating flexible beams under harmonic driving.

We study an extreme active matter system, which is essentially a dense assembly of athermal, soft and infinitely persistent active particles. Using extensive numerical simulations we obtain jammed configurations of this system in two dimensions and probe the stability of such structures under increasing active forcing magnitude. We show that the critical active forcing magnitude for the jammed phase to yield scales with virial pressure as $f_c\sim p^α$, with $α=1.17$, describing the yielding line. Using a Laplacian framework, we redistribute the active forces into a modified contact force network. By analysing the statistics of these redistributed forces, we obtain a very robust scaling law consistent with the passive limit, not just near the unjamming line, but in the entire jammed active phase. The probability distribution of the magnitude of the contact force deviates from the power-law form found in passive systems for values smaller than the active force. Moreover, within the jammed phase, the system displays elastic, plastic, and yielding events with increasing active forcing. This active plasticity appears abruptly and can not be captured by the continuous softening of the Hessian spectrum. However, we demonstrate that the Hessian still retains the ability to predict relaxation times. These results clarify how activity modifies force distributions and leads to deformation, plasticity and yielding in dense, jammed, infinitely persistent active matter.

Junction spectroscopy techniques (JSTs) are powerful tools for investigating electrically active defects in semiconductors. Originally developed to study point-like defects in bulk semiconductors, JSTs have since been extended to increasingly complex systems, providing valuable insights into defect energetics and interactions. This review paper outlines the fundamental principles of JSTs and critically examines their application to emerging materials, such as perovskite solar cells and two-dimensional (2D) materials. By highlighting both the capabilities and limitations of JSTs in these non-classical systems, the review demonstrates their continued relevance and important role in advancing next-generation semiconductor materials and devices.

We study elementary semiconductors and insulators that are symmetric under spatial inversion: silicon, diamond, germanium, and black phosphorene. These materials are ideal candidates for realizing obstructed atomic insulators, which differ from trivial atomic insulators by a quantized spatial shift of their electronic Wannier centers with respect to the atomic lattice. We use symmetry indicator invariants that allow the prediction of non-trivial responses to crystal dislocations in these materials. We find that edge dislocations generically exhibit a non-trivial response, while screw dislocations always display a trivial response. With the aid of numerical simulations of realistic tight-binding models, we confirm the presence of mid-gap polarization bands localized along dislocations in silicon, diamond, and germanium.

Spin manipulation by hopping has recently emerged as a promising strategy to control hole spins in quantum dots using exclusively baseband control, thereby mitigating power dissipation and high-frequency management constraints in large-scale architectures. Unlike conventional approaches such as electron dipole spin resonance (EDSR), this mechanism exploits dot-to-dot variations of the spin precession axes to enable spin rotations. However, it is intrinsically disorder-dependent: in the absence of sufficient variability, the precession axes remain aligned and spin manipulation becomes ineffective. This fundamental reliance on disorder raises concerns regarding its compatibility with the long-term evolution of spin-qubit platforms toward improved material quality, cleaner interfaces, and enhanced device reproducibility. Here, we numerically assess the viability of spin manipulation by hopping as a function of disorder strength and demonstrate that its implementation is indeed increasingly constrained as disorder is reduced. To overcome this limitation, we propose an alternative strategy based on hopping between intentionally squeezed quantum dots. This approach retains the advantages of baseband control while being independent of disorder and robust against moderate variability, thereby offering improved prospects for scalable hole-spin quantum computing architectures.

We report the experimental observation of memory effects in a vertically vibrated thin granular layer. Following a quench in the input acceleration, the granular temperature exhibits an anomalous Kovacs memory effect confined to the initial fast relaxation stage. This memory vanishes shortly thereafter, yielding a time-dependent memoryless regime governed solely by the instantaneous temperature before the system reaches its final steady state. We develop a kinetic theory framework that quantitatively captures these features by identifying the initial memory and subsequent memoryless regimes with the kinetic and hydrodynamic states, respectively (that are well established in kinetic theory). Our analysis reveals that memory emerges during fast transients through coupling between horizontal and vertical temperatures, a mechanism that fundamentally constrains the accessible memory phenomenology and precludes observation of the standard Kovacs effect in this system. Molecular dynamics simulations provide independent confirmation of all experimental and theoretical findings.

Tipping points are one of the hot topics in modern physics of complex systems. But what is a tipping point? A generic definition declares it as ``a state of the system where a small change in its parameters can lead to a significant change in its properties''. Additional ingredients that often enter the definition of tipping process are the abruptness of the resulting change and its irreversibility, i.e. it is impossible to recover the initial state if one reverses the protocol of change of the parameters. However, there exists a number of different mathematical structures that can show this behavior, the one that was originally suggested as a tipping point (nowadays usually referred to as bifurcation induced tipping) is just one of many. Different preconditions and/or different level of details included into the model, reflecting also different environmental forcing, can lead to a variety of tipping mechanisms. Furthermore, in a spatially extended system and/or a system with multiple scales, different parts can react to a change in environmental conditions differently or at a different time, interacting with each other to create a tipping cascade. In this paper, using ecosystems as a paradigm of complex nonlinear open systems, we provide a critical overview of the progress made in tipping point science over the last 15 years. We highlight the main findings, identify gaps in our knowledge, and outline a roadmap for further progress.

We prove a finite entanglement length for the Gibbs state of any local Hamiltonian on a spin chain at any finite temperature: After removing an interval of size at least equal to the entanglement length, the remaining left and right half-chains are in a separable state.

Studying shape changing thick surfaces induced by differential growth helps us understand morphogenesis in biology and offers opportunities for device design. While ideal 2D differential growth maps have been well studied for both isotropic and anisotropic growth, scenarios involving gradients in thickness growth are far less explored. In this paper, we focus on a bilayer system in which the two layers undergo independent but incompatible growth. We examine how the strength of the growth patterns and the aspect ratio of the bilayer influence the resulting shapes. We first investigate the effect of global area difference in the bilayer. Next, we make one of the two layers active and program it with positive or negative, localized or uniform curvature. We then present examples involving competition between two active surfaces with opposite curvature signs or different curvature distributions and understand how the final configurations follow from the principles identified earlier. Finally, we demonstrate the ability to program biologically inspired shapes, such as dental epithelium and the ventral furrow in Drosophila, using a bilayer with simple, distinct deformation gradients.

We present a simple model that recalls two different patterns depending on the temperature. To realize a change in recall pattern due to temperature change, we embed two patterns to different graphs: the first pattern into a fully connected graph and the second pattern into a sparse graph. Because a fully connected graph is more resistant to thermal fluctuations than a sparse graph, we can realize a change in recall pattern by tuning relative weights of the two patterns properly. We demonstrate by equilibrium Monte-Carlo simulations that such a temperature-dependent change in recall patterns does occur in our model. Simulation results strongly indicate that the system undergoes a first-order phase transition when the change in recall patterns occurs. It is also demonstrated by annealing simulations that the system fails to recall the pattern embedded in the sparse graph at low temperatures if the free-energy barrier is too high to overcome within the given simulation timescale.

We study the breakdown of Kondo screening by a local magnetic field from the perspective of a measurement-driven entanglement transition in a monitored quantum system. Here, the Kondo coupling leads to the growth in entanglement of an impurity spin with it's fermionic environment, while the local field plays the role of a continuous observer. Using a non-perturbative Unitary Renormalization Group (URG) approach, we derive coupled renormalization-group flow equations for the Kondo exchange and the local field, and obtain a field-dependent RG phase diagram. The RG flows separate a low-energy Kondo-screened phase, where the impurity is absorbed into the Fermi sea and forms an entangled singlet with the conduction bath, from a polarized local-moment phase in which screening is frustrated and impurity-bath entanglement is suppressed. We identify the fixed-point Hamiltonians governing the two phases and the critical regime, and relate the transition to the emergence of a novel non-Fermi liquid. Various impurity signatures such as the spectral function and thermalisation of impurity observables are used to characterise this entanglement transition. These results offer insight into the interplay of decoherence and measurement in governing the dynamics of a prototypical quantum system.

Analytical treatments of tunneling in bilayer graphene have typically relied on minimal models including only the vertical interlayer hopping $γ_1$ and have been restricted to the weak interlayer bias regime $2\varepsilon \ll γ_1$. These simplifications limit the ability of analytical theories to describe lattice deformations and strong electric-field effects. In this work, we present an analytical theory of evanescent states in electrically gapped bilayer graphene that overcomes both limitations. Specifically, our approach explicitly incorporates the skew interlayer hoppings $γ_3$ and $γ_4$ and remains valid even when the interlayer bias $2\varepsilon$ is comparable to $γ_1$. Focusing on low-energy electronic states near the charge neutrality point, we analytically derive the complex longitudinal wave numbers, the gap width, and the sublattice pseudospin inside the electric-field-induced gap, and systematically analyze their dependence on the interlayer shear displacement $\vecδ=(δ_x,δ_y)$. The analytical expressions quantitatively reproduce exact numerical calculations, demonstrating that skew interlayer hoppings, in particular $γ_3$, play an essential role. Taking the zigzag direction as the longitudinal (transport) $x$ direction, the wave vector becomes complex along $x$ while remaining real along the transverse $y$ direction. For a monolayer--bilayer--monolayer junction with transport along this direction, we find that $δ_y$ has a significantly stronger impact on the conductance than $δ_x$. This anisotropic response is quantitatively explained by the analytical expressions. Furthermore, we identify a shear-induced phase proportional to $δ_y$ that appears universally in the analytical expressions for the gap width, the sublattice pseudospin, and the decay length.

In this work, we numerically verify the Jarzynski equality and Crook fluctuation theorem for a Brownian particle diffusing in a heterogeneous thermal bath and hence having a non-Gaussian position distribution. We use the diffusing-diffusivity model to take the account of heterogeneity of the thermal bath where the mobility is considered as a fluctuating quantity. The Brownian particle is confined by a time-dependent harmonic potential. By changing the stiffness coefficient, we perform an isothermal process. We use the stochastic thermodynamics framework to calculate the work. We find that the Jarzynski equality and the Crook fluctuation theorem are convincingly satisfied for a non-Gaussion system. We also find that the work distribution is non-Gaussian for diffusing-diffusivity system even at a larger process time.

Micron scale imaging of magnetic fields is an important tool for understanding the evolution of magnetism through phase transitions and as a result of interactions inside of heterostructures. However, most imaging platforms, like the nitrogen-vacancy (NV) centre in diamond, are restricted to applying magnetic fields along the quantisation axis of the quantum sensor. This greatly restricts the utility of these systems for exploring materials that emit strong fields or exhibit variable response with respect to the applied field direction. Here we explore an alternative approach using weakly coupled spin-pairs in hBN that exhibit a spin-1/2-like behaviour and an isotropic response to magnetic field. We demonstrate that the spin-pair system can operate in the presence of strong fields from a thin film magnet which were incompatible with NV diamond imaging even with applied fields along the quantisation axis. Further, we demonstrate that using this platform allows for imaging with an arbitrary applied magnetic field direction, allowing us to probe the anisotropy and spin-reorientation transition in the ferrimagnet TbMn$_6$Sn$_6$. Finally, we propose an improved geometry for imaging small anisotropy contributions such as crystalline anisotropy. These results demonstrate how this or similar spin-1/2 systems might be used for imaging magnetic materials that are incompatible with other techniques despite the reduction in sensitivity compared with NV in diamond imaging.

Motivated by the recent proposal of two-step transduction from microwave to optical domain using magnetic and elastic intermediate stages arXiv:2205.05088, we consider the coupling between resonant magnetic and elastic modes within a simple axially-symmetric nanodevice designed to host high-quality-factor acoustic modes: A suspended YIG ring structure supported by a central stem, fabricated from a continuous single-crystal film. We study the modes of the system with our custom finite element solvers. We identify the lowest order ``breathing'' mode of a magnetic vortex and the lowest order elastic breathing mode as having the largest mode overlap. For this pair of modes, the external out-of-plane magnetic bias field is critical for bringing them into resonance; however, we show that at the same time it also affects the strength of the coupling. To counteract this, we optimize the radius of the ring at fixed thickness. For the 100 nm-thick film the resonant coupling is maximized at $g/2π= 8\text{MHz}$ at $R\approx1.7μ\text{m}$, indicating that the overlap integral approaches the idealized limit assumed in previous order-of-magnitude estimates. Our results pave the way for the design of tunable frequency-conversion devices based on magnetoelastics.

We systematically investigate the nonlinear photogalvanic effect in few-layer graphene with various stacking orders, including AA- and AB-stacked bilayers, and AAA-, ABA-, and ABC-stacked trilayers. Using a tight-binding model to describe the electronic states, the shift current conductivity and jerk current conductivity are calculated over a broad spectral range from terahertz to visible frequencies. Our symmetry analysis reveals that a nonvanishing shift current emerges only in ABA-stacked trilayer graphene due to its broken inversion symmetry, with a peak conductivity reaching approximately $1.21 \times 10^{-13}$ A$\cdot$m/V$^2$ at optimal doping. In contrast, the jerk current, permitted in all structures, requires an in-plane static electric field and exhibits pronounced spectral tunability with chemical potential. These findings establish a comprehensive symmetry-band-field coupling paradigm for nonlinear photocurrents in layered graphene and provide design principles for tunable, polarization-sensitive photodetection and energy-harvesting devices based on van der Waals heterostructures.

The dynamic fragmentation of residually stressed solids involves a complex interplay between stored elastic energy, stress wave propagation, and crack instabilities. In this work, we investigate the fracture mechanics of chemically toughened glass through high-velocity projectile impact experiments and a novel micromechanical network model. We rigorously incorporate residual stress into the discrete lattice framework via a prescribed inelastic strain (eigenstrain) distribution, formulated as equivalent body and surface forces to ensure mesh-independent fracture paths. Our experiments and simulations demonstrate that while the fracture topology shifts from coarse to fine with increasing impact energy, the cumulative fragment size distribution consistently follows an exponential decay. Crucially, we reveal a universal scaling law: fragment size distributions from diverse loading conditions and stress profiles collapse onto a single master curve when normalized by the mean fragment area. Furthermore, the model elucidates the determinants of fragmentation, showing that the resulting fragment size is governed not only by the magnitude of residual stress but also by the steepness of the stress gradient. At the microscopic scale, we identify a mechanism for dynamic instability where non-sequential bond breaking ahead of the crack tip leads to apparent local crack speeds exceeding the Rayleigh wave speed ($c_r$). These arrested micro-branches, analogous to the Burridge-Andrews mechanism, provide a physical explanation for the "tongue-like" features and hackle zones observed in post-mortem fractography.

Carbyne-the one-dimensional sp-hybridised allotrope of carbon-has long been predicted to exhibit unique properties, yet its synthesis remains elusive. To probe its behaviour, finite sp-carbon chains such as cumulenes and polyynes have been studied, but work to date has focused almost exclusively on short, linear systems far from the infinite carbyne limit and without considering topology. Here, we investigate long (48-carbon) linear and cyclic polyynes using steady-state and ultrafast, temperature- and polarization-resolved optical and vibrational spectroscopy. We find highly delocalized ground states in both topologies, with Peierls distortions markedly weaker than in short chains. In contrast, excited states undergo rapid self-localisation, with the localisation pathway and subsequent intersystem crossing strongly dependent on chain length and topology. Unlike shorter polyynes, excited-state structural rearrangements are minimal, and comparison with theoretical predictions shows that properties, such as Huang-Rhys factors, have plateaued by 48 carbons. Our results reveal how topology influences the electronic dynamics of long polyynes and refines our understanding of sp-carbon systems approaching the carbyne limit

Understanding and predicting how complex systems respond to external perturbations is a central challenge in nonequilibrium statistical physics. Here we consider continuous-time Markov networks, which we subject to perturbations along a single edge. We find that in steady state the probabilities of any two states are linearly related to one another. We show that this mutual linearity of probabilities extends to a broad class of observables, including currents but also generic counting and state-dependent observables. Moreover, we derive an exact relation between the relative response of any state's probability and the ratio of two steady-state probabilities. Leveraging the Markov chain tree theorem, we further show that probabilities and the considered observables are constrained by the topological and kinetic properties of the network and provide analytical expressions in terms of spanning tree polynomials. Our results are general, holding for arbitrary rate parameterizations and extending far from equilibrium.

While nucleation in typical active and driven fluids often appears equilibrium-like, striking departures emerge when large-scale fluctuations are strongly suppressed. Here, we investigate nucleation in nonequilibrium hyperuniform fluids by projecting the full density-field dynamics onto relevant collective variables. We demonstrate that nucleation is governed by a nonequilibrium quasi-potential rather than the reversible work of formation. Surprisingly, because of the reduced hyperuniform fluctuations, the nucleation probability no longer separates into the usual surface and volume contributions. Furthermore, accounting for capillary waves reveals a clear breakdown of detailed balance driven by nonreciprocal dynamics. More broadly, our framework can be readily extended to identify nonequilibrium signatures in conventional active fluids.

We approach the 3-SAT satisfiability problem with the quantum-inspired method of imaginary time propagation (ITP) applied to matrix product states (MPS) on a classical computer. This ansatz is fundamentally limited by a quantum entanglement barrier that emerges in imaginary time, reflecting the exponential hardness expected for this NP-complete problem. Strikingly, we argue based on careful analysis of the structure imprinted onto the MPS by the 3-SAT instances that this barrier arises from classical computational complexity. To reveal this connection, we elucidate with stochastic models the specific relationship between the classical hardness of the $\sharp$P $\supseteq$ NP-complete counting problem $\sharp$3-SAT and the entanglement properties of the quantum state. Our findings illuminate the limitations of this quantum-inspired approach and demonstrate how purely classical computational complexity can manifest in quantum entanglement. Furthermore, we present estimates of the non-stabilizerness required by the protocol, finding a similar resource barrier. Specifically, the necessary amount of non-Clifford operations scales superlinearly in system size, thus implying extensive resource requirements of ITP on different architectures such as Clifford circuits or gate-based quantum computers.

An applied magnetic field affects a superconductor in two ways -- by promoting pairing fluctuations, and by inducing topological defects called vortices that carry quantized magnetic flux. A quantitative characterization of the resultant field-induced superconducting dynamics with spatio-temporal resolution remains challenging, particularly in two-dimensional materials. In this work, we analyze magnetic noise measured by the depolarization rate of a proximate single spin-qubit as a non-invasive probe of such dynamical fluctuations. We demonstrate that the temperature dependence of the magnetic noise spectrum near $T_c$ deviates from predictions based on quasiparticle excitations due to critical superconducting fluctuations, which in turn are enhanced by a weak applied field. By analyzing the magnetic noise due to vortex dynamics, we further show that noise spectroscopy is not only able to distinguish between different vortex phases, but also extract key physical quantities of interest, such as oscillation frequencies of pinned vortices, phonon dispersion of vortex lattices and vortex diffusivity in a vortex liquid. Complementing recent work on noise magnetometry of quasiparticle excitations and Berezinskii-Kosterlitz-Thouless transitions in two-dimensional superconductors without an applied field, our work highlights the ability of noise spectroscopy to reveal a wealth of superconducting dynamical phenomena in an applied field.

We present a general hydrodynamic theory for active fluids, capable of describing living matter, that conserve center of mass or dipole moment. Imposition of dipole or center-of-mass conservation has been reported to yield peculiar behavior: breaking Galilean invariance in classical systems and potentially enabling exotic immobile excitations in quantum settings. In passive fluids, dipole conservation has been shown to cause a breakdown of linear hydrodynamics in all experimentally relevant dimensions. We show that introducing activity changes this picture: it can either restore or break linear hydrodynamics depending on spatial dimensions. Using our formulation, we predict universal dynamical scaling exponents for single-component active fluids in $d=1,2,3$ dimensions and find agreement with microscopic lattice-field simulations. Strikingly, for $d\geq 2$, activity revives linear hydrodynamics, while for $d<2$ it leads to a breakdown; both cases flow to previously unexplored universality classes. Our results suggest that dipole-conserving active fluids are far more experimentally accessible than their passive counterparts.

We develop a systematic theory of spectral decimation for quantum many-body Hamiltonians and show that it provides a quantitative probe of emergent symmetries in statistically mixed spectra. Building on an analytical description of statistical mixtures, we derive an explicit expression for the size of a characteristic symmetry sector (CSS), defined as the largest subsequence of levels exhibiting non-Poissonian correlations. The CSS dimension is shown to be the size-biased average of the underlying symmetry sectors, establishing a direct link between spectral statistics and Hilbert-space structure. We apply this framework to two paradigmatic settings: Hilbert-space fragmentation and disorder-induced many-body localization (MBL). In fragmented systems, the CSS reproduces the mixture prediction and isolates correlated subsectors even when the full spectrum appears nearly Poissonian. In the disordered Heisenberg chain, spectral decimation reveals the gradual emergence of integrability through a shrinking CSS, whose statistics exhibit signatures consistent with local integrals of motion. We introduce a characteristic symmetry entropy (CSE) as a finite-size scaling observable and extract, within accessible system sizes, the crossover exponents. Our results establish spectral decimation as a controlled, unbiased and computationally inexpensive diagnostic of hidden structure in many-body spectra, capable of distinguishing between chaotic dynamics, statistical mixtures, and emergent integrability.

We develop a hydrodynamic theory of charge and heat currents induced by traveling waves, such as surface acoustic waves, in graphene devices near charge neutrality. The currents depend on the intrinsic conductivity and viscosity of the electron liquid, the disorder strength, and the geometry of the device. We obtain analytic expressions for the heat and charge currents to second order in the wave amplitude for Hall-bar devices. At charge neutrality and in the absence of DC bias, the heat content is entrained by the wave in the absence of net charge transfer. At the same time, device conductance is enhanced by the wave. Away from charge neutrality, the transport charge current induced by the wave arises in the absence of a DC bias.

Emergent electrodynamics arising from spatially and temporally varying magnetic textures provides a framework for spin control in quantum materials. While this principle is established for ferromagnetic and antiferromagnetic textures, its consequences for altermagnets -- magnetic orders with vanishing net magnetization but finite spin splitting -- remain largely unexplored. In this work, we develop an effective low-energy theory of itinerant electrons coupled to smoothly varying altermagnetic spin textures. In the adiabatic regime, we show that altermagnetic textures generate additional emergent electromagnetic fields and quantum-geometric effects that are absent in conventional magnetic systems. These effects include emergent Zeeman fields that encode the structure of the altermagnetic order parameter, enabling local spin manipulation and a way to distinguish different altermagnetic orders. Moreover, we demonstrate a quantum-metric-induced, spin-dependent electron lensing effect that provides a mechanism for spin filtering, and discuss the local admixture of effective odd-parity magnetic components. Our results suggest that textured altermagnets could serve as a versatile resource for spintronics functionalities and a probe of altermagnetism.

The Casimir-Polder interaction energy between a unipolarizable point atom and a unipolarizable dielectric ring has been limited, until now, to the case when the atom is confined on the axis of symmetry of the ring. We find the generalized analytical expression for any position of the atom relative to the ring in terms of complete elliptic integrals. This is aided by the construction of a class of integrals of a Jacobian elliptic function as a linear combination of complete elliptic integrals. Our expression for the interaction energy allows us to investigate the instability of the atom even for the equilibrium points which exists off the axis of symmetry.

Copolymerization is a powerful tool to tune polymers' glass formation behavior in order to improve properties such as ion conductivity, thermal stability, and mechanical response. Classically, mixing rules such as the Fox equation and its variants have been employed to predict and explain glass transition temperature (Tg) variations with copolymer composition. However, a large number of copolymers exhibit significant excursions from these mixing rules. Increasingly, it also appears that these excursions can be tuned by copolymer sequence - an effect that is not predicted by classical mixing rules and thus far remains poorly understood. Here, we perform molecular dynamics simulations to probe the interplay between copolymer sequence, liquid-liquid phase equilibria, and Tg. We find that the direction of Tg shifts, and their dependence on sequence, are predicted by the type of liquid-liquid phase boundary towards which the comonomers tend. Monomers tending towards Upper Critical Solution Temperature behavior exhibit Tg reductions that become stronger with increasing monomer alternation; monomers tending towards Lower Critical Solution Temperature behavior exhibit Tg enhancements that become stronger with increasing monomer alternation. This behavior fundamentally emerges from a forced mixing phenomenon wherein monomer alternation induces close contact between monomer types that would otherwise tend towards phase separation. These results point towards strategies for rationally varying copolymer Tg, at fixed chemical composition, via judicious design of polymer chain sequence.

Artificial agents capable of understanding and aligning with others' intentions are essential for safe and socially robust artificial intelligence. We introduce a computational framework for empathy in active inference agents, grounded in explicit perspective-taking via self-other model transformation. We instantiate this framework in a multi-agent Iterated Prisoner's Dilemma and show that empathic perspective-taking induces robust cooperation without explicit communication or reward shaping. Cooperation emerges only when empathy is reciprocated, while asymmetric empathy leads to systematic exploitation. Beyond equilibrium outcomes, empathic agents exhibit synchronized behavior, rapid recovery from stochastic defections, and joint intentional dynamics resembling apology-forgiveness cycles. Near empathy symmetry, interactions display long transients and elevated variance, consistent with critical dynamics near regime boundaries. We further examine a learning-enabled variant in which agents infer opponent type via Bayesian updating. While opponent models converge rapidly, long-run cooperation remains primarily determined by the empathy parameter, indicating that cooperation is driven by empathic structure rather than learned reciprocity. Empathy functions as a structural prior over social interaction, shaping coordination stability, robustness, and temporal dynamics. The proposed framework highlights active inference as a principled foundation for socially aligned artificial agents that coordinate through internal simulation rather than behavioral mimicry.

We consider the half-wave maps equation $$ \partial_t \mathbf{u} = \mathbf{u} \times |D| \mathbf{u} $$ for $\mathbf{u} : \mathbb{R} \times \mathbb{T} \to \mathbb{S}^2$, where $\mathbb{T}=\mathbb{R}/2 π\mathbb{Z}$ is the one-dimensional torus and $\mathbb{S}^2 \subset \mathbb{R}^3$ denotes the unit sphere. By extension from rational initial data, we construct a unique and continuous flow map for data in the critical energy space $H^{1/2}(\mathbb{T}; \mathbb{S}^2)$. Moreover, we show almost periodicity in time of these solutions. For the dense subset of rational initial data, we establish quasi-periodicity in time and a-priori bounds on $\| \mathbf{u}(t) \|_{H^s(\mathbb{T})}$ for any $s >0$. Our analysis relies crucially on an explicit formula arising from the Lax pair structure acting on a Hardy space of vector-valued holomorphic functions on the unit disk. As a central ingredient, we develop a general {\em stability principle} for explicit formulae associated with completely integrable PDEs possessing a Lax pair structure on Hardy spaces, including the Benjamin--Ono equation, Calogero--Sutherland DNLS, and the half-wave-maps equation posed on $\mathbb{T}$. Our results extend to the matrix-valued half-wave maps equation $$ \partial_t \mathbf{U} = -\frac{i}{2} [ \mathbf{U}, |D| \mathbf{U} ] $$ with target manifold given by the complex Grassmannians $\mathsf{Gr}_k(\mathbb{C}^d)$, thereby generalizing the special case $\mathbb{S}^2 \cong \mathbb{CP}^1 \cong \mathsf{Gr}_1(\mathbb{C}^2)$. In a companion work, we prove global well-posedness for the half-wave maps equation posed on $\mathbb{R}$ in the scaling-critical energy space $\dot{H}^{1/2}$, by establishing a stability principle for explicit formulae on Hardy spaces in the complex half-plane $\mathbb{C}_+$.

In this work we revisit the geometric approach to chaos in Hamiltonian dynamics, by means of the Jacobi-Levi-Civita equation (JLCE). We inspect numerically two low-dimensional dynamical systems; show that, along chaotic orbits, the exponential divergence between nearby trajectories quantified by the JLCE does not unfold in a continuous manner, rather is closer to a multiplicative discrete process: in correspondence of each turning point, where the trajectory bounces away from the boundary of the energetically allowed region, the relative separation increases sharply and abruptly. We highlight through analytical and numerical arguments that the chaotic rather than regular nature of the trajectory is determined by the details of the scattering with the boundary, and interpret these results in terms of parametric resonance theory, and specifically the Mathieu equation.

This review offers a retrospective of the development of the theory of coupled oscillators and synchronization over the past half century. Among the various works made by myself during this period, the following three specific works will be focused on, serving as some key points to illustrate the field's evolution. They are the derivation of (1) a simple partial differential equation exhibiting spatio-tempoeral chaos (Kuramoto-Sivashinsky equaiton), (2) a solvable mathematical model describing synchronization phase transition (Kuramoto model), and the discovery of (3) coexistence of coherence and incoherence in nonlocally coupled oscillators (chimera states). It is emphasized that all these works resulted fron the phase reduction of the complex Ginzburg-Landau equation (or its variants), the equation which was derived with a coworker in 1974 from a certain reaction-diffusion model. A quick overview will also be made on how the above three works influenced the subsequent development of the field of coupled oscillators and synchronization. Finally, a few comments will be made on how the methods of dynamical reduction, such as the center-manifold reduction and phase reduction, are crucial for exploring this field in depth. This article is a largely faithful reproduction of the content presented in my award lecture.

Guided acoustic waves in two-dimensional materials are a key channel for energy transport and dissipation, yet their generation and propagation in transition metal dichalcogenides remain poorly understood. Here, we employ in-situ and spatially decoupled ultra-fast optical pump-probe techniques to investigate guided waves in MoS2 flakes with thicknesses between 90 and 410 nm. We observe a propagating acoustic excitation with a constant velocity of (6.7 +/- 0.8) km/s, independent of thickness. Finite element simulations and symmetry analysis reveal that these vibrations deviate from the classical Lamb wave model and are better described as a superposition of decoupled longitudinal and shear modes. We show that their optical detectability is governed by the Poisson effect: longitudinal components modulate the flake thickness and generate a measurable signal, whereas shear motion remains largely optically invisible. An intrinsic attenuation length of approximately 3.3 microns indicates that dissipation is dominated by material-specific mechanisms rather than geometric spreading. Finally, we demonstrate remote excitation across a nanometric step, enabling acoustic generation in optically inaccessible regions. These results provide a foundation for nanoscale phononic circuits and engineered in-plane energy transport in 2D-based optomechanical and quantum acoustic devices.

While established semiconductor measurement techniques such as four-point probe or capacitance-voltage measurements require a physical contact to the material, terahertz spectroscopy is completely contact-free. Its capability to measure the doping of semiconductors is well known, yet the exact doping ranges that are accessible to terahertz spectroscopy are not obvious. Therefore, we introduce a sensitivity metric to clarify whether a semiconductor sample can be characterized in principle by reflection terahertz time-domain spectroscopy. This quantity takes into account the semiconductor material with a certain layer thickness, doping type, and doping level and is based on numerical simulations. In this work, we calculate this sensitivity value for relevant semiconductor materials (SiC, Si, GaN) in realistic layer stacks with up to three layers. It is used to create meaningful heat maps depending on the thicknesses and charge carrier densities of the sample structures of interest. The plausibility of the sensitivity is validated by mapping a variety of measurements with terahertz techniques from us and from other groups onto these heat maps. Based on these, the accessible range of charge carrier densities for terahertz spectroscopy spans roughly from 10$^{15}$ cm$^{-3}$ to 10$^{20}$ cm$^{-3}$, but with dependencies on material, doping type, and sample thickness. Furthermore, the sensitivity value allows for a substantiated assessment of the possible benefits future improvements of photoconductive antennas and terahertz systems could have, which is demonstrated by simulations based on varied bandwidths.

The group stage of a sports tournament is often made more appealing by introducing additional constraints in the group draw that promote an attractive and balanced group composition. For example, the number of intra-regional group matches is minimised in several World Cups. However, under such constraints, the traditional draw procedure may become non-uniform, meaning that the feasible allocations of the teams into groups are not equally likely to occur. Our paper quantifies this non-uniformity of the 2026 FIFA World Cup draw for the official draw procedure, as well as for 47 reasonable alternatives implied by all permutations of the four pots and two group labelling policies. We show why simulating with a recursive backtracking algorithm is intractable, and propose a workable implementation using integer programming. The official draw mechanism is found to be optimal based on four measures of non-uniformity. Nonetheless, non-uniformity can be more than halved if the organiser aims to treat the best teams drawn from the first pot equally.

The discovery of the galaxy ring known as the Council of Giants (CoG) highlights the need to explain such structures in the Local Universe. In the first paper of this series we presented HINORA - a code to locate (ring-like) structures in 3D point sets - and used it to identify the CoG in the most complete observations of the Local Volume. Here, in Part II, we apply the same method to cosmological simulations to quantify the possible existence of such objects in the LCDM model of structure formation. We analyze DM-only simulations with random and constrained initial conditions, selecting regions that reproduce the properties of the Local Group and Volume, respectively. In order to use the same selection criteria as previsouly done for observations, we relate K-band luminosities to halo masses through semi-empirical relations. After confirming that the selected regions from the simulations match the observed mass function and density of the Local Universe, we use HINORA to search for ring-like structures in them. We find that the existence of CoGs in LCDM simulations is a rather unusual phenomenon. The observed CoG represents an anomaly of more than 2.7 sigma from what is expected in the distribution of massive galaxies in LCDM. These results hint that the CoG could either be a rare chance configuration or the imprint of physical processes at intermediate scales that standard DM-only simulations fail to capture.

Capturing ultrafast spin and charge photocurrents on nanoscopic scales is essential for fundamental research in physics and engineering, as well as for future applications, such as novel spinorbitronic devices. Accessing the fundamental dynamics driven by changes in electronic energy, linear momentum, and angular momentum requires probing that simultaneously resolves their native spatiotemporal scales: femtoseconds and nanometers. However, experimental approaches achieving this simultaneous resolution remain scarce and instrumentally demanding. While near-field probing offers promising platforms to combine ultrafast and nanometer resolution, a major challenge is the observation of in-plane propagation dynamics, required for studying many application-relevant, high-speed phenomena from carrier transport in 2D materials to spin-to-charge conversion in spintronic terahertz emitters (STEs). However, near-field scanning probes are primarily sensitive to out-of-plane electric fields. Here we present a solution to this challenge by performing terahertz (THz) emission nanoscopy (TEN) of a photoexcited fiber-coupled STE using a scanning-probe microscope. We uncover a counterintuitive, dipolar evolution of the near-field THz signal, which we show originates from the scanning-probe's inherent sensitivity to local out-of-plane electric fields that emerge in our case from spatially confined in-plane charge currents. Our findings resolve ambiguities regarding the sensitivity of near-field probes to ultrafast in-plane charge transport and establish TEN as a fully vectorial probe for the spatiotemporal imaging of coupled nanoscale THz charge and spin dynamics.

The quantum cryptographic conferencing (QCC) protocol, which distributes identical secure keys to user groups, is a crucial component of the quantum network. Previous experimental works have implemented the measurement-device-independent (MDI) QCC, of which the key rate in an $N$-user network scales down as $R\sim O(η^N)$, respectively. Building on the MDI QCC protocol, the asynchronous MDI (AMDI) QCC protocol theoretically integrates the mode pairing scheme into QCC, significantly boosting the key rate to $R\sim O(η)$, which is independent of the number of users, and thus demonstrating greater application potential. Experimentally, in this work, we implement the three-user AMDI QCC network without global phase tracking by adopting the fast Fourier transform-based frequency difference estimation and the phase drift compensation technique. Finally, we achieve a key rate of about $4.470\times10^{-9}$ bits per pulse under a maximum overall loss of about 59.6 dB. This work provides a scalable solution for the development of large-scale quantum communication networks in the future.

Photonic computing promises ultrafast and energy-efficient artificial intelligence. However, existing photonic neural networks (PNNs) remain functionally shallow and difficult to scale. Here we establish a theory-guided framework showing that power stability and complex-field correlation are the fundamental prerequisites for scalable, coherent PNNs. Building on these macroscopic principles, we introduce the Coherent, Compensated and Cross-connected (C3) unit - an architecture that integrates coherent nonlinearity, active loss compensation and native optical residual connectivity. Implemented on a silicon-on-insulator platform, the C3 unit provides reconfigurable activation functions and dynamic energy stabilization without external amplification. We validate this framework using a width-constrained spiral benchmark, in which the C3 unit substantially improves parameter utilization and power robustness relative to incoherent nonlinearities. In a high-complexity 1,623-class recognition task, our C3-enabled coherent residual network (CoP-ResNet) achieves a top-1 accuracy of 77.92%, whereas non-residual architectures fail to converge. Together, these results offer a physically grounded, theory-guided pathway toward greater optical processing depth, laying the foundation for next-generation, large-scale photonic computing architectures.

This work develops photoacoustic-based experimental methods for comprehensive characterization of multispecies mass transport from donor compartments to thin-membrane acceptor systems in perfect contact, supported by a dedicated mass transfer modeling framework. Multibeam configurations are implemented in photoacoustic setups operating in front-side detection and diffuse-reflection geometries. The setups are calibrated and adjusted prior to measurements by means of transmission-mode photoacoustic experiments conducted under steady-state conditions. Finally, the methodologies were applied to a model system undergoing photoinduced decay, enabling characterization of bulk transport kinetics as well as interfacial equilibration monitored through time-dependent changes in interfacial reflectivity. For the analysis of bulk transport data, a lumped electrical-circuit (EC) model is introduced. The model is formulated in tableau form, linking species population dynamics to an interaction matrix representing mass-transport couplings consistent with the underlying diffusion-reaction framework. A simplified approximation of the model is further proposed and validated against experimental results. The combined experimental-modeling framework provides an effective approach for quantitative analysis of coupled diffusion, reaction, and interfacial processes in thin-membrane systems.

In a social system individual actions have the potential to trigger spontaneous collective reactions. The way and extent to which the activity (number of actions$-A$) of an individual causes or is connected to the response (number of reactions$-R$) of the system is still an open question. We measure the relationship between activity and response with the distribution of efficiency, a metric defined as $η=R/A$. Generalizing previous results, we show that the efficiency distribution presents a universal structure in three systems of different nature: Twitter, Wikipedia and the scientific citations network. To understand this phenomenon, we develop a theoretical framework composed of three minimal statistical models that contemplate different levels of dependence between $A$ and $R$. The models not only are able to reproduce the empirical activity-response data but also can serve as baselines or null models for more elaborated and domain-specific approaches.

Dynamically tunable terahertz (THz) photonics requires low-loss dielectric platforms with practical, continuous control of refractive index. Here we present a mechanically reconfigurable THz photonic material platform: a monolithic, all-silicon (Si) stretchable effective medium whose refractive index is tuned by deformation. A 200 micrometer-thick high-resistivity single-crystal Si slab was patterned into a subwavelength spiral-spring through-hole lattice, rendering bulk Si mechanically compliant while preserving its low-loss dielectric response. THz time-domain spectroscopy demonstrates high transmission below 0.6 THz and reveals a monotonic decrease in the effective refractive index under uniaxial stretching. At 12.6% elongation, the effective index decreases by 6% and 8% for polarizations perpendicular and parallel to the stretch direction, respectively, thereby demonstrating deformation-induced, controllable anisotropy without a detectable increase in extinction. This structurally engineered bulk-Si approach offers a process-compatible route to mechanically tunable, low-loss THz components for adaptive wavefront and polarization control.

We have retrieved and analyzed several millions of Twitter messages corresponding to the Spanish General elections held on the 20th of December 2015 and repeated on the 26th of June 2016. The availability of data from two electoral campaigns that are very close in time allows us to compare collective behaviors of two analogous social systems with a similar context. By computing and analyzing the time series of daily activity, we have found a significant linear correlation between both elections. Additionally, we have revealed that the daily number of tweets, retweets and mentions follow a power law with respect to the number of unique users that take part in the conversation. Furthermore, we have verified that the topologies of the networks of mentions and retweets do not change from one election to the other, indicating that their underlying dynamics are robust in the face of a change in social context. Hence, in the light of our results, there are several recurrent collective behavioral patterns that exhibit similar and consistent properties in different electoral campaigns.

Political polarization generates strong effects on society, driving controversial debates and influencing the institutions. Territorial disputes are one of the most important polarized scenarios and have been consistently related to the use of language. In this work, we analyzed the opinion and language distributions through Twitter data of a particular territorial dispute around the independence of Catalonia. We infer a continuous opinion distribution by applying a model based on retweet interactions, previously detecting elite users with fixed and antagonist opinions. The resulting distribution presents a mainly bimodal behavior with an intermediate third pole that shows a less polarized society with the presence of not only antagonist opinions. We find that the more active, engaged and influential users hold more extreme positions. Also we prove that there is a clear relationship between political positions and the use of language, showing that against independence users speak mainly Spanish while pro-independence users speak Catalan and Spanish almost indistinctly. However, the third pole, closer in political opinion to the pro-independence pole, behaves similarly to the against-independence one concerning the use of language.

3D nano-printing through two-photon polymerization enables monolithic manufacturing of mechanical and freeform micro-optical elements with high inherent alignment accuracy. However, viscoelasticity and temperature-dependent stiffness of the photopolymer lead to hysteresis and drift, which significantly degrade open-loop position accuracy and long-term stability in quasi-static operation. Therefore, closed-loop control with integrated displacement sensing is essential for 3D nano-printed optical microsystems in practical precision positioning applications. Here, we present a closed-loop control system for a such a lens actuator that uses a commercial 3-axis Hall sensor for position tracking. A NdFeB micromagnet encircling the integrated microlens provides both the actuation force and a position-dependent sensing signal. The Hall sensor, located between an anti-Helmholtz-like coil pair that drives the scanner bidirectionally, measures the combined field due to the micro-magnet and the coil pair. Calibration-based subtraction separates the coil field and sensor offset contributions to recover the magnet field and thus the axial magnet displacement. Closed-loop operation yields a mean absolute accuracy of 0.86 um and a precision of 0.49 um over a displacement range of 150 um while eliminating viscoelastic creep, suppressing hysteresis, and minimizing temperature-induced displacement drift under coil self-heating. This sensing approach requires no additional microfabrication steps and provides a practical path toward stable and repeatable positioning for monolithically 3D nano-printed optical microsystems.

This paper investigates the potential of near-field (NF) indoor communications in the FR2 frequency bands using fully dielectric structures to generate orbital angular momentum (OAM) waves. All-dielectric platforms based on distributions of T-shaped unit cells are employed for this purpose. The unit cell design is based on a circuital approach and analytical formulations, where phase shifts necessary for OAM generation are achieved by varying the dielectric-to-air ratio within the structure. Based on this unit cell, a set of transmitarrays (TAs) are designed to produce specific OAM modes. These TAs are fabricated in-house using stereolithographic 3D printing and experimentally tested. The tests evaluate two key features of OAM beams: the orthogonality of distinct vortex modes, as characterized by their electric field distributions, and their object-avoidance capability, enabled by the central null characteristic of the wavefront. In addition, a field-test within an indoor environment is conducted emulating a real wireless system. A bit error rate lower than 10\textsuperscript{$-$6} is observed for solidary modes in Tx and Rx, whereas orthogonal modes produces an increment in 4 order of magnitude. The obtained results reveals that the prototype is suitable for short-range scenarios, enabling techniques such as OAM-multiplexation or physical-layer security thanks to the effective orthogonality beteween modes.

Thermodynamic Integration (TI) is the state-of-the-art computational technique for accurate Gibbs free energy predictions of solids. Conventional TI schemes start from an NVT harmonic reference and require three successive corrections to recover the Gibbs free energy of the real crystal in the NPT ensemble. However, the NVT-to-NPT correction neglects full cell flexibility. Here, we present a rigorous (and only) two-step TI scheme that operates entirely in the NPT ensemble, eliminating the need for the approximate NVT-to-NPT step. The key methodological advancement is the novel NPT reference that explicitly accounts for full cell fluctuations. The new approach is compared with the conventional one via two complementary case studies. For ice polymorphs, having simple cell-shape distributions, the new approach reproduces conventional TI results with excellent agreement. For CsPbI3, whose black phase exhibits complex cell-shape behavior, we demonstrate that our novel method provides more accurate Gibbs free energy differences than the conventional one. Moreover, the proposed framework maintains comparable computational cost while offering a simplified workflow. Overall, the new NPT TI scheme provides rigorous and direct Gibbs free energy calculations for solids.

We model financial transactions as random walks on activity-driven temporal networks. By enforcing fund conservation, our framework analytically derives heavy-tailed distributions for the stationary balances and transaction sizes. Crucially, the latter is driven by variance in the spending propensity of individuals. Calibrated with empirical data from a closed, digital currency community, the model also reproduces observed correlations between inflows and outflows. Our findings provide a path for understanding emergent properties of the circulation of money.

We propose a photonic native procedure for computing the QR factorization of a matrix using a programmable unitary interferometer mesh. The method configures the mesh through a sequence of local power routing steps within tunable two mode interferometric elements, while reading out the resulting upper triangular factor directly from the optical outputs. The number of physical operations grows quadratically ($O(N^2)$) with matrix size, matching the information content of a dense input and reducing runtime relative to standard digital QR routines, which scale cubically ($O(N^3)$). Beyond single factorizations, the same architecture supports iterative spectral computations by reusing the configured interferometer in a mirrored arrangement that implements the core update step of the QR eigenvalue algorithm. We also describe related optical procedures for Hessenberg reduction and Bidiagonalization, serving as compatible preprocessors for QR and SVD workflows. Finally, we analyze the impact of finite control resolution on the computed factors.

We investigate the propagation of surface acoustic waves (SAWs) in a layered half-space system comprising a continuous, sub-wavelength-thick layer weakly adhering to a substrate. Using finite element simulations, we demonstrate that this configuration - without requiring surface structuration - gives rise to frequency ranges bounded in k-space and characterized by strong SAW attenuation, which we term adhesion-induced resonant attenuation zones. We show that these attenuation zones closely mimic the resonant behavior typically observed in locally resonant metamaterials and can be understood through a mass-spring analogy, where the adhesion between the layer and substrate governs the frequency and width of the attenuation zones. As a practical demonstration, we propose a bilayer configuration as a practical route to experimentally realize adhesion-induced resonant attenuation of SAWs, where a soft and thin interfacial film serves as an intermediate adhesive bonding between the layer and substrate, providing a realistic and tunable interfacial stiffness. Our findings offer a simplified route to achieving SAW manipulation through continuous layered media with tunable adhesion, providing a practical alternative to complex structural designs in SAW-based devices across a broad frequency range.

We introduce a system architecture that addresses a fundamental challenge in deploying language-model agents for autonomous control of scientific instrumentation: ensuring reliability in safety-critical environments. The framework integrates probabilistic reasoning by domain-specialized language models with deterministic execution layers that enforce constraints through structured validation and modular orchestration. By separating intent interpretation, experimental planning, and command verification, the architecture translates high-level scientific goals into verifiable experimental actions. We demonstrate this approach in real-time atomic-resolution scanning probe microscopy experiments operated at room temperature, where the system autonomously generates control strategies, invokes corrective modules, and maintains stable operation under experimentally challenging conditions. Quantitative evaluations show that domain-adapted small language models achieve high routing robustness and command accuracy while operating on consumer-grade hardware. Beyond a specific instrument, the framework establishes a general computational principle for deploying language-model agents in safety-critical experimental workflows, providing a pathway toward scalable autonomous laboratories.

Wave energy, as a renewable energy source, is widely distributed and possesses substantial reserves. However, many existing wave energy harvesters exhibit motion constraints under irregular wave conditions, which limits their energy conversion efficiency. In this study, an omnidirectional wave energy gimbal-based electromagnetic generator (OWG-EMG) is proposed to achieve stable power generation under irregular wave motions. By integrating a three-axis gimbal mechanism with an inertial mass regulator, the proposed harvester effectively converts complex and varying wave-induced motions into relative rotational motion between gimbal frames. A planetary gear transmission is further employed to increase the rotational speed of the electromagnetic generator, enabling efficient omnidirectional wave energy harvesting. Experimental results demonstrate that the device delivers an average output power of up to 0.24 W at a frequency of 1 Hz, corresponding to an average power density of 40 W/m3. Importantly, in addition to its high energy conversion efficiency, the proposed device offers notable advantages including low cost, structural simplicity, and high reliability, highlighting its significant potential for applications in ocean energy harvesting and utilization.

Half-metallic Heusler compounds are of significant interest for spintronics. For device fabrication, compounds that can be epitaxially grown on III-V semiconductors are particularly attractive. We present a first-principles investigation of four Co-based and two Ni-based Heusler compounds that are lattice-matched to InAs. The results of density functional theory (DFT) using semi-local and hybrid functionals are compared to quasiparticle self-consistent $GW$ (QPGW). We also consider DFT with machine-learned Hubbard $U$ corrections [npj Computational Materials 6, 180 (2020)] with a new Bayesian optimization (BO) objective function to determine the $U$ values that yield the closest agreement with the QPGW band structure and magnetic moments. We find that DFT+U(BO) can adequately reproduce the key QPGW features in most cases. Our results reveal a strong method dependence of the degree of spin polarization at the Fermi level and, in some cases, even the dominant spin channel (majority or minority). Of the materials studied here, Co$_2$TiSn and Co$_2$ZrAl are the most likely to be half-metals, and Co$_2$MnIn is likely to be a near-half-metal.

Quantized transport is a prominent feature in topological physics, with canonical examples being the quantum Hall effect and adiabatic Thouless pump, which are based on the Chern number, a topological invariant of 2D systems. Going beyond the Chern-number-based paradigms, quantized transports can also arise from k-direction quasienergy winding unique to periodically driven (Floquet) systems, which are free of dimensionality and adiabaticity limitations. However, lattices displaying winding of their quasienergy bands require asymmetric long-range couplings that are difficult to achieve in lattices of real-space coupled sites. Here, by leveraging photonic synthetic dimensions we construct asymmetric long-range-couplings in a one-dimensional temporal quantum walk based on three coupled fiber loops. We demonstrate quantized transport arising from the winding of quasienergy bands in k direction. We show that the average group velocity of an initial wave packet is proportional to the winding number, which leads to a quantized transport displacement. To better visualize this quantized displacement, we cascade two regions with flipped nearest/long-range couplings and observe a focusing effect with a quantized spatial shift in the focusing point. We also probe the robust properties of quantized transport against obstacles and disorders. The study initiates quasienergy-winding-based topological transports, which can feature applications in precise and robust imaging and information processing.

Bound states in the continuum (BICs), known for their theoretically infinite quality (Q) factors and strong field localization, hold great promise for high-performance photonic devices. However, conventional true BICs typically rely on infinitely periodic structures, and their realization in finite-sized compact structures faces fundamental challenges, which severely limits device miniaturization and integration. In this work, a compact BIC design method based on optical conformal mapping is proposed, where a conventionally infinite periodic structure extended along one direction is mapped into a finite-sized annular structure. This symmetry transition, i.e., from translational to rotational, enables structural miniaturization while fully preserving the eigenvalues and BIC type of the original system. These transformations require only the adjustment of background permittivity and source distribution, without introducing extreme material parameters. As a concrete example, we show through theoretical and numerical analysis that a transformed compact all-dielectric structure, consisting of a double annular dielectric grating embedded in a gradient-index dielectric background, can support true BICs in a finite region. This work provides a simple and general strategy for achieving true BICs in compact all-dielectric structures, paving the way toward miniaturized high-Q photonic devices.

Dissipative microcavity solitons offer a route to integrate comb-based metrology systems on photonic chips. However, integrated solitons generally lack agile control of comb parameters, particularly pulse timing control, hindering their application in quantum-limited metrology. Here we introduce dynamical soliton trapping to enable optically timing-engineered microcombs (OTEM). By injecting an auxiliary laser to anchor one of the microcomb lines, we create a potential well to trap and steer the soliton. Thus, soliton timing can be engineered by phase modulating the injected laser. Theory and measurement reveal the fast response bandwidth of the OTEM, which enabled a soliton slew rate of 31.3 ps/$μ$s, surpassing existing time-programmable fiber laser combs by more than two orders of magnitude. Leveraging the timing scan, we used a single OTEM for single-pixel and parallel ranging by retrieving the phase of the multi-heterodyne beat spectrum. Picometer-scale ranging precision was achieved, establishing new record for optical absolute ranging. Our work can transform timing-engineering of microcavity solitons and endow integrated dual-microcomb metrology systems with enhanced precision.

Multimode fibers (MMFs) provide a compact, high-throughput platform for minimally invasive imaging and information transmission. However, their utility is fundamentally constrained by mode mixing, which renders image transmission spatially disrupted and sensitive to external perturbations. Current imaging methods typically rely on transmission matrix measurement or deep learning models that are fragile to fiber movement, necessitating frequent, time-consuming calibrations and re-calibrations that are easily disrupted and fail to generalize across different fiber configurations, let alone across entirely distinct fibers. Here, we propose a calibration and feedback-free MMF coherent imaging paradigm, that we termed Spatial Harmonic Invariant Nonlinear Encoding (SHINE). By leveraging the angle-dependent phase-matching conditions of second-harmonic generation, we encode spatial features into broadband spectral signatures that possess intrinsic insensitivity not only to modal scrambling but also to fiber bending, movement, as well as structural variations. This spectral representation enables a deep learning model to robustly reconstruct images in real time despite dynamic perturbations and even generalizes well to distinct MMFs without recalibration or feedback. We achieve experimentally an average Pearson correlation coefficient (PCC) of 0.82 for image reconstruction tasks on Fashion-MNIST and a classification accuracy of 92.3% on HERLEV biomedical dataset. Uniquely, our method exhibits remarkable cross-fiber generalization: a model trained on a single MMF successfully reconstructs images transmitted through entirely distinct, previously unseen MMFs with a PCC of 0.74. These results establish a robust, calibration-free framework for imaging through MMFs in real time, paving the way for practical, resilient optical diagnostics that operate without distal-end feedback.

Emulating long-distance light propagation on a laboratory scale is essential for the ground-based testing of intersatellite optical systems. To address this challenge, we propose and analyze a novel optical system called the Range Emulator (RE) to reproduce the spatial propagation effects of a long-distance beam within a compact apparatus. Our analysis identifies that three lenses are required as the minimum number of lenses to implement the RE. Through a numerical exploration, we quantify the fundamental trade-off between system compactness and manufacturing precision. This work provides a practical framework for designing compact optical testbeds for future multi-satellite laser link technologies.

Optical levitation of nano-scale systems offers a pathway to highly sensitive rotation measurements, which are critical for advancing gyroscopic technologies. While prior studies have primarily focused on controlling rotational degrees of freedom of optically levitated particles via modulation of optical power, polarization, and ambient pressure, here we demonstrate wavelength-controlled rotation of the "coin-flip" mode in optically levitated NaYF hexagonal prisms doped with erbium by modulating the wavelength of a secondary pump beam. By switching the pump light wavelength, we precisely modulate the particle's rotation rate in a binary fashion, encoding the ASCII message "hello" in its rotational frequency. Finally, we observe long-term bimodal periodic dynamics in the rotational motion of a levitated prism that are suggestive of a Dzhanibekov (or tennis-racket)-like effect.

The ability to significantly enhance energy transfer processes at the nanoscale requires the simultaneous optimization of molecular scale orientation and macroscopic photonic enhancement between multiple quantum emitters. However, achieving this dual control has remained a significant experimental challenge, often limited by the stochastic arrangement of emitter assemblies and spatially non-uniform electromagnetic fields in conventional photonic platforms. In this work, we demonstrate a unified architecture that achieves this synergy by combining the structural precision of DNA nanotechnology with the unique field environment generated by epsilon-near-zero (ENZ) materials. Using DNA molecular beacons as programmable emitter scaffolds, we establish fixed donor-acceptor separations and emitter orientations (Atto425/Cy3.5) in two well-defined conformational states: closed hairpin (emitter separation 2 nm), and open extended (7.2 nm) configurations. These structures are then embedded in the near-field of a multilayer ENZ metamaterial substrate, which facilitates spatially uniform, enhanced electromagnetic field coupling. Time-resolved photoluminescence measurements demonstrate a significant increase in FRET efficiency for DNA-programmed emitter pairs in the ENZ environment, compared to those on a glass substrate, corresponding to increased donor quenching and shortened donor lifetime. These results establish a scalable experimental pathway for engineering light-matter interactions at molecular scales, enabled by the structural precision of DNA paired with ENZ mediated redistribution of the local density of optical states (LDOS) to amplify near-filed coupling between quantum emitters, with applications in next-generation bio-sensing and quantum photonic technologies.

Halide perovskite quantum dots exhibit cooperative optical phenomena that are absent in conventional semiconductor nanocrystals, including exciton superradiance, superabsorption, and biexciton superradiance within individual dots. Here we develop a microscopic theory that identifies the physical origin of these Dicke effects and establishes how they can be controlled by materials parameters. The central result is that cooperative emission emerges from a competition between collective coupling of optical transition dipoles and lattice-induced disorder, with the balance governed by the Raman-derived phonon spectral density and the excitonic oscillator strength. At elevated temperature, strong Fröhlich coupling and glassy lattice dynamics produce dynamic disorder that suppresses dipole synchronization and yields incoherent emission. Upon cooling, lattice fluctuations freeze and cooperative coherence emerges when the collective coupling exceeds residual static disorder, defining size- and composition-dependent crossover temperatures that we map as phase diagrams. Extending the framework to biexcitons, we show that a confined biexciton constitutes a single correlated charge distribution dressed by a shared lattice configuration, enabling pathway-indistinguishable decay and cooperative radiative enhancement. The theory quantitatively accounts for observed size, composition, and temperature trends in radiative-rate constant ratios and biexciton binding energies, while explaining why full Dicke saturation is not universal. More broadly, the results establish Raman spectral weight and oscillator strength as design parameters for engineering cooperative quantum-optical behavior in quantum materials.

Flexomagnetism is the coupling between magnetism and strain gradients and is a technologically relevant phenomenon. We present a theory of elastic wave propagation in a linear elastic flexomagnetic material with microstructure and strain gradient elastic interactions. The expressions of frequency, phase velocity, and group velocity of longitudinal and transverse waves are derived and are shown to depend on the flexomagnetic coefficient and microstructure. We also show that the effect of flexomagnetism and microstructure can lead to some interesting phenomena in wave propagation, which are not observed in classical linear elasticity theory of waves. Specifically, in contrast to classical linear elastic materials, where wave propagation is non-dispersive, flexomagnetic materials with microstructure can exhibit both normal and abnormal dispersion. It is also noteworthy that, in flexomagnetic materials with gradient elasticity, the phase velocities of transverse waves can exceed those of longitudinal waves, which is atypical in classical elasticity. Furthermore, waves can also attenuate for a certain range of wavenumbers that depend on the flexomagnetic coefficient and microstructural parameters. Finally, we explore the possibility of waves exhibiting zero group velocity modes, where waves are non-propagating but have strong local energy confinement, negative group velocity modes, where the wave packet moves in the opposite direction to that of wave propagation, and the phenomenon of wave freezing, where a propagating wave stops in space without diffusing or spreading.

We report progress toward a CMOS-integrated quantum diamond biosensing platform that combines nitrogen-vacancy (NV) centers in diamond with a custom 40 nm CMOS Single-Photon Avalanche Diode (SPAD) array. The system integrates on-chip active quenching and digital readout with external FPGA-based photon counting, compact microwave delivery, and practical optical excitation and collection schemes to support widefield optically detected magnetic resonance (ODMR). System-level design considerations spanning fluorescence collection efficiency, detector count-rate capability, and microwave homogeneity are analyzed with biological compatibility and scalability in mind. Using superparamagnetic iron oxide nanoparticle (SPION)-labeled HEK293T cells as a representative use case, simple dipole-field estimates indicate that sub-$μ$T sensitivity is required to resolve ODMR shifts within typical ensemble linewidths. Based on the proposed architecture and efficiency analysis, a magnetic field sensitivity of approximately 90 nT/$\sqrt{\mathrm{Hz}}$ per pixel is estimated. These results outline a practical path from optics-heavy quantum diamond microscopes toward compact, CMOS-integrated NV-based biosensors for quantitative magnetic imaging in complex biological environments.

Silicon-organic hybrid (SOH) platforms exhibit exceptional electro-optic (EO) properties, including high-speed operation, low energy consumption, and compact footprints. However, the absence of a scalable poling method for EO polymers, combined with the slow switching speeds characteristic of liquid crystals, has impeded the integration and compatibility of these materials with commercial silicon photonic foundries. On the other hand, the realization of very-large-scale photonic integrated circuits (PICs) in the native silicon photonics platform itself is impeded by the complexities associated with the wavelength and thermal stabilization for microring modulators (MRMs). This study establishes the foundation for a poling-free, CMOS-compatible SOH MRM platform by exploiting simultaneous AC phase shifts in ferroelectric nematic liquid crystals (FN-LCs). We present the first demonstration of an MRM coated with FN-LC, with both the RF signal and the DC bias applied to the same electrodes, taking advantage of the dual phase shift uniquely available in FN-LC. An EO bandwidth of f$_{-6dB}\approx$ 7.8 GHz is achieved \textendash~ the highest reported value for an SOH MRM to date. As a proof of concept, we demonstrate an approximately linear resonance shift across the full width at half maximum ($\approx$ 150 pm/V), with a static power efficiency of $\approx$ 4.5 nW/$π$ for an MRM occupying a total footprint of $\approx$ 0.084 mm$^2$ and exhibiting an on-chip optical insertion loss of $\approx$ 0.78 dB. Successful infiltration of FN-LC, selectively patterned on top of phase shifters, along with optical input/output channels established using free-form photonic wire bonds, is demonstrated in the proposed PIC.

We present an on-line tuning strategy for the ISAC post-accelerator that pre-sets machine optics with a digital twin and then performs Bayesian optimization for steering under online operation with beam. The model computes end-to-end tunes in seconds and interfaces with the control system under device bounds, slew-rate limits, and loss interlocks. We report three experimental case studies demonstrating that decoupling optics from steering yields faster and more reliable convergence than a fully Bayesian optics-plus-steering baseline under identical conditions. Across these cases, iterations to high transmission tunes are reduced by a factor of 4-6, with final average transmissions in the mid- to high-90% range. By factorizing optics from steering, the dimensionality of the parameter space is reduced, convergence becomes more predictable, and operational safeguards are easier to enforce.

We present trainsum, a versatile Python package for doing computations with multidimensional quantics tensor trains: https://github.com/fh-igd-iet/trainsum. Using the Array API standard together with opt_einsum, trainsum allows the effortless approximation of tensors or functions by tensor trains independent of their shape or dimensionality. Once approximated, our package can perform normal arithmetic operations with quantics tensor trains, including addition, Einstein summations and element-wise transformations. It can be therefore used for generic computations with applications in simulation, data compression, machine learning and data analysis.

The basic building blocks of many forms of optical topologies are particle-like singularities in phase and polarisation, giving rise to lines of darkness that weave complex threads in 3D space. Although known for half a century since seminal work on dislocations in wave trains, their behaviour in complex media remains under debate, especially with respect to their relative stability. Here we show that polarisation and phase vortices behave identically in one-sided turbulent complex channels. We perform complementary numerical and experimental studies using atmospheric turbulence as a test case, demonstrating agreement and equivalent dynamics. Our work addresses open questions on optical topologies and will be relevant to their harnessing for applications such as sensing, communication, imaging, and information transfer in noisy or complex environments.

Visualizing the spatial profile of light beams is essential for evaluating irradiance, characterizing beam quality, and achieving precise alignment. In the optical spectral range, this is readily performed using silicon-based CCD and CMOS cameras. In the terahertz (THz) range, however, it typically requires specialized detectors with prohibitive costs. Here, we show that an infrared (IR) camera can be used outside of its labeled specifications to achieve similar performance as a dedicated microbolometric THz camera, at under 1% of the THz camera's cost. We compared the cameras by characterizing THz beam profiles from two sources: a pulsed broadband THz beam produced through optical rectification in organic crystals, and a narrowband quasi-continuous-wave (quasi-CW) THz beam emitted by a quantum cascade laser. For the broadband THz radiation, the beam width measured by the two cameras differed by only ~ 6%, well within the pixel resolution limit, and in the narrowband quasi-CW case by just ~ 1.3%. Additionally, both cameras show linear responsivity over a comparable irradiance range. These results expand the applicability of conventional IR cameras to the THz range, suggesting that they will become routine tools for high-fidelity THz beam diagnostics and imaging in scientific and industrial applications.

Dengue remains a major global public health concern due to its high mortality and economic burden. Mathematical modeling is essential to understand its transmission mechanisms and for evaluating intervention strategies. In this paper, we formulate a vector host model in which the human population is structured by age, and vaccinated individuals are further described by time since vaccination. The mosquito population is coupled to the host dynamics and reduced under a quasi steady state assumption. By integrating over vaccination age, we obtain a nonlinear steady state formulation and express the endemic equilibrium as a fixed point problem for the infected mosquito population. Using Lipschitz estimates and a contraction argument, we establish existence and uniqueness of the equilibrium under a weak transmission condition. The analysis highlights the influence of age dependent vaccination on long term dengue dynamics.

Biological systems like long-lived clonal organisms, holobionts and clades challenge traditional evolutionary thinking since they adapt without populations or reproduction. This paper aims to provide an overarching theoretical framework which encompasses standard Darwinian evolution as well as other processes of adaptation. This framework is cumulative selection and I provide a general `recipe' for it to occur. Lewontin's recipe for evolution by natural selection is shown to be a particular example of cumulative selection, but not the only one. Similarly, reproduction, inheritance and populations are just one way to perform cumulative selection. I discuss several other examples of cumulative selection including clonal organisms, dioecious populations, Gaia and neural networks.

Integrating high-dimensional biological data into data-driven mechanistic modeling requires rigorous practical identifiability to ensure interpretability and generalizability. However, coordinate identifiability analysis often suffers from numerical instabilities near singular local minimizers. We present a computational framework that uncovers fundamental scaling laws governing practical identifiability through asymptotic analysis. By synthesizing Fisher information with perturbed Hessian matrices, we establish a hierarchical approach to quantify coordinate identifiability and inform uncertainty quantification within non-identifiable subspaces across different orders. Supported by rigorous mathematical analysis and validated on synthetic and real-world data, our framework was applied to HIV-host dynamics and spatiotemporal amyloid-beta propagation. These applications demonstrate the framework's efficiency in elucidating critical mechanisms underlying HIV diagnostics and Alzheimer's disease progression. In the era of large-scale mechanistic digital twins, our framework provides the scaling laws for data-driven modeling in terms of both parameter identifiability and uncertainty, ensuring that data-driven inferences are grounded in verifiable biological reality.

We construct a new class of infinite-dimensional diffusions taking values in a generalized Kingman simplex. Our model describes the temporal evolution of the relative frequencies of infinitely-many types which are "labeled" by an arbitrary finite number of marks or colors, but "unlabeled" within each mark. We start with a finite-dimensional construction which extends to Wright-Fisher diffusions a self-similarity property known for Dirichlet distributions, and corresponds to a multiple skew-product representation of the Wright-Fisher diffusion relative to the marks in the population. After ranking decreasingly the frequencies within each mark, we identify the limit in distribution of the resulting diffusion when the number of types for each mark goes to infinity, and describe its infinitesimal operator. The limiting process reduces to a diffusion in the Thoma simplex in the special case of only two marks, whereas the infinitely-many-neutral-alleles model is recovered when all frequencies have the same mark. The stationary measure of the limit diffusion is shown to be the recently introduced multiple Poisson-Dirichlet distribution, which extends Kingman's Poisson-Dirichlet distribution and is the de Finetti representing measure for a family of random partitions whose elements are marked.

Purpose: To determine whether targeted T2 fluid-attenuated inversion recovery (T2-FLAIR) dropout training improves glioblastoma MRI tumor segmentation robustness to missing T2-FLAIR without degrading performance when T2-FLAIR is available. Materials and Methods: This retrospective multi-dataset study developed nnU-Net models on BraTS 2021 (n=848) and externally tested them on UPenn-GBM glioblastoma MRI (n=403; 2006-2018; age 18-89 years; 60% male). Models were trained with no dropout or targeted T2-FLAIR dropout (probability rate r=0.35 or 0.50) by replacing only the T2-FLAIR channel with zeros. Inference used T2-FLAIR-present and T2-FLAIR-absent scenarios (T2-FLAIR set to zero). The primary endpoint was Dice similarity coefficient (DSC); secondary endpoints were 95th percentile Hausdorff distance and Bland-Altman whole-tumor volume bias. Equivalence was assessed with two one-sided tests using +/-1.5 DSC percentage points, and noninferiority versus HD-GLIO used a -1.5-point margin. Results: With T2-FLAIR present, median overall DSC was 94.8% (interquartile range, 90.0%-97.1%) with dropout and 95.0% (interquartile range, 90.3%-97.1%) without dropout (equivalence supported, p<0.001). With T2-FLAIR absent, median overall DSC improved from 81.0% (interquartile range, 75.1%-86.4%) without dropout to 93.4% (interquartile range, 89.1%-96.2%) with dropout (r=0.35); edema DSC improved from 14.0% to 87.0%, edema 95th percentile Hausdorff distance improved from 22.44 mm to 2.45 mm, and whole-tumor volume bias improved from -45.6 mL to 0.83 mL. Dropout was noninferior to HD-GLIO under T2-FLAIR-present (all p<0.001). Conclusion: Targeted T2-FLAIR dropout preserved segmentation performance when T2-FLAIR was available and reduced segmentation error and whole-tumor volume bias when T2-FLAIR was absent.

The Euler characteristic transform (ECT) is an emerging and powerful framework within topological data analysis for quantifying the geometry of shape. The applicability of ECT has been limited due to its sensitivity to noisy data. Here, we introduce SampEuler, a novel ECT-based shape descriptor designed to achieve enhanced robustness to perturbations. We provide a theoretical analysis establishing the stability of SampEuler and validate these properties empirically through pairwise similarity analyses on a benchmark dataset and showcase it on a thymus dataset. The thymus is a primary lymphoid organ that is essential for the maturation and selection of self-tolerant T cells, and within the thymus, thymic epithelial cells are organized in complex three-dimensional architectures, yet the principles governing their formation, functional organization, and remodeling during age-related involution remain poorly understood. Addressing these questions requires robust and informative shape descriptors capable of capturing subtle architectural changes across developmental stages. We develop and apply SampEuler to a newly generated two-dimensional imaging dataset of mouse thymi spanning multiple age groups, where SampEuler outperforms both persistent homology--based methods and deep learning models in detecting subtle, localized morphological differences associated with aging. To facilitate interpretation, we develop a vectorization and visualization framework for SampEuler, which preserves rich morphological information and enables identification of structural features that distinguish thymi across age groups. Collectively, our results demonstrate that SampEuler provides a robust and interpretable approach for quantifying thymic architecture and reveals age-dependent structural changes that offer new insights into thymic organization and involution.

The correlated color temperature (CCT) and color rendering index (CRI) of artificial light sources are important because they have implications for human biology and professional applications. Although CCT information is generally available for commercial lamps, CRI is commonly not reported. In addition, devices measuring these parameters are difficult to access as they require a spectrophotometer, a commonly expensive device. In this context, the present work designs and builds a meter in detail, from the structural part of the equipment, interface with sensors, and the calculation to the compensation algorithms implementation, aiming to build the dedicated functionalities of a spectrophotometer, which is designed without the use of optical lenses. In addition to simplifying the device, this approach allows measurements free from dispersions caused by chromatic aberrations typical of optical lenses. The prototype obtained proved to be effective, capturing the spectral power distributions of various light sources and calculating their CCT and CRI.

The signal-to-interference-plus-noise ratio (SINR) is central to performance optimization in user-centric beamforming for satellite-based non-terrestrial networks (NTNs). Its assessment either requires the transmission of dedicated pilots or relies on computing the beamforming matrix through minimum mean squared error (MMSE)-based formulations beforehand, a process that introduces significant computational overhead. In this paper, we propose a low-complexity SINR estimation framework that leverages multi-head self-attention (MHSA) to extract inter-user interference features directly from either channel state information or user location reports. The proposed dual MHSA (DMHSA) models evaluate the SINR of a scheduled user group without requiring explicit MMSE calculations. The architecture achieves a computational complexity reduction by a factor of three in the CSI-based setting and by two orders of magnitude in the location-based configuration, the latter benefiting from the lower dimensionality of user reports. We show that both DMHSA models maintain high estimation accuracy, with the root mean squared error typically below 1 dB with priority-queuing-based scheduled users. These results enable the integration of DMHSA-based estimators into scheduling procedures, allowing the evaluation of multiple candidate user groups and the selection of those offering the highest average SINR and capacity.

This paper proposes a novel optimization framework for enhancing the security resilience of cell-free massive multiple-input multiple-output (CF-mMIMO) networks with multi-antenna access points (APs) and protective partial zero-forcing (PPZF) under active eavesdropping. Based on the main principles of absorption, adaptation, and recovery, we formulate a security-aware resilience metric to quantify the system performance during and after a security outage. A multi-user service priority-aware power allocation problem is formulated to minimize the mean squared error (MSE) between real-time and desired security efficiency, thereby enabling a trade-off between the target user's secrecy performance and multi-user quality of service (QoS). To solve this non-convex problem, a security-aware iterative algorithm based on the successive convex approximation (SCA) is employed. The proposed algorithm determines the optimal power allocation strategy by balancing solution quality against recovery time. At each iteration, it evaluates the overall resilience score and selects the strategy that achieves the highest value. Simulation results confirm that the proposed framework significantly improves the resilience of CF-mMIMO networks, allowing flexible adaptation between rapid recovery and high-quality recovery, depending on system requirements.

We introduce the broadband reconfigurable intelligent surface (RIS) channel (BRISC) dataset. The dataset comprises measurements of channel state information (CSI) collected at 5.53 GHz using a 256-element RIS with binary states. In the measurement campaign, the transmitter and receiver are two software defined radios (SDRs), phase-synchronized via an OctoClock, where the transmitter (receiver) is equipped with one (two) antenna(s). To manage complexity, the RIS elements are grouped into blocks of different sizes, where all elements within a block share the same state. CSIs have been captured for multiple a) transmitter positions (and fixed receiver location), b) pilot block sizes, and c) state configurations. Furthermore, we calibrated the parameters of state-of-the-art RIS channel models to fit the measured CSI. With approximately 10000 configurations explored per transmitting position, BRISC serves as a robust benchmark in communication applications. We also show here an example of its use for physical-layer authentication.

This paper proposes a distributed continuous aperture array (D CAPA) to support simultaneous wireless information and power transfer (SWIPT) to multiple information users (IUs) and energy users (EUs). Each metasurface supports continuous surface currents that radiate electromagnetic (EM) waves for information and energy transmission to the users. These waves propagate through continuous EM channels characterized by the dyadic Green function. We formulate a system power consumption (PC) minimization problem subject to spectral efficiency and energy harvesting quality of service (QoS) requirements, where the QoS requirements are derived under the equal power allocation (EPA) scheme. An efficient two layer optimization algorithm is developed to solve this problem by optimizing the power allocation subject to the QoS violation penalties using augmented Lagrangian transformation. Our numerical results show that well optimized current distributions over each metasurface in the proposed D CAPA achieve up to 65% and 61% reductions in overall system PC compared to the EPA and colocated CAPA (C CAPA) cases, while maintaining the same total aperture size and transmission power.

Over-the-air computation (OAC) enables low-latency aggregation over multiple-access channels (MACs) by exploiting the superposition property of the wireless medium to compute functions efficiently in distributed networks. A critical but often overlooked challenge is that electromagnetic interference in practical radio channels frequently exhibits heavy-tailed behavior, causing strong impulsive noise that severely degrades computation performance. This work studies digital OAC with QAM-based signaling under heavy-tailed interference modeled by a Cauchy distribution (lacking a finite second moment). We seek QAM-like constellations that minimize the mean-squared error (MSE) of sum aggregation subject to an average-power constraint. The problem is formulated as a constrained optimization, whose solution yields unique optimality conditions. Numerical results confirm the effectiveness of the proposed design. Notably, the framework extends naturally to nomographic functions, broader constellation families, and alternative noise models.

This study explores a next-generation multiple access (NGMA) framework for cell-free massive MIMO (CF-mMIMO) systems enhanced by stacked intelligent metasurfaces (SIMs), aiming to improve simultaneous wireless information and power transfer (SWIPT) performance. A fundamental challenge lies in optimally selecting the operating modes of access points (APs) to jointly maximize the received energy and satisfy spectral efficiency (SE) quality-of-service constraints. Practical system impairments, including a non-linear harvested energy model, pilot contamination (PC), channel estimation errors, and reliance on long-term statistical channel state information (CSI), are considered. We derive closed-form expressions for both the achievable SE and the average sum harvested energy (sum-HE). A mixed-integer non-convex optimization problem is formulated to jointly optimize the SIM phase shifts, APs mode selection, and power allocation to maximize average sum-HE under SE and average harvested energy constraints. To solve this problem, we propose a centralized training, decentralized execution (CTDE) framework based on deep reinforcement learning (DRL), which efficiently handles high-dimensional decision spaces. A Markovian environment and a normalized joint reward function are introduced to enhance the training stability across on-policy and off-policy DRL algorithms. Additionally, we provide a two-phase convex-based solution as a theoretical robust performance. Numerical results demonstrate that the proposed DRL-based CTDE framework achieves SWIPT performance comparable to convexification-based solution, while significantly outperforming baselines.

Recent advances in neuromorphic signal processing have introduced time encoding machines as a promising alternative to conventional uniform sampling for low-power communication receivers. In this paradigm, analog signals are converted into event timings by an integrate-and-fire circuit, allowing information to be represented through spike times rather than amplitude samples. While event-driven sampling eliminates the need for a fixed-rate clock, receivers equipped with integrate-and-fire time encoding machines, called time encoding receivers, often assume perfect symbol synchronization, leaving the problem of symbol timing recovery unresolved. This paper presents a joint timing recovery and data detection framework for integrate-and-fire time encoding receivers. The log-likelihood function is derived to capture the dependence between firing times, symbol timing offset, and transmitted sequence, leading to a maximum likelihood formulation for joint timing estimation and sequence detection. A practical two-stage receiver is developed, consisting of a timing recovery algorithm followed by a zero-forcing detector. Simulation results demonstrate accurate symbol timing offset estimation and improved symbol error rate performance compared to existing time encoding receivers.

Traditional protection systems for microgrids, which rely on high fault currents and continuous communication, struggle to keep up with the changing dynamics and cybersecurity concerns of decentralized networks. In this study, we introduce a novel biologically inspired protection system based on neuromorphic principles, where each distributed energy resource (DER) functions as a simple neuron. These neurons process local changes in voltage, current signals, and converting them into spike patterns that represent the severity of disturbances. Just as neurons communicate via synapses in biological systems, we exploit transmission cables to coordinate between DERs, enabling them to share information and respond to faults collectively. Fault detection and circuit breaker activation are driven by a First-To-Spike (FTTS) mechanism, similar to the concept of traveling wave protection, but without needing GPS synchronization or communication links. A key innovation is the ability to use the timing of spikes to locally determine the nature of a fault, offering an intelligent, adaptive response to disturbances. Performance shows tripping latency of 10-58 ms, surpassing conventional relays and even traveling-wave methods (60 ms), while maintaining detection accuracy above 98% and spatial selectivity over 97%, enabling real-time, communication-free, scalable protection for plug-and-play microgrids.

Affine frequency division multiplexing (AFDM) has recently emerged as a resilient waveform candidate for high-mobility next-generation wireless systems. However, current literature mostly focuses on discrete time (DT) models, often overlooking effects and hardware non-idealities of actual continuous time (CT) signal generation. In this paper, we bridge this gap by developing a CT-analytical framework based on the affine Fourier series (AFS) representation, which allows us to demonstrate that strictly bandlimited pulses and subcarrier suppression strategies are essential to maintain the multicarrier structure of the signal. In addition, we derive the analytical power spectral density of AFDM and evaluate its spectral characteristics in comparison with other multicarrier schemes, considering the impact of realistic truncated pulse-shaping. Furthermore, we analyze the sensitivity of the CT model to phase noise, carrier frequency offset, and sampling jitter, providing a theoretical analysis of communication performance. Finally, we derive closed-form Cramér-Rao bounds for channel parameter estimation, showing that the chirped modulation peculiar of AFDM increases the estimation variance but enables the resolution of Doppler ambiguities. Our findings provide the necessary theoretical and practical foundations for the implementation of AFDM in realistic wireless transceivers.

Physical-layer characteristics, such as channel state information (CSI) and transmitter noise induced by hardware impairments, are often uniquely associated with a transmitter. This paper investigates transmitter anonymity at the physical layer from a signal design perspective. We consider an anonymous communication problem where the receiver should reliably decode the signal from the transmitter but should not make use of the signal to infer the transmitter's identity.Transmitter anonymity is quantified using a Kullback-Leibler divergence (KLD)-based metric, which enables the formulation of explicit anonymity constraints in the precoder design.We then propose an anonymous symbol-level precoding strategy that preserves reliable communication under spatial multiplexing while preventing transmitter identification. The proposed framework employs a partitioned equal-gain combining (P-EGC) scheme that leverages receiver diversity without requiring transmitter-specific CSI. Simulation results demonstrate anonymity-reliability tradeoffs across different signal-to-noise ratios (SNRs) and numbers of data streams. Moreover, the results reveal opposite trends of anonymity with respect to transmitter-dependent noise variations in the low-SNR and high-SNR regimes.

The graph fractional Fourier transform (GFRFT) generalizes the graph Fourier transform (GFT) but suffers from a significant computational bottleneck: determining the optimal transform order requires expensive eigendecomposition and matrix multiplication, leading to $O(N^3)$ complexity. To address this issue, we propose a fast GFRFT (FGFRFT) algorithm for unitary GFT matrices based on Fourier series approximation and an efficient caching strategy. FGFRFT reduces the complexity of generating transform matrices to $O(2LN^2)$ while preserving differentiability, thereby enabling adaptive order learning. We validate the algorithm through theoretical analysis, approximation accuracy tests, and order learning experiments. Furthermore, we demonstrate its practical efficacy for image and point cloud denoising and present the fractional specformer, which integrates the FGFRFT into the specformer architecture. This integration enables the model to overcome the limitations of a fixed GFT basis and learn optimal fractional orders for complex data. Experimental results confirm that the proposed algorithm significantly accelerates computation and achieves superior performance compared with the GFRFT.

In continuous aperture arrays (CAPAs), careful consideration of the underlying physics is essential, among which electromagnetic (EM) mutual coupling plays a critical role in beamforming performance. Building on a physically consistent mutual coupling model, the beamforming design is formulated as a functional optimization whose optimality condition leads to a Fredholm integral equation. The incorporation of the coupling model, however, substantially increases computational complexity, necessitating efficient and accurate integral equation solvers. In this letter, we propose two efficient solvers: 1) a coordinate-transformation-based kernel approximation that preserves the operator structure while alleviating discretization demands, and 2) a direct lower-upper (LU)-based solver that stably handles the Nyström-discretized system. Numerical results demonstrate improved accuracy and reduced computational overhead compared to conventional methods, with the LU-based solver emerging as an efficient and scalable solution for large-scale CAPA optimization via offline factorization.

We study the recovery of an unknown three-dimensional band-limited signal from multiple noisy observations that are randomly rotated by latent elements of SO(3), where the rotations are drawn from an unknown, non-uniform distribution. Because the rotations are unobserved, only the signal orbit under the rotation group can be recovered. We show that the signal orbit and the rotation distribution are jointly identifiable from the first and second moments. This yields an improved high-noise sample complexity that scales quadratically with the noise variance, rather than cubically as in the uniform-rotation case. We further develop a provable, computationally efficient reconstruction algorithm that recovers the 3-D signal by successively solving a sequence of well-conditioned linear systems. The algorithm is validated through extensive numerical experiments. Our results provide a principled and tractable framework for high-noise 3-D orbit recovery, with potential relevance to cryo-electron microscopy and cryo-electron tomography modeling, where molecules are observed in unknown orientations.

Beyond diagonal reconfigurable intelligent surface (BD-RIS) architectures offer superior beamforming gain over conventional diagonal RISs. However, the channel estimation overhead is the main hurdle for reaping the above gain in practice. This letter addresses this issue for group-connected BDRIS aided uplink communication from multiple multi-antenna users to one multi-antenna base station (BS). We first reveal that within each BD-RIS group, the cascaded channel associated with one user antenna and one BD-RIS element is a scaled version of that associated with any other user antenna and BD-RIS element due to the common RIS-BS channel. This insight drastically reduces the dimensionality of the channel estimation problem. Building on this property, we propose an efficient two-phase channel estimation protocol. In the first phase, the reference cascaded channels for all groups are estimated in parallel based on common received signals while determining the scaling coefficients for a single reference antenna. In the second phase, the scaling coefficients for all remaining user antennas are estimated. Numerical results demonstrate that our proposed framework achieves substantially lower estimation error with fewer pilot signals compared to state-of-the-art benchmark schemes.

Continuous signal representations are naturally suited for inverse problems, such as magnetic resonance imaging (MRI) and computed tomography, because the measurements depend on an underlying physically continuous signal. While classical methods rely on predefined analytical bases like B-splines, implicit neural representations (INRs) have emerged as a powerful alternative that use coordinate-based networks to parameterize continuous functions with implicitly defined bases. Despite their empirical success, direct comparisons of their intrinsic representation capabilities with conventional models remain limited. This preliminary empirical study compares a positional-encoded INR with a cubic B-spline model for continuous function fitting from sparse random samples, isolating the representation capacity difference by only using coefficient-domain Tikhonov regularization. Results demonstrate that, under oracle hyperparameter selection, the INR achieves a lower normalized root-mean-squared error, yielding sharper edge transitions and fewer oscillatory artifacts than the oracle-tuned B-spline model. Additionally, we show that a practical bilevel optimization framework for INR hyperparameter selection based on measurement data split effectively approximates oracle performance. These findings empirically support the superior representation capacity of INRs for sparse data fitting.

The joint optimization of the integer matrix $\mathbf{A}$ and the power scaling matrix $\mathbf{D}$ is central to achieving the capacity-approaching performance of Integer-Forcing (IF) precoding. This problem, however, is known to be NP-hard, presenting a fundamental computational bottleneck. In this paper, we reveal that the solution space of this problem admits a intrinsic geometric structure: it can be partitioned into a finite number of conical regions, each associated with a distinct full-rank integer matrix $\mathbf{A}$. Leveraging this decomposition, we transform the NP-hard problem into a search over these regions and propose the Multi-Cone Nested Stochastic Pattern Search (MCN-SPS) algorithm. Our main theoretical result is that MCN-SPS finds a near-optimal solution with a computational complexity of $\mathcal{O}\left(K^4\log K\log_2(r_0)\right)$, which is polynomial in the number of users $K$. Numerical simulations corroborate the theoretical analysis and demonstrate the algorithm's efficacy.

The increasing demand for wireless connectivity necessitates advanced spectrum modeling to enable efficient spectrum sharing for next-generation aerial communications. While traditional models often overlook vertical variations in signal behavior, this paper proposes a height-dependent propagation model using a helikite-mounted software-defined radio (SDR). We collected extensive measurement data across the 88 MHz to 6 GHz range in both urban and rural environments. As a case study to validate our methodology, we focus on the FM radio band, which allows us to use publicly available transmitter locations and transmit power levels to facilitate comparisons between analytical with measurement results. We identify a clear transition from non-line-of-sight (NLoS) to line-of-sight (LoS) regimes at a specific altitude threshold and propose an altitude-dependent path loss model that incorporates this transition. Our results demonstrate that the proposed model significantly outperforms the standard free space path loss (FSPL) model in complex urban topologies, providing a more accurate framework for altitude-aware spectrum prediction and management across emerging aerial wireless technologies and bands.

Nowadays, waveforms of integrated sensing and communication (ISAC) are almost based on conventional communication and sensing signal, which bounds both the communication and sensing performance. To deal with this issue, in this paper, a novel waveform design is presented for the partial-time superimposed (PTS) ISAC system. At the base station (BS), a parameter-adjustable linear frequency modulation (LFM) pulse signal and a continuous communication orthogonal frequency division multiplexing (OFDM) signal are employed to broadcast public information and perform sensing tasks, respectively, using a PTS scheme. Pulse compression gain enhances the system's long-range sensing capability, while OFDM ensures the system's high-speed data transmission capability. Meanwhile, the LFM signal is utilized as superimposed pilot for channel estimation, which has higher time-frequency resource utilization and stronger real-time performance compared to orthogonal pilots. We present an accurate parameter estimation method of multi-path sensing signal for reconstructing and interference cancellation in communication users. Additionally, a cyclic maximum likelihood method is introduced for channel estimation and the Cramer-Rao lower bound (CRLB) of channel estimation is derived. Simulations demonstrate the accuracy and robustness of the proposed parameter estimation algorithm as well as the improved channel estimation performance over traditional methods. The proposed waveform design method can achieve reliable data transmission and accurate target sensing.

In this paper, we explore a cooperative integrated sensing and communication (ISAC) framework that utilizes orthogonal frequency division multiplexing (OFDM) waveforms. Under the control of a central processing unit (CPU), multiple access points (APs) collaboratively perform multistatic sensing while providing communication service in a cell-free multiple-input multiple-output (MIMO) system. Achieving high sensing accuracy requires the collection of global sensing information at the CPU, which can lead to significant fronthaul signaling overhead due to the feedback of the sensing signals from each AP. To tackle this issue, we propose a collaborative processing scheme in which the APs locally compress and quantize the received sensing signals before forwarding them to the CPU. The CPU then aggregates the information from all APs to estimate the location and velocity of the targets. We develop a distributed vector-quantized variational autoencoder (D-VQVAE) to enable an end-to-end implementation of this scheme. D-VQVAE consists of distributed encoders at the APs to locally encode the received sensing signals, codebooks for quantizing the encoded results, and a decoder at the CPU for location and velocity estimation. It effectively reduces the amount of data transmitted from each AP to the CPU while maintaining a high sensing accuracy. We employ a collaborative learning-assisted scheme to train D-VQVAE in an end-to-end manner. Simulation results show that the proposed D-VQVAE network outperforms the baseline schemes in sensing accuracy and reduces fronthaul signaling overhead by 99% when compared with the centralized sensing approach.

Motivated by recent advances in digital quantum emulation using noisy intermediate-scale quantum (NISQ) devices and an increased interest in topological Green's function zeros in condensed matter systems, we here study Green's function zeros in topological Floquet systems. We concentrate on interacting Kitaev-like Floquet chains (or equivalently transverse field Ising circuits) and introduce Floquet Green's-function-based topological invariants for the corresponding symmetry class BDI. In the vicinity of special points in the free fermion phase diagram and using tailor-made interactions which lead to the Floquet version of symmetric mass generation, we analytically calculate both edge and bulk Green's functions. Just as in the case of continuum time evolution, topological bands of Green's function zeros may also contribute to the topological invariant. However, contrary to the case of continuum time evolution, Floquet Green's functions can have zeros even in the absence of interactions. Finally, we also discuss an implementation of this Floquet system in a digital quantum emulator: We present a circuit which encodes the interaction under consideration and pinpoint the observables carrying information about the topological Green's function boundary zeros.

We build an exact framework to evaluate heat, energy, and particle transport between Gaussian reservoirs mediated by a quadratic quantum system. By combining full counting statistics with newly developed non-Markovian master equation approaches, we introduce an effective master equation whose solution can be used to generate arbitrary moments of the heat statistics for any number of reservoirs. This theory applies equally to fermionic and bosonic systems, holds at arbitrarily strong coupling, and resolves out-of-equilibrium transient dynamics determined by the system's initial state. In the steady-state, weak-coupling limit, we recover results analogous to those of the well-known Landauer-Büttiker formalism. We conclude our discussion by demonstrating an application of the method to a prototypical fermionic system. Our results uncover a regime of transient negative heat conductance contingent upon the initial system preparation, providing a clear signature of non-trivial out-of-equilibrium dynamics.

Recent DESI DR2 BAO measurements, when combined with CMB and SNeIa data, exhibit a $3.2σ$-$3.4σ$ preference for dynamical dark energy (DE) described by the CPL-parametrized equation of state. A particularly striking feature of these reconstructions is an apparent transition from an early-time phantom-like regime to a late-time quintessence-like behavior. For positive-definite DE densities, this transition is often phrased as a crossing of the phantom divide line (PDL) at $w(a)=-1$. Allowing the DE density to become negative, however, renders the PDL (in the sense of $w(a)=-1$) non-diagnostic as a global separator: the physically meaningful criterion is instead the null energy condition boundary (NECB), $ρ_{\rm DE}+p_{\rm DE}=0$. We therefore test whether the data-driven preference for NECB-crossing in CPL reconstructions persists once alternative realizations of phantom behavior are admitted, specifically through sign-switching DE densities. To this end, we introduce and constrain two controlled phenomenological extensions of the CPL framework featuring a negative DE phase in the past. In the CPL$\to-Λ$ model, the switching epoch is tied to the CPL-inferred NECB-crossing scale factor, yielding an early-time negative cosmological-constant phase, while the post-switch evolution follows the CPL branch. In the sCPL model, the CPL equation of state is maintained at all times, while the sign switch in the energy density occurs at an independent transition redshift. We find that late-time BAO and SNeIa data drive the negative-density phase beyond their effective redshift coverage, and that this requirement is the primary driver of the inferred parameter behavior. While both models are statistically disfavored relative to the baseline CPL, admitting a negative DE phase generally reduces the significance of deviations from a cosmological constant.

Hamiltonian formulations of lattice field theories provide access to real-time dynamics, but their simulation is difficult to implement efficiently. Trotter-Suzuki decompositions are at the center of time evolution computation, either on quantum hardware or classically, for instance with the use of tensor networks. While low-order Trotterizations remain the standard choice due to their simplicity, higher-order schemes offer the potential for improved efficiency. In this work we outline a short guide to Trotter-Suzuki schemes and their implementations in general. To help with this, we highlight new efficient schemes found by our optimization framework, and demonstrate their performance on the Heisenberg model.

Phenomena concerning altermagnets have opened up a window for unconventional analysis of the momentum space spin polarization (MSSP) of antiferromagnetic materials. Taking the example of one of the widely investigated altermagnets, CrSb, we explore the underlying mechanisms leading to the formation or breaking of altermagnetism. With the aid of DFT calculation and symmetry analysis, we study the behavior of MSSP in the altermagnetic bands of pristine CrSb, along with a few model structures designed from the pristine one by hypothetical vacancy engineering and interstitial doping. We show that the six-fold rotational symmetry of the pristine CrSb can be reduced to a two-fold rotational symmetry via vacancy and doping engineering. We discover the formation of fragmented nodal curves (FNCs) across the Brillouin zone when in an altermagnetic material when the symmetry is restricted to two-fold rotation. Unlike the typical nodal planes and axes, the location of the FNCs in the momentum space is found to be band-specific. The formation of FNCs is further validated by introducing uniaxial strain to CrSb and by examining the band structure of RbMnPO$_4$, as they both exhibit a two-fold rotational symmetry responsible for altermagnetism. We observe that, unlike the pristine case, these FNCs have the potential to manifest anomalous Hall conductivities (AHC), while the Néel vector orients along both in-plane and out-of-plane directions. This flexibility of the AHC will pave the way for the application of altermagnets in the futuristic quantum devices.

Decoding Low-Density Parity-Check (LDPC) codes is a fundamental problem in coding theory, and Belief Propagation (BP) is one of the most popular methods for LDPC code decoding. However, BP may encounter convergence issues and suboptimal performance, especially for short-length codes and in high-noise channels. The Quantum Approximate Optimization Algorithm (QAOA) is a type of Variational Quantum Algorithm (VQA) designed to solve combinatorial optimization problems by minimizing a problem-specific cost function. In this paper, we present a QAOA-based decoding framework for LDPC codes by formulating a decoding cost function that incorporates both parity-check constraints and soft channel reliability information. The resulting optimization problem is solved using QAOA to search for low-energy configurations corresponding to valid codewords. We test the proposed method through extensive numerical experiments and compare its performance with BP decoding. The experimental results demonstrate that the QAOA-based decoder achieves a higher probability of correctly recovering the transmitted codeword than BP across multiple experimental settings.

Hydrodynamic reformulations of the Schrödinger equation suggest an interpretation of quantum mechanics in terms of a fluid flowing on configuration space. In the discrete hydrodynamic view, this fluid is not fundamental but emerges from many underlying microscopic fluid components whose collective behavior reproduces quantum phenomena. The most developed realization of this idea is the discrete many interacting worlds (MIW) framework, in which discrete particle-like worlds interact via inter-world forces and quantum probabilities are grounded in direct world counting. But there is also an older, continuous version of MIW. After reviewing the hydrodynamic and MIW formalisms, and emphasizing some of their interpretational advantages over the Everettian Many Worlds and Bohmian approaches, we argue that all discrete hydrodynamic models face a generic structural difficulty, which we call the problem of sparse ontology. Because wavefunctions typically branch under decoherence, the discrete components of the fluid are repeatedly partitioned into sub-ensembles, thereby thinning their density in configuration space and driving the dynamics away from the quantum regime once the components become sufficiently sparse. We conclude that successful hydrodynamic completions of quantum mechanics plausibly require an essentially continuous ontology.

Reconstructing DNA sequences without a reference, known as de novo assembly, is a complex computational task involving the alignment of overlapping fragments. To address this problem, a usual strategy is to map the assembly to a Quadratic Unconstrained Binary Optimization (QUBO) formulation, which can be solved by different quantum algorithms. In this work, we focus on three versions of the Feedback-based Algorithm, a protocol that eliminates classical optimization loops via measurement feedback. We analyze long-read DNA fragments from SARS-CoV-2 and human mitochondrial DNA using standard FALQON, second-order FALQON (SO-FALQON), and time-rescaled FALQON (TR-FALQON). Numerical results show that both variants improve convergence to the ground state and increase success probabilities at reduced circuit depths. These findings indicate that enhanced feedback-driven dynamics are effective for solving combinatorial problems on near-term quantum hardware.

The performance of a given quantum error correction (QEC) code depends upon the noise model that is assumed. Independent Pauli noise, applied after each quantum operation, is a simplistic noise model that is easy to simulate and understand in the context of stabilizer codes. Although such a noise model is artificial, it is equivalent to independent, random, unbiased qubit rotations. What about spatially or temporally correlated qubit rotations? Such a noise model is applicable to global operations (e.g., NMR or ESR), common control sources (e.g., lasers), or slow drift (e.g., charge or magnetic noise) in various qubit technologies. In the worst case, such errors can combine constructively and result in a post-correction failure rate that increases with the number of error correction cycles. However, we show that this worst case does not generally arise unless taking active corrective actions while performing QEC. That is, by employing virtual Pauli frame updates ("passive" error correction) rather than physical corrections ("active" error correction), coherent errors do not compound appreciably. Starting in a random Pauli frame is also advantageous. In fact, through perturbation theory arguments and supporting numerical simulations, we show that the logical qubit performance beyond distance 3 for correlated single-qubit Hamiltonian noise models (i.e., global errant qubit rotations), when employing these "lazy" strategies, essentially matches the performance of Pauli noise model with the same process fidelity (fidelity after one application). In a more general circuit model of noise, correlations may add constructively within syndrome extraction rounds but Pauli frame randomization from passive error correction mitigates this effect across multiple rounds.

We study quantum coherence in top-antitop production at the LHC by comparing Standard Model predictions with CMS data across different kinematic regimes. Theory and experiment are statistically consistent in the near-threshold and boosted central regions, confirming that the spin-density framework captures the dominant helicity-interference structure. The intermediate-mass window shows a noticeable deviation, indicating enhanced sensitivity to radiative QCD effects. This work reinterprets measured spin-correlation data within a quantum-coherence framework, thereby introducing coherence as a complementary and experimentally grounded probe of the Standard Model spin structure and a potentially sensitive diagnostic of new-physics effects in the top-quark sector.

Recently, the coprecession of both the accretion disk and the jet formed following the tidal disruption event associated with the optical transient AT2020afhd, driven by a supermassive black hole of almost ten million solar masses, were independently measured in both the X and radio bands, respectively, showing a periodicity of nearly 20 days over about 300 days. An analytical model of the general relativistic gravitomagnetic Lense-Thirring precession of the effective orbit of a fictitious test particle revolving about a spinning primary can explain the observed precessional features. It yields allowed regions in the system's parameter space which, as far as the hole's dimensionless spin parameter is concerned, are essentially in agreement with those obtained in the literature with general relativistic magnetohydrodynamic simulations. The present analytical approach can be extended to include the precession due to the hole's quadrupole mass moment as well. It breaks the degeneracy in the allowed regions occurring for negative and positive values of the spin parameter when only the Lense-Thirring effect is considered. The best estimate for the hole's mass yields the range $0.185-0.215$ for the dimensionless spin parameter. Using the same strategy with the gravitomagnetic frequency for an extended disk of finite size with a parameterized power-law mass density yields to distinct, generally non-overlapping allowed regions for each value of the power-law index adopted.

Advanced quantum communication protocols require high-visibility quantum interference between photons generated at distant nodes, which places stringent demands on optical synchronization. Conventionally, synchronization of optical wave packets relies on pulsed sources and precise optical path stabilization. An alternative approach employs continuous-wave (CW) photon-pair sources, where temporal indistinguishability is enforced by post-selecting detection events within a coincidence window $τ_w$ shorter than the photon coherence time $T_c$. Despite its conceptual simplicity, the quantitative relation between relevant time scales, achievable interference visibility, and usable multi-photon rates has remained unclear. Here, we develop in detail and experimentally validate a theoretical framework that quantitatively describes time-resolved multi-photon interference in the CW regime. We explicitly incorporate detector timing jitter, photon coherence time, and temporal post-selection. The model is verified using four-photon Hong-Ou-Mandel interference measurements. Based on this validated framework, we determine the coincidence window that maximizes usable four-photon rates for a target visibility. Finally, we compare CW and pulsed SPDC sources under equivalent indistinguishability constraints and show that CW operation can achieve comparable rates while relaxing optical synchronization requirements.

In this paper, we present a Maxwell extension of kinematical Lie algebras by promoting the contraction method underlying the Bacry and Lévy-Leblond cube to a semigroup expansion framework. Within this approach, we show that both non- and ultra-relativistic Maxwell algebras admitting non-degenerate invariant bilinear forms can be systematically obtained from different parent algebras through a unified expansion scheme, leading to a Maxwellian kinematical cube. This construction is further generalized to an infinite hierarchy of kinematical algebras. The expansion method naturally provides the corresponding invariant tensors, allowing for the systematic construction of three-dimensional Chern-Simons gravity theories.

Detection of gravitational waves from compact binaries involving at least one intermediate mass black hole, and component mass ratios in the range $0.1$-$10^{-4}$, are among the primary sources for future space detectors with target strain sensitivities in the deci-Hertz (dHz) band. Tuned to the waveform requirements for analyzing such sources, a hybrid model is obtained by combining waveforms from the post-Newtonian (PN) and black hole perturbation (BHP) theory. Components of the binary are assumed to be nonspinning and on eccentric orbits. This hybrid model is 3PN accurate in terms of results from PN theory and 5PN in results from BHP theory. In terms of eccentricity, corrections through the order $\mathcal{O}(e^{10})$ are included. Further, using number of gravitational wave cycles estimates for a few representative binaries observable in the dHz band, we demonstrate the significance of the mass ratio information from the PN approach, of contributions from the BHP theory at high PN orders, and also of higher order eccentricity corrections. In particular, we find an almost 10-fold increase in number of gravitational wave cycles for a fixed mass ratio of 0.1 and $e_0\sim0.3$ (evaluated at $0.01$Hz) when contributions beyond the leading order in eccentricity are accounted for. We also confirm the requirement to go beyond 5PN order in the circular part of the waveform from BHP theory.

We investigate bipartite entanglement in qubit hypergraph states across an arbitrary fixed bipartition. Using the real equally weighted (REW) representation, the Schmidt rank across the cut can be computed as the real rank of a phase-cleaned cross-cut sign matrix. Whereas graph states admit an exact cut-rank rule, because the cross-cut phase is purely bilinear, hypergraph states typically contain higher-degree cross-cut interactions, for which the cut-rank rule fails. Our approach certifies entanglement by fixing a single computational-basis assignment on a subset of qubits, thereby selecting a submatrix on an active slice. When this restriction removes all higher-degree cross-cut residues, the remaining cross-cut phase becomes bilinear up to cut-local terms. We call the resulting submatrices residual-free bilinear cores and show that they yield an exponential Schmidt-rank lower bound in terms of the $\mathbb{F}_2$-rank of an exposed core matrix. We further give a combinatorial sufficient condition, phrased as a disjoint bridge matching, that guarantees the existence of large full-rank cores for broad families of CCZ-type bridge patterns, and we present a search-and-verify procedure that constructs and certifies such cores directly from the hyperedge description.

The Quantum Internet will allow clients to delegate quantum workloads to remote servers over heterogeneous networks, but choosing the server that minimizes end-to-end execution time is difficult because server processing, feedforward classical communication, and entanglement distribution can overlap in protocol-dependent ways and shift the runtime bottleneck. We propose $T_{\max}$, a lightweight runtime score that sums coarse telemetry from multiple layers to obtain a conservative ranking for online server selection without calibrating weights for each deployment. Using NetSquid discrete-event simulations of a modified parameter-blind VQE (PB-VQE) workload, we evaluate $T_{\max}$ on pools of 10,000 heterogeneous candidates (selecting among up to 100 per decision) across crossover and bottleneck-dominated regimes, including temporal jitter scenarios and jobs with multiple shots. $T_{\max}$ achieves single-digit mean regret normalized by the oracle (below 10%) in both regimes and remains in the single-digit range under classical communication latency jitter for multi-shot jobs, while performance degrades for single-shot jobs under severe jitter. To connect performance to deployment planning, we derive an operating map based on requirements relating distance and entanglement rate requirements to protocol level counts, quantify how simple multiuser contention shifts the crossover, and use Sobol global sensitivity analysis to identify regime-dependent bottlenecks. These findings suggest that simple cross-layer telemetry can enable practical server selection while providing actionable provisioning guidance for emerging Quantum Internet services.

The coupling between internal degrees of freedom of quantum systems and their overall motion in an external gravitational field plays a central role in multiple extensions of Einstein's equivalence principle to quantum physics. While previous models of such effects were predominantly restricted to linearized gravity and often required the motion of quantum particles to follow prescribed world-lines, this letter shows how such phenomena can be understood using generally covariant semiclassical approximations in the framework of quantum field theory in curved space-times. This method provides a unification and generalization of previously established results, but also predicts new effects such as an influence of internal energies on field amplitudes, as well as correction terms to the internal Schrödinger equation that give rise to Berry phases.

Classifying the causal structure of quantum channels is essential for verifying quantum networks and certifying quantum resources. We introduce a characterization-free protocol enabling two isolated players, Alice and Bob, to classify and identify the definite-order strategy adopted by an unknown environment mediating their channels. Without assuming knowledge of their devices or the environment, the players infer the causal order solely from input-output statistics by testing Markovian conditions that we prove are necessary and sufficient for each strategy class. Remarkably, we prove that even with a minimal random channel consisting of two-outcome POVMs and two-state preparations, the protocol retains full performance with probability one. We experimentally demonstrate the protocol on an optical platform, reliably distinguishing between several strategies. Our results provide a strong and robust tool for causal inference in quantum networks.

Quantum many-body devices suffer from imperfections that destabilize dynamics and limit scalability. We show that the dynamical growth of entanglement can intrinsically protect generic quantum dynamics against coherent and perturbative noise. Through rigorous theoretical analysis of general quantum dynamics and numerical simulations of spin chains and fermionic lattices, we prove that entanglement-entropy growth confines the influence of local Hamiltonian perturbations, thereby suppressing errors in dynamical errors. The degree of protection correlates quantitatively with the entanglement entropy of subsystems on which the perturbations act, and applies broadly to both analog quantum simulators and real-time control protocols. This entanglement-induced resilience is conceptually distinct from quantum error correction or dynamical decoupling: it passively leverages native many-body correlations without additional qubits, measurements, or control overhead. Our results reveal a generic mechanism linking entanglement growth to dynamical stability and provide practical guidelines for designing noise-resilient quantum devices.

In this work we investigate the influence of weak acceleration on the confinement-deconfinement phase transition in gluodynamics. Our study is carried out within lattice simulation in the comoving reference frame of accelerated observer which is parameterized by the Rindler coordinates. We find that finite temperature confinement-deconfinement phase transition turns into spatial crossover in the Rindler spacetime. In other words, spatially separated confinement and deconfinement phases can coexist in the Rindler spacetime within certain intervals of temperature and acceleration. We determine the position of the boundary between the phases as a function of temperature for several accelerations and find that it can be described by the Tolman-Ehrenfest law with rather good accuracy although a minor deviation takes place. Moreover, the critical temperature of the system in the weak acceleration regime is found to remain unchanged as that of the standard homogeneous gluodynamics. Our results imply that the spatial confinement-deconfinement transition might take place in the vicinity of the Schwarzschild black hole horizon.

Precise measurements of both the arrival time and carrier frequency of light pulses are essential for time-frequency-encoded quantum technologies. Quantum mechanics, however, imposes fundamental limits on the simultaneous determination of these quantities. In this work, we derive and experimentally verify the quantum uncertainty bounds governing joint time-frequency measurements. We show that when detection is restricted to finite time windows, the problem is naturally described by a quantum rotor, rendering the commonly used Heisenberg uncertainty relation inapplicable. We further propose an optimal detection scheme that saturates these fundamental limits. By sampling the Q-function, we demonstrate the reconstruction of the Wigner function beyond the harmonic oscillator. Using an experimental implementation based on a quantum pulse gate, we confirm that the proposed scheme approaches the ultimate quantum limit for simultaneous time-frequency measurements. These results provide a new framework for joint time-frequency detection with direct implications for precision measurements and quantum information processing.

Performing a quantum measurement yields two different results: a classical outcome drawn from a probability distribution, according to Born's rule, and a quantum outcome corresponding to the post-measurement state. Quantum devices that provide both outcomes can be described through quantum instruments. In a realistic scenario, one can expect that the observer's obtained classical and quantum outcomes are non-ideal: this can be due to experimental limitations, but could also be explained by adversarial interference, that is, a second party that disturbs the device through a concealed measurement to obtain information. The second scenario can be interpreted through quantum compatibility, as it implies that both the observer's instrument and the adversary's measurement can be performed simultaneously. In this work, we show how the noise of the observer's device relates to the amount of information that the adversary can obtain. We study scenarios in which the adversary aims to acquire information on the same basis as the observer's measurement, or on a mutually unbiased basis with respect to the observer's basis. In both cases, we derive necessary and sufficient conditions for the compatibility of a single qubit non-ideal quantum instrument and a noisy meter, from which we obtain the maximum amount of information that the adversary can extract in terms of the noise parameters of the observer's instrument. Finally, we provide the device implementation from the adversary's point of view for the same basis scenario.

Standard optomechanical sensors operating in the low-temperature regime often face fundamental precision limits imposed by vacuum fluctuations. Here, we demonstrate that moving beyond conventional radiation-pressure interactions and exploiting quadratic coupling can surpass these limits, generating intrinsic squeezing and non-Gaussian features in the probe state. We study quantum thermometry and magnetometry in a coupled two-resonator system, focusing on the estimation of a thermal bath temperature and an external magnetic field. The resonators are assumed to be in thermal equilibrium with a common bath, while a weak magnetic field acts on one of the resonators. We perform measurements on a single resonator, which serves as the probe for estimating both parameters. We compute the quantum Fisher information of the probe for two different interaction models between the resonators. Our results show that the counter-rotating terms in the quadratic interaction naturally induce squeezing at intermediate coupling and strong non-Gaussian correlations as the coupling increases further. These effects yield orders-of-magnitude enhancement in sensitivity in the low-temperature and weak-field regimes compared to standard radiation-pressure couplings. Finally, we investigate multiparameter estimation and find that, although the optimal measurements remain compatible, statistical correlations between parameters prevent the simultaneous estimation of temperature and magnetic field from attaining single-parameter precision.

We prove that random 1D Clifford brickwork circuits form (in expectation) good approximate quantum error correction codes in logarithmic depth. Our proof makes use of the statistical mechanics techniques for random circuits developed by Dalzell et al. [PRX Quantum 3, 010333], adapted extensively to our own purpose. We also consider exact error correction, where we give matching upper and lower bounds for the required depth in which random 1D Clifford brickwork circuits become error correcting.

We present and benchmark a type of variational quantum eigensolver (VQE), which we denote the $σ$-VQE. It is designed to target mid-spectrum eigenstates and prepare quantum many-body scar states. The approach leverages the fact that noisy intermediate-scale quantum devices are limited in their ability to generate generic highly-entangled states. This modified VQE pairs a low-depth circuit with an energy-selective objective that explicitly penalizes energy variance around a chosen target energy. The cost function exploits the limited expressibility of the shallow circuit as atypical low-entanglement eigenstates such as scar states are preferentially selected. We validate this mechanism across two complementary families of models that contain many-body scar states: the Shiraishi-Mori embedding approach, and the matrix-product state parent Hamiltonian construction. We define an unbiased estimation scheme for the nonlinear cost function that is compatible with qubit-wise commuting grouping and bitstring reuse. A proof-of-principle demonstration using a small-system instance was carried out on IBM Fez (Heron r2 QPU). These results motivate its use both as a practical "scar detector" and as a state-preparation primitive for initializing nonthermal eigenstate-supported dynamics.

Matter interacts through two long range forces: gravity and electromagnetism. While all matter contributes to the gravitational potential, electromagnetic effects were traditionally expected to cancel in large systems because positive and negative charges balance. Yet astrophysical objects clearly show long range electromagnetic behavior, so the cancellation cannot be perfect. This paper develops a model for stationary aggregation of matter into a star that consistently includes angular momentum and electromagnetic effects. We reformulate the standard polytropic stellar model as a variational problem and extend it to include the kinetic energy of rigid rotation and the electromagnetic interaction energy between oppositely charged baryonic matter. The electromagnetic contribution to the action is taken to be the minimal energy required to generate the stellar magnetic dipole moment. This energy has two parts: the pure electromagnetic contribution, expressible as a surface integral, and the free energy difference between magnetized and unmagnetized matter, obtained by analyzing a degenerate electron gas in a background of cold ions. Differential forms provide a convenient mathematical framework. The resulting model incorporates electromagnetic effects into stellar structure in a way consistent with linearized general relativity. Although the full system forms a complex open boundary problem, exact solutions exist under simplifying assumptions. The phase diagram predicted by the simplified model shows patterns that may motivate further study of the balance between matter, gravitation, and electromagnetism

Coherent spin resonance techniques, such as nuclear and electron spin resonance spectroscopy, have revolutionized non-invasive imaging by providing spectrally resolved information about spin dynamics. Motivated by the recent emergence of electron microscopy methods capable of sensing microwave-excitations, we establish a theoretical framework for Spin Resonance Spectroscopy (SRS) in transmission electron microscopy (TEM). This technique combines microwave pump fields with focused electron probe beams to enable state-selective spin imaging at the atomic scale. Using scattering theory, we model the interaction between free-space electrons and electron spin systems, capturing both elastic and inelastic processes. The strongest effect of the spin system on the free electron is a magnetic phase shift. Our simulations demonstrate that phase shifts from individual electron spins are detectable in both image mode and diffraction mode. In principle, differential measurements under microwave control allow the extraction of local resonance frequencies that are influenced by the surrounding spin environment. By evaluating the Classical Fisher Information (CFI), we identify imaging conditions that maximize the signal-to-noise ratio (SNR), showing how defocus and beam width affect the measurement sensitivity. These findings establish a foundation for integrating SRS with high-resolution TEM, bridging spin spectroscopy and atomic-scale imaging, and enabling new capabilities in quantum spin research and nanoscale materials characterization.

We study a nonrelativistic system made of two quantum particles constrained to move on a line and a spin located at a fixed point of the line. Initially the two particles are in a maximally entangled state and the spin is down. The first particle interacts with the spin while the second particle is free, i.e., it does not interact neither with the first particle nor with the spin. We rigorously prove that there is a correlation between the state of the spin and the state of the second particle. More precisely, we show that, in a suitable scaling limit, if the first particle flips the spin, then the second particle possesses a definite momentum in the direction opposite to the spin.

Manipulating cold atoms in traps is a key tool for numerous realizations of quantum simulators and quantum sensors. They require accurate modeling and characterization of the underlying trapping potentials. We introduce a technique based on the Mach-Zehnder interferometer for in-situ characterization of weakly anharmonic potentials. By simulating the interferometer in an optical dipole trap, we can accurately determine its trap frequency and upper bounds onto anharmonicity magnitudes.

We focus our attention on two-qubit mixed states as initial states, and apply the geometric measure of quantum discord to investigate quantum discord properties in the background of a Schwarzschild black hole under phase damping, phase flip and bit flip channels, respectively. Several analytical complementary relationships based on quantum discords for bipartite subsystems are proposed. For the three channel noises, the behaviors of discords are similar, the accessible discords always degrade as the Hawking acceleration rising, but sudden death never occurs, while the inaccessible discords increase from zero monotonically. Interestingly, in the case of the bit flip channel and phase flip channel, the discords perform symmetrically with the decay probability rising.

We focus our attention on tripartite mixed states as initial states, and apply coherence concurrence to investigate quantum coherence properties in the background of a Schwarzschild black hole under phase damping, phase flip and bit flip channels, respectively. Several analytic complementary relationships based on coherence concurrence for tripartite subsystems are proposed. In the case of the bit flip channel, the behavior of the coherence concurrence is similar to the one of the phase damping channel, the accessible coherence concurrence always degrades as the Hawking acceleration rising, but sudden death never occurs, while the inaccessible coherence increases from zero monotonically. Interestingly, the coherence concurrence is decreasing at first and then increasing as the decay probability rising under phase flip channel. Unlike the case of tripartite pure states, the coherence concurrence of mixed state with X shape is equal to $l_1$ -norm of coherence.

Along with the development of quantum technology, finding useful applications of quantum computers has been a central pursuit. Despite various quantum algorithms have been developed, many of them often require strong input assumptions, which is hardware demanding. In particular, recent advances on dequantization have revealed that the quantum advantage is more of a mere artifact of strong input assumptions. In this work, we propose a variant of these algorithms, leveraging both classical and quantum resources. Provided the classical knowledge (the entries) of the matrix/vector of interest, a classical procedure is used to pre-process this information. Then they are fed into a quantum circuit which is shown to be a block encoding of the matrix of interest. From this block-encoding, we show how to use it to tackle a wide range of problems, including principal component analysis, linear equation solving, Hamiltonian simulation, preparing ground state, and data fitting. We also analyze our protocol, showing that both the classical and quantum procedure can achieve logarithmic complexity in the input dimension, thus implying its potential for near term realization. We then discuss several implications and corollaries of our result. First,, our results suggest there are certain matrices/Hamiltonians where our method can provide exponential improvement compared to the existing ones with respect to the sparsity. Regarding dense linear systems, our method achieves exponential speed-up with respect to the inverse of error tolerance, compared to the best previously known quantum algorithm for dense systems. Last, and most importantly, regarding quantum data fitting, we show how the output of our quantum algorithms can be leveraged to predict unseen data. Thus, it provides an end-to-end application, which has been an open aspect of the previous quantum data fitting algorithm.

We investigate signatures of non-ergodic behavior in the real-time evolution of a one-dimensional Bose-Hubbard model, where the initial state is a doubly occupied density-wave state. We show that the occupation dynamics at strong interactions is dominated by doublon-holon exchange which leads to a domain wall excitation and propagation. The latter manifests as a negated staggered pattern in the density-density correlations. While the single-particle and the pair correlation functions show highly localized correlations that decay rapidly away from the nearest neighbor. We show that the time scale of the domain-wall excitations depends on the inverse of the interaction strength and therefore dictates the slow relaxation dynamics. In the presence of a parabolic trap, the occupation dynamics at the edges become frozen and further suppresses the propagation of correlations. This suppression happens even for trap strengths weaker than the tunneling rate. We also show that the model can be mapped to an antiferromagnetic transverse-field Ising model in the limit of strong interactions and that the correlation-propagation velocity in the original model is well captured by the group velocity of the spin-wave excitation in the effective spin model.

The advancement of diverse generative deep learning models and their variants has furnished substantial insights for investigating quantum many-body problems. In this work, we design two models based on the foundational architecture of generative adversarial networks (GANs) to investigate the ground-state properties and phase transition characteristics of the two-dimensional quantum rotor model (QRM). Within a semi-supervised learning framework, we incorporate multiple layers of transposed convolutions in the generator, enabling the conditional GAN to more efficiently extract low-dimensional encoded information. Analysis of one-dimensional latent variables associated with ground-state samples for different system sizes allows us to pinpoint the location of the critical point. In addition, we introduce dynamically adaptive weighting factors related to the distributional characteristics into the loss function of the deep convolutional GAN, and utilize upsampling techniques to enlarge the generated sample sizes. Comparisons of the optimization processes for mean magnetization and potential energy density across different magnetization regimes of QRM demonstrate that our model can efficiently generate valid ground-state samples, significantly reducing computational time. Our results highlight the promising potential of generative deep learning in quantum phase transition research, especially in critical point identification and the auxiliary generation of simulation data for quantum many-body models.

The experimental detection of quantum entanglement is of great importance in quantum information processing. We present two separability criteria based on the generalized realignment moments. By incorporating additional parameters, these criteria prove to be more flexible and stronger than some of existing ones. Detailed examples are given to demonstrate their availability and feasibility for entanglement detection.

Gravitational wave radiation from periodic orbits in some standard regular black hole spacetimes is studied, primarily using known methods (numerical and analytic). We demonstrate specific differences with the singular Schwarzschild geometry by analysing orbit characteristics, gravitational wave strain profiles, and the corresponding power spectrum density, for different values of the regularising parameter `$g$'. Further, we assess our results vis-a-vis the LISA sensitivity curves and show how our results may be useful while developing templates for detecting regular black holes as viable alternatives to the singular ones. The appendices to our article contain details on errors in our estimates and provide for the first time, some exact analytical expressions on gravitational wave radiation from different types of periodic orbits in Schwarzschild spacetime.

We demonstrate that quantum entanglement between primordial gravitons in dynamically decoupled gravitational sectors can parametrically enhance the tensor power spectrum during inflation. Unlike standard mechanisms relying on classical dynamics or modified actions, this enhancement originates from the reduced density matrix of the observable sector after tracing over a hidden gravitational reservoir. This framework allows for a sizable tensor-to-scalar ratio r > 0.01 consistent with sub-Planckian inflaton excursions, providing a purely quantum mechanical evasion of the Lyth bound. The resulting mixed state leaves a distinctive "quantum birthmark" in the form of oscillatory features in the power spectrum and a characteristic violation of the single-field consistency relation, manifesting as a scale-dependent enhancement of the squeezed-limit bispectrum. Furthermore, we forecast that this entanglement may manifest as a late-time stochastic noise enhancement in gravitational wave interferometers, offering a novel experimental window into the quantum nature of spacetime.

We report the theoretical prediction and measurement of a tune-out wavelength for the ground state of the thulium atom in a linearly polarized optical dipole trap with a wavelength of approximately 576 nm. The measurements were conducted using a combination of trap frequency and RF loss spectroscopy, thus making it possible to separate the scalar and tensor parts of the total polarizability without measurements in the range of negative total polarizability. The calculated tune-out wavelength is consistent with the measured one of $575.646_{-0.014}^{+0.016}$ nm in air. The existence of the zero in the polarizability for the Tm ground state was confirmed by the trap loss experiment, which also made it possible to refine the tune-out wavelength to $575.646_{-0.004}^{+0.004}$. Despite the presence of an imaginary part of the polarizability at some wavelengths, it was experimentally demonstrated that, with a proper choice of the dipole trap polarization, it was possible to achieve Bose-Einstein condensation of thulium atoms in the range from 575.348 to 575.689 nm, covering the tune-out wavelength.

Measurement-Based Quantum Networks (MBQNs) rely on multipartite pre-shared entanglement resources to satisfy entanglement requests. Traditional designs optimize these resources for individual tasks, neglecting that multiple tasks may arrive concurrently and compete for the same entanglement. We introduce compatibility as a design-level metric, capturing whether concurrent tasks can be satisfied by the same entanglement resources. We define a worst-case notion of compatibility where nodes are prevented from coordinating after task arrival and illustrate why tasks may be incompatible. Furthermore, we explore compatibility extensions that account for stochastic arrivals and the capability to supplement the pre-shared entanglement with additional entanglement on-demand, and show that incompatibility differs structurally dependent on the set of concurrent tasks. We argue that compatibility should be used for resource state design, building the foundation for determining which task pairs the network should support with pre-shared entanglement and which require execution-time coordination. Numerical simulations demonstrate this potential, with $(G,1)$-compatibility achieving a 40%-55% gain in simultaneously supported tasks relative to the single-task baseline. By incorporating compatibility as a fundamental design objective, quantum networks can move beyond single-task optimization towards scalable, robust architectures that effectively balance proactive entanglement distribution and supplemental reactive coordination.

In this work we extend the connection between Quantum Error Correction (QEC) and Lattice Gauge Theories (LGTs) by showing that a $\mathbb{Z}_N$ gauge theory with prime dimension $N$ coupled to dynamical matter can be expressed as a qudit stabilizer code. Using the stabilizer formalism we show how to formulate an exact mapping of the encoded $\mathbb{Z}_N$ gauge theory onto two different bosonic models, uncovering a logical duality generated by error correction itself. From this perspective, quantum error correction provides a unifying language to expose dual descriptions of lattice gauge theories. In addition, we generalize earlier $\mathbb{Z}_2$ constructions on qubits to $\mathbb{Z}_N$ on $N$-level qudits and demonstrate how universal fault-tolerant gates can be implemented via state injection between compatible qudit codes.

This work presents a digital quantum simulation of a three-level atomic system interacting with a single-mode electromagnetic field based on the Jaynes-Cummings model, implemented on IBM Quantum superconducting processors. A qutrit is encoded using two physical qubits to represent the atomic states, while an additional qubit encodes the truncated field mode, enabling the realization of effective $Λ$-type atomic dynamics.The continuous-time light-matter interaction is implemented in a digital form by discretizing the evolution using Suzuki-Trotter decomposition. In contrast to an analog realization, the digital simulation replaces the continuous evolution with a sequence of quantum gates whose parameters are explicitly controlled. Phase evolution arising from the interaction Hamiltonian is digitally encoded using calibrated $R_Z$ gates, whose rotation angles are fixed by the physically relevant coupling scale and the chosen Trotter time step.State preparation is achieved using Hadamard and parametrized rotation gates, while the interaction dynamics are implemented through controlled operations. A comparative analysis between first- and second-order Trotter implementations reveals a trade-off between digital accuracy and hardware-induced noise. Overall, the results demonstrate that calibrated gate operations and noise-aware circuit design enable reliable digital simulation of multi-level light-matter interactions on noisy intermediate-scale quantum platforms.

A cluster state is a strongly entangled state, which is a source of measurement-based quantum computation. It is generated by applying controlled-Z (CZ) gates to the state $\left\vert ++\cdots +\right\rangle $. It is protected by the $\mathbb{Z}_{2}^{\text{even}}\times \mathbb{Z}_{2}^{ \text{odd}}$ symmetry. By applying general quantum gates to the state $ \left\vert ++\cdots +\right\rangle $, we systematically obtain a general short-range entangled cluster state. If we use a non-Clifford gate such as the controlled phase-shift gate, we obtain a non-Clifford cluster state. Furthermore, if we use the controlled-controlled Z (CCZ) gate instead of the CZ gate, we obtain non-Clifford cluster states with five-body entanglement. We generalize it to the C$^{N}$Z gate, where $(2N+1)$-body entangled states are generated. The $\mathbb{Z}_{2}^{\text{even}}\times \mathbb{Z}_{2}^{\text{odd}}$ symmetry is non-Clifford for $N\geq 3$. We demonstrate that there emerge $2^{2N}$ fold degenerate ground states for an open chain, indicating the emergence of $N$ free spins at each edge. They can be used as an $N$-qubit input and an $N$-qubit output in measurement-based quantum computation. We also study the non-invertible symmetry, the Kennedy-Tasaki transformation and the string-order parameter in addition to the $\mathbb{Z}_{2}^{\text{even}}\times \mathbb{Z}_{2}^{\text{odd}}$ symmetry in these models.

Designing quantum circuits for ground state preparation is a fundamental task in quantum information science. However, standard Variational Quantum Algorithms (VQAs) are often constrained by limited ansatz expressivity and difficult optimization landscapes. To address these issues, we adopt a geometric perspective, formulating the problem as the minimization of an energy cost function directly over the unitary group. We establish a retraction-based Riemannian optimization framework for this setting, ensuring that all algorithmic procedures are implementable on quantum hardware. Within this framework, we unify existing randomized gradient approaches under a Riemannian Random Subspace Gradient Projection (RRSGP) method. While recent geometric approaches have predominantly focused on such first-order gradient descent techniques, efficient second-order methods remain unexplored. To bridge this gap, we derive explicit expressions for the Riemannian Hessian and show that it can be estimated directly on quantum hardware via parameter-shift rules. Building on this, we propose the Riemannian Random Subspace Newton (RRSN) method, a scalable second-order algorithm that constructs a Newton system from measurement data. Numerical simulations indicate that RRSN achieves quadratic convergence, yielding high-precision ground states in significantly fewer iterations compared to both existing first-order approaches and standard VQA baselines. Ultimately, this work provides a systematic foundation for applying a broader class of efficient Riemannian algorithms to quantum circuit design.

We investigate axial gravitational perturbations and quasinormal modes of regular black holes supported by an Einasto distribution of matter. Halo matter not only removes the central singularity but also modifies the quasinormal spectrum. We show that the resulting quasinormal spectrum deviates systematically from the Schwarzschild case, with shifts in both oscillation frequencies and damping rates that grow with the halo scale parameter and the Einasto index. In near-extremal configurations, the damping rate can be significantly suppressed, leading to long-lived modes. The effects of regularity and environmental factors on the spectrum are found to be substantially larger than the estimated numerical uncertainty, as confirmed independently by high-order WKB calculations with Padé resummation and time-domain integration.

The approximately 11-year jet precession period observed in M87* strongly suggests that the supermassive rotating black hole with a tilted accretion disk, which could provide a powerful constraint for confining the parameters of black hole. In this paper, our aim is to utilize the observations of M87* to preliminarily constrain the parameters of the rotating black hole in Bumblebee gravity by modeling the motion of the tilted accretion disk particle with the spherical orbits. We compute spherical orbits and ISSOs, demonstrating that the conserved quantities energy $\mathcal{E}$, angular momentum $\mathcal{L}$, and Carter constan $\mathcal{K}$ depend on $(r,a,\ell,ζ)$, exhibiting distinct behaviors for prograde and retrograde orbits. For prograde orbits, the ISSO radius $r_{ISSO}$ decreases with spin parameter $a$ and LSB parameter $\ell$ and increases with the tilt angular $ζ$, whereas the opposite trends occur for retrograde orbits. Angular analysis shows that $θ$ oscillates within $(π/2-ζ, π/2+ζ)$, while $φ$ increases approximately linearly, enabling the determination of the oscillation period $T_θ$, azimuthal accumulation $φ(T_θ)/π$, and precession angular velocity $ω_t$. Using the observed jet precession period $T=11.24 \pm 0.47$ years with a fixed tilt $ζ=1.25^\circ$, the warp radius $r/M$ ranges from $(5.73,25.15)$ for prograde and $(6.16,26.46)$ for retrograde orbits, increasing with $a$ or $\ell$. Comparisons with Kerr limits ($r/M=14.12$ prograde, $16.1$ retrograde) suggest that $r/M>16$ may indicate a non-vacuum Bumblebee vector field. Incorporating the EHT shadow $θ_{sh}=42\pm3μ$as further constrains $r/M$ to $(5.82,22.61)$ and $(6.17,24.74)$, with discrepancies of $0.05\sim1.96$.

Oscillators are often employed as a model of radiation fields, which may couple to an atom and play an important role for creating and manipulating nonclassical states in quantum metrology, quantum simulation, and quantum information. Aging transitions in coupled oscillators have been studied extensively in both the classical and quantum contexts. It is well known that the onset of aging transitions can be modulated by the dissipative coupling between oscillators. In this study, we propose an alternative way to modulate the aging transition through coherent couplings between a two-level atom and the oscillators. Our findings reveal that, compared to atom-free systems in both classical and quantum regimes, the atom-oscillator coherent interaction reduces the inactive-to-total oscillator ratio required for aging transitions. Analytical results of the transition for both the classical oscillators and quantum oscillators suggest that the decay rate of the atom and the atom-oscillator coupling strength jointly change the aging transition point. The physics behind the observation is also elucidated in this article. Our research introduces a readily implementable strategy for manipulating aging transitions in more intricate systems, thereby advancing the control and understanding of these critical transitions in quantum technologies.

The geometry of the Newman-Unti-Tamburino (NUT) vacuum solution is characterized as the unique Petrov Type D vacuum metric such that the two double principal null directions form an integrable distribution. We study expanding and twisting non-vacuum Type D metrics in this geometry, with the additional assumption $Φ_{01}=Φ_{12}=0$. We prove that these conditions determine the solutions up to a freedom in $Φ_{11}\pm 3Λ$.

In 2013, Clader, Jacobs, and Sprouse developed a quantum computing algorithm that solves electromagnetic scattering problems exponentially faster than the best known classical algorithm for that problem. We examine this quantum algorithm's potential practical feasibility for modeling a target's radar cross section. Doing so could be important for modeling and predicting radar behavior against emerging targets.

Magic State Distillation is considered to be one of the promising methods for supplying the non-Clifford resources required to achieve universal fault tolerance. Conventional MSD protocols implemented in surface codes often require multiple code blocks and lattice surgery rounds, resulting in substantial qubit overhead, especially at low target error rates. In this work, we present practical magic state distillation factories on Bivariate Bicycle (BB) codes that execute Pauli-measurement-based Clifford circuits inside a single BB code block. We formulate distillation circuit design as a joint optimization of logical qubit mapping, gate scheduling, measurement nativization, and protocol compression via qubit recycling. Based on detailed resource analysis and simulations, our BB factories have space-time volume comparable to that of leading distillation factories while delivering lower target error at a smaller qubit footprint, and are particularly compelling as second-round distillers following magic state cultivations.

The aging transition refers to the shift from an oscillatory state to a globally ceased state due to some forms of deterioration in classical physics. Similar behavior has also been observed in quantum oscillators. Although it has received extensive attention in coupled oscillator systems, it has not yet been studied in coupled qubits. In this manuscript, we explore the aging transition in a network of coupled qubits. Our model describes {numerous} qubits driven by a laser, with both dissipative and coherent qubit-qubit couplings. The ratio of inactive qubits to total qubits and the population in the excited state of the qubits are employed to characterize the aging transition. We find a transition where the population in the excited states suddenly drops when the ratio exceeds a threshold. This behavior is intriguing and contrasts with coupled oscillators, where no sudden drop is observed. Additionally, we demonstrate how the couplings and driving laser influence the threshold. The underlying physics of the sudden drop is elucidated. The region where the aging transition occurs is determined based on stability analysis theory.

Black holes permit matter to cross their event horizon in only one direction. We show that interacting bosons in optical lattices with asymmetric barrier exhibit an analogous phenomenon, creating unidirectional quantum transport without external driving or dissipation. This directionality emerges purely from many-body interactions, which cause asymmetric projection of the initial state onto transport-enabled or transport-forbidden sectors. The resulting dynamics create an effective one-way boundary in Hilbert space, forming a quantum analog of a black-hole event horizon. Our results establish interactions as a fundamentally new route to directional transport, enabling coherent rectification in atomtronic circuits by the use of intrinsic properties of the system only.

Quantum machine learning (QML) sits at the intersection of quantum computing and classical machine learning, offering the prospect of new computational paradigms and advantages for processing complex data. This chapter introduces the fundamentals of QML for readers from both communities, establishing a shared conceptual foundation. We connect the worst-case, adversarial perspective from theoretical computer science with the physical principles of quantum systems, highlighting how superposition, entanglement, and measurement collapse influence learning and robustness. Special attention is given to adversarial robustness, understood as the ability of QML models to resist inputs designed to cause failure. We motivate the study of QML in adversarial settings, outlining distinctions between classical and quantum data and computations when the adversary is a core element. This chapter serves as a starting point to adversarial and robust quantum machine learning in subsequent chapters.

We analytically investigate scalar superradiant scattering in the rotating dyonic Kerr-Sen black hole of Einstein-Maxwell-Dilaton-Axion theory. Starting from the separable Klein-Gordon equation for a massive neutral scalar field, we work in the low-frequency and slow-rotation regime and employ the analytical asymptotic matching (AAM) method to compute the reflection coefficient and the associated superradiant amplification factor. Since an exact global scattering solution is not available in this four-charge geometry, the AAM framework enables a controlled analytic treatment of the near-and far-region dynamics. We provide detailed and systematic derivations of the matching procedure leading to the closed-form amplification formula. The superradiant condition is obtained explicitly and we demonstrate that energy extraction occurs exclusively for co-rotating modes satisfying $Ω< m Ω_H$. We show that the presence of electric and magnetic charges suppresses the amplification relative to the Kerr limit, whereas lighter co-rotating scalar fields broaden the superradiant window and enhance the efficiency of rotational energy extraction.

Sidney Coleman has noted that superluminal particles or observers would be able to go back in time and have no definite trajectory according to subluminal observers, while not violating Lorentz invariance [1]. Recently, Dragan and Ekert have significantly developed similar ideas even further, which lead to formulation of ``quantum principle of relativity'' that intimately links the two theories [2]. However, field theory descriptions of an on-shell tachyon, described by scalar field $φ$ with negative mass squared parameter, lead to violation of basic principles of relativity or quantum mechanics. In this work, we investigate whether purely virtual tachyons can be consistent within the fakeon framework-the only known viable formulation of purely virtual particles. We identify two fatal obstructions. First, Lorentz boosts mix creation and annihilation operators, rendering the canonical commutation relations non-invariant despite formal invariance of the vacuum. Second, the real part of the tachyon Feynman propagator and Wheeler propagator have disjoint support, preventing application of both the fakeon prescription and Wheeler-Feynman absorber mechanism. Interactions with stable Standard Model fields further violate Lorentz invariance and the equivalence principle, and we provide quantitative limit on coupling strength of such scenario. Our analysis excludes possibility of formulating covariant quantum field theory of interacting virtual tachyons.

We present a universal fault-tolerant quantum computing architecture based on identical particle qubits (IPQs), where we find that the first-order IPQ - bath interaction fundamentally differs from the conventional first-order qubit-bath interaction. This key distinction necessitates a redesign of existing strategies to fight decoherence. We propose that the simplest quantum error correction code can be realized directly within the physical qubit, provided that conventional correction and restoration are generalized beyond unitary operations to employ physically implementable reversal operations -- naturally placing logical and physical qubits on equal footing. We further demonstrate that dynamical decoupling (DD) remains effective within this unified framework, and that a decoherence-free subspace (DFS) -- like structure emerges. Unlike previous approximate treatments, our analytically solvable IPQ-Bath model enables rigorous testing of these strategies, with numerical simulations validating their effectiveness.

We revisit Jacobson's thermodynamic derivation of gravitational dynamics in the presence of generalized, non-extensive horizon entropies. Working within a local Rindler-wedge framework, we formulate the Clausius relation as the stationarity condition of a Massieu functional at fixed Unruh temperature, which identifies the entropy slope as the parameter controlling the effective gravitational coupling. For area-type entropies with constant slope, the construction reproduces Einstein's equations with $G_{eff} = 1/(4s_0)$, while curvature-dependent entropy densities supplemented by an internal entropy-production term yield the field equations of $f(R)$ gravity. Motivated by group-entropic considerations and long-range correlations, we model the entropy of horizon cross sections by a power law $S(A) = η(A/4G)^δ$ and analyze its local and global implications. To fix the otherwise arbitrary coarse-graining scale entering the entropy slope, we introduce a Topological Calibration Principle that ties the reference area to intrinsic geometric data through the Gauss-Bonnet theorem. For compact two-dimensional sections, this selects a canonical calibration area and leads to a topology-dependent effective coupling $G_{eff}(χ) \propto |χ|^{1-δ}$ where $χ$ represents the Euler characteristic. Consistency across scales and topologies yields logarithmic bounds on $|1-δ|$, while the associated scale dependence induces a characteristic modulation of the gravitational coupling in cosmology. The framework thus provides a controlled route to confront non-extensive horizon thermodynamics with both theoretical consistency requirements and observational constraints.

Fractional statistics give rise to quantum behaviors that differ fundamentally from those of bosons and fermions. While two-dimensional anyons play a major role in strongly correlated systems and topological quantum computing, the nature of their one-dimensional (1D) counterparts remains the subject of intense debate, with renewed interest fueled by recent experimental progress. Theoretically, 1D anyons are predicted to host exotic many-body phases and quantum phase transitions, yet experimental signatures have remained elusive. Using ultracold atoms in an optical lattice, we prepare two-body ground states of the 1D anyon-Hubbard model by combining Hamiltonian engineering via quasiperiodic drives and adiabatic state manipulation. We uncover the effects of statistical interactions that lead to pseudo-fermionization and to the formation of chiral bound states when particles remain close together. Our results establish a link between lattice and continuum realizations of anyon models, and mark important steps towards the precise control of 1D anyons in both equilibrium and out-of-equilibrium settings.

The zero-point energy of a trapped ultracold Fermi gas at unitarity is investigated in relation to the combined effects of the Heisenberg uncertainty principle and the Pauli principle. This lowest allowed quantum state is a superfluid state which has been studied extensively both experimentally and theoretically. The method used for the current investigation is based on a recent series of papers that proposed microscopic dynamics based on normal modes to describe superfluidity instead of real-space Cooper pairs. This approach yielded excellent agreement with experimental data for multiple properties and allowed the microscopic behavior underlying these results as well as the basis of universal behavior to be analyzed in detail using the group theoretic basis of this general N-body approach. This microscopic picture is now used to illucidate the roles played by the uncertainty principle and the Pauli principle in determining the energy and character of the lowest allowed quantum state including the squeezed character of this superfluid state and the suppression of the Pauli principle.

Quantum sensing has drawn considerable attention as a means to overcome the fundamental limitations in classical sensing. In practice, however, quantum sensing has been strongly constrained by the photon loss, the achievable photon number N in N00N states, and by a finite squeezing level in squeezed states. These limitations are particularly critical to photon-loss-sensitive applications such as LiDAR as well as to general sensing platforms that require large effective N, such as ring-laser gyroscopes. Recently, fundamentally different sensing platforms have been reported to overcome both classical and quantum constraints in a practical regime. One such approach exploits the coherence de Broglie wavelength (CBW) realized in an anti-symmetrically coupled Mach-Zehnder interferometer (MZI) architecture. Here, a pure quantum mechanical analysis of the CBW is presented for a loss-free sensing mechanism of superresolution with enhanced sensitivity. A proof-of-principle demonstration of CBW is also presented.

Limited circuit depth remains a central constraint for quantum optimization in the noisy intermediate-scale quantum (NISQ) regime, where shallow unitary dynamics may fail to sufficiently concentrate probability on low-energy configurations. We introduce Measurement-Guided Initialization (MGI), an iterative strategy that uses measurement outcomes from previous executions to update the initialization of subsequent runs. The method extracts single-qubit marginal probabilities from dominant measurement outcomes and prepares a biased product-state initialization, allowing information obtained during optimization to be reused without introducing classical parameter optimization. We implement this approach in the context of the Feedback-Based Algorithm for Quantum Optimization (FALQON) and evaluate its performance on weighted MaxCut instances. Numerical results show that measurement-guided initialization improves the performance of shallow-depth circuits and enables iterative refinement toward high-quality solutions while preserving the non-variational structure of the algorithm. These results indicate that measurement statistics can be exploited to improve shallow quantum optimization protocols compatible with NISQ devices.

A quantum description of hairy black holes within gravitational decoupling (GD) is developed using coherent graviton states. Classical geometry emerges as the mean-field limit of a finite graviton condensate, while quantum fluctuations introduce a short-distance regulator. We construct a coherent quantum GD hairy black hole metric by Gaussian smearing the gravitational potential, with an effective width that encodes the size of the quantum core. The resulting geometry is free of curvature singularities over appropriate parameter ranges and exhibits modified horizon structures. The effective stress-energy tensor is derived, the energy conditions are analyzed, and the Misner-Sharp mass is computed. We investigate geodesics in the coherent quantum GD hairy black hole geometry, with relevant deviations from the RN case. Photon rings, critical impact parameters, and light deflection exhibit measurable strong-field modifications, allowing observational constraints on deviations from general relativity.

We investigate an inflationary scenario driven by a complex scalar field nonminimally coupled to gravity and subject to a non-symmetric complex potential. The real part of the potential controls the cosmological background and realizes a plateau-type inflation compatible with $α$-attractor $\mathrm{T}$-models, while the imaginary part acts as an effective non-Hermitian deformation encoding dissipative effects. Working in the Jordan frame and imposing ghost-free conditions on the effective Planck mass, we derive the background equations and define a complex equation-of-state parameter whose real part governs the expansion and whose imaginary part quantifies departures from conservative dynamics. Numerical integration shows that the duration of inflation is primarily controlled by the nonminimal coupling $ζ$, whereas the complex asymmetry parameter $Δ\varepsilon$ has a negligible impact on the real background: the real energy density and pressure vary by less than $10^{-5}$ as $Δ\varepsilon$ is scanned over its allowed range. Mapping the two-field dynamics to an effective single-field description in the Einstein frame, we obtain a spectral index $n_s\simeq 0.968-0.971$ and a tensor-to-scalar ratio $r<10^{-3}$, fully consistent with Planck 2018 bounds. We introduce a relevance parameter and show that non-Hermitian effects remain strongly suppressed during slow roll but grow to $\mathcal{O}(1)$ near the end of inflation, triggering an efficient reheating phase without additional fields or {\it ad hoc} friction terms. In this sense, the imaginary sector behaves as an effective $\mathcal{PT}$-symmetric channel for energy transfer, providing a geometrical mechanism for inflation and its exit within a non-Hermitian scalar-tensor framework.

Quantum machine learning (QML) is rapidly transitioning from theoretical promise to practical relevance across data-intensive scientific domains. In this Review, we provide a structured overview of recent advances that bridge foundational quantum learning principles with real-world applications. We survey foundational QML paradigms, including variational quantum algorithms, quantum kernel methods, and neural-network quantum states, with emphasis on their applicability to complex quantum systems. We examine neural-network quantum states as expressive variational models for correlated matter, non-equilibrium dynamics, and open quantum systems, and discuss fundamental challenges associated with training and sampling. Recent advances in quantum-enhanced sampling and diagnostics of learning dynamics, including information-theoretic tools, are reviewed as mechanisms for improving scalability and trainability. The Review further highlights application-driven QML frameworks in drug discovery, cancer biology, and agro-climate modeling, where data complexity and constraints motivate hybrid quantum-classical approaches. We conclude with a discussion of federated quantum machine learning as a route to distributed, privacy-preserving quantum intelligence. Overall, this Review presents a unified perspective on the opportunities and limitations of QML for complex systems.

This brief note is written with a somewhat similar purpose to Earman's 2011 paper on the conceptual problems on the Unruh effect. However, we confine ourselves to Sewell's modular approach to the Unruh effect, which is based on the theorems of Tomita-Takesaki and Bisognano-Wichmann. This approach is rigorous, and has an advantage of being model-independent. However, we will see that a number of conceptual problems remain unsolved on this approach.

The unification of quantum mechanics and general relativity remains among the most profound challenges in fundamental physics. Here we investigate a novel quantum probe of strong-field gravity: the gravitomagnetic Aharonov-Bohm (AB) effect for Cooper pairs propagating in Kerr spacetime. The frame-dragging induced by a rotating black hole (BH) generates an effective vector potential through the off-diagonal metric component $g_{tφ}$, which couples to the macroscopic phase of the superconducting condensate. We derive the gauge-invariant AB phase shift $Δθ= (4πm^* Ma/\hbar)(1/r_2 - 1/r_1)$ for an interferometer with arms at radii $r_1$ and $r_2$, where $m^* = 2m_e$ is the Cooper pair mass and $a$ is the BH spin parameter. Remarkably, the predicted phases reach $|Δθ| \sim 10^{24}$ radians for Sgr~A* and $\sim 10^{27}$ radians for M87*, reflecting the enormous gravitomagnetic flux near supermassive BHs. We analyze the dependence on interferometer geometry, demonstrate that tidal disruption of Cooper pairs is negligible at distances $r \gtrsim 10\,r_s$, and establish connections to the geometric Berry phase. Although direct experimental realization remains beyond current technology due to the vast distances involved, our framework provides quantitative predictions linking quantum coherence to spacetime curvature, complementing recent observations of gravitational AB phases in atom interferometry.

Hilbert space fundamentalism (HSF) states that everything about the physical world is encoded in the Hamiltonian operator and the state vector (as a unit vector, not a wavefunction, which requires additional specification of a configuration space, a position basis, or the position observables). That all structures needed to describe reality, including subsystems, space, fields, emerge from these. I show that HSF can't account for our observations that the physical world changes in time.

An atomic Josephson junction realized with dipolar bosons in a double-well potential can be described by an extended Bose-Hubbard model in which dipolar interactions generate an effective on-site interaction and nearest-neighbor pair tunneling. Using mean-field theory and exact diagonalization, we investigate how this correlated process affects zero-temperature equilibrium and dynamical properties of the system. In equilibrium, we show that pair tunneling induces ground-state parity modulations and significantly reshapes the phase diagram, producing qualitative changes in the quantum phase transitions toward NOON and phase-NOON states, as well as quantitative shifts of the critical points. Out of equilibrium, we demonstrate that it modifies the conditions for macroscopic quantum self-trapping, and assess its impact by comparing mean-field and fully quantum evolution, including the emergence of dynamical quantum phase transitions.

Accurate calibration of coherent optical receivers is essential for reliable performance assessment in coherent communications, precision and quantum sensing, and continuous-variable quantum key distribution (CV-QKD), where the effective detection efficiency directly impacts channel parameter estimation. We present a methodology traceable to the International System of Units (SI) to determine the effective heterodyne detection efficiency of balanced receivers using shot-noise-referenced measurements. The protocol relies on two observables acquired with an electrical spectrum analyzer: the heterodyne beat-note power and the local oscillator shot-noise variance, with explicit treatment of the analyzer's equivalent noise bandwidth (ENBW). The photon flux in the signal path is referenced to SI units via calibrated radiometric standards. We first validate the protocol on a free-space receiver, demonstrating consistency with an independently constructed optical loss chain across a wide range of signal powers and under controlled, calibrated attenuation. Extending the same estimator to a fiber-coupled, polarization-maintaining balanced receiver confirms that the protocol is robust for practical coherent-receiver architectures and intermediate frequencies in the MHz range. These results establish a traceable, uncertainty-bounded framework for real-time receiver calibration, providing a practical route for CV-QKD and other coherent optical systems.

As gravitational-wave catalogs grow, they will become increasingly computationally expensive to analyze in their entirety, especially when inferring astrophysical source populations with high-dimensional, flexible models. Bayesian statistics offers a natural remedy, letting us update our knowledge of physical models as new data arrive, without re-analyzing existing data. However, doing so requires the posterior probability density of model parameters for previous observations, which is typically intractable. Here, we use variational neural posterior estimation to rapidly update the inferred population of binary black holes as data are observed in gravitational-wave detectors. We apply this approach to real and simulated catalogs analyzed with both low- and high-dimensional population models, testing the reliability of three update cadences: with new catalogs of sources, month by month during an observing run, and as each new signal arrives. We investigate the success and failure modes of neural sequential updates, finding that the robustness of updating is sensitive to the information contained in each update and that updating is most effective when performed with larger segments of data. We outline one additional scientific application enabled by Bayesian updating: identification of events that are individually informative about the population. Neural Bayesian updates to astrophysical population models also provide efficient likelihood representations for joint analyses with other data, e.g., standard-siren cosmology, and similar methods can be used to perform Bayesian stochastic background searches.

Minimizing the time required for quantum state preparation is crucial to mitigate decoherence and enable practical quantum algorithms on near-term hardware. In this work, we introduce a technique for quantum state preparation in transmon-qubit systems using optimized electromagnetic pulse sequences rather than discrete quantum gates. By directly targeting quantum correlations instead of specific target states, we identify minimal-time pulse protocols that optimize relevant entanglement resources, such as concurrence and the three-tangle for two and three qubit systems, respectively. For the figures of merit considered, this approach successfully achieves maximal entanglement in each case: Bell, GHZ and W like states. Beyond state preparation, the resource-oriented nature of the approach leads to a reduced effective expressivity of the control scheme, a feature that represents an advantage in algorithmic settings where excessive control freedom is known to hinder performance.

Resonant inelastic X-ray scattering (RIXS) is the workhorse experimental technique for probing the structural degradation of higher-capacity cathode materials. However, the interpretation of experimental spectra is challenging due to the lack of accurate simulations. In this work, we propose a quantum algorithm for simulating the RIXS spectrum of molecular clusters hypothesized to form in Li-excess cathodes. The algorithm uses quantum phase estimation to sample the spectrum from a state encoding the scattering transition amplitudes of the cluster valence excitations. We prepare this state in the quantum computer using a block-encoding of the dipole operator and quantum signal processing to implement the Green's function propagator over intermediate core-excited states. To showcase the algorithm, we use a model cluster proposed in recent experimental works consisting of an oxygen dimer bonded to a manganese atom. Using the PennyLane software platform, we report resource estimation for simulating RIXS spectra for chemically motivated active spaces of increasing sizes. For a classically challenging active space with 20 orbitals, the algorithm requires $2.0 \times 10^{10}$ Toffoli gates and $414$ logical qubits.

Protecting information against decoherence in open quantum systems remains a central challenge for quantum computing. In particular, passive error correction schemes have so far been limited to static memories rather than dynamical qubits. We demonstrate that a driven-dissipative bosonic system can encode a persistently oscillating qubit within a noiseless subsystem, realized explicitly in the Bose-Hubbard dimer (BHD). The strong parity symmetry of the model leads to degenerate stationary states. This symmetry is further broken into non-stationary states in the thermodynamic limit, which exhibit persistent oscillations. As the driving force increases, the Liouvillian spectrum of these states features a phase transition. Above the transition point, the non-stationary state encodes quantum information, preserving it in a noiseless subsystem. In addition to global loss that affects both bosonic modes identically, we further add global dephasing and show that the oscillating qubit is preserved. Finally, in order to gain additional physical insight, we study the effect of phase perturbation to both modes and observe that likewise they are passively protected, returning approximately to their initial configurations. These results establish dissipative time-crystalline dynamics as a mechanism for passive protection of dynamical quantum information, enabling autonomously stabilized oscillating qubits.

We construct a new class of perturbative asymptotically Anti-de Sitter pp-wave spacetimes by performing a long-wavelength expansion of Kaigorodov metrics in arbitrary spacetime dimensions. Holographically, these spacetimes are described by a null fluid hydrodynamic expansion around null states in the conformal field theory, which can be obtained as zero temperature and infinite momentum limits of finite temperature states. Building on this, we explicitly show that special cases of this null fluid/gravity correspondence can be obtained as an ultra-relativistic limit of the usual fluid/gravity correspondence in which the temperature tends to zero while the flow approaches the speed of light. We also extend these results to the context of the blackfold approach in which the corresponding pp-wave spacetimes are asymptotically flat and can be obtained as infinite temperature limits of boosted black branes.

We construct a pseudo-Newtonian potential (PNP) corresponding to a rotating black hole solution in a modified gravity (MGR) framework using a metric-based prescription. The motivation is to enable realistic accretion disk studies in MGR, where full relativistic MHD simulations remain computationally prohibitive. Effective potentials and the underlying Newtonian-like forces are derived for both massive and massless particles in the equatorial plane, relevant for disk dynamics. The reliability of the PNP is tested by comparing key orbital properties -- marginally stable, marginally bound, photon orbits and energies at marginally stable orbit radii -- with exact MGR predictions. The PNP reproduces the marginally stable and photon orbits exactly, while marginally bound orbits and specific energies deviate by less than about 7-10%. The influence of the MGR parameter on particle dynamics and effective potentials is analyzed, revealing non-trivial departures from simple Newtonian intuition. The study demonstrates that the proposed PNP accurately captures essential spatial properties of MGR spacetime and provides an efficient, physically consistent tool for investigating accretion phenomena and strong-gravity astrophysics beyond general relativity.

We investigate the extent to which pseudo-Newtonian gravitational potentials can reproduce classic tests of general relativity without resorting to full general relativistic formalisms. This is useful for the researchers seeking intuitive insight into relativistic gravity. Focusing on the perihelion precession of Mercury, gravitational redshift, and gravitational light bending, we derive analytical expressions for orbital precession and demonstrate that, with suitable physically acceptable parameters, pseudo-Newtonian approaches can accurately reproduce the observed perihelion advance and gravitational redshift. However, we confirm that no single potential consistently captures all relativistic effects. In particular, while certain parameters yield agreement with general relativity for planetary motion and redshift, they fail to reproduce gravitational lensing over a broad range of impact parameters. Our results highlight both the usefulness and limitations of pseudo-Newtonian methods in modeling gravitational phenomena. Although the pseudo-Newtonian approach cannot serve as universal substitutes for general relativity, especially in strong-field regimes, it provides valuable semi-analytical insight and pedagogical simplicity. Our results indicate the usefulness of the pseudo-Newtonian approach to uncover more complicated phenomena involved with strong field gravity in possibly modification to general relativity.

We extend the first order dissipative relativistic hydrodynamics model of Bemfica-Disconzi-Noronha- Kovtun (BDNK) in order to include the charge number current in full first order expansion with out-of-equilibrium contribution proportional to the evolution equation of the ideal fluid. We obtain a fully second order system of partial differential equation (PDE) that can be casted in a fully conservative way. We analyze the hyperbolicity of this model coupled to Einstein field equations using a newly developed technique that allows for hyperbolicity studies without explicit first order reduction. Furthermore, we identify a frame choice where our formulation is causal, stable and with positive entropy generation for a wide range of equations of state (EoS). Our analysis shows that the inclusion of an out-of-equilibrium correction to the charge current, plays an important role in guaranteeing the strong hyperbolicity and, therefore, the well-posedness of the system. If such correction is not applied, an extra frame restriction must be added to the present in the literature in order to obtain a strongly hyperbolic system.

Quantum field theory predicts that dynamical curved spacetimes can spontaneously excite particle pairs from the quantum vacuum, a phenomenon extensively studied in expanding universes and in scenarios involving gravitational collapse. In this article, we explore particle creation driven by radial oscillations of 3+1-dimensional spherically symmetric compact objects, such as neutron stars, using a massless, minimally coupled scalar field as a reference model. We employ a toy model to describe the oscillatory dynamics and its coupling to the field modes, focusing on the resulting effects in the exterior spacetime of the star. The Bogoliubov coefficients relating the in and out vacua are computed non-perturbatively using high-precision numerical methods, without relying on weak-field, small-amplitude or small-velocity expansions. This allows us to determine the full particle spectrum and the total particle number in the strong-field and fully relativistic regime. Our analysis confirms the existence of particle creation in this setting and, crucially, reveals a distinct resonance structure in the spectrum.

Light crossing dark domain walls that source a top form coupled to gauge Chern--Simons terms mixing visible and dark $U(1)$ gauge fields generically converts into dark photons. The effect is entirely localized on the wall and requires no additional ingredients. The conversion rate is a sharp function of the photon frequency in the wall rest frame, vanishing above the ultraviolet cutoff of the top form sector. Partial cloaking may also induce a rotation of the polarization of transmitted light of order $Δ\vartheta \sim 10^{-3}$ radians, modify the cosmic microwave background power spectrum, and violate Etherington's reciprocity relation at low frequencies. These effects can impact cosmological determinations of the Hubble rate.

Fast decoders that achieve strong error suppression are essential for fault-tolerant quantum computation (FTQC) from both practical and theoretical perspectives. The union-find (UF) decoder for the surface code is widely regarded as a promising candidate, offering almost-linear time complexity and favorable empirical error suppression supported by numerical evidence. However, the lack of a rigorous threshold theorem has left open whether the UF decoder can achieve fault tolerance beyond the error models and parameter regimes tested in numerical simulations. Here, we provide a rigorous proof of a finite threshold for the UF decoder on the surface code under the circuit-level local stochastic error model. To this end, we develop a refined error-clustering framework that extends techniques previously used to analyze cellular-automaton and renormalization-group decoders, by showing that error clusters can be separated by substantially larger buffers, thereby enabling analytical control over the behavior of the UF decoder. Using this guarantee, we further prove a quasi-polylogarithmic upper bound on the average runtime of a parallel UF decoder in terms of the code size. We also show that this framework yields a finite threshold for the greedy decoder, a simpler decoder with lower complexity but weaker empirical error suppression. These results provide a solid theoretical foundation for the practical use of UF-based decoders in the development of fault-tolerant quantum computers, while offering a unified framework for studying fault tolerance across these practical decoders.

Extreme Ultraviolet (EUV) lithography is the state-of-the-art process in semiconductor fabrication, yet its spatial resolution is fundamentally limited by the ``blur'' originating from absorption of photons at 92 eV, which induce physical and chemical changes in the photoresist via excited state processes and electron cascades. Accurate modeling of these phenomena requires precise ab initio data for high-energy decay channels, specifically photoabsorption and photoelectron emission. These are computationally difficult for classical methods due to prohibitive scaling in simulating electron dynamics, or due to the inability to resolve the ionization continuum in an efficient manner. In this work, we present quantum simulation algorithms to compute these key observables. First, we introduce a coherent time-domain spectroscopy algorithm optimized to resolve the photoabsorption cross-section at the 92 eV operating frequency. Second, we develop a first-quantized plane-wave simulation to compute the photoelectron kinetic energy spectrum, utilizing real-time dynamics and energy windowing to treat bound and delocalized scattering states on equal footing. Additionally, we provide logical resource estimation for a model photoresist monomer, 4-iodo-2-methylphenol (IMePh), and demonstrate that 92 eV absorption sensitivity can be resolved using roughly $200$ logical qubits and $10^{9}$ total non-Clifford gates per circuit with approximately $10^3$ shots for the smallest instance. The more sophisticated photoemission algorithm that models the continuum explicitly, incurs gate costs of $\geq 10^{13}$ total non-Clifford gates per circuit, $10^4$ shots, and requires a few thousand logical qubits. These results establish high-fidelity quantum simulations as a key component to parameterize the multi-scale macroscopic models required to overcome the electron blur bottleneck in semiconductor miniaturization.

Quantum computers show promise to solve select problems otherwise intractable on classical computers. However, noisy intermediate-scale quantum (NISQ) era devices are currently prone to various sources of error. Quantum error correction (QEC) shows promise as a path towards fault tolerant quantum computing. Surface codes, in particular, have become ubiquitous throughout literature for their efficacy as a quantum error correcting code, and can execute quantum circuits via lattice surgery operations. Lattice surgery also allows for logical qubits to maneuver around the architecture, if there is space for it. Hardware used for near-term demonstrations have both spatially and temporally varying error results in logical qubits. By maneuvering logical qubits around the topology, an average logical error rate (LER) can be enforced. We propose cyclic qubit mappings (CQM), a dynamic remapping technique implemented during compilation to mitigate hardware heterogeneity by expanding and contracting logical qubits. In addition to LER averaging, CQM shows initial promise given it's minimal execution time overhead and effective resource utilization.

We study one-class scoring for resolvable-source detection in confusion-limited LISA time-series data represented as continuous-wavelet-transform (CWT) scalograms. With data generation and preprocessing held fixed, we benchmark geometry-style scoring against likelihood-style latent-density scoring, while also evaluating morphology-augmented and contrastive variants. Geometry-only and geometry+morphology methods provide modest gains over the reconstruction baseline, and contrastive variants do not show stable improvement. Likelihood scoring on AE latents is consistently stronger: across three seeds, latent-only likelihood reaches ROC-AUC $0.8555\pm 0.0181$ and PR-AUC $0.9219 \pm 0.0118$, versus ROC-AUC $0.7663 \pm 0.0450$ and PR-AUC $0.8667 \pm 0.0255$ for AE+manifold. These results indicate that explicit latent density modeling can outperform local off-manifold distance in this confusion-limited regime. We provide seed-based comparisons, unified ROC/PR visual summaries, and reproducible experimental artifacts to support follow-on work in LISA anomaly detection.

We present a symplectic linear-algebraic proof of the Quantum Singleton Bound for stabiliser quantum error-correcting codes together with a Lean4 formalisation of the linear-algebraic argument. The proof is formulated in the language of finite-dimensional symplectic vector spaces modelling Pauli operators and relies on distance-based erasure correctability and the cleaning lemma. Using a dimension-counting argument within the symplectic stabiliser framework, we derive the bound \( k + 2(d - 1) \le n \) for any [[n, k, d]] stabiliser code. This approach isolates the algebraic structure underlying the bound and avoids the heavier analytic machinery that appears in entropy-based proofs, while remaining well-suited to formal verification.

The study of highly charged electronic and muonic hydrogen-like ions, provides an intriguing way to probe the internal structure of their atomic nuclei. In this work, we use nuclear structure calculations to accurately calculate the hyperfine splitting of electronic and muonic hydrogen-like ions, focusing in particular on the incorporation of finite-volume corrections, such as Bohr-Weisskopf and Breit-Rosenthal, due to the penetration of the electron and muon wavefunction into the nuclear electric charge and magnetic dipole densities. These corrections are essential for refining our understanding of the nuclear magnetic dipole and electric quadrupole moments. Our simulations use a Skyrme-Hartree-Fock-BCS model known for its effectiveness in modeling well-deformed nuclei such as ${}^{159}\mathrm{Tb}^{64+}$ and ${}^{165}\mathrm{Ho}^{66+}$, with particular emphasis on ${}^{161,163}\mathrm{Dy}^{65+}$ isotopes. It can also be generalised to multi-electron ions by studying the hyperfine anomaly between two isotopes.

The wavefunction of the universe, as studied in perturbative quantum field theory, is a rational function whose singularities and factorization properties encode a rich underlying combinatorial structure. We define and study a broad generalization of such wavefunctions that can be associated to any lattice. We obtain these wavefunctions as the Laplace transform of a polyhedral fan, the universe fan, whose cones are defined by positivity conditions reflecting a notion of causality in the lattice, and we describe its face lattice. In the matroid case, the universe fan projects to the nested set fan, and the wavefunctions we define recover the matroid amplitudes introduced by Lam as residues. Moreover, in the case relevant for physics, the positivity conditions give a novel way to study the wavefunction, and we show how it is related to the cosmological polytopes of Arkani-Hamed, Benincasa, Postnikov. Finally, we study refinements of the universe fan induced by piecewise linear (tropical) functions. The resulting subdivisions project to refinements of the nested set fan and correspond dually to blow-ups of matroid polytopes, generalizing the cosmohedron polytope.

We propose a generic empirical formula for total inelastic cross-sections for various target nuclei scattered by a proton at different energies, which is applicable over a wide range of energy from $15 ~MeV$ to $1~ TeV$. The proposed model is parameterized based on the fitting of extensively studied experimental cross-section data for the Aluminium and Carbon nucleus targets, considering factorization over high-energy and low-energy regimes. The parameters in high-energy formula are determined by the fitting of the high-energy saturation value of the inelastic scattering cross-section data with mass numbers. The universality of the empirical formula is investigated by comparing the model prediction with the experimental data of inelastic proton-nucleus scattering over a wide range from light elements such as Deuterium to heavy elements such as Uranium. A detailed comparison with the existing models and GEANT4 simulation is also presented.

Recent measurements of the CP asymmetry in the decay $D^0 \rightarrow K_S^0 K_S^0$ by the CMS collaboration, $A_{CP}(K_S^0 K_S^0) = (6.2 \pm 3.0 \pm 0.2 \pm 0.8)\%$, and by LHCb, $A_{CP}(D^0 \to K_S^0 K_S^0) = (1.86 \pm 0.23 \pm 0.11)\%$, suggest possible deviations from Standard Model (SM) expectations, which predict asymmetries below the percent level. This singly Cabibbo-suppressed decay is particularly sensitive to new physics, as the leading amplitudes vanish in the exact U-spin symmetry limit and the process is dominated by W-exchange topologies. We investigate scalar diquark contributions to this decay, comparing color-sextet and color-triplet representations. We find that the color-sextet diquark, characterized by a symmetric color structure $(C_1^{\mathrm{NP}} = C_2^{\mathrm{NP}})$, avoids color suppression and can generate CP asymmetries in the range $0.5\%$--$1.5\%$ for a diquark mass of order 1~TeV. In contrast, the color-triplet contribution is strongly suppressed due to destructive interference from its antisymmetric color structure. We further show that a flavor hierarchy in the sextet couplings, with $λ_{ud} > λ_{us}$, can simultaneously account for the observed deviation from the U-spin sum rule in $D^0 \to K^+ K^-$ and $D^0 \to π^+ π^-$ and the measured CP asymmetry in $D^0 \to K_S^0 K_S^0$. These results identify color-sextet scalar diquarks as viable candidates for explaining enhanced CP violation in charm decays.

The dual-phase xenon time projection chamber (TPC) is a powerful technology to detect rare interactions such as scatters of dark matter particles on nuclei. In particular, the built-in gain of ionization signals in a dual-phase TPC makes it sensitive to events in the few-electron regime, as expected from low-mass dark matter interactions. The pursuit of this low-energy sensitivity through ionization-only signal detection has so far been hindered by excessive electron backgrounds observed across experiments. Much of this background is attributed to the plate-out of $^{222}$Rn decay chain isotopes on the high voltage electrode grid surfaces that span the full cross section of the TPC. This work presents a first-principle model constructed for this background, the predictions of which are consistent with data from the LZ and LUX experiments. We then discuss mitigation strategies of this background in future dual-phase TPCs and the possibility of applying this grid background model to ionization-only dark matter searches.

Currently, the kinetic mass is defined in terms of the pole mass and operators at order $1/m_Q^2$, which are known to N$^3$LO accuracy in $α_s$. At the same time, the Heavy Quark Expansion (HQE) for inclusive semileptonic decays is known up to and including terms of order $1/m_Q^5$. Therefore, it is desirable to extend the definition of the kinetic mass to higher orders in $1/m_Q$. The original kinetic mass is based on the hadron-mass formula in Heavy Quark Effective Theory (HQET). However, the HQE is formulated in terms of matrix elements defined in full QCD to avoid the appearance of non-local matrix elements. To avoid this, we develop a definition of the kinetic mass rooted in full QCD. Starting from the hadron-mass formula derived from the energy-momentum tensor of full QCD, we define a relation between a general mass and the pole mass. Using a simple cut-off scheme, we compute a generalized kinetic mass at one loop to all powers of $1/m_Q$, which reproduces the well-known results for the kinetic mass up to $1/m_Q^2$. Our approach opens the road to a consistent use of the kinetic mass at higher-orders in the heavy quark expansion.

We develop a continuum framework for confining planar QCD ($N_c\to\infty$) by quantizing the Fermi-string (1981) degrees of freedom on the rigid Hodge-dual minimal surface of Geometric QCD I. Worldsheet Majorana fermions provide the algebraic mechanism required by the Makeenko-Migdal loop equations: the Pauli principle cancels non-planar self-intersections, preserving planar factorization. The Liouville instability of random surfaces is avoided since the bulk geometry is rigidly fixed holographically by the boundary loop, without summing over fluctuating metrics. Formulating the theory in momentum loop space integrates out coordinate-space cusp singularities, yielding a finite local limit for quark-loop amplitudes. Changing variables in the phase-space path integral, after fixing reparametrization symmetry (Virasoro constraint), parametrizes the reduced measure by boundary twistors and an induced Jacobian. This yields a twistor-string integral derived directly from the planar QCD amplitude. In the local limit, the theory becomes an analytic twistor string: a boundary sigma model $S^1\to (S^3\times S^3)/S^1$ coupled to a holographically determined Liouville field $ρ(z,\bar z)=\log(|λ(z)||μ(z)|)$ obtained by analytic continuation of boundary twistor data. The meson spectrum is encoded in a 1D functional integral over the boundary twistor trajectory $λ(z),μ(z)$ on $|z|=1$. While evaluated numerically via a reweighted Complex Langevin approach (extracting linear Regge trajectories as a proof of concept), our ultimate finding is analytical. Analyzing complexified effective action monodromies reveals the discrete mass spectrum is exactly governed by Catastrophe Theory and classified by the topological number of twistor poles inside the unit circle. The 1-pole sector exactly yields $m^2 = πσ(n + 1/12)$.

We present a classification of all scalar Lie point symmetries of the Standard Model with one or two real gauge-singlet scalars (SM+S and SM+2S). By analyzing the associated field equations, we identify all realizable and inequivalent Lie point symmetry algebras of these models, distinguishing strict variational, variational (including divergence symmetries), and Euler--Lagrange cases. In addition, we devise efficient algorithms that, for any given numerical instance of the models, determine the Lie point symmetry algebra in each of the three categories by a parameter-based decision procedure using affine reparametrizations and simple parameter tests, thereby avoiding explicit symmetry analysis and the need to derive and solve the determining equations. Finally, we prove several relevant general results, including a characterization of the three disjoint types of Lie point symmetry generators -- strict variational, divergence, and non-variational -- for a broad class of Lagrangians with potentials, including the SM+S and SM+2S.

Hamiltonians ${\cal H}^{a}_k$ of new integrable systems associated with the integer rays $(1,a)$ (commutative subalgebras) of Ding-Iohara-Miki (DIM) algebra in the $N$-body representation are closely related to commuting twisted Cherednik Hamiltonians $\mathfrak{C}_i^{(a)}$, ${\cal H}^{a}_k = \sum_{i=1}^N (\mathfrak{C}_i^{(a)})^k$. Moreover, symmetric combinations of eigenfunctions in the twisted Cherednik system were recently shown to produce the DIM Hamiltonian eigenstates. We explicitly construct these twisted Cherednik eigenfunctions recurrently by action of some (creation and permutation) operations. It resembles of a far-going generalization of Kirillov-Noumi operators, but exact relation remains to be specified.

We develop theoretical diagnostics for the breakdown of mean-field theory, demonstrate how spatial structure and finite interaction ranges enter the effective description, and show how these scales qualitatively modify the renormalization-group flow.

We prove that massless scalar three-point amplitudes are self-dual under Fourier transformation. This implies that the momentum space amplitude can be expressed as the position space amplitude of the same graph with transformed edge-weights (not the dual graph) if external vertices are labeled accordingly. In particular, a massless scalar three-point integral can be expressed as a graphical function. The result follows immediately from a theorem by M. Golz, E. Panzer and the author on parametric representations of position space integrals (2015), but it was only observed by X. Jiang in 2025 in the context of four-dimensional Super-Yang-Mills theory. We generalize Jiang's result and discuss the consequences of the self-duality in the context of graphical functions. In particular, we derive a new identity for graphical functions and a new twist relation for scalar integrals (Feynman periods) in $φ^4$ theory.

We review the original argument of Lehmann et al (LSZ) that relates the flat space S-matrix to the correlation function of field operators and clarify some confusing issues. Next we discuss the origin of the $iε$ prescriptions following Weinberg, but without assuming that the vacuum is free at asymptotic times. Then we discuss the corresponding argument in inflationary cosmology, emphasizing that unitarity violating contour deformations are unnecessary.

In this work, we derive a set of boost-weighted $w$ functionals of the metric, with $w\in\{2,1,0,-1,-2\}$, which transform semi-covariantly under the action of the near-horizon symmetry group. In particular, we demonstrate that the knowledge of the $w=-2$ metric functional and its behaviour under the near-horizon symmetry transformations allows us to derive the expressions and the properties of the remaining boost-weighted functionals via a recursive pattern. A similar recursive pattern also appears when evaluating the action of the near-horizon symmetry group on the evolution equations of these boost-weighted functionals. Again, the knowledge of the evolution equation of the boost $w=-2$ functional and its behaviour under symmetry transformations allows the remaining evolution equations to be determined using symmetry arguments. We also emphasize the role played by the duality symmetry in the characterization of the phase space of a null boundary and in the evaluation of the equation of motion. In conclusion, we derive the sub-leading Noether charges in the Einstein-Cartan-Holst formulation of gravity, showing that the imaginary part of the Weyl scalar $Ψ_2$ appears at the sub-leading order in the Holst charge.

In this work, we propose a cohomological approach to studying perturbative anomalies in quantum mechanics. The Hamiltonian $\hat{H}$ together with the symmetry generator $\hat{S}$ forms a unitary representation of the two-dimensional Abelian Lie algebra $g\cong \mathbb{R}^{2}$ on the Hilbert space $V$. We show that perturbations of such a system are related to the first Chevalley-Eilenberg cohomology group $H^{1}_{CE}(\mathbb{R}^{2},\mathfrak{u}(V))$. In turn, the perturbative anomalies of the symmetry $\hat{S}$ are related to the second cohomology group $H^{2}_{CE}(\mathbb{R}^{2},\mathfrak{u}(V))$.

We derive the equations of motion of relativistic magnetohydrodynamics from the Boltzmann equation using the method of moments. We consider a locally electrically neutral system composed of two particle species with opposite charges, with vanishing dipole moment or spin, so that the fluid has vanishing magnetization and polarization. We find that the dynamics of this fluid changes dramatically in the presence of a magnetic field. The shear stress tensor no longer adheres to a single differential equation; instead, it splits into three non-degenerate components, each evolving according to distinct dynamical equations. Exploring these equations in a Bjorken flow scenario, we find that for large magnetic fields, our theory predicts oscillatory behavior beyond the scope of an Israel-Stewart-like theory.

We investigate in detail a new class (GOOFy) of transformations for bosonic and fermionic fields that leave the Lagrangian density unchanged. The transformations act upon complex scalar fields Φand Φ^\dagger employing generalized charge conjugation (C) transformation in a non-consistent manner, i.e. allowing for Φ\dagger \to (Φ\dagger)^\prime \neq (Φ^\prime)^\dagger. Requiring invariance of the kinetic terms under such transformations specifies the form of (Φ\dagger)^\prime. An analogous strategy is also adopted for fermionic fields. This offers a systematic way to construct new GOOFy-invariant field-theoretical models. It turns out that theories which are invariant with respect these GOOFy transformations satisfy relations among parameters that are found to be RGE-stable up to two and three loop orders, thus constituting fixed-points under running of the RGE. This has been verified for various theories containing different numbers of bosonic and fermionic fields. In particular it has been shown that the Standard Model (SM) can not be a viable electroweak theory if demanding invariance under GOOFy transformations. However, the two-Higgs-Doublet Model (2HDM) may be invariant under GOOFy transformations (Yukawa couplings included), providing an interesting phenomenological example of physics beyond the SM. The most striking aspect of this study is the RGE stability of new relations between model parameters in a wide class of field theories. We also present a set of new relations between 2HDM potential parameters that constitute a fixed point under the running of the RGE up to at least three loop order.

We compare the holographic timelike entanglement entropy with the pseudo-entropy arising from a two-qubit quantum mechanical system. In this model, we consider transitions from an initial thermal state to a final thermal state at fixed temperature under the influence of an external magnetic field. Our findings highlight significant discrepancies between the two quantities, which display markedly different behavior.

For vector/AdS and dS holography we establish the structure of the emergent Hilbert space. This is done through implementation of finite $N$ trace relations on the infinite collective space. For fermionic theories a finite Hilbert space is established, while for bosonic theories a space of freely acting primaries multiplied by a finite set of secondaries emerges. The Hilbert space of states obey finite $N$ cut off bounds, implying finiteness of traces and entropy.

We investigate the impact of a semi-decoupled dark sector with a temperature hierarchy relative to the Standard Model plasma on the electroweak phase transition and its associated gravitational wave signal. Working within a minimal Higgs-portal extension, we allow the dark sector to possess a higher temperature at the electroweak epoch while remaining consistent with cosmological bounds on additional relativistic degrees of freedom. The temperature hierarchy modifies the thermal structure of the effective potential and alters nucleation dynamics without requiring large portal couplings or extreme supercooling. Within the cosmologically allowed window, we find a monotonic enhancement of the gravitational wave amplitude by more than an order of magnitude compared to the standard thermal case, accompanied by a shift of the peak frequency within the millihertz regime. The resulting stochastic background moves substantially closer to the projected sensitivity of future space-based interferometers. Our results demonstrate that hidden-sector temperature hierarchies can leave observable imprints on electroweak-scale phase transitions even in minimal and perturbative frameworks.

Precision spectroscopy has long played a central role in testing the foundations of physics, from the early insights that led to the development of quantum mechanics to the validation of quantum electrodynamics and the determination of fundamental constants. Today, advances in atomic and molecular spectroscopy enable sensitive searches for physics beyond the Standard Model. A broad class of well-motivated extensions predicts new light degrees of freedom with feeble couplings to electrons, muons, and nucleons, giving rise to tiny spin-independent interactions that can be probed at low energies. In this review, we present a unified overview of spectroscopic searches for such interactions. We discuss the effective theoretical framework connecting fundamental interactions to atomic and nuclear observables, survey the key experimental and theoretical strategies, and review the atomic and molecular systems providing the strongest sensitivity. We conclude with updated spectroscopic constraints on representative benchmark models, highlighting the unique and complementary role of precision spectroscopy in exploring new fundamental interactions.

We study the nature of near-threshold eigenstates in systems with the attractive Coulomb plus short-range interactions. Using a model providing the Coulomb-modified effective range expansion, we analyze pole trajectories and the internal structure of near-threshold states characterized by the compositeness. We find that bound state and resonance poles are disconnected in the presence of the attractive Coulomb interaction, in contrast to the repulsive Coulomb force. Near the threshold, bound states become almost purely composite, while the behavior of the compositeness near unityis controlled by the competition between the Coulomb and short-range interactions, characterized by the Bohr radius and the Coulomb effective range.

We construct an explicit local operator realization that reproduces Dirac fermion correlation functions in three spacetime dimensions within an Abelian Chern-Simons framework and use it to examine the conjectured operator and deformation maps of fermion-boson duality directly at the level of correlation functions. We show that the predicted relation between bosonic and fermionic mass deformations, including the relative sign, is realized quantitatively, and that the fermionic U(1) current coincides with the topological gauge current inside correlation functions at the infrared fixed point. These results provide a direct correlator-level verification of key features of Abelian Chern-Simons dualities, going beyond arguments based solely on phase structure, anomaly matching, or large-N considerations.

The quantization of the Seiberg-Witten curve of ${\cal N}=2$ super Yang-Mills theory leads to a deformation of one-dimensional quantum mechanics with unconventional behavior. Most notably, quantum tunneling is suppressed at special points in parameter space. In this paper we examine these deformed models in the case of double-well and cubic potentials, and we find that they have a rich phase structure. In what we call the strong coupling phase, the theory behaves like conventional quantum mechanics, instantons are real, and tunneling is not suppressed. In the weak coupling phase, the instantons responsible for tunneling become complex, and tunneling suppression takes place at the so-called Toda lattice points. At the critical point between the two phases, which corresponds to a monopole point in super Yang-Mills theory, the non-perturbative amplitudes display an anomalous scaling as a function of $\hbar$. This phase structure reflects the physics of the underlying super Yang-Mills theory and can be regarded as a physical manifestation of wall-crossing behavior of the BPS spectrum, which we determine in our problem by using resurgent techniques.

We propose a minimal and predictive framework for asymmetric sterile neutrino dark matter (DM) produced via freeze-in. The Standard Model (SM) is extended by a gauge-singlet Dirac sterile neutrino carrying a conserved dark charge, a real scalar mediator, and an auxiliary singlet fermion. DM is generated through the out-of-equilibrium decay of the mediator, which simultaneously produces a particle{antiparticle asymmetry in the sterile sector controlled by a CP-violating parameter. We show that the observed relic abundance can be naturally reproduced without thermal equilibration with the SM plasma. The resulting non-thermal momentum distribution is colder than a thermal Fermi{Dirac spectrum, ensuring consistency with structure formation constraints. Combining relic density, Lyman-α, Higgs invisible decay, and big bang nucleosynthesis (BBN) bounds, we identify correlated and predictive regions of the parameter space characterized by non-trivial relations among the sterile neutrino mass and the decay parameters. This scenario provides a self-consistent realization of Dirac asymmetric sterile neutrino DM within an asymmetric freeze-in (AFI) framework, offering a constrained and testable alternative to conventional production mechanisms.

We present an updated analysis of the angular distribution for $ψ(3686) \to p\bar{p}$ decay, taking into account transverse beam polarization, to investigate potential sources of forward-backward asymmetry and azimuthal modulation beyond the simple $1+α\cos^2θ$ form. We focus on the interference between the $ψ(3686)$ resonance and the two-photon exchange continuum process, as well as the background from initial-state-final-state radiation interference. A maximum-likelihood fit to the $\cosθ$ distribution of $ψ(3686)\to p\bar{p}$ yields $α= 1.00 \pm 0.03$, consistent with previous results. The fitted contributions from the two-photon interference are small but non-negligible, while the ISR-FSR background is negligible. Our model predicts a significant $\sin(2φ)$ modulation in the azimuthal angle, indicating the influence of transverse beam polarization. These findings motivate future two-dimensional angular analyses to fully disentangle the polarization and interference dynamics in charmonium decays to baryon pairs.

A search for $CP$ violation with an entangled system of $Ξ^-$-$\barΞ^+$ pairs is performed, using $(10,087\pm44)\times10^{6}$ $J/ψ$ events collected with the BESIII experiment. A nine-dimensional helicity amplitude is used to fit $e^+e^-\to J/ψ\toΞ^-\barΞ^+$ and its subsequent decays. The $Ξ^-$ and $\barΞ^+$ decay parameters are determined with higher precision compared to the best results reported so far. Furthermore, the strong phase difference, $(δ_P-δ_S)=(0.3\pm1.2\pm0.2)\times10^{-2}~\text{rad}$, and the weak phase difference, $(ξ_P-ξ_S)=(-0.2\pm1.2\pm0.1)\times10^{-2}~\text{rad}$, are directly determined. These are the most precise measurements to date. The $CP$ asymmetry observables $A_{CP}^Ξ=(-7.8\pm4.8\pm0.8)\times10^{-3}$ and $Δφ_{CP}^Ξ=(0.6\pm5.1\pm0.2)\times10^{-3}~\text{rad}$ are determined, which are consistent with $CP$ conservation. In addition, independent measurements of the $Λ$ decay parameter and $CP$ asymmetry $A^Λ_{CP}=(-2.9\pm4.3\pm0.7)\times10^{-3}$ are also obtained, which are in agreement with the previous measurements, but with much improved precision.

A search for the pair production of heavy spin-1/2 or spin-3/2 resonances (t$^*$) in proton-proton collisions at $\sqrt{s}$ = 13 TeV is presented. Data collected with the CMS detector at the CERN LHC from 2016 to 2018 corresponding to an integrated luminosity of 138 fb$^{-1}$ are used. The analysis targets benchmark signal scenarios where one t$^*$ decays into a top quark (t) and a photon ($γ$), and the other into a t quark and a gluon (g), i.e., pp $\to$ t$^*\bar{\mathrm{t}}^*$ $\to$ tt$γ$g. All-hadronic final states from the t pair decay chain are selected using jet substructure techniques. The signal is probed as a function of the t$^*$ candidate mass, which is reconstructed using the photon and a top quark candidate jet. No significant deviation from the background-only hypothesis is found. Observed (expected) upper limits on the signal cross section at 95% confidence level are set, excluding masses of spin-1/2 t$^*$ particles below 930 (930) GeV and spin-3/2 t$^*$ particles below 1330 (1390) GeV. This analysis marks the first search for heavy resonances in the $\mathrm{t\bar{t}}γ$g channel. Exploiting the high-energy photon to reduce the backgrounds, this search achieves sensitivity competitive with pp $\to$ t$^*\mathrm{\bar{t}}^*$ $\to$ $\mathrm{t\bar{t}}γ$g searches for spin-1/2 t$^*$ despite the small expected t$^*$ $\to$ t$γ$ branching fraction.

Motivated by the successful application of Heavy Quark Effective Field Theory in describing decays from heavy to light mesons, this work explores its applicability to the semileptonic decays of charmed mesons. So in this paper we investigate the $D_s^+\to η^{(\prime)} \ell^+ ν_\ell$ transition form factors using the light-cone sum rules approach within the framework of heavy-quark effective field theory. To address the large uncertainties arsing from the $η^{(\prime)}$-meson twist-3 distribution amplitudes, we employ the right-handed chiral correlation function. By applying the converging simplified series expansion method, we extrapolate the form factors to the entire physical $q^2$-region. Our analysis yields the branching fractions precise predictions for semi-leptonic decays $D_s^+\to η^{(\prime)}\ell^+ν_\ell$ with : $\mathcal{B}(D_s^+\toη\ell^+ν_\ell)=2.300^{+0.230}_{-0.227}\%$ ($\ell = e$) and $2.249_{-0.206}^{+0.209}\%$ ($\ell = μ$); $\mathcal{B}(D_s^+\toη^\prime \ell^+ν_\ell)=0.861^{+0.095}_{-0.093}\%$ ($\ell = e$) and $0.821^{+0.082}_{-0.080}\%$ ($\ell = μ$). The derived lepton flavor universality ratios $R^η_{μ,e}=0.977^{+0.008}_{-0.006}$ and $R^{η^{\prime}}_{μ,e} = 0.953^{+0.011}_{-0.009}$ are consistent with lasted BESIII experimental measurements. Additionally, the forward-backward asymmetry parameters $\langle \mathcal{A}^η_{\rm FB}\rangle=-0.034^{+0.003}_{-0.003}$ and $\langle \mathcal{A}^{η^\prime}_{\rm FB}\rangle=-0.073^{+0.007}_{-0.008}$ suggest that no significant violation of lepton flavor universality in this decay process.

Quintessential inflation provides a unified description of the early and late accelerated phases of the Universe, linking the inflationary epoch to the present-day dark energy-dominated era through a single scalar degree of freedom. In this work, we explore the implications of this unification for cosmological constraints on the sum of neutrino masses. Focusing on the $α$-attractor scenario, we implement the model in a modified version of the Boltzmann solver CLASS to compute the relevant cosmological observables and perform a Bayesian parameter estimation analysis using data from the cosmic microwave background (CMB), baryon acoustic oscillations (BAOs), and Type Ia supernovae. The model naturally breaks the degeneracy between the dark energy equation of state and the total neutrino mass, yielding tight upper bounds of $\sum m_ν< 0.067$ eV for flat spatial geometry and $\sum m_ν< 0.116$ eV when curvature is included. We also provide forecasts for future probes, showing that the Simons Observatory, LiteBIRD, and Euclid configurations may reduce the uncertainty on $\sum m_ν$ by $\approx 9\%$, while the precision on the quintessential parameter $α_{QI}$ is improved by $\approx 72\%$. These results highlight the importance of consistently accounting for neutrino mass when assessing the viability of extensions to the standard cosmological model.

A study of two-particle differential number ($B$) and transverse momentum ($P_{2}^\mathrm{CD}$) balance functions in photon-proton ($γp$) and proton-proton (pp) collisions at $\sqrt{s}=$ 5.36 TeV is presented. The analysis focuses on inclusive charged hadrons within the pseudorapidity coverage $|η|<2.4$ and the transverse momentum interval $0.3 < p_\mathrm{T} < 3.0$ GeV and examines their correlations in terms of relative pseudorapidity ($Δη$) and relative azimuthal angle ($Δφ$). The correlation functions are evaluated for same- and opposite-sign pairs, and their combinations are used to extract charge-dependent (CD) and charge-independent (CI) components. The evolution of the near-side peak of the CD correlations is investigated in terms of $Δη$ and $Δφ$ as a function of charged-particle multiplicity ($N_\mathrm{ch}$) for $γp$ collisions and compared to pp collisions at a similar multiplicity range. A clear multiplicity dependence of the balance function width is obtained. The width is found systematically lower in $γp$ events than in pp collisions. This study provides valuable information on particle correlations and production mechanisms in low-$N_\mathrm{ch}$ regimes for upcoming measurements in small systems.

Known holographic dictionaries, especially AdS/CFT, rely on symmetry matching between the bulk and the boundary. We take a step toward a holographic dictionary with no symmetry requirement and without assuming the geometry being asymptotically AdS. Starting from any interacting Majorana generalized free field on a $(0+1)$d boundary and its two-point function data, we derive a concise analytic formula for the dual $(1+1)$d bulk geometry, borrowing techniques from unitary matrix integral and inverse scattering. Using this formula, we compute the near-horizon curvature, give conditions for positive versus negative curvature, and identify simple boundary models with de Sitter or anti-de Sitter near-horizon duals. We also study the large-$q$ SYK model, finding an unusual temperature dependence of the near-horizon curvature, related to the discrepancy between physical temperature and the ``fake disk'' temperature. We also construct, directly from boundary operators, approximate algebras generated by null translations and boost that become exact at the bifurcate horizon.

We consider correlation functions of topologically twisted, $\mathcal{N}=2$ supersymmetric Yang-Mills theory with gauge group ${\rm SU}(2)$ and $N_f\leq 3$ massive hypermultiplets in the fundamental representation. For a smooth, compact, oriented four-manifold $X$ with $b_2^+>1$, the correlation functions are expressed in terms of a finite set of universal functions. The mass dependence of these functions encodes intersection numbers of the moduli space of instantons. We determine closed expressions for the universal functions by combining techniques of the Seiberg-Witten geometry, $u$-plane integral and the blowup formula. If $X$ is specialised to a complex algebraic surface $S$, the correlation functions can be identified with generating functions of Segre invariants for moduli spaces of sheaves on $S$. We verify that our results agree with the results by Göttsche and Kool for these generating functions.

We show that the near horizon of a generic non-extremal black hole (BH) can be understood in terms of a Carrollian geometry with two null directions, also called a String-Carroll (SC) geometry. The base space of this fibre-bundle structure is a sphere (or a plane for a black brane) and the fibre is the two-dimensional Rindler spacetime. We launch a detailed study of probes in this geometry. We study particle geodesics and scalar fields. The first part of the paper constructs geodesics and probe scalar fields directly in the SC geometry. We then look at a wide class of examples, including the Schwarzschild BH and the Kerr BH in asymptotically flat spacetimes, the BTZ BH and the black brane in AdS spacetimes, as well as Lifshitz black holes and construct the explicit maps to the SC geometry to obtain results specific to each case. These results are reproduced by considering the probe particles and fields in the original BH background and taking the near-horizon limit of the solutions. Our encyclopedia of examples establishes the notion of SC geometries as near-horizon geometries of non-extremal black objects, paving the way for a detailed, intricate future analysis of the quantum aspects of this geometry.

We unify one of the most widely studied frameworks to explain the hierarchical structure of the flavor sector in the Standard Model, the Froggatt-Nielsen mechanism, with cosmic inflation. We propose that the complex scalar field, the so-called flavon, which breaks the Froggatt-Nielsen $U(1)$ symmetry and generates the Yukawa couplings of the Standard Model, to also drive inflation, which we dub as Inflavon. After inflation ends, the decay of the inflavon reheats the Universe, establishing a novel link between early Universe cosmology and flavor physics. As concrete examples, we present realizations where the inflavon potential is described by an $α$-attractor potential. We then compute the resulting CMB observables, specifically the spectral index ($n_s$), the tensor-to-scalar ratio ($r$), and the amplitude of scalar perturbations ($A_s$) as functions of the underlying Froggatt-Nielsen model parameters. We identify the parameter space in Froggatt-Nielsen models involving the scale of Flavor symmetry breaking $Λ_{\rm FN}$ and FN charges which is ruled out by Planck and ACT data, as well as the region that could be probed by next-generation CMB experiments like CMB-S4, SO and LiteBIRD. We also discuss inflavon as dark matter and its isocurvature constraints.

Laboratory searches for ultralight axion dark matter (DM) have traditionally assumed the terrestrial density of axions is equal to the average density of DM in the solar system. However, quadratic couplings to matter introduce a non-trivial field profile near the Earth. In this work, we present the first dedicated experimental implementation of this environment-aware axion DM wind search framework. Leveraging the extreme sensitivity of a K--Rb--$^{21}$Ne comagnetometer to pseudo-magnetic fields induced by axion DM, we analyzed our data in the context of the massively enhanced local gradient of axions due to interactions with matter, though no signal candidates were found. Consequently, we have set the most stringent limits on axion-neutron derivative interactions for masses $m_a \in [0.041, ~28.9]~\rm feV$, improving from previous experiments that ignore terrestrial matter effects by as much as three orders of magnitude for certain masses. Our work highlights the necessity of accounting for environmental modifications in precision frontier experiments and demonstrates how geophysical variations can be harnessed to act as a natural amplifier for DM possibly enabling future detection in parts of the parameter space that were previously beyond reach.

Cosmic-ray scatterings on the cosmic neutrino background induce a flux of gamma-rays and X-rays from boosted meson decays and charged lepton processes. Here we present the first estimate of this flux and its cumulative cosmological contribution. Confronting expectations with Fermi-LAT diffuse gamma-ray data, we find a limit on the cosmic neutrino background overdensity at the level of $η\lesssim 2 \times 10^{4}$ for a lightest neutrino mass of $m_ν\gtrsim0.1$ eV, orders of magnitude stronger than current direct laboratory probes, and comparable to constraints on the cosmic neutrino background obtained with IceCube. We further show that X-ray synchrotron emission from cascade electron-positron pairs in intergalactic magnetic fields provides a complementary, albeit weaker, constraint. We discuss how anisotropic signatures and future gamma-ray data from CTA could further improve bounds on the relic neutrino overdensity, approaching in sensitivity the $Λ$CDM expected density.

We present a theoretical study of the azimuthal decorrelation $δφ$ and transverse momentum imbalance $q_T$ in dijet production at the LHC, offering intriguing insights into the dynamics of quantum chromodynamics. We define the jet axes using the recoil-free winner-take-all (WTA) recombination scheme. For the azimuthal decorrelation $δφ$, this axis choice eliminates non-global logarithms (NGLs) entirely. For the transverse momentum imbalance $q_T$, NGLs emerge specifically in the small jet radius limit ($R \ll 1$). In this regime, the WTA scheme simplifies the theoretical framework by restricting jet radius logarithms to the soft sector. We derive factorization formulae for both observables within soft-collinear effective theory. To address the small-$R$ NGLs in the $q_T$ distribution, we refactorize the soft function into global soft, collinear-soft, and ultra-collinear-soft modes. We perform the resummation of global large logarithms $\ln(δφ)$ and $\ln(q_T/Q)$ up to next-to-next-to-leading logarithmic accuracy. For the $q_T$ distribution, this is combined with a leading-logarithmic resummation of the non-global $\ln R$ terms. We match our predictions to leading fixed-order $O(α_s^3)$ calculations. We also numerically investigate the structure of the first subleading power corrections. Comparisons with PYTHIA8 simulations demonstrate that the observables we consider are robust against non-perturbative multi-parton interactions and hadronization effects.

Recent studies of Galactic surveys, such as Gaia, have revealed that the Milky Way's gravitational potential comes from a matter distribution that is triaxial and rotated with respect to the Galactic center-Sun axis. This, in turn, could mean that the dark matter halo also shares these properties. In this work, by fitting to the Fermi-LAT gamma-ray observations, we test the compatibility of the morphology of the Galactic Center Excess (GCE) from dark matter annihilation with a triaxial dark matter halo. In particular, we consider both untilted triaxial halos and halos whose principal axes are tilted with respect to the Galactic disk. In our fits of the Fermi-LAT data, by testing over a large library of galactic diffuse emission models, we quantify how the halo triaxiality and tilt affect the line-of-sight-integrated annihilation signal and, consequently, the preferred GCE spatial templates. We find that the GCE spectrum and inner cuspiness are robust against variations in the triaxiality and tilt of the dark matter halo. However, in terms of its overall morphology, the GCE in the gamma-ray data can discriminate between choices for the dark matter halo's triaxiality and tilt. Finally, we find that the GCE is more compatible with originating from a triaxial and tilted halo of dark matter than originating from a triaxial and tilted halo of stars, a result important for understanding the GCE's origin.

State-of-the-art calculations motivated by dark matter exhibit severe violation of partial-wave unitarity in the non-relativistic regime in radiative bound-state-formation processes. It has been recently shown, in a model-independent fashion, that unitarity is restored by the proper resummation of the inelastic contributions to the self-energy of the incoming state. In this work, we first derive Kramers-like formulae for individual partial waves, demonstrating that existing calculations of bound-state formation severely violate unitarity. We then discuss how unitarity is restored through the resummation of the absorptive contributions to the incoming-state self-energy, generated by bound-state formation processes, taking into account their analytic structure in the complex momentum plane. Our results can be generalized in a variety of theories and employed in phenomenological studies, such as dark-matter freeze-out, indirect detection and self-interactions.

We study warm Higgs portal dark matter (DM) in the framework of freeze-in at stronger coupling. This scenario assumes that the Standard Model thermal bath temperature has always been relatively low, which suppresses dark matter production. As a result, a significant DM-Higgs coupling is allowed, enabling warm dark matter detection via Higgs decay at colliders. We find that the Lyman-α bound on the DM mass is particularly strong, excluding masses below 50-100 keV, depending on further details. The shape of the DM momentum distribution is highly non-thermal, with low momenta being effectively cut off, and not captured by the common αβγ-parametrization.

The phenomenology of a charged Higgs present in a model with two Higgs SU(2) doublets and a $Z_3$ flavor symmetry is analyzed. It is shown that it is possible to generate an enhancement of its flavor changing coupling to c and b quarks and also to reproduce the ATLAS excess associated to the process $H^\pm \to bc$ for a charge Higgs mass of $130$~GeV. Furthermore, by considering the possibility of a search at the future LHeC, the analysis suggests viability for its detection.

First-order phase transitions produce gravitational waves and primordial black holes. They always occur in field theories where symmetries are radiatively broken and masses are correspondingly generated. These theories predict a period of supercooling: phase transitions become effective at temperatures much smaller than the symmetry-breaking scale. This paper reviews a model-independent approach to study phase transitions in this scenario, which can be adopted if supercooling is strong enough. Perturbative methods can be used to determine the effective action and such model-independent approach allows us to obtain ready-to-use formulas that can be applied to any specific model of this sort.

We investigate a minimal extension of the Leptogenesis framework that simultaneously explains the observed baryon asymmetry and dark matter (DM) abundance through the decay of a heavy Majorana neutrino. In this scenario, CP violation arises from complex Yukawa couplings, enabling the generation of asymmetries in both the Standard Model (SM) and DM sectors. We explore two regimes: (i) wash-in, where an initial dark asymmetry is transferred to SM leptons by $2 \leftrightarrow 2$ scattering processes; and (ii) co-genesis, featuring a hierarchical coupling structure that allows enhanced CP violation while supporting a low-scale seesaw mechanism at order $\mathcal{O}(2)$ TeV. This setup not only links light neutrino masses to the Majorana mass term but also suggests that lepton-number violation may occur at experimentally accessible energy scales. In the co-genesis scenario, we show spin-independent cross sections for DM heavier than 10 GeV that can be tested in current direct detection experiments and motivate the exploration of cross sections inside the neutrino fog for lighter DM masses, establishing asymmetric leptogenesis as a predictive benchmark framework for direct-detection experiments and identifying a new hierarchical-coupling regime enabling TeV-scale leptogenesis.

Given the non-negligible interplay between parton distribution functions (PDFs) at large x and potential New Physics (NP) effects in the high-energy tails of hadron collider observables, a central question is which PDFs can be reliably employed in beyond-the-Standard-Model (BSM) analyses. In this work, we examine the fine balance between using PDF sets with small uncertainties in the large-x region -- crucial for maximising BSM sensitivity -- and adopting conservative PDF fits that exclude high-energy data potentially contaminated by unaccounted NP contributions. We systematically assess a range of conservative PDF fitting strategies designed to mitigate such biases and provide a recommendation for the class of PDFs best suited for robust BSM searches. In addition, we investigate the alternative approach of performing simultaneous fits of Standard Model Effective Field Theory (SMEFT) Wilson coefficients and PDFs, thereby consistently accounting for their mutual correlations. Starting from a toy model to illustrate the underlying mechanisms, we then analyse two realistic NP scenarios: one modifying high-mass Drell-Yan production and another affecting the high-invariant-mass tail of top-quark pair production. Both cases are representative of measurements that will be probed with high precision during the High-Luminosity phase of the LHC.

We introduce a novel multi-parameter rating methodology for objective comparison of theoretical models with experimental data in heavy-ion collisions, addressing fundamental limitations of the traditional global $χ^2$/ndf criterion. The methodology divides phase space into seven physically motivated kinematic regions reflecting distinct production mechanisms from thermal freeze-out to perturbative QCD. Each region receives a quality score $Q_i \in [10, 1000]$ via logarithmic transformation of local $χ^2_i/ν_i$ statistics, ranging from 10 (very poor agreement) to 1000 (perfect agreement). A composite rating $R$ aggregates weighted average, geometric mean, and minimum scores with dispersion penalty, preventing compensation of poor agreement in one region by artificial success elsewhere. Demonstrated on real ALICE and representative LHCb data for $K^0_S$ mesons and $Λ$ hyperons at $\sqrt{s_{NN}} = 5.02$ TeV in p-Pb collisions, the methodology reveals that no universal model exists: PYTHIA8 excels for mesons ($R=879$) via nuclear PDFs, while models with coalescence are potentially superior for baryons. The baryon anomaly peak serves as critical discriminator, with highest quality scores $Q_i \sim 950\text{--}1000$ in zones $3\text{--}4$. The methodology is transparent, reproducible, fully algorithmic, and ready for integration into standard analysis frameworks.

This document presents a synthesis of the theory contributions and discussions from the 2nd EIC-France Workshop, held at IJCLab (Orsay) on 1-3 December 2025. The workshop brought together members of the French hadron-physics community to review recent theoretical developments relevant to the future Electron-Ion Collider (EIC) and to coordinate national efforts in preparation for its early physics program. The report first summarizes the collider's initial running conditions and luminosity performance, as outlined in the EIC Early Science Matrix. It then provides concise overviews of the theoretical presentations on inclusive, semi-inclusive, exclusive, heavy-flavor, and small-x physics. Based on these discussions, two measurements emerged as especially well suited for early EIC operation and strongly aligned with areas of established French expertise: inclusive diffraction and inclusive quarkonium production. These channels offer clean signatures, robust theoretical interpretability, and direct sensitivity to fundamental QCD phenomena such as gluon saturation, heavy-quark dynamics, and the small-x structure of hadrons and nuclei. In addition, the workshop identified longer-term physics opportunities that will benefit from the full capabilities of the EIC after its ramp-up phase. These include accessing the three-dimensional structure of the pion through the Sullivan process and a broader program of exclusive three-body final states, both of which represent high-impact avenues for exploring hadronic structure and non-perturbative QCD. Together, the elements summarized in this report provide a coherent overview of the strategic priorities and scientific ambitions shaping the French community's contribution to the EIC physics program.

We prove that the 4-dimensional Galois representations associated with a certain Calabi-Yau threefold are reducible, with 2-dimensional composition factors coming from specific modular forms of weights 2 and 4, both level 14. This was essentially conjectured by Meyer and Verrill. It was revisited in its present form by Candelas, de la Ossa, Elmi and van Straten, whose computations of Euler factors in a whole pencil of Calabi-Yau threefolds highlighted this fibre as one of three overwhelmingly likely to be ``rank-2 attractors''.

Modern measurements in flavour physics rely on accurate simulations of signal and background processes, provided by a wide range of general-purpose and specialised Monte-Carlo event generators. Due to the inclusion of a larger amount of specialised decays of heavy hadrons, EvtGen is often the tool of choice for many scenarios. We investigate the phase-space sampling algorithm of EvtGen and demonstrate that it generates unphysical features in kinematic distributions of semileptonic $B$ decays involving resonances, originating from neglected phase-space factors. We provide a short-term solution to correct the affected simulated samples through reweighting of the hadronic invariant mass distribution.

RR Lyrae stars are a common, dependable Population II distance indicator, and provide an independent tracer of early star formation. Here, we utilize archival HST/ACS and JWST/NIRCam observations of the nearby dwarf star-forming galaxy WLM to study RR Lyrae in JWST filters. We independently identify RR Lyrae in HST and JWST imaging in order to evaluate JWST's efficacy at characterizing RR Lyrae in the near-IR. We use an MCMC template-fitting technique to obtain periods, amplitudes, and mean magnitudes from the RR Lyrae time-series data. The spatially overlapping HST and JWST observations allow us to directly compare the same sources observed with the instruments, and calibrate the NIRCam F090W and F150W RR Lyrae period-Wesenheit-metallicity (PWZ) relation to the Gaia-consistent HST PWZ. We additionally assess the epoch-to-epoch consistency of NIRCam photometry, and find evidence of burn-in. We conclude that the zero-point offset is negligible compared to the uncertainties from the template fitting. We conduct an MCMC fit of the PWZ with both HST and JWST data. Our results are three-fold. First, we find that we can reliably identify RR Lyrae in NIRCam data, but light-curve template fitting proves difficult on short-baseline observations. Second, the HST PWZ fit yields a distance modulus to WLM of $μ= 24.85\pm0.05$ ($0.93\pm0.02$ Mpc). This is closer than previous measurements, primarily attributed to consistency with the Gaia scale. Lastly, although the JWST PWZ fit has large uncertainties and a poorly-constrained slope, it represents a first-of-its-kind PWZ calibration in NIRCam filters.

Scattering of stars by interstellar clouds or massive clumps increases the stellar velocity dispersion and promotes a radial disk profile that is exponential. Here we show that such scattering reaches a steady-state distribution function of stellar eccentricity, after which eccentricity increases and decreases occur at equal rates. The implication is that clump/cloud scattering recircularizes eccentric stellar orbits, keeping the stellar velocity dispersion in a limited range. This re-circularization regulates disk heating and maintains kinematic coherence, contributing to the longevity of disk structures. The eccentricity distribution function and the presence of recircularizing cloud-star interactions are independent of cloud mass but the timescale to reach equilibrium decreases with increasing mass. The calculations are made in the simplest possible disk system to highlight the effects of scattering without contamination from spiral waves, star formation, and other processes. The calculations also reveal a bifurcation in the disk evolutions whereby in a minority of cases temporary asymmetries in the clump spatial distribution drive the disks to an end state of increased velocity dispersion and orbital eccentricity corresponding to early type disks. Overall the models emphasize an important physical process that can make and maintain an exponential stellar disk in all galaxies with a cloudy interstellar medium.

Cosmic rays (CRs) are known to play a key role in many astrophysical environments: they can modify shock dynamics, influence the thermochemistry and the ionization of the interstellar medium, regulate galaxy mass content by driving galactic winds, and be released by jets from active galactic nuclei. They also serve as important observational tracers through $γ$-ray emission, radio synchrotron, and secondary particle production. Since CR particles follow power-law distributions in momentum space spanning many decades in energy, and because diffusion and radiative losses further shape these spectra, it is crucial to model spectrally resolved CRs in numerical simulations and to assess the impact of this modeling on gas dynamics and observational signatures. We present a consistent multi-group spectral method in momentum space for CR protons called RAMSES-MCR in the adaptive mesh refinement code RAMSES, based on the two-moment formalism that evolves both CR energy and number density in momentum space, together with their associated flux. The modeled CR processes include advection, anisotropic/isotropic diffusion, streaming instability, Coulomb and hadronic losses, adiabatic changes, and feedback onto the gas. We also show that the method can be naturally extended to CR electrons (e.g. including synchrotron losses) and generalized to multiple CR species. The implementation is validated against a suite of standard multi-dimensional tests. We finally apply RAMSES-MCR to the three-dimensional expansion of a supernova remnant including CRs with anisotropic diffusion and energy losses, and demonstrate how CR energy redistributes in a momentum-dependent manner and modifies the gas momentum during the snowplough phase.

Carbon disulfide (CS$_2$) is one of the sulfur-bearing species expected to be present in the interstellar medium (ISM). In this study, we investigated the surface reactions of solid CS$_2$ with hydrogen (H) atoms on amorphous solid water (ASW) using laboratory experiments supported by computational calculations. Our results show that CS$_2$ reacts with H atoms through quantum tunneling in the initial step, followed by successive H addition reactions, with or without activation barriers, on icy surfaces. These processes lead to the formation of several sulfur-bearing species, including hydrogen sulfide (H$_2$S), methyl mercaptan (CH$_3$SH), and small amounts of dithioformic acid (HC(S)SH) and methanedithiol (CH$_2$(SH)$_2$). The observed reactivity of CS$_2$ with H atoms provides a plausible explanation for the non-detection of CS$_2$ in interstellar ices. Furthermore, the efficient hydrogenation of the complex molecules derived from CS$_2$, namely HC(S)SH and CH$_2$(SH)$_2$, suggests that these species could be easily undergone with H atoms to produce other S-bearing species under ISM conditions.

Recent JWST observations of HII regions in the Central Molecular Zone have shown a highly filamentary morphology distinct from HII regions in other parts of the galaxy. We present magnetohydrodynamic (MHD) simulations of strongly magnetized (plasma-$β$ << 1) HII region evolution that investigate and describe the formation of these ionized gas filaments. HII region evolution has been simulated in a 30 pc$^3$ box, in distinct models with pre-placed overdensities in the ambient medium and overdensities that have been generated with driven turbulence. We find that when these overdensities are seeded in the ambient medium before the birth of the ionizing source, the photoionized plasma stripped off of these dense blobs is funneled into long filaments along the magnetic field lines. The length and emission measure of these ionized gas filaments are similar to the filaments observed in the Central Molecular Zone. Given that these filaments are effectively magnetically-confined flows of photoionized gas, their density and curvature are influenced by the density of the blob and the geometry of the configuration.

Upcoming galaxy surveys with the SKA Observatory will detect neutral hydrogen (HI) across unprecedented volumes, and their scientific return will crucially depend on predictive models for HI observables. In this work, we present a framework to simulate the neutral hydrogen 21cm emission line in such large-scale HI galaxy surveys. This framework is developed as a modular layer that builds on semi-analytical models. In particular we use as bases the Galaxy Evolution and Assembly (GAEA) and L-Galaxies semi-analytical models, coupled to merger trees from the Millennium Simulation suite. We validate our framework against local Universe observations, demonstrating consistency with velocity functions, and generalised Tully-Fisher relations. Predictions based on GAEA and L-Galaxies exhibit mutual consistency despite the distinct underlying physical prescriptions. We construct mock galaxy catalogues that incorporate forward-modelled selection functions, inclination effects, and redshift broadening, reproducing the statistical distributions of HI-selected galaxies in the ALFALFA survey. Finally, we present redshift distribution forecasts for future SKA Observatory HI galaxy surveys. This framework offers a flexible tool for interpreting upcoming large-scale radio surveys and studying HI line observables as cosmological probes.

We present a catalog of 486 Ba stars identified in the GALAH DR4 survey using high-resolution spectra and precise abundance measurements. The sample was selected based on s-process enrichment criteria involving the abundances of Ba and La relative to Eu, and further refined using the signed-distance method, resulting in the largest sample of Ba stars to date, including five newly identified Ba dwarfs. Using astrometric and kinematic parameters from Gaia DR3 and StarHorse, we derived the Galactic velocity components for 367 Ba stars and found that most belong to the thin or thick disk, while 18 exhibit halo-like kinematics. Chemical abundance analysis suggests that most Ba stars are of in situ origin, whereas two stars (4077588766331013248 and 6692980582560946304) display signatures of accreted populations. The E-Lz diagram further shows that star 4077588766331013248 lies within the region of substructure ED-8. The observed decline of [Ba/Fe] and [La/Fe] with increasing metallicity implies that s-process elements originate from nucleosynthetic sites distinct from those of iron-peak elements, while the decreasing [hs/ls] ratio toward higher metallicity indicates higher neutron capture efficiency at lower metallicity. Both giant and dwarf Ba stars-except for 6053735173729807872-are confirmed binaries hosting white dwarf companions. Moreover, we estimated the masses of the former asymptotic giant branch (AGB) companions to these Ba stars based on the good agreement between the FRUITY AGB model predictions and their observed abundance patterns.

I examine high-quality CO and SiO molecular maps of the core-collapse supernova (CCSN) remnant SN 1987A from the literature and find that the molecular gas exhibits a bipolar structure, correlated with the visible bipolar morphology (termed the keyhole) and the bipolar morphology of the iron emission map. The keyhole has a morphology similar to that of many jet-shaped pairs of bubbles in cooling-flow clusters of galaxies and planetary nebulae. Therefore, the findings of this study, which make the bipolar structure of SN 1987A robust, strengthen the claim that a pair of energetic jets shaped the keyhole and its surroundings. According to the jittering-jets explosion mechanism (JJEM), this pair of jets was the most energetic of several pairs that exploded SN 1987A. This study adds to the accumulating evidence that the JJEM is the primary explosion mechanism of CCSNe, responsible for the majority, or even all, CCSNe.

The existence of massive, compact galaxies (Little Red Dots, LRDs) at $z \sim 2$ challenges early structure formation models, suggesting rapid stellar and black hole (BH) assembly. While LRDs are efficient environments for BH growth, many show no X-ray evidence of strong AGN emission. We utilize a subsample of X-ray non-detected LRDs to test the compatibility of collision-based BH formation scenarios and constrain physical parameters like metallicity and column density.Our results indicate LRDs are ideal birthplaces for massive BHs, particularly given a mass-radius relation $R_{gal} \propto M_{gal}^{0.6}$. Collision-based models suggest seed masses larger than those in the local Universe, consistent with high-redshift BH mass-radius relations. We modeled BH seed formation and X-ray emission (0.3-7 keV) against observed upper limits. We find that mass-radius exponents $> 0.55$ favor the collision-based scenario; however, consistency with stacked X-ray analysis requires specific accretion and obscuration parameters. Constant or increasing SFR scenarios with high Eddington ratios are feasible but necessitate larger column densities or higher metal enrichment. Alternatively, moderate sub-Eddington accretion reconciles massive seeds with observed masses and X-ray weakness. We conclude that even if LRDs began as starbursts, they should eventually evolve into AGNs.

In this study, we analysed about $\sim$13 years of publicly available data from MAXI and Swift/BAT to examine the long-term source evolution of 42 transient low-mass X-ray binaries. The sample consists of 11 confirmed black hole X-ray binaries (BHXBs), 10 black hole candidates (BHC), and 21 neutron star X-ray binaries (NSXBs). Outbursts and flaring activities studied over 13 years show that 19/21 NSXBs spend significantly longer time in the hard state (observations for which hardness ratio is $\geq$ 0.2) while 15/21 BHXB+XRC spend substantially longer time in the soft state (observations for which hardness ratio is $<$ 0.2). The frequency distribution of the hardness ratio clearly shows two distinct distributions for BHXBs and NSXBs, with their peaks separated: NSXBs prefer harder values, while BHXBs prefer softer values of hardness. Our model-independent analysis for 42 transient sources shows that statistically NSXBs do not prefer to show a canonical high soft state as observed in BHXBs. Additionally, the probability distribution of the duration of the 2-20 keV X-ray outburst is observed to peak at a significantly longer duration ($>$100 days) for BHXBs than for NSXBs (15-60 days). Our analysis shows that among candidate sources, Swift J1728.9-3613, MAXI J1535-571, MAXI J1659-152, EXO 1846-031 show a `q' diagram in the HID and prefer to align with the HID frequency distribution of BHXBs that show `q' diagram, MAXI J1305-704 and MAXI J1836-194 align with frequency distribution of black hole sources without `q' diagram while MAXI J1848-015 shows the HID distribution similar to NSXBs, indicating a neutron star accretor. Therefore, a long-term statistical study of MAXI and Swift/BAT X-ray outbursts from a large sample of transient sources may be used to distinguish BHXB from NSXB.

It has been shown that the turbulence acceleration in AGN coronae can account for 1-10 TeV neutrinos from some AGNs, such as the Seyfert galaxy NGC 1068. Based on this, there are attempts to explain the diffuse neutrinos observed by IceCube with the accumulated contribution from a population of AGNs, but it is found that the maximum neutrino energy is less than tens of TeV and, as a result, additional source classes are needed to explain the high-energy component above this energy. Recently, motivated by the detection of >100 TeV neutrinos from the Seyfert galaxy NGC 7469, it was shown that the turbulence acceleration in the corona can explain >100 TeV neutrinos given a larger magnetization parameter ($σ\sim 1$) in the corona, which leads to a larger maximum proton energy and a hard proton spectrum. In this paper, we extend this assumption to the population of AGNs and study whether the population of AGNs with a wide range of magnetization can explain the entire diffuse neutrinos. We find that AGN coronae could account for the diffuse neutrinos up to PeV energies if a significant fraction of AGNs have magnetization as large as $σ\sim 1-10$. This conclusion is insensitive to the shape of the magnetization parameter distribution as long as the range of the magnetization parameter is sufficiently wide and the distribution is flat towards high magnetization. Interestingly, this model can also explain the peak of the diffuse neutrino spectrum at 30 TeV.

Gjergo & Kroupa (2025) have proposed a new foreground CMBR component produced by dust associated with the progenitor clouds at $z=15-20$ that led to the formation of massive early-type galaxies, calculating that a minimum of 1.4% and a maximum of 100% of the whole CMBR radiation would be produced by this mechanism. Here, I check how much dust emission is compatible with the spectrum of the CMBR monopole flux measurements within the corresponding errors. COBE-FIRAS monopole is fitted to different models with different dust emissivity spectral indexes ($β$) and dispersion of redshifts of the progenitor clouds ($σ_d$). Within the realistic values of $β\ge 0.5$, $σ_d\ge 0.8$, the contamination of CMBR flux by high $z$ dust should be $<1.3$% (95% CL).

We present the first clear detection of ionized Fe-K emission and absorption components in the nearby radio galaxy Centaurus A, revealed by the high-resolution XRISM/Resolve detector. In the 6.5-6.9 keV band, XRISM reveals multiple Fe XXV and Fe XXVI emission components. One is a broad (with a width of sigma = 3000 km/s) and redshifted (+3400 km/s) component, originating at D = 0.02 pc from the central black hole. The other two components are narrow (with a width of sigma = 500 km/s) and exhibit redshifted and blueshifted velocities (+2600 km/s and -1500 km/s), originating from more distant regions (D = 0.1 pc). The photo-ionized model explains the broader component, while the two narrower components can be explained by either photo-ionization or collisional ionization. One interpretation is that the broader component is an outflow at ~10^2 R_S (R_S; Schwarzschild radius) and the narrow component is a shock-heated plasma close to the torus, with a possible connection to the JWST-discovered outflow outside the torus. Two blueshifted absorption lines are detected at ~7.1 keV (~10^4 km/s) and ~10.6 keV (~10^5 km/s). The line significance of the 10.6 keV line is above 98%. The absorption line components might be attributed to the broad emission component. These results demonstrate the high potential of XRISM/Resolve to characterize ionized emission and absorption features in the Fe-K band. Our findings establish a new benchmark in the study of circumnuclear environments in low-luminosity radio galaxies, thereby contributing to a broader understanding of AGN unification.

Magnetically arrested accretion disks (MADs) in black hole jet launching simulations are very successful in modelling low-luminosity active galactic nuclei (AGN) like M87*. The Fishbone-Moncrief torus is well established for this purpose in numerical astrophysics. The extent of the magnetic vector potential inside the torus that we coin the filling factor has not been studied before in the case of MAD simulations. We employ five 3D general relativistic magneto-hydrodynamics (GRMHD) simulations initialized with large-scale tori, that are immersed in weak, poloidal magnetic fields. To study the impact of the spatial extent of the initial magnetic field, hence the magnetic energy content in the torus, we scale it with the filling factor w.r.t. the poloidal geometric area of the mass density distribution. A common choice of the filling factor is complimented and investigated in terms of altered energetics and angular momentum transport. Further, we investigate the polarized, radiative imprints of synchrotron emission on M87 at 86 GHz, comparing them with VLBI observations. Our simulations show that elevated filling factors significantly increase the electromagnetic energy contributions and outward angular momentum transport in the jet, due to the initially increased magnetic energy-content in the torus. High magnetic fillings exhibit increased linear polarization fractions, agreeing with the observed 15$\%$ in M87*. We find the jet morphology more prone to disk-vertical flux tubes generated by MAD events. We show, that GRMHD simulations bracket the jet width measurements at the jet base in M87*. Increased magnetic filling of the torus produces jets that are noticeably brighter downstream compared to our reference models, hence, we find high fillings well suited for extended GRMHD jet models of other low-luminosity AGN, as well.

We present a strong lensing analysis of the double source plane lens J0946+1006 (colloquially "Jackpot" lens) to measure the inner dark matter density profile, the stellar-to-halo mass ratio, and the stellar initial mass function normalisation using a two component stellar plus dark matter mass model. The stellar mass follows a multi-Gaussian expansion light model with a free global mass-to-light ratio and an allowed radial $M/L$ gradient, while the dark matter is described by an elliptical generalised NFW halo. The double-source-plane geometry provides additional leverage against the mass-sheet transformation and helps constrain the radial mass profile. Despite allowing both a radial stellar $M/L$ gradient and a generalised NFW halo, the data prefer the canonical picture: an approximately constant stellar mass-to-light ratio with a Salpeter-like IMF normalisation, and a dark matter halo consistent with NFW. We infer $M_{\star} = 4.4^{+0.25}_{-0.39}\times 10^{11}\,M_{\odot}$ and an inner halo slope $γ_{\rm in}^{\rm halo} = 1.04^{+0.10}_{-0.14}$. The halo mass is $M_{200}^{\rm halo} = 1.11^{+0.37}_{-0.32}\times 10^{13}\,M_{\odot}$, implying $\log_{10}(M_{200}/M_{\star})=1.41^{+0.13}_{-0.14}$. At fixed halo mass, the inferred stellar mass lies $\sim0.1$ dex above typical literature stellar halo mass relations at similar redshift, which is comparable to the intrinsic scatter of these relations. We expect this approach to provide a practical template for future dark matter studies with the large double-source-plane lens samples from Euclid.

Nearby pulsars within $\sim1\,{\rm kpc}$ are considered to be possible sources of 10-500 GeV cosmic-ray positron excess measured by PAMELA and AMS-02. A TeV halo around Geminga is detected by HAWC, and the measurements of its surface brightness profile indicate a slow particle diffusion surrounding the source. This result challenges the pulsar interpretation of the positron excess. The observations at GeV energies provide direct information on the electron/positron density in the GeV nebula, which can offer more direct constraints on the origin of the positron excess. Two previous works have performed analyses on the GeV emission of the pulsar halo, but focused on the energy band above 8 GeV. In this work, we use a longer dataset from the Fermi Large Area Telescope (LAT) to re-analyze the GeV halo emission of Geminga, extending the analysis to cover the energy range of 1-1000 GeV. We find that the analysis in this wider energy range results in a low significance of the halo emission. This can be attributed to the Galactic interstellar emission model being unable to perfectly fit the background over this broader energy range, and due to the low measured halo flux at $<$ 10 GeV energies leading to a mismatch between the observation and model expectation. We also derive the spectral energy distribution of the tentative halo emission, which shows a very hard spectrum in the 1-10 GeV range.

The Fermi Large Area Telescope (LAT) provides advantages for radio pulsar searches by enabling efficient target selection. We can confidently point radio telescopes to the positions of Fermi unidentified gamma-ray sources that have a high probability of hosting a pulsar. As part of Transients and Pulsars with MeerKAT (TRAPUM), we conducted a survey of Fermi-LAT sources using the Ultra High Frequency (UHF; 544-1088 MHz) receiver of the MeerKAT radio telescope. We observed 79 sources that were identified as pulsar-like candidates using a random forest technique from the Fermi-LAT Fourth Source Catalogue. We observed each target for 10 minutes at two separate epochs. As a result, we discovered nine new millisecond pulsars (MSPs) and six slow pulsars. Based on the radio discoveries, we also searched for gamma-ray pulsations, confirming that seven of the newly discovered MSPs are associated with Fermi-LAT sources, and performed joint radio and gamma-ray pulsar timing. Companion mass estimates and evidence of radio eclipses indicate that among the nine MSPs there are three black widows and three redbacks. Lastly, we compared the discovered pulsars in the MeerKAT UHF survey against the previous Fermi sources TRAPUM survey at L band, concluding the superiority of UHF observations in sensitivity to fainter pulsars and in detection rate than L band for finding new gamma-ray MSPs.

The global topology of the Universe can affect the long-range gravitational forces through the boundary conditions. To study non-trivial topologies in detail, simulations that natively adopt such geometries are required. Cosmological $N$-body simulations typically evolve matter in a periodic cubic box. While numerically convenient, this imposes a non-trivial 3-torus topology that affects long-range gravitational forces, potentially biasing large-scale statistics. We introduce a compactified simulation framework that is only periodic along a single axis, while having infinite topology with isotropic boundary conditions towards the perpendicular directions, i.e. an $\mathrm{S}^1\times\mathbb{R}^2$ ("slab") topology. This new simulation geometry is ideal for simulating systems with cylindrical symmetries like filaments or certain anisotropic cosmological models. We compactify comoving space via inverse stereographic projection along the radial direction of a periodic cylinder, and evolve particles with Newtonian dynamics. A smoothly varying spatial and mass resolution with radius suppresses edge artefacts at the free outer boundary. Our implementation in the StePS (STEreographically Projected cosmological Simulations) framework uses a direct $\mathcal{O}(N^2)$ force calculation that maps efficiently to GPUs, and Octree $\mathcal{O}(N \log N)$ force calculation that can be used on large CPU-clusters. The cylindrical domain's topology enables fully self-consistent simulations in the $\mathrm{S}^1\times\mathbb{R}^2$ manifold and mitigates periodic-image artefacts for targets whose symmetries are mismatched to a cubic box. The main trade-off is radially varying resolution with distinct systematics and analysis requirements. We demonstrate the accuracy of the new simulation method in a standard $Λ$CDM cosmological simulation.

We present a multi-wavelength study of the nearby starburst galaxy M 82 by combining high-resolution Far-ultraviolet (FUV) imaging from the Ultra-Violet Imaging Telescope (UVIT) onboard AstroSat and archival Chandra X-ray observations. Using FUV flux measurements, we estimate a spatially-resolved star formation rate (SFR) across several star-forming clumps within a radius of $\sim$3.6 kpc, finding a total SFR of 0.022 M$_{\odot}$ yr$^{-1}$. The H$_α$ recombination line flux yields an SFR of $\sim$0.010 M$_{\odot}$ yr$^{-1}$, while the infrared-based SFR derived from 24 $μ\mathrm{m}$ emission is significantly higher at 16 - 18 M$_{\odot}$ yr$^{-1}$, suggesting that a substantial fraction of star formation in M 82 is heavily dust-obscured. Morphological comparison of FUV, H$_α$, mid-infrared, and soft X-ray emission reveals a strong spatial correlation, tracing multi-phase outflows along the galaxy's minor axis. X-ray spectral analysis using a three-temperature $\texttt{VAPEC}$ model shows enhanced abundances of Ne, Mg, Si, and S, consistent with enrichment from Type-II supernovae. These results demonstrate the importance of combining UV, optical, IR, and X-ray observations to probe both obscured and unobscured star formation, the metal enrichment, and the outflow-driven evolution of starburst galaxies.

We present spectroscopic and photometric observations and analysis of SN 2024abvb, a peculiar transitional Type Ibn/Icn supernova located at an unusually large projected distance from its host galaxy (21.5 kpc). SN 2024abvb displays an extended rise time in the $g$- and $o$-bands (10.1 and 10.6 days respectively), followed by a linear decline in all photometric bands. Comparisons with other supernova subclasses show that the photometric and spectroscopic evolution of SN 2024abvb are distinct from Type~Ibn and Type Icn events, with a higher peak $r$-band luminosity and lower blackbody temperatures. Spectra reveal an initial blue continuum and narrow P-Cygni profiles, with C {\sc ii} $λ$5890 dominating in emission, persisting at late phases, and showing a rapid decline in the expansion velocity. Weak He {\sc i} $λ$5876 features are tentatively detected at early times. Analysis of progenitor scenarios rules out thermonuclear origins based on incompatible light curve shapes and spectral signatures. A rare massive star progenitor appears unlikely given the low local star formation rate. The most plausible origin is an ultra-stripped supernova scenario involving a binary system; this best explains the observed separation from the host, the low circumstellar material mass, the fast photometric evolution and the low nickel production, although a discrepancy in model versus observed ejecta mass remains. These results reinforce the classification of SN 2024abvb as a distinctive Type Ibn/Icn event and highlight the diversity of progenitor channels for interacting supernovae.

Be/X-ray binaries are the most numerous group of high-mass X-ray binaries. Their long-term optical and infrared variability reflects the evolution of the circumstellar disk around the luminous companion. This variability manifests photometrically as an excess of flux that increases with wavelength and spectroscopically as line emission. The disk is also expected to generate linear polarization. We present a systematic study of the optical long-term polarimetric variability of Be/X-ray binaries on data collected over 10 years. Our aim is to characterize the polarimetric properties of these systems and to probe the structure of their circumstellar disks. We have been monitoring Be/X-ray binaries visible from the Northern hemisphere with the RoboPol polarimeter. Optical polarimetric variability is a common trait in Be/X-ray binaries. The variability can be attributed to the Be star's circumstellar disk. Our polarization analysis confirms previous claims based on spectroscopic data that the circumstellar disks in BeXBs are, on average, smaller and denser than those in Be stars in non-binary systems. Our data also confirms the presence of highly distorted disks before giant X-ray outbursts, although this result is still affected by the lack of simultaneous and well-sampled observations during major X-ray outbursts.

Context. Compact jets from active galactic nuclei (AGN) are commonly assumed to be in equipartition between particle and magnetic-field energy densities at the regions where the radio emission dominates at centimetre wavelengths. This assumption has significant implications for both jet physics and the accuracy of VLBI-based astrometry and geodesy. Aims. We tested the validity of the energy equipartition hypothesis in AGN cores at centimetre wavelengths by analysing the blazar 1803+784 using simultaneous broadband full-polarization observations with the VLBI Global Observing System (VGOS). Methods. We present VGOS observations of the blazar 1803+784 covering the 3-11 GHz frequency range. The data were processed using a dedicated calibration pipeline, followed by model fitting and multi-frequency imaging analysis. We measured the frequency-dependent core shift and mapped the spectral index and rotation measure (RM) across the source. Results. We find a core-shift power-law index of $k_r = 0.73^{+0.12}_{-0.19}$, significantly deviating from the expected equipartition value of $k_r = 1$. This indicates that either the equipartition condition or the conical jet geometry, or both, are not fulfilled in the centimetre-wavelength core region. The wide frequency coverage of VGOS also allows us to decouple the Faraday rotation of the core into an internal component (${\rm RM}_I = 121 \pm 8$ rad m$^{-2}$, produced in the core region) and an external component (${\rm RM}_E = -44 \pm 9$ rad m$^{-2}$, associated with a distant, extended medium that may also affect the polarization in downstream regions of the jet at larger scales). Conclusions. These results demonstrate the power of VGOS for high-fidelity simultaneous multi-frequency polarimetric studies of compact AGN jets, and underline the need to account for non-equipartition effects in both jet astrophysics and geodetic VLBI.

Galaxies evolve within the intricate geometry of the cosmic web, yet the distinct roles of its primary components - nodes and filaments remain incompletely understood. Using the EAGLE cosmological hydrodynamical simulation, we investigate how distances to filament spines and cluster-scale nodes jointly and independently influence galaxy evolution. Galaxies are classified into red, green, and blue populations through a fully data-driven entropic thresholding technique, and the nodes and filaments are identified using DisPerSE. We find that red galaxies preferentially reside near filament cores and nodes, whereas blue galaxies dominate the outskirts. This spatial segregation reveals two characteristic transition scales: a node-related crossover at $\sim 2.5~\mathrm{Mpc}$ and a filament-related crossover at $\sim 0.75~\mathrm{Mpc}$. To further quantify environmental influence, we adopt an information-theoretic approach and measure the normalised mutual information between dominant mass component and galaxy colour across the $d_{\mathrm{f}}$-$d_{\mathrm{n}}$ plane. The mass-colour coupling increases with distance from nodes at fixed filament proximity, indicating a weakening of this relation in cluster-scale environments and a stronger coupling within filamentary regions. This behaviour is strongly mass dependent, with low-mass galaxies exhibiting a more pronounced environmental modulation than high-mass systems. These results support a scale-dependent view of galaxy evolution across the cosmic web, highlighting the distinct and complementary influence of nodes and filaments.

We investigate how misalignment between the core angular momentum and the large-scale magnetic field affects protostellar outflows, and whether a single protostellar system can drive multiple outflow components. We perform three-dimensional nonideal magnetohydrodynamic simulations of magnetized rotating cores, focusing on the formation of a protostar, a circumstellar disk, and magnetically driven outflows. The initial angle between the core angular-momentum vector and the magnetic field is systematically varied from $0^\circ$ to $90^\circ$. All models launch a classical magnetocentrifugal disk wind (DW) roughly along the local disk normal. For large misalignment, the system also develops a spiralflow (SF) component that propagates parallel to the disk plane. In a representative model with a $60^\circ$ misalignment, the outflow transitions from a DW-dominated to an SF-dominated phase, with the SF becoming more massive and more extended than the DW, and the two components intermittently coexisting. Across the model suite, the maximum mass and size ratios of SF to DW, as well as the relative lifetimes of the two components, increase for misalignment angles $\gtrsim60^\circ$. We propose that secondary, misaligned outflows (or their fossil remnants) observed in some protostellar systems can be interpreted as the SF component, while the main bipolar outflow traces the DW from the same misaligned system.

The first generations of stars ionised and enriched their host galaxies and seeded the growth of massive black holes. Models often assume that Pop III stellar masses in different minihaloes are stochastic realisations of a single universal initial mass function (IMF). We use 138 cosmological zoom-in hydrodynamics simulations to test this assumption and to map the first-star IMF across redshift, halo mass, and baryon-dark matter streaming velocity (SV). We construct a dense-cloud merger tree and assign first-star masses by mapping the radial gas accretion-rate profile to stellar mass, yielding per-halo stellar mass functions without imposing any a priori IMF. The high-mass tail and multiplicity increase systematically with redshift, halo mass, and SV. Low-mass, low-SV haloes form only one or a few first stars, whereas massive, high-SV haloes host rich first star clusters and commonly produce very massive ($\gtrsim10^3$-$10^4\,{\rm M}_\odot$) first stars. Even in a fixed environment, halo-to-halo scatter remains substantial. Nevertheless, group-averaged IMFs converge to well-defined forms, ruling out a single universal IMF at the halo level across the range of environments probed here. Mapping our seeds onto the redshift-mass plane, we show that high-SV and massive haloes preferentially populate the heavy-seed regime relevant to luminous high-redshift sources. At the same time, low-SV environments are consistent with single/few-event enrichment signatures in metal-poor stars. Our results deliver a practical, physically motivated prescription for per-halo IMF.

Fast blue optical transients (FBOTs) represent one of the most exotic astrophysical transients, exhibiting unusually strong emission across X-ray, optical, and radio wavelengths. Their physical origins remain highly debated, with proposed explanations ranging from stellar explosion to tidal disruption event (TDE). Here we report observations of the most luminous FBOT, AT 2024wpp whose post-peak luminosity rebrightens in X ray and becomes flattening in optical in a manner follows the decay rate characteristic of TDEs ($L_{\rm bol} \propto t^{-5/3}$). This invokes energy contribution of accretion by a central compact object, getting further corroborations from hardening of X-ray spectral index and detection of outflow inferred from the emission lines at similar phase. Detailed modeling of luminsoity evolution favors a coalesce explosion of a 34 M$_{\odot}$ Wolf-Rayet star with a 15 M$_{\odot}$ black hole (BH), demonstrating that some FBOTs may be associated with TDE of a stellar blackhole.

We report the discovery of a slowly evolving, extragalactic radio transient, ASKAP J005512.2--255834 (hereafter ASKAP J0055-2558), identified using the Australian SKA Pathfinder in a search for orphan afterglows associated with archival gravitational wave events. Although discovered in this context, there is no evidence that the transient is associated with any known gravitational wave event. Nonetheless, this source exhibits a 20-fold increase in flux density over $<250$ days, and it remains in a declining, detectable state more than 1000 days after the initial detection. Follow-up observations from 0.3 to 9 GHz reveal an evolving spectrum consistent with synchrotron emission. ASKAP J0055-2558 is spatially coincident with a low-mass, star-forming galaxy at redshift $z = 0.116$ ($d_{\rm L}$= 543 Mpc), placing its peak radio luminosity at $νL_ν\sim 10^{39}\,\rm erg\,s^{-1}$. Analysis of its radio light curve, inferred blastwave velocity, energetics, host galaxy properties and the absence of counterparts at other wavelengths suggest that ASKAP J0055-2558 is most consistent with either the late-time phase of an orphan long gamma-ray burst afterglow or a tidal disruption event involving an intermediate-mass black hole spatially offset from the galaxy nucleus. The radio discovery of either of these phenomena is extremely rare, with only a few or no confirmed examples to date.

Free Fractionally Charged Particles (FCPs) are predicted by some theories beyond or extended to the standard model. FCPs have been widely searched for by underground and space-based experiments based on the assumption of heavy lepton-like particles. However, there is a paucity of research focusing on light-mass FCPs (LFCPs) in the sub-MeV mass range. In this work, we report the LFCPs in primary high energy cosmic rays, based on observational data from the Dark Matter Particle Explorer (DAMPE) satellite. This study utilized ten years on-orbit data of DAMPE to search for LFCPs with a charge of $\frac{2}{3}~e$. No LFCP candidate was observed. Upper flux limit of LFCPs with a mass of 0.511 MeV$/c^{2}$ and a charge of $\frac{2}{3}~e$ is determined to be $\rm 5.0 \times 10^{-11}\,cm^{-2}sr^{-1}s^{-1}$ at the $\rm 90\%$ confidence level.

The putative dusty torus is a key ingredient of the unification scheme of active galactic nuclei (AGN), but its origin remains a mystery. Here we put forward a new physical model to explain how a large number of small dusty gas clumps form and they collectively appear as a geometrically thick dynamic dusty torus. The circumnuclear hot gas flows towards the central black hole (BH) and forms a rotating disk on sub-pc scales. A fraction of inflowing hot gas condenses to form small cold clumps due to thermal instabilities, when the accretion rate is sufficiently high. These cold dusty gas clumps are irradiated by the central accretion disk and re-radiate as dust emission mostly in the infrared. We propose that the dusty torus in AGN consists of such cold clumps vertically supported by the radiation force against gravity. For clumps with suitable column density, the vertical component of the BH gravity is in quasi-static equilibrium with the infrared radiation force together with the vertical component of the disk radiation force. Our model is robust in the sense that for any reasonable range of parameters concerning clump vertical dynamical equilibrium a torus exists. We further show that the hot gas in the rotating flow condenses to cold clumps only if its accretion rate is higher than about one percent of the Eddington rate. The radiation force is unable to lift the cold gas clumps up away from the mid-plane when the luminosity of the disk surrounding the BH is lower than 0.1 percent of the Eddington luminosity. These two features of our model may provide a physical explanation for the lack of evidence of dusty tori in low-luminosity AGNs.

Context: We present the first Cosmological Parameter inferences from eROSITA X-ray observations of galaxy clusters using a Machine Learning algorithm. Methods: We train a Random Forest using mock catalogs of clusters from Magneticum multi-cosmology hydrodynamical simulations. We apply the trained ML algorithm to observed X-ray features (gas luminosity, mass, and temperature) at different redshifts from the eROSITA eFEDS and eRASS1 catalogs. Results: We obtain cosmological constraints with precision comparable to those from standard analyses, such as weak lensing and cluster abundances. We infer $Ω_{\rm m}=0.30^{+0.03}_{-0.02}$, $σ_8=0.81\pm0.01$, and $h_0=0.710\pm0.004$. The recovered parameters show no tension in the $Ω_{\rm m}-σ_8$ space, but a significant deviation of $h_0$ from the Planck estimates. These inferences remain rather stable against variations of the input observable set and parameter space coverage. These results indicate that correlations among intracluster properties contain cosmological information beyond that encoded in the cluster abundance alone, which can be captured by machine learning trained on multi-cosmology simulations. Conclusions: ML algorithms trained on multi-cosmology hydrodynamical simulations can effectively infer cosmological parameters directly from galaxy cluster data. This is a change of paradigm in the context of cosmological parameter inferences. This approach complements traditional cluster-count analyses and is particularly suited to large upcoming surveys, where systematic uncertainties in mass calibration may otherwise dominate the error budget. It also highlights the potential of large-scale X-ray surveys to deliver independent tests of the standard cosmological model.

Many astrophysical fluids are magnetized and turbulent. Such fluids can be often described by magnetohydrodynamics (MHD). In this review, we mainly consider MHD turbulence with a strong mean magnetic field whose energy density is greater than or equal to the local kinetic energy density. In these fluids, the MHD waves, especially Alfvén waves, play a dominant dynamical role. Alfvén waves travel along magnetic field lines and collisions of opposite-traveling Alfvén wave packets are essential for turbulence cascade. We focus on strong turbulence regime, where nonlinear interaction during the collision is sufficiently strong and thus one collision is enough to complete turbulence cascade. We will cover the following types of turbulence. First, we review strong Alfvénic MHD turbulence. If the mean magnetic field is very strong, wave packets move very fast and duration of collision is too short to complete turbulence cascade. Even in this case we will show that strong turbulence regime emerges on a small scale. Second, we will consider small-scale MHD turbulence, where interaction of small-scale variants of Alfvén waves (i.e., whistler waves) is important. Third, we review scaling relations in strong relativistic force-free MHD turbulence, where interaction of relativistic Alfvén waves is important. Finally, we briefly discuss scaling relations in compressible MHD turbulence, where interaction of Alfvén waves is still important.

The existence and prevalence of planar, co-rotating distributions of satellite galaxies around L* host galaxies in the local universe remains a subject of ongoing debate. Despite numerous observational efforts over the past decade, a statistically robust sample of "satellite planes" across the diversity of host galaxy environments is lacking. To guide future observing strategies, we construct a controlled suite of mock observations of on-sky positions and line-of-sight (LOS) velocities of isolated L* host galaxies and their satellite systems, based on samples drawn from the Illustris TNG100-1 cosmological simulation to build a statistical sample. In these mock systems, satellite planes are defined by three key parameters: the number of satellites ($N_{\mathrm{sat}}$), the fraction residing in a thin co-rotating plane ($f_{p}$), and the orientation angle relative to the observer ($θ_{\mathrm{rot}}$). We evaluate the sensitivity of three observational metrics, $N_{\mathrm{cor}}$ (number of co-rotating satellites), $b/a$ (projected flattening of the satellite distribution), and $v_\mathrm{los}$ (mean absolute LOS velocity), to the presence of such planes. Our results show that detection rates are strongly dependent on $θ_{\mathrm{rot}}$ and $N_{\mathrm{sat}}$. Satellite planes that are viewed nearly edge-on or face-on, are the most readily detected. In contrast, intermediate orientations and systems with fewer satellites yield low detection success rates. Generally, only satellite planes with $N_{\mathrm{sat}}>20$ have high chances of being detected. These findings provide a practical framework for prioritising observational targets and designing future surveys aimed at detecting and characterising satellite planes.

We release the intermediate-width spectral window data from the ALMA Central Molecular Zone Exploration Survey (ACES) Large Program, which covers SiO(2-1), SO(2_2-1_1), H13CO+(1-0), H13CN(1-0), HN13C(1-0), and HC15N (1-0), among other molecular line transitions, with an angular resolution of ~2 arcsec and a velocity resolution of 1.7 km s-1 . The full cubes of the two spectral windows as well as the key data products will be available to the community. We also present the integrated brightness, peak brightness, centroid velocity, and Galactic longitude-velocity maps of the six lines. We briefly discuss morphological correlations between the continuum and the molecular line emission, and brightness ratios between pairs of isotopologue or isotopomer lines. We highlight features and trends in the data that will be followed up in upcoming ACES science papers.

During the final stages of a galaxy merger, dynamical friction acting on the supermassive black holes (SMBHs) in the post-merger remnant can lead to the formation of a gravitationally bound binary SMBH. In the event that at least one of these SMBHs is actively accreting, the system can appear phenomenologically as an active galactic nucleus (AGN) with a broad line region (BLR) kinematically offset from the host galaxy rest frame. Such velocity offsets have been interpreted as signatures of binary SMBHs, recoiling SMBHs, or BLR gas dynamics within a single-SMBH system. We present deep Chandra X-ray observations of five nearby (0.1 < z < 0.2) Sloan Digital Sky Survey quasars whose broad emission lines are Doppler-shifted relative to their host galaxies' systemic velocities, along with archival Chandra observations of 11 additional sources from the same sample. Using our Chandra data, we constrain SMBH masses with multiple independent techniques. We find systematic, method-dependent differences among black hole mass estimates, with masses inferred from the fundamental plane of black hole activity generally lower and single-epoch virial masses typically higher than those obtained using other methods. We also compare the X-ray photon indices and optical-to-X-ray spectral indices of our quasars to the broader quasar population. While we find no strong differences in optical-to-X-ray spectral indices, we do find systematically harder X-ray photon indices than typically observed in comparable quasars. These results constrain competing physical models but do not provide conclusive evidence for or against a binary SMBH origin of the velocity-offset BLRs.

We develop a unified analytic framework for modeling the real-space dark matter bispectrum, including both auto and cross bispectra constructed from pre- and post-reconstruction density fields, based on Unified Lagrangian Perturbation Theory (ULPT). ULPT reorganizes the standard Lagrangian approach by separating the Jacobian deviation, which generates the nonlinear source bispectrum, from the displacement-mapping effect that determines how long-wavelength bulk flows distort observed configurations. Within this structure, we derive general one-loop ULPT expressions for the bispectrum and analyze their infrared (IR) behavior, demonstrating exact, nonperturbative IR cancellation and the natural emergence of an IR-resummed description of baryon acoustic oscillation (BAO) damping. In particular, ULPT enables a detailed and fully analytic treatment of the wiggle-wiggle contribution to the IR-resummed bispectrum, whose structure has remained comparatively unexplored in previous approaches. We further construct IR-resummed models for the cross bispectra of all pre- and post-reconstruction combinations. For configurations in which the displacement fields differ, ULPT captures not only the overall exponential damping but also the more intricate BAO-scale modulation characteristic of mixed pre/post bispectra. Our results clarify the physical origin of nonlinear BAO suppression and provide a compact theoretical framework in which the full ULPT bispectrum, once implemented numerically through displacement-mapping convolution integrals, will automatically encode all nonlinear BAO and IR effects. The framework developed here therefore offers a unified and IR-safe foundation for interpreting next-generation bispectrum measurements.

Emission-line galaxies (ELGs) are an important tracer of baryon acoustic oscillations (BAO) and large-scale structure (LSS) at $z > 1$. In this work, we investigate the feasibility of using deep wide-area multi-band imaging (e.g., from the Rubin Observatory) to efficiently select high redshift ELGs. Using Hyper Supreme-Cam $grizy$ photometry and COSMOS2020 many-band photometric redshifts, we designed simple color cuts guided by a probabilistic random forest classifier to select galaxies at $z = 1.1$--$1.6$. We then empirically tested and refined these color cuts using two samples of galaxies with deep spectroscopy and broad color coverage obtained with the Dark Energy Spectroscopic Instrument (DESI). Compared to DESI ELGs at $z = 1.1$--$1.6$, we achieve a higher redshift measurement success rate (89% versus 69%), a much higher correct redshift range success rate (84% versus 34%), and a far higher net surface density yield (1372 $\mathrm{deg^{-2}}$ versus 660 $\mathrm{deg^{-2}}$). Combining our sample with current DESI ELGs would increase the net ELG number density by a factor of $\sim3$, moving it out of the shot-noise limited regime and reducing the uncertainties on the BAO scale parameter at $z = 1.1$--$1.6$ by a factor of $\sim 2$.

We investigate the relationship between star formation and X-ray emission from the hot interstellar medium (ISM) on $\sim$kpc scales in NGC 4254 (M99) by combining spatially resolved star formation histories (SFHs) and Bayesian X-ray spectral fitting. We measure sub-galactic star formation rates (SFR) by modeling spectrophotometric UV-IR data with flexible SFHs, and we produce point-source-subtracted maps of the diffuse X-ray emission using Chandra data. We extract and fit the spectra of 5 regions selected by their SFR density $Σ_{\rm SFR}$, deriving hot gas luminosities and plasma temperatures. We examine the sub-galactic $kT-Σ_{\rm SFR}$ and $L^{\rm gas}_X-Σ_{\rm SFR}$ scaling relations in NGC 4254, and compare to predictions from simple models of the feedback into the ISM from core collapse supernovae (CCSNe). The hot gas emission from NGC 4254 is consistent with thermalization of $\approx 40-50\%$ of the energy from CCSNe in the ISM, and mass-loading of the CCSNe ejecta which decreases as $Σ_{\rm SFR}^{-1/3}$. Our optimized model implies a temperature and X-ray production efficiency that scale as $kT = (0.72^{+0.26}_{-0.18}~{\rm keV}) Σ_{\rm SFR}^{0.34\pm0.10}$ and $η= (0.03^{+0.02}_{-0.01}) Σ_{\rm SFR}^{0.34\pm0.18}$, respectively, for $Σ_{\rm SFR} = 0.01-0.13~{\rm M_{\odot}~yr^{-1}~kpc^{-2}}$. We also compare the properties of the hot ISM to other ISM phases using data from the PHANGS program. The diffuse X-ray emission of a given region is on average 200 times fainter than the H$α$ emission, and we see evidence that the hot ISM is over-pressurized compared to the large-scale dynamical equilibrium pressure of the galaxy, consistent with expansion of the hot ISM into the ambient medium.

Large dust grains in thermal equilibrium dominate the far-infrared and contribute to the millimetre continuum of star-forming galaxies, but constraining their properties is difficult due to free-free and synchrotron contamination. We study spatial variations in the dust spectral index, mass, and grain properties in the nearby face-on spiral galaxy M 99. We use new 1.15 and 2 mm continuum observations from NIKA2 on the IRAM 30 m telescope (IMEGIN Guaranteed Time Large Programme) combined with multiwavelength data from UV to radio. The infrared-to-radio SED is decomposed into dust, free-free, and synchrotron components using the hierarchical Bayesian code HerBIE. Dust is modelled via a modified blackbody (MBB) with variable millimetre spectral index beta and the THEMIS dust model with fixed beta. We perform spatially-resolved analysis at scales ~1.75 kpc (~25''), covering the centre, spiral arms, and inter-arm regions. MBB fits reveal beta variations from ~1.6-1.7 in diffuse regions to ~2.3-2.5 in dense star-forming areas, likely due to grain coagulation and changes in silicate-to-carbonaceous ratios. Dust masses from variable beta are up to ~4x higher than fixed-beta models, which systematically bias dust-to-stellar and dust-to-gas ratios. The small grain fraction rises from ~10% in the centre to ~15% in the diffuse disc, anti-correlated with the interstellar radiation field; gas-phase metallicity plays a minor role within 8 kpc. Synchrotron spectral index ranges from ~0.6-0.7 in star-forming regions to ~1.2 in the diffuse medium, consistent with cosmic-ray electron ageing.

A central topic in extragalactic astronomy is understanding the formation and evolutionary histories of galaxies. These systems often comprise multiple structural components with distinct physical and dynamical properties, making it challenging to disentangle their individual contributions. Aiming at investigating the true structure of the inner stellar disk, we have developed a comprehensive pipeline for the chronochemical and dynamical analysis of galaxies (Glance: Galactic archaeoLogy via chronochemicAl & dyNamiCal modElling). The presented pipeline employs several state-of-the-art techniques by integrating them into a single, automated pipeline, enabling streamlined analysis of integral-field spectroscopy data, by allowing users to easily and directly extract valuable information on stellar populations, kinematics, dynamics, and gas properties. It automates multiple analysis techniques, including stellar population synthesis (Fado, Starlight, post-processing with RemoveYoung, kinematic extraction (pPXF, Bayes-LOSVD), and dynamical modelling (Dynamite). It handles tasks such as Galactic extinction correction, de-redshifting, Voronoi binning, and nebular continuum correction, while offering extensive customization options. Parallel processing significantly reduces computational time. When applied to MUSE data sampling the central region of NGC 1566, this methodology reveals that its stellar disk significantly deviates from the conventional exponential model, challenging the assumption of universality in disk morphology. In summary, this work presents a powerful, publicly available pipeline for conducting galactic archaeology, designed to advance our understanding of the formation and evolution of galaxies.

CO$_2$ is an important, stable, and abundant molecule in the Universe, but it is very difficult to detect because it has no observable pure rotational transitions. The unique sensitivity and resolution of the James Webb Space Telescope (JWST) provide a fresh way to investigate it. CO$_2$ is typically found in the solid phase (ice) on grain mantles in dense molecular clouds, but is less commonly detected in the gas phase (compared to common molecules such as CO and H$_2$O) and has mostly been found in protostellar and proto-planetary environments. Here, we report and characterize the first observations of gas-phase CO$_2$ absorption toward two IR-bright regions of the Galactic Center, thanks to the high sensitivity of JWST. Using an LTE model we find a CO$_2$ gas excitation temperature between 20 and 50~K, a column density around 2$\times$10$^{15}$~cm$^{-2}$ and a radial velocity consistent with 0. We also report: 1) simultaneous detections of C$_2$H$_2$ and HCN absorption bands (near 13.7 and 14.0 $μ$m, respectively), with column densitiy ratios of 1:3 and 3:2 with respect to gas-phase CO$_2$, and 2) CO$_2$ ice absorption with a ice-to-gas ratio of 90, consistent with previous findings. We conclude that the absorbing medium is likely in the foreground, most likely from one or more somewhat clumpy cloud(s), located between 0.15 and 4~kpc away from Earth. Additionally, we detected point-like CO$_2$ emission likely associated with a Galactic Center star (IRS~11SW), which is also spatially coincident with a previously reported X-ray source, raising the possibility that the system is a symbiotic binary.

Reliable, versatile galaxy activity diagnostics are essential for understanding galaxy evolution. Traditional methods frequently necessitate extensive preprocessing, such as starlight subtraction and emission line deblending (e.g., Hα and [N II]), which can introduce substantial biases and uncertainties due to their model-dependent nature. In this work we developed an automated, diagnostic tool capable of distinguishing between star-forming (SF), active galactic nuclei (AGN), low-ionization nuclear emission-line regions (LINERs), composite, and passive galaxies. We developed a diagnostic tool based on a support vector machine trained on data from optical emission-line ratios and color selection criteria. From literature studies and exploring combinations of discriminatory feature schemes, we found that the equivalent widths of Hβ, [O III]λ5007, and Hα+[N II]λ6548,84 as key diagnostic features. Additionally, galaxies classified as AGN can be distinguished into broad- and narrow-line AGN by measuring the full quarter at the half-maximum of Hα and [N II] complex. We have developed a diagnostic tool that encompasses all activities of galaxies while achieving high performance scores across all of them. Our diagnostic achieves overall accuracy of 83% and recall of 79% for SF, 94% for AGN, 85% for LINER, 77% for composite, and 96% for passive galaxies. Our diagnostic tool significantly improves upon existing diagnostics as it eliminates the need for preprocessing (i.e., starlight subtraction or flux calibration) and spectral line fitting, includes all activity classes under one scheme, and distinguishes the two main AGN types. In addition, omitting starlight subtraction does not significantly reduce performance. Furthermore, Its narrow wavelength requirement enables use across a wide redshift range, making it ideal for high-z studies.

The mass flows and energy cycles within the inner regions of galaxies exert a powerful influence on the evolution of the galaxy population. The centre of the Milky Way is the only galactic nucleus for which it is possible to resolve the physical mechanisms that drive these cycles, namely star formation and feedback, while also tracing global (>100 pc) processes which determine where and when star formation and feedback occur. We present an overview of ACES, the 'Atacama Large Millimeter/submillimeter Array (ALMA) CMZ Exploration Survey', a ~1.5" angular resolution, 0.2-3 km/s spectral resolution ALMA Band 3 (85-102 GHz), survey of the 'Central Molecular Zone' (CMZ) -- the inner-100 pc of the Galaxy (l = 359.4 deg to 0.8 deg). ACES spectral setup is tuned to observe optimal tracers of the physical, chemical, and kinematic conditions in over 70 spectral features (e.g. HCO+, HNCO, SiO, H40alpha, complex molecules) of the gas in the CMZ, to derive the properties of all potentially star-forming Galactic Centre gas, from global scales (100 pc) to dense ~0.05 pc structures that are expected to host individual star-forming cores, down to sub-sonic (<0.4 km/s) velocity resolution. In this overview paper, we provide the scientific justification for the ACES survey, explain the choice of observational setup, and describe the data legacy products. Finally, we show some of the initial ACES data which highlight the power of ACES' combination of high angular resolution, unprecedented spatial dynamic range, sensitivity, spectral resolution and spectral bandwidth as an illustration of how ACES aims to understand how global processes set the location, intensity, and timescales for star formation and feedback in the CMZ.

We aim to increase the limited number of quasars behind M31, necessary for probing the chemical content of the gas and for proper motion reference, with reliable and homogeneous redshift measurements from emission lines. We carried out spectroscopic follow up of 32 quasar candidates. We confirm 23 quasars: two are new discoveries (J004029.727+403705.68 and J004215.489+412031.52) and the rest were reported elsewhere, but with somewhat deficient analysis; 16 spectra are published for the first time. We report new homogeneous redshifts for 34 quasars (from 40 spectra, adding 17 from archives) and summarize all available information about bona-fide quasars with reliable redshift, bringing their number to 124 within the mu_B=26 mag/arcsec^2 isophote. We carried out a comparison of redshifts from different sources and excluded some objects with redshifts derived from low-resolution spectra. We derive the reddening for them from the color excess with respect to dereddened counterparts with similar redshifts in the field. Comparisons of our reddenings with M31 reddening maps found no significant correlations. Most QSOs behind M31 show low reddening and do not probe high-extinctions underlining the need to identify fainter quasars behind nearby galaxies, especially behind higher extinction regions -- probably due to a bias towards following up brighter and less extinct candidates. Finally, the redshifts derived from low-resolution spectra must be treated with caution, because they can contain significant errors.

X-ray variability is a well-established characteristic of active galactic nuclei (AGN), known to correlate inversely with both the supermassive black hole mass and luminosity, although the degree of each remains a topic of debate. The potential of X-ray variability as a proxy for MBH or for intrinsic LX has led to proposals to use AGN as standard candles to test cosmological models. However, the large intrinsic dispersion in these relations has limited their practical applications. In this work, we investigate the dependence of X-ray variability on AGN physical properties using a sample of 134 Seyfert 1 galaxies from the BAT AGN Spectroscopic Survey (BASS), which is the largest sample to date, more than three times larger than those used in previous studies. Contrary to earlier findings, we observe that X-ray variability correlates with luminosity just as strongly as with MBH. Furthermore, we still do not find evidence for the expected anti-correlation between variability and Eddington ratio, even when using refined bolometric luminosities from SED fitting to compute the Eddington ratio. From a cosmological perspective, the increased sample size reduces the scatter in the log(L)-log(exvar) relation to ~0.63 dex - a significant improvement over previous results, but still too large to serve as competitive standard candles, when compared to SNIa (uncertainties on distances of ~5-10%) or the L(X)-L(UV) relation in quasars (uncertainties of 10-12%). We tested including the width of broad emission lines as additional parameters, but found that this does not significantly lower the observed dispersion, contrary to previous studies on smaller samples. Finally, we discuss how future X-ray missions such as AXIS and NewAthena will improve this scenario by enabling precise variability measurements for thousands of AGN up to redshift z~3, thereby enabling it as a new cosmological probe.

The ALMA Central molecular zone Exploration Survey (ACES) large program has observed the inner ~ 200 pc of the Milky Way at 3 mm (Band 3) using ALMA's 12m, 7m, and Total Power arrays. With an angular resolution of ~ 2", ACES provides a contiguous, multi-scale view of the Central Molecular Zone (CMZ) via the dust continuum and a suite of molecular lines. We present an overview of the molecular line data processing for ACES and describe the first data release. We showcase the HNCO (4-3) and HCO$^{+}$ (1-0) data, which were targeted at high spectral resolution (0.2 km s$^{-1}$) to trace the kinematics of the molecular gas in the CMZ. The HNCO and HCO$^{+}$ maps are compared with previous single-dish CMZ surveys and discrete ALMA observations of CMZ clouds to demonstrate the quality of the data. We highlight the ubiquity of parsec-scale, linear absorption features traced by HCO$^{+}$. Their origin is unknown, and ACES provides the first opportunity to study these enigmatic features throughout the CMZ. We release the HNCO and HCO$^{+}$ cubes for all 45 ACES fields, along with the full cube mosaics which combine all fields into a contiguous mosaic of the CMZ. We additionally provide advanced products of these full mosaics, including integrated and peak intensity, noise, and position-velocity maps. These products provide substantial legacy value for the community, offering an unparalleled view of the physical and kinematic structure of the dense gas in the CMZ.

The ultra-faint Milky Way satellite Willman 1 (W1; $M_V = -2.6$; $r_{\rm half} \sim27$ pc) was the first stellar over-density found via resolved stars in the Sloan Digital Sky Survey, yet its classification as either a dwarf galaxy or star cluster remains ambiguous. Using new Keck/DEIMOS spectroscopy, HST/ACS photometry, and orbital modeling, we re-examine the nature of W1. From our updated sample of 57 member stars, we find that past analyses included four binaries and seven non-members, identified here using Gaia proper motions and updated velocities. We continue to find a velocity dispersion consistent with previous analyses, measuring $σ_v = 4.7^{+1.5}_{-1.3}$ km s$^{-1}$ from 49 stars out to $3~r_{\rm half}$. If W1 is in equilibrium, this suggests a dynamical mass of $5.9^{+3.7}_{-3.4} \times 10^5~M_{\odot}$ and a mass-to-light ratio of $(M/L)_V = 660 \pm 590$. Based on Ca II triplet measurements, we estimate an iron abundance of [Fe/H] $= -2.45^{+0.12}_{-0.13}$ and a metallicity dispersion of $σ_{\rm [Fe/H]} = 0.30^{+0.15}_{-0.11}$ dex. We confirm that W1 does not exhibit mass segregation inside $\sim1~r_{\rm half}$. Our best-fit orbital model predicts that W1 is at apocenter, implying that W1 has been closer to the Milky Way in the recent past, reaching a pericentric distance $\lesssim 25$ kpc from the Galactic center $\sim0.3$ Gyr ago. Given its internal kinematics, metallicity spread, and lack of mass segregation, we conclude that W1 is a galaxy. However, given its orbit and structural properties, which suggest that W1 might be tidally disrupted, and the difficulty of identifying a pure member sample, we caution that the measured internal velocity dispersion may not accurately reflect the dynamical mass of this system.

The Central Molecular Zone (CMZ) of the Milky Way is the way station that primarily controls how much gas flows from the disk of the Galaxy towards the central nucleus. While this region is well documented to have extreme gas properties that clearly distinguish it from the rest of the Galaxy, the properties of the bulk molecular gas at high angular resolution are relatively unexplored. Band 3 data from the ALMA (Atacama Large Millimeter/Submillimeter Array) large program ACES (ALMA CMZ Exploration Survey) reveal the highly filamentary nature of CMZ molecular gas at high resolution (3" or 0.1pc) across the entire CMZ. Visual inspection of these data suggests that there are at least two general classes of elongated structures, which we identify as: i) large-scale (10 pc) filamentary structures (LFs) and ii) a ubiquitous population of small-scale (about 1 pc) filamentary structures (SFs). We present detailed morphological and kinematic properties towards three structures in each category, as well as their association with magnetic fields and the correlation of HNCO 4(0,4)-3(0,3) with other molecular species. Our investigation reveals that these structures are largely coherent in position-position-velocity space. The alignment with the magnetic field structure is mixed, with some parallel, some perpendicular, and some intermediate alignments. We find that LFs likely trace pieces of contiguous CMZ orbital structures and are a manifestation of global CMZ dynamics. The second class, SFs, are pervasive and may be the result of complicated turbulence and shearing dynamics in the CMZ gas flows, as seen in numerical simulations.

Stellar halos are the only major stellar component of disk galaxies that lack systematic observational characterization, yet they encode critical information about galaxy merger histories. We present the first systematic census of stellar halos in a large, flux-limited sample of 169 high-inclination central galaxies with stellar masses 7.3 <= log Mstar/Msun <= 11.0 and redshift z < 0.1, using HSC-SSP Deep optical images. Stellar halos are detected in 93 galaxies, primarily through their low isophotal ellipticities in the outskirts, improving upon conventional methods of stellar halo identification. The halo detection rate reaches ~ 50% at log Mstar/Msun > 9.9 and >= 70% for Milky Way (MW)-mass galaxies. We derive halo surface brightness profiles, colors, and masses, finding that stellar halos generally follow power-law radial profiles. Higher-mass galaxies, on average, exhibit smaller power-law indices and larger halo mass fractions, indicating more extended halos and more active merger histories. A significant stellar halo color-mass correlation, driven mainly by the mass-metallicity relation, suggests dominance by a few massive accretion events. MW-mass galaxies have a median stellar halo fraction of 10% +/- 5%. Among nearby galaxies with halo measurements within 25 Mpc, two thirds (including the MW) lie below the mean stellar halo fraction-galaxy mass relation. Overall, the nearby galaxies show a median halo deficit of ~ 0.3 dex, implying unusually quiescent merger histories. We show that this deficit follows a broader trend in which typical halo fractions increase with heliocentric distance, tracking the gradual rise in matter density toward the cosmic average by z <= 0.07.

A significant fraction of millisecond pulsars (MSPs) in globular clusters (GCs) are observed as isolated objects, despite the widely accepted scenario in which MSPs are formed through recycling in compact binary systems. The origin of these isolated objects therefore remains an open problem. In this Letter, we propose a physically motivated encounter rate per binary, $Λ\propto n a/(σa_H)$, incorporating the local stellar density $n$, velocity dispersion $σ$, binary separation $a$, and the Heggie--Hills ionization radius $a_H$. Combined with companion ablation by the MSP, this rate successfully predicts the observed fraction of isolated MSPs in GCs, that is $\mathcal{F}_i \proptoΛ\propto a_H^{-1}$, establishing dynamical ionization as the primary channel for producing isolated MSPs. We quantitatively test this model against a null hypothesis in which $\mathcal{F}_i$ is independent of $a_H$, and find that the ionization-driven model is 220 times more likely than the null hypothesis. Our framework naturally explains the observed overabundance of isolated MSPs in $ω$ Centauri and establishes binary ionization as the primary mechanism responsible for the production of isolated MSPs in GCs.

We compile JWST/NIRSpec prism and MIRI data for 249 Little Red Dots (LRDs) at $2.3<z<9.3$, forming a representative spectroscopic subset of NIRCam-selected LRDs. We derive a median stacked spectrum covering rest-frame 0.09-1.2 $μ$m, with MIRI photometry extending the spectral energy distribution to 4 $μ$m. Four additional stacks for subsamples defined by optical-to-UV luminosity ratios show that LRDs form a heterogeneous population spanning diverse continuum slopes and line properties. Assuming LRDs host super-massive black holes (BHs) surrounded by dense gas clouds, and stars accompany this core, we infer masses of $M_{BH}\sim10^{6.0-6.5}$~M$_\odot$ and $M_\bigstar\sim10^{8.3}$ M$_\odot$, corresponding to BH-to-stellar mass ratios of 1-2%. The stacks show ubiquitous UV and optical FeII emission, indicating a direct view of the broad-line region and high (but sub-Eddington) accretion ($λ_{Edd}=0.6\pm0.2$). We find a significant stellar contribution in the far-UV, reaching $\sim80$% in the bluest systems. Possible Wolf-Rayet features (HeII$λ$4687, nitrogen lines) are identified, tracing a young (3-7 Myr) compact starburst event. We also detect strong Balmer breaks and atypical Balmer, Paschen, [OIII], and optical and near-infrared HeI line ratios, and an absorption at $\sim4550$ Angstrom (probably linked to FeII), all consistent with radiative-transfer effects in high-density gas with warm temperatures (4000-7000 K). We find a diversity of LRD flavors modulated by the luminosity ratio between between a short ($\lesssim20$ Myr) and intense phase of BH activity, the most extreme stage lasting $\sim3-7$ Myr, characterized by near-Eddington-limit radiation, and a nuclear and compact starburst dominated by massive stars (even super-massive, $\mathrm{M}_\mathrm{SMS}\sim10^{5}$ M$_\odot$), all embedded in dense gas with modest dust content producing a variety of optical depths.

Gaseous halos around galaxies play an important role in galaxy evolution. The exchange of metals from the interstellar medium (ISM) to the circumgalactic medium (CGM) are caused by the formation, feedback, and/or merging history of galaxies. We study the variation in chemical composition between the ISM ($\lesssim3\,$kpc) and CGM ($\sim5-10\,$kpc) for a sample of $M_{\star}>10^{9.5}\,M_{\odot}$ main-sequence galaxies at $4<z<6$ with both JWST and ALMA observations. Using JWST/NIRSpec integral field spectroscopy, we derive the optical line ratios from the ISM and the CGM for our sample focusing on the typical optical lines used for metallicity studies. Our comparison shows that the ISM and the CGM have similar chemical abundances. This indicates that the CGM of these typical $4<z<6$ galaxies is enriched to the level of their ISM in the early universe. Using statistical tests, we find that some of the line ratios show marginal differences between the ISM and CGM. Combined with \texttt{Cloudy} modeling, our results suggest that a difference in ionization level (higher for the ISM) is the dominant reason for the observed ratio difference of oxygen line ratios. There is also indication of a deficit in the nitrogen abundance with respect to oxygen in the CGM, which suggests a delay in redistribution of secondary nitrogen. Finally, an enhanced $F_{[\rm CII]\rm 158μm}/F_{\rm Hα}$ ratio is observed in the CGM, suggesting that feedback and/or mergers play a key role in metal mixing.

We study binary stars moving through a uniform dark matter background and experiencing dynamical friction. The centre-of-mass motion of the pairs is taken into account. We derive formulas and timescales for the secular evolution of the orbital parameters for both wide and close binaries. We apply these results to environments typical of dark matter dominated ultra-faint dwarf galaxies and show that some binaries undergo significant eccentricity oscillations, while their semi-major axes decrease more gradually. We consider a simple binary star population and find that dynamical friction, notably, can enhance the bias from unresolved binaries in velocity dispersion measurements. With future, more detailed theoretical studies and improving observational capabilities, binary stars may serve as a tool to probe the dark matter content of some of the faintest galaxies.

The ALMA Central Molecular Zone Exploration Survey, ACES, has mapped $\gtrsim1000$ square arcminutes at 3 mm toward the center of our Galaxy. ACES provides the first large-scale, high-resolution ($\sim2.5$") view of the central $\sim200$ parsecs of the Milky Way. In this work, we describe the continuum data processing and present the continuum data products. In the combined mosaic of 45 individual ALMA mosaics, the typical RMS noise achieved is $\sim0.1$ mJy per $\sim2.5$" beam, though there is a tail of substantially higher noise toward regions with bright continuum structure, especially around Sgr A* and Sgr B2. In-band spectral indices are measurable for a small fraction of the brightest and most compact sources, enabling distinction between dust-dominated and free-free- or synchrotron-dominated sources. To recover emission on large angular scales, we present the GBT MUSTANG-2 Three millimeter Extended Nucleus Survey (TENS), a new 10"resolution survey of the CMZ, which we combine with the ACES image by feathering. To demonstrate the quality and reliability of the ACES data, we compare to previously-published ALMA data obtained with higher resolution and sensitivity, finding overall good agreement with past results, but some disagreement toward the brightest sources.

Considering the physics of radiation-driven winds of massive stars, the wind properties should depend on the metal content of the stellar atmosphere. Therefore, studying the winds of massive stars in different metallicities provides a sanity check on prescriptions that are widely used in evolutionary calculations. We obtained the stellar and wind properties of a sample of 20 late-O and B supergiants in the Small Magellanic Cloud (SMC) from a quantitative combined UV and optical spectroscopic analysis using CMFGEN. By comparing these properties with those of a Large Magellanic Cloud counterpart study, which has a similar sample and data, and employed the same modelling techniques used in this study, We derived a metallicity-dependent recipe for wind momentum, which is applicable for $5.4 \leq \log{L_{\rm bol}/L_{\odot}} \leq6.1$ and $14 \leq T_{\rm eff}/{\rm kK} \leq 32$. We find a significant dependence of the wind momentum on the metallicity, which is largely due to the mass-loss rates. We do not find any evidence of a discontinuity in either the mass-loss rate or the ratio of the terminal wind velocity to the escape velocity between $25$ and $21$~kK, which could be attributed to the bi-stability jump. Stellar parameters are consistent across different methods and radiative transfer codes, whereas mass-loss rates differ significantly, with our values being generally lower. We find a discrepancy between the evolutionary and spectroscopic masses in $40\%$ of our sample, with the evolutionary mass usually being systematically higher. The mass-loss rates of blue supergiants are far too low to strip the stellar envelope and the subsequent formation of classical Wolf-Rayet (WR) stars, leading to the conclusion that luminous blue variable eruptions or binary interactions are necessary to explain the characteristics of the WR population in the SMC.

We present a radio polarimetric study of the Boomerang pulsar wind nebula G106.65+2.96 with VLA observations at the 6 GHz band. Our high-resolution image discovers new small-scale features in the nebula, including an elliptical core of $40''\times20''$ surrounding the central pulsar and a $2'$-long arc wrapping around the core in the north. The latter shows a clear gap from the core, and it consists of a bright lobe in the northwest and a tongue-like structure in the northeast. These could be resulting from the pulsar wind interaction with the environment. Our polarization measurement reveals a highly ordered magnetic field with toroidal geometry. The small scale features are all highly linearly polarized. In particular, the lobe has a polarization fraction of $\sim$60%, close to the synchrotron limit. This is also much higher than the value measured at a lower frequency, implying significant depolarization. We show that this can be explained by Faraday rotation in the nebula, and we constructed a simple 3D model accordingly to infer a magnetic field strength of $\sim$50-105$μ$G.

Open clusters (OCs) usually evolve gradually as the number of their members changes, which can be manifested in their morphological characteristics. We aim to investigate the morphological stability of 1,490 OCs and further explore the potential change of morphological stability of the OCs at different spatial positions, using the OC catalog from the literature. We define for the first time a new morphological stability parameter Ncore/Nouter, a ratio of member numbers between cluster core and outer areas within tidal radii, which has a significant positive correlation against N, with a slope of 1.140$\pm$0.039, significantly steeper than the 0.720$\pm$0.026 measured for Score/Souter. This demonstrates that the stellar density in the core is a more sensitive tracer for morphological stability than geometry. Spatially, the radial sample OCs have larger slopes of Ncore/Nouter and Score/Souter against N, with 1.083$\pm$0.116 and 0.733$\pm$0.080, respectively, whereas those in the tangential direction 1.013$\pm$0.110 and 0.529$\pm$0.075, respectively, which means that the impact on sample OCs from tidal forces directed toward the Galactic center is possibly stronger than that from the shear force caused by the differential rotation of the Galactic disk. Besides, the sample OCs younger than 30 Myr display a shallow slope of 0.751$\pm$0.166, with those older than 800 Myr (1.442$\pm$0.128), reflecting that young OCs likely endure both internal disruptions, such as early dynamical heating weakening core binding and more severe external disturbances, compared to older OCs.

Active Galactic Nuclei (AGNs) are characterized by strong temporal flux density variability across the electromagnetic spectrum, offering insights into the complex physical processes governing accretion and plasma outflows. To systematically investigate AGNs flux density variability in radio bands, a long-term program was initiated in late 2016: quasi-Simultaneous Multiwavelength Monitoring of gamma-ray-loud AGNs with the Nanshan 26-m radio telescope (SMMAN). This work presents the first data release of the SMMAN program, spanning over eight years from 2016 to 2024 with observations at 4.8 and 23.6 GHz. The SMMAN sample includes 131 northern ($δ>\sim0^{\circ}$) sources selected from the Fermi Large Area Telescope third source catalog. The characteristics of variability, spectral index, luminosity, and $γ$-ray loudness factor are examined for different AGN classes within the sample. Target sources exhibit stronger variability at 23.6 GHz compared to 4.8 GHz, with BL Lac objects being more variable than flat-spectrum radio quasars (FSRQs). BL Lacs generally have flatter radio spectra, while FSRQs, blazar candidates of uncertain type (BCUs), and radio galaxies (RDGs) span a wider range from flat to steep. FSRQs are more radio-luminous than BL Lacs and other classes, with BCUs intermediate and RDGs generally fainter. FSRQs and BL Lacs have higher $γ$-ray loudness factors than RDGs, while BCUs have intermediate values. The SMMAN dataset, incorporated with other historical and ongoing monitoring programs, will provide a unique opportunity to investigate the evolution of spectral energy distributions, search for quasi-periodic oscillations, and analyze supermassive black hole binary systems.