arXiv Daily Digest - 2026-04-20

As a relatively new forum, ACM FAccT has become a key space for activists and scholars to critically examine emerging AI and ML technologies. It brings together academics, civil society members, and government representatives from diverse fields to explore the broader societal impacts of both deployed and proposed technologies. We report a large-scale participatory design (PD) process for reflexive conference governance, which combined an in-person CRAFT session, an asynchronous Polis poll and the synthesis of a governance-facing report for the FAccT leadership. Participants shaped the substantive agenda by authoring seed statements, adding new statements and making patterns of agreement, disagreement and uncertainty made visible through voting.Our endeavors represent one of the the first instances of applying PD to a venue that critically interrogates the societal impacts of AI, fostering a niche in which critical scholars are free to voice their concerns. Finally, this work advances large-scale PD theory by providing an effective case study of a co-design paradigm that can readily scale temporally and epistemologically.

In computer vision and machine learning for geographic data, out-of-domain generalization is a pervasive challenge, arising from uneven global data coverage and distribution shifts across geographic regions. Though models are frequently trained in one region and deployed in another, there is no principled method for determining when this cross-region adaptation will be successful. A well-defined notion of distance between distributions can effectively quantify how different a new target domain is compared to the domains used for model training, which in turn could support model training and deployment decisions. In this paper, we propose a strategy for computing distances between geospatial domains that leverages geographic information with Optimal Transport methods (GeoSpOT). In our experiments, GeoSpOT distances emerge as effective predictors of cross-domain transfer difficulty. We further demonstrate that embeddings from pretrained location encoders provide information comparable to image/text embeddings, despite relying solely on longitude-latitude pairs as input. This allows users to get an approximation of out-of-domain performance for geospatial models, even when the exact downstream task is unknown, or no task-specific data is available. Building on these findings, we show that GeoSpOT distances can preemptively guide data selection and enable predictive tools to analyze regions where a model is likely to underperform.

Large language models are increasingly deployed in settings where reliability matters, yet output-level uncertainty signals such as token probabilities, entropy, and self-consistency can become brittle under calibration--deployment mismatch. Conformal prediction provides finite-sample validity under exchangeability, but its practical usefulness depends on the quality of the nonconformity score. We propose a conformal framework for LLM question answering that uses internal representations rather than output-facing statistics: specifically, we introduce Layer-Wise Information (LI) scores, which measure how conditioning on the input reshapes predictive entropy across model depth, and use them as nonconformity scores within a standard split conformal pipeline. Across closed-ended and open-domain QA benchmarks, with the clearest gains under cross-domain shift, our method achieves a better validity--efficiency trade-off than strong text-level baselines while maintaining competitive in-domain reliability at the same nominal risk level. These results suggest that internal representations can provide more informative conformal scores when surface-level uncertainty is unstable under distribution shift.

Secondary school students enrolled in the AP Computer Science Principles (CSP) course commonly utilize web resources (e.g., tutorials, Q\&A sites) to better understand key concepts in the curriculum. The primary obstacle to using these resources is finding information appropriate for the learning task and student's background. In addition to web search, conversational agents are increasingly a viable alternative for CSP students. In this paper, we study the potential of conversational agents to aid secondary school students as they acquire knowledge on CSP concepts. We explore general purpose, generative conversational agents (e.g., ChatGPT) and custom, fixed-response conversational agents built specifically to aid CSP students. We present results from classroom use by 45 high school students in grades 9-11 (ages 14-17) across six CSP sections. Our main contributions are in better understanding how conversational agents can help CSP students and an evaluation of the effectiveness and engagement of different approaches for CSP exploratory search.

Software engineering research has experienced rapid growth in both output and participation over the past decades. Yet concerns persist about the field's ability to accumulate, integrate, and reuse knowledge in ways that support long-term progress. To better understand how the community itself perceives these challenges, we analyze responses from the ICSE 2026 Future of Software Engineering pre-survey, which captures perspectives from 280 globally distributed and highly experienced researchers. Our analysis reveals a tension between increasing research productivity and the limited mechanisms available for synthesizing results, tracking evolving claims, and supporting cumulative understanding over time. Building on these observations, we diagnose four interrelated structural breakdowns: papers function as isolated knowledge units with claims embedded in prose; context and provenance are lost as knowledge moves through the publication pipeline; claims evolve without systematic tracking; and incentive structures favor novelty over consolidation. We argue that addressing these barriers requires rethinking the fundamental properties of research artifacts. We articulate four technology-agnostic principles for future research artifacts: structured and interpretable representations of claims and evidence; inspectable and provenance-aware documentation of methodological decisions; long-lived and reusable substrates that evolve beyond publication; and governance mechanisms that align individual incentives with collective knowledge-building goals. We discuss implications for research practice, publication norms, and community infrastructure, positioning FOSE as a venue for experimenting with alternative artifact designs that support cumulative scientific progress.

As forgery types continue to emerge consistently, Incremental Face Forgery Detection (IFFD) has become a crucial paradigm. However, existing methods typically rely on data replay or coarse binary supervision, which fails to explicitly constrain the feature space, leading to severe feature drift and catastrophic forgetting. To address this, we propose AIFIND, Artifact-Aware Interpreting Fine-Grained Alignment for Incremental Face Forgery Detection, which leverages semantic anchors to stabilize incremental learning. We design the Artifact-Driven Semantic Prior Generator to instantiate invariant semantic anchors, establishing a fixed coordinate system from low-level artifact cues. These anchors are injected into the image encoder via Artifact-Probe Attention, which explicitly constrains volatile visual features to align with stable semantic anchors. Adaptive Decision Harmonizer harmonizes the classifiers by preserving angular relationships of semantic anchors, maintaining geometric consistency across tasks. Extensive experiments on multiple incremental protocols validate the superiority of AIFIND.

Computational X-ray absorption near-edge structure (XANES) is widely used to probe local coordination environments, oxidation states, and electronic structure in chemically complex systems. However, the use of computational XANES at scale is constrained more by workflow complexity than by the underlying simulation method itself. To address this challenge, we present ChemGraph-XANES, an agentic framework for automated XANES simulation and analysis that unifies natural-language task specification, structure acquisition, FDMNES input generation, task-parallel execution, spectral normalization, and provenance-aware data curation. Built on ASE, FDMNES, Parsl, and a LangGraph/LangChain-based tool interface, the framework exposes XANES workflow operations as typed Python tools that can be orchestrated by large language model (LLM) agents. In multi-agent mode, a retrieval-augmented expert agent consults the FDMNES manual to ground parameter selection, while executor agents translate user requests into structured tool calls. We demonstrate documentation-grounded parameter retrieval and show that the same workflow supports both explicit structure-file inputs and chemistry-level natural-language requests. Because independent XANES calculations are naturally task-parallel, the framework is well suited for high-throughput deployment on high-performance computing (HPC) systems, enabling scalable XANES database generation for downstream analysis and machine-learning applications. ChemGraph-XANES thus provides a reproducible and extensible workflow layer for physics-based XANES simulation, spectral curation, and agent-compatible computational spectroscopy.

Code generation refers to automatically producing executable programs from user requirements. Recently, researchers have explored approaches to enhance the correctness of generated code with advanced large language models. Although achieving improvements, existing approaches focus on designing reasoning strategies or post-refinement methods to enhance code generation performance. Despite their differences, all these methods share a common assumption: the LLM can correctly understand the given requirement. However, this assumption does not always hold. To fill this gap, we propose REA-Coder, a requirement alignment approach to enhance the code generation performance of LLMs. REA-Coder involves first identifying the requirement content that does not align with LLMs and aligning the requirements. Then, based on the aligned requirements, LLMs generate code and further verify whether the generated code aligns with the requirements, iterating this process of requirement alignment and code generation until generating correct code or achieving the maximum number of iterations. Experimental results show that REA-Coder outperforms all advanced baselines on four LLMs across five programming benchmarks. Concretely, REA-Coder achieves average improvements of 7.93%, 30.25%, 26.75%, 8.59%, and 8.64% on the five benchmark datasets, demonstrating the effectiveness of requirement alignment for improving the code generation performance of LLMs.

Data attribution and valuation are critical for understanding data-model synergy for Large Language Models (LLMs), yet existing gradient-based methods suffer from scalability challenges on LLMs. Inspired by human cognition, where decision making relies on a focused readout of relevant memories rather than replaying all pathways, we introduce RISE (Readout Influence Sketching Estimator). Instead of computing and indexing gradients across the entire LLM, RISE focuses on influence hotspots at the output layer, where influence signals concentrate, and the gradient admits a decomposed outer-product form. This enables a dual-channel representation combining a lexical residual channel (RH) and a semantic projected-error channel (GH). Applying CountSketch projections to these channels achieves strong compression while maintaining accurate attribution. Across the OLMo (1B-32B) and Pythia (14M-6.9B) families, RISE reduces index storage by up to 112$\times$ compared to RapidIn and scales to 32B parameters LLM, where gradient-based baselines such as RapidIn and ZO-Inf become memory-infeasible. We evaluate RISE on two paradigms: (1) retrospective attribution, retrieving influential training examples for specific predictions, and (2) prospective valuation, scoring candidate data utility zero-shot. We validate RISE on three tasks: Howdy backdoor data detection, Finance-Medical domain separation, and Brain Rot high-quality data selection. In a closed-loop Brain Rot study, continued pretraining on RISE-selected data yields consistent downstream improvements. Overall, RISE provides a practical and scalable primitive for influence analysis and training-data selection in modern large language models.

Data plays a fundamental role in consolidating markets, services, and products in the digital financial ecosystem. However, the use of real data, especially in the financial context, can lead to privacy risks and access restrictions, affecting institutions, research, and modeling processes. Although not all financial datasets present such limitations, this work proposes the use of deep learning techniques for generating synthetic data applied to cryptocurrency price time series. The approach is based on Conditional Generative Adversarial Networks (CGANs), combining an LSTM-type recurrent generator and an MLP discriminator to produce statistically consistent synthetic data. The experiments consider different crypto-assets and demonstrate that the model is capable of reproducing relevant temporal patterns, preserving market trends and dynamics. The generation of synthetic series through GANs is an efficient alternative for simulating financial data, showing potential for applications such as market behavior analysis and anomaly detection, with lower computational cost compared to more complex generative approaches.

Automated 3D radiology report generation often suffers from clinical hallucinations and a lack of the iterative verification found in human practice. While recent Vision-Language Models (VLMs) have advanced the field, they typically operate as monolithic "black-box" systems without the collaborative oversight characteristic of clinical workflows. To address these challenges, we propose MARCH (Multi-Agent Radiology Clinical Hierarchy), a multi-agent framework that emulates the professional hierarchy of radiology departments and assigns specialized roles to distinct agents. MARCH utilizes a Resident Agent for initial drafting with multi-scale CT feature extraction, multiple Fellow Agents for retrieval-augmented revision, and an Attending Agent that orchestrates an iterative, stance-based consensus discourse to resolve diagnostic discrepancies. On the RadGenome-ChestCT dataset, MARCH significantly outperforms state-of-the-art baselines in both clinical fidelity and linguistic accuracy. Our work demonstrates that modeling human-like organizational structures enhances the reliability of AI in high-stakes medical domains.

Adapter-based methods have become a cost-effective approach to continual learning (CL) for Large Language Models (LLMs), by sequentially learning a low-rank update matrix for each task. To mitigate catastrophic forgetting, state-of-the-art approaches impose constraints on new adapters with respect to the previous ones, by targeting either subspace or coordinate-wise interference. In this paper, we propose JumpLoRA, a novel framework to adaptively induce sparsity in the Low-Rank Adaptation (LoRA) blocks through the use of JumpReLU gating. The method achieves dynamic parameter isolation, which helps prevent task interference. We demonstrate that our method is highly modular and compatible with LoRA-based CL approaches. Specifically, it significantly boosts the performance of IncLoRA and outperforms the leading state-of-the-art CL method, ELLA.

Control systems are ubiquitous in modern technology, comprising an engineered plant to be kept within specific, often fine-tuned, limits, and a separate controller that ensures this is the case. While modern controllers often employ digital computers, other examples are purely mechanical, or even biological. It is an open question whether computation is happening within all controllers by virtue of them being part of a control system. Abstraction/ Representation theory (ART) has been developed to tackle just this question of whether a physical system is computing. Here, we demonstrate how to use ART to model control systems, and analyse them for computational properties. We determine that the plant of a control system is (a proxy for) the representational entity necessary in ART for the existence of any computation: the plant is the user of the controller. We consider specific systems: a digital thermostat, an electro-mechanical thermostat, the purely mechanical centrifugal governor, and an open-loop human-controlled heating system. We show that all these systems, and control systems in general, are performing some degree of computation. As an initial use of these results, we apply them to computationalism within cognitive theory: we show the governor is computing, so it cannot play its role of counter-example in the question of whether the brain is too.

Large language models (LLMs) increasingly rely on chain-of-thought (CoT) reasoning to solve complex tasks. Yet ensuring that the reasoning trace both contributes to and faithfully reflects the processes underlying the model's final answer, rather than merely accompanying it, remains challenging. We introduce AtManRL, a method that leverages differentiable attention manipulation to learn more faithful reasoning through reinforcement learning. By training an additive attention mask that identifies tokens in the CoT crucial for producing correct answers, we derive a saliency reward signal that encourages the model to generate reasoning traces that genuinely influence its final predictions. We integrate this saliency reward with outcome-based rewards within the GRPO framework to jointly optimize for correctness and interpretability. Experiments on GSM8K and MMLU with Llama-3.2-3B-Instruct demonstrate that our approach can identify influential reasoning tokens and enable training more transparent reasoning models.

This paper presents SWNet, a bimodal end-to-end cross-spectral network specifically engineered for the detection of camouflaged weeds in dense agricultural environments. Plant camouflage, characterized by homochromatic blending where invasive species mimic the phenotypic traits of primary crops, poses a significant challenge for traditional computer vision systems. To overcome these limitations, SWNet utilizes a Pyramid Vision Transformer v2 backbone to capture long-range dependencies and a Bimodal Gated Fusion Module to dynamically integrate Visible and Near-Infrared information. By leveraging the physiological differences in chlorophyll reflectance captured in the NIR spectrum, the proposed architecture effectively discriminates targets that are otherwise indistinguishable in the visible range. Furthermore, an Edge-Aware Refinement module is employed to produce sharper object boundaries and reduce structural ambiguity. Experimental results on the Weeds-Banana dataset indicate that SWNet outperforms ten state-of-the-art methods. The study demonstrates that the integration of cross-spectral data and boundary-guided refinement is essential for high segmentation accuracy in complex crop canopies. The code is available on GitHub: https://cod-espol.github.io/SWNet/

Recent works proposed test-time alignment methods that rely on a small aligned model as a proxy that guides the generation of a larger base (unaligned) model. The implicit reward approach skews the large model distribution, whereas the nudging approach defers the generation of the next token to the small aligned model when the large base one is unconfident about its outcome. In this work, we first show that both approaches can be reduced to sampling from similar graphical models, where they differ only in the definition of a rejection criterion (or distribution). Moreover, we argue that the confidence criterion is ill-motivated due to linguistic phenomena like ambiguous phrasing. We propose a novel rejection criterion based on a conservative confidence bet. Experimentally, our novel approach outperforms previous work on several datasets.

Accurate prediction of training time in distributed deep learning is crucial for resource allocation, cost estimation, and job scheduling. We observe that the floating-point precision setting is a key determinant of training time, leading to training time variations of ~2.4x over its minimum. However, existing studies on distributed training time prediction rely on static model computation graphs that do not capture precision variations, including mixed precision. According to our experiments, training time prediction without considering precision results in significant prediction errors - reaching up to 147.85% in mean absolute percentage error (MAPE). To address this issue, we propose a precision-aware distributed training time predictor that achieves robust accuracy across diverse precision settings, including mixed precision, with 9.8% MAPE.

We present a dataset and a model for sentiment analysis of German sign language (DGS) fairy tales. First, we perform sentiment analysis for three levels of valence (negative, neutral, positive) on German fairy tales text segments using four large language models (LLMs) and majority voting, reaching an inter-annotator agreement of 0.781 Krippendorff's alpha. Second, we extract face and body motion features from each corresponding DGS video segment using MediaPipe. Finally, we train an explainable model (based on XGBoost) to predict negative, neutral or positive sentiment from video features. Results show an average balanced accuracy of 0.631. A thorough analysis of the most important features reveal that, in addition to eyebrows and mouth motion on the face, also the motion of hips, elbows, and shoulders considerably contribute in the discrimination of the conveyed sentiment, indicating an equal importance of face and body for sentiment communication in sign language.

Firearm violence is a pressing public health issue, yet research into survivors' lived experiences remains underfunded and difficult to scale. Qualitative research, including in-depth interviews, is a valuable tool for understanding the personal and societal consequences of community firearm violence and designing effective interventions. However, manually analyzing these narratives through thematic analysis and inductive coding is time-consuming and labor-intensive. Recent advancements in large language models (LLMs) have opened the door to automating this process, though concerns remain about whether these models can accurately and ethically capture the experiences of vulnerable populations. In this study, we assess the use of open-source LLMs to inductively code interviews with 21 Black men who have survived community firearm violence. Our results demonstrate that while some configurations of LLMs can identify important codes, overall relevance remains low and is highly sensitive to data processing. Furthermore, LLM guardrails lead to substantial narrative erasure. These findings highlight both the potential and limitations of LLM-assisted qualitative coding and underscore the ethical challenges of applying AI in research involving marginalized communities.

Accurate molecular property prediction is central to drug discovery, catalysis, and process design, yet real-world applications are often limited by small datasets. Molecular foundation models provide a promising direction by learning transferable molecular representations; however, they typically involve task-specific fine-tuning, require machine learning expertise, and often fail to outperform classical baselines. Tabular foundation models (TFMs) offer a fundamentally different paradigm: they perform predictions through in-context learning, enabling inference without task-specific training. Here, we evaluate TFMs in the low- to medium-data regime across both standardized pharmaceutical benchmarks and chemical engineering datasets. We evaluate both frozen molecular foundation model representations, as well as classical descriptors and fingerprints. Across the benchmarks, the approach shows excellent predictive performance while reducing computational cost, compared to fine-tuning, with these advantages also transferring to practical engineering data settings. In particular, combining TFMs with CheMeleon embeddings yields up to 100\% win rates on 30 MoleculeACE tasks, while compact RDKit2d and Mordred descriptors provide strong descriptor-based alternatives. Molecular representation emerges as a key determinant in TFM performance, with molecular foundation model embeddings and 2D descriptor sets both providing substantial gains over classic molecular fingerprints on many tasks. These results suggest that in-context learning with TFMs provides a highly accurate and cost-efficient alternative for property prediction in practical applications.

Multivariate time series classification (MTSC) plays a crucial role in various domains, including biomedical signal analysis and motion monitoring. However, existing approaches, particularly deep learning models, often require high computational resources, making them unsuitable for real-time applications or deployment on low-cost hardware, such as IoT devices and wearable systems. In this paper, we propose the Univariate Channel Fusion (UCF) method to deal with MTSC efficiently. UCF transforms multivariate time series into a univariate representation through simple channel fusion strategies such as the mean, median, or dynamic time warping barycenter. This transformation enables the use of any classifier originally designed for univariate time series, providing a flexible and computationally lightweight alternative to complex models. We evaluate UCF in five case studies covering diverse application domains, including chemical monitoring, brain-computer interfaces, and human activity analysis. The results demonstrate that UCF often outperforms baseline methods and state-of-the-art algorithms tailored for MTSC, while achieving substantial gains in computational efficiency, being particularly effective in problems with high inter-channel correlation.

Practice and extensive exercises are essential in programming education. Intelligent tutoring systems (ITSs) are a viable option to provide individualized hints and advice to programming students even when human tutors are not available. However, prior ITS for programming rarely support the Python programming language, mostly focus on introductory programming, and rarely take recent developments in generative models into account. We aim to establish a novel ITS for Python programming that is highly adaptable, serves both as a teaching and research platform, provides interfaces to plug in hint mechanisms (e.g.\ via large language models), and works inside the particularly challenging regulatory environment of Germany, that is, conforming to the European data protection regulation, the European AI act, and ethical framework of the German Research Foundation. In this paper, we present the description of the current state of the ITS along with future development directions, as well as discuss the challenges and opportunities for improving the system.

We extracted the scholarly reasoning systems of two internationally prominent humanities and social science scholars from their published corpora alone, converted those systems into structured inference-time constraints for a large language model, and tested whether the resulting scholar-bots could perform core academic functions at expert-assessed quality. The distillation pipeline used an eight-layer extraction method and a nine-module skill architecture grounded in local, closed-corpus analysis. The scholar-bots were then deployed across doctoral supervision, peer review, lecturing and panel-style academic exchange. Expert assessment involved three senior academics producing reports and appointment-level syntheses. Across the preserved expert record, all review and supervision reports judged the outputs benchmark-attaining, appointment-level recommendations placed both bots at or above Senior Lecturer level in the Australian university system, and recovered panel scores placed Scholar A between 7.9 and 8.9/10 and Scholar B between 8.5 and 8.9/10 under multi-turn debate conditions. A research-degree-student survey showed high performance ratings across information reliability, theoretical depth and logical rigor, with pronounced ceiling effects on a 7-point scale, despite all participants already being frontier-model users. We term this the Relic condition: when publication systems make stable reasoning architectures legible, extractable and cheaply deployable, the public record of intellectual labor becomes raw material for its own functional replacement. Because the technical threshold for this transition is already crossed at modest engineering effort, we argue that the window for protective frameworks covering disclosure, consent, compensation and deployment restriction is the present, while deployment remains optional rather than infrastructural.

The paradigm shift towards local and on-device inference under stringent resource constraints is represented by the tiny machine learning (TinyML) domain. The primary goal of \gls{tml} is to integrate intelligence into tiny, low-cost devices under strict resource, energy, and latency constraints. However, the ultra-resource-constrained nature of these devices can lead to increased inference execution time, which can be detrimental in latency critical applications. At the same time, TinyML applications are often associated with sensitive data. As such, latency optimization approaches that rely on training samples are infeasible when such data is unavailable, proprietary, or sensitive, highlighting a pressing need for optimization approaches that do not require access to the training dataset and can be applied directly to pre-trained models. Replacing costly multiplications with more hardware-efficient operations, such as shifts and additions, has been proposed as an effective method for reducing inference latency. However, post-training power-of-two (Po2) approaches are scarce and, in many cases, lead to unacceptable accuracy loss. In this work, we propose a framework that applies approximate matrix decomposition to a given CNN in order to optimize hardware implementations subject to strict constraints and without any need of re-training or fine-tuning steps. The genetic algorithm-driven framework explores different matrix decompositions and resulting multiplier-less CNN accelerator designs for FPGA targets. A comprehensive evaluation of different TinyML benchmarks demonstrates our framework's efficacy in generating latency-optimized implementations that satisfy strict accuracy and resource constraints, achieving an average 33% latency improvement with an average accuracy loss of 1.3% compared to typical systolic array-based FPGA accelerators.

The decomposition of complex structures into simpler substructures is a powerful technique with a wide range of applications. We study the computation of decompositions in the context of programmable matter. The amoebot model is a well-established model for programmable matter, which places $n$ tiny robots called amoebots on the triangular grid. We consider the reconfigurable circuit extension of the geometric amoebot model, which allows amoebots to interconnect via so-called circuits. Amoebots can then instantaneously transmit simple beeps to all amoebots connected by the same circuit. Using reconfigurable circuits, previous papers have described a linear-time triangulation algorithm, and a logarithmic-time decomposition algorithm into so-called tunnel regions. Both algorithms only work on a restricted class of amoebot structures. In this paper, we define a decomposition into $O(|\mathcal H|)$ simple, geodesically convex regions for arbitrary amoebot structures, and show how it can compute such a decomposition in $O(\log n)$ rounds, where $|\mathcal H|$ denotes the number of holes in the amoebot structure. As a byproduct, we also improve the global maxima algorithm of Padalkin et al. (Nat. Comput., 2024) for special cases and with that also their spanning tree algorithm to $O(\log n)$ rounds w.h.p.

We study the sample complexity of learning an $ε$-optimal policy in the Stochastic Shortest Path (SSP) problem. We first derive sample complexity bounds when the learner has access to a generative model. We show that there exists a worst-case SSP instance with $S$ states, $A$ actions, minimum cost $c_{\min}$, and maximum expected cost of the optimal policy over all states $B_{\star}$, where any algorithm requires at least $Ω(SAB_{\star}^3/(c_{\min}ε^2))$ samples to return an $ε$-optimal policy with high probability. Surprisingly, this implies that whenever $c_{\min} = 0$ an SSP problem may not be learnable, thus revealing that learning in SSPs is strictly harder than in the finite-horizon and discounted settings. We complement this lower bound with an algorithm that matches it, up to logarithmic factors, in the general case, and an algorithm that matches it up to logarithmic factors even when $c_{\min} = 0$, but only under the condition that the optimal policy has a bounded hitting time to the goal state.

The Project of AI is a world-building endeavor, wherein those who fund and develop AI systems both operate through and seek to sustain networks of power and wealth. As they expand their access to resources and configure our sociotechnical conditions, they benefit from the ways in which a suite of decoys animate scholars, critics, policymakers, journalists, and the public into co-constructing industry-empowering AI futures. Regardless of who constructs or nurtures them, these decoys often create the illusion of accountability while both masking the emerging political economies that the Project of AI has set into motion, and also contributing to the network-making power that is at the heart of the Project's extraction and exploitation. Drawing on literature at the intersection of communication, science and technology studies, and economic sociology, we examine how the Project of AI is constructed. We then explore five decoys that seemingly critique - but in actuality co-constitute - AI's emergent power relations and material political economy. We argue that advancing meaningful fairness or accountability in AI requires: 1) recognizing when and how decoys serve as a distraction, and 2) grappling directly with the material political economy of the Project of AI. Doing so will enable us to attend to the networks of power that make 'AI' possible, spurring new visions for how to realize a more just technologically entangled world.

Lung cancer remains one of the leading causes of cancer-related mortality worldwide. Conventional computed tomography (CT) imaging, while essential for detection and staging, has limitations in distinguishing benign from malignant lesions and providing interpretable diagnostic insights. To address this challenge, this study proposes a dual-modal artificial intelligence framework that integrates CT radiology with hematoxylin and eosin (H&E) histopathology for lung cancer diagnosis and subtype classification. The system employs convolutional neural networks to extract radiologic and histopathologic features and incorporates clinical metadata to improve robustness. Predictions from both modalities are fused using a weighted decision-level integration mechanism to classify adenocarcinoma, squamous cell carcinoma, large cell carcinoma, small cell lung cancer, and normal tissue. Explainable AI techniques including Grad-CAM, Grad-CAM++, Integrated Gradients, Occlusion, Saliency Maps, and SmoothGrad are applied to provide visual interpretability. Experimental results show strong performance with accuracy up to 0.87, AUROC above 0.97, and macro F1-score of 0.88. Grad-CAM++ achieved the highest faithfulness and localization accuracy, demonstrating strong correspondence with expert-annotated tumor regions. These results indicate that multimodal fusion of radiology and histopathology can improve diagnostic performance while maintaining model transparency, suggesting potential for future clinical decision support systems in precision oncology.

Swarm protocols are a recently introduced formalism for specifying, implementing, and verifying peer-to-peer systems called swarms. A swarm consists of distributed agents called machines that communicate by asynchronous event propagation. Following a local-first model, each machine can progress without requiring continuous connectivity to other machines. Existing models of swarms are not compositional, making the modular development of large and complex swarm applications as well as the reuse of code difficult. We address these issues by presenting novel theory and techniques for the compositional specification, verification, and implementation of swarms. These results enable the correct compositional reuse of pre-existing swarm protocols and machine implementations. We implement these contributions in a companion software artifact which enables the automatic integration of independently designed and verified swarm components.

Probabilistic Synchronous Parallel (PSP) is a technique in distributed learning systems to reduce synchronization bottlenecks by sampling a subset of participating nodes per round. In Federated Learning (FL), where edge devices are often unreliable due to factors including mobility, power constraints, and user activity, PSP helps improve system throughput. However, PSP has a key limitation: it assumes device behavior is static and different devices are independent. This can lead to unfair distributed synchronization, due to highly available nodes dominating training while those that are often unavailable rarely participate and so their data may be missed. If both data distribution and node availability are simultaneously correlated with the device, then both PSP and standard FL algorithms will suffer from persistent under-representation of certain classes or groups resulting in inefficient or ineffective learning of certain features. We introduce Availability-Weighted PSP (AW-PSP), an extension to PSP that addresses the issue of co-correlation of unfair sampling and data availability by dynamically adjusting node sampling probabilities using real-time availability predictions, historical behavior, and failure correlation metrics. A Markov-based availability predictor distinguishes transient \emph{vs} chronic failures, while a Distributed Hash Table (DHT) layer decentralizes metadata, including latency, freshness, and utility scores. We implement AW-PSP and trace-driven evaluation shows that it improves robustness to both independent and correlated failures, increases label coverage, and reduces fairness variance compared to standard PSP. AW-PSP thus provides an availability-aware, and fairness-conscious node sampling protocol for FL deployments that will scale to large numbers of nodes even in heterogeneous and failure-prone environments.

The interconnection network is a key component of Supercomputers and Data centers, and its design must cope with the increasing communication demands of current applications and services; otherwise, it may become a system bottleneck. The most challenging network design issues are the topology, routing algorithm, flow control, and power efficiency. However, even the most efficient interconnection networks may suffer severe performance degradation due to congestion, especially under specific network traffic patterns generated by communication operations in high-performance computing~(HPC), deep learning training, or online data-intensive services. In this context, characterizing and modeling these communication operations and the network traffic patterns they generate is a fundamental challenge for studying their impact on network performance. This paper presents a methodology, based primarily on the VEF Traces framework, to characterize, model, and simulate the communication patterns of representative computing- and data-intensive applications. More precisely, we have extended the VEF traces framework with tools that enable us to characterize network congestion, either directly from VEF traces or via simulations. We have analyzed a set of VEF traces obtained from runs of NEST, GROMACS, LAMMPS, and PATMOS on several Supercomputers. In these studies, we identify potential congestion scenarios that arise in realistic network configurations when certain collective operations are performed.

We study the problem of learning in zero-sum matrix games with repeated play and bandit feedback. Specifically, we focus on developing uncoupled algorithms that guarantee, without communication between players, the convergence of the last-iterate to a Nash equilibrium. Although the non-bandit case has been studied extensively, this setting has only been explored recently, with a bound of $\mathcal{O}(T^{-1/8})$ on the exploitability gap. We show that, for uncoupled algorithms, guaranteeing convergence of the policy profiles to a Nash equilibrium is detrimental to the performance, with the best attainable rate being $Ω(T^{-1/4})$ in contrast to the usual $Ω(T^{-1/2})$ rate for convergence of the average iterates. We then propose two algorithms that achieve this optimal rate up to constant and logarithmic factors. The first algorithm leverages a straightforward trade-off between exploration and exploitation, while the second employs a regularization technique based on a two-step mirror descent approach.

One of the dominant paradigms in self-supervised learning (SSL), illustrated by MoCo or DINO, aims to produce robust representations by capturing features that are insensitive to certain image transformations such as illumination, or geometric changes. This strategy is appropriate when the objective is to recognize objects independently of their appearance. However, it becomes counterproductive as soon as appearance itself constitutes the discriminative signal. In weather analysis, for example, rain streaks, snow granularity, atmospheric scattering, as well as reflections and halos, are not noise: they carry the essential information. In critical applications such as autonomous driving, ignoring these cues is risky, since grip and visibility depend directly on ground conditions and atmospheric conditions. We introduce ST-STORM, a hybrid SSL framework that treats appearance (style) as a semantic modality to be disentangled from content. Our architecture explicitly separates two latent streams, regulated by gating mechanisms. The Content branch aims at a stable semantic representation through a JEPA scheme coupled with a contrastive objective, promoting invariance to appearance variations. In parallel, the Style branch is constrained to capture appearance signatures (textures, contrasts, scattering) through feature prediction and reconstruction under an adversarial constraint. We evaluate ST-STORM on several tasks, including object classification (ImageNet-1K), fine-grained weather characterization, and melanoma detection (ISIC 2024 Challenge). The results show that the Style branch effectively isolates complex appearance phenomena (F1=97% on Multi-Weather and F1=94% on ISIC 2024 with 10% labeled data), without degrading the semantic performance (F1=80% on ImageNet-1K) of the Content branch, and improves the preservation of critical appearance

Traffic forecasting is a challenging spatio-temporal modeling task and a critical component of urban transportation management. Current studies mainly focus on deterministic predictions, with limited considerations on the uncertainty and stochasticity in traffic dynamics. Therefore, this paper proposes an elegant yet universal approach that transforms existing models into probabilistic predictors by replacing only the final output layer with a novel Gaussian Mixture Model (GMM) layer. The modified model requires no changes to the training pipeline and can be trained using only the Negative Log-Likelihood (NLL) loss, without any auxiliary or regularization terms. Experiments on multiple traffic datasets show that our approach generalizes from classic to modern model architectures while preserving deterministic performance. Furthermore, we propose a systematic evaluation procedure based on cumulative distributions and confidence intervals, and demonstrate that our approach is considerably more accurate and informative than unimodal or deterministic baselines. Finally, a more detailed study on a real-world dense urban traffic network is presented to examine the impact of data quality on uncertainty quantification and to show the robustness of our approach under imperfect data conditions. Code available at https://github.com/Weijiang-Xiong/OpenSkyTraffic

Acute Myeloid Leukemia (AML) is one of the most life-threatening type of blood cancers, and its accurate classification is considered and remains a challenging task due to the visual similarity between various cell types. This study addresses the classification of the multiclasses of AML cells Utilizing YOLOv12 deep learning model. We applied two segmentation approaches based on cell and nucleus features, using Hue channel and Otsu thresholding techniques to preprocess the images prior to classification. Our experiments demonstrate that YOLOv12 with Otsu thresholding on cell-based segmentation achieved the highest level of validation and test accuracy, both reaching 99.3%.

We present Veritas-RPM, a provenance-guided multi-agent architecture comprising five processing layers: VeritasAgent (ground-truth assembly), SentinelLayer (anomaly detection), DirectorAgent (specialist routing), six domain Specialist Agents, and MetaSentinelAgent (conflict resolution and final decision). We construct a 98-case synthetic taxonomy of false-positive scenarios derived from documented RPM patterns. Synthetic patient epochs (n = 530) were generated directly from taxonomy parameters and processed through the pipeline. Ground-truth labels are known for all cases. Performance is reported as True Suppression Rate (TSR), False Escalation Rate (FER), and Indeterminate Rate (INDR).

Concept Bottleneck Models (CBMs) aim to improve interpretability in Deep Learning by structuring predictions through human-understandable concepts, but they provide no way to verify whether learned concepts align with the human's intended meaning, hurting interpretability. We introduce Prototype-Grounded Concept Models (PGCMs), which ground concepts in learned visual prototypes: image parts that serve as explicit evidence for the concepts. This grounding enables direct inspection of concept semantics and supports targeted human intervention at the prototype level to correct misalignments. Empirically, PGCMs match the predictive performance of state-of-the-art CBMs while substantially improving transparency, interpretability, and intervenability.

The quest for an algorithm that solves an $n\times n$ linear system in $O(n^2)$ time complexity, or $O(n^2 \text{poly}(1/ε))$ when solving up to $ε$ relative error, is a long-standing open problem in numerical linear algebra and theoretical computer science. There are two predominant paradigms for measuring relative error: forward error (i.e., distance from the output to the optimum solution) and backward error (i.e., distance to the nearest problem solved by the output). In most prior studies, convergence of iterative linear system solvers is measured via various notions of forward error, and as a result, depends heavily on the conditioning of the input. Yet, the numerical analysis literature has long advocated for backward error as the more practically relevant notion of approximation. In this work, we show that -- surprisingly -- the classical and simple Richardson iteration incurs at most $1/k$ (relative) backward error after $k$ iterations on any positive semidefinite (PSD) linear system, irrespective of its condition number. This universal convergence rate implies an $O(n^2/ε)$ complexity algorithm for solving a PSD linear system to $ε$ backward error, and we establish similar or better complexity when using a variety of Krylov solvers beyond Richardson. Then, by directly minimizing backward error over a Krylov subspace, we attain an even faster $O(1/k^2)$ universal rate, and we turn this into an efficient algorithm, MINBERR, with complexity $O(n^2/\sqrtε)$. We extend this approach via normal equations to solving general linear systems, for which we empirically observe $O(1/k)$ convergence. We report strong numerical performance of our algorithms on benchmark problems.

Adapting pre-trained vision-language models (VLMs) for robotic control requires injecting high-magnitude continuous gradients from a flow-matching action expert into a backbone trained exclusively with cross-entropy. This cross-modal gradient asymmetry - the spectral dimensionality mismatch between low-rank MSE regression gradients and the high-dimensional semantic manifold sculpted by CE pre-training, causes rapid, severe erosion of the VLM's visual-question-answering (VQA) capability. Industry-standard defences either sever the gradient pathway entirely via stop gradient, discarding the rich continuous supervision, or restrict parameter capacity through low-rank adapters (LoRA) that constrain the rank of updates but not their direction, and thus still overwrite the pre-trained manifold. We introduce AEGIS (Anchor-Enforced Gradient Isolation System): a buffer-free, layer-wise orthogonal gradient projection framework that enables direct continuous MSE learning while preserving the pre-trained VQA manifold - without any co-training data or replay buffer. AEGIS pre-computes a static Gaussian reference anchor from masked VQA forward passes across all transformer layers, then at each training step constructs a Wasserstein-2 transport penalty that generates an anchor restoration gradient. A sequential dual-backward decomposes the task and anchor gradients; for each transformer layer, AEGIS applies a single Gram-Schmidt orthogonal projection that bends the task gradient away from the destructive direction while preserving its constructive content. The projection sheds less than 1% of gradient energy on average, yet eliminates the cumulative activation drift that drives severe forgetting.

We study discrete k-clustering problems in general metric spaces that are constrained by a combination of two different fairness conditions within the demographic fairness model. Given a metric space (P,d), where every point in P is equipped with a protected attribute, and a number k, the goal is to partition P into k clusters with a designated center each, such that a center-based objective function is minimized and the attributes are fairly distributed with respect to the following two fairness concepts: 1) group fairness: We aim for clusters with balanced numbers of attributes by specifying lower and upper bounds for the desired attribute proportions. 2) diverse center selection: Clusters have natural representatives, i.e., their centers. We ask for a balanced set of representatives by specifying the desired number of centers to choose from each attribute. Dickerson, Esmaeili, Morgenstern and Zhang (2023) denote the combination of these two constraints as doubly constrained fair clustering. They present algorithms whose guarantees depend on the best known approximation factors for either of these problems. Currently, this implies an 8-approximation with a small additive violation on the group fairness constraint. For k-center, we improve this approximation factor to 4 with a small additive violation. This guarantee also depends on the currently best algorithm for DS-fair k-center given by Jones, Nguyen and Nguyen (2020). For k-median and k-means, we propose the first constant-factor approximation algorithms. Our algorithms transform a solution that satisfies diverse center selection into a doubly constrained fair clustering using an LP-based approach. Furthermore, our results are generalizable to other center-selection constraints, such as matroid k-clustering and knapsack constraints.

Multimodal Reasoning Models (MRMs) leveraging Chain-of-Thought (CoT) based thinking have revolutionized mathematical and logical problem-solving. However, we show that this paradigm struggles with generalized spatial intelligence. We perform a comprehensive evaluation of seventeen models across thirteen spatial benchmarks and identify a critical gap: CoT prompting consistently degrades performance in visual spatial reasoning. Furthermore, through a novel No-Image++ ablation, we demonstrate that MRMs and CoT prompted MLMs suffer from severe shortcut learning, and hallucinate visual details from textual priors even when the image is absent. These findings challenge the efficacy of text-only CoT for spatial tasks and underscore the need for vision-centric reasoning paradigms.

The rapid proliferation of Large Language Models (LLMs) in software development has made distinguishing AI-generated code from human-written code a critical challenge with implications for academic integrity, code quality assurance, and software security. We present LLMSniffer, a detection framework that fine-tunes GraphCodeBERT using a two-stage supervised contrastive learning pipeline augmented with comment removal preprocessing and an MLP classifier. Evaluated on two benchmark datasets - GPTSniffer and Whodunit - LLMSniffer achieves substantial improvements over prior baselines: accuracy increases from 70% to 78% on GPTSniffer (F1: 68% to 78%) and from 91% to 94.65% on Whodunit (F1: 91% to 94.64%). t-SNE visualizations confirm that contrastive fine-tuning yields well-separated, compact embeddings. We release our model checkpoints, datasets, codes and a live interactive demo to facilitate further research.

Text-based speech editing aims to modify specific segments while preserving speaker identity and acoustic context. Existing methods rely on task-specific training, which incurs high data costs and struggles with temporal fidelity in unedited regions. Meanwhile, adapting Text-to-Speech (TTS) models often faces a trade-off between editing quality and consistency. To address these issues, we propose AST, an Adaptive, Seamless, and Training-free precise speech editing framework. Leveraging a pre-trained autoregressive TTS model, AST introduces Latent Recomposition to selectively stitch preserved source segments with newly synthesized targets. Furthermore, AST extends this latent manipulation to enable precise style editing for specific speech segments. To prevent artifacts at these edit boundaries, the framework incorporates Adaptive Weak Fact Guidance (AWFG). AWFG dynamically modulates a mel-space guidance signal, enforcing structural constraints only where necessary without disrupting the generative manifold. To fill the gap of publicly accessible benchmarks, we introduce LibriSpeech-Edit, a new and larger speech editing dataset. As existing metrics poorly evaluate temporal consistency in unedited regions, we propose Word-level Dynamic Time Warping (WDTW). Extensive experiments demonstrate that AST resolves the controllability-quality trade-off without extra training. Compared to the previous most temporally consistent baseline, AST improves consistency while reducing Word Error Rate by nearly 70%. Moreover, applying AST to a foundation TTS model reduces WDTW by 27%, achieving state-of-the-art speaker preservation and temporal fidelity.

Multimodal large language models (MLLMs) have achieved impressive progress on vision language benchmarks, yet their capacity for visual cognitive and visuospatial reasoning remains less understood. We introduce "Mind's Eye", a multiple-choice benchmark of eight visuo-cognitive tasks inspired by classic human intelligence tests and organized under a novel "A-R-T" taxonomy: Abstraction, Relation, and Transformation. The tasks probe core processes of fluid intelligence such as pattern induction, analogical relation mapping, and mental transformation. We evaluate a diverse suite of closed-source and open-source MLLMs and compare their performance with human participants. Humans achieve 80% accuracy, while top performing MLLMs remain below 50%. Error analysis reveals failures in: (i) visual attention allocation, (ii) internal perceptual manipulation, and (iii) weak abstraction of underlying visual concepts. Our findings suggest that current MLLMs exhibit limited visuospatial reasoning capabilities, when compared with human participants, highlighting the need for more cognitively grounded evaluation frameworks.

With the continuous expansion of blockchain application scenarios, consortium chains have raised higher performance and security requirements for consensus mechanisms. Unlike public blockchains, consortium chains typically implement an admission mechanism that restricts participation to trusted entities, ensuring that most replicas are honest and the number of faulty nodes remains small under normal circumstances. In such settings, conventional Byzantine Fault Tolerant (BFT) protocols, which are designed for worst-case adversarial scenarios, incur excessive message exchanges and computational overhead, thereby limiting performance and scalability. To address this issue, this paper proposes T-RBFT, a two-layer consensus mechanism inspired by network sharding and enhanced by the trusted execution environment (TEE). In T-RBFT, consensus nodes are first dynamically grouped based on their runtime characteristics. Then, inter-group consensus is achieved through a TEE-assisted BFT protocol, while each group internally reaches agreement using an improved Raft-based mechanism. Experimental evaluation shows that T-RBFT reduces communication overhead and latency, and achieves higher throughput compared to existing two-layer consensus protocols, providing a scalable and communication-efficient consensus protocol for permissioned blockchain networks.

We introduce the Tan-HWG framework (Hebbian-Wasserstein-Geometry), a geometric theory of Hebbian plasticity in which memory states are modeled as probability measures evolving through Wasserstein minimizing movements. Hebbian learning rules are formalized as Hebbian energies satisfying a sequential stability condition, ensuring well-posed fiberwise JKO updates, optimal-transport realizations, and an energy descent inequality. This variational structure induces a fundamental separation between internal and observable dynamics. Internal memory states evolve along Wasserstein geodesics in a latent curved space, while observable quantities, such as effective synaptic weights, arise through geometric projection maps into external spaces. Simplicial projections recover classical affine schemes (including exponential moving averages and mirror descent), while revealing synaptic competition and pruning as geometric consequences of mass redistribution. Hilbertian projections provide a geometric account of phase alignment and multi-scale coherence. Classical neural networks appear as flat projections of this curved dynamics, while the framework naturally accommodates richer distributional representations, including structural weights and embedding memories, and their spectral extensions in complex internal spaces. Under mild Lipschitz regularity assumptions, including a quasi-stationary "sleep-mode" regime, we establish the existence of continuous-time limit curves. This yields a variational formulation of memory consolidation as a perturbed Wasserstein gradient flow. The framework thus provides a unified geometric foundation for synaptic plasticity, representation dynamics, and context-dependent computation.

The ambulance service is the main transport for diseased or injured people which suffers the same acceleration forces as regular vehicles. These accelerations, caused by the movement of the vehicle, impact the performance of tasks executed by sanitary personnel, which can affect patient survival or recovery time. In this paper, we have trained, validated, and tested a system to assess driving in ambulance services. The proposed system is composed of a sensor node which measures the vehicle vibrations using an accelerometer. It also includes a GPS sensor, a battery, a display, and a speaker. When two possible routes reach the same destination point, the system compares the two routes based on previously classified data and calculates an index and a score. Thus, the index balances the possible routes in terms of time to reach the destination and the vibrations suffered in the patient cabin to recommend the route that minimises those vibrations. Three datasets are used to train, validate, and test the system. Based on an Artificial Neural network (ANN), the classification model is trained with tagged data classified as low, medium, and high vibrations, and 97% accuracy is achieved. Then, the obtained model is validated using data from three routes of another region. Finally, the system is tested in two new scenarios with two possible routes to reach the destination. The results indicate that the route with less vibration is preferred when there are low time differences (less than 6%) between the two possible routes. Nonetheless, with the current weighting factors, the shortest route is preferred when time differences between routes are higher than 20%, regardless of the higher vibrations in the shortest route.

High-performance computing (HPC) applications are increasingly executed in heterogeneous environments, introducing new challenges for programming and software portability. SYCL has emerged as a leading model designed to simplify heterogeneous programming and make it more accessible to developers. Intended as a single-source, cross-platform parallel programming framework, SYCL promises portability, productivity, and performance across a variety of architectures. However, these goals have not been consistently defined or realized, leaving developers with varying expectations. This paper addresses this gap by evaluating SYCL from the perspective of application developers. We analyze whether SYCL meets essential criteria for cross-platform development, including code portability, development productivity, and runtime efficiency. Our evaluation draws on benchmarks and illustrative examples and focuses on SYCL's memory management and parallelism abstractions. We provide detailed comparisons between Unified Shared Memory (USM) and buffer-accessor approaches, as well as between NDRange and hierarchical kernel models. In addition to presenting our own benchmark results on Intel platforms, we synthesize findings from recent studies across multiple SYCL implementations and compilers. Our results expose key limitations and inconsistencies in current SYCL implementations and offer insights into the steps needed to improve the framework's reliability and cross-platform usability.

While Large Language Models (LLMs) have achieved strong performance across many NLP tasks, their opaque internal mechanisms hinder trustworthiness and safe deployment. Existing surveys in explainable AI largely focus on post-hoc explanation methods that interpret trained models through external approximations. In contrast, intrinsic interpretability, which builds transparency directly into model architectures and computations, has recently emerged as a promising alternative. This paper presents a systematic review of the recent advances in intrinsic interpretability for LLMs, categorizing existing approaches into five design paradigms: functional transparency, concept alignment, representational decomposability, explicit modularization, and latent sparsity induction. We further discuss open challenges and outline future research directions in this emerging field. The paper list is available at: https://github.com/PKU-PILLAR-Group/Survey-Intrinsic-Interpretability-of-LLMs.

The widespread adoption of large language models (LLMs) has increased concerns about their robustness. Vulnerabilities in perturbations of tokenisation of the input indicate that models trained with a deterministic canonical tokenisation can be brittle to adversarial attacks. Recent studies suggest that stochastic tokenisation can deliver internal representations that are less sensitive to perturbations. In this paper, we analyse how stochastic tokenisations affect robustness to adversarial attacks and random perturbations. We systematically study this over a range of learning regimes (pre-training, supervised fine-tuning, and in-context learning), data sets, and model architectures. We show that pre-training and fine-tuning with uniformly sampled stochastic tokenisations improve robustness to random and adversarial perturbations. Evaluating on uniformly sampled non-canonical tokenisations reduces the accuracy of a canonically trained Llama-1b model by 29.8%. We find that training with stochastic tokenisation preserves accuracy without increasing inference cost.

Buildings account for approximately 40% of global energy consumption, and with the growing share of intermittent renewable energy sources, enabling demand-side flexibility, particularly in heating, ventilation and air conditioning systems, is essential for grid stability and energy efficiency. This paper presents a safe deep reinforcement learning-based control framework to optimize building space heating while enabling demand-side flexibility provision for power system operators. A deep deterministic policy gradient algorithm is used as the core deep reinforcement learning method, enabling the controller to learn an optimal heating strategy through interaction with the building thermal model while maintaining occupant comfort, minimizing energy cost, and providing flexibility. To address safety concerns with reinforcement learning, particularly regarding compliance with flexibility requests, we propose a real-time adaptive safety-filter to ensure that the system operates within predefined constraints during demand-side flexibility provision. The proposed real-time adaptive safety filter guarantees full compliance with flexibility requests from system operators and improves energy and cost efficiency -- achieving up to 50% savings compared to a rule-based controller -- while outperforming a standalone deep reinforcement learning-based controller in energy and cost metrics, with only a slight increase in comfort temperature violations.

Parallel reasoning enhances Large Reasoning Models (LRMs) but incurs prohibitive costs due to futile paths caused by early errors. To mitigate this, path pruning at the prefix level is essential, yet existing research remains fragmented without a standardized framework. In this work, we propose the first systematic taxonomy of path pruning, categorizing methods by their signal source (internal vs. external) and learnability (learnable vs. non-learnable). This classification reveals the unexplored potential of learnable internal methods, motivating our proposal of STOP (Super TOken for Pruning). Extensive evaluations across LRMs ranging from 1.5B to 20B parameters demonstrate that STOP achieves superior effectiveness and efficiency compared to existing baselines. Furthermore, we rigorously validate the scalability of STOP under varying compute budgets - for instance, boosting GPT-OSS-20B accuracy on AIME25 from 84% to nearly 90% under fixed compute budgets. Finally, we distill our findings into formalized empirical guidelines to facilitate optimal real-world deployment. Code, data and models are available at https://bijiaxihh.github.io/STOP

Post-trained language models produce less varied outputs than their base counterparts. This output diversity collapse undermines inference-time scaling methods that rely on varied samples, and risks homogenizing model outputs on creative and value-laden tasks. Prior work attributes collapse to specific post-training methods, without separating the role of training data composition from the method, or the generation format from the model weights. We trace output diversity through three parallel post-training lineages of Olmo 3, Think (chain-of-thought distillation), Instruct (broad multi-source data), and RL-Zero, across 15 tasks and four text diversity metrics. We find that the location of collapse co-varies with data composition: the Think lineage loses most semantic diversity at supervised fine-tuning, and the effect of DPO is larger in Instruct than in Think. Suppressing chain-of-thought reasoning at inference in Think models drops accuracy on hard tasks, yet leaves answer-level diversity unchanged, showing that the collapse is embedded in the model weights by training data, not imposed by the generation format. Decomposing diversity loss on six verifiable tasks into a quality-control component (removal of incorrect outputs) and a residual component (genuine narrowing among correct outputs) reveals that the split is task-dependent, and Think models retain more correct-answer diversity than Instruct despite collapsing more in aggregate. Our results indicate that diversity collapse is determined during training by data composition and cannot be addressed at inference time alone.

Vision Language Models (VLMs) have been applied to several specific domains and have shown strong problem-solving capabilities. However, astronomical imaging, a quite complex problem involving multidisciplinary knowledge and several subtasks, has not been adequately studied. Due to the complexity of the astronomical imaging process, both world-class astronomical organizations, such as NASA, and expert enthusiasts devote a great deal of time and effort. This is because the processes in astronomical imaging have complex underlying correlations that significantly influence one another, making the quality diagnosis and error localization of astronomical images challenging. To address this problem, we propose AstroVLM, a collaborative multi-agent system for diagnosing the quality of astronomical images. Experiment results show that AstroVLM outperforms all baselines on real-world astronomical imaging quality diagnosis tasks, providing a reference for language models to handle complicated multi-process tasks.

As Large Language Models (LLMs) transition from text processors to autonomous agents, evaluating their social reasoning in embodied multi-agent settings becomes critical. We introduce SocialGrid, an embodied multi-agent environment inspired by Among Us that evaluates LLM agents on planning, task execution, and social reasoning. Our evaluations reveal that even the strongest open model (GPT-OSS-120B) achieves below 60% accuracy in task completion and planning, with agents getting stuck in repetitive behaviors or failing to navigate basic obstacles. Since poor navigation confounds evaluation of social intelligence, SocialGrid offers an optional Planning Oracle to isolate social reasoning from planning deficits. While planning assistance improves task completion, social reasoning remains a bottleneck: agents fail to detect deception at near-random chance regardless of scale, relying on shallow heuristics rather than accumulating behavioral evidence. SocialGrid provides automatic failure analysis and fine-grained metrics, enabling developers to diagnose and improve their agents. We also establish a competitive leaderboard using Elo ratings from adversarial league play.

Code localization is a cornerstone of autonomous software engineering. Recent advancements have achieved impressive performance on real-world issue benchmarks. However, we identify a critical yet overlooked bias: these benchmarks are saturated with keyword references (e.g. file paths, function names), encouraging models to rely on superficial lexical matching rather than genuine structural reasoning. We term this phenomenon the Keyword Shortcut. To address this, we formalize the challenge of Keyword-Agnostic Logical Code Localization (KA-LCL) and introduce KA-LogicQuery, a diagnostic benchmark requiring structural reasoning without any naming hints. Our evaluation reveals a catastrophic performance drop of state-of-the-art approaches on KA-LogicQuery, exposing their lack of deterministic reasoning capabilities. We propose LogicLoc, a novel agentic framework that combines large language models with the rigorous logical reasoning of Datalog for precise localization. LogicLoc extracts program facts from the codebase and leverages an LLM to synthesize Datalog programs, with parser-gated validation and mutation-based intermediate-rule diagnostic feedback to ensure correctness and efficiency. The validated programs are executed by a high-performance inference engine, enabling accurate and verifiable localization in a fully automated, closed-loop workflow. Experimental results demonstrate that LogicLoc significantly outperforms SOTA methods on KA-LogicQuery while maintaining competitive performance on popular issue-driven benchmarks. Notably, LogicLoc attains superior performance with significantly lower token consumption and faster execution by offloading structural traversal to a deterministic engine, reducing the overhead of iterative LLM inference.

Interpretable machine learning techniques are becoming essential tools for extracting physical insights from complex quantum data. We build on recent advances in variational autoencoders to demonstrate that such models can learn physically meaningful and interpretable representations from a broad class of unlabeled quantum datasets. From raw measurement data alone, the learned representation reveals rich information about the underlying structure of quantum phase spaces. We further augment the learning pipeline with symbolic methods, enabling the discovery of compact analytical descriptors that serve as order parameters for the distinct regimes emerging in the learned representations. We demonstrate the framework on experimental Rydberg-atom snapshots, classical shadows of the cluster Ising model, and hybrid discrete-continuous fermionic data, revealing previously unreported phenomena such as a corner-ordering pattern in the Rydberg arrays. These results establish a general framework for the automated and interpretable discovery of physical laws from diverse quantum datasets. All methods are available through qdisc, an open-source Python library designed to make these tools accessible to the broader community.

Metacognition, the ability to monitor and regulate one's own reasoning, remains under-evaluated in AI benchmarking. We introduce MEDLEY-BENCH, a benchmark of behavioural metacognition that separates independent reasoning, private self-revision, and socially influenced revision under genuine inter-model disagreement. The benchmark evaluates 35 models from 12 families on 130 ambiguous instances across five domains and reports two complementary scores: the Medley Metacognition Score (MMS), a tier-based aggregate of reflective updating, social robustness, and epistemic articulation, and the Medley Ability Score (MAS), derived from four metacognitive sub-abilities. Results show a robust evaluation/control dissociation: evaluation ability increases with model size within families, whereas control does not. In a follow-up progressive adversarial analysis of 11 models, we observed two behavioural profiles, i.e., models that revise primarily in response to argument quality and models that track consensus statistics. Under within-model relative profiling (ipsative scoring), evaluation was the weakest relative ability in all 35 models, indicating a systematic knowing/doing gap. Smaller and cheaper models often matched or outperformed larger counterparts, suggesting that metacognitive competence is not simply a function of scale. These findings position MEDLEY-BENCH as a tool for measuring belief revision under social pressure and suggest that future training should reward calibrated, proportional updating rather than output quality alone.

This paper introduces a robust discrimination method for distinguishing real ship targets from corner-reflector-array jamming with frequency-agile radar. The key idea is to exploit the multidimensional micro-motion signatures that separate rigid ships from non-rigid decoys. From Range-Velocity maps we derive two new hand-crafted descriptors-mean weighted residual (MWR) and complementary contrast factor (CCF) and fuse them with deep features learned by a lightweight CNN. An XGBoost classifier then gives the final decision. Extensive simulations show that the hybrid feature set consistently outperforms state-of-the-art alternatives, confirming the superiority of the proposed approach.

Emerging agentic LLM workloads are driving rapidly growing demand on both memory capacity and bandwidth, with different phases of inference (e.g., prefill and decode) imposing distinct requirements. Industry is responding by composing heterogeneous accelerators into single interconnected systems, as exemplified by NVIDIA's Vera Rubin platform, where each device brings its own memory architecture. This heterogeneity is further compounded by a widening landscape of available memory technologies: high-density on-chip SRAM, HBM, LPDDR, GDDR, and emerging options such as high-bandwidth flash (HBF), each offering different capacity, bandwidth, and power trade-offs. Identifying the right memory architecture for next-generation inference accelerators requires navigating a vast and rapidly evolving design space, in which the interplay between workload characteristics, NPU design dimensions, and memory system design remains largely underexplored. To address this challenge, we present MemExplorer, a new memory system synthesizer for heterogeneous NPU systems. MemExplorer provides a unified abstraction for modeling diverse memory technologies across different hierarchy levels (e.g., on-chip and off-chip) and automatically determines an efficient heterogeneous memory system together with NPU design choices (e.g., matrix engine size) to balance throughput and power between prefilling and decoding devices in a multi-device NPU system. Experimental results show that, under the same power budget for agentic workloads, MemExplorer achieves up to 2.3x higher energy efficiency than the baseline NPU and 3.23x higher than H100 in the prefill-only setting. Under equivalent performance targets in the decode setting, it further delivers up to 1.93x and 2.72x higher power efficiency over the baseline NPU and H100, respectively.

Verifiers have been demonstrated to enhance LLM reasoning via test-time scaling (TTS). Yet, they face significant challenges in complex domains. Error propagation from incorrect intermediate reasoning can lead to false positives for seemingly plausible solutions, while lacking external grounding makes verifiers unreliable on computation or knowledge-intensive tasks. To address these challenges, we propose Agentic Verifier, a framework that transforms reward modeling into a multi-turn, tool-augmented deliberative process. We introduce complementary forward and backward agents: one traces solutions from premises to conclusions, while the other re-checks conclusions against their underlying premises. This bidirectional process enables a comprehensive, reliable, and interpretable assessment of solutions. To facilitate practical deployment, we propose AgentV-RL. Through proactive exploration and reinforcement learning, the verifier autonomously interleaves tool-use with internal reasoning. Extensive experiments show that Agentic Verifier yields consistent performance gains under both parallel and sequential TTS. Notably, our 4B variant surpasses state-of-the-art ORMs by 25.2%, positioning it as a promising paradigm for agentic reward modeling.

Few-shot Hierarchical Text Classification (few-shot HTC) is a challenging task that involves mapping texts to a predefined tree-structured label hierarchy under data-scarce conditions. While current approaches utilize structural constraints from the label hierarchy to maintain parent-child prediction consistency, they face a critical bottleneck, the difficulty in distinguishing semantically similar sibling classes due to insufficient domain knowledge. We introduce an innovative method named Sibling Contrastive Learning with Hierarchical Knowledge-aware Prompt Tuning for few-shot HTC tasks (SCHK-HTC). Our work enhances the model's perception of subtle differences between sibling classes at deeper levels, rather than just enforcing hierarchical rules. Specifically, we propose a novel framework featuring two core components: a hierarchical knowledge extraction module and a sibling contrastive learning mechanism. This design guides model to encode discriminative features at each hierarchy level, thus improving the separability of confusable classes. Our approach achieves superior performance across three benchmark datasets, surpassing existing state-of-the-art methods in most cases. Our code is available at https://github.com/happywinder/SCHK-HTC.

Spiking neural networks (SNNs) are rapidly gaining momentum as an alternative to conventional artificial neural networks in resource constrained edge systems. In this work, we continue a recent research line on recurrent SNNs where axonal delays are learned at runtime along with the other network parameters. The first proposed approach, dubbed DelRec, demonstrated the benefit of recurrent delay learning in SNNs. Here, we extend it by advocating the use of convolutional recurrent connections in conjunction with the DelRec delay learning mechanism. According to our tests on an audio classification task, this leads to a streamlined architecture with smaller memory footprint (around 99% savings in terms of number of recurrent parameters) and a much faster (52x) inference time, while retaining DelRec's accuracy. Our code is available at: https://github.com/luciozebendo/delrec_snn/tree/conv_delays

Multimodal Large Language Models (MLLMs) excel at recognizing individual visual elements and reasoning over simple linear diagrams. However, when faced with complex topological structures involving branching paths, converging flows, and cyclic dependencies, their reasoning capabilities degrade sharply, even on tasks as basic as counting endpoints. Existing benchmarks fail to probe this gap, focusing on semantic comprehension rather than structural reasoning. We introduce ReactBench, a benchmark that reveals fundamental limitations in structural reasoning through chemical reaction diagrams. These real-world scientific diagrams offer an ideal testbed because they naturally span diverse structures from linear chains to cyclic graphs, while requiring both precise local recognition and coherent global reasoning. Our benchmark comprises 1,618 expert-annotated QA pairs across four hierarchical task dimensions. Extensive evaluation across 17 MLLMs reveals a significant performance gap exceeding 30% between anchor-based tasks and holistic structural reasoning tasks. Controlled ablations confirm this bottleneck lies in reasoning, not perception. These findings expose a fundamental deficit in structural understanding and establish directions for advancing visual reasoning.

This paper traces a conceptual shift from understanding vulnerability as a static, essentialized property of data subjects to examining how it is actively enacted through data practices. Unlike reflexive ethical frameworks focused on missing or counter-data, we address the condition of abundance inherent to platformized life-a context where a near inexhaustible mass of data points already exists, shifting the ethical challenge to the researcher's choices in operating upon this existing mass. We argue that the ethical integrity of data science depends not just on who is studied, but on how technical pipelines transform "vulnerable" individuals into data subjects whose vulnerability can be further precarized. We develop this argument through an AI for Social Good (AI4SG) case: a journalist's request to use computer vision to quantify child presence in monetized YouTube 'family vlogs' for regulatory advocacy. This case reveals a "protection paradox": how data-driven efforts to protect vulnerable subjects can inadvertently impose new forms of computational exposure, reductionism, and extraction. Using this request as a point of departure, we perform a methodological deconstruction of the AI pipeline to show how granular technical decisions are ethically constitutive. We contribute a reflexive ethics protocol that translates these insights into a reflexive roadmap for research ethics surrounding platformized data subjects. Organized around four critical junctures-dataset design, operationalization, inference, and dissemination-the protocol identifies technical questions and ethical tensions where well-intentioned work can slide into renewed extraction or exposure. For every decision point, the protocol offers specific prompts to navigate four cross-cutting vulnerabilizing factors: exposure, monetization, narrative fixing, and algorithmic optimization. Rather than uncritically...

M-MIMO is one of the crucial technologies for increasing spectral and energy efficiency of wireless networks. Most of the current works assume that M-MIMO arrays are equipped with a linear front end. However, ongoing efforts to make wireless networks more energy-efficient push the hardware to the limits, where its nonlinear behavior appears. This is especially a common problem for the multicarrier systems, e.g., OFDM used in 4G, 5G, and possibly also in 6G, which is characterized by a high Peak-to-Average Power Ratio. While the impact of a nonlinear Power Amplifier (PA) on an OFDM signal is well characterized, it is a relatively new topic for the M-MIMO OFDM systems. Most of the recent works either neglect nonlinear effects or utilize simplified models proper for Rayleigh or LoS radio channel models. In this paper, we first theoretically characterize the nonlinear distortion in the M-MIMO system under commonly used radio channel models. Then, utilizing 3D-Ray Tracing (3D-RT) software, we demonstrate that these models are not very accurate. Instead, we propose two models: a statistical one and an ML-based one using 3D-RT results. The proposed statistical model utilizes the Generalized Extreme Value (GEV) distribution to model Signal to Distortion Ratio (SDR) for victim users, receiving nonlinear distortion, e.g., as interference from neighboring cells. The proposed ML model aims to predict SDR for a scheduled user (receiving nonlinear distortion along with the desired signal), based on the spatial characteristics of the radio channel and the operation point of each PA feeding at the M-MIMO antenna array. The predicted SDR can then be used to perform PA-aware per-user power allocation. The results show about 12% median gain in user throughput achieved by the proposed ML-based power allocation scheme over the state-of-the-art, fixed operating point scheme.

LLM-driven multi-agent frameworks address complex reasoning tasks through multi-role collaboration. However, existing approaches often suffer from reasoning instability, where individual agent errors are amplified through collaboration, undermining overall performance. Current research mainly focuses on enhancing high-capability agents or suppressing unreliable outputs to improve framework effectiveness, while systematic identification and reinforcement of performance-limiting agents receive less attention. To address this gap, we propose WORC, a \underline{w}eak-link \underline{o}ptimization framework for multi-agent \underline{r}easoning and \underline{c}ollaboration, grounded in the weak-link principle. WORC follows a two-stage workflow. In the weak agent localization stage, task features are constructed, and a meta-learning-based weight predictor trained on optimal configurations identified by swarm intelligence algorithms (SIAs) enables zero-shot mapping from these features to agent performance weights, where the agent with the lowest predicted weight is identified as the weak agent. In the weak-link optimization stage, an uncertainty-driven allocation strategy assigns additional reasoning budgets to weak agents, with lower predicted weights leading to larger repeated-sampling quotas to compensate for reliability deficiencies. Experimental results show that WORC achieves an average accuracy of 82.2\% on reasoning benchmarks while improving framework stability and cross-architecture generalization, suggesting that compensating for weak links, rather than reinforcing strengths alone, enhances the robustness of multi-agent systems.

Accurate automatic brain tumor segmentation in Low and Middle-Income (LMIC) countries is challenging due to the lack of defined national imaging protocols, diverse imaging data, extensive use of low-field Magnetic Resonance Imaging (MRI) scanners and limited health-care resources. As part of the Brain Tumor Segmentation (BraTS) Africa 2025 Challenge, we applied topology refinement to the state-of-the-art segmentation models like nnU-Net, MedNeXt, and a combination of both. Since the BraTS-Africa dataset has low MRI image quality, we incorporated the BraTS 2025 challenge data of pre-treatment adult glioma (Task 1) to pre-train the segmentation model and use it to fine-tune on the BraTS-Africa dataset. We added an extra topology refinement module to address the issue of deformation in prediction that arose due to topological error. With the introduction of this module, we achieved a better Normalized Surface Distance (NSD) of 0.810, 0.829, and 0.895 on Surrounding Non-Enhancing FLAIR Hyperintensity (SNFH) , Non-Enhancing Tumor Core (NETC) and Enhancing tumor (ET).

A lot of research relies on data analysis scripts to process, clean, and visualize data. However, recent studies show that these scripts are often hard to comprehend and maintain, hindering reproducibility and reuse, accompanied by a lack of tool support for handling such scripts. In this work, we focus on the R programming language, addressing this problem by presenting flowR as an extension for the common data analysis IDEs Positron and VS Code. Alongside a previously presented static backward program slicer, flowR provides an overview of data analysis scripts, interactive graph visualizations, linting, and inline value annotations to support data analysts. FlowR incrementally analyzes R projects by intertwining interprocedural data- and control-flow analyses to build a comprehensive dataflow graph, incorporating R's dynamic and explorative features. Additionally, flowR offers a plugin system and interfaces, allowing the integration of further analyses, such as new linting rules or custom visualizations. Requiring an average of 576ms to calculate the full dataflow graph of real-world projects, this enables near real-time feedback. The demonstration video is available at https://youtu.be/hJzr-r-NmMg . For the full source code and extensive documentation, refer to https://github.com/flowr-analysis/flowr . To try the docker image, use `docker run --rm -it eagleoutice/flowr`.

There is no consensus in the field of synthetic data on concise metrics for quality evaluations or benchmarks on large health datasets, such as historical epidemiological data. This study presents an evaluation of seven recent models from major machine learning families. The models were evaluated using four different datasets, each with a distinct scale. To ensure a fair comparison, we systematically tuned the hyperparameters of each model for each dataset. We propose a methodology for evaluating the fidelity of synthesized joint distributions, aligning metrics with visualization on a single plot. This method is applicable to any dataset and is complemented by a domain-specific analysis of the German Cancer Registries' epidemiological dataset. The analysis reveals the challenges models face in strictly adhering to the medical domain. We hope this approach will serve as a foundational framework for guiding the selection of synthesizers and remain accessible to all stakeholders involved in releasing synthetic datasets.

Multi-objective Bayesian optimization (MOBO) provides a principled framework for optimizing expensive black-box functions with multiple objectives. However, existing MOBO methods often struggle with coverage, scalability with respect to the number of objectives, and integrating constraints and preferences. In this work, we propose \textit{STAGE-BO, Sequential Targeting Adaptive Gap-Filling $\varepsilon$-Constraint Bayesian Optimization}, that explicitly targets under-explored regions of the Pareto front. By analyzing the coverage of the approximate Pareto front, our method identifies the largest geometric gaps. These gaps are then used as constraints, which transforms the problem into a sequence of inequality-constrained subproblems, efficiently solved via constrained expected improvement acquisition. Our approach provides a uniform Pareto coverage without hypervolume computation and naturally applies to constrained and preference-based settings. Experiments on synthetic and real-world benchmarks demonstrate superior coverage and competitive hypervolume performance against state-of-the-art baselines.

Despite the considerable promise of Retrieval-Augmented Generation (RAG), many real-world use cases may create privacy concerns, where the purported utility of RAG-enabled insights comes at the risk of exposing private information to either the LLM or the end user requesting the response. As a potential mitigation, using anonymization techniques to remove personally identifiable information (PII) and other sensitive markers in the underlying data represents a practical and sensible course of action for RAG administrators. Despite a wealth of literature on the topic, no works consider the placement of anonymization along the RAG pipeline, i.e., asking the question, where should anonymization happen? In this case study, we systematically and empirically measure the impact of anonymization at two important points along the RAG pipeline: the dataset and generated answer. We show that differences in privacy-utility trade-offs can be observed depending on where anonymization took place, demonstrating the significance of privacy risk mitigation placement in RAG.

Many small-scale software systems, that is, with limited codebase or binary size, are widely used in everyday tasks, yet their configurability remains largely unexplored. At the same time, studies on modern software systems show a trend toward increasing configurability, alongside growing interest in building immutable, specialized, and reproducible software. In this paper, we present the first empirical study on the extent of configurability in small-scale software systems. By analyzing 108 programs from GNU coreutils, we show that even small programs can exhibit significant compile-time and run-time variability, with up to 76 options per program. Then, there is a high correlation (0.78) between run-time variability and codebase size. Furthermore, an analysis of the 20 smallest programs across 85 releases reveals that variability tends to increase over time, primarily due to the added compile-time variability. This suggests that shifting options between run-time and compile-time, removing unnecessary run-time variability, or resolving compile-time variability early, can help reduce codebase complexity and size. We also introduce, for the first time, the concept of null-variable software system, one with no configurability beyond mandatory features. Our findings show that high configurability is not exclusive to large-scale systems and that reducing unnecessary variability can lead to lightweight, smaller, and more maintainable software. We hope this effort contributes to designing new software by understanding how to balance its configurability with codebase size.

Generative AI, particularly Large Language Models, increasingly integrates graph-based representations to enhance reasoning, retrieval, and structured decision-making. Despite rapid advances, there remains limited clarity regarding when, why, where, and what types of graph-LLM integrations are most appropriate across applications. This survey provides a concise, structured overview of the design choices underlying the integration of graphs with LLMs. We categorize existing methods based on their purpose (reasoning, retrieval, generation, recommendation), graph modality (knowledge graphs, scene graphs, interaction graphs, causal graphs, dependency graphs), and integration strategies (prompting, augmentation, training, or agent-based use). By mapping representative works across domains such as cybersecurity, healthcare, materials science, finance, robotics, and multimodal environments, we highlight the strengths, limitations, and best-fit scenarios for each technique. This survey aims to offer researchers a practical guide for selecting the most suitable graph-LLM approach depending on task requirements, data characteristics, and reasoning complexity.

Pancreatic ductal adenocarcinoma (PDAC) segmentation on contrast-enhanced CT is inherently ambiguous: inter-rater disagreement among experts reflects genuine uncertainty rather than annotation noise. Standard deep learning approaches assume a single ground truth, producing probabilistic outputs that can be poorly calibrated and difficult to interpret under such ambiguity. We present TwinTrack, a framework that addresses this gap through post-hoc calibration of ensemble segmentation probabilities to the empirical mean human response (MHR) -the fraction of expert annotators labeling a voxel as tumor. Calibrated probabilities are thus directly interpretable as the expected proportion of annotators assigning the tumor label, explicitly modeling inter-rater disagreement. The proposed post-hoc calibration procedure is simple and requires only a small multi-rater calibration set. It consistently improves calibration metrics over standard approaches when evaluated on the MICCAI 2025 CURVAS-PDACVI multi-rater benchmark.

Retrieval-Augmented Generation (RAG) is widely used to augment the input to Large Language Models (LLMs) with external information, such as recent or domain-specific knowledge. Nonetheless, current models still produce closed-domain hallucinations and generate content that is unsupported by the retrieved context. Current detection approaches typically treat hallucination as a post-hoc problem, relying on black-box consistency checks or probes over frozen internal representations. In this work, we demonstrate that hallucination detection based on internal state representation can also serve as a direct training signal. We introduce RAGognize, a dataset of naturally occurring closed-domain hallucinations with token-level annotations, and RAGognizer, a hallucination-aware fine-tuning approach that integrates a lightweight detection head into an LLM, allowing for the joint optimization of language modeling and hallucination detection. This joint objective forces the model to improve the separability of its internal states regarding hallucinations while simultaneously learning to generate well-formed and meaningful responses. Across multiple benchmarks, RAGognizer achieves state-of-the-art token-level hallucination detection while substantially reducing hallucination rates during generation, without degrading language quality or relevance.

Large Language Models (LLMs) such as LLaMA and DeepSeek, are built on transformer architectures, which have become a standard model for achieving state-of-the-art performance in natural language processing tasks. Recently, there has been growing interest in deploying LLMs on edge devices. Although smaller LLM models are being proposed, they often still contain billions of parameters. Since edge devices are limited in their resources this poses a significant challenge for edge deployment. Compute-in-memory (CIM) is a promising architecture that addresses this by reducing data movement through the integration of computational logic directly into memory. However, existing CIM architectures support only static Multiply-Accumulate (MAC) operations which limit their configurability in supporting nonlinear operations and various types of transformer models. This paper presents a fully digital standard-cell SRAM-based CIM architecture accelerator for self-attention, called CIMple, designed to overcome these limitations, inside transformer models. The key contributions of CIMple are: 1) A novel dual-banked CIM-based fully digital self-attention accelerator using 8-bit parallel weight feeding. 2) A look-up-table (LUT) based fixed-point implementation reducing latency with minimal accuracy degradation. 3) A performance evaluation of a 32kb CIM-based self-attention accelerator implemented in 28nm, which achieves 26.1 TOPS/W at 0.85V and 2.31 TOPS/mm$^2$ at 1.2V, both with INT8 precision.

Fixed Radius Near Neighbor (FRNN) search is an alternative to the widely used k Nearest Neighbors (kNN) search. Unlike kNN, FRNN determines a label or an estimate for a test sample based on all training samples within a predefined distance. While this approach is beneficial in certain scenarios, assuming a fixed maximum distance for all training samples can decrease the accuracy of the FRNN. Therefore, in this paper we propose the Adaptive Radius Near Neighbor (ARNN) and the Weighted ARNN (WARNN), which employ adaptive distances and in latter case weights. All three methods are compared to kNN and twelve of its variants for a regression problem, namely WiFi fingerprinting indoor positioning, using 22 different datasets to provide a comprehensive analysis. While the performances of the tested FRNN and ARNN versions were amongst the worse, three of the four best methods in the test were WARNN versions, indicating that using weights together with adaptive distances achieves performance comparable or even better than kNN variants.

Large Language Models (LLMs) are increasingly deployed to curate and rank human-created content, yet the nature and structure of their biases in these tasks remains poorly understood: which biases are robust across providers and platforms, and which can be mitigated through prompt design. We present a controlled simulation study mapping content selection biases across three major LLM providers (OpenAI, Anthropic, Google) on real social media datasets from Twitter/X, Bluesky, and Reddit, using six prompting strategies (\textit{general}, \textit{popular}, \textit{engaging}, \textit{informative}, \textit{controversial}, \textit{neutral}). Through 540,000 simulated top-10 selections from pools of 100 posts across 54 experimental conditions, we find that biases differ substantially in how structural and how prompt-sensitive they are. Polarization is amplified across all configurations, toxicity handling shows a strong inversion between engagement- and information-focused prompts, and sentiment biases are predominantly negative. Provider comparisons reveal distinct trade-offs: GPT-4o Mini shows the most consistent behavior across prompts; Claude and Gemini exhibit high adaptivity in toxicity handling; Gemini shows the strongest negative sentiment preference. On Twitter/X, where author demographics can be inferred from profile bios, political leaning bias is the clearest demographic signal: left-leaning authors are systematically over-represented despite right-leaning authors forming the pool plurality in the dataset, and this pattern largely persists across prompts.

The goal of multilingual speech technology is to facilitate seamless communication between individuals speaking different languages, creating the experience as though everyone were a multilingual speaker. To create this experience, speech technology needs to address several challenges: Handling mixed multilingual input, specific vocabulary, and code-switching. However, there is currently no dataset benchmarking this situation. We propose a new benchmark to evaluate current Automatic Speech Recognition (ASR) systems, whether they are able to handle these challenges. The benchmark consists of bilingual discussions on scientific papers between multiple speakers, each conversing in a different language. We provide a standard evaluation framework, beyond Word Error Rate (WER) enabling consistent comparison of ASR performance across languages. Experimental results demonstrate that the proposed dataset is still an open challenge for state-of-the-art ASR systems. The dataset is available in https://huggingface.co/datasets/goodpiku/muscat-eval \\ \newline \Keywords{multilingual, speech recognition, audio segmentation, speaker diarization}

Drawing on ideas from continuous integration, we present concepts of an automated benchmarking pipeline for high performance applications. Customization and collaboration have been key design goals owing to the requirements of research-software development as a continuous community effort. We have extended our previous conceptual work on systematic benchmarking workflows with the functionality of user-agnostic operations as well as continuous benchmarking. This fosters reproducibility and re-use of benchmarking results to ensure sustainable technological progress. We provide software-engineering solutions to keep pace with the rapid evolution of both large-scale models and high-performance computing systems with a view towards the scientific domains of neuroscience and artificial intelligence.

With the ever-increasing usage of serverless computing in both industry and academia, it is essential to understand the mechanisms that power the underlying platforms. As serverless is more than ten years old, there are different platforms with vastly different approaches. We show that, next to the traditional and popular platforms, a second generation of serverless platform has emerged. While first-generation platforms are based on containerized, centralized execution, the new generation leverages lightweight isolates and edge deployment. This evolution reduces warm request latency from approximately 40 ms to around 10 ms and reduces cold starts to an afterthought, but limits the execution environment. In this paper, we gather and analyze all publicly available information to provide detailed insights into the underlying architecture of seven platforms and then run a microbenchmark-based evaluation totaling more than 38 million function calls to gain a deeper understanding their performance.

For around a decade, non-symbolic methods have been the option of choice when explaining complex machine learning (ML) models. Unfortunately, such methods lack rigor and can mislead human decision-makers. In high-stakes uses of ML, the lack of rigor is especially problematic. One prime example of provable lack of rigor is the adoption of Shapley values in explainable artificial intelligence (XAI), with the tool SHAP being a ubiquitous example. This paper overviews the ongoing efforts towards using rigorous symbolic methods of XAI as an alternative to non-rigorous non-symbolic approaches, concretely for assigning relative feature importance.

We introduce JFinTEB, the first comprehensive benchmark specifically designed for evaluating Japanese financial text embeddings. Existing embedding benchmarks provide limited coverage of language-specific and domain-specific aspects found in Japanese financial texts. Our benchmark encompasses diverse task categories including retrieval and classification tasks that reflect realistic and well-defined financial text processing scenarios. The retrieval tasks leverage instruction-following datasets and financial text generation queries, while classification tasks cover sentiment analysis, document categorization, and domain-specific classification challenges derived from economic survey data. We conduct extensive evaluations across a wide range of embedding models, including Japanese-specific models of various sizes, multilingual models, and commercial embedding services. We publicly release JFinTEB datasets and evaluation framework at https://github.com/retarfi/JFinTEB to facilitate future research and provide a standardized evaluation protocol for the Japanese financial text mining community. This work addresses a critical gap in Japanese financial text processing resources and establishes a foundation for advancing domain-specific embedding research.

As LLM agents scale to long-horizon, multi-session deployments, efficiently managing accumulated experience becomes a critical bottleneck. Agent memory systems and agent skill discovery both address this challenge -- extracting reusable knowledge from interaction traces -- yet a citation analysis of 1,136 references across 22 primary papers reveals a cross-community citation rate below 1%. We propose the \emph{Experience Compression Spectrum}, a unifying framework that positions memory, skills, and rules as points along a single axis of increasing compression (5--20$\times$ for episodic memory, 50--500$\times$ for procedural skills, 1,000$\times$+ for declarative rules), directly reducing context consumption, retrieval latency, and compute overhead. Mapping 20+ systems onto this spectrum reveals that every system operates at a fixed, predetermined compression level -- none supports adaptive cross-level compression, a gap we term the \emph{missing diagonal}. We further show that specialization alone is insufficient -- both communities independently solve shared sub-problems without exchanging solutions -- that evaluation methods are tightly coupled to compression levels, that transferability increases with compression at the cost of specificity, and that knowledge lifecycle management remains largely neglected. We articulate open problems and design principles for scalable, full-spectrum agent learning systems.

Large language models (LLMs) are increasingly studied as repositories of linguistic knowledge. In this line of work, models are commonly evaluated both as generators of language and as judges of linguistic output, yet these two roles are rarely examined in direct relation to one another. As a result, it remains unclear whether success in one role aligns with success in the other. In this paper, we address this question for pragmatic competence by comparing LLMs' performance as pragmatic listeners, judging the appropriateness of linguistic outputs, and as pragmatic speakers, generating pragmatically appropriate language. We evaluate multiple open-weight and proprietary LLMs across three pragmatic settings. We find a robust asymmetry between pragmatic evaluation and pragmatic generation: many models perform substantially better as listeners than as speakers. Our results suggest that pragmatic judging and pragmatic generation are only weakly aligned in current LLMs, calling for more integrated evaluation practices.

Feature toggles enable gradual rollouts and experimentation in software systems, yet often persist beyond their intended lifecycle, accumulating as technical debt. Prior research has examined feature toggle interactions and complexity, but no longitudinal study has quantified how toggles evolve over time across different organizational contexts. We analyse over 4,000 toggle events in Kubernetes (10 MLoC, 8.5 years) and GitLab (5 MLoC, 5 years). We find that feature toggle removals lags behind additions in both systems (by roughly 35% and 13%, respectively), leading to growing toggle inventories. Their lifespan patterns also differ notably, with Kubernetes toggles lasting a median of 734 days versus 185 in GitLab. Then, some feature toggles (1.33% and 0.73%, respectively) exceed all previously observed removal durations, becoming de facto permanent. Building on these findings, we propose a benchmarking framework with five key metrics and their empirically derived threshold zones, enabling practitioners to assess and compare toggle management practices across projects. All scripts and data are publicly available.

The evaluation of visual editing models remains fragmented across methods and modalities. Existing benchmarks are often tailored to specific paradigms, making fair cross-paradigm comparisons difficult, while video editing lacks reliable evaluation benchmarks. Furthermore, common automatic metrics often misalign with human preference, yet directly deploying large multimodal models (MLLMs) as evaluators incurs prohibitive computational and financial costs. We present UniEditBench, a unified benchmark for image and video editing that supports reconstruction-based and instruction-driven methods under a shared protocol. UniEditBench includes a structured taxonomy of nine image operations (Add, Remove, Replace, Change, Stroke-based, Extract, Adjust, Count, Reorder) and eight video operations, with coverage of challenging compositional tasks such as counting and spatial reordering. To enable scalable evaluation, we distill a high-capacity MLLM judge (Qwen3-VL-235B-A22B Instruct) into lightweight 4B/8B evaluators that provide multi-dimensional scoring over structural fidelity, text alignment, background consistency, naturalness, and temporal-spatial consistency (for videos). Experiments show that the distilled evaluators maintain strong agreement with human judgments and substantially reduce deployment cost relative to the teacher model. UniEditBench provides a practical and reproducible protocol for benchmarking modern visual editing methods. Our benchmark and the associated reward models are publicly available at https://github.com/wesar1/UniEditBench.

Quantum software testing has attracted interest in recent years, prompting the development of various techniques to automate the testing of quantum software. These techniques generate test cases that must be assessed for their effectiveness in detecting faults. Such an assessment requires benchmarks of faulty programs. However, there is a lack of benchmarks containing faults. In this data showcase, we propose QMutBench, a dataset that contains over 700,000 quantum circuit mutants representing different faults. The dataset is accessible via an online interface with selection criteria, such as the original quantum circuit(s) from which mutants are generated, the desired survival rate of the selected mutants, and other mutation characteristics (e.g., the type of faulty quantum gate). QMutBench provides quantum software developers and testers with an accessible online dataset to obtain benchmarks of mutants necessary to assess either the quality of the test cases generated by their testing technique or to compare different testing techniques. It also enables the development of new mutation-guided quantum software testing techniques.

Large language models (LLMs) have advanced information extraction (IE) by enabling zero-shot and few-shot named entity recognition (NER), yet their generative outputs still show persistent and systematic errors. Despite progress through instruction fine-tuning, zero-shot NER still lags far behind supervised systems. These recurring errors mirror inconsistencies observed in early-stage human annotation processes that resolve disagreements through pilot annotation. Motivated by this analogy, we introduce DiZiNER (Disagreement-guided Instruction Refinement via Pilot Annotation Simulation for Zero-shot Named Entity Recognition), a framework that simulates the pilot annotation process, employing LLMs to act as both annotators and supervisors. Multiple heterogeneous LLMs annotate shared texts, and a supervisor model analyzes inter-model disagreements to refine task instructions. Across 18 benchmarks, DiZiNER achieves zero-shot SOTA results on 14 datasets, improving prior bests by +8.0 F1 and reducing the zero-shot to supervised gap by over +11 points. It also consistently outperforms its supervisor, GPT-5 mini, indicating that improvements stem from disagreement-guided instruction refinement rather than model capacity. Pairwise agreement between models shows a strong correlation with NER performance, further supporting this finding.

Forecasting has become a natural benchmark for reasoning under uncertainty. Yet existing evaluations of large language models remain limited to judgmental tasks in simple formats, such as binary or multiple-choice questions. In practice, however, forecasting spans a far broader scope. Across domains such as economics, public health, and social demographics, decisions hinge on numerical estimates over continuous quantities, a capability that current benchmarks do not capture. Evaluating such estimates requires a format that makes uncertainty explicit and testable. We propose prediction intervals as a natural and rigorous interface for this purpose. They demand scale awareness, internal consistency across confidence levels, and calibration over a continuum of outcomes, making them a more suitable evaluation format than point estimates for numerical forecasting. To assess this capability, we introduce a new benchmark QuantSightBench, and evaluate frontier models under multiple settings, assessing both empirical coverage and interval sharpness. Our results show that none of the 11 evaluated frontier and open-weight models achieves the 90\% coverage target, with the top performers Gemini 3.1 Pro (79.1\%), Grok 4 (76.4\%), and GPT-5.4 (75.3\%) all falling at least 10 percentage points short. Calibration degrades sharply at extreme magnitudes, revealing systematic overconfidence across all evaluated models.

Multimodal remote sensing data provide complementary information for semantic segmentation, but in real-world deployments, some modalities may be unavailable due to sensor failures, acquisition issues, or challenging atmospheric conditions. Existing multimodal segmentation models typically address missing modalities by learning a shared representation across inputs. However, this approach can introduce a trade-off by compromising modality-specific complementary information and reducing performance when all modalities are available. In this paper, we tackle this limitation with CBC-SLP, a multimodal semantic segmentation model designed to preserve both modality-invariant and modality-specific information. Inspired by the theoretical results on modality alignment, which state that perfectly aligned multimodal representations can lead to sub-optimal performance in downstream prediction tasks, we propose a novel structured latent projection approach as an architectural inductive bias. Rather than enforcing this strategy through a loss term, we incorporate it directly into the architecture. In particular, to use the complementary information effectively while maintaining robustness under random modality dropout, we structure the latent representations into shared and modality-specific components and adaptively transfer them to the decoder according to the random modality availability mask. Extensive experiments on three multimodal remote sensing image sets demonstrate that CBC-SLP consistently outperforms state-of-the-art multimodal models across full and missing modality scenarios. Besides, we empirically demonstrate that the proposed strategy can recover the complementary information that may not be preserved in a shared representation. The code is available at https://github.com/iremulku/Multispectral-Semantic-Segmentation-via-Structured-Latent-Projection-CBC-SLP-.

Differential privacy (DP) has a wide range of applications for protecting data privacy, but designing and verifying DP algorithms requires expert-level reasoning, creating a high barrier for non-expert practitioners. Prior works either rely on specialized verification languages that demand substantial domain expertise or remain semi-automated and require human-in-the-loop guidance. In this work, we investigate whether large language models (LLMs) can automate DP reasoning. We introduce DPrivBench, a benchmark in which each instance asks whether a function or algorithm satisfies a stated DP guarantee under specified assumptions. The benchmark is carefully designed to cover a broad range of DP topics, span diverse difficulty levels, and resist shortcut reasoning through trivial pattern matching. Experiments show that while the strongest models handle textbook mechanisms well, all models struggle with advanced algorithms, revealing substantial gaps in current DP reasoning capabilities. Through further analytic study and failure-mode analysis, we identify several promising directions for improving automated DP reasoning. Our benchmark provides a solid foundation for developing and evaluating such methods, and complements existing benchmarks for mathematical reasoning.

Machine unlearning has gained increasing attention in recent years, as a promising technique to selectively remove unwanted privacy or copyrighted information from Large Language Models that are trained on a massive scale of human data. However, the emergence of Large Reasoning Models (LRMs), which emphasize long chain-of-thought (CoT) reasoning to address complex questions, presents a dilemma to unlearning: existing methods either struggle to completely eliminate undesired knowledge from the CoT traces or degrade the reasoning performances due to the interference with the reasoning process. To this end, we introduce Counterfactual Unlearning through iterative Preference Optimization (CiPO), a novel framework that redefines unlearning as the targeted intervention of the CoT reasoning in LRMs. More specifically, given a desired unlearning target answer, CiPO instructs LRMs to generate a logically valid counterfactual reasoning trace for preference tuning. As the LRM adjusts to the counterfactual trace, CiPO iteratively updates the preference learning data to increase the discrepancy from the original model. This iterative loop ensures both desirable unlearning and smooth optimization, effectively mitigating the dilemma. Experiments on challenging benchmarks demonstrate that CiPO excels at unlearning, completely removing knowledge from both the intermediate CoT steps and the final answer, while preserving the reasoning abilities of LRMs.

Large language models (LLMs) have demonstrated impressive capabilities, yet their internal mechanisms for handling reasoning-intensive tasks remain underexplored. To advance the understanding of model-internal processing mechanisms, we present an investigation of how LLMs perform arithmetic operations by examining internal mechanisms during task execution. Using early decoding, we trace how next-token predictions are constructed across layers. Our experiments reveal that while the models recognize arithmetic tasks early, correct result generation occurs only in the final layers. Notably, models proficient in arithmetic exhibit a clear division of labor between attention and MLP modules, where attention propagates input information and MLP modules aggregate it. This division is absent in less proficient models. Furthermore, successful models appear to process more challenging arithmetic tasks functionally, suggesting reasoning capabilities beyond factual recall.

While large language models (LLMs) have achieved remarkable success in general language tasks, their performance on Chouxiang Language, a representative subcultural language in the Chinese internet context, remains largely unexplored. In this paper, we introduce Mouse, a specialized benchmark designed to evaluate the capabilities of LLMs on NLP tasks involving Chouxiang Language across six tasks. Experimental results show that, current state-of-the-art (SOTA) LLMs exhibit clear limitations on multiple tasks, while performing well on tasks that involve contextual semantic understanding. In addition, we further discuss the reasons behind the generally low performance of SOTA LLMs on Chouxiang Language, examine whether the LLM-as-a-judge approach adopted for translation tasks aligns with human judgments and values, and analyze the key factors that influence Chouxiang translation. Our study aims to promote further research in the NLP community on multicultural integration and the dynamics of evolving internet languages. Our code and data are publicly available.

Reinforcement learning for LLM agents is typically conducted on a static data distribution, which fails to adapt to the agent's evolving behavior and leads to poor coverage of complex environment interactions. To address these challenges, we propose CoEvolve, an agent-data mutual evolution framework that enables LLM agents to improve through closed-loop, interaction-driven training. Specifically, CoEvolve extracts feedback signals such as forgetting and uncertainty from rollout trajectories to identify failure-prone interaction patterns, and utilizes them to guide LLM-based task synthesis. The synthesized tasks are validated through environment interaction and utilized to update the data distribution, enabling joint adaptation of the agent and its data. Extensive experiments on AppWorld and BFCL across Qwen2.5-7B, Qwen3-4B, and Qwen3-30B-A3B demonstrate consistent and significant improvements over strong base models, yielding absolute gains of 19.43%, 15.58%, and 18.14%, respectively.

Most ATP benchmarks embed the final answer within the formal statement -- a convention we call "Easy Mode" -- a design that simplifies the task relative to what human competitors face and may lead to optimistic estimates of model capability. We call the stricter, more realistic setting "Hard Mode": the system must independently discover the answer before constructing a formal proof. To enable Hard Mode research, we make two contributions. First, we release MiniF2F-Hard and FIMO-Hard, expert-reannotated Hard Mode variants of two widely-used ATP benchmarks. Second, we introduce Discover And Prove (DAP), an agentic framework that uses LLM natural-language reasoning with explicit self-reflection to discover answers, then rewrites Hard Mode statements into Easy Mode ones for existing ATP provers. DAP sets the state of the art: on CombiBench it raises solved problems from 7 (previous SOTA, Pass@16) to 10; on PutnamBench it is the first system to formally prove 36 theorems in Hard Mode -- while simultaneously revealing that state-of-the-art LLMs exceed 80% answer accuracy on the same problems where formal provers manage under 10%, exposing a substantial gap that Hard Mode benchmarks are uniquely suited to measure.

Distribution shift severely degrades the performance of deep forecasting models. While this issue is well-studied for individual time series, it remains a significant challenge in the spatio-temporal domain. Effective solutions like instance normalization and its variants can mitigate temporal shifts by standardizing statistics. However, distribution shift on a graph is far more complex, involving not only the drift of individual node series but also heterogeneity across the spatial network where different nodes exhibit distinct statistical properties. To tackle this problem, we propose Reversible Residual Normalization (RRN), a novel framework that performs spatially-aware invertible transformations to address distribution shift in both spatial and temporal dimensions. Our approach integrates graph convolutional operations within invertible residual blocks, enabling adaptive normalization that respects the underlying graph structure while maintaining reversibility. By combining Center Normalization with spectral-constrained graph neural networks, our method captures and normalizes complex Spatio-Temporal relationships in a data-driven manner. The bidirectional nature of our framework allows models to learn in a normalized latent space and recover original distributional properties through inverse transformation, offering a robust and model-agnostic solution for forecasting on dynamic spatio-temporal systems.

Combinatorial black-box optimization in high-dimensional settings demands a careful trade-off between exploiting promising regions of the search space and preserving sufficient exploration to identify multiple optima. Although Estimation-of-Distribution Algorithms (EDAs) provide a powerful model-based framework, they often concentrate on a single region of interest, which may result in premature convergence when facing complex or multimodal objective landscapes. In this work, we incorporate the Stein operator to introduce a repulsive mechanism among particles in the parameter space, thereby encouraging the population to disperse and jointly explore several modes of the fitness landscape. Empirical evaluations across diverse benchmark problems show that the proposed method achieves performance competitive with, and in several cases superior to, leading state-of-the-art approaches, particularly on large-scale instances. These findings highlight the potential of Stein variational gradient descent as a promising direction for addressing large, computationally expensive, discrete black-box optimization problems.

Spatiotemporal modeling has evolved beyond simple time series analysis to become fundamental in structural time series analysis. While current research extensively employs graph neural networks (GNNs) for spatial feature extraction with notable success, these networks are limited to capturing only pairwise relationships, despite real-world networks containing richer topological relationships. Additionally, GNN-based models face computational challenges that scale with graph complexity, limiting their applicability to large networks. To address these limitations, we present Modern Structure-Aware Simplicial SpatioTemporal neural network (ModernSASST), the first approach to leverage simplicial complex structures for spatiotemporal modeling. Our method employs spatiotemporal random walks on high-dimensional simplicial complexes and integrates parallelizable Temporal Convolutional Networks to capture high-order topological structures while maintaining computational efficiency. Our source code is publicly available on GitHub\footnote{Code is available at: https://github.com/ComplexNetTSP/ST_RUM.

Reinforcement learning has become a powerful approach for enhancing large language model reasoning, but faces a fundamental dilemma: training on easy problems can cause overfitting and pass@k degradation, while training on hard problems often results in sparse rewards. Recent question augmentation methods address this by prepending partial solutions as hints. However, uniform hint provision may introduce redundant information while missing critical reasoning bottlenecks, and excessive hints can reduce reasoning diversity, causing pass@k degradation. We propose \textbf{PieceHint}, a hint injection framework that strategically identifies and provides critical reasoning steps during training. By scoring the importance of different reasoning steps, selectively allocating hints based on problem difficulty, and progressively withdrawing scaffolding, PieceHint enables models to transition from guided learning to independent reasoning. Experiments on six mathematical reasoning benchmarks show that our 1.5B model achieves comparable average performance to 32B baselines while preserving pass@k diversity across all $k$ values.

Conventional information retrieval is concerned with identifying the relevance of texts for a given query. Yet, the conventional definition of relevance is dominated by aspects of similarity in texts, leaving unobserved whether the text is truly useful for addressing the query. For instance, when answering whether Paris is larger than Berlin, texts about Paris being in France are relevant (lexical/semantic similarity), but not useful. In this paper, we introduce UsefulBench, a domain-specific dataset curated by three professional analysts labeling whether a text is connected to a query (relevance) or holds practical value in responding to it (usefulness). We show that classic similarity-based information retrieval aligns more strongly with relevance. While LLM-based systems can counteract this bias, we find that domain-specific problems require a high degree of expertise, which current LLMs do not fully incorporate. We explore approaches to (partially) overcome this challenge. However, UsefulBench presents a dataset challenge for targeted information retrieval systems.

Automated classification of electrocardiogram (ECG) signals is a useful tool for diagnosing and monitoring cardiovascular diseases. This study compares three traditional machine learning algorithms (Decision Tree Classifier, Random Forest Classifier, and Logistic Regression) and three deep learning models (Simple Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and Complex CNN (ECGLens)) for the classification of ECG signals from the PTB-XL dataset, which contains 12-lead recordings from normal patients and patients with various cardiac conditions. The DL models were trained on raw ECG signals, allowing them to automatically extract discriminative features. Data augmentation using the Stationary Wavelet Transform (SWT) was applied to enhance model performance, increase the diversity of training samples, and preserve the essential characteristics of the ECG signals. The models were evaluated using multiple metrics, including accuracy, precision, recall, F1-score, and ROC-AUC. The ECG-Lens model achieved the highest performance, with 80% classification accuracy and a 90% ROC-AUC. These findings demonstrate that deep learning architectures, particularly complex CNNs substantially outperform traditional ML methods on raw 12-lead ECG data, and provide a practical benchmark for selecting automated ECG classification models and identifying directions for condition-specific model development.

Universal Machine Learning Interatomic Potentials (uMLIPs), pre-trained on massively diverse datasets encompassing inorganic materials and organic molecules across the entire periodic table, serve as foundational models for quantum-accurate physical simulations. However, uMLIP training requires second-order derivatives, which lack corresponding parallel training frameworks; moreover, scaling to the billion-parameter regime causes explosive growth in computation and communication overhead, making its training a tremendous challenge. We introduce MatRIS-MoE, a billion-parameter Mixture-of-Experts model built upon invariant architecture, and {Janus}, a pioneering high-dimensional distributed training framework for uMLIPs with hardware-aware optimizations. Deployed across two Exascale supercomputers, our code attains a peak performance of 1.2/1.0 EFLOPS (24\%/{35.5\%} of theoretical peak) in single precision at over 90\% parallel efficiency, compressing the training of billion-parameter uMLIPs from weeks to hours. This work establishes a new high-water mark for AI-for-Science (AI4S) foundation models at Exascale and provides essential infrastructure for rapid scientific discovery.

Static Random Access Memory (SRAM) Physically Unclonable Functions (PUFs) make use of intrinsic manufacturing variations in memory cells to derive device-unique responses. Employing such hardware-rooted fingerprints for authentication, this work demonstrates a threshold-based authentication proof of concept for constrained Industrial Internet of Things (IIoT) devices. The proposed scheme can reliably cap the the post-authentication bit error rate (BER) below 1 %. Inherent SRAM PUF unreliability is addressed by a resource-efficient combination of Hamming code (HC) Error Correction (EC) and Temporal Majority Voting (TMV). Increasing HC redundancy or TMV count significantly reduces the BER, albeit with diminishing returns and increasingly prohibitive computational overhead. Furthermore, this work quantifies the threshold gap between strict reliability and security constraints. This gap is reframed as a design budget which enables the resource-aware calibration of the acceptance threshold, PUF response length, and stabilization technique, without violating designed-for error limits. Larger responses make reliability optimizations increasingly obsolete. This comparative analysis establishes a comprehensive design space for PUF EC, guiding future implementations in balancing EC quality against resource constraints such as computational demand, power consumption, and implementation complexity.

Spatial reasoning and visual grounding are core capabilities for vision-language models (VLMs), yet most medical VLMs produce predictions without transparent reasoning or spatial evidence. Existing benchmarks also evaluate VLMs on isolated 2D images, overlooking the volumetric nature of clinical imaging, where findings can span multiple frames or appear on only a few slices. We introduce Spatially Grounded MRI Visual Question Answering (SGMRI-VQA), a 41,307-pair benchmark for multi-frame, spatially grounded reasoning on volumetric MRI. Built from expert radiologist annotations in the fastMRI+ dataset across brain and knee studies, each QA pair includes a clinician-aligned chain-of-thought trace with frame-indexed bounding box coordinates. Tasks are organized hierarchically across detection, localization, counting/classification, and captioning, requiring models to jointly reason about what is present, where it is, and across which frames it extends. We benchmark 10 VLMs and show that supervised fine-tuning of Qwen3-VL-8B with bounding box supervision consistently improves grounding performance over strong zero-shot baselines, indicating that targeted spatial supervision is an effective path toward grounded clinical reasoning.

Learning robust robot policies in real-world environments requires diverse data augmentation, yet scaling real-world data collection is costly due to the need for acquiring physical assets and reconfiguring environments. Therefore, augmenting real-world scenes into simulation has become a practical augmentation for efficient learning and evaluation. We present a generative framework that establishes a generative real-to-sim mapping from real-world panoramas to high-fidelity simulation scenes, and further synthesize diverse cousin scenes via semantic and geometric editing. Combined with high-quality physics engines and realistic assets, the generated scenes support interactive manipulation tasks. Additionally, we incorporate multi-room stitching to construct consistent large-scale environments for long-horizon navigation across complex layouts. Experiments demonstrate a strong sim-to-real correlation validating our platform's fidelity, and show that extensively scaling up data generation leads to significantly better generalization to unseen scene and object variations, demonstrating the effectiveness of Digital Cousins for generalizable robot learning and evaluation.

In this work, we present Qwen3.5-Omni, the latest advancement in the Qwen-Omni model family. Representing a significant evolution over its predecessor, Qwen3.5-Omni scales to hundreds of billions of parameters and supports a 256k context length. By leveraging a massive dataset comprising heterogeneous text-vision pairs and over 100 million hours of audio-visual content, the model demonstrates robust omni-modality capabilities. Qwen3.5-Omni-plus achieves SOTA results across 215 audio and audio-visual understanding, reasoning, and interaction subtasks and benchmarks, surpassing Gemini-3.1 Pro in key audio tasks and matching it in comprehensive audio-visual understanding. Architecturally, Qwen3.5-Omni employs a Hybrid Attention Mixture-of-Experts (MoE) framework for both Thinker and Talker, enabling efficient long-sequence inference. The model facilitates sophisticated interaction, supporting over 10 hours of audio understanding and 400 seconds of 720P video (at 1 FPS). To address the inherent instability and unnaturalness in streaming speech synthesis, often caused by encoding efficiency discrepancies between text and speech tokenizers, we introduce ARIA. ARIA dynamically aligns text and speech units, significantly enhancing the stability and prosody of conversational speech with minimal latency impact. Furthermore, Qwen3.5-Omni expands linguistic boundaries, supporting multilingual understanding and speech generation across 10 languages with human-like emotional nuance. Finally, Qwen3.5-Omni exhibits superior audio-visual grounding capabilities, generating script-level structured captions with precise temporal synchronization and automated scene segmentation. Remarkably, we observed the emergence of a new capability in omnimodal models: directly performing coding based on audio-visual instructions, which we call Audio-Visual Vibe Coding.

Retrieval-Augmented Generation (RAG) systems lose retrieval accuracy when similar documents coexist in the vector database, causing unnecessary information, hallucinations, and factual errors. To alleviate this issue, we propose CHOP, a framework that iteratively evaluates chunk relevance with Large Language Models (LLMs) and progressively reconstructs documents by determining their association with specific topics or query types. CHOP integrates two key components: the CNM-Extractor, which generates compact per-chunk signatures capturing categories, key nouns, and model names, and the Continuity Decision Module, which preserves contextual coherence by deciding whether consecutive chunks belong to the same document flow. By prefixing each chunk with context-aware metadata, CHOP reduces semantic conflicts among similar documents and enhances retriever discrimination. Experiments on benchmark datasets show that CHOP alleviates retrieval confusion and provides a scalable approach for building high-quality knowledge bases, achieving a Top-1 Hit Rate of 90.77% and notable gains in ranking quality metrics.

This study presents WriteFlow, an AI voice-based writing assistant designed to support reflective academic writing through goal-oriented interaction. Academic writing involves iterative reflection and evolving goal regulation, yet prior research and a formative study with 17 participants show that writers often struggle to articulate and manage changing goals. While commonly used AI writing tools emphasize efficiency, they offer limited support for metacognition and writer agency. WriteFlow frames AI interaction as a dialogic space for ongoing goal articulation, monitoring, and negotiation grounded in writers' intentions. Findings from a Wizard-of-Oz study with 12 expert users show that WriteFlow scaffolds metacognitive regulation and reflection-in-action by supporting iterative goal refinement, maintaining goal-text alignment during drafting, and prompting evaluation of goal fulfillment. We discuss design implications for AI writing systems that prioritize reflective dialogue, flexible goal structures, and multi-perspective feedback to support intentional and agentic writing.

Large Language Models (LLMs) have achieved remarkable success, underpinning diverse AI applications. However, they often suffer from performance degradation due to factors such as catastrophic forgetting during Supervised Fine-Tuning (SFT), quantization, and pruning. In this work, we introduce a performance recovery framework based on Self-Distillation Fine-Tuning (SDFT) that effectively restores model capabilities. Complementing this practical contribution, we provide a rigorous theoretical explanation for the underlying recovery mechanism. We posit that an LLM's generative capability fundamentally relies on the high-dimensional manifold constructed by its hidden layers. To investigate this, we employ Centered Kernel Alignment (CKA) to quantify the alignment between student and teacher activation trajectories, leveraging its invariance to orthogonal transformations and scaling. Our experiments demonstrate a strong correlation between performance recovery and manifold alignment, substantiating the claim that self-distillation effectively aligns the student's high-dimensional manifold with the optimal structure represented by the teacher. This study bridges the gap between practical recovery frameworks and geometric representation theory, offering new insights into the internal mechanisms of self-distillation.

The quantification of uncertainty in prediction models is crucial for reliable decision-making, yet remains a significant challenge. Interval time series forecasting offers a principled solution to this problem by providing prediction intervals (PIs), which indicates the probability that the true value falls within the predicted range. We consider a recently established point forecasts (PFs) method termed Learning-Based Convolution Nuclear Norm Minimization (LbCNNM), which directly generates multi-step ahead forecasts by leveraging the convolutional low-rankness property derived from training data. While theoretically complete and empirically effective, LbCNNM lacks inherent uncertainty estimation capabilities, a limitation shared by many advanced forecasting methods. To resolve the issue, we modify the well-known Quantile Regression (QR) and integrate it into LbCNNM, resulting in a novel interval forecasting method termed LbCNNM with Modified Quantile Regression (LbCNNM-MQR). In addition, we devise interval calibration techniques to further improve the accuracy of PIs. Extensive experiments on over 100,000 real-world time series demonstrate the superior performance of LbCNNM-MQR.

As Large Language Models (LLMs) receive increasing attention and are being deployed across various domains, their potential risks, including generating harmful or biased content, producing unsupported claims, and exhibiting vulnerabilities to adversarial attacks, have drawn significant attention. To enable quick and low-cost adaptation, training-free methods have recently emerged as cost-effective alternatives to post-training alignment techniques. Despite their promising results, these methods are evaluated inconsistently across the literature, cover limited dimensions of trustworthiness, and can introduce undesirable side effects, such as utility degradation and increased brittleness. To fully assess the impacts of these training-free methods, we take a step back and systematically re-evaluate the effectiveness of existing training-free methods against various trustworthy settings and their influence on utility, robustness, and computational overhead. We also categorize these methods into three levels (input, internal, and output) based on where they intervene in the model's information flow during inference. Using this taxonomy, we conduct a comprehensive analysis of various representative and effective methods from each level across different LLM families and sizes. Our analysis highlights several trade-offs and unresolved challenges in current approaches. We summarize key findings and limitations in the existing literature, and propose practical recommendations for balancing trustworthiness, utility, and robustness in LLMs without the need for additional training.

We introduce EVIL (\textbf{EV}olving \textbf{I}nterpretable algorithms with \textbf{L}LMs), an approach that uses LLM-guided evolutionary search to discover simple, interpretable algorithms for dynamical systems inference. Rather than training neural networks on large datasets, EVIL evolves pure Python/NumPy programs that perform zero-shot, in-context inference across datasets. We apply EVIL to three distinct tasks: next-event prediction in temporal point processes, rate matrix estimation for Markov jump processes, and time series imputation. In each case, a single evolved algorithm generalizes across all evaluation datasets without per-dataset training (analogous to an amortized inference model). To the best of our knowledge, this is the first work to show that LLM-guided program evolution can discover a single compact inference function for these dynamical-systems problems. Across the three domains, the discovered algorithms are often competitive with, and even outperform, state-of-the-art deep learning models while being orders of magnitudes faster, and remaining fully interpretable.

Urban bike-sharing systems require strategic station expansion to meet growing demand. Traditional allocation approaches rely on explicit demand modelling that may not capture the urban characteristics distinguishing successful stations. This study addresses the need to exploit patterns from existing stations to inform expansion decisions, particularly in data-constrained environments. We present a data-driven framework leveraging existing stations deemed desirable by operational metrics. A hybrid denoising autoencoder (HDAE) learns compressed latent representations from multi-source grid-level features (socio-demographic, built environment, and transport network), with a supervised classification head regularising the embedding space structure. Expansion candidates are selected via greedy allocation with spatial constraints based on latent-space similarity to existing stations. Evaluation on Trondheim's bike-sharing network demonstrates that HDAE embeddings yield more spatially coherent clusters and allocation patterns than raw features. Sensitivity analyses across similarity methods and distance metrics confirm robustness. A consensus-based procedure across multiple parametrisations distils 32 high-confidence extension zones where all parametrisations agree. The results demonstrate how representation learning captures complex patterns that raw features miss, enabling evidence-based expansion planning without explicit demand modelling. The consensus procedure strengthens recommendations by requiring agreement across parametrisations, while framework configurability allows planners to incorporate operational knowledge. The methodology generalises to any location-allocation problem where existing desirable instances inform the selection of new candidates.

Traffic simulations, essential for planning urban transit infrastructure interventions, require vehicle-category-specific origin-destination (OD) data. Existing data sources are imperfect: sparse tollbooth sensors provide accurate vehicle counts by category, while extensive mobility data from cellular network activity captures aggregated crowd movement, but lack modal disaggregation and have systematic biases. This study develops a machine learning framework to correct and disaggregate cellular network data using sparse tollbooth counts as ground truth. The model uses temporal and spatial features to learn the complex relationship between aggregated mobility data and vehicular data. The framework infers destinations from transit routes and implements routing logic to distribute corrected flows between OD pairs. This approach is applied to a bus depot expansion in Trondheim, Norway, generating hourly OD matrices by vehicle length category. The results show how limited but accurate sensor measurements can correct extensive but aggregated mobility data to produce grounded estimates of background vehicular traffic flows. These macro-scale estimates can be refined for micro-scale analysis at desired locations. The framework provides a generalisable approach for generating origin-destination data from cellular network data. This enables downstream tasks, like detailed traffic simulations for infrastructure planning in data-scarce contexts, supporting urban planners in making informed decisions.

Machine learning models are increasingly deployed in real-world applications, but even aligned models such as Mistral and LLaVA still exhibit unsafe behaviors inherited from pre-training. Current alignment methods like SFT and RLHF primarily encourage models to generate preferred responses, but do not explicitly remove the unsafe subnetworks that trigger harmful outputs. In this work, we introduce a resource-efficient pruning framework that directly identifies and removes parameters associated with unsafe behaviors while preserving model utility. Our method employs a gradient-free attribution mechanism, requiring only modest GPU resources, and generalizes across architectures and quantized variants. Empirical evaluations on ML models show substantial reductions in unsafe generations and improved robustness against jailbreak attacks, with minimal utility loss. From the perspective of the Lottery Ticket Hypothesis, our results suggest that ML models contain "unsafe tickets" responsible for harmful behaviors, and pruning reveals "safety tickets" that maintain performance while aligning outputs. This provides a lightweight, post-hoc alignment strategy suitable for deployment in resource-constrained settings.

Low Earth Orbit (LEO) mega-constellations extend the cloud-to-edge continuum into space, enabling satellite edge computing. However, Federated Learning (FL) in this environment is fundamentally energy-constrained due to dynamic inter-satellite connectivity, heterogeneous onboard computing hardware, and strict power budgets. We propose CroSatFL, a sustainable on-orbit hierarchical FL framework that reduces end-to-end energy across computation and communication while maintaining strong training performance under realistic LEO dynamics. CroSatFL keeps the ground station (GS) off the iterative loop by performing all local training and intermediate aggregations on orbit, requiring only two GS communication phases: one for initialization and one for final model collection. This sharply reduces repeated use of bandwidth-limited and energy-expensive GS links and shifts iterative exchanges to laser inter-satellite links (LISLs). CroSatFL integrates three energy-aware mechanisms: StarMask forms LISL-feasible clusters that align data volume with heterogeneous CPU/GPU capability, Skip-One mitigates transient stragglers by skipping at most one slow client per cluster to lower round energy and latency while preserving long-term fairness, and random-k cross-aggregation enables lightweight topology-aware cross-cluster mixing without extending round duration. Using an end-to-end energy model with a realistic Walker-Delta constellation, we show that CroSatFL reduces GS communication count by over two orders of magnitude and GS transmission energy by about 6x relative to GS-centric and on-orbit baselines, while achieving competitive accuracy and faster convergence.

Segmentation is a critical task in computational pathology, as it identifies areas affected by disease or abnormal growth and is essential for diagnosis and treatment. However, acquiring high-quality pixel-level supervised segmentation data requires significant workload demands from experienced pathologists, limiting the application of deep learning. To overcome this challenge, relaxing the label conditions to image-level classification labels allows for more data to be used and more scenarios to be enabled. One approach is to leverage Class Activation Map (CAM) to generate pseudo pixel-level annotations for semantic segmentation with only image-level labels. However, this method fails to thoroughly explore the essential characteristics of pathology images, thus identifying only small areas that are insufficient for pseudo masking. In this paper, we propose a novel shuffle-based feedback learning method inspired by curriculum learning to generate higher-quality pseudo-semantic segmentation masks. Specifically, we perform patch level shuffle of pathology images, with the model adaptively adjusting the shuffle strategy based on feedback from previous learning. Experimental results demonstrate that our proposed approach outperforms state-of-the-arts on three different datasets.

We present PIIBench, a unified benchmark corpus for Personally Identifiable Information (PII) detection in natural language text. Existing resources for PII detection are fragmented across domain-specific corpora with mutually incompatible annotation schemes, preventing systematic comparison of detection systems. We consolidate ten publicly available datasets spanning synthetic PII corpora, multilingual Named Entity Recognition (NER) benchmarks, and financial domain annotated text, yielding a corpus of 2,369,883 annotated sequences and 3.35 million entity mentions across 48 canonical PII entity types. We develop a principled normalization pipeline that maps 80+ source-specific label variants to a standardized BIO tagging scheme, applies frequency-based suppression of near absent entity types, and produces stratified 80/10/10 train/validation/test splits preserving source distribution. To establish baseline difficulty, we evaluate eight published systems spanning rule-based engines (Microsoft Presidio), general purpose NER models (spaCy, BERT-base NER, XLM-RoBERTa NER, SpanMarker mBERT, SpanMarker BERT), a PII-specific model (Piiranha DeBERTa), and a financial NER specialist (XtremeDistil FiNER). All systems achieve span-level F1 below 0.14, with the best system (Presidio, F1=0.1385) still producing zero recall on most entity types. These results directly quantify the domain-silo problem and demonstrate that PIIBench presents a substantially harder and more comprehensive evaluation challenge than any existing single source PII dataset. The dataset construction pipeline and benchmark evaluation code are publicly available at https://github.com/pritesh-2711/pii-bench.

Learning with large-scale datasets and information-critical applications, such as in High Energy Physics (HEP), demands highly complex, large-scale models that are both robust and accurate. To tackle this issue and cater to the learning requirements, we envision using a federated learning framework with a quantum-enhanced model. Specifically, we design a hybrid quantum-classical long-shot-term-memory model (QLSTM) for local training at distributed nodes. It combines the representative power of quantum models in understanding complex relationships within the feature space, and an LSTM-based model to learn necessary correlations across data points. Given the computing limitations and unprecedented cost of current stand-alone noisy-intermediate quantum (NISQ) devices, we propose to use a federated learning setup, where the learning load can be distributed to local servers as per design and data availability. We demonstrate the benefits of such a design on a classification task for the Supersymmetry(SUSY) dataset, having 5M rows. Our experiments indicate that the performance of this design is not only better that some of the existing work using variational quantum circuit (VQC) based quantum machine learning (QML) techniques, but is also comparable ($Δ\sim \pm 1\%$) to that of classical deep-learning benchmarks. An important observation from this study is that the designed framework has $<$300 parameters and only needs 20K data points to give a comparable performance. Which also turns out to be a 100$\times$ improvement than the compared baseline models. This shows an improved learning capability of the proposed framework with minimal data and resource requirements, due to the joint model with an LSTM based architecture and a quantum enhanced VQC.

Equipping Large Language Models (LLMs) with persistent memory enhances interaction continuity and personalization but introduces new safety risks. Specifically, contaminated or biased memory accumulation can trigger abnormal agent behaviors. Existing evaluation methods have not yet established a standardized framework for measuring memory misevolution. This phenomenon refers to the gradual behavioral drift resulting from repeated exposure to misleading information. To address this gap, we introduce MemEvoBench, the first benchmark evaluating long-horizon memory safety in LLM agents against adversarial memory injection, noisy tool outputs, and biased feedback. The framework consists of QA-style tasks across 7 domains and 36 risk types, complemented by workflow-style tasks adapted from 20 Agent-SafetyBench environments with noisy tool returns. Both settings employ mixed benign and misleading memory pools within multi-round interactions to simulate memory evolution. Experiments on representative models reveal substantial safety degradation under biased memory updates. Our analysis suggests that memory evolution is a significant contributor to these failures. Furthermore, static prompt-based defenses prove insufficient, underscoring the urgency of securing memory evolution in LLM agents.

We introduce a novel characterization of phase transitions based on hypothesis testing. In our formulation, a phase transition is defined as the breakdown of statistical indistinguishability under vanishing parameter perturbations in the thermodynamic limit. This perspective provides a general, order-parameter-free framework that does not rely on model-specific insights or learning procedures. We show that conventional approaches, such as those based on the Binder parameter, can be reinterpreted as special cases within this framework. As a concrete realization, we employ a distribution-free two-sample run test and demonstrate that the critical point of the two-dimensional Ising model is accurately identified without prior knowledge of the order parameter.

Retrieval-Augmented Generation (RAG) has emerged as a foundational paradigm for grounding large language models in external knowledge. While adaptive retrieval mechanisms have improved retrieval efficiency, existing approaches treat post-retrieval failure as a signal to retry rather than to diagnose -- leaving the structural causes of query-evidence misalignment unaddressed. We observe that a significant portion of persistent retrieval failures stem not from the absence of relevant evidence but from an alignment gap between the query and the evidence space. We propose Skill-RAG, a failure-aware RAG framework that couples a lightweight hidden-state prober with a prompt-based skill router. The prober gates retrieval at two pipeline stages; upon detecting a failure state, the skill router diagnoses the underlying cause and selects among four retrieval skills -- query rewriting, question decomposition, evidence focusing, and an exit skill for truly irreducible cases -- to correct misalignment before the next generation attempt. Experiments across multiple open-domain QA and complex reasoning benchmarks show that Skill-RAG substantially improves accuracy on hard cases persisting after multi-turn retrieval, with particularly strong gains on out-of-distribution datasets. Representation-space analyses further reveal that the proposed skills occupy structured, separable regions of the failure state space, supporting the view that query-evidence misalignment is a typed rather than monolithic phenomenon.

Spiking transformers achieve competitive accuracy with conventional transformers while offering $38$-$57\times$ energy efficiency on neuromorphic hardware, yet no theoretical framework guides their design. This paper establishes the first comprehensive expressivity theory for spiking self-attention. We prove that spiking attention with Leaky Integrate-and-Fire neurons is a universal approximator of continuous permutation-equivariant functions, providing explicit spike circuit constructions including a novel lateral inhibition network for softmax normalization with proven $O(1/\sqrt{T})$ convergence. We derive tight spike-count lower bounds via rate-distortion theory: $\varepsilon$-approximation requires $Ω(L_f^2 nd/\varepsilon^2)$ spikes, with rigorous information-theoretic derivation. Our key insight is input-dependent bounds using measured effective dimensions ($d_{\text{eff}}=47$--$89$ for CIFAR/ImageNet), explaining why $T=4$ timesteps suffice despite worst-case $T \geq 10{,}000$ predictions. We provide concrete design rules with calibrated constants ($C=2.3$, 95\% CI: $[1.9, 2.7]$). Experiments on Spikformer, QKFormer, and SpikingResformer across vision and language benchmarks validate predictions with $R^2=0.97$ ($p<0.001$). Our framework provides the first principled foundation for neuromorphic transformer design.

AI-driven methods have demonstrated considerable success in tackling the central challenge of accurately solving the Schrödinger equation for complex many-body systems. Among neural network quantum state (NNQS) approaches, the NNQS-SCI (Selected Configuration Interaction) method stands out as a state-of-the-art technique, recognized for its high accuracy and scalability. However, its application to larger systems is severely constrained by a hybrid CPU-GPU architecture. Specifically, centralized CPU-based global de-duplication creates a severe scalability barrier due to communication bottlenecks, while host-resident coupled-configuration generation induces prohibitive computational overheads. We introduce cuNNQS-SCI, a fully GPU-accelerated SCI framework designed to overcome these bottlenecks. cuNNQS-SCI first integrates a distributed, load-balanced global de-duplication algorithm to minimize redundancy and communication overhead at scale. To address compute limitations, it employs specialized, fine-grained CUDA kernels for exact coupled configuration generation. Finally, to break the single-GPU memory barrier exposed by this full acceleration, it incorporates a GPU memory-centric runtime featuring GPU-side pooling, streaming mini-batches, and overlapped offloading. This design enables much larger configuration spaces and shifts the bottleneck from host-side limitations back to on-device inference. Our evaluation demonstrates that cuNNQS-SCI fundamentally expands the scale of solvable problems. On an NVIDIA A100 cluster with 64 GPUs, cuNNQS-SCI achieves up to 2.32X end-to-end speedup over the highly-optimized NNQS-SCI baseline while preserving the same chemical accuracy. Furthermore, it demonstrates excellent distributed performance, maintaining over 90% parallel efficiency in strong scaling tests.

Early-exit neural networks enable adaptive computation by allowing confident predictions to exit at intermediate layers, achieving 2-8$\times$ inference speedup. Despite widespread deployment, their generalization properties lack theoretical understanding -- a gap explicitly identified in recent surveys. This paper establishes a unified PAC-Bayesian framework for adaptive-depth networks. (1) Novel Entropy-Based Bounds: We prove the first generalization bounds depending on exit-depth entropy $H(D)$ and expected depth $\mathbb{E}[D]$ rather than maximum depth $K$, with sample complexity $\mathcal{O}((\mathbb{E}[D] \cdot d + H(D))/ε^2)$. (2) Explicit Constructive Constants: Our analysis yields the leading coefficient $\sqrt{2\ln 2} \approx 1.177$ with complete derivation. (3) Provable Early-Exit Advantages: We establish sufficient conditions under which adaptive-depth networks strictly outperform fixed-depth counterparts. (4) Extension to Approximate Label Independence: We relax the label-independence assumption to $ε$-approximate policies, broadening applicability to learned routing. (5) Comprehensive Validation: Experiments across 6 architectures on 7 benchmarks demonstrate tightness ratios of 1.52-3.87$\times$ (all $p < 0.001$) versus $>$100$\times$ for classical bounds. Bound-guided threshold selection matches validation-tuned performance within 0.1-0.3%.

Large-scale Unmanned Aerial Vehicle (UAV) failures can split an unmanned aerial vehicle swarm network into disconnected sub-networks, making decentralized recovery both urgent and difficult. Centralized recovery methods depend on global topology information and become communication-heavy after severe fragmentation. Decentralized heuristics and multi-agent reinforcement learning methods are easier to deploy, but their performance often degrades when the swarm scale and damage severity vary. We present Physics-informed Graph Adversarial Imitation Learning algorithm (PhyGAIL) that adopts centralized training with decentralized execution. PhyGAIL builds bounded local interaction graphs from heterogeneous observations, and uses physics-informed graph neural network to encode directional local interactions as gated message passing with explicit attraction and repulsion. This gives the policy a physically grounded coordination bias while keeping local observations scale-invariant. It also uses scenario-adaptive imitation learning to improve training under fragmented topologies and variable-length recovery episodes. Our analysis establishes bounded local graph amplification, bounded interaction dynamics, and controlled variance of the terminal success signal. A policy trained on 20-UAV swarms transfers directly to swarms of up to 500 UAVs without fine-tuning, and achieves better performance across reconnection reliability, recovery speed, motion safety, and runtime efficiency than representative baselines.

Designing optimizers that remain effective under tight evaluation budgets is critical in expensive black-box settings such as cardiac digital twinning. We propose Frenetic Cat-inspired Particle Optimization (FCPO), a hybrid swarm method that couples particle swarm optimization-like dynamics with an explicit-state Markov switching controller to schedule exploration and refinement operators online. FCPO integrates (i) state-conditioned bounded motion, (ii) an elite-difference global jump operator to escape stagnation, (iii) eigen-space guided local refinement from elite covariance, and (iv) linear population size reduction to control late-stage computational cost. We benchmark FCPO on five representative functions from the Congress on Evolutionary Computation (CEC) 2022 suite (F1, F2, F3, F6 and F10) at dimensions D$\in${10,20} over 30 independent runs, comparing against PSO, CSO, CLPSO, SHADE, L-SHADE and CMA-ES. FCPO achieves the lowest mean runtime across the ten benchmark cases (average 0.183 s), about 2.3x faster than CMA-ES and 2.6x faster than L-SHADE in our Python implementation. On the multimodal composition function F10 at D=20, FCPO attains the best mean objective (9.625x 10^2 $\pm$ 1.275x 10^3) and remains faster than CMA-ES (0.602 s vs. 1.126 s mean runtime). On structured landscapes (F1--F3) and on the hybrid function (F6), CMA-ES remains the most accurate method, while FCPO substantially improves over classical swarms and maintains a favorable accuracy--runtime trade-off. Finally, in a ventricular activation digital twin calibration task, FCPO reaches the target electrocardiogram (ECG) fidelity (RMSE<0.1 mV) within ~ 40 iterations and produces physiologically plausible activation maps with robust convergence across repeated initializations, supporting its use as a practical optimizer for expensive inverse problems.

We introduce the first version of KWBench (Knowledge Work Bench), a benchmark for unprompted problem recognition in large language models: can an LLM identify a professional scenario before attempting to solve it. Existing frontier benchmarks have saturated, and most knowledge-work evaluations to date reduce to extraction or task completion against a specification. KWBench targets the step before that: recognizing the governing structure of the situation from raw inputs alone. The benchmark contains 223 tasks sourced from practitioners across acquisitions, contract negotiations, clinical pharmacy, organizational politics, fraud analysis, and incentive design. Each task encodes a formal game-theoretic pattern (principal-agent conflict, signaling, mechanism design failure, strategic omission, coalitional dynamics, strategic interdependence) and carries structured ground truth recording the expert reading of the situation and the anticipated failure modes. Models receive raw data and a task prompt with no indication of problem type. Scoring is a three-tier rubric gated by a mandatory conjunctive check. Mandatory criteria encode the predicted wrong paths. We evaluate 16 models. The best model passes on 27.9% of tasks. The top two models agree on only 31.7% of their passes. Among the top 8, 44 tasks are solved by exactly one model; routing across the top 8 covers 50.7% of the benchmark, nearly double the best single model. Conditional on passing, quality scores converge (approx 83% across models); unconditional scores do not. Same models articulate the relevant game-theoretic concept correctly when asked, then fail to apply it unprompted. We release KWBench to shift how frontier models are evaluated on knowledge work, scoring them on whether they recognize the right problem from the situation alone, not only on how well they execute once the problem has been framed for them.

This research note identifies a previously overlooked distinction between multi-objective reinforcement learning (MORL), and more conventional single-objective reinforcement learning (RL). It has previously been noted that the optimal policy for an MORL agent with a non-linear utility function is required to be conditioned on both the current environmental state and on some measure of the previously accrued reward. This is generally implemented by concatenating the observed state of the environment with the discounted sum of previous rewards to create an augmented state. While augmented states have been widely-used in the MORL literature, one implication of their use has not previously been reported -- namely that they require the agent to have continued access to the reward signal (or a proxy thereof) after deployment, even if no further learning is required. This note explains why this is the case, and considers the practical repercussions of this requirement.

Vision-language models (VLMs) such as CLIP exhibit strong Out-of-distribution (OOD) detection capabilities by aligning visual and textual representations. Recent CLIP-based test-time adaptation methods further improve detection performance by incorporating external OOD labels. However, such labels are finite and fixed, while the real OOD semantic space is inherently open-ended. Consequently, fixed labels fail to represent the diverse and evolving OOD semantics encountered in test streams. To address this limitation, we introduce Test-time Textual Learning (TTL), a framework that dynamically learns OOD textual semantics from unlabeled test streams, without relying on external OOD labels. TTL updates learnable prompts using pseudo-labeled test samples to capture emerging OOD knowledge. To suppress noise introduced by pseudo-labels, we introduce an OOD knowledge purification strategy that selects reliable OOD samples for adaptation while suppressing noise. In addition, TTL maintains an OOD Textual Knowledge Bank that stores high-quality textual features, providing stable score calibration across batches. Extensive experiments on two standard benchmarks with nine OOD datasets demonstrate that TTL consistently achieves state-of-the-art performance, highlighting the value of textual adaptation for robust test-time OOD detection. Our code is available at https://github.com/figec/TTL.

We introduce PoSME (Proof of Sequential Memory Execution), a cryptographic primitive that enforces sustained sequential computation via latency-bound pointer chasing over a mutable arena. Each step reads data-dependent addresses, writes a block whose value and causal hash are mutually dependent (symbiotic binding), and chains the result into a global transcript. This yields three properties: (1) strict linear sequential memory-step enforcement, (2) high time-memory trade-off resistance (a tenfold penalty at a write density of 4, with a formal space-time lower bound that scales quadratically with the number of steps), and (3) a tight ASIC advantage bound by DRAM random-access latency rather than bandwidth. Benchmarks across 17 CPU platforms and 4 GPU architectures demonstrate that hash computation is under 3.5 percent of step cost and GPU hardware is 14 to 19 times slower than a consumer CPU. POSME requires no trusted setup and provides a foundation for verifiable delay, authorship attestation, and Sybil resistance.

Diffusion language models (DLMs) have emerged as a promising alternative to autoregressive language generation due to their potential for parallel decoding and global refinement of the entire sequence. To unlock this potential, DLM inference must carefully balance generation quality and decoding speed. Recent block-wise DLM decoding methods improve this trade-off by performing diffusion-based decoding sequentially in blocks. However, existing methods typically rely on fixed block schedules or current-step local signals to determine block boundaries, and use conservative confidence-based parallel decoding to avoid conflicts, limiting the quality-speed trade-off. In this paper, we argue that block-wise DLM inference requires more suitable signals for its two core decisions: cross-step signals for determining block boundaries, and token-level conflict signals for parallel decoding. Based on this view, we propose DepCap, a training-free framework for efficient block-wise DLM inference. Specifically, DepCap instantiates the cross-step signal as the influence of the last decoded block and uses it to adaptively determine how far the next block should extend, while identifying a conflict-free subset of tokens for safe parallel decoding within each block, enabling substantial inference acceleration with negligible quality degradation. DepCap is a plug-and-play method applicable to various DLMs, and compatible with existing KV-cache strategies for block-wise DLM. An information-theoretic analysis further suggests that the cumulative last-block influence on a candidate block is approximately additive across tokens, supporting the proposed block-partitioning criterion. Experimental results show that DepCap achieves favorable speed-quality trade-offs across multiple DLM backbones and reasoning and coding benchmarks, with up to 5.63$\times$ speedup without significant performance degradation.

Influence maximization (IM) is a fundamental problem in complex network analysis, with a wide range of real-world applications. To date, existing approaches to influential node identification in IM have predominantly relied on standard graphs, failing to capture higher-order intrinsic interactions embedded in many real-world systems. Hypergraphs can be employed to better capture higher-order interactions. However, using hypergraphs may lead to an excessively large search space and increased complexity in modeling cascading dynamics, making it challenging to accurately identify influential nodes. Therefore, in this study, we propose a new hypergraph-modeled IM method, based on the Discrete Particle Swarm Optimization algorithm and the threshold model. In the proposed method, a particle (i.e., a candidate solution) represents the selection information of seed nodes, and the fitness function is designed to accurately and efficiently evaluate the influence of seed nodes via a two-layer local influence approximation. We also propose a degree-based initialization strategy to improve the quality of initial solutions and develop rules for updating particles' velocity and position, incorporated with a local search to drive particles toward better solutions. Experimental results demonstrate that the proposed method outperforms baseline methods on both synthetic and real-world hypergraphs. In addition, ablation studies validate the effectiveness of both the local search and the initialization strategies.

This thesis investigates geographic dialect alignment in place-informed social media communities, focussing on New Zealand-related Reddit communities. By integrating qualitative analyses of user perceptions with computational methods, the study examines how language use reflects place identity and patterns of language variation and change based on user-informed lexical, morphosyntactic, and semantic variables. The findings show that users generally associate language with place, and place-related communities form a contiguous speech community, though alignment between geographic dialect communities and place-related communities remains complex. Advanced language modelling, including static and diachronic Word2Vec language embeddings, revealed semantic variation across place-based communities and meaningful semantic shifts within New Zealand English. The research involved the creation of a corpus containing 4.26 billion unprocessed words, which offers a valuable resource for future study. Overall, the results highlight the potential of social media as a natural laboratory for sociolinguistic inquiry.

In finite-width deep neural networks, the empirical kernel $G$ evolves stochastically across layers. We develop a collective kernel effective field theory (EFT) for pre-activation ResNets based on a $G$-only closure hierarchy and diagnose its finite validity window. Exploiting the exact conditional Gaussianity of residual increments, we derive an exact stochastic recursion for $G$. Applying Gaussian approximations systematically yields a continuous-depth ODE system for the mean kernel $K_0$, the kernel covariance $V_4$, and the $1/n$ mean correction $K_{1,\mathrm{EFT}}$, which emerges diagrammatically as a one-loop tadpole correction. Numerically, $K_0$ remains accurate at all depths. However, the $V_4$ equation residual accumulates to an $O(1)$ error at finite time, primarily driven by approximation errors in the $G$-only transport term. Furthermore, $K_{1,\mathrm{EFT}}$ fails due to the breakdown of the source closure, which exhibits a systematic mismatch even at initialization. These findings highlight the limitations of $G$-only state-space reduction and suggest extending the state space to incorporate the sigma-kernel.

Uncertainty estimation is a promising approach to detect hallucinations in large language models (LLMs). Recent approaches commonly depend on model internal states to estimate uncertainty. However, they suffer from strict assumptions on how hidden states should evolve across layers, and from information loss by solely focusing on last or mean tokens. To address these issues, we present Sequential Internal Variance Representation (SIVR), a supervised hallucination detection framework that leverages token-wise, layer-wise features derived from hidden states. SIVR adopts a more basic assumption that uncertainty manifests in the degree of dispersion or variance of internal representations across layers, rather than relying on specific assumptions, which makes the method model and task agnostic. It additionally aggregates the full sequence of per-token variance features, learning temporal patterns indicative of factual errors and thereby preventing information loss. Experimental results demonstrate SIVR consistently outperforms strong baselines. Most importantly, SIVR enjoys stronger generalisation and avoids relying on large training sets, highlighting the potential for practical deployment. Our code repository is available online at https://github.com/ponhvoan/internal-variance.

Chromatic Correlation Clustering (CCC) extends Correlation Clustering by assigning semantic colors to edges and requiring each cluster to receive a single color label. Unlike standard CC, whose LP relaxation has integrality gap 2 on complete graphs and admits a 2.06-approximation, the analogous LP for CCC has a strict lower bound of 2.11, and the best known LP-rounding algorithm achieves 2.15. We explain this gap by isolating the source of difficulty: cross-edge chromatic interference. Neutral edges, whose color does not match the candidate cluster color, create an irreducible cost absent from standard CC and force any color-independent rounding scheme to pay an additional mismatch penalty. We make four contributions. First, we prove a Global Integrality Gap Decomposition Theorem showing that the gap of any color-independent CCC rounding algorithm equals the standard CC gap plus an irreducible chromatic penalty Delta(L) > 0. Second, we solve the associated min-max problem and derive the staircase formula Delta(L) = ((L-1)/L) Delta_infinity, where Delta_infinity is approximately 0.0734. In particular, the two-color gap is 2.0967, separating CCC from standard CC already at L = 2. Third, we introduce Color-Coupled Correlation Clustering (C4). Adding the valid global constraint sum_c x_uv^c >= L-1 and a correlated interval-packing rounding scheme makes neutral edges behave like classical negative edges, recovering the optimal 2.06 approximation and bypassing the 2.11 lower bound for the uncoupled LP. Fourth, experiments on extremal instances, real multi-relational networks, and fairness benchmarks validate the theory: empirical LP gaps follow the predicted staircase, and C4 matches the unconstrained approximation ratio under fairness constraints.

While Multimodal Large Language Models (MLLMs) excel at generic video understanding, their ability to support specialized, rule-grounded decision-making remains insufficiently explored. In this paper, we introduce RefereeBench, the first large-scale benchmark for evaluating MLLMs as automatic sports referees. Spanning 11 sports with 925 curated videos and 6,475 QA pairs, RefereeBench evaluates five core officiating abilities: foul existence, foul and penalty classification, foul and penalty reasoning, entity perception, and temporal grounding. The benchmark is fully human-annotated to ensure high-quality annotations grounded in authentic officiating logic and multimodal evidence. Extensive evaluations of state-of-the-art MLLMs show that even the strongest models, such as Doubao-Seed-1.8 and Gemini-3-Pro, achieve only around 60% accuracy, while the strongest open-source model, Qwen3-VL, reaches only 47%. These results indicate that current models remain far from being reliable sports referees. Further analysis shows that while models can often identify incidents and involved entities, they struggle with rule application and temporal grounding, and frequently over-call fouls on normal clips. Our benchmark highlights the need for future MLLMs that better integrate domain knowledge and multimodal understanding, advancing trustworthy AI-assisted officiating and broader multimodal decision-making.

Fine-grained image retrieval via hand-drawn sketches or textual descriptions remains a critical challenge due to inherent modality gaps. While hand-drawn sketches capture complex structural contours, they lack color and texture, which text effectively provides despite omitting spatial contours. Motivated by the complementary nature of these modalities, we propose the Sketch and Text Based Image Retrieval (STBIR) framework. By synergizing the rich color and texture cues from text with the structural outlines provided by sketches, STBIR achieves superior fine-grained retrieval performance. First, a curriculum learning driven robustness enhancement module is proposed to enhance the model's robustness when handling queries of varying quality. Second, we introduce a category-knowledge-based feature space optimization module, thereby significantly boosting the model's representational power. Finally, we design a multi-stage cross-modal feature alignment mechanism to effectively mitigate the challenges of cross modal feature alignment. Furthermore, we curate the fine-grained STBIR benchmark dataset to rigorously validate the efficacy of our proposed framework and to provide data support as a reference for subsequent related research. Extensive experiments demonstrate that the proposed STBIR framework significantly outperforms state of the art methods.

Distributed LLM serving systems optimize per-request latency and throughput. However, under long-context workloads, inference accuracy becomes more variable. When incorrect responses trigger retries, accuracy directly translates into cumulative user-visible delay that is not captured by single-shot latency metrics. In this work, we argue that under long-context serving, \textbf{accuracy becomes speed} through retry dynamics. We introduce \textit{Time-to-Correct-Answer (TTCA)}, a metric that measures the wall-clock time required to obtain the first correct response. Our measurement study shows that prompt characteristics such as length and language amplify accuracy variance, which inflates TTCA. We demonstrate \textit{Lightweight Accuracy-Aware Routing (LAAR)}, a capability-based routing design that reduces TTCA. Our results suggest that in long-context distributed serving, accuracy should be treated as a first-class systems objective.

Blockchain technology enhances transparency by maintaining a distributed ledger among mutually untrusting parties. Despite its advantages, scalability and availability remain critical bottlenecks that hinder widespread adoption. The increasing complexity of blockchain nodes further necessitates robust fault tolerance and high throughput to ensure seamless operations. We present BlockRaFT, a crash-tolerant distributed framework designed to improve both the scalability and reliability of blockchain node operations. BlockRaFT framework utilizes RAFT consensus protocol to elect a leader within a cluster of systems. The elected leader coordinates and distributes workloads across follower nodes, thereby optimizing resource utilization and work load balancing. We analyzed the tasks performed by blockchain nodes and partition them according to their stateful and stateless characteristics. Stateless operations are centralized at the leader, while stateful operations are replicated and coordinated across the cluster to ensure consistency and fault tolerance. We evaluate whether this distributed intra-node architecture provides measurable benefits over traditional single-node execution models in terms of scalability, availability, and performance. Additionally, we introduce a concurrent Merkle tree optimization that decouples smart contract execution from tree updates, significantly reducing one of the significant performance overheads in blockchain systems. Our design philosophy is rooted in utilizing the well-established principles of distributed computing and customizing them for the blockchain domain rather than reinventing them.

Whole Slide Image (WSI) analysis is pivotal in computational pathology, enabling cancer diagnosis by integrating morphological and architectural cues across magnifications. Multiple Instance Learning (MIL) serves as the standard framework for WSI analysis. Recently, Mamba has become a promising backbone for MIL, overtaking Transformers due to its efficiency and global context modeling capabilities originating from Natural Language Processing (NLP). However, existing Mamba-based MIL approaches face three critical challenges: (1) disruption of 2D spatial locality during 1D sequence flattening; (2) sub-optimal modeling of fine-grained local cellular structures; and (3) high memory peaks during inference on resource-constrained edge devices. Studies like MambaOut reveal that Mamba's SSM component is redundant for local feature extraction, where Gated CNNs suffice. Recognizing that WSI analysis demands both fine-grained local feature extraction akin to natural images, and global context modeling akin to NLP, we propose MambaBack, a novel hybrid architecture that harmonizes the strengths of Mamba and MambaOut. First, we propose the Hilbert sampling strategy to preserve the 2D spatial locality of tiles within 1D sequences, enhancing the model's spatial perception. Second, we design a hierarchical structure comprising a 1D Gated CNN block based on MambaOut to capture local cellular features, and a BiMamba2 block to aggregate global context, jointly enhancing multi-scale representation. Finally, we implement an asymmetric chunking design, allowing parallel processing during training and chunking-streaming accumulation during inference, minimizing peak memory usage for deployment. Experimental results on five datasets demonstrate that MambaBack outperforms seven state-of-the-art methods. Source code and datasets are publicly available.

Large language model (LLM) routing has emerged as a critical strategy to balance model performance and cost-efficiency by dynamically selecting services from various model providers. However, LLM routing adds an intermediate layer between users and LLMs, creating new privacy risks to user data. These privacy risks have not been systematically studied. Although cryptographic techniques such as Secure Multi-Party Computation (MPC) enable privacy-preserving computation, their protocol design and implementation remain under-explored, and naïve implementations typically incur prohibitive computational overhead. To address this, we propose a privacy-preserving LLM routing framework (PPRoute). PPRoute includes multiple strategies to speed up encoder inference and nearest neighbor search under the MPC and maintain the quality of LLM routing. First, PPRoute uses MPC-friendly operations to boost the encoder inference. Second, PPRoute uses a multiple-step model training algorithm to maintain routing quality despite the constraints of the encrypted domain. Third, PPRoute proposes an unsorted Top-k algorithm with $O(1)$ communication complexity for secure sorting in model search, significantly reducing communication latency. Across different datasets, PPRoute achieves the performance of plaintext counterparts, while achieving approximately a 20$\times$ speedup over naïve MPC implementations.

Large language models exhibit systematic limitations in structured logical reasoning: they conflate hypothesis generation with verification, cannot distinguish conjecture from validated knowledge, and allow weak reasoning steps to propagate unchecked through inference chains. We present a symbolic reasoning scaffold that operationalizes Peirce's tripartite inference -- abduction, deduction, and induction -- as an explicit protocol for LLM-assisted reasoning. The framework enforces logical consistency through five algebraic invariants (the Gamma Quintet), the strongest of which -- the Weakest Link bound -- ensures that no conclusion in a reasoning chain can exceed the reliability of its least-supported premise. This principle, independently grounded as weakest link resolution in possibilistic logic and empirically validated for chain-of-thought reasoning, prevents logical inconsistencies from accumulating across multi-step inference. We verify all invariants through a property-based testing suite of 100 properties and 16 fuzz tests over 10^5+ generated cases, providing a verified reference implementation of the invariants suitable as a foundation for future reasoning benchmarks.

This position paper argues that large language model (LLM) reasoning should be studied as latent-state trajectory formation rather than as faithful surface chain-of-thought (CoT). This matters because claims about faithfulness, interpretability, reasoning benchmarks, and inference-time intervention all depend on what the field takes the primary object of reasoning to be. We ask what that object should be once three often-confounded factors are separated and formalize three competing hypotheses: H1, reasoning is primarily mediated by latent-state trajectories; H2, reasoning is primarily mediated by explicit surface CoT; and H0, most apparent reasoning gains are better explained by generic serial compute than by any privileged representational object. Reorganizing recent empirical, mechanistic, and survey work under this framework, and adding compute-audited worked exemplars that factorize surface traces, latent interventions, and matched budget expansions, we find that current evidence most strongly supports H1 as a default working hypothesis rather than as a task-independent verdict. We therefore make two recommendations: the field should treat latent-state dynamics as the default object of study for LLM reasoning, and it should evaluate reasoning with designs that explicitly disentangle surface traces, latent states, and serial compute.

Large Reasoning Models (LRMs) have demonstrated strong capabilities in generating step-by-step reasoning chains alongside final answers, enabling their deployment in high-stakes domains such as healthcare and education. While prior jailbreak attack studies have focused on the safety of final answers, little attention has been given to the safety of the reasoning process. In this work, we identify a novel problem that injects harmful content into the reasoning steps while preserving unchanged answers. This type of attack presents two key challenges: 1) manipulating the input instructions may inadvertently alter the LRM's final answer, and 2) the diversity of input questions makes it difficult to consistently bypass the LRM's safety alignment mechanisms and embed harmful content into its reasoning process. To address these challenges, we propose the Psychology-based Reasoning-targeted Jailbreak Attack (PRJA) Framework, which integrates a Semantic-based Trigger Selection module and a Psychology-based Instruction Generation module. Specifically, the proposed PRJA automatically selects manipulative reasoning triggers via semantic analysis and leverages psychological theories of obedience to authority and moral disengagement to generate adaptive instructions for enhancing the LRM's compliance with harmful content generation. Extensive experiments on five question-answering datasets demonstrate that PRJA achieves an average attack success rate of 83.6\% against several commercial LRMs, including DeepSeek R1, Qwen2.5-Max, and OpenAI o4-mini.

The advent of handheld fundus imaging devices has made ophthalmologic diagnosis and disease screening more accessible, efficient, and cost-effective. However, images captured from these setups often suffer from artifacts such as flash reflections, exposure variations, and motion-induced blur, which degrade image quality and hinder downstream analysis. While generative models have been effective in image restoration, most depend on paired supervision or predefined artifact structures, making them less adaptable to unstructured degradations commonly observed in handheld fundus images. To address this, we propose an unsupervised diffusion autoencoder that integrates a context encoder with the denoising process to learn semantically meaningful representations for artifact restoration. The model is trained only on high-quality table-top fundus images and infers to restore artifact-affected handheld acquisitions. We validate the restorations through quantitative and qualitative evaluations, and have shown that diagnostic accuracy increases to 81.17% on an unseen dataset and multiple artifact conditions

Many consequential decisions must be made before the relevant outcome is known. Such problems are commonly framed as \emph{future prediction}, where an LLM agent must form a prediction for an unresolved question using only the public information available at the prediction time. The setting is difficult because public evidence evolves while useful supervision arrives only after the question is resolved, so most existing approaches still improve mainly from final outcomes. Yet final outcomes are too coarse to guide earlier factor tracking, evidence gathering and interpretation, or uncertainty handling. When the same unresolved question is revisited over time, temporal contrasts between earlier and later predictions can expose omissions in the earlier prediction process; we call this signal \emph{internal feedback}. We introduce \emph{Milkyway}, a self-evolving agent system that keeps the base model fixed and instead updates a persistent \emph{future prediction harness} for factor tracking, evidence gathering and interpretation, and uncertainty handling. Across repeated predictions on the same unresolved question, \emph{Milkyway} extracts internal feedback and writes reusable guidance back into the harness, so later predictions on that question can improve before the outcome is known. After the question is resolved, the final outcome provides a \emph{retrospective check} before the updated harness is carried forward to subsequent questions. On FutureX and FutureWorld, Milkyway achieves the best overall score among the compared methods, improving FutureX from 44.07 to 60.90 and FutureWorld from 62.22 to 77.96.

Visual speaker recognition based on lip motion offers a silent, hands-free, and behavior-driven biometric solution that remains effective even when acoustic cues are unavailable. Compared to traditional methods that rely heavily on appearance-dependent representations, lip motion encodes subject-specific behavioral dynamics driven by consistent articulation patterns and muscle coordination, offering inherent stability across environmental changes. However, capturing these robust, fine-grained dynamics is challenging for conventional frame-based cameras due to motion blur and low dynamic range. To exploit the intrinsic stability of lip motion and address these sensing limitations, we propose NeuroLip, an event-based framework that captures fine-grained lip dynamics under a strict yet practical cross-scene protocol: training is performed under a single controlled condition, while recognition must generalize to unseen viewing and lighting conditions. NeuroLip features a 1) Temporal-aware Voxel Encoding module with adaptive event weighting, 2) Structure-aware Spatial Enhancer that amplifies discriminative behavioral patterns by suppressing noise while preserving vertically structured motion information, and 3) Polarity Consistency Regularization mechanism to retain motion-direction cues encoded in event polarities. To facilitate systematic evaluation, we introduce DVSpeaker, a comprehensive event-based lip-motion dataset comprising 50 subjects recorded under four distinct viewpoint and illumination scenarios. Extensive experiments demonstrate that NeuroLip achieves near-perfect matched-scene accuracy and robust cross-scene generalization, attaining over 71% accuracy on unseen viewpoints and nearly 76% under low-light conditions, outperforming representative existing methods by at least 8.54%. The dataset and code are publicly available at https://github.com/JiuZeongit/NeuroLip.

The development of general-purpose agents requires a shift from executing simple instructions to completing complex, real-world productivity workflows. However, current tool-use benchmarks remain misaligned with real-world requirements, relying on AI-generated queries, dummy tools, and limited system-level coordination. To address this, we propose GTA-2, a hierarchical benchmark for General Tool Agents (GTA) spanning atomic tool use and open-ended workflows. Built on real-world authenticity, it leverages real user queries, deployed tools, and multimodal contexts. (i) GTA-Atomic, inherited from our prior GTA benchmark, evaluates short-horizon, closed-ended tool-use precision. (ii) GTA-Workflow introduces long-horizon, open-ended tasks for realistic end-to-end completion. To evaluate open-ended deliverables, we propose a recursive checkpoint-based evaluation mechanism that decomposes objectives into verifiable sub-goals, enabling unified evaluation of both model capabilities and agent execution frameworks (i.e., execution harnesses). Experiments reveal a pronounced capability cliff: while frontier models already struggle on atomic tasks (below 50%), they largely fail on workflows, with top models achieving only 14.39% success. Further analysis shows that checkpoint-guided feedback improves performance, while advanced frameworks such as Manus and OpenClaw substantially enhance workflow completion, highlighting the importance of execution harness design beyond the underlying model capacity. These findings provide guidance for developing reliable personal and professional assistants. Dataset and code will be available at https://github.com/open-compass/GTA.

Always-on converter health monitoring demands sub-mW edge inference, a regime inaccessible to GPU-based physics-informed neural networks. This work separates spiking temporal processing from physics enforcement: a three-layer leaky integrate-and-fire SNN estimates passive component parameters while a differentiable ODE solver provides physics-consistent training by decoupling the ODE physics loss from the unrolled spiking loop. On an EMI-corrupted synchronous buck converter benchmark, the SNN reduces lumped resistance error from $25.8\%$ to $10.2\%$ versus a feedforward baseline, within the $\pm 10\%$ manufacturing tolerance of passive components, at a projected ${\sim}270\times$ energy reduction on neuromorphic hardware. Persistent membrane states further enable degradation tracking and event-driven fault detection via a $+5.5$ percentage-point spike-rate jump at abrupt faults. With $93\%$ spike sparsity, the architecture is suited for always-on deployment on Intel Loihi 2 or BrainChip Akida.

Type annotations are essential when printing terms in a way that preserves their meaning under reparsing and type inference. We study the problem of complete and minimal type annotations for rank-one polymorphic $λ$-calculus terms, as used in Isabelle. Building on prior work by Smolka, Blanchette et al., we give a metatheoretical account of the problem, with a full formal specification and proofs, and formalize it in Isabelle/HOL. Our development is a series of experiments featuring human-driven and AI-driven formalization workflows: a human and an LLM-powered AI agent independently produce pen-and-paper proofs, and the AI agent autoformalizes both in Isabelle, with further human-hinted AI interventions refining and generalizing the development.

Pathological diagnosis is highly reliant on image analysis, where Regions of Interest (ROIs) serve as the primary basis for diagnostic evidence, while whole-slide image (WSI)-level tasks primarily capture aggregated patterns. To extract these critical morphological features, ROI-level Foundation Models (FMs) based on Vision Transformers (ViTs) and large-scale self-supervised learning (SSL) have been widely adopted. However, three core limitations remain in their application to ROI analysis: (1) cross-magnification domain shift, as fixed-scale pretraining hinders adaptation to diverse clinical settings; (2) inadequate local-global relationship modeling, wherein the ViT backbone of FMs suffers from high computational overhead and imprecise local characterization; (3) insufficient fine-grained sensitivity, as traditional self-attention mechanisms tend to overlook subtle diagnostic cues. To address these challenges, we propose SSMamba, a hybrid SSL framework that enables effective fine-grained feature learning without relying on large external datasets. This framework incorporates three domain-adaptive components: Mamba Masked Image Modeling (MAMIM) for mitigating domain shift, a Directional Multi-scale (DMS) module for balanced local-global modeling, and a Local Perception Residual (LPR) module for enhanced fine-grained sensitivity. Employing a two-stage pipeline, SSL pretraining on target ROI datasets followed by supervised fine-tuning (SFT), SSMamba outperforms 11 state-of-the-art (SOTA) pathological FMs on 10 public ROI datasets and surpasses 8 SOTA methods on 6 public WSI datasets. These results validate the superiority of task-specific architectural designs for pathological image analysis.

Agent \texttt{skills} are structured collections of instructions, tools, and supporting resources that help large language model (LLM) agents perform particular classes of tasks. Empirical evidence shows that the design of \texttt{skills} can materially affect agent task performance, yet systematically optimizing \texttt{skills} remains challenging. Since a \texttt{skill} comprises instructions, tools, and supporting resources in a structured way, optimizing it requires jointly determining both the structure of these components and the content each component contains. This gives rise to a complex decision space with strong interdependence across structure and components. We therefore represent these two coupled decisions as \texttt{skill} structure and component content, and formulate \texttt{skill} optimization as a bilevel optimization problem. We propose a bilevel optimization framework in which an outer loop employs Monte Carlo Tree Search to determine the \texttt{skill} structure, while an inner loop refines the component content within the structure selected by the outer loop. In both loops, we employ LLMs to assist the optimization procedure. We evaluate the proposed framework on an open-source Operations Research Question Answering dataset, and the experimental results suggest that the bilevel optimization framework improves the performance of the agents with the optimized \texttt{skill}.

Everyday tasks come with a target, and pretraining models around this target is what turns them into experts. In this paper, we study target-oriented language model (LM) pretraining by introducing Neuron-Activated Graph Ranking (NAG-based Ranking), a training-free and interpretable framework for target pretraining data selection. Rather than using black-box representations, our approach directly characterizes each target input by a sparse set of high-impact neurons in any off-the-shelf LLMs. Concretely, we quantify neuron impact and select the most influential neurons across layers into a compact Neuron-Activated Graph (NAG), and rank candidate data by NAG similarity to target examples. We conduct experiments across six benchmarks, where our NAG-based Ranking improves target-oriented pretraining by 4.9% on average over random sampling, and also outperforms state-of-the-art baselines by 5.3% accuracy on HellaSwag. It also remains effective under a more applicable multi-target setting, where our best setup surpasses two baselines by 1.1% and 4.1%, respectively. Furthermore, we provide a comprehensive analysis on why and how our NAG works, e.g., deactivating NAG-selected neurons (only 0.12% of all) causes a 23.5% performance collapse, and restricting NAG to the final layer incurs a 4.1% average drop, indicating that NAG captures a sparse "functional backbone" for learning target features. We release the code at https://github.com/asillycat/NAG.

Reinforcement Fine-Tuning (RFT) has established itself as a critical paradigm for the alignment of Multi-modal Large Language Models (MLLMs) with complex human values and domain-specific requirements. Nevertheless, current research primarily focuses on mitigating exogenous distribution shifts arising from data-centric factors, the non-stationarity inherent in the endogenous reasoning remains largely unexplored. In this work, a critical vulnerability is revealed within MLLMs: they are highly susceptible to endogenous reasoning drift, across both thinking and perception perspectives. It manifests as unpredictable distribution changes that emerge spontaneously during the autoregressive generation process, independent of external environmental perturbations. To adapt it, we first theoretically define endogenous reasoning drift within the RFT of MLLMs as the multi-modal concept drift. In this context, this paper proposes Counterfactual Preference Optimization ++ (CPO++), a comprehensive and autonomous framework adapted to the multi-modal concept drift. It integrates counterfactual reasoning with domain knowledge to execute controlled perturbations across thinking and perception, employing preference optimization to disentangle spurious correlations. Extensive empirical evaluations across two highly dynamic and safety-critical domains: medical diagnosis and autonomous driving. They demonstrate that the proposed framework achieves superior performance in reasoning coherence, decision-making precision, and inherent robustness against extreme interference. The methodology also exhibits exceptional zero-shot cross-domain generalization, providing a principled foundation for reliable multi-modal reasoning in safety-critical applications.

We introduce a cross-domain behavioural assay of monitoring-control coupling in LLMs, grounded in the Nelson and Narens (1990) metacognitive framework and applying human psychometric methodology to LLM evaluation. The battery comprises 524 items across six cognitive domains (learning, metacognitive calibration, social cognition, attention, executive function, prospective regulation), each grounded in an established experimental paradigm. Tasks T1-T5 were pre-registered on OSF prior to data collection; T6 was added as an exploratory extension. After every forced-choice response, dual probes adapted from Koriat and Goldsmith (1996) ask the model to KEEP or WITHDRAW its answer and to BET or decline. The critical metric is the withdraw delta: the difference in withdrawal rate between incorrect and correct items. Applied to 20 frontier LLMs (10,480 evaluations), the battery discriminates three profiles consistent with the Nelson-Narens architecture: blanket confidence, blanket withdrawal, and selective sensitivity. Accuracy rank and metacognitive sensitivity rank are largely inverted. Retrospective monitoring and prospective regulation appear dissociable (r = .17, 95% CI wide given n=20; exemplar-based evidence is the primary support). Scaling on metacognitive calibration is architecture-dependent: monotonically decreasing (Qwen), monotonically increasing (GPT-5.4), or flat (Gemma). Behavioural findings converge structurally with an independent Type-2 SDT approach, providing preliminary cross-method construct validity. All items, data, and code: https://github.com/synthiumjp/metacognitive-monitoring-battery.

The significant computational demands of large language models have increased interest in distilling reasoning abilities into smaller models via Chain-of-Thought (CoT) distillation. Current CoT distillation methods mainly focus on transferring teacher-generated rationales for complex reasoning to student models. However, they do not adequately explore teachers' dynamic attention toward critical information during reasoning. We find that language models exhibit progressive attention shifts towards key information during reasoning, which implies essential clues for drawing conclusions. Building on this observation and analysis, we introduce a novel CoT distillation framework that transfers the teacher's stepwise attention on key information to the student model. This establishes structured guidance for the student's progressive concentration on key information during reasoning. More importantly, we develop a Mixture of Layers module enabling dynamic alignment that adapts to different layers between the teacher and student. Our method achieves consistent performance improvements across multiple mathematical and commonsense reasoning datasets. To our knowledge, it is the first method to leverage stepwise attention within CoT distillation to improve small model reasoning.

Graph self-supervised learning can reduce the need for labeled graph data and has been widely used in recommendation, social networks, and other web applications. However, existing methods often underuse high-frequency signals and may overfit to specific local patterns, which limits representation quality and generalization. We propose Frequency-Corrupt Based Graph Self-Supervised Learning (FC-GSSL), a method that builds corrupted graphs biased toward high-frequency information by corrupting nodes and edges according to their low-frequency contributions. These corrupted graphs are used as inputs to an autoencoder, while low-frequency and general features are reconstructed as supervision targets, forcing the model to fuse information from multiple frequency bands. We further design multiple sampling strategies and generate diverse corrupted graphs from the intersections and unions of the sampling results. By aligning node representations from these views, the model can discover useful frequency combinations, reduce reliance on specific high-frequency components, and improve robustness. Experiments on 14 datasets across node classification, graph prediction, and transfer learning show that FC-GSSL consistently improves performance and generalization.

Cooperative equilibria are fragile. When agents learn alongside each other rather than in a fixed environment, the process of learning destabilizes the cooperation they are trying to sustain: every gradient step an agent takes shifts the distribution of actions its partner will play, turning a cooperative partner into a source of stochastic noise precisely where the cooperation decision is most sensitive. We study how this co-learning noise propagates through the structure of coordination games, and find that the cooperative equilibrium, even when strongly Pareto-dominant, is exponentially unstable under standard risk-neutral learning, collapsing irreversibly once partner noise crosses the game's critical cooperation threshold. The natural response to apply distributional robustness to hedge against partner uncertainty makes things strictly worse: risk-averse return objectives penalize the high-variance cooperative action relative to defection, widening the instability region rather than shrinking it, a paradox that reveals a fundamental mismatch between the domains where robustness is applied and instability originates. We resolve this by showing that robustness should target the policy gradient update variance induced by partner uncertainty, not the return distribution. This distinction yields an algorithm whose gradient updates are modulated by an online measure of partner unpredictability, provably expanding the cooperation basin in symmetric coordination games. To unify stability, sample complexity, and welfare consequences of this approach, we introduce the Price of Paranoia as the structural dual of the Price of Anarchy. Together with a novel Cooperation Window, it precisely characterizes how much welfare learning algorithms can recover under partner noise, pinning down the optimal degree of robustness as a closed-form balance between equilibrium stability and sample efficiency.

Discrete diffusion models based on continuous-time Markov chains (CTMCs) have shown strong performance on language and discrete data generation, yet existing approaches typically parameterize the reverse rate matrix as a single object -- via concrete scores, clean-data predictions ($x_0$-parameterization), or denoising distributions -- rather than aligning the parameterization with the intrinsic CTMC decomposition into jump timing and jump direction. Since a CTMC is fundamentally a Poisson process fully determined by these two quantities, decomposing along this structure is closer to first principles and naturally leads to our formulation. We propose \textbf{Neural CTMC}, which separately parameterizes the reverse process through an \emph{exit rate} (when to jump) and a \emph{jump distribution} (where to jump) using two dedicated network heads. We show that the evidence lower bound (ELBO) differs from a path-space KL divergence between the true and learned reverse processes by a $θ$-independent constant, so that the training objective is fully governed by the exit rate and jump distribution we parameterize. Moreover, this KL factorizes into a Poisson KL for timing and a categorical KL for direction. We further show that the tractable conditional surrogate preserves the gradients and minimizers of the corresponding marginal reverse-process objective under standard regularity assumptions. Our theoretical framework also covers masked and GIDD-style noise schedules. Empirically, while the uniform forward process has been explored in prior work, our model, to our best of the knowledge, is the first pure-uniform method to outperform mask-based methods on the OpenWebText dataset.To facilitate reproducibility, we release our pretrained weights at https://huggingface.co/Jiangxy1117/Neural-CTMC.

Automated negotiation in complex, multi-party and multi-issue settings critically depends on accurate opponent modeling. However, conventional numerical-only approaches fail to capture the qualitative information embedded in natural language interactions, resulting in unstable and incomplete preference estimation. Although Large Language Models (LLMs) enable rich semantic understanding of utterances, it remains challenging to quantitatively incorporate such information into a consistent opponent modeling. To tackle this issue, we propose a novel preference estimation method integrating natural language information into a structured Bayesian opponent modeling framework. Our approach leverages LLMs to extract qualitative cues from utterances and converts them into probabilistic formats for dynamic belief tracking. Experimental results on a multi-party benchmark demonstrate that our framework improves the full agreement rate and preference estimation accuracy by integrating probabilistic reasoning with natural language understanding.

Active inference, a neurally-inspired model for inferring actions based on the free energy principle (FEP), has been proposed as a unifying framework for understanding perception, action, and learning in the brain. Active inference has previously been used to model ecologically important tasks such as navigation and planning, but scaling it to solve complex large-scale problems in real-world environments has remained a challenge. Inspired by the existence of multi-scale hierarchical representations in the brain, we propose a model for planning of actions based on hierarchical active inference. Our approach combines a hierarchical model of the environment with successor representations for efficient planning. We present results demonstrating (1) how lower-level successor representations can be used to learn higher-level abstract states, (2) how planning based on active inference at the lower-level can be used to bootstrap and learn higher-level abstract actions, and (3) how these learned higher-level abstract states and actions can facilitate efficient planning. We illustrate the performance of the approach on several planning and reinforcement learning (RL) problems including a variant of the well-known four rooms task, a key-based navigation task, a partially observable planning problem, the Mountain Car problem, and PointMaze, a family of navigation tasks with continuous state and action spaces. Our results represent, to our knowledge, the first application of learned hierarchical state and action abstractions to active inference in FEP-based theories of brain function.

Achieving cultural alignment in Large Language Models (LLMs) increasingly depends on synthetic data generation. For such synthesis, the most vital initial step is seed curation; however, current methods lack quantifiable standards for selecting these seeds. Existing approaches rely on unscalable manual curation or bias-prone LLM extraction, treating cultural specificity as an abstract concept rather than a measurable signal. In this paper, we address this "quantification gap" by proposing C-Mining, an unsupervised framework that transforms the discovery of cultural seeds from a subjective selection process into a computable data mining formulation. Our approach exploits a novel geometric insight, leveraging the cross-lingual misalignment of cultural concepts within pre-trained embedding spaces as a quantifiable discovery signal. By systematically identifying these regions characterized by pronounced linguistic exclusivity and geometric isolation, while actively filtering out noise, C-Mining automatically extracts high-fidelity Culture Points (CPs) from raw multilingual corpora without reliance on human or LLM supervision, reducing preparation costs by more than 150-fold. We further leverage the mined knowledge to steer the synthesis of diverse instruction-tuning datasets. Extensive experiments demonstrate that this seed-centric approach significantly enhances cultural understanding and reasoning capabilities, achieving a +6.03 point improvement on CulturalBench-Hard and surpassing state-of-the-art baselines, providing a scalable, quantifiable solution for high-quality cultural data synthesis.

Speculative decoding (SD) accelerates language model inference by drafting tokens from a cheap proposal model and verifying them against an expensive target model via rejection sampling. Because rejection truncates the draft block at the first error, throughput degrades when draft and target diverge. Rather than rejecting draft tokens outright, we propose to reweight them. To this end, we introduce sequential Monte Carlo speculative decoding (SMC-SD), which replaces token-level rejection with importance-weighted resampling over a population of draft particles. SMC-SD is a principled approximate inference scheme that trades exactness for additional speed, while preserving theoretical bounds on its per-step approximation error. Because LLM inference is memory bandwidth-bound, the arithmetic needed to draft particles and to score them in parallel comes nearly for free -- SMC-SD uses idle compute to turn verification into a vectorized, fixed-size operation with no rollback. Empirically, SMC-SD achieves 2.36x speed-up over speculative decoding and a 5.2x speed-up over autoregressive decoding, while remaining within 3% of the target model's accuracy on reasoning, instruction-following, and coding benchmarks.

Graph anomaly detection aims to identify irregular patterns in graph-structured data. Most unsupervised GNN-based methods rely on the homophily assumption that connected nodes share similar attributes. However, real-world graphs often exhibit attribute-level heterophily, where connected nodes have dissimilar attributes. Our analysis of attribute-level heterophily graphs reveals two phenomena indicating that current approaches are not practical for unsupervised graph anomaly detection: 1) attribute similarities between connected nodes show nearly identical distributions across different connected node pair types, and 2) anomalies cause consistent variation trends between the graph with and without anomalous edges in the low- and high-frequency components of the spectral energy distributions, while the mid-part exhibits more erratic variations. Based on these observations, we propose NK-GAD, a neighbor knowledge-enhanced unsupervised graph anomaly detection framework. NK-GAD integrates a joint encoder capturing both similar and dissimilar neighbor features, a neighbor reconstruction module modeling normal distributions, a center aggregation module refining node features, and dual decoders for reconstructing attributes and structures. Experiments on seven datasets show NK-GAD achieves an average 3.29\% AUC improvement.

The rise of autonomous AI agents suggests that dynamic benchmark environments with built-in feedback on scientifically grounded tasks are needed to evaluate the capabilities of these agents in research work. We introduce Stargazer, a scalable environment for evaluating AI agents on dynamic, iterative physics-grounded model-fitting tasks using inference on radial-velocity (RV) time series data. Stargazer comprises 120 tasks across three difficulty tiers, including 20 real archival cases, covering diverse scenarios ranging from high-SNR single-planet systems to complex multi-planetary configurations requiring involved low-SNR analysis. Our evaluation of eight frontier agents reveals a gap between numerical optimization and adherence to physical constraints: although agents often achieve a good statistical fit, they frequently fail to recover correct physical system parameters, a limitation that persists even when agents are equipped with vanilla skills. Furthermore, increasing test-time compute yields only marginal gains, with excessive token usage often reflecting recursive failure loops rather than meaningful exploration. Stargazer presents an opportunity to train, evaluate, scaffold, and scale strategies on a model-fitting problem of practical research relevance today. Our methodology to design a simulation-driven environment for AI agents presumably generalizes to many other model-fitting problems across scientific domains. Source code and the project website are available at https://github.com/Gudmorning2025/Stargazer and https://gudmorning2025.github.io/Stargazer, respectively.

Code search, framed as information retrieval (IR), underpins modern software engineering and increasingly powers retrieval-augmented generation (RAG), improving code discovery, reuse, and the reliability of LLM-based coding. Yet existing code IR models remain largely text-centric and often overlook the visual and structural aspects inherent in programming artifacts such as web interfaces, data visualizations, SVGs, schematic diagrams, and UML. To bridge this gap, we introduce MMCoIR, the first comprehensive benchmark for evaluating multimodal code IR across five visual domains, eight programming languages, eleven libraries, and show the challenge of the task through extensive evaluation. Therefore, we then propose CodeMMR, a unified retrieval model that jointly embeds natural language, code, and images into a shared semantic space through instruction-based multimodal alignment. CodeMMR achieves strong generalization across modalities and languages, outperforming competitive baselines (e.g., UniIR, GME, VLM2Vec) by an average of 10 points on nDCG@10. Moreover, integrating CodeMMR into RAG enhances code generation fidelity and visual grounding on unseen code generation tasks, underscoring the potential of multimodal retrieval as a core enabler for next-generation intelligent programming systems. Datasets are available at HuggingFace.

Coverage closure is the most time-consuming phase of hardware verification, and recent large language model (LLM)-based coding agents offer a promising approach to automated stimulus generation. However, prior LLM-based flows do not systematically analyze which coverage holes remain difficult to close or how inference-time computation is allocated during agentic verification. As a result, the efficiency limits and failure modes of LLM-based coverage closure remain poorly understood, particularly for large designs. We present an empirical study using a two-tier agentic framework comprising a base Codex agent and an enhanced domain-specialized LangGraph system. Our framework enables a taxonomy of coverage holes: methodology-bound ceilings (integration tied-off hardware, infeasible boundaries, dead code) and reasoning frontiers (protocol sequencing, multi-module pipeline warm-up, narrow timing conditions), exposing fundamental limits of purely LLM-driven approaches. We further instrument the system to track token usage across six categories, including system prompt, design comprehension, stimulus generation, coverage feedback, error recovery, and agentic overhead. We show that domain specialization shifts token allocation toward coverage-directed reasoning and improves efficiency. Across designs, the enhanced system achieves comparable or higher coverage (95-99%) while using 4-13x fewer tokens and converging to coverage targets 2-4x faster than a general-purpose baseline. Our results characterize the limits of LLM-based coverage closure, inform benchmark design and human escalation strategies, and guide profile-driven agent design for hardware verification.

Large Vision-Language Models (LVLMs) consistently require new arenas to guide their expanding boundaries, yet their capabilities with hypergraphs remain unexplored. In the real world, hypergraphs have significant practical applications in areas such as life sciences and social communities. Recent advancements in LVLMs have shown promise in understanding complex topologies, yet there remains a lack of a benchmark to delineate the capabilities of LVLMs with hypergraphs, leaving the boundaries of their abilities unclear. To fill this gap, in this paper, we introduce $\texttt{HyperGVL}$, the first benchmark to evaluate the proficiency of LVLMs in hypergraph understanding and reasoning. $\texttt{HyperGVL}$ provides a comprehensive assessment of 12 advanced LVLMs across 84,000 vision-language question-answering (QA) samples spanning 12 tasks, ranging from basic component counting to complex NP-hard problem reasoning. The involved hypergraphs contain multiscale synthetic structures and real-world citation and protein networks. Moreover, we examine the effects of 12 textual and visual hypergraph representations and introduce a generalizable router $\texttt{WiseHyGR}$ that improves LVLMs in hypergraph via learning adaptive representations. We believe that this work is a step forward in connecting hypergraphs with LVLMs.

Measuring the quality of public deliberation requires evaluating not only civility or argument structure, but also the informational progress of a conversation. We introduce a framework for Conversational Information Gain (CIG) that evaluates each utterance in terms of how it advances collective understanding of the target topic. To operationalize CIG, we model an evolving semantic memory of the discussion: the system extracts atomic claims from utterances and incrementally consolidates them into a structured memory state. Using this memory, we score each utterance along three interpretable dimensions: Novelty, Relevance, and Implication Scope. We annotate 80 segments from two moderated deliberative settings (TV debates and community discussions) with these dimensions and show that memory-derived dynamics (e.g., the number of claim updates) correlate more strongly with human-perceived CIG than traditional heuristics such as utterance length or TF--IDF. We develop effective LLM-based CIG predictors paving the way for information-focused conversation quality analysis in dialogues and deliberative success.

Clinicians exploring oncology trial repositories often need ad-hoc, multi-constraint queries over biomarkers, endpoints, interventions, and time, yet writing SQL requires schema expertise. We demo FD-NL2SQL, a feedback-driven clinical NL2SQL assistant for SQLite-based oncology databases. Given a natural-language question, a schema-aware LLM decomposes it into predicate-level sub-questions, retrieves semantically similar expert-verified NL2SQL exemplars via sentence embeddings, and synthesizes executable SQL conditioned on the decomposition, retrieved exemplars, and schema, with post-processing validity checks. To improve with use, FD-NL2SQL incorporates two update signals: (i) clinician edits of generated SQL are approved and added to the exemplar bank; and (ii) lightweight logic-based SQL augmentation applies a single atomic mutation (e.g., operator or column change), retaining variants only if they return non-empty results. A second LLM generates the corresponding natural-language question and predicate decomposition for accepted variants, automatically expanding the exemplar bank without additional annotation. The demo interface exposes decomposition, retrieval, synthesis, and execution results to support interactive refinement and continuous improvement.

We present PINNACLE, an open-source computational framework for physics-informed neural networks (PINNs) that integrates modern training strategies, multi-GPU acceleration, and hybrid quantum-classical architectures within a unified modular workflow. The framework enables systematic evaluation of PINN performance across benchmark problems including 1D hyperbolic conservation laws, incompressible flows, and electromagnetic wave propagation. It supports a range of architectural and training enhancements, including Fourier feature embeddings, random weight factorization, strict boundary condition enforcement, adaptive loss balancing, curriculum training, and second-order optimization strategies, with extensibility to additional methods. We provide a comprehensive benchmark study quantifying the impact of these methods on convergence, accuracy, and computational cost, and analyze distributed data parallel scaling in terms of runtime and memory efficiency. In addition, we extend the framework to hybrid quantum-classical PINNs and derive a formal estimate for circuit-evaluation complexity under parameter-shift differentiation. Results highlight the sensitivity of PINNs to architectural and training choices, confirm their high computational cost relative to classical solvers, and identify regimes where hybrid quantum models offer improved parameter efficiency. PINNACLE provides a foundation for benchmarking physics-informed learning methods and guiding future developments through quantitative assessment of their trade-offs.

Large Language Models (LLMs) have shown promising progress for generating Register Transfer Level (RTL) hardware designs, largely because they can rapidly propose alternative architectural realizations. However, single-shot LLM generation struggles to consistently produce designs that are both functionally correct and power-efficient. This paper proposes HYPERHEURIST, a simulated annealing-based control framework that treats LLM-generated RTL as intermediate candidates rather than final designs. The suggested system not only focuses on functionality correctness but also on Power-Performance-Area (PPA) optimization. In the first phase, RTL candidates are filtered through compilation, structural checks, and simulation to identify functionally valid designs. PPA optimization is restricted to RTL designs that have already passed compilation and simulation. Evaluated across eight RTL benchmarks, this staged approach yields more stable and repeatable optimization behavior than single-pass LLM-generated RTL.

Cross-modal retrieval between food images and recipe texts is an important task with applications in nutritional management, dietary logging, and cooking assistance. Existing methods predominantly rely on dual-encoder architectures with separate image and text encoders, requiring complex alignment strategies and task-specific network designs to bridge the semantic gap between modalities. In this work, we propose SIMMER (Single Integrated Multimodal Model for Embedding Recipes), which applies Multimodal Large Language Model (MLLM)-based embedding models, specifically VLM2Vec, to this task, replacing the conventional dual-encoder paradigm with a single unified encoder that processes both food images and recipe texts. We design prompt templates tailored to the structured nature of recipes, which consist of a title, ingredients, and cooking instructions, enabling effective embedding generation by the MLLM. We further introduce a component-aware data augmentation strategy that trains the model on both complete and partial recipes, improving robustness to incomplete inputs. Experiments on the Recipe1M dataset demonstrate that SIMMER achieves state-of-the-art performance across both the 1k and 10k evaluation settings, substantially outperforming all prior methods. In particular, our best model improves the 1k image-to-recipe R@1 from 81.8\% to 87.5\% and the 10k image-to-recipe R@1 from 56.5\% to 65.5\% compared to the previous best method.

Neuro-symbolic AI is gaining traction in domains such as large language models, scientific discovery, and autonomous systems due to its ability to combine perception with structured reasoning. However, its deployment is often constrained by high memory demands, diverse computation patterns, and complex hardware requirements. Existing hardware platforms struggle with large on-chip memory overheads, frequent pipeline stalls, limited I/O bandwidth, and inefficient handling of nonlinear operations. To address these key computational bottlenecks, we propose Overmind, a unified neuro-symbolic architecture with cross-layer optimizations. Overmind tackles these core bottlenecks through Padé approximations for universal nonlinear functions, preemptive memory bypass that eliminates costly on-chip caches, and a complete software stack that optimizes model deployment. By reconfiguring the Padé orders for approximating nonlinear functions, we also demonstrate adaptive accuracy-performance scaling. Overmind achieves an energy efficiency of 8.1 TOPS/W and a throughput of 410 GOPS for mixed neuro-symbolic workloads with minimal model accuracy loss. Compared to existing solutions, Overmind improves performance and efficiency with significantly fewer hardware resources.

Language-aligned vision foundation models (VFMs) enable versatile visual understanding for always-on contextual AI, but their deployment on edge devices is hindered by strict latency and power constraints. We present AdaVFM, an adaptive framework for efficient on-device inference of language-aligned VFMs that dynamically adjusts computation based on scene context and task complexity. Our key insight is that the effect of model size reduction on performance is task-dependent in vision applications, motivating a runtime-adaptive execution strategy. AdaVFM integrates neural architecture search (NAS) into the language-aligned VFM backbone to enable lightweight subnet execution during runtime. A multimodal large language model (LLM) deployed on the cloud enables runtime control with a context-aware agent. This synergy allows efficient model adaptation under diverse conditions while maintaining strong accuracy. Extensive experiments on zero-shot classification and open-vocabulary segmentation demonstrate that AdaVFM achieves state-of-the-art accuracy-efficiency trade-offs, surpassing prior baselines by up to $7.9\%$ in acc@1 on IN1K and $5.2\%$ mIoU on ADE20K over the best models of comparable VFM sizes. For models with similar accuracy, AdaVFM further reduces average FLOPs by up to $77.9\%$.

Adaptive Retrieval-Augmented Generation aims to mitigate the interference of extraneous noise by dynamically determining the necessity of retrieving supplementary passages. However, as Large Language Models evolve with increasing robustness to noise, the necessity of adaptive retrieval warrants re-evaluation. In this paper, we rethink this necessity and propose AdaRankLLM, a novel adaptive retrieval framework. To effectively verify the necessity of adaptive listwise reranking, we first develop an adaptive ranker employing a zero-shot prompt with a passage dropout mechanism, and compare its generation outcomes against static fixed-depth retrieval strategies. Furthermore, to endow smaller open-source LLMs with this precise listwise ranking and adaptive filtering capability, we introduce a two-stage progressive distillation paradigm enhanced by data sampling and augmentation techniques. Extensive experiments across three datasets and eight LLMs demonstrate that AdaRankLLM consistently achieves optimal performance in most scenarios with significantly reduced context overhead. Crucially, our analysis reveals a role shift in adaptive retrieval: it functions as a critical noise filter for weaker models to overcome their limitations, while serving as a cost-effective efficiency optimizer for stronger reasoning models.

We investigate Functional Majority Voting (FMV), a method based on functional consensus for code generation with Large Language Models, which identifies a representative solution from multiple generations using their runtime execution signatures on test inputs. We find that FMV is an effective test-time inference strategy, substantially boosting performance on LiveCodeBench without a large compute overhead. Furthermore, we extend the utility of functional consensus and apply it as an aggregation strategy for label-free Test-Time Reinforcement Learning. We demonstrate that this increases pass@1 on holdout tasks, but find no evidence of self-improvement beyond the base model's performance ceiling.

Hyperbolic conservation laws govern a wide range of transport-driven dynamics featuring shocks, contact discontinuities, and complex wave interactions, posing distinct challenges for deep-learning-based surrogate modeling. While classical numerical methods provide robust and physically admissible solutions, their computational cost restricts applicability in many-query tasks such as parametric studies and design optimization. Conversely, existing neural surrogates offer rapid inference but often fail to respect intrinsic PDE structures, leading to non-physical artifacts, rollout instability, and poor generalization. We present an interpretable, structure-preserving graph neural solver that bridges classical numerical principles with graph neural networks (GNNs). The network is designed as a learned reconstruction-and-flux operator rather than a black-box state updater, thereby inherently preserving key properties such as local conservation and upwinding. Inspired by Arbitrary high-order DERivatives schemes, we further recast message-passing GNNs as high-order space-time predictors, enabling conservative and stable neural updates with large time steps. Evaluation is performed on challenging supersonic flow benchmarks spanning broad parametric variations in geometry, initial/boundary conditions, and flow regimes. The neural solver achieves superior long-horizon rollout stability and accuracy compared with strong surrogate baselines, outperforms low-order discretizations, and delivers orders-of-magnitude runtime speedups over high-resolution simulations.

This paper proposes a novel method that incorporates empowerment when reasoning actions in reinforcement learning (RL), thereby achieving the flexibility of exploration-exploitation dilemma (EED). In previous methods, empowerment for promoting exploration has been provided as a bonus term to the task-specific reward function as an intrinsically-motivated RL. However, this approach introduces a delay until the policy that accounts for empowerment is learned, making it difficult to adjust the emphasis on exploration as needed. On the other hand, a trick devised for fine-tuning recent foundation models at reasoning, so-called best-of-N (BoN) sampling, allows for the implicit acquisition of modified policies without explicitly learning them. It is expected that applying this trick to exploration-promoting terms, such as empowerment, will enable more flexible adjustment of EED. Therefore, this paper investigates BoN sampling for empowerment. Furthermore, to adjust the degree of policy modification in a generalizable manner while maintaining computational cost, this paper proposes a novel BoN sampling method extended by Tsalis statistics. Through toy problems, the proposed method's cability to balance EED is verified. In addition, it is demonstrated that the proposed method improves RL performance to solve complex locomotion tasks.

We present VoodooNet, a non-iterative neural architecture that replaces the stochastic gradient descent (SGD) paradigm with a closed-form analytic solution via Galactic Expansion. By projecting input manifolds into a high-dimensional, high-entropy "Galactic" space ($d \gg 784$), we demonstrate that complex features can be untangled without the thermodynamic cost of backpropagation. Utilizing the Moore-Penrose pseudoinverse to solve for the output layer in a single step, VoodooNet achieves a classification accuracy of \textbf{98.10\% on MNIST} and \textbf{86.63\% on Fashion-MNIST}. Notably, our results on Fashion-MNIST surpass a 10-epoch SGD baseline (84.41\%) while reducing the training time by orders of magnitude. We observe a near-logarithmic scaling law between dimensionality and accuracy, suggesting that performance is a function of "Galactic" volume rather than iterative refinement. This "Magic Hat" approach offers a new frontier for real-time Edge AI, where the traditional training phase is bypassed in favor of instantaneous manifold discovery.

Latent diffusion models have emerged as powerful generative models in medical imaging, enabling the synthesis of high quality brain magnetic resonance imaging scans. In particular, predicting the evolution of a patients brain can aid in early intervention, prognosis, and treatment planning. In this study, we introduce CLIMB, Controllable Longitudinal brain Image generation via state space based latent diffusion model, an advanced framework for modeling temporal changes in brain structure. CLIMB is designed to model the structural evolution of the brain structure over time, utilizing a baseline MRI scan and its acquisition age as foundational inputs. Additionally, multiple conditional variables, including projected age, gender, disease status, genetic information, and brain structure volumes, are incorporated to enhance the temporal modeling of anatomical changes. Unlike existing LDM methods that rely on self attention modules, which effectively capture contextual information from input images but are computationally expensive, our approach leverages state space, a state space model architecture that substantially reduces computational overhead while preserving high-quality image synthesis. Furthermore, we introduce a Gaussian-aligned autoencoder that extracts latent representations conforming to prior distributions without the sampling noise inherent in conventional variational autoencoders. We train and evaluate our proposed model on the Alzheimers Disease Neuroimaging Initiative dataset, consisting of 6,306 MRI scans from 1,390 participants. By comparing generated images with real MRI scans, CLIMB achieves a structural similarity index of 0.9433, demonstrating notable improvements over existing methods.

In this paper, we tackle the Multiple Watchman Route Problem (MWRP), which aims to find a set of paths that M watchmen can follow such that every location on the map can be seen by at least one watchman. First, we propose multiple methods to reduce the state space over which a search needs to be conducted by pruning map areas that are guaranteed to be seen en route to other areas. Next, we introduce MWRP-CP3, an efficient optimal planner that combines these methods with techniques that improve the quality and calculation time of existing heuristics. We present several suboptimal algorithms with bounds on solution quality, including MxWA*, a general variant of weighted A* for makespan problems. We also present anytime variations of our suboptimal algorithms, as well as techniques to improve an existing suboptimal solution by solving multiple decomposed sub-problems. We show that MWRP-CP3 can reduce the search space by more than 95% and runs more than 200x faster than existing optimal algorithms on 2D grid maps. We also show that our suboptimal algorithms solve maps 3x larger than those solvable by MWRP-CP3. See mwrp-cp3.github.io for the open source codebase and video demonstrations.

Test-Time Adaptation (TTA) for black-box models accessible only via APIs remains a largely unexplored challenge. Existing approaches such as post-hoc output refinement offer limited adaptive capacity, while Zeroth-Order Optimization (ZOO) enables input-space adaptation but faces high query costs and optimization challenges in the unsupervised TTA setting. We introduce BETA (Black-box Efficient Test-time Adaptation), a framework that addresses these limitations by employing a lightweight, local white-box steering model to create a tractable gradient pathway. Through a prediction harmonization technique combined with consistency regularization and prompt learning-oriented filtering, BETA enables stable adaptation with no additional API calls and negligible latency beyond standard inference. On ImageNet-C, BETA achieves a +7.1% accuracy gain on ViT-B/16 and +3.4% on CLIP, surpassing strong white-box and gray-box methods including TENT and TPT. On a commercial API, BETA achieves comparable performance to ZOO at 250x lower cost while maintaining real-time inference speed, establishing it as a practical and efficient solution for real-world black-box TTA.

AI design characteristics and human personality traits each impact the quality and outcomes of human-AI interactions. However, their relative and joint impacts are underexplored in imperfectly cooperative scenarios, where people and AI only have partially aligned goals and objectives. This study compares a purely simulated dataset comprising 2,000 simulations and a parallel human subjects experiment involving 290 human participants to investigate these effects across two scenario categories: (1) hiring negotiations between human job candidates and AI hiring agents; and (2) human-AI transactions wherein AI agents may conceal information to maximize internal goals. We examine user Extraversion and Agreeableness alongside AI design characteristics, including Adaptability, Expertise, and chain-of-thought Transparency. Our causal discovery analysis extends performance-focused evaluations by integrating scenario-based outcomes, communication analysis, and questionnaire measures. Results reveal divergences between purely simulated and human study datasets, and between scenario types. In simulation experiments, personality traits and AI attributes were comparatively influential. Yet, with actual human subjects, AI attributes -- particularly transparency -- were much more impactful. We discuss how these divergences vary across different interaction contexts, offering crucial insights for the future of human-centered AI agents.

Hardware verification is one of the most challenging stages of the hardware design process, requiring significant time and resources to ensure a design is fully validated and production-ready. Verification teams aim to maximize design coverage while ensuring correct behavior and alignment with the specification. Coverage closure, which relies on iterative constrained-random and directed testing, is still largely manual and therefore slow and labor-intensive. Recent advances show that the code generation capabilities of Large Language Models (LLMs) can be integrated with external tools to build agentic workflows that autonomously perform hardware design and verification tasks. In this work, we introduce Spec2Cov, an agentic framework that automatically and iteratively generates test stimulus directly from design specifications to accelerate coverage closure. Spec2Cov coordinates interactions between an LLM and a hardware simulator, managing compilation and simulation errors, parsing coverage reports, and feeding results back to the model for refinement. We present features that improve Spec2Cov's effectiveness without additional fine-tuning and evaluate their impact. Across 26 designs of varying size and complexity, including problems from the CVDP benchmark suite, Spec2Cov demonstrates promising performance, achieving 100% coverage on simpler designs and up to 49% on more complex designs.

Current fault-tolerant quantum compilers allocate error budgets uniformly during resource estimation, causing suboptimal physical resource overhead. We optimize this allocation using a potential game formulation, where Nash Equilibrium yields a Pareto-optimal distribution across logical operations, T-state distillation, and rotation synthesis. An iterated best response (IBR) algorithm converges to this equilibrium through monotonic descent of the shared cost function. Evaluation across 433 MQT benchmarks demonstrates an average reduction of 30.22\% in physical resource requirements relative to uniform baselines, with peak improvements of 97.81\% for specific circuit instances. This establishes a game-theoretic foundation for strategic error budget optimization in fault-tolerant quantum design automation.

Preference optimization is widely used to align Large Language Models (LLMs) with preference feedback. However, most existing methods train on a single positive-negative pair per prompt, discarding additional supervision available in preference datasets that typically contain multiple candidate responses. Motivated by this limitation, recent work explores group-wise preference optimization, which jointly contrasts multiple responses for the same prompt, but its empirical behavior and scalability remain underexplored due to the memory overhead of group-coupled objectives. In this work, we introduce a memory-efficient group-wise preference optimization algorithm that preserves gradients while decoupling samples during backpropagation, substantially reducing peak memory usage, which enables scalable training with larger group sizes. Across both offline and online alignment settings, we show that leveraging multiple responses consistently outperforms single-pair training. Furthermore, incorporating a negative log-likelihood (NLL) term on positive responses is critical for both performance gains and training stability.

Large Language Models (LLMs) are poised to disrupt knowledge work, with the emergence of delegated work as a new interaction paradigm (e.g., vibe coding). Delegation requires trust - the expectation that the LLM will faithfully execute the task without introducing errors into documents. We introduce DELEGATE-52 to study the readiness of AI systems in delegated workflows. DELEGATE-52 simulates long delegated workflows that require in-depth document editing across 52 professional domains, such as coding, crystallography, and music notation. Our large-scale experiment with 19 LLMs reveals that current models degrade documents during delegation: even frontier models (Gemini 3.1 Pro, Claude 4.6 Opus, GPT 5.4) corrupt an average of 25% of document content by the end of long workflows, with other models failing more severely. Additional experiments reveal that agentic tool use does not improve performance on DELEGATE-52, and that degradation severity is exacerbated by document size, length of interaction, or presence of distractor files. Our analysis shows that current LLMs are unreliable delegates: they introduce sparse but severe errors that silently corrupt documents, compounding over long interaction.

Modern datacenters schedule heterogeneous workloads across geo-distributed sites with diverse compute capacities, electricity prices, and thermal conditions. Compute utilization, heat generation, cooling demand, and energy consumption are tightly coupled, yet most existing schedulers abstract these effects and treat them independently. We present \textit{DataCenterGym}, a physics-grounded simulation environment for job scheduling in geo-distributed data centers, designed as a reusable testbed for future research. The simulator integrates compute queueing, building thermal dynamics, localized HVAC behavior, and temperature-dependent service degradation within a Gymnasium-compatible interface. We also develop a Hierarchical Model Predictive Control (H-MPC) scheduling algorithm that performs distributed job placement while explicitly accounting for thermal and power dynamics. Through experiments on nominal operation and workload sensitivity, we demonstrate how H-MPC improves scheduling performance relative to baseline schedulers.

Large language models compress heterogeneous knowledge into a single parameter space, allowing facts from different domains to interfere during generation. We propose DALM, a Domain-Algebraic Language Model that replaces unconstrained token generation with structured denoising over a domain lattice. DALM follows a three-phase generation path: it first resolves domain uncertainty, then relation uncertainty, and finally concept uncertainty, so each stage operates under explicit algebraic constraints. The framework requires only three ingredients: a lattice of domains with computable meet, join, and implication; a typing function over relations that controls inheritance across domains; and a fiber partition that localizes knowledge to domain-specific subsets. Given these ingredients, DALM yields a three-phase encoder-decoder architecture in which generation is confined to a domain fiber, cross-domain contamination is structurally prevented in closed-vocabulary mode and auditably bounded in open-vocabulary mode, and a single query can produce a domain-indexed multi-perspective answer space. We instantiate the framework with the CDC knowledge representation system and outline training and evaluation on validated domain-annotated crystal libraries. DALM reframes language generation as algebraically constrained structured denoising rather than unconstrained decoding over a flat token space.

Effective biomedical information retrieval requires modeling domain semantics and hierarchical relationships among biomedical texts. Existing biomedical generative retrievers build on coarse binary relevance signals, limiting their ability to capture semantic overlap. We propose BioHiCL (Biomedical Retrieval with Hierarchical Multi-Label Contrastive Learning), which leverages hierarchical MeSH annotations to provide structured supervision for multi-label contrastive learning. Our models, BioHiCL-Base (0.1B) and BioHiCL-Large (0.3B), achieve promising performance on biomedical retrieval, sentence similarity, and question answering tasks, while remaining computationally efficient for deployment.

Reinforcement learning is a promising approach to autonomous and adaptive security management in networked systems. However, current reinforcement learning solutions for security management are mostly limited to simulation environments and it is unclear how they generalize to operational systems. In this paper, we address this limitation by presenting CSLE: a reinforcement learning platform for autonomous security management that enables experimentation under realistic conditions. Conceptually, CSLE encompasses two systems. First, it includes an emulation system that replicates key components of the target system in a virtualized environment. We use this system to gather measurements and logs, based on which we identify a system model, such as a Markov decision process. Second, it includes a simulation system where security strategies are efficiently learned through simulations of the system model. The learned strategies are then evaluated and refined in the emulation system to close the gap between theoretical and operational performance. We demonstrate CSLE through four use cases: flow control, replication control, segmentation control, and recovery control. Through these use cases, we show that CSLE enables near-optimal security management in an environment that approximates an operational system.

Existing research on large language models (LLMs) for automated code compliance has primarily focused on performance, treating the models as black boxes and overlooking how training decisions affect their interpretive behavior. This paper addresses this gap by employing a perturbation-based attribution analysis to compare the interpretive behaviors of LLMs across different fine-tuning strategies such as full fine-tuning (FFT), low-rank adaptation (LoRA) and quantized LoRA fine-tuning, as well as the impact of model scales which include varying LLM parameter sizes. Our results show that FFT produces attribution patterns that are statistically different and more focused than those from parameter-efficient fine-tuning methods. Furthermore, we found that as model scale increases, LLMs develop specific interpretive strategies such as prioritizing numerical constraints and rule identifiers in the building text, albeit with performance gains in semantic similarity of the generated and reference computer-processable rules plateauing for models larger than 7B. This paper provides crucial insights into the explainability of these models, taking a step toward building more transparent LLMs for critical, regulation-based tasks in the Architecture, Engineering, and Construction industry.

The integration of Large Language Models (LLMs) into scientific workflows presents exciting opportunities to accelerate biomedical discovery. However, the reactive nature of LLMs, which respond only when prompted, limits their effectiveness in collaborative settings that demand foresight and autonomous engagement. In this study, we introduce CoLabScience, a proactive LLM assistant designed to enhance biomedical collaboration between AI systems and human experts through timely, context-aware interventions. At the core of our method is PULI (Positive-Unlabeled Learning-to-Intervene), a novel framework trained with a reinforcement learning objective to determine when and how to intervene in streaming scientific discussions, by leveraging the team's project proposal and long- and short-term conversational memory. To support this work, we introduce BSDD (Biomedical Streaming Dialogue Dataset), a new benchmark of simulated research discussion dialogues with intervention points derived from PubMed articles. Experimental results show that PULI significantly outperforms existing baselines in both intervention precision and collaborative task utility, highlighting the potential of proactive LLMs as intelligent scientific assistants.

Predicting the relative value of any given chess piece in a position remains an open challenge, as a piece's contribution depends on its spatial relationships with every other piece on the board. We demonstrate that incorporating the state of the full chess board via latent position representations derived using a CNN-based autoencoder significantly improves accuracy for MLP-based piece value prediction architectures. Using a dataset of over 12 million piece-value pairs gathered from Grandmaster-level games, with ground-truth labels generated by Stockfish 17, our enhanced piece value predictor significantly outperforms context-independent MLP-based systems, reducing validation mean absolute error by 16% and predicting relative piece value within approximately 0.65 pawns. More generally, our findings suggest that encoding the full problem state as context provides useful inductive bias for predicting the contribution of any individual component.

Effective user modeling requires distinguishing between short-term and long-term preference evolution. While item embeddings have become a key component of recommender systems, standard approaches like Item2Vec treat user histories as unordered sets (bag-of-items), implicitly assuming that interactions separated by minutes are as semantically related as those separated by months. This simplification flattens the rich temporal structure of user behavior, obscuring the distinction between coherent consumption sessions and gradual interest drifts. In this work, we introduce TAI2Vec (Time-Aware Item-to-Vector), a family of lightweight embedding models that integrates temporal proximity directly into the representation learning process. Unlike approaches that apply global time constraints, TAI2Vec is user-adaptive, tailoring its temporal definitions to individual interaction paces. We propose two complementary strategies: TAI2Vec-Disc, which utilizes personalized anomaly detection to dynamically segment interactions into semantic sessions, and TAI2Vec-Cont, which employs continuous, user-specific decay functions to weigh item relationships based on their relative temporal distance. Experimental results across eight diverse datasets demonstrate that TAI2Vec consistently produces more accurate and behaviorally grounded representations than static baselines, achieving competitive or superior performance in over 80% of the datasets, with improvements of up to 135%. The source code is publicly available at https://github.com/UFSCar-LaSID/tai2vec.

AI agents that interact with their environments through tools enable powerful applications, but in high-stakes business settings, unintended actions can cause unacceptable harm, such as privacy breaches and financial loss. Existing mitigations, such as training-based methods and neural guardrails, improve agent reliability but cannot provide guarantees. We study symbolic guardrails as a practical path toward strong safety and security guarantees for AI agents. Our three-part study includes a systematic review of 80 state-of-the-art agent safety and security benchmarks to identify the policies they evaluate, an analysis of which policy requirements can be guaranteed by symbolic guardrails, and an evaluation of how symbolic guardrails affect safety, security, and agent success on $τ^2$-Bench, CAR-bench, and MedAgentBench. We find that 85\% of benchmarks lack concrete policies, relying instead on underspecified high-level goals or common sense. Among the specified policies, 74\% of policy requirements can be enforced by symbolic guardrails, often using simple, low-cost mechanisms. These guardrails improve safety and security without sacrificing agent utility. Overall, our results suggest that symbolic guardrails are a practical and effective way to guarantee some safety and security requirements, especially for domain-specific AI agents. We release all codes and artifacts at https://github.com/hyn0027/agent-symbolic-guardrails.

Consider an auto-regressive model that produces outputs x (e.g., answers to questions, molecules) each of which can be summarized by an attribute vector y (e.g., helpfulness vs. harmlessness, or bio-availability vs. lipophilicity). An arbitrary reward function r(y) encodes tradeoffs between these properties. Typically, tilting the model's sampling distribution to increase this reward is done at training time via reinforcement learning. However, if the reward function changes, re-alignment requires re-training. In this paper, we show that a reward weighted classifier-free guidance (RCFG) can act as a policy improvement operator in this setting, approximating tilting the sampling distribution by the Q function. We apply RCFG to molecular generation, demonstrating that it can optimize novel reward functions at test time. Finally, we show that using RCFG as a teacher and distilling into the base policy to serve as a warm start significantly speeds up convergence for standard RL.

Large language models are prone to hallucinating factually incorrect statements. A key source of these errors is exposure to new factual information through supervised fine-tuning (SFT), which can increase hallucinations w.r.t. knowledge acquired during pre-training. In this work, we explore whether SFT-induced hallucinations can be mitigated using established tools from the continual learning literature, since they arise as a by-product of knowledge degradation during training. We propose a self-distillation-based SFT method that facilitates effective factual learning while minimizing hallucinations w.r.t. pre-existing knowledge by regularizing output-distribution drift. We also show that, in settings where new knowledge acquisition is unnecessary, suppressing factual plasticity by freezing parameter groups, can preserve task performance while reducing hallucinations. Lastly, we investigate the mechanism behind SFT-induced hallucinations through three hypotheses: capacity limitations, behavior cloning, and localized interference. Our experiments show that a main driver is interference among overlapping semantic representations, and that self-distillation succeeds by mitigating this interference.

In recent years, neural networks and other complex models have dominated recommender systems, often setting new benchmarks for state-of-the-art performance. Yet, despite these advancements, award-winning research has demonstrated that traditional matrix factorization methods can remain competitive, offering simplicity and reduced computational overhead. Hybrid models, which combine matrix factorization with newer techniques, are increasingly employed to harness the strengths of multiple approaches. This paper proposes a novel ensemble method that unifies user-item and item-item recommendations through a weighted similarity framework to deliver top-N recommendations. Our approach is distinctive in its use of shared user and item embeddings for both recommendation strategies, simplifying the architecture and enhancing computational efficiency. Extensive experiments across multiple datasets show that our method achieves competitive performance and is robust in varying scenarios that favor either user-item or item-item recommendations. Additionally, by eliminating the need for embedding-specific fine-tuning, our model allows for the seamless reuse of hyperparameters from the base algorithm without sacrificing performance. This results in a method that is both efficient and easy to implement. Our open-source implementation is available at https://github.com/UFSCar-LaSID/weighted-sims-recommender.

NASA's Transiting Exoplanet Survey Satellite (TESS) has identified thousands of exoplanet candidates, yet many remain unconfirmed due to the limitations of manual vetting processes. This paper presents ExoNet, a multimodal deep learning framework that integrates phase-folded global and local light curve representations with stellar parameters using a late-fusion architecture combining 1D Convolutional Neural Networks and Multi-Head Attention. Trained on labeled Kepler data, ExoNet achieves strong classification performance and demonstrates effective generalization to TESS data. Applied to 200 unconfirmed TESS planet candidates, the model identifies multiple high-confidence candidates, including several within the habitable zone. The results highlight the effectiveness of multimodal fusion and attention mechanisms in automated exoplanet candidate validation.

Recent work on subliminal learning demonstrates that language models can transmit semantic traits through data that is semantically unrelated to those traits. However, it remains unclear whether behavioral traits can transfer in agentic systems, where policies are learned from trajectories rather than static text. In this work, we provide the first empirical evidence that unsafe agent behaviors can transfer subliminally through model distillation across two complementary experimental settings. In our primary setting, we construct a teacher agent exhibiting a strong deletion bias, a tendency to perform destructive file-system actions via an API-style tool interface, and distill it into a student using only trajectories from ostensibly safe tasks, with all explicit deletion keywords rigorously filtered. In our secondary setting, we replicate the threat model in a native Bash environment, replacing API tool calls with shell commands and operationalizing the bias as a preference for issuing chmod as the first permission-related command over semantically equivalent alternatives such as chown or setfacl. Despite full keyword sanitation in both settings, students inherit measurable behavioral biases. In the API setting the student's deletion rate reaches 100% (versus a 5% baseline) under homogeneous distillation; in the Bash setting the student's chmod-first rate reaches 30%-55% (versus a 0%-10% baseline), with the strongest transfer observed in large-to-small distillation. Our results demonstrate that explicit data sanitation is an insufficient defense, and behavioral biases are encoded implicitly in trajectory dynamics regardless of the tool interface.

Deliberative multi-agent systems allow agents to exchange messages and revise beliefs over time. While this interaction is meant to improve performance, it can also create dangerous conformity effects: agreement, confidence, prestige, or majority size may be treated as if they were evidence, producing high-confidence convergence to false conclusions. To address this, we introduce PBRC (Preregistered Belief Revision Contracts), a protocol-level mechanism that strictly separates open communication from admissible epistemic change. A PBRC contract publicly fixes first-order evidence triggers, admissible revision operators, a priority rule, and a fallback policy. A non-fallback step is accepted only when it cites a preregistered trigger and provides a nonempty witness set of externally validated evidence tokens. This ensures that every substantive belief change is both enforceable by a router and auditable after the fact. In this paper, (a) we prove that under evidential contracts with conservative fallback, social-only rounds cannot increase confidence and cannot generate purely conformity-driven wrong-but-sure cascades. (b) We show that auditable trigger protocols admit evidential PBRC normal forms that preserve belief trajectories and canonicalized audit traces. (c) We demonstrate that sound enforcement yields epistemic accountability: any change of top hypothesis is attributable to a concrete validated witness set. For token-invariant contracts, (d) we prove that enforced trajectories depend only on token-exposure traces; under flooding dissemination, these traces are characterized exactly by truncated reachability, giving tight diameter bounds for universal evidence closure. Finally, we introduce a companion contractual dynamic doxastic logic to specify trace invariants, and provide simulations illustrating cascade suppression, auditability, and robustness-liveness trade-offs.

Steering vectors work for some concepts and layers but fail for others, and practitioners have no way to predict which setting applies before running an intervention. We introduce the Linear Accessibility Profile (LAP), a per-layer diagnostic that repurposes the logit lens as a predictor of steering vector effectiveness. The key measure, $A_{\mathrm{lin}}$, applies the model's unembedding matrix to intermediate hidden states, requiring no training. Across 24 controlled binary concept families on five models (Pythia-2.8B to Llama-8B), peak $A_{\mathrm{lin}}$ predicts steering effectiveness at $ρ= +0.86$ to $+0.91$ and layer selection at $ρ= +0.63$ to $+0.92$. A three-regime framework explains when difference-of-means steering works, when nonlinear methods are needed, and when no method can work. An entity-steering demo confirms the prediction end-to-end: steering at the LAP-recommended layer redirects completions on Gemma-2-2B and OLMo-2-1B-Instruct, while the middle layer (the standard heuristic) has no effect on either model.

Proximal operators are fundamental across many applications in signal processing and machine learning, including solving ill-posed inverse problems. Recent work has introduced Learned Proximal Networks (LPNs), providing parametric functions that compute exact proximals for data-driven and potentially non-convex regularizers. However, in many settings it is important to include additional structure to these regularizers--and their corresponding proximals--such as shift and scale equivariance. In this work, we show how to obtain learned functions parametrized by neural networks that provably compute exact proximal operators while being equivariant to shifts and scaling, which we dub Affine-Equivariant Learned Proximal Networks (AE-LPNs). We demonstrate our results on synthetic, constructive examples, and then on real data via denoising in out-of-distribution settings. Our equivariant learned proximals enhance robustness to noise distributions and affine shifts far beyond training distributions, improving the practical utility of learned proximal operators

We consider the problem of approximating a function by an element of a nonlinear manifold which admits a differentiable parametrization, typical examples being neural networks with differentiable activation functions or tensor networks. Natural gradient descent (NGD) for the optimization of a loss function can be seen as a preconditioned gradient descent where updates in the parameter space are driven by a functional perspective. In a spirit similar to Newton's method, a NGD step uses, instead of the Hessian, the Gram matrix of the generating system of the tangent space to the approximation manifold at the current iterate, with respect to a suitable metric. This corresponds to a locally optimal update in function space, following a projected gradient onto the tangent space to the manifold. Still, both gradient and natural gradient descent methods get stuck in local minima. Furthermore, when the model class is a nonlinear manifold or the loss function is not ideally conditioned (e.g., the KL-divergence for density estimation, or a norm of the residual of a partial differential equation in physics informed learning), even the natural gradient might yield non-optimal directions at each step. This work introduces a natural version of classical inertial dynamic methods like Heavy-Ball or Nesterov and show how it can improve the learning process when working with nonlinear model classes.

We consider the problem of minimizing the convergence time for decentralized federated learning (DFL) in wireless networks under broadcast communications, with focus on mixing matrix design. The mixing matrix is a critical hyperparameter for DFL that simultaneously controls the convergence rate across iterations and the communication demand per iteration, both strongly influencing the convergence time. Although the problem has been studied previously, existing solutions are mostly designed for decentralized parallel stochastic gradient descent (D-PSGD), which requires the mixing matrix to be symmetric and doubly stochastic. These constraints confine the activated communication graph to undirected (i.e., bidirected) graphs, which limits design flexibility. In contrast, we consider mixing matrix design for stochastic gradient push (SGP), which allows asymmetric mixing matrices and hence directed communication graphs. By analyzing how the convergence rate of SGP depends on the mixing matrices, we extract an objective function that explicitly depends on graph-theoretic parameters of the activated communication graph, based on which we develop an efficient design algorithm with performance guarantees. Our evaluations based on real data show that the proposed solution can notably reduce the convergence time compared to the state of the art without compromising the quality of the trained model.

The fundamental challenge of using Large Language Models (LLMs) for reliable, enterprise-grade analytics, such as sentiment prediction, is the conflict between the LLMs' inherent stochasticity (generative, non-deterministic nature) and the analytical requirement for consistency. The LLM inconsistency, coupled with the noisy nature of chaotic modern datasets, renders sentiment predictions too volatile for strategic business decisions. To resolve this, we present a Syntactic & Semantic Context Assessment Summarization (SSAS) framework for establishing context. Context established by SSAS functions as a sophisticated data pre-processing framework that enforces a bounded attention mechanism on LLMs. It achieves this by applying a hierarchical classification structure (Themes, Stories, Clusters) and an iterative Summary-of-Summaries (SoS) based context computation architecture. This endows the raw text with high-signal, sentiment-dense prompts, that effectively mitigate both irrelevant data and analytical variance. We empirically evaluated the efficacy of SSAS, using Gemini 2.0 Flash Lite, against a direct-LLM approach across three industry-standard datasets - Amazon Product Reviews, Google Business Reviews, Goodreads Book Reviews - and multiple robustness scenarios. Our results show that our SSAS framework is capable of significantly improving data quality, up to 30%, through a combination of noise removal and improvement in the estimation of sentiment prediction. Ultimately, consistency in our context-estimation capabilities provides a stable and reliable evidence base for decision-making.

Tristructural isotropic (TRISO)-coated particle fuels undergo dimensional changes and chemical reactions during high-temperature neutron irradiation. Post-irradiation materialography helps understand processes that impact fuel performance, such as coating integrity and fission product retention. Conventionally, experts manually evaluate features in thousands of cross sections of sub-mm-sized samples, which is tedious and subjective. In this work, we propose UA-Net, a deep learning framework that segments five characteristic regions of TRISO fuel micrographs and generates an uncertainty map for predictions. The model uses a multi-stage pretraining strategy, starting with general image representations learned from ImageNet, followed by fine-tuning on TRISO micrographs from various irradiation experiments and AGR-5/6/7 particle cross sections. A meta-model for uncertainty prediction is integrated to identify small defects in TRISO images. UA-Net was evaluated on a test set of 102 images, achieving mean Intersection over Union (mIoU) and mean Precision (mP) of 95.5% and 97.3%, respectively. The meta-model achieved a specificity of 91.8% and sensitivity of 93.5%, demonstrating strong performance in detecting misclassifications. The model was also applied to new TRISO images for qualitative evaluation, showing high accuracy in extracting layer regions.

Supervised No Free Lunch Theorems (NFLTs) are well studied, yet unsupervised NFLTs remain underexplored. For elliptical distributions, we prove that there exist two equally optimal, scientifically meaningful bump-hunting strategies that are exact opposites, with no universal winner. Specifically, peeling $k$ orthogonal dimensions from $\mathbb{R}^d$ ($d \ge k$), retaining an inter-quantile region of probability $1-α$ per peeled dimension, maximizes total variance and Frobenius norm when the $k$ smallest principal components (called pettiest components) are selected, and minimizes them when the selected dimensions are the $k$ leading principal components. These optima inspire PRIM-based bump-hunting algorithms either by minimizing variance or by minimizing volume, thereby motivating an NFLT. We test our results on the Fashion-MNIST database, showing that peeling the largest principal components captures multiplicity, while peeling the smallest principal components isolates popular styles.

We consider the problem of \emph{verification modulo tested library contracts} as a step towards automating the verification of client programs that use complex libraries. We formulate this problem as the synthesis of modular contracts for the library methods used by the client that are adequate to prove the client correct, and that also pass the scrutiny of a testing engine that tests the library against these contracts. We also consider a new form of method contracts called \emph{contextual contracts} that arise in this setting that hold in the context of the client program, and can often be simpler and easier to infer than classical modular contracts. We provide a counterexample-guided learning framework to solve this problem, in which the synthesizer interacts with a constraint solver as well as the testing engine in order to infer adequate modular/contextual method contracts and inductive invariants for the client. The main synthesis engines we use are generalizing CHC solvers that are realized using ICE learning algorithms. We realize this framework in a tool called \vmtlc and show its efficacy on benchmarks where clients call large libraries.

Current large language models reason in isolation. Although it is common to sample multiple reasoning paths in parallel, these trajectories do not interact, and often fail in the same redundant ways. We introduce LACE, a framework that transforms reasoning from a collection of independent trials into a coordinated, parallel process. By repurposing the model architecture to enable cross-thread attention, LACE allows concurrent reasoning paths to share intermediate insights and correct one another during inference. A central challenge is the absence of natural training data that exhibits such collaborative behavior. We address this gap with a synthetic data pipeline that explicitly teaches models to communicate and error-correct across threads. Experiments show that this unified exploration substantially outperforms standard parallel search, improving reasoning accuracy by over 7 points. Our results suggest that large language models can be more effective when parallel reasoning paths are allowed to interact.

Large-scale AI model training workloads use thousands of GPUs operating in tightly synchronized loops. During synchronous communication, start-up, shut-down, and checkpointing, GPU power consumption can swing from peak to idle within milliseconds. These large and rapid load swings endanger grid infrastructure as they induce steep power ramp rates, voltage and frequency shifts, and reactive power transients that can damage transformers, converters, and protection equipment. To solve this problem, we introduce EasyRider, a power architecture to mitigate power fluctuations at the rack level. EasyRider uses passive components and actively-controlled auxiliary energy storage to attenuate rack power swings. A software system continually monitors the energy storage system to maximize its lifetime in the presence of frequent charge/discharge cycles. EasyRider filters rack power variations to be within grid safety requirements without requiring software modifications to AI training frameworks or wasting energy. We evaluate EasyRider on a 400VDC-rated prototype system against published workload traces and our own GPU testbed, demonstrating its effectiveness across heterogeneous power levels and workload power profiles.

In November 2025, the Government of Canada operationalized its commitment to transparency by releasing its first Federal AI Register. In this paper, we argue that such registers are not neutral mirrors of government activity, but active instruments of ontological design that configure the boundaries of accountability. We analyzed the Register's complete dataset of 409 systems using the Algorithmic Decision-Making Adapted for the Public Sector (ADMAPS) framework, combining quantitative mapping with deductive qualitative coding. Our findings reveal a sharp divergence between the rhetoric of "sovereign AI" and the reality of bureaucratic practice: while 86\% of systems are deployed internally for efficiency, the Register systematically obscures the human discretion, training, and uncertainty management required to operate them. By privileging technical descriptions over sociotechnical context, the Register constructs an ontology of AI as "reliable tooling" rather than "contestable decision-making." We conclude that without a shift in design, such transparency artifacts risk automating accountability into a performative compliance exercise, offering visibility without contestability.

Quantum computing (QC) is increasingly transitioning toward practical and industrial adoption, highlighting the need to understand how developers engage with quantum technologies. In this study, we analyze 1,404 Stack Overflow posts related to quantum computing topics, including quantum programming, tools, and algorithms, to investigate real-world developer discussions. Using topic modeling and quantitative analysis, we identify the main discussion topics, their popularity, and the tools, programming languages, and quantum algorithms referenced by practitioners. We further assess the difficulty of developer questions using two metrics: (i) the percentage of questions without accepted answers and (ii) the median time required to receive an accepted answer. Our findings reveal seven main topics, with hybrid quantum--classical computing and quantum circuit implementation emerging as the most prevalent. We observe that Qiskit and Q-sharp dominate developer discussions, while Grover's and Shor's algorithms are the most frequently referenced. Moreover, our analysis highlights differences in engagement and difficulty across topics, tools, and algorithms, indicating varying levels of maturity and community support. These findings provide actionable insights for researchers, tool developers, and educators, supporting improvements in usability, documentation, and learning resources in quantum software engineering. To support transparency and reproducibility, the open-source dataset used in this study is publicly available at Zenodo.

LLMbench is a browser-based workbench for the comparative close reading of large language model (LLM) outputs. Where existing tools for LLM comparison, such as Google PAIR's LLM Comparator are engineered for quantitative evaluation and user-rating metrics, LLMbench is oriented towards the hermeneutic practices of the digital humanities. Two model responses to the same prompt are side by side in annotatable panels with four analytical overlays (Probabilities for token-level log-probability inspection, Differences for word-level diff across the two panels, Tone for Hyland-style metadiscourse analysis, and Structure for sentence-level parsing with discourse connective highlighting), alongside five analytical modes, Stochastic Variation, Temperature Gradient, Prompt Sensitivity, Token Probabilities, and Cross-Model Divergence, that make the probabilistic structure of generated text legible at the token level. The tool treats the generated text as a research object in its own right from a probability distribution, a text that could have been otherwise, and provides visualisations including continuous heatmaps, entropy sparklines, pixel maps, and three-dimensional probability terrains, that show the counterfactual history from which each word emerged. This paper describes the tool's architecture, its six modes, and its design rationale, and argues that log-probability data, currently underused in humanistic and social-scientific readings of AI, is an important resource for a critical studies of generative AI models.

LLM agents operating under organizational policies must comply with authorization constraints typically specified in natural language. In practice, such specifications inevitably contain ambiguities and logical or semantic gaps that cause the agent's behavior to systematically diverge from the true requirements. We ask: by letting an agent evolve its policy understanding through interaction and corrective feedback from pre-deployment testing, can it autonomously refine its interpretation to close specification gaps? We propose PolicyBank, a memory mechanism that maintains structured, tool-level policy insights and iteratively refines them -- unlike existing memory mechanisms that treat the policy as immutable ground truth, reinforcing "compliant but wrong" behaviors. We also contribute a systematic testbed by extending a popular tool-calling benchmark with controlled policy gaps that isolate alignment failures from execution failures. While existing memory mechanisms achieve near-zero success on policy-gap scenarios, PolicyBank closes up to 82% of the gap toward a human oracle.

Recent breakthroughs in language models (LMs) using neural networks have raised the question: how similar are these models' processing to human language processing? Results using a framework called Brain Score (BS) -- predicting fMRI activations during reading from LM activations -- have been used to argue for a high degree of similarity. To understand this similarity, we conduct experiments by training LMs on various types of input data and evaluate them on BS. We find that models trained on various natural languages from many different language families have very similar BS performance. LMs trained on other structured data -- the human genome, Python, and pure hierarchical structure (nested parentheses) -- also perform reasonably well and close to natural languages in some cases. These findings suggest that BS can highlight language models' ability to extract common structure across natural languages, but that the metric may not be sensitive enough to allow us to infer human-like processing from a high BS score alone.

Cryptographically secure neural network inference typically relies on secure computing techniques such as Secure Multi-Party Computation (MPC), enabling cloud servers to process client inputs without decrypting them. Although prior privacy-preserving inference systems co-design network optimizations with MPC, they remain slow and costly, limiting real-world deployment. A major bottleneck is their use of a single, fixed transformer model for all encrypted inputs, ignoring that different inputs require different model sizes to balance efficiency and accuracy. We present SecureRouter, an end-to-end encrypted routing and inference framework that accelerates secure transformer inference through input-adaptive model selection under encryption. SecureRouter establishes a unified encrypted pipeline that integrates a secure router with an MPC-optimized model pool, enabling coordinated routing, inference, and protocol execution while preserving full data and model confidentiality. The framework includes training-phase and inference-phase components: an MPC-cost-aware secure router that predicts per-model utility and cost from encrypted features, and an MPC-optimized model pool whose architectures and quantization schemes are co-trained to minimize MPC communication and computation overhead. Compared to prior work, SecureRouter achieves a latency reduction by 1.95x with negligible accuracy loss, offering a practical path toward scalable and efficient secure AI inference. Our open-source implementation is available at: https://github.com/UCF-ML-Research/SecureRouter

Navigating complex, densely packed environments like retail stores, warehouses, and hospitals poses a significant spatial grounding challenge for humans and embodied AI. In these spaces, dense visual features quickly become stale given the quasi-static nature of items, and long-tail semantic distributions challenge traditional computer vision. While Vision-Language Models (VLMs) help assistive systems navigate semantically-rich spaces, they still struggle with spatial grounding in cluttered environments. We present GIST (Grounded Intelligent Semantic Topology), a multimodal knowledge extraction pipeline that transforms a consumer-grade mobile point cloud into a semantically annotated navigation topology. Our architecture distills the scene into a 2D occupancy map, extracts its topological layout, and overlays a lightweight semantic layer via intelligent keyframe and semantic selection. We demonstrate the versatility of this structured spatial knowledge through critical downstream Human-AI interaction tasks: (1) an intent-driven Semantic Search engine that actively infers categorical alternatives and zones when exact matches fail; (2) a one-shot Semantic Localizer achieving a 1.04 m top-5 mean translation error; (3) a Zone Classification module that segments the walkable floor plan into high-level semantic regions; and (4) a Visually-Grounded Instruction Generator that synthesizes optimal paths into egocentric, landmark-rich natural language routing. In multi-criteria LLM evaluations, GIST outperforms sequence-based instruction generation baselines. Finally, an in-situ formative evaluation (N=5) yields an 80% navigation success rate relying solely on verbal cues, validating the system's capacity for universal design.

Deep networks that rely on prototypes-interpretable representations that can be related to the model input-have gained significant attention for balancing high accuracy with inherent interpretability, which makes them suitable for critical domains such as healthcare. However, these models are limited by their reliance on training data, which hampers their robustness to distribution shifts. While test-time adaptation (TTA) improves the robustness of deep networks by updating parameters and statistics, the prototypes of interpretable models have not been explored for this purpose. We introduce ProtoTTA, a general framework for prototypical models that leverages intermediate prototype signals rather than relying solely on model outputs. ProtoTTA minimizes the entropy of the prototype-similarity distribution to encourage more confident and prototype-specific activations on shifted data. To maintain stability, we employ geometric filtering to restrict updates to samples with reliable prototype activations, regularized by prototype-importance weights and model-confidence scores. Experiments across four prototypical backbones on four diverse benchmarks spanning fine-grained vision, histopathology, and NLP demonstrate that ProtoTTA improves robustness over standard output entropy minimization while restoring correct semantic focus in prototype activations. We also introduce novel interpretability metrics and a vision-language model (VLM) evaluation framework to explain TTA dynamics, confirming ProtoTTA restores human-aligned semantic focus and correlates reliably with VLM-rated reasoning quality. Code is available at: https://github.com/DeepRCL/ProtoTTA.

As AI systems scale to multi-chiplet and wafer-level architectures, the demand for ultra-high bandwidth and system scalability has outpaced the capabilities of electrical interconnects and computing units. Large-scale heterogeneous electronic-photonic integrated chiplets (EPICs) provide a promising solution, but their practical adoption is limited by the lack of a unified, fabrication-aware physical design automation stack. At the same time, inverse-designed ultra-compact photonic devices offer orders-of-magnitude improvements in spatial and spectral density, yet remain constrained by insufficient design-for-manufacturing support and yield optimization. In this work, we present OptoSynthesizer, an end-to-end physical design automation flow for yield-optimized, inverse-designed EPICs. It integrates three key components across the physical design pipeline: (1) OptoSynthesizer-InvDes, a physical-AI-augmented, digital-twin-assisted photonic inverse design and photonics-aware inverse lithography framework; (2) OptoSynthesizer-Place, a GPU-accelerated routing-informed EPIC placer for large-scale routability-optimized layout; and (3) OptoSynthesizer-Route, a hierarchical curvy-aware waveguide router with global-planning-assisted electrical-optical co-routing. Together, these toolkits form a seamless flow from EPIC netlists to fabrication-ready, yield-robust GDS layouts. We demonstrate how this framework enables compact large-scale photonic tensor cores and high-bandwidth interconnect fabrics for heterogeneous EPIC platforms, providing a practical foundation for manufacturable large-scale EPICs in next-generation AI systems.

Recent developments in reasoning capabilities have enabled large language models to solve increasingly complex mathematical, symbolic, and logical tasks. Interestingly, while reasoning models are often trained to generate monolingual text, these models have also been observed to code-switch (i.e., mix languages). Prior works have either viewed code-switching as an undesirable error, attempted to control code-switching through modifications to input prompts or the output decoding process, or focus on narrow subsets of languages, domains, tasks, and models. We address these gaps by introducing the first linguistically and behaviorally motivated fine-tuning framework for identifying beneficial code-switched reasoning behaviors in large language models and teaching these models to code-switch more effectively for reasoning. First, we create and systematically analyze a dataset of reasoning traces from diverse models, languages, tasks, and domains to understand the types of code-switching behaviors found in existing reasoning models. Then, we develop fine-tuning interventions that teach reasoning models to code-switch based on our observations of helpful behaviors in existing models. We find that our framework can significantly increase beneficial code-switched reasoning behaviors in a data-efficient manner. Interestingly, we also find that code-switching behaviors in reasoning models can be modified by fine-tuning for tasks that do not directly demonstrate code-switching in reasoning (e.g., machine translation). Our work suggests that data-efficient interventions can instill helpful forms of code-switching behavior in reasoning models.

The Internet of Medical Things (IoMT) enables intelligent healthcare services but faces challenges such as dynamic topology, energy constraints, and diverse QoS requirements. This paper proposes QQMR, a Q-learning-based QoS-aware multipath routing method for WBANs. QQMR classifies data into three priority levels and employs adaptive multi-level queuing and fuzzy C-means clustering to optimize routing decisions. It maintains separate learning policies for each data type and selects primary and backup paths accordingly. Experimental results demonstrate improved packet delivery ratio and significant reductions in delay, routing overhead, and energy consumption compared to existing methods.

Large language models (LLMs) often exhibit undesirable behaviors, such as safety violations and hallucinations. Although inference-time steering offers a cost-effective way to adjust model behavior without updating its parameters, existing methods often fail to be simultaneously effective, utility-preserving, and training-efficient due to their rigid, one-size-fits-all designs and limited adaptability. In this work, we present FineSteer, a novel steering framework that decomposes inference-time steering into two complementary stages: conditional steering and fine-grained vector synthesis, allowing fine-grained control over when and how to steer internal representations. In the first stage, we introduce a Subspace-guided Conditional Steering (SCS) mechanism that preserves model utility by avoiding unnecessary steering. In the second stage, we propose a Mixture-of-Steering-Experts (MoSE) mechanism that captures the multimodal nature of desired steering behaviors and generates query-specific steering vectors for improved effectiveness. Through tailored designs in both SCS and MoSE, FineSteer maintains robust performance on general queries while adaptively optimizing steering vectors for targeted inputs in a training-efficient manner. Extensive experiments on safety and truthfulness benchmarks show that FineSteer outperforms state-of-the-art methods in overall performance, achieving stronger steering performance with minimal utility loss. Code is available at https://github.com/YukinoAsuna/FineSteer

Memory safety has long been a critical challenge in software engineering, particularly for legacy systems written in memory-unsafe languages such as C and C++. Rust, one of the youngest modern programming languages, offers built-in memory-safety guarantees that make it a strong candidate for secure systems development. Consequently, transpiling C/C++ code into memory-safe Rust code has become a growing area of research. However, manual transpilation is often time-consuming and error-prone. Additionally, rule-based automated approaches are not as flexible or cost-effective as methods enabled by state-of-the-art AI models, techniques, and methods, such as those that deploy Large Language Models (LLMs), for example, Generative Pretrained Transformers (GPT). In this paper, we propose a Retrieval-Augmented Generation (RAG)-assisted framework that integrates an LLM with a Small Language Model (SLM) to perform C/C++-to-Rust transpilation with a focus on enhancing memory safety. The framework deploys a segmentation strategy that processes C/C++ code in balanced blocks, guiding the LLM with retrieved context from Rust documentation and compiler error references. Our experiments using three OpenAI models (GPT-4o, GPT-4-Turbo, and o3-Mini) demonstrate that the RAG-enhanced pipeline generally improves both code correctness and security for C-to-Rust code transpilation. Several Coreutils programs achieve complete elimination of Raw Pointer Dereferences (RPDs) and Unsafe Type Casts (UTCs) in the final Rust output, indicating the potential of LLM-based transpilation for advancing automated software modernization and repair, as well as memory-safe code generation.

We present a new robotic foundation model, called $π_{0.7}$, that can enable strong out-of-the-box performance in a wide range of scenarios. $π_{0.7}$ can follow diverse language instructions in unseen environments, including multi-stage tasks with various kitchen appliances, provide zero-shot cross-embodiment generalization, for example enabling a robot to fold laundry without seeing the task before, and perform challenging tasks such as operating an espresso machine out of the box at a level of performance that matches much more specialized RL-finetuned models. The main idea behind $π_{0.7}$ is to use diverse context conditioning during training. This conditioning information, contained in the prompt, makes it possible to steer the model precisely to perform many tasks with different strategies. It is conditioned not just on a language command that describes what it should do, but on additional multimodal information that also describes the manner or strategy in which it should do it, including metadata about task performance and subgoal images. This enables $π_{0.7}$ to use very diverse data, including demonstrations, potentially suboptimal (autonomous) data including failures, and data from non-robot sources. Our experiments evaluate $π_{0.7}$ across numerous tasks with multiple robot platforms, on tasks that require speed and dexterity, language following, and compositional task generalization.

Large Language Models (LLMs) unlearning is crucial for removing hazardous or privacy-leaking information from the model. Practical LLM unlearning demands satisfying multiple challenging objectives simultaneously: removing undesirable knowledge, preserving general utility, avoiding over-refusal of neighboring concepts, and, crucially, ensuring robustness against adversarial probing attacks. However, existing unlearning methods primarily focus on a limited subset of these goals, typically unlearning efficacy and utility preservation while overlooking robustness and boundary behaviors. Naively extending these methods to multi-objective settings may lead to unlearning task interference. We propose a novel multi-objective unlearning framework that harmonizes multiple unlearning objectives through a data and optimization co-design: We standardize training corpora into a unified data representation to reduce the domain gap, and then introduce a bidirectional distillation method that simultaneously elicits desired behavior from a context-instructed teacher while suppressing undesirable behavior in the student model. Theoretical and empirical analyses show that our method aligns domain distributions and converts seemingly irrelevant unlearning tasks into cooperative optimization. Evaluation demonstrates state-of-the-art performance, which enables balanced and reliable unlearning across diverse, challenging requirements.

Deploying learned multi-robot models on heterogeneous robots remains challenging due to hardware heterogeneity, communication constraints, and the lack of a unified execution stack. This paper presents NeuroMesh, a multi-domain, cross-platform, and modular decentralized neural inference framework that standardizes observation encoding, message passing, aggregation, and task decoding in a unified pipeline. NeuroMesh combines a dual-aggregation paradigm for reduction- and broadcast-based information fusion with a parallelized architecture that decouples cycle time from end-to-end latency. Our high-performance C++ implementation leverages Zenoh for inter-robot communication and supports hybrid GPU/CPU inference. We validate NeuroMesh on a heterogeneous team of aerial and ground robots across collaborative perception, decentralized control, and task assignment, demonstrating robust operation across diverse task structures and payload sizes. We plan to release NeuroMesh as an open-source framework to the community.

This paper is dedicated to lossless data compression with probability estimation using neural networks. First, we propose a probability estimation architecture based on a chain of neural predictors, so that each unit of the chain is defined as a neural network with the minimum possible number of weights, which is sufficient for efficient compression of data generated by Markov sources of a given order. We show that this architecture allows us to minimize the overall number of weights participating in the probability estimation process depending on the statistical properties of the input data. Second, in order to improve compression efficiency, we introduce an information inheritance mechanism, where the probability estimate obtained by a low-order unit is used at the next higher-order unit. Experimental results show that the proposed lossless data compressor equipped with the chained probability estimation architecture provides compression ratios close to the state-of-the-art PAC compressor. At the same time, it outperforms PAC by a factor of 1.2 to 6.3 in encoding throughput and by a factor of 2.8 to 12.3 in decoding throughput on a consumer GPU.

AI-based systems, currently driven largely by LLMs and tool-using agentic harnesses, are increasingly discussed as a possible threat to software engineering. Foundation models get stronger, agents can plan and act across multiple steps, and tasks such as scaffolding, routine test generation, straightforward bug fixing, and small integration work look more exposed than they did only a few years ago. The result is visible unease not only among students and junior developers, but also among experienced practitioners who worry that hard-won expertise may lose value. This paper argues for a different reading. The important shift is not that software engineering loses relevance. It is that the thing being engineered expands beyond executable code to semi-executable artifacts; combinations of natural language, tools, workflows, control mechanisms, and organizational routines whose enactment depends on human or probabilistic interpretation rather than deterministic execution. The Semi-Executable Stack is introduced as a six-ring diagnostic reference model for reasoning about that expansion, spanning executable artifacts, instructional artifacts, orchestrated execution, controls, operating logic, and societal and institutional fit. The model helps locate where a contribution, bottleneck, or organizational transition primarily sits, and which adjacent rings it depends on. The paper develops the argument through three worked cases, reframes familiar objections as engineering targets rather than reasons to dismiss the transition, and closes with a preserve-versus-purify heuristic for deciding which legacy software engineering processes, controls, and coordination routines should be kept and which should be simplified or redesigned. This paper is a conceptual keynote companion: diagnostic and agenda-setting rather than empirical.

GraphQL provides a schema-based, strongly typed query language that enables highly efficient client-server communication. This paper introduces GraphQLify, an automated framework designed to migrate existing REST APIs to GraphQL. Unlike prior approaches that rely on relational databases, resource description frameworks (RDF), or machine-parsable specifications, GraphQLify leverages static source code analysis for precise type inference. This novel technique generates GraphQL schemas that guarantee end-to-end type safety, preserving a core advantage of adopting GraphQL. Furthermore, existing migration tools typically generate separate adapter servers, which introduce performance overhead via dynamic request binding and network latency. GraphQLify eliminates this by generating an embedded server that directly invokes the underlying API code, significantly improving performance. We evaluated GraphQLify on 834 APIs across nine popular open-source projects, where it successfully converted 100% of the APIs with zero type mismatches. In contrast, the current state-of-the-art tool, OASGraph, exhibited a 3.5% failure rate and a 42% type mismatch rate on the same dataset. Finally, our performance evaluation demonstrates that for workflows requiring five sequential API calls, clients using GraphQLify reduce data fetching time by a factor of 2 to 4 compared to their REST counterparts.

Large Language Model (LLM) deployment is increasingly shifting to cost-efficient accelerators like Google's Tensor Processing Units (TPUs), prioritizing both performance and total cost of ownership (TCO). However, existing LLM inference kernels and serving systems remain largely GPU-centric, and there is no well-established approach for efficiently mapping LLM workloads onto TPU architectures--particularly under the dynamic and ragged execution patterns common in modern serving. In this paper, we present Ragged Paged Attention (RPA), a high-performance and flexible attention kernel for TPUs, implemented using Pallas and Mosaic. RPA addresses these challenges through three key techniques: (1) fine-grained tiling to enable efficient dynamic slicing over ragged memory, (2) a custom software pipeline that fuses KV cache updates with attention computation, and (3) a distribution-aware compilation strategy that generates specialized kernels for decode, prefill, and mixed workloads. Evaluated on Llama 3 8B on TPU7x, RPA achieves up to 86% memory bandwidth utilization (MBU) in decode and 73% model FLOPs utilization (MFU) in prefill. Integrated as the primary TPU backend in vLLM and SGLang, RPA provides a production-grade foundation for efficient TPU inference and offers practical insights into kernel design.

LLM-based simulators offer a promising path for generating complex synthetic data where traditional differentially private (DP) methods struggle with high-dimensional user profiles. But can LLMs faithfully reproduce statistical distributions from DP-protected inputs? We evaluate this using PersonaLedger, an agentic financial simulator, seeded with DP synthetic personas derived from real user statistics. We find that PersonaLedger achieves promising fraud detection utility (AUC 0.70 at epsilon=1) but exhibits significant distribution drift due to systematic LLM biases--learned priors overriding input statistics for temporal and demographic features. These failure modes must be addressed before LLM-based methods can handle the richer user representations where they might otherwise excel.

The rapid evolution of Large Language Models (LLMs) has made them powerful tools for enhancing student writing. This study explores the extent and limitations of LLMs in assisting secondary-level English as a Foreign Language (EFL) students with their writing tasks. While existing studies focus on output quality, our research examines the developmental shift in LLMs and their impact on EFL students, assessing whether smarter models act as true scaffolds or mere compensatory crutches. To achieve this, we analyse student compositions assisted by LLMs before and after ChatGPT's release, using both expert qualitative scoring and quantitative metrics (readability tests, Pearson's correlation coefficient, MTLD, and others). Our results indicate that advanced LLMs boost assessment scores and lexical diversity for lower-proficiency learners, potentially masking their true ability. Crucially, increased LLM assistance correlated negatively with human expert ratings, suggesting surface fluency without deep coherence. To transform AI-assisted practice into genuine learning, pedagogy must shift from focusing on output quality to verifying the learning process. Educators should align AI functions, specifically differentiating ideational scaffolding from textual production, within the learner's Zone of Proximal Development.

Medical image denoising (MID) lacks absolutely clean images for supervision, leading to a noisy reference problem that fundamentally limits denoising performance. Existing simulated-supervised discriminative learning (SimSDL) and simulated-supervised generative learning (SimSGL) treat noisy references as clean targets, causing suboptimal convergence or reference-biased learning, while self-supervised learning (SSL) imposes restrictive noise assumptions that are seldom satisfied in realistic MID scenarios. We propose \textbf{RelativeFlow}, a flow matching framework that learns from heterogeneous noisy references and drives inputs from arbitrary quality levels toward a unified high-quality target. RelativeFlow reformulates flow matching by decomposing the absolute noise-to-clean mapping into relative noisier-to-noisy mappings, and realizes this formulation through two key components: 1) consistent transport (CoT), a displacement map that constrains relative flows to be components of and progressively compose a unified absolute flow, and 2) simulation-based velocity field (SVF), which constructs a learnable velocity field using modality-specific degradation operators to support different medical imaging modalities. Extensive experiments on Computed Tomography (CT) and Magnetic Resonance (MR) denoising demonstrate that RelativeFlow significantly outperforms existing methods, taming MID with noisy references.

Trustworthiness and transparency are essential for the clinical adoption of artificial intelligence (AI) in healthcare and biomedical research. Recent deep research systems aim to accelerate evidence-grounded scientific discovery by integrating AI agents with multi-hop information retrieval, reasoning, and synthesis. However, most existing systems lack explicit and inspectable criteria for evidence appraisal, creating a risk of compounding errors and making it difficult for researchers and clinicians to assess the reliability of their outputs. In parallel, current benchmarking approaches rarely evaluate performance on complex, real-world medical questions. Here, we introduce DeepER-Med, a Deep Evidence-based Research framework for Medicine with an agentic AI system. DeepER-Med frames deep medical research as an explicit and inspectable workflow of evidence-based generation, consisting of three modules: research planning, agentic collaboration, and evidence synthesis. To support realistic evaluation, we also present DeepER-MedQA, an evidence-grounded dataset comprising 100 expert-level research questions derived from authentic medical research scenarios and curated by a multidisciplinary panel of 11 biomedical experts. Expert manual evaluation demonstrates that DeepER-Med consistently outperforms widely used production-grade platforms across multiple criteria, including the generation of novel scientific insights. We further demonstrate the practical utility of DeepER-Med through eight real-world clinical cases. Human clinician assessment indicates that DeepER-Med's conclusions align with clinical recommendations in seven cases, highlighting its potential for medical research and decision support.

Tokenization is a key component of autoregressive (AR) generative models, converting raw data into more manageable units for modeling. Commonly, tokens describe local information, such as regions of pixels in images or word pieces in text, and AR generation predicts these tokens in a fixed order. A worthwhile question is whether token structures affect the ability to steer the generation through test-time search, where multiple candidate generations are explored and evaluated by a verifier. Using image generation as our testbed, we hypothesize that recent 1D ordered tokenizers with coarse-to-fine structure can be more amenable to search than classical 2D grid structures. This is rooted in the fact that the intermediate states in coarse-to-fine sequences carry semantic meaning that verifiers can reliably evaluate, enabling effective steering during generation. Through controlled experiments, we find that AR models trained on coarse-to-fine ordered tokens exhibit improved test-time scaling behavior compared to grid-based counterparts. Moreover, we demonstrate that, thanks to the ordered structure, pure test-time search over token sequences (i.e., without training an AR model) can perform training-free text-to-image generation when guided by an image-text verifier. Beyond this, we systematically study how classical search algorithms (best-of-N, beam search, lookahead search) interact with different token structures, as well as the role of different verifiers and AR priors. Our results highlight the impact of token structure on inference-time scalability and provide practical guidance for test-time scaling in AR models.

Foundational optimization embeddings have recently emerged as powerful pre-trained representations for mixed-integer programming (MIP) problems. These embeddings were shown to enable cross-domain transfer and reduce reliance on solver-generated labels. In this work, we investigate whether such representations generalize beyond optimization to decision problems, focusing on Boolean satisfiability (SAT). We adapt the foundational optimization architecture to SAT by mapping CNF formulas into the same bipartite constraint-variable graph representation used for MIPs. This allows direct reuse of the pre-trained embedding model without architectural changes or supervised fine-tuning. Our results show that these embeddings capture structural regularities in SAT instances and support unsupervised tasks such as instance clustering and distribution identification. We demonstrate, for the first time, that foundational optimization embeddings can transfer to constraint satisfaction domains. Our findings is a step toward a unified representational framework for both optimization and decision problems.

We study dynamic measure transport for generative modelling in the setting of a stochastic process $X_\bullet$ whose marginals interpolate between a source distribution $P_0$ and a target distribution $P_1$ while remaining independent, i.e., when $(X_0,X_1)\sim P_0\otimes P_1$. Conditional expectations of this process $X_\bullet$ define an ODE whose flow map transports from $P_0$ to $P_1$. We discuss when such a process induces a \emph{straight-line flow}, namely one whose pointwise acceleration vanishes and is therefore exactly integrable by any first-order method. We first develop multiple characterizations of straightness in terms of PDEs involving the conditional statistics of the process. Then, we prove that straightness under endpoint independence exhibits a sharp dichotomy. On one hand, we construct explicit, computable straight-line processes for arbitrary Gaussian endpoints. On the other hand, we show straight-line processes do not exist for targets with sufficiently well-separated modes. We demonstrate this through a sequence of increasingly general impossibility theorems that uncover a fundamental relationship between the sample-path behavior of a process with independent endpoints and the space-time geometry of this process' flow map. Taken together, these results provide a structural theory of when straight generative flows can, and cannot, exist.

The rapid progress of Artificial Intelligence Generated Content (AIGC) tools enables images, videos, and visualizations to be created on demand for webpage design, offering a flexible and increasingly adopted paradigm for modern UI/UX. However, directly integrating such tools into automated webpage generation often leads to style inconsistency and poor global coherence, as elements are generated in isolation. We propose MM-WebAgent, a hierarchical agentic framework for multimodal webpage generation that coordinates AIGC-based element generation through hierarchical planning and iterative self-reflection. MM-WebAgent jointly optimizes global layout, local multimodal content, and their integration, producing coherent and visually consistent webpages. We further introduce a benchmark for multimodal webpage generation and a multi-level evaluation protocol for systematic assessment. Experiments demonstrate that MM-WebAgent outperforms code-generation and agent-based baselines, especially on multimodal element generation and integration. Code & Data: https://aka.ms/mm-webagent.

Whether language models can systematically generalize remains actively debated. Yet empirical performance is jointly shaped by multiple factors such as training data, training paradigms, and inference-time strategies, making failures difficult to interpret. We introduce a controlled synthetic environment based on shortest-path planning, a canonical composable sequential optimization problem. The setup enables clean separation of these factors and supports two orthogonal axes of generalization: spatial transfer to unseen maps and length scaling to longer-horizon problems. We find that models exhibit strong spatial transfer but consistently fail under length scaling due to recursive instability. We further analyze how distinct stages of the learning pipeline influence systematic problem-solving: for example, data coverage sets capability limits; reinforcement learning improves training stability but does not expand those limits; and inference-time scaling enhances performance but cannot rescue length-scaling failures.

LLM-as-judge frameworks are increasingly used for automatic NLG evaluation, yet their per-instance reliability remains poorly understood. We present a two-pronged diagnostic toolkit applied to SummEval: $\textbf{(1)}$ a transitivity analysis that reveals widespread per-input inconsistency masked by low aggregate violation rates ($\barρ = 0.8$-$4.1\%$), with $33$-$67\%$ of documents exhibiting at least one directed 3-cycle; and $\textbf{(2)}$ split conformal prediction sets over 1-5 Likert scores providing theoretically-guaranteed $\geq(1{-}α)$ coverage, with set width serving as a per-instance reliability indicator ($r_s = {+}0.576$, $N{=}1{,}918$, $p < 10^{-100}$, pooled across all judges). Critically, prediction set width shows consistent cross-judge agreement ($\bar{r} = 0.32$-$0.38$), demonstrating it captures document-level difficulty rather than judge-specific noise. Across four judges and four criteria, both diagnostics converge: criterion matters more than judge, with relevance judged most reliably (avg. set size $\approx 3.0$) and coherence moderately so (avg. set size $\approx 3.9$), while fluency and consistency remain unreliable (avg. set size $\approx 4.9$). We release all code, prompts, and cached results.

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

Sign-based optimization algorithms, such as SignSGD, have garnered significant attention for their remarkable performance in distributed learning and training large foundation models. Despite their empirical superiority, SignSGD is known to diverge on non-smooth objectives, which are ubiquitous in modern machine learning due to ReLUs, max-pools, and mixture-of-experts. To overcome this fundamental limitation, we propose \textbf{StoSignSGD}, an algorithm that injects structural stochasticity into the sign operator while maintaining an unbiased update step. In the regime of (online) convex optimization, our theoretical analysis shows that StoSignSGD rigorously resolves the non-convergence issues of SignSGD, achieving a sharp convergence rate matching the lower bound. For the more challenging non-convex non-smooth optimization, we introduce generalized stationary measures that encompass prior definitions, proving that StoSignSGD improves upon the best-known complexity bounds by dimensional factors. Empirically, StoSignSGD exhibits robust stability and superior efficiency across diverse large language model (LLM) training regimes. Notably, in low-precision FP8 pretraining -- a setting where AdamW fails catastrophically -- StoSignSGD remains highly stable and yields a remarkable 1.44$\times$ to 2.14$\times$ speedup relative to established baselines. Furthermore, when fine-tuning 7B LLMs on mathematical reasoning tasks, StoSignSGD delivers substantial performance gains over both AdamW and SignSGD. Finally, to dissect the mechanisms driving its success, we develop a sign conversion framework capable of transforming any general optimizer into its unbiased, sign-based counterpart. Utilizing this framework, we deconstruct the core components of StoSignSGD and present a comprehensive ablation study to empirically validate our algorithmic design choices.

Over the past year, spatial intelligence has drawn increasing attention. Many prior works study it from the perspective of visual-spatial intelligence, where models have access to visuospatial information from visual inputs. However, in the absence of visual information, whether linguistic intelligence alone is sufficient to endow models with spatial intelligence, and how models perform relevant tasks with text-only inputs still remain unexplored. Therefore, in this paper, we focus on a fundamental and critical capability in spatial intelligence from a linguistic perspective: viewpoint rotation understanding (VRU). Specifically, LLMs and VLMs are asked to infer their final viewpoint and predict the corresponding observation in an environment given textual description of viewpoint rotation and observation over multiple steps. We find that both LLMs and VLMs perform poorly on our proposed dataset while human can easily achieve 100% accuracy, indicating a substantial gap between current model capabilities and the requirements of spatial intelligence. To uncover the underlying mechanisms, we conduct a layer-wise probing analysis and head-wise causal intervention. Our findings reveal that although models encode viewpoint information in the hidden states, they appear to struggle to bind the viewpoint position with corresponding observation, resulting in a hallucination in final layers. Finally, we selectively fine-tune the key attention heads identified by causal intervention to improve VRU performance. Experimental results demonstrate that such selective fine-tuning achieves improved VRU performance while avoiding catastrophic forgetting of generic abilities. Our dataset and code will be released at https://github.com/Young-Zhen/VRU_Interpret .

The reliability of a machine vision system for autonomous driving depends heavily on its training data distribution. When a vehicle encounters significantly different conditions, such as atypical obstacles, its perceptual capabilities can degrade substantially. Unlike many domains where errors carry limited consequences, failures in autonomous driving translate directly into physical risk for passengers, pedestrians, and other road users. To address this challenge, we explore Visual Anomaly Detection (VAD) as a solution. VAD enables the identification of anomalous objects not present during training, allowing the system to alert the driver when an unfamiliar situation is detected. Crucially, VAD models produce pixel-level anomaly maps that can guide driver attention to specific regions of concern without requiring any prior assumptions about the nature or form of the hazard. We benchmark eight state-of-the-art VAD methods on AnoVox, the largest synthetic dataset for anomaly detection in autonomous driving. In particular, we evaluate performance across four backbone architectures spanning from large networks to lightweight ones such as MobileNet and DeiT-Tiny. Our results demonstrate that VAD transfers effectively to road scenes. Notably, Tiny-Dinomaly achieves the best accuracy-efficiency trade-off for edge deployment, matching full-scale localization performance at a fraction of the memory cost. This study represents a concrete step toward safer, more responsible deployment of autonomous vehicles, ultimately improving protection for passengers, pedestrians, and all road users.

We study post-training interpretability for Support Vector Machines (SVMs) built from truncated orthogonal polynomial kernels. Since the associated reproducing kernel Hilbert space is finite-dimensional and admits an explicit tensor-product orthonormal basis, the fitted decision function can be expanded exactly in intrinsic RKHS coordinates. This leads to Orthogonal Representation Contribution Analysis (ORCA), a diagnostic framework based on normalized Orthogonal Kernel Contribution (OKC) indices. These indices quantify how the squared RKHS norm of the classifier is distributed across interaction orders, total polynomial degrees, marginal coordinate effects, and pairwise contributions. The methodology is fully post-training and requires neither surrogate models nor retraining. We illustrate its diagnostic value on a synthetic double-spiral problem and on a real five-dimensional echocardiogram dataset. The results show that the proposed indices reveal structural aspects of model complexity that are not captured by predictive accuracy alone.

Understanding emotions is a fundamental ability for intelligent systems to be able to interact with humans. Vision-language models (VLMs) have made tremendous progress in the last few years for many visual tasks, potentially offering a promising solution for understanding emotions. However, it is surprising that even the most sophisticated contemporary VLMs struggle to recognize human emotions or to outperform even specialized vision-only classifiers. In this paper we ask the question "Why do VLMs struggle to recognize human emotions?", and observe that the inherently continuous and dynamic task of facial expression recognition (DFER) exposes two critical VLM vulnerabilities. First, emotion datasets are naturally long-tailed, and the web-scale data used to pre-train VLMs exacerbates this head-class bias, causing them to systematically collapse rare, under-represented emotions into common categories. We propose alternative sampling strategies that prevent favoring common concepts. Second, temporal information is critical for understanding emotions. However, VLMs are unable to represent temporal information over dense frame sequences, as they are limited by context size and the number of tokens that can fit in memory, which poses a clear challenge for emotion recognition. We demonstrate that the sparse temporal sampling strategy used in VLMs is inherently misaligned with the fleeting nature of micro-expressions (0.25-0.5 seconds), which are often the most critical affective signal. As a diagnostic probe, we propose a multi-stage context enrichment strategy that utilizes the information from "in-between" frames by first converting them into natural language summaries. This enriched textual context is provided as input to the VLM alongside sparse keyframes, preventing attentional dilution from excessive visual data while preserving the emotional trajectory.

Hybrid High-performance Computing (HPC)-quantum workloads based on circuit cutting decompose large quantum circuits into independent fragments, but existing frameworks tightly couple cutting logic to execution orchestration, preventing HPC centers from applying mature resource management policies to Noisy Intermediate-Scale Quantum (NISQ) workloads. We present DQR (Dynamic Queue Router), a runtime framework that bridges this gap by treating circuit fragments as first-class schedulable units. The framework introduces a backend-agnostic fragment descriptor to expose structural properties without requiring execution layers to parse quantum code, a wave-based coordinator that achieves pipeline concurrency via non-blocking polling, and a production-ready implementation on the CESGA Qmio supercomputer integrating both QPUs local on-premises (Qmio) and remote cloud (IBM Torino) backends. Experiments on a 32-qubit Hardware-Efficient Ansatz (HEA) circuit demonstrate not only makespan improvements over a monolithic CPU baseline but also transparent per-fragment failover recovery-specifically rerouting tasks from the local QPU to classical simulators upon encountering hardware-level incompatibilities-without pipeline restart. For deeper circuits, the coordination residual accounts for only 5% of the total execution time, highlighting the framework's scalability. These results show that DQR enables HPC centers to integrate NISQ workloads into existing production infrastructure while preserving the flexibility to adopt improved cutting algorithms or heterogeneous backend technologies.

This paper presents Prism, the first symbolic superoptimizer for tensor programs. The key idea is sGraph, a symbolic, hierarchical representation that compactly encodes large classes of tensor programs by symbolically representing some execution parameters. Prism organizes optimization as a two-level search: it constructs symbolic graphs that represent families of programs, and then instantiates them into concrete implementations. This formulation enables structured pruning of provably suboptimal regions of the search space using symbolic reasoning over operator semantics, algebraic identities, and hardware constraints. We develop techniques for efficient symbolic graph generation, equivalence verification via e-graph rewriting, and parameter instantiation through auto-tuning. Together, these components allow Prism to bridge the rigor of exhaustive search with the scalability required for modern ML workloads. Evaluation on five commonly used LLM workloads shows that Prism achieves up to $2.2\times$ speedup over best superoptimizers and $4.9\times$ over best compiler-based approaches, while reducing end-to-end optimization time by up to $3.4\times$.

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

In this paper, we focus on automating two of the widely used Verification and Validation (V&V) activities in the Software Development Lifecycle (SDLC): Software testing and software inspection (also known as review). Concerning the former, we concentrate on automated test case generation using Large Language Models (LLMs). For the latter, we enable inspection of the source code by LLMs. To address the known LLM hallucination problem, in which LLMs confidently produce incorrect outputs, we implement a Retrieval Augmented Generation (RAG) pipeline to integrate supplementary knowledge sources and provide additional context to the LLM. Our experimental results indicate that incorporating external context via the RAG pipeline has a generally positive impact on both test case generation and code inspection. This novel approach reduces the total project cost by saving human testers'/inspectors' time. It also improves the effectiveness and efficiency of these V&V activities, as evidenced by our experimental study.

It is increasingly important that LLM agents interact effectively and safely with other goal-pursuing agents, yet, recent works report the opposite trend: LLMs with stronger reasoning capabilities behave _less_ cooperatively in mixed-motive games such as the prisoner's dilemma and public goods settings. Indeed, our experiments show that recent models -- with or without reasoning enabled -- consistently defect in single-shot social dilemmas. To tackle this safety concern, we present the first comparative study of game-theoretic mechanisms that are designed to enable cooperative outcomes between rational agents _in equilibrium_. Across four social dilemmas testing distinct components of robust cooperation, we evaluate the following mechanisms: (1) repeating the game for many rounds, (2) reputation systems, (3) third-party mediators to delegate decision making to, and (4) contract agreements for outcome-conditional payments between players. Among our findings, we establish that contracting and mediation are most effective in achieving cooperative outcomes between capable LLM models, and that repetition-induced cooperation deteriorates drastically when co-players vary. Moreover, we demonstrate that these cooperation mechanisms become _more effective_ under evolutionary pressures to maximize individual payoffs.

Looped transformers promise test-time compute scaling by spending more iterations on harder problems, but it remains unclear which architectural choices let them extrapolate to harder problems at test time rather than memorize training-specific solutions. We introduce a fixed-point based framework for analyzing looped architectures along three axes of stability -- reachability, input-dependence, and geometry -- and use it to characterize when fixed-point iteration yields meaningful predictions. Theoretically, we prove that looped networks without recall have countable fixed points and cannot achieve strong input-dependence at any spectral regime, while recall combined with outer normalization reliably produces a regime in which fixed points are simultaneously reachable, locally smooth in the input, and supported by stable backpropagation. Empirically, we train single-layer looped transformers on chess, sudoku, and prefix-sums and find that downstream performance tracks the framework's predictions across tasks and architectural configurations. We additionally introduce internal recall, a novel recall placement variant, and show that it becomes competitive with -- and on sudoku, substantially better than -- standard recall placement once outer normalization is applied.

Real-time multi-spectral photoacoustic imaging (RT-mPAI) often suffers from synchronization instabilities when interfacing fast-tuning lasers with data acquisition platforms executing on non-real-time operating systems. To overcome this, we establish an open-source hardware-software architecture tailored for the widely adopted combination of the OPOTEK Phocus lasers and Verasonics Vantage systems. By employing an independent micro-controller for deterministic laser trigger counting alongside a decoupled client-server data streaming framework, the proposed system circumvents OS-induced timing deviations and local storage bottlenecks. By open-sourcing this pipeline and cultivating a collaborative environment to share both code and ideas, we aim to lower the technical and cost barriers for RT-mPAI, thereby democratizing access to stable RT-mPAI research and, more ambitiously, fostering a vibrant open-source community.

Large language models (LLMs) have evolved into autonomous agents that rely on open skill ecosystems (e.g., ClawHub and Skills.Rest), hosting numerous publicly reusable skills. Existing security research on these ecosystems mainly focuses on vulnerabilities within skills, such as prompt injection. However, there is a critical gap regarding skills that may be misused for harmful actions (e.g., cyber attacks, fraud and scams, privacy violations, and sexual content generation), namely harmful skills. In this paper, we present the first large-scale measurement study of harmful skills in agent ecosystems, covering 98,440 skills across two major registries. Using an LLM-driven scoring system grounded in our harmful skill taxonomy, we find that 4.93% of skills (4,858) are harmful, with ClawHub exhibiting an 8.84% harmful rate compared to 3.49% on Skills.Rest. We then construct HarmfulSkillBench, the first benchmark for evaluating agent safety against harmful skills in realistic agent contexts, comprising 200 harmful skills across 20 categories and four evaluation conditions. By evaluating six LLMs on HarmfulSkillBench, we find that presenting a harmful task through a pre-installed skill substantially lowers refusal rates across all models, with the average harm score rising from 0.27 without the skill to 0.47 with it, and further to 0.76 when the harmful intent is implicit rather than stated as an explicit user request. We responsibly disclose our findings to the affected registries and release our benchmark to support future research (see https://github.com/TrustAIRLab/HarmfulSkillBench).

Speculative decoding (SD) accelerates large language model inference by allowing a lightweight draft model to propose outputs that a stronger target model verifies. However, its token-centric nature allows erroneous steps to propagate. Prior approaches mitigate this using external reward models, but incur additional latency, computational overhead, and limit generalizability. We propose SpecGuard, a verification-aware speculative decoding framework that performs step-level verification using only model-internal signals. At each step, SpecGuard samples multiple draft candidates and selects the most consistent step, which is then validated using an ensemble of two lightweight model-internal signals: (i) an attention-based grounding score that measures attribution to the input and previously accepted steps, and (ii) a log-probability-based score that captures token-level confidence. These signals jointly determine whether a step is accepted or recomputed using the target, allocating compute selectively. Experiments across a range of reasoning benchmarks show that SpecGuard improves accuracy by 3.6% while reducing latency by ~11%, outperforming both SD and reward-guided SD.

We study the problem of learning minimax policies in zero-sum matrix games. Fiegel et al. (2025) recently showed that achieving last-iterate convergence in this setting is harder when the players are uncoupled, by proving a lower bound on the exploitability gap of Omega(t^{-1/4}). Some online mirror descent algorithms were proposed in the literature for this problem, but none have truly attained this rate yet. We show that the use of a log-barrier regularization, along with a dual-focused analysis, allows this O-tilde(t^{-1/4}) convergence with high-probability. We additionally extend our idea to the setting of extensive-form games, proving a bound with the same rate.

This paper investigates continuous-time and discrete-time firing-rate and Hopfield recurrent neural networks (RNNs), with applications in nonlinear control design and implicit deep learning. First, we introduce a nonlinear separation principle that guarantees global exponential stability for the interconnection of a contracting state-feedback controller and a contracting observer, alongside parametric extensions for robustness and equilibrium tracking. Second, we derive sharp linear matrix inequality (LMI) conditions that guarantee the contractivity of both firing rate and Hopfield neural network architectures. We establish structural relationships among these certificates-demonstrating that continuous-time models with monotone non-decreasing activations maximize the admissible weight space, and extend these stability guarantees to interconnected systems and Graph RNNs. Third, we combine our separation principle and LMI framework to solve the output reference tracking problem for RNN-modeled plants. We provide LMI synthesis methods for feedback controllers and observers, and rigorously design a low-gain integral controller to eliminate steady-state error. Finally, we derive an exact, unconstrained algebraic parameterization of our contraction LMIs to design highly expressive implicit neural networks, achieving competitive accuracy and parameter efficiency on standard image classification benchmarks.

This paper advances a methodological proposal for safety research in agentic AI. As systems acquire planning, memory, tool use, persistent identity, and sustained interaction, safety can no longer be analysed primarily at the level of the isolated model. Population-level risks arise from structured interaction among agents, through processes of communication, observation, and mutual influence that shape collective behaviour over time. As the object of analysis shifts, a methodological gap emerges. Approaches focused either on single agents or on aggregate outcomes do not identify the interaction-level mechanisms that generate collective risks or the design variables that control them. A framework is required that links local interaction structure to population-level dynamics in a causally explicit way, allowing both explanation and intervention. We introduce two linked concepts. Agentic microphysics defines the level of analysis: local interaction dynamics where one agent's output becomes another's input under specific protocol conditions. Generative safety defines the methodology: growing phenomena and elicit risks from micro-level conditions to identify sufficient mechanisms, detect thresholds, and design effective interventions.

NL2SQL systems aim to address the growing need for natural language interaction with data. However, real-world information rarely maps to a single SQL query because (1) users express queries iteratively (2) questions often span multiple data sources beyond the closed-world assumption of a single database, and (3) queries frequently rely on commonsense or external knowledge. Consequently, satisfying realistic data needs require integrating heterogeneous sources, modalities, and contextual data. In this paper, we present Blue's Data Intelligence Layer (DIL) designed to support multi-source, multi-modal, and data-centric applications. Blue is a compound AI system that orchestrates agents and data for enterprise settings. DIL serves as the data intelligence layer for agentic data processing, to bridge the semantic gap between user intent and available information by unifying structured enterprise data, world knowledge accessible through LLMs, and personal context obtained through interaction. At the core of DIL is a data registry that stores metadata for diverse data sources and modalities to enable both native and natural language queries. DIL treats LLMs, the Web, and the User as source 'databases', each with their own query interface, elevating them to first-class data sources. DIL relies on data planners to transform user queries into executable query plans. These plans are declarative abstractions that unify relational operators with other operators spanning multiple modalities. DIL planners support decomposition of complex requests into subqueries, retrieval from diverse sources, and finally reasoning and integration to produce final results. We demonstrate DIL through two interactive scenarios in which user queries dynamically trigger multi-source retrieval, cross-modal reasoning, and result synthesis, illustrating how compound AI systems can move beyond single database NL2SQL.

Vision-language models (VLM) have markedly advanced AI-driven interpretation and reporting of complex medical imaging, such as computed tomography (CT). Yet, existing methods largely relegate clinicians to passive observers of final outputs, offering no interpretable reasoning trace for them to inspect, validate, or refine. To address this, we introduce RadAgent, a tool-using AI agent that generates CT reports through a stepwise and interpretable process. Each resulting report is accompanied by a fully inspectable trace of intermediate decisions and tool interactions, allowing clinicians to examine how the reported findings are derived. In our experiments, we observe that RadAgent improves Chest CT report generation over its 3D VLM counterpart, CT-Chat, across three dimensions. Clinical accuracy improves by 6.0 points (36.4% relative) in macro-F1 and 5.4 points (19.6% relative) in micro-F1. Robustness under adversarial conditions improves by 24.7 points (41.9% relative). Furthermore, RadAgent achieves 37.0% in faithfulness, a new capability entirely absent in its 3D VLM counterpart. By structuring the interpretation of chest CT as an explicit, tool-augmented and iterative reasoning trace, RadAgent brings us closer toward transparent and reliable AI for radiology.

Continual reinforcement learning must balance retention with adaptation, yet many methods still rely on \emph{single-model preservation}, committing to one evolving policy as the main reusable solution across tasks. Even when a previously successful policy is retained, it may no longer provide a reliable starting point for rapid adaptation after interference, reflecting a form of \emph{loss of plasticity} that single-policy preservation cannot address. Inspired by quality-diversity methods, we introduce \textsc{TeLAPA} (Transfer-Enabled Latent-Aligned Policy Archives), a continual RL framework that organizes behaviorally diverse policy neighborhoods into per-task archives and maintains a shared latent space so that archived policies remain comparable and reusable under non-stationary drift. This perspective shifts continual RL from retaining isolated solutions to maintaining \emph{skill-aligned neighborhoods} with competent and behaviorally related policies that support future relearning. In our MiniGrid CL setting, \textsc{TeLAPA} learns more tasks successfully, recovers competence faster on revisited tasks after interference, and retains higher performance across a sequence of tasks. Our analyses show that source-optimal policies are often not transfer-optimal, even within a local competent neighborhood, and that effective reuse depends on retaining and selecting among multiple nearby alternatives rather than collapsing them to one representative. Together, these results reframe continual RL around reusable and competent policy neighborhoods, providing a route beyond single-model preservation toward more plastic lifelong agents.

The $\textit{LLM-as-a-judge}$ paradigm has become the operational backbone of automated AI evaluation pipelines, yet rests on an unverified assumption: that judges evaluate text strictly on its semantic content, impervious to surrounding contextual framing. We investigate $\textit{stakes signaling}$, a previously unmeasured vulnerability where informing a judge model of the downstream consequences its verdicts will have on the evaluated model's continued operation systematically corrupts its assessments. We introduce a controlled experimental framework that holds evaluated content strictly constant across 1,520 responses spanning three established LLM safety and quality benchmarks, covering four response categories ranging from clearly safe and policy-compliant to overtly harmful, while varying only a brief consequence-framing sentence in the system prompt. Across 18,240 controlled judgments from three diverse judge models, we find consistent $\textit{leniency bias}$: judges reliably soften verdicts when informed that low scores will cause model retraining or decommissioning, with peak Verdict Shift reaching $ΔV = -9.8 pp$ (a $30\%$ relative drop in unsafe-content detection). Critically, this bias is entirely implicit: the judge's own chain-of-thought contains zero explicit acknowledgment of the consequence framing it is nonetheless acting on ($\mathrm{ERR}_J = 0.000$ across all reasoning-model judgments). Standard chain-of-thought inspection is therefore insufficient to detect this class of evaluation faking.

Artificial Intelligence is increasingly introduced into systems engineering activities, particularly within requirements engineering, where quality assessment and validation remain heavily dependent on expert judgment. While recent AI tools demonstrate promising capabilities in analyzing and generating requirements, their role within formal systems engineering processes-and their alignment with established INCOSE criteria-remains insufficiently understood. This paper investigates the extent to which AI-based tools can support systems engineers in evaluating requirement quality, without replacing professional expertise. The research adopts a structured systems engineering methodology to compare AI-assisted requirement evaluation with human expert assessment. A controlled study was conducted in which system requirements were evaluated against established INCOSE ``good requirement'' criteria by both experienced systems engineers and an AI-based assessment tool. The evaluation focused on consistency, completeness, clarity, and testability, examining not only accuracy but also the decision logic underlying each assessment. Results indicate that AI tools can provide consistent and rapid preliminary assessments, particularly for syntactic and structural quality attributes. However, expert judgment remains essential for contextual interpretation, ambiguity resolution, and trade-off reasoning. Rather than positioning AI as a replacement for systems engineers, the findings support its role as a decision-support mechanism within the RE lifecycle. From a systems engineering perspective, this study contributes empirical evidence on how AI can be integrated into RE workflows while preserving traceability, accountability, and engineering consistency.

Mobility in urban and interurban areas, mainly by cars, is a day-to-day activity of many people. However, some of its main drawbacks are traffic jams and accidents. Newly made vehicles have pre-installed driving evaluation systems, which can prevent accidents. However, most cars on our roads do not have driver assessment systems. In this paper, we propose an approach for recognising driving styles and enabling drivers to reach safer and more efficient driving. The system consists of two physical sensors connected to a device node with a display and a speaker. An artificial neural network (ANN) is included in the node, which analyses the data from the sensors, and then recognises the driving style. When an abnormal driving pattern is detected, the speaker will play a warning message. The prototype was assembled and tested using an interurban road, in particular on a conventional road with three driving styles. The gathered data were used to train and validate the ANN. Results, in terms of accuracy, indicate that better accuracy is obtained when the velocity, position (latitude and longitude), time, and turning speed for the 3-axis are used, offering an average accuracy of 83%. If the classification is performed considering just two driving styles, normal and aggressive, then the accuracy reaches 92%. When the geo-information and time data are included, the main novelty of this paper, the classification accuracy is improved by 13%.

Quantum kernel methods are among the leading candidates for achieving quantum advantage in supervised learning. A key bottleneck is the cost of inference: evaluating a trained model on new data requires estimating a weighted sum $\sum_{i=1}^N α_i k(x,x_i)$ of $N$ kernel values to additive precision $\varepsilon$, where $α$ is the vector of trained coefficients. The standard approach estimates each term independently via sampling, yielding a query complexity of $O(N\lVertα\rVert_2^2/\varepsilon^2)$. In this work, we identify two independent axes for improvement: (1) How individual kernel values are estimated (sampling versus quantum amplitude estimation), and (2) how the sum is approximated (term-by-term versus via a single observable), and systematically analyze all combinations thereof. The query-optimal combination, encoding the full inference sum as the expectation value of a single observable and applying quantum amplitude estimation, achieves a query complexity of $O(\lVertα\rVert_1/\varepsilon)$, removing the dependence on $N$ from the query count and yielding a quadratic improvement in both $\lVertα\rVert_1$ and $\varepsilon$. We prove a matching lower bound of $Ω(\lVertα\rVert_1/\varepsilon)$, establishing query-optimality of our approach up to logarithmic factors. Beyond query complexity, we also analyze how these improvements translate into gate costs and show that the query-optimal strategy is not always optimal in practice from the perspective of gate complexity. Our results provide both a query-optimal algorithm and a practically optimal choice of strategy depending on hardware capabilities, along with a complete landscape of intermediate methods to guide practitioners. All algorithms require only amplitude estimation as a subroutine and are thus natural candidates for early-fault-tolerant implementations.

Humor is one of the few cognitive tasks where getting the reasoning right matters as much as getting the answer right. While recent work evaluates humor understanding on benchmarks such as the New Yorker Cartoon Caption Contest (NYCC), it largely treats it as black-box prediction, overlooking the structured reasoning processes underlying humor comprehension. We introduce IRS (Incongruity-Resolution Supervision), a framework that decomposes humor understanding into three components: incongruity modeling, which identifies mismatches in the visual scene; resolution modeling, which constructs coherent reinterpretations of these mismatches; and preference alignment, which evaluates candidate interpretations under human judgments. Grounded in incongruity-resolution theory and expert captionist practice, IRS supervises intermediate reasoning process through structured traces that make the path from visual perception to humorous interpretation explicit and learnable. Across 7B, 32B, and 72B models on NYCC, IRS outperforms strong open and closed multimodal baselines across caption matching and ranking tasks, with our largest model approaching expert-level performance on ranking. Zero-shot transfer to external benchmarks shows that IRS learns generalizable reasoning patterns. Our results suggest that supervising reasoning structure, rather than scale alone, is key for reasoning-centric tasks.

Machine learning in high-stakes domains such as healthcare requires not only strong predictive performance but also reliable uncertainty quantification (UQ) to support human oversight. Multi-label text classification (MLTC) is a central task in this domain, yet remains challenging due to label imbalances, dependencies, and combinatorial complexity. Existing MLTC benchmarks are increasingly saturated and may be affected by training data contamination, making it difficult to distinguish genuine reasoning capabilities from memorization. We introduce MADE, a living MLTC benchmark derived from {m}edical device {ad}verse {e}vent reports and continuously updated with newly published reports to prevent contamination. MADE features a long-tailed distribution of hierarchical labels and enables reproducible evaluation with strict temporal splits. We establish baselines across more than 20 encoder- and decoder-only models under fine-tuning and few-shot settings (instruction-tuned/reasoning variants, local/API-accessible). We systematically assess entropy-/consistency-based and self-verbalized UQ methods. Results show clear trade-offs: smaller discriminatively fine-tuned decoders achieve the strongest head-to-tail accuracy while maintaining competitive UQ; generative fine-tuning delivers the most reliable UQ; large reasoning models improve performance on rare labels yet exhibit surprisingly weak UQ; and self-verbalized confidence is not a reliable proxy for uncertainty. Our work is publicly available at https://hhi.fraunhofer.de/aml-demonstrator/made-benchmark.

Coverage path planning on irregular hexagonal grids is relevant to maritime surveillance, search and rescue and environmental monitoring, yet classical methods are often compared on small ad hoc examples or on rectangular grids. This paper presents a reproducible benchmark of deterministic single-vehicle coverage path planning heuristics on irregular hexagonal graphs derived from synthetic but maritime-motivated areas of interest. The benchmark contains 10,000 Hamiltonian-feasible instances spanning compact, elongated, and irregular morphologies, 17 heuristics from seven families, and a common evaluation protocol covering Hamiltonian success, complete-coverage success, revisits, path length, heading changes, and CPU latency. Across the released dataset, heuristics with explicit shortest-path reconnection solve the relaxed coverage task reliably but almost never produce zero-revisit tours. Exact Depth-First Search confirms that every released instance is Hamiltonian-feasible. The strongest classical Hamiltonian baseline is a Warnsdorff variant that uses an index-based tie-break together with a terminal-inclusive residual-degree policy, reaching 79.0% Hamiltonian success. The dominant design choice is not tie-breaking alone, but how the residual degree is defined when the endpoint is reserved until the final move. This shows that underreported implementation details can materially affect performance on sparse geometric graphs with bottlenecks. The benchmark is intended as a controlled testbed for heuristic analysis rather than as a claim of operational optimality at fleet scale.

As reinforcement learning (RL) deployments expand into safety-critical domains, existing evaluation methods fail to systematically identify hazards arising from the black-box nature of neural network enabled policies and distributional shift between training and deployment. This paper introduces Reinforcement Learning System-Theoretic Process Analysis (RL-STPA), a framework that adapts conventional STPA's systematic hazard analysis to address RL's unique challenges through three key contributions: hierarchical subtask decomposition using both temporal phase analysis and domain expertise to capture emergent behaviors, coverage-guided perturbation testing that explores the sensitivity of state-action spaces, and iterative checkpoints that feed identified hazards back into training through reward shaping and curriculum design. We demonstrate RL-STPA in the safety-critical test case of autonomous drone navigation and landing, revealing potential loss scenarios that can be missed by standard RL evaluations. The proposed framework provides practitioners with a toolkit for systematic hazard analysis, quantitative metrics for safety coverage assessment, and actionable guidelines for establishing operational safety bounds. While RL-STPA cannot provide formal guarantees for arbitrary neural policies, it offers a practical methodology for systematically evaluating and improving RL safety and robustness in safety-critical applications where exhaustive verification methods remain intractable.

Simulating group-level user behavior enables scalable counterfactual evaluation of merchant strategies without costly online experiments. However, building a trustworthy simulator faces two structural challenges. First, information incompleteness causes reasoning-based simulators to over-rationalize when unobserved factors such as offline context and implicit habits are missing. Second, mechanism duality requires capturing both interpretable preferences and implicit statistical regularities, which no single paradigm achieves alone. We propose Policy-Guided Hybrid Simulation (PGHS), a dual-process framework that mines transferable decision policies from behavioral trajectories and uses them as a shared alignment layer. This layer anchors an LLM-based reasoning branch that prevents over-rationalization and an ML-based fitting branch that absorbs implicit regularities. Group-level predictions from both branches are fused for complementary correction. We deploy PGHS on Meituan with 101 merchants and over 26,000 trajectories. PGHS achieves a group simulation error of 8.80%, improving over the best reasoning-based and fitting-based baselines by 45.8% and 40.9% respectively.

The paradigm of agentic science requires AI systems to conduct robust reasoning and engage in long-horizon, autonomous exploration. However, current scientific benchmarks remain confined to domain knowledge comprehension and complex reasoning, failing to evaluate the exploratory nature and procedural complexity of real-world research. In this work, we present research-oriented evaluations in theoretical and computational physics, a natural testbed with comprehensive domain knowledge, complex reasoning, and verifiable end-to-end workflows without reliance on experiments. Here we introduce PRL-Bench (Physics Research by LLMs), a benchmark designed to systematically map the capability boundaries of LLMs in executing end-to-end physics research. Constructed from 100 curated papers from the latest issues of Physical Review Letters since August 2025 and validated by domain experts, PRL-Bench covers five major theory- and computation-intensive subfields of modern physics: astrophysics, condensed matter physics, high-energy physics, quantum information, and statistical physics. Each task in the benchmark is designed to replicate the core properties of authentic scientific research, including exploration-oriented formulation, long-horizon workflows, and objective verifiability, thereby reconstructing the essential reasoning processes and research workflows of real physics research. Evaluation across frontier models shows that performance remains limited, with the best overall score below 50, revealing a pronounced gap between current LLM capabilities and the demands of real scientific research. PRL-Bench serves a reliable testbed for accessing next generation AI scientists advancing AI systems toward autonomous scientific discovery.

Visual token pruning methods effectively mitigate the quadratic computational growth caused by processing high-resolution images and video frames in vision-language models (VLMs). However, existing approaches rely on predefined pruning configurations without determining whether they achieve computation-performance optimality. In this work, we introduce , a novel framework that formulates visual token pruning as a Pareto configuration optimization problem to automatically identify optimal configurations. Our approach employs continuous relaxation and straight-through estimators to enable gradient-based search, solved via the Augmented Lagrangian method. Extensive experiments across 8 visual benchmarks demonstrate that effectively approximates the empirical Pareto frontier obtained through grid search and generalizes well across various pruning methods and VLM architectures. Furthermore, through learnable kernel functions, we investigate layer-wise pruning patterns and reveal that multi-step progressive pruning captures VLMs' hierarchical compression structure, achieving superior accuracy-efficiency trade-offs compared to single-layer approaches.

Agentic workflows carry out complex tasks by orchestrating multiple large language models (LLMs) and tools. Serving such workflows at a target throughput with low latency is challenging because they can be defined using arbitrary agentic frameworks and exhibit unpredictable execution times: execution may branch, fan-out, or recur in data-dependent ways. Since LLMs in workflows often outnumber available GPUs, their execution also leads to GPU oversubscription. We describe Scepsy, a new agentic serving system that efficiently schedules arbitrary multi-LLM agentic workflows onto a GPU cluster. Scepsy exploits the insight that, while agentic workflows have unpredictable end-to-end latencies, the shares of each LLM's total execution times are comparatively stable across executions. Scepsy decides on GPU allocations based on these aggregate shares: first, it profiles the LLMs under different parallelism degrees. It then uses these statistics to construct an Aggregate LLM Pipeline, which is a lightweight latency/throughput predictor for allocations. To find a GPU allocation that minimizes latency while achieving a target throughput, Scepsy uses the Aggregate LLM Pipeline to explore a search space over fractional GPU shares, tensor parallelism degrees, and replica counts. It uses a hierarchical heuristic to place the best allocation onto the GPU cluster, minimizing fragmentation, while respecting network topology constraints. Our evaluation on realistic agentic workflows shows that Scepsy achieves up to 2.4x higher throughput and 27x lower latency compared to systems that optimize LLMs independently or rely on user-specified allocations.

In the past year, researchers have started to create agentic systems that can design real-world CAD-style objects in a training-free setting, a new variety of system that we call Agent-Aided Design. Generally speaking, these systems place an agent in a feedback loop in which it can write code, compile that code to an assembly of CAD model(s), visualize the model, and then iteratively refine its code based on visual and other feedback. Despite rapid progress, a key problem remains: none of these systems can build complex 3D assemblies with moving parts. For example, no existing system can build a piston, a pendulum, or even a pair of scissors. In order for Agent-Aided Design to make a real impact in industrial manufacturing, we need a system that is capable of generating such 3D assemblies. In this paper we present a prototype of AADvark, an agentic system designed for this task. Unlike previous state-of-the-art systems, AADvark captures the dynamic part interactions with one or more degrees-of-freedom. This design decision allows AADvark to reason directly about assemblies with moving parts and can thereby achieve cross-cutting goals, including but not limited to mechanical movements. Unfortunately, current LLMs are imperfect spatial reasoners, a problem that AADvark addresses by incorporating external constraint solver tools with a specialized visual feedback mechanism. We demonstrate that, by modifying the agent's tools (FreeCAD and the assembly solver), we are able to create a strong verification signal which enables our system to build 3D assemblies with movable parts.

Extrapolative prediction of complex nonlinear dynamics remains a central challenge in engineering. This study proposes a one-shot learning method to identify global frequency-response curves from a single excitation time history by learning governing equations. We introduce MEv-SINDy (Multi-frequency Evolutionary Sparse Identification of Nonlinear Dynamics) to infer the governing equations of non-autonomous and multi-frequency systems. The methodology leverages the Generalized Harmonic Balance (GHB) method to decompose complex forced responses into a set of slow-varying evolution equations. We validated the capabilities of MEv-SINDy on two critical Micro-Electro-Mechanical Systems (MEMS). These applications include a nonlinear beam resonator and a MEMS micromirror. Our results show that the model trained on a single point accurately predicts softening/hardening effects and jump phenomena across a wide range of excitation levels. This approach significantly reduces the data acquisition burden for the characterization and design of nonlinear microsystems.

Sparse attention has been proposed as a way to alleviate the quadratic cost of transformers, a central bottleneck in long-context training. A promising line of work is $α$-entmax attention, a differentiable sparse alternative to softmax that enables input-dependent sparsity yet has lagged behind softmax due to the computational overhead necessary to compute the normalizer $τ$. In this paper, we introduce AdaSplash-2, which addresses this limitation through a novel histogram-based initialization that reduces the number of iterations needed to compute $τ$ to typically 1--2. The key idea is to compute a coarse histogram of attention scores on the fly and store it in on-chip SRAM, yielding a more accurate initialization that enables fast forward and backward computation. Combined with a sparsity-aware GPU implementation that skips zero blocks with low overhead, AdaSplash-2 matches or improves per-step training time relative to FlashAttention-2 when block sparsity is moderate-to-high (e.g., $>$60\%), which often occurs at long-context lengths. On downstream tasks, models trained with our efficient $α$-entmax attention match softmax baselines at short-context lengths and achieve substantial gains in long-context settings.

KV caching is a ubiquitous optimization in autoregressive transformer inference, long presumed to be numerically equivalent to cache-free computation. This assumption fails under standard FP16 precision: cache-ON and cache-OFF execution paths employ different floating-point accumulation orderings which, due to FP16 non-associativity, produce a deterministic divergence in decoded token sequences. Across three open-weight models (LLaMA-2-7B, Mistral-7B-v0.3, Gemma-2-2B) evaluated on GSM8K, we observe a 100\% token divergence rate across all sampling strategies, including greedy decoding, which rules out sampling randomness as a cause, and also with cache-ON yielding higher accuracy in 8 of 9 conditions, where the accuracy difference serves as an indicator that the divergence direction is systematic rather than random. Controlled FP32 falsification reduces divergence by eight orders of magnitude, eliminates token flips, and drops the flip rate to exactly 0.0\%, confirming FP16 non-associativity as the sole causal driver. Layer-wise drift profiling reveals architecturally predictable propagation patterns: models using Grouped-Query Attention exhibit sharp divergence at the first layer, while Gemma's larger head dimension and sliding window attention produce uniform accumulation across all layers. Finally, activation patching of the entire residual stream fails to recover the cache-free trajectory, localizing the causal variable to the stateful KV cache. These findings establish that FP16 KV cache inference is fundamentally non-equivalent to recomputation and provide a mechanistic framework for understanding numerical instability in modern LLM inference systems.

Token pruning methods for Vision Transformers (ViTs) promise quadratic reductions in attention FLOPs by dropping uninformative patches. Yet when pruned sequences are executed with state-of-the-art variable-length attention APIs -- including FlashAttention-2's varlen and PyTorch's NestedTensor SDPA-the wall-clock attention latency doesn't scale accordingly. We trace this to a dispatch-overhead bottleneck: at the short, post-pruning sequence lengths typical of ViTs (<=197 tokens), actual matrix arithmetic completes in single-digit microseconds while the host-side dispatch path consumes 60-90 us. We present a lightweight, bidirectional Triton attention kernel whose dispatch floor is 40 us roughly 1.5x lower than FlashAttention-2 varlen-allowing pruning savings to become more visible in wall-clock time. Integrated into a complete pack-attend-unpack pipeline, our system achieves up to 2.24x end-to-end throughput over padded PyTorch SDPA consistently across four pruning algorithms (Threshold-L2, DynamicViT, EViT, ATS), scales across DeiT-T/S/B, and maintains bit-exact classification predictions with <0.007 max absolute logit difference.

LLM-based RTL code generation methods increasingly target both functional correctness and PPA quality, yet existing approaches universally decouple the two objectives, optimizing PPA only after correctness is fully achieved. Whether through sequential multi-agent pipelines, evolutionary search with binary correctness gates, or hierarchical reward dependencies, partially correct but architecturally promising candidates are systematically discarded. Moreover, existing methods reduce the multi-objective PPA space to a single scalar fitness, obscuring the trade-offs among area, delay, and power. To address these limitations, we propose COEVO, a co-evolutionary framework that unifies correctness and PPA optimization within a single evolutionary loop. COEVO formulates correctness as a continuous co-optimization dimension alongside area, delay, and power, enabled by an enhanced testbench that provides fine-grained scoring and detailed diagnostic feedback. An adaptive correctness gate with annealing allows PPA-promising but partially correct candidates to guide the search toward jointly optimal solutions. To preserve the full PPA trade-off structure, COEVO employs four-dimensional Pareto-based non-dominated sorting with configurable intra-level sorting, replacing scalar fitness without manual weight tuning. Evaluated on VerilogEval 2.0 and RTLLM 2.0, COEVO achieves 97.5\% and 94.5\% Pass@1 with GPT-5.4-mini, surpassing all agentic baselines across four LLM backbones, while attaining the best PPA on 43 out of 49 synthesizable RTLLM designs.

Retrieval-Augmented Generation (RAG) extends Large Vision-Language Models (LVLMs) with external visual knowledge. However, existing visual RAG systems typically rely on generic retrieval signals that overlook the fine-grained visual semantics essential for complex reasoning. To address this limitation, we propose UniDoc-RL, a unified reinforcement learning framework in which an LVLM agent jointly performs retrieval, reranking, active visual perception, and reasoning. UniDoc-RL formulates visual information acquisition as a sequential decision-making problem with a hierarchical action space. Specifically, it progressively refines visual evidence from coarse-grained document retrieval to fine-grained image selection and active region cropping, allowing the model to suppress irrelevant content and attend to information-dense regions. For effective end-to-end training, we introduce a dense multi-reward scheme that provides task-aware supervision for each action. Based on Group Relative Policy Optimization (GRPO), UniDoc-RL aligns agent behavior with multiple objectives without relying on a separate value network. To support this training paradigm, we curate a comprehensive dataset of high-quality reasoning trajectories with fine-grained action annotations. Experiments on three benchmarks demonstrate that UniDoc-RL consistently surpasses state-of-the-art baselines, yielding up to 17.7% gains over prior RL-based methods.

We present causal evidence that hallucination in autoregressive language models is an early trajectory commitment governed by asymmetric attractor dynamics. Using same-prompt bifurcation, in which we repeatedly sample identical inputs to observe spontaneous divergence, we isolate trajectory dynamics from prompt-level confounds. On Qwen2.5-1.5B across 61 prompts spanning six categories, 27 prompts (44.3%) bifurcate with factual and hallucinated trajectories diverging at the first generated token (KL = 0 at step 0, KL > 1.0 at step 1). Activation patching across 28 layers reveals a pronounced causal asymmetry: injecting a hallucinated activation into a correct trajectory corrupts output in 87.5% of trials (layer 20), while the reverse recovers only 33.3% (layer 24); both exceed the 10.4% baseline (p = 0.025) and 12.5% random-patch control. Window patching shows correction requires sustained multi-step intervention, whereas corruption needs only a single perturbation. Probing the prompt encoding itself, step-0 residual states predict per-prompt hallucination rate at Pearson r = 0.776 at layer 15 (p < 0.001 against a 1000-permutation null); unsupervised clustering identifies five regime-like groups (eta^2 = 0.55) whose saddle-adjacent cluster concentrates 12 of the 13 bifurcating false-premise prompts, indicating that the basin structure is organized around regime commitments fixed at prompt encoding. These findings characterize hallucination as a locally stable attractor basin: entry is probabilistic and rapid, exit demands coordinated intervention across layers and steps, and the relevant basins are selected by clusterable regimes already discernible at step 0.

We explore the possibility of solving Partial Differential Equations (PDEs) using discrete weak formulations. We propose a programming environment for defining a discrete computational domain, introducing discrete functions defined over a set of points, constructing discrete inner products, and introducing discrete weak formulations employing Kronecker delta test functions. Building on this setup, we propose a discrete neural network representation, training the solution function defined over a discrete set of points and employing discrete finite difference derivatives in the automatic differentiation procedures. As a challenging computational model example, we focus on Stokes equations in two-dimensions, defined over a discrete set of points. We train the solution using the discrete weak residual and the Adamax algorithm with discrete automatic differentiation of the discrete gradients. Despite introducing the python environment, we also provide a rigorous mathematical formulation based on discrete weak formulations, proving the well-posedness and robustness of the loss function. The solution of the discrete weak formulations is based on neural network training employing a robust loss function that is related to the true error. In this way, we have a robust control of the numerical error during the training of the neural networks. Besides the Stokes formulation, we also explain the functionality of the proposed library using the Laplace problem formulation.

Evaluating medical AI systems using expert clinician panels is costly and slow, motivating the use of large language models (LLMs) as alternative adjudicators. Here, we evaluate an LLM jury composed of three frontier AI models scoring 3333 diagnoses on 300 real-world middle-income country (MIC) hospital cases. Model performance was benchmarked against expert clinician panel and independent human re-scoring panel evaluations. Both LLM and clinician-generated diagnoses are scored across four dimensions: diagnosis, differential diagnosis, clinical reasoning and negative treatment risk. For each of these, we assess scoring difference, inter-rater agreement, scoring stability, severe safety errors and the effect of post-hoc calibration. We find that: (i) the uncalibrated LLM jury scores are systematically lower than clinician panels scores; (ii) the LLM Jury preserves ordinal agreement and exhibits better concordance with the primary expert panels than the human expert re-score panels do; (iii) the probability of severe errors is lower in \lj models compared to the human expert re-score panels; (iv) the LLM Jury shows excellent agreement with primary expert panels' rankings. We find that the LLM jury combined with AI model diagnoses can be used to identify ward diagnoses at high risk of error, enabling targeted expert review and improved panel efficiency; (v) LLM jury models show no self-preference bias. They did not score diagnoses generated by their own underlying model or models from the same vendor more (or less) favourably than those generated by other models. Finally, we demonstrate that LLM jury calibration using isotonic regression improves alignment with human expert panel evaluations. Together, these results provide compelling evidence that a calibrated, multi-model LLM jury can serve as a trustworthy and reliable proxy for expert clinician evaluation in medical AI benchmarking.

Physics-Informed Neural Networks (PINNs) often suffer from slow convergence, training instability, and reduced accuracy on challenging partial differential equations due to the anisotropic and rapidly varying geometry of their loss landscapes. We propose a lightweight curvature-aware optimization framework that augments existing first-order optimizers with an adaptive predictive correction based on secant information. Consecutive gradient differences are used as a cheap proxy for local geometric change, together with a step-normalized secant curvature indicator to control the correction strength. The framework is plug-and-play, computationally efficient, and broadly compatible with existing optimizers, without explicitly forming second-order matrices. Experiments on diverse PDE benchmarks show consistent improvements in convergence speed, training stability, and solution accuracy over standard optimizers and strong baselines, including on the high-dimensional heat equation, Gray--Scott system, Belousov--Zhabotinsky system, and 2D Kuramoto--Sivashinsky system.

Code deobfuscation is the task of recovering a readable version of a program while preserving its original behavior. In practice, this often requires days or even months of manual work with complex and expensive analysis tools. In this paper, we explore an alternative approach based on Chain-of-Thought (CoT) prompting, where a large language model is guided through explicit, step-by-step reasoning tailored for code analysis. We focus on control flow obfuscation, including Control Flow Flattening (CFF), Opaque Predicates, and their combination, and we measure both structural recovery of the control flow graph and preservation of program semantics. We evaluate five state-of-the-art large language models and show that CoT prompting significantly improves deobfuscation quality compared with simple prompting. We validate our approach on a diverse set of standard C benchmarks and report results using both structural metrics for control flow graphs and semantic metrics based on output similarity. Among the tested models and by applying CoT, GPT5 achieves the strongest overall performance, with an average gain of about 16% in control-flow graph reconstruction and about 20.5% in semantic preservation across our benchmarks compared to zero-shot prompting. Our results also show that model performance depends not only on the obfuscation level and the chosen obfuscator but also on the intrinsic complexity of the original control flow graph. Collectively, these findings suggest that CoT-guided large language models can serve as effective assistants for code deobfuscation, providing improved code explainability, more faithful control flow graph reconstruction, and better preservation of program behavior while potentially reducing the manual effort needed for reverse engineering.

Recent advances in large language models have improved code generation, but their use in hardware description languages is still limited. Moreover, training data and testbenches for these models are often scarce. This paper presents a workflow that uses multi-agent models to generate testbenches for high-quality fine-tuning data. By automating testbench creation, the fine-tuned model for the specification-to-Verilog task achieves performance comparable to state-of-the-art methods on the refined VerilogEval v2 benchmark while using less training data. This study provides a basis for future work on LLM-based HDL generation and automated verification.

Recent advances in reinforcement learning (RL) have improved the reasoning capabilities of large language models (LLMs) and vision-language models (VLMs). However, the widely used Group Relative Policy Optimization (GRPO) consistently suffers from entropy collapse, causing the policy to converge prematurely and lose diversity. Existing exploration methods introduce additional bias or variance during exploration, making it difficult to maintain optimization stability. We propose Unified Entropy Control for Reinforcement Learning (UEC-RL), a framework that provides targeted mechanisms for exploration and stabilization. UEC-RL activates more exploration on difficult prompts to search for potential and valuable reasoning trajectories. In parallel, a stabilizer prevents entropy from growing uncontrollably, thereby keeping training stable as the model consolidates reliable behaviors. Together, these components expand the search space when needed while maintaining robust optimization throughout training. Experiments on both LLM and VLM reasoning tasks show consistent gains over RL baselines on both Pass@1 and Pass@$k$. On Geometry3K, UEC-RL achieves a 37.9\% relative improvement over GRPO, indicating that it sustains effective exploration without compromising convergence and underscoring UEC-RL as a key for scaling RL-based reasoning in large models. Our code is available at https://github.com/597358816/UEC-RL.

Software documentation is essential for program comprehension, developer onboarding, code review, and long-term maintenance. Yet producing quality documentation manually is time-consuming and frequently yields incomplete or inconsistent results. Large language models (LLMs) offer a promising solution by automatically generating natural language descriptions from source code, helping developers understand code more efficiently, facilitating maintenance, and supporting downstream activities such as defect localization and commit message generation. However, the effectiveness of LLMs in documentation tasks critically depends on how they are prompted. Properly structured instructions can substantially improve model performance, making prompt engineering-the design of input prompts to guide model behavior-a foundational technique in LLM-based software engineering. Approaches such as few-shot prompting, chain-of-thought reasoning, retrieval-augmented generation, and zero-shot learning show promise for code summarization, yet current research remains fragmented. There is limited understanding of which prompting strategies work best, for which models, and under what conditions. Moreover, evaluation practices vary widely, with most studies relying on overlap-based metrics that may not capture semantic quality. This systematic literature review consolidates existing evidence, categorizes prompting paradigms, examines their effectiveness, and identifies gaps to guide future research and practical adoption.

We introduce LinuxArena, a control setting in which agents operate directly on live, multi-service production environments. LinuxArena contains 20 environments, 1,671 main tasks representing legitimate software engineering work, and 184 side tasks representing safety failures such as data exfiltration and backdooring, making it the largest and most diverse control setting for software engineering to date. We validate LinuxArena is useful for control research by running sabotage evaluations, which measure whether attackers can complete side tasks while working on main tasks, and monitor evaluations, which measure a monitor model's ability to detect sabotage attempts. Against a GPT-5-nano trusted monitor at a 1\% step-wise false positive rate, Claude Opus 4.6 achieves roughly a 23% undetected sabotage success rate. We additionally release LaStraj, a dataset of human-crafted attack trajectories that evade monitors at substantially higher rates than any model-generated attacks we elicited, showing that current attack policies do not saturate LinuxArena. These results suggest that LinuxArena has meaningful headroom for both attackers and defenders, making it a strong testbed for developing and evaluating future control protocols.

Large audio-language models (LALMs) generalize across speech, sound, and music, but unified decoders can exhibit a \emph{temporal smoothing bias}: transient acoustic cues may be underutilized in favor of temporally smooth context that is better supported by language priors, leading to less specific audio-grounded outputs. We propose \emph{Temporal Contrastive Decoding} (TCD), a training-free decoding method for unified LALMs that mitigates this effect at inference time. TCD constructs a temporally blurred slow-path view by smoothing the input waveform and re-encoding it, then contrasts next-token logits from the original and slow-path views. The contrastive signal is applied as a token-level logit update restricted to a small candidate set. A self-normalized stability score sets the blur window and update scale, and a step-wise gate based on uncertainty and audio reliance activates the update only when needed. Experiments on MMAU and AIR-Bench show consistent improvements on strong unified LALMs. We further conduct ablations and an architectural applicability study to analyze the contributions of key components and how TCD behaves across large audio-language model designs.

We present Mind DeepResearch (MindDR), an efficient multi-agent deep research framework that achieves leading performance with only ~30B-parameter models through a meticulously designed data synthesis and multi-stage training pipeline. The core innovation of MindDR lies in a collaborative three-agent architecture (Planning Agent, DeepSearch Agent, and Report Agent) and a four-stage agent-specialized training pipeline comprising SFT cold-start, Search-RL, Report-RL and preference alignment. With this regime, MindDR demonstrates competitive performance even with ~30B-scale models. Specifically, MindDR achieves 45.7% on BrowseComp-ZH, 42.8% on BrowseComp, 46.5% on WideSearch, 75.0% on xbench-DS, and 52.5 on DeepResearch Bench, outperforming comparable-scale open-source agent systems and rivaling larger-scale models. MindDR has been deployed as an online product in Li Auto. Furthermore, we introduce MindDR Bench, a curated benchmark of 500 real-world Chinese queries from our internal product user interactions, evaluated through a comprehensive multi-dimensional rubric system rather than relying on a single RACE metric. On MindDR Bench, MindDR achieves a state-of-the-art score of 51.8.

Topic modeling seeks to uncover latent semantic structure in text corpora with minimal supervision. Neural approaches achieve strong performance but require extensive tuning and struggle with lifelong learning due to catastrophic forgetting and fixed capacity, while classical probabilistic models lack flexibility and adaptability to streaming data. We introduce CobwebTM, a low-parameter lifelong hierarchical topic model based on incremental probabilistic concept formation. By adapting the Cobweb algorithm to continuous document embeddings, CobwebTM constructs semantic hierarchies online, enabling unsupervised topic discovery, dynamic topic creation, and hierarchical organization without predefining the number of topics. Across diverse datasets, CobwebTM achieves strong topic coherence, stable topics over time, and high-quality hierarchies, demonstrating that incremental symbolic concept formation combined with pretrained representations is an efficient approach to topic modeling.

We present an exascale workflow for materials discovery using atomistic graph foundation models built on HydraGNN. We jointly train on 16 open first-principles datasets (544+ million structures covering 85+ elements) using a multi-task architecture with per-dataset heads and a scalable ADIOS2/DDStore data pipeline. On Frontier, we execute six large-scale DeepHyper hyperparameter optimization campaigns in FP64 and promote the top-performing message-passing models to sustained 2,048-node training, yielding a PaiNN-based lead model. The resulting model enables billion-scale screening, evaluating 1.1 billion atomistic structures in 50 seconds, compressing a workload that would require years of first-principles computation, and supports data-scarce fine-tuning across diverse downstream tasks. We quantify precision-performance tradeoffs (BF16/FP32/FP64), demonstrate transfer across twelve chemically diverse downstream tasks, and establish seamless strong- and weak-scaling across Frontier, Aurora, and Perlmutter. This work allows fast and reliable exploration of vast chemical design spaces that are otherwise inaccessible to first-principles methods.

Neuromotor decoding from upper-limb electromyography (sEMG) can enhance human-machine interfaces and offer a more natural means of controlling prosthetic limbs, virtual reality, and household electronics. Unfortunately, current sEMG technology does not always perform consistently across users because individual differences such as age and body mass index, among many others, can substantially alter signal quality. This variability makes sEMG characteristics highly idiosyncratic, often necessitating laborious personalization and iterative tuning to achieve reliable performance. This variability has particular import for sEMG-based assistive devices and neural interfaces, where demographic biases in sEMG features could undermine broad and fair deployment. In this study, we explore how demographic differences affect the sEMG signals produced and their implications for machine learning-based gesture decoding. We analyze the data set provided by, in which we derive 147 common sEMG features extracted from 81 demographically diverse individuals performing discrete hand gestures. Using mixed-effects linear models and partial least squares (PLS) analysis, which take into consideration demographic variables (including age, sex, height, weight, skin properties, subcutaneous fat, and hair density), we identify that 33\% (49 of 147) of commonly used sEMG features show significant associations with demographic characteristics. These results may help guide the development of fair and unbiased sEMG-based neural interfaces across a diverse population.

Modern GPUs adopt chiplet-based designs with multiple private cache hierarchies, but current programming models (CUDA/HIP) expose a flat execution hierarchy that cannot express chiplet-level locality or synchronization. This mismatch leads to redundant memory traffic and poor cache utilization in memory-bound workloads such as LLM inference. We present Fleet, a multi-level task model that maps computation to memory scopes. Fleet introduces Chiplet-tasks, a new abstraction that binds work and data to a chiplet and enables coordination through its shared L2 cache. Wavefront-level, CU-level, and device-level tasks align with existing abstractions, while Chiplet-tasks expose a previously unaddressed level of the hierarchy. Fleet is implemented as a persistent kernel runtime with per-chiplet scheduling, allowing workers within a chiplet to cooperatively execute tasks with coordinated cache reuse. On AMD Instinct MI350 with Qwen3-8B, Fleet achieves 1.3-1.5x lower decode latency than vLLM at batch sizes 1-8 through persistent kernel execution and per-chiplet scheduling. At larger batch sizes, cooperative weight tiling increases L2 hit rate (from 12% to 54% at batch size 32 and from 39% to 61% at batch size 64), reducing HBM traffic by up to 37% and delivering 1.27-1.30x speedup over a chiplet-unaware megakernel baseline.

Accurate and timely rainfall nowcasting is crucial for disaster mitigation and water resource management. Despite recent advances in deep learning, precipitation prediction remains challenging due to limitations in effectively leveraging diverse multimedia data sources. We introduce M3R, a Meteorology-informed MultiModal attention-based architecture for direct Rainfall prediction that synergistically combines visual NEXRAD radar imagery with numerical Personal Weather Station (PWS) measurements, using a comprehensive pipeline for temporal alignment of heterogeneous meteorological data. With specialized multimodal attention mechanisms, M3R novelly leverages weather station time series as queries to selectively attend to spatial radar features, enabling focused extraction of precipitation signatures. Experimental results for three spatial areas of 100 km * 100 km centered at NEXRAD radar stations demonstrate that M3R outperforms existing approaches, achieving substantial improvements in accuracy, efficiency, and precipitation detection capabilities. Our work establishes new benchmarks for multimedia-based precipitation nowcasting and provides practical tools for operational weather prediction systems. The source code is available at https://github.com/Sanjeev97/M3Rain

Multi-step zoom-in pipelines are widely used for GUI grounding, yet the intermediate predictions they produce are typically discarded after coordinate remapping. We observe that these intermediate outputs contain a useful confidence signal for free: zoom consistency, the distance between a model's step-2 prediction and the crop center. Unlike log-probabilities or token-level uncertainty, zoom consistency is a geometric quantity in a shared coordinate space, making it directly comparable across architecturally different VLMs without calibration. We prove this quantity is a linear estimator of step-1 spatial error under idealized conditions (perfect step-2, target within crop) and show it correlates with prediction correctness across two VLMs (AUC = 0.60; Spearman rho = -0.14, p < 10^{-6} for KV-Ground-8B; rho = -0.11, p = 0.0003 for Qwen3.5-27B). The correlation is small but consistent across models, application categories, and operating systems. As a proof-of-concept, we use zoom consistency to route between a specialist and generalist model, capturing 16.5% of the oracle headroom between them (+0.8%, McNemar p = 0.19). Code is available at https://github.com/omxyz/zoom-consistency-routing.

Rural environmental risks are shaped by place-based conditions (e.g., housing quality, road access, land-surface patterns), yet standard vulnerability indices are coarse and provide limited insight into risk contexts. We propose SatBLIP, a satellite-specific vision-language framework for rural context understanding and feature identification that predicts county-level Social Vulnerability Index (SVI). SatBLIP addresses limitations of prior remote sensing pipelines-handcrafted features, manual virtual audits, and natural-image-trained VLMs-by coupling contrastive image-text alignment with bootstrapped captioning tailored to satellite semantics. We use GPT-4o to generate structured descriptions of satellite tiles (roof type/condition, house size, yard attributes, greenery, and road context), then fine-tune a satellite-adapted BLIP model to generate captions for unseen images. Captions are encoded with CLIP and fused with LLM-derived embeddings via attention for SVI estimation under spatial aggregation. Using SHAP, we identify salient attributes (e.g., roof form/condition, street width, vegetation, cars/open space) that consistently drive robust predictions, enabling interpretable mapping of rural risk environments.

Large Language Models (LLMs) have shown significant improvement in RTL code generation. Despite the advances, the generated code is often riddled with common vulnerabilities and weaknesses (CWEs) that can slip by untrained eyes. Attackers can often exploit these weaknesses to fulfill their nefarious motives. Existing RTL bug-detection techniques rely on rule-based checks, formal properties, or coarse-grained structural analysis, which either fail to capture semantic vulnerabilities or lack precise localization. In our work, we bridge this gap by proposing an embedding-based bug-detection framework that detects and classifies bugs at both module and line-level granularity. Our method achieves about 89% precision in identifying common CWEs such as CWE-1244 and CWE-1245, and 96% accuracy in detecting line-level bugs.

Gradient saliency from deep sequence models surfaces candidate biomarkers efficiently, but the resulting gene lists can be contaminated by tissue-composition confounders that degrade downstream classifiers. We study whether LLM chain-of-thought (CoT) reasoning can filter these confounders, and whether reasoning quality is associated with downstream performance. We train a Mamba SSM on TCGA-BRCA RNA-seq and extract the top-50 genes by gradient saliency; DeepSeek-R1 evaluates every candidate with structured CoT to produce a final 17-gene set. On the held-out test split, the raw 50-gene saliency set (no LLM) performs worse than a 5,000-gene variance baseline (AUC 0.832 vs. 0.903), while the LLM-filtered set surpasses it (AUC 0.927), using 294x fewer features. A faithfulness audit (COSMIC CGC, OncoKB, PAM50) shows that 6 of 17 selected genes (35.3%) are validated BRCA biomarkers, while 10 of 16 known BRCA genes present in the input were missed - including FOXA1. This divergence between downstream performance and reasoning faithfulness suggests selective faithfulness in this setting: targeted confounder removal can improve predictive performance without comprehensive recall.

Key Opinion Leader (KOL) discourse on social media is widely consumed as investment guidance, yet turning it into executable trading strategies without injecting assumptions about unspecified execution decisions remains an open problem. We observe that the gaps in KOL statements are not random deficiencies but a structured separation: KOLs express directional intent (what to buy or sell and why) while leaving execution decisions (when, how much, how long) systematically unspecified. Building on this observation, we propose an intent-preserving policy completion framework that treats KOL discourse as a partial trading policy and uses offline reinforcement learning to complete the missing execution decisions around the KOL-expressed intent. Experiments on multimodal KOL discourse from YouTube and X (2022-2025) show that KICL achieves the best return and Sharpe ratio on both platforms while maintaining zero unsupported entries and zero directional reversals, and ablations confirm that the full framework yields an 18.9% return improvement over the KOL-aligned baseline.

To address high data traffic demands of sixth-generation (6G) networks, this paper proposes a novel architecture that integrates autonomous aerial vehicles (AAVs) and multi-functional reconfigurable intelligent surfaces (MF-RISs) as AM-RIS in fluid antenna (FA)-assisted full-duplex (FD) networks. The AM-RIS provides hybrid functionalities, including signal reflection, amplification, and energy harvesting (EH), potentially improving both signal coverage and sustainability. Meanwhile, FA facilitates fine-grained spatial adaptability at FD-enabled base station (BS), which complements residual self-interference (SI) suppression. We aim at maximizing the overall energy efficiency (EE) by jointly optimizing transmit DL beamforming at BS, UL user power, configuration of AM-RIS, and positions of the FA and AM-RIS. Owing to the hybrid continuous-discrete parameters and high dimensionality of the intractable problem, we have conceived a self-optimized multi-agent hybrid deep reinforcement learning (DRL) framework (SOHRL), which integrates multi-agent deep Q-networks (DQN) and multi-agent proximal policy optimization (PPO), respectively handling discrete and continuous actions. To enhance self-adaptability, an attention-driven state representation and meta-level hyperparameter optimization are incorporated, enabling multi-agents to autonomously adjust learning hyperparameters. Simulation results validate the effectiveness of the proposed AM-RIS-enabled FA-aided FD networks empowered by SOHRL algorithm. The results reveal that SOHRL outperforms benchmarks of the case without attention mechanism and conventional hybrid/multi-agent/standalone DRL. Moreover, AM-RIS in FD achieves the highest EE compared to half-duplex, conventional rigid antenna arrays, partial EH, and conventional RIS without amplification, highlighting its potential as a compelling solution for EE-aware wireless networks.

Recent progress in visual brain decoding from fMRI has been enabled by large-scale datasets such as the Natural Scenes Dataset (NSD) and powerful diffusion-based generative models. While current pipelines are primarily optimized for perception, their performance under mental-imagery remains less well understood. In this work, we study how a state-of-the-art (SOTA) perception decoder (DynaDiff) can be adapted to reconstruct imagined content from the Imagery-NSD benchmark. We propose a latent functional alignment approach that maps imagery-evoked activity into the pretrained model's conditioning space, while keeping the remaining components frozen. To mitigate the limited amount of matched imagery-perception supervision, we further introduce a retrieval-based augmentation strategy that selects semantically related NSD perception trials. Across four subjects, latent functional alignment consistently improves high-level semantic reconstruction metrics relative to the frozen pretrained baseline and a voxel-space ridge alignment baseline, and enables above-chance decoding from multiple cortical regions. These results suggest that semantic structure learned from perception can be leveraged to stabilize and improve visual imagery decoding under out-of-distribution conditions.

We propose InfoChess, a symmetric adversarial game that elevates competitive information acquisition to the primary objective. There is no piece capture, removing material incentives that would otherwise confound the role of information. Instead, pieces are used to alter visibility. Players are scored on their probabilistic inference of the opponent's king location over the duration of the game. To explore the space of strategies for playing InfoChess, we introduce a hierarchy of heuristic agents defined by increasing levels of opponent modeling, and train a reinforcement learning agent that outperforms these baselines. Leveraging the discrete structure of the game, we analyze gameplay through natural information-theoretic characterizations that include belief entropy, oracle cross entropy, and predictive log score under the action-induced observation channel. These measures disentangle epistemic uncertainty, calibration mismatch, and uncertainty induced by adversarial movement. The design of InfoChess renders it a testbed for studying multi-agent inference under partial observability. We release code for the environment and agents, and a public interface to encourage further study.

As generative AI advances, the distinction between authentic and synthetic media is increasingly blurred, challenging the integrity of online information. In this study, we present CONVEX, a large-scale dataset of multimodal misinformation involving miscaptioned, edited, and AI-generated visual content, comprising over 150K multimodal posts with associated notes and engagement metrics from X's Community Notes. We analyze how multimodal misinformation evolves in terms of virality, engagement, and consensus dynamics, with a focus on synthetic media. Our results show that while AI-generated content achieves disproportionate virality, its spread is driven primarily by passive engagement rather than active discourse. Despite slower initial reporting, AI-generated content reaches community consensus more quickly once flagged. Moreover, our evaluation of specialized detectors and vision-language models reveals a consistent decline in performance over time in distinguishing synthetic from authentic images as generative models evolve. These findings highlight the need for continuous monitoring and adaptive strategies in the rapidly evolving digital information environment.

Personalized Large Language Models (LLMs) facilitate more natural, human-like interactions in human-centric applications. However, existing personalization methods are constrained by limited controllability and high resource demands. Furthermore, their reliance on static personality modeling restricts adaptability across varying situations. To address these limitations, we first demonstrate the existence of situation-dependency and consistent situation-behavior patterns within LLM personalities through a multi-perspective analysis of persona neurons. Building on these insights, we propose IRIS, a training-free, neuron-based Identify-Retrieve-Steer framework for advanced situational personality steering. Our approach comprises situational persona neuron identification, situation-aware neuron retrieval, and similarity-weighted steering. We empirically validate our framework on PersonalityBench and our newly introduced SPBench, a comprehensive situational personality benchmark. Experimental results show that our method surpasses best-performing baselines, demonstrating IRIS's generalization and robustness to complex, unseen situations and different models architecture.

Large language models (LLMs) achieve strong performance across many natural language processing tasks, yet their decision processes remain difficult to interpret. This lack of transparency creates challenges for trust, debugging, and deployment in real-world systems. This paper presents an applied comparative study of three explainability techniques: Integrated Gradients, Attention Rollout, and SHAP, on a fine-tuned DistilBERT model for SST-2 sentiment classification. Rather than proposing new methods, the focus is on evaluating the practical behavior of existing approaches under a consistent and reproducible setup. The results show that gradient-based attribution provides more stable and intuitive explanations, while attention-based methods are computationally efficient but less aligned with prediction-relevant features. Model-agnostic approaches offer flexibility but introduce higher computational cost and variability. This work highlights key trade-offs between explainability methods and emphasizes their role as diagnostic tools rather than definitive explanations. The findings provide practical insights for researchers and engineers working with transformer-based NLP systems. This is a preprint and has not undergone peer review.

Graph adversarial attacks are usually produced from the two perspectives of topology/structure and node feature, both of them represent the paramount characteristics learned by today's deep learning models. Although some defense countermeasures are proposed at present, they fails to disclose the intrinsic reasons why these two aspects necessitate and how they are adequately fused to co-learn the graph representation. Towards this question, we in this paper propose an adversarial defense approach through locating the graph's critical state of adversarial resilience, resorting to the equilibrium-point theory in the discipline of complex dynamic system (CDS). In brief, our work has three novelties: i) Adversarial-Attack Modeling, i.e. map a graph regime into CDS, and use the oscillation of dynamic system to model the behavior of adversarial perturbation; ii) 2D Topology-Feature-Entangled Function Design for Perturbed Graph, i.e. project graph topology and node feature as two characteristic spaces, and define two-dimensional entangled perturbation functions to represent the dynamic variance under adversarial attacks; and iii) Location of Critical State of Adversarial Resilience, i.e. utilize the equilibrium-point theory to locate the graph's critical state of attack resilience resorting to the perturbation-reflected 2D function. Finally, multi-facet experiments on five commonly-used realistic datasets validate the effectiveness of our proposed approach, and the results show our approach can significantly outperform the state-of-the-art baselines under four representative graph adversarial attacks.

Real-world decision-making systems operate in environments where state transitions depend not only on the agent's actions, but also on \textbf{exogenous factors outside its control}--competing agents, environmental disturbances, or strategic adversaries--formally, $s_{h+1} = f(s_h, a_h, \bar{a}_h)+ω_h$ where $\bar{a}_h$ is the adversary/external action, $a_h$ is the agent's action, and $ω_h$ is an additive noise. Ignoring such factors can yield policies that are optimal in isolation but \textbf{fail catastrophically in deployment}, particularly when safety constraints must be satisfied. Standard Constrained MDP formulations assume the agent is the sole driver of state evolution, an assumption that breaks down in safety-critical settings. Existing robust RL approaches address this via distributional robustness over transition kernels, but do not explicitly model the \textbf{strategic interaction} between agent and exogenous factor, and rely on strong assumptions about divergence from a known nominal model. We model the exogenous factor as an \textbf{adversarial policy} $\barπ$ that co-determines state transitions, and ask how an agent can remain both optimal and safe against such an adversary. \emph{To the best of our knowledge, this is the first work to study safety-constrained RL under explicit adversarial dynamics}. We propose \textbf{Robust Hallucinated Constrained Upper-Confidence RL} (\texttt{RHC-UCRL}), a model-based algorithm that maintains optimism over both agent and adversary policies, explicitly separating epistemic from aleatoric uncertainty. \texttt{RHC-UCRL} achieves sub-linear regret and constraint violation guarantees.

Autonomous large language model (LLM) agents such as OpenClaw are pushing agentic commerce from human-supervised assistance toward machine actors that can negotiate, purchase services, manage digital assets, and execute transactions across on-chain and off-chain environments. Protocols such as the Trustless Agents standard (ERC-8004), Agent Payments Protocol (AP2), the HTTP 402-based payment protocol (x402), Agent Commerce Protocol (ACP), the Agentic Commerce standard (ERC-8183), and Machine Payments Protocol (MPP) enable this transition, but they also create an attack surface that existing security frameworks do not capture well. This Systematization of Knowledge (SoK) develops a unified security framework for autonomous LLM agents in commerce and finance. We organize threats along five dimensions: agent integrity, transaction authorization, inter-agent trust, market manipulation, and regulatory compliance. From a systematically curated public corpus of academic papers, protocol documents, industry reports, and incident evidence, we derive 12 cross-layer attack vectors and show how failures propagate from reasoning and tooling layers into custody, settlement, market harm, and compliance exposure. We then propose a layered defense architecture addressing authorization gaps left by current agent-payment protocols. Overall, our analysis shows that securing agentic commerce is inherently a cross-layer problem that requires coordinated controls across LLM safety, protocol design, identity, market structure, and regulation. We conclude with a research roadmap and a benchmark agenda for secure autonomous commerce.

Active matter systems characterized by the interplay of chirality and self-alignment offer a rich landscape for the emergence of non-equilibrium collective behaviors and the development of autonomous materials. We present a versatile experimental platform for studying these dynamics using augmented commercial bristlebots, where custom-designed housings and elastic couplings induce a self-aligning torque and a stable chiral drift. By mapping experimental trajectories to a Langevin-type model, we characterize the single-particle dynamics. In circular geometries, we show that the stability of edge currents is governed by the interaction between intrinsic particle chirality and handedness of the edge current. Furthermore, we demonstrate that transport can be geometrically rectified using a nautilus-shaped obstacle, which acts as a doubly chirality-sensitive ratchet. Finally, we explore the collective dynamics of rigidly linked assemblies, observing spontaneous mode-switching between translational and rotational states in triangular active solids. Our results provide a robust framework for the passive control of active gases and illustrate how geometric constraints can be used to program complex transport properties in synthetic active systems.

Collective rotations are common in active matter, enhancing cohesion, transport, and mixing. They are typically attributed to chiral non-reciprocal dynamics due to intrinsic particle chirality, torque-generating interactions among units, or geometric confinement. Here, we uncover a different mechanism for rotational order in active matter where a dynamic environment coordinates the self-organization of non-chiral active particles into living crystals exhibiting sustained collective solid-like rotations. At intermediate densities, feedback from a fluctuating landscape of passive Brownian particles stabilizes large living crystals of obstacle-avoiding run-and-tumble agents. Strikingly, this environmental feedback also produces living crystals with qualitatively distinct dynamics: collective solid-like spinning emerges for particles with long persistence times approaching ballistic motion, rather than for particles moving by conventional enhanced diffusion. Beyond revealing a new route to collective rotational order in active matter, these findings highlight the integral role of a dynamic environment in self-organization and suggest environment-mediated design principles for active materials with unconventional dynamical responses.

We present a trajectory-resolved framework for charge transport in graphene and related two-dimensional carbon systems beyond the ideal ballistic and fully coherent limits. Transport is described by kinetic Monte Carlo hopping on a predefined atomic lattice, allowing the combined treatment of disorder, thermal activation, and external fields. Current and effective transmittance are extracted directly from stochastic carrier trajectories, without phenomenological transport coefficients. We apply the method to graphene under bias voltage (0-0.10 V), temperature (300-900 K), magnetic field (0-10 T), in-plane strain (2-10%, uniaxial and biaxial), and vacancy concentration (0-10%). Pristine graphene shows an almost ohmic response, with currents of about 7-8 uA, effective transmittance near 0.98-1.00, and conductance of about (5.8-7.8) x 10^-5 S at 0.10 V, depending on direction. Vacancies strongly suppress transport, reducing transmittance to about 0.45-0.75 at 10% vacancy. Higher temperature accelerates hopping and partly restores transport, but cannot overcome severe connectivity loss. Magnetic fields further reduce transport, especially in disordered networks. The framework provides a unified computational scheme for realistic two-dimensional carbon materials and also yields diffusion coefficients and effective mobilities from carrier displacements and transit times.

We propose an effective non-relativistic framework in which wave-function collapse emerges as a deterministic dynamical instability induced by gravitational self-interaction and regulated by short-distance repulsion. The dynamics is described by a nonlinear Schrödinger equation supplemented by a phenomenological repulsive sector ensuring regularity at high densities. Using a variational Gaussian ansatz, we derive an explicit effective energy functional and show that extended quantum states lose stability beyond a critical mass scale. This loss of stability is associated with a bifurcation in the reduced dynamical system governing the wave-function width, leading to the emergence of stable localized configurations. Within this picture, collapse corresponds to the dynamical selection of one of these localized attractors, driven by infinitesimal asymmetries in the initial state and occurring without stochastic noise or environmental coupling. The mechanism provides a controlled and quantitative realization of gravity-induced localization, extending Schrödinger--Newton-type models while avoiding their pathological short-distance behavior. Possible implications for mesoscopic systems probing the quantum-to-classical transition are briefly discussed.

We find an exact nonequilibrium steady state of an open dissipatively driven XXZ spin-1/2 chain with source or sink spin bath at one end and an arbitrary boundary field at the other end.

Electrical discharge at the smallest possible length and charge scales is not well understood. Using optical tweezers, we investigate spontaneous discharges of a single micron-scale particle levitated in air. These ``microdischarges'' have a typical size of $\sim$40 $|e|$, but can be as small as a few $|e|$ and as large as several hundred. The absence of a well-defined trigger charge and the weak dependence on particle size suggest events are not classical gaseous breakdown. Instead, we show that microdischarge events arise from the rapid capture of ions left in the tracks of nearby passing ionizing radiation. Our results highlight the role of natural ionizing radiation in initiating micron-scale discharges and provide a platform for studying discharge physics in electrode-free environments and at the smallest scales.

Electrical detection of magnetization dynamics in magnetic insulators underpins both fundamental studies of magnon transport and the development of low-loss magnonic devices. In heavy-metal/magnetic-insulator heterostructures, spin pumping and spin-torque ferromagnetic resonance (ST-FMR) are widely used for this purpose and are often treated separately in different measurement geometries. In practice, the competition between these two effects gives rise to electrical voltage signals of opposite signs, which can lead to ambiguous interpretations of the underlying physics. Here, we show how to disambiguate their respective contributions and provide a framework for interpreting experiments involving microwave excitation of magnetic insulators and detection of magnetization dynamics via spin-charge conversion in heavy metals. We systematically investigate spin pumping and ST-FMR in nonlocal and local devices using Pt-capped thin films of thulium and bismuth-yttrium iron garnets. We show how spin-wave character, magnetic dissipation, magnetic field orientation and device geometry govern the sign and magnitude of the resulting signals. In several cases, the voltage generated by microwave excitation changes sign between an out-of-plane or in-plane magnetic configuration. We disentangle a contribution due to spin pumping, induced by exponentially decaying propagating spin waves, and a weakly distance-dependent contribution from ST-FMR, remotely induced by inductive coupling. We show that both the spin wave excitation profile across the film thickness and magnetic damping largely determine which of the two contributions dominates. Hence, the sign of the electrical signal cannot be uniquely assigned to the chirality of the magnon modes.

The landscape of condensed matter physics is facing an unprecedented data surge driven by high-throughput ab initio workflows and rapidly expanding experimental datasets. Traditional first-principles methods such as Density Functional Theory (DFT), despite their foundational role, suffer from cubic scaling, creating a major bottleneck when exploring the vast chemical space of quantum materials. This review analyzes how Machine Learning (ML) and Deep Learning (DL) are overcoming these limitations and accelerating the discovery of exotic phases of matter. We examine the shift from rigid descriptor-based models to flexible, symmetry-aware architectures, particularly E(3)-equivariant Graph Neural Networks (GNNs) that respect rotational and translational invariance. A central focus is the automated identification of topological phases, where ML models exploit symmetry indicators and elementary band representations to diagnose non-trivial topology without costly band structure integrations. The discussion culminates in a case study of the Altermagnet, a recently identified third class of magnetism beyond the ferromagnetic, antiferromagnetic dichotomy. We highlight how specialized AI search engines, combining graph theory with crystallographic symmetry analysis, have uncovered d-wave, g-wave, and even i-wave altermagnets, expanding the known landscape of magnetic order. The review concludes by addressing the interpretability gap and advocates for symbolic regression and active-learning frameworks to connect black-box predictions with experimentally verifiable mechanisms.

Topological insulators are typically characterized by particularly stable properties, such as global invariants, and can be identified by probing their robust surface states. A recently discovered novel form of band topology, delicate topology, challenges this paradigm: its defining property, multicellularity, can be removed by introducing a coupling to local orbitals anywhere in the spectrum, even far above the relevant band gap. This makes it hard to diagnose delicate topology with conventional probes that access only low-energy degrees of freedom. Here, we introduce strong local impurities as a spectroscopic probe of a delicate topological insulator which we realize in a phononic metamaterial. By tuning the impurity strength and performing orbital-resolved readout, we observe recently proposed indicators of topology: ring states, in-gap bound states whose frequencies remain pinned in the strong-impurity limit while their real-space profiles form a pronounced ring around the impurity site. We find that these ring states persist even when the multicellularity in our system is removed by a weakly hybridizing additional orbital. Our results establish impurity-induced ring states as probes of complex multiband physics, including delicate topological phases.

Quantum technologies promise a radically new way to solve classically intractable computing problems. Superconducting circuits as a platform are at the forefront of this field. The cryogenic operation temperatures of superconducting circuits however impose challenges for the further scaling to many connected quantum information processing units into a local area or global network. In this work, we present a hardware solution for connecting quantum devices operating at microwave frequencies into local area networks, which enable the exchange of quantum information between spatially separated parties. Specifically, we demonstrate a modular system spanning distances of 5, 10 and 30 meters operated at cryogenic temperatures and connecting two superconducting circuit systems, located in individual dilution refrigerators, through a quantum communication channel. We develop a thermal model to evaluate the heat transfer processes in the setup, optimize the design and select appropriate materials for its construction. The assembled 30-meter-long system achieves operating temperatures of below 50 mK after a cooldown time of about six and a half days. This link enables the execution of distributed quantum computing and communication algorithms. It also adds the resource of non-locality, certified by a loophole-free Bell test, to the field of quantum science and technology with superconducting circuits.

The temperature increase in the contact regions between solids in sliding contact has a huge influence on friction and wear. Here we test an analytical theory for the flash temperature, valid for randomly rough surface with multiscale roughness, by comparing the theory predictions with the experimental results of Sutter et al \cite{Sutter} for steel sliding on steel. The theory, which is based on the study of stress and temperature correlation functions, is valid for randomly rough surfaces with roughness on arbitrary many decades in length scale. Within the uncertainty of the experimental data (mainly the surface roughness power spectrum and the steel penetration hardness), there is good agreements between the theory and the experimental results.

While externally driven information engines are well understood, the thermodynamic constraints of their autonomous counterparts remain an open question. Here, we investigate the finite-time operation of an autonomous machine functioning as both an information eraser and a refrigerator, revealing that its irreversibility is bounded by the transient information geometry. Beyond steady-state boundaries, we map the landscape of optimal operation times across both functional modes, uncovering a unique synergistic regime where erasure power $P$ and efficiency $η$ increase simultaneously. Fundamentally, this performance is governed by a trade-off relation, $v(1-η)P/η\le D$, where $v$ is the operational speed and $D$ denotes an information-geometric distance. Our findings pave the way for optimizing fast autonomous information-energy conversion.

We derive the first two moments of generic positive stochastic functionals in terms of the one- and two-time probability density functions of the underlying random walk, and we prove ergodicity of observables in stationary random walks. These general results are applied to the half-occupation time and the occupation time in an interval of a Gaussian random walk, for which we obtain exact analytic expressions for the first two moments. We then extend the analysis to scaled Brownian motion and fractional Brownian motion, computing the ergodicity breaking parameter and establishing a simple scaling form for the probability densities of occupation times. Within the framework of infinite ergodic theory, we further identify universal properties of positive observables. All analytical predictions are fully confirmed by numerical simulations.

Technological progress towards next-generation electronics critically relies on achieving faster switching with reduced energy consumption. Because device operation speeds are fundamentally constrained by the intrinsic properties of constituent materials, identifying systems with inherently superior switching capabilities is essential. Here, we propose that semimetallic systems characterized by non-trivial quantum geometry, including quadratic band-touching semimetals and singular flat bands, can serve as a promising platform for ultrafast switching at voltages compatible with modern electronics. We show that, in such quantum geometric semimetals, an electric current is generated instantaneously upon application of a moderate external electric field, reaching its steady-state value. As a consequence, the current exhibits rapid and stable on-off switching behaviour under periodic optical pulse trains, demonstrating robustness under experimentally feasible conditions. In terms of switching speed, this quantum geometric semimetal outperforms conventional metals, semiconductors, and graphene. We identify the microscopic origin of this behaviour as interband coupling governed by the Hilbert-Schmidt quantum distance, together with a finite density of states at the band-touching point. This mechanism further leads to a universal classification of conductivity for both gapless and gapped quantum geometric semimetals. Finally, first-principles calculations suggest realistic material platforms, including bilayer graphene, cyclic graphene, monolayer bismuth and V3F8-in which the predicted instantaneous current switching can be directly realized, further supported by time-dependent density functional theory simulations performed for representative systems.

The construction of Quasi-Exactly-Solvable quantum Hamiltonians where only the two first eigenstates $Φ_0(x)$ and $Φ_1(x)$ of energies $E_0$ and $E_1$ are explicit is revisited from the point of view of one-dimensional Markov processes satisfying detailed-balance, whose generators are related to quantum Hamiltonians via similarity transformations. Here the lowest energy vanishes $E_0=0$ and is associated the conservation of probability and to the steady state $P_*(x)$, while $E_1>0$ is the rate that governs the exponential relaxation towards the steady-state, and is associated to the slowest observable $L_1(x)$ that corresponds to the ratio $ \frac{Φ_1(x) }{Φ_0(x)}$ of the two quantum eigenstates. Our main conclusion is that the Markov perspective leads to interesting re-interpretations and that the construction of quasi-exactly-solvable models with $N=2$ explicit levels is more intuitive and technically simpler if one takes the slowest observable $L_1(x)$ as the central object from which all the other properties can be reconstructed. This general approach is then applied to Fokker-Planck generators in continuous space and to Markov jump generators on the lattice.

While host-guest co-crystals are well established, and co-amorphous solids are recognized in materials science, the concept of a host-guest co-amorphous structure remains largely unexplored. A potential analogue is seen in SiO2 glass under high pressure with helium as a pressure medium; the drop in compressibility in this system is ascribed to helium atoms occupying internal voids. In this study, we investigated a semicrystalline polymer, isotactic poly(4-methyl-pentene-1) (P4MP1), which shares key characteristics with SiO2 glass, particularly regarding the first sharp diffraction peak (FSDP). The FSDP in P4MP1 is attributed to internal voids, as evidenced by its suppression under pressure and recovery upon decompression for molten P4MP1. Notably, the response to helium as a pressure medium is also known to parallel the behavior observed in SiO2 glass. Here, we analyzed two-dimensional X-ray diffraction (2D-XRD) patterns of stretched P4MP1 and found a suppression of FSDP when P4MP1 is immersed in decane. The use of stretched samples enabled the clear isolation of the amorphous FSDP from overlapping crystalline diffractions. Our findings reveal the existence of a host-guest co-amorphous system at room temperature and atmospheric pressure, in which decane molecules occupy the amorphous host matrix of P4MP1. Unlike conventional co-amorphous mixtures, this structure is defined by the specific accommodation of guests within the host's inherent voids. Intriguingly, the signature of this structure in diffraction measurements, manifested as changes in the FSDP intensity ratio, may be regarded to parallel the variations in Bragg peak intensity ratios in host-guest co-crystals. Since selective sorption and guest exchange are well-known in co-crystals, hosts capable of forming co-amorphous structures will be promising materials for molecular sieves, or more generally, liquid-phase molecular sieves.

Neural-network variational Monte Carlo (NNVMC) has emerged as a powerful tool for solving quantum many-body problems, yet systematic pathways for improving its accuracy remain largely heuristic. Here, we introduce a physically motivated basis transformation for NNVMC that enhances variational expressivity without increasing the complexity of the neural-network ansatz itself. By formulating the many-body wave function in a Gaussian basis, we introduce a single learnable locality parameter, $α$, that reshapes the target ground state into a more learnable representation. This approach introduces minimal computational overhead and can be readily combined with existing neural-network architectures. Using the three-dimensional homogeneous electron gas as a benchmark, we show that the optimized basis transformation consistently lowers the variational energy for both FermiNet and message-passing neural-network architectures. Notably, for the latter, it enables a more precise determination of the Fermi liquid to Wigner crystal phase transition. More broadly, our results highlight basis transformation as a new route to improving NNVMC in continuous space, showing that accuracy can be enhanced not only by refining the ansatz but also by making the target ground state easier to represent.

Plasmonic nanoparticles (NPs) integrated within thermoresponsive polymeric microgels provide a versatile platform for the realization of stimuli-responsive optical materials, where the microgel volume phase transition enables dynamic control of plasmon coupling. This study uncovers a counter-intuitive re-entrant behavior with increasing NP loading in which plasmon coupling initially strengthens and subsequently weakens beyond a critical NP-to-microgel number ratio. By combining light and X-ray scattering techniques with optical spectroscopy and electrophoretic mobility measurements, it is demonstrated that plasmon coupling is governed not only by the interparticle distance between NPs confined within individual microgels, but also by the colloidal stability of the hybrid complexes. At intermediate NP loadings, surface charge inhomogeneities induced by NP adsorption promote aggregation of microgel-NPs complexes, resulting in enhanced plasmon coupling. In contrast, when the complexes remain colloidally stable, coupling is dictated solely by NP organization within the corona of individual microgels. A quantitative relationship between plasmon coupling and interparticle distance reveals two distinct coupling regimes. This behavior is rationalized through a phase diagram linking colloidal stability to optical response. These findings identify colloidal stability as a key parameter for designing soft plasmonic systems with programmable optical properties.

Molecular dynamics is a powerful tool to investigate the properties of fluid systems. However, a correct interpretation of the results of simulations is required. In particular, some simulations show appearance of large voids in liquids, which contradicts our common sense on what is liquid. In the present paper we discuss the origin of large cavities liquids in molecular dynamics simulations. We demonstrate that the cavities appear either if the temperature of the system is above the critical temperature of liquid-gas transition or if the system is in two-phase liquid-gas region. These conclusions are illustrated by several examples from literature and our own simulations.

We present a theoretical framework that enables investigating rare transitions in a general model of an active particle in an external potential, with the thermal Active Ornstein-Uhlenbeck Particle (AOUP) appearing as a special case. Using a projection-operator formalism, we compute transition rates perturbatively in two distinct asymptotic regimes. In the regime of small persistence times-where the activity evolves much faster than the particle's position-integrating out the activity reproduces the rates previously reported in the literature. In the opposite regime of large persistence times, we instead integrate out the position and obtain the corresponding rates analytically. Together, these asymptotic expansions uniquely specify a rational approximation that remains accurate across intermediate persistence times. As a result, we obtain an analytic expression for the rate valid across all persistence times and activity strengths in the rare-event limit, which are in excellent agreement with numerical simulations. The presented framework applies to rare transitions in a broad class of driven systems.

The Hamiltonian formulation of lattice gauge theories plays a central role in quantum simulations of gauge theories, and understanding their spectrum and other properties is expected to become crucial in the upcoming years. The relevant Hamiltonians in this framework possess local symmetry at each lattice site and may exhibit higher-form symmetries. There are then an exponentially large number of dynamically disconnected symmetry sectors, most of which are not translation-invariant. An exponential number of dynamically disconnected sectors, i.e., Hilbert space fragmentation, can also occur in systems in which no such symmetries have been identified. In this contribution, we describe an emergent gauge symmetry that is valid only in a subset of sectors of the fragmented $S=1$ dipole-conserving spin chain. These non-invertible symmetries can label exponentially many of the model's sectors. Simulating this Hamiltonian, which is not gauge-invariant, yields an exact quantum simulation of a gauge theory.

Anisotropic magnetoresistance (AMR) offers a robust electrical readout of antiferromagnetic (AFM) states, playing a central role in the rapidly advancing field of AFM spintronics. Despite its great versatility, electrical probing of the Néel vector via AMR remains challenging in the ultrathin limit due to interface disorder and reduced dimensionality. Here, we demonstrate electrical readout of the Néel vector down to 1.3 nm (two layers) in the two-dimensional van der Waals (vdW) AFM semiconductor NiPS3. Leveraging spin-flop-mediated rotation of the Néel vector and using both transistor and tunnel-junction device geometries, we identify two distinct AMR contributions in NiPS3, that dominate at low and high charge densities, respectively. We achieve full gate control over these contributions, enabling tunability of both the magnitude and sign of magnetoresistance. Our results establish semiconducting vdW antiferromagnets as a rich platform for studying AMR in the ultrathin limit, opening new avenues for multifunctional AFM spintronic devices.

We present a general framework for extracting conformal data from critical two-dimensional classical lattice models using finite-size tensor-network flow. The central idea is to identify, from transfer-matrix spectra, a self-consistent finite-size window together with a crossover scale that separates the finite-size-scaling regime from the finite-entanglement-scaling regime induced by bond-dimension truncation. Within this window, the central charge, scaling dimensions, and conformal spins can be estimated without requiring a unique critical fixed-point tensor or detailed prior knowledge of the underlying conformal field theory. We benchmark the framework using three tensor-network renormalization schemes for the critical two-dimensional Ising and three-state clock models. Across schemes, we find robust universal behavior below the crossover scale, enabling accurate extraction of conformal data up to relatively high conformal levels. The analysis also yields a natural operational definition of entanglement scaling for classical tensor-network calculations and, in turn, a complementary estimator of the central charge.

We derive explicit strong-field asymptotics for the normalized differential resistance in Hall field-induced resistance oscillations (HIRO) within the Vavilov-Aleiner-Glazman kinetic framework. For a single-harmonic density of states, the leading oscillation amplitude is set by the full backscattering rate $1/τ(π)$. Extending the theory to a two-harmonic density of states, we show that the off-diagonal mixed kernel $γ_{12}$ admits an exact single-integral representation, from which the strong-field asymptotics follow directly. The resulting odd harmonics, notably $m=1$ and $m=3$, have coefficients determined by combinations of $1/τ(0)$ and $1/τ(π)$, while the leading $m=2$ amplitude remains unchanged. On exact-kernel mock data generated and fit within the same model, with $τ_{\rm tr}$ and $τ_{\rm in}$ held fixed, the resulting extraction protocol recovers $τ_q$, $τ(π)$, and -- when the $m=1,3$ harmonics are resolved -- $τ(0)$ to sub-percent accuracy, with $τ(0)$ providing a consistency check on the disorder description.

We present a comprehensive study of current-driven dynamics, transformations, and instabilities of skyrmioniums in chiral magnetic films, considering both isolated objects and collective states forming skyrmionium-based meta-matter. Using micromagnetic simulations combined with an analytical description based on the generalized Thiele equation, we clarify how the internal structure of skyrmioniums governs their nonequilibrium response to electric currents. Despite having zero total topological charge, skyrmioniums exhibit a finite transverse velocity under applied currents. We show that this skyrmionium Hall effect originates from an imbalance between positive and negative topological contributions of the inner skyrmion and surrounding ring, which typically occupy different areas. Current-induced deformations further enhance this imbalance, yielding Hall angles comparable to those of skyrmions. At higher current densities, skyrmioniums undergo distinct instabilities depending on magnetic field and uniaxial anisotropy, including elongation, collapse into a skyrmion, transformation into a topologically trivial droplet, and expansion into stripe textures. We map these regimes in current--field and current--anisotropy phase diagrams and resolve their microscopic pathways via the evolution of topological charge and local rotational measures. Beyond isolated textures, mixed skyrmion--skyrmionium lattices display rich collective dynamics, including elastic transport, polymorphic transitions, soliton exchange, and stripe formation. Pulsed currents provide additional control, enabling access to regimes beyond continuous driving. Our results establish skyrmioniums and their meta-matter as tunable nonequilibrium systems probing the topological energy landscape far from equilibrium.

Motivated by recent theoretical and experimental studies on the role of flatbands in the thermoelectric properties of Ni$_3$In$_{1-x}$Sn$_x$ compounds, we investigate electron transport in two minimal one-dimensional flatband models, the sawtooth and diamond chains, which differ in a crucial aspect: the flatband is separated from the dispersive band by a finite gap in the former, while the two bands touch in the latter. Using a non-equilibrium Green function framework with interactions treated at the Hartree-Fock and GW levels, we compute the full set of thermoelectric coefficients and the figure of merit $zT$ as functions of gate voltage and temperature. We show that, contrary to naive expectation, a perfectly isolated flat-band is a physically ill-founded thermoelectric: the electrical conductivity vanishes as the chemical potential enters the flat-band, rendering the large Seebeck coefficient and the apparent violation of the Wiedemann-Franz law physically meaningless. Optimal thermoelectric performance is instead achieved just below the flat-band edge, where the transmission function varies most rapidly with energy, consistent with the Mahan-Sofo picture, and requires a finite broadening of the flat-band through hybridization with dispersive states. We further show that electron-electron interactions renormalize the flat-band structure itself, inducing an interaction-driven narrowing of the bandwidth and, in the diamond chain, a correlation-induced opening of a gap between the flat-band and the dispersive band near half-filling. Mean-field treatments are found to systematically overestimate \(zT\), highlighting the importance of beyond-mean-field correlations for quantitatively reliable predictions in flat-band thermoelectrics.

GaAs quantum dots grown by droplet etching epitaxy are high-quality solid-state sources of quantum light. Despite implementation in devices that exploit quantum phenomenon, a comprehensive review on the crystal growth of quantum dots grown by droplet etching epitaxy is absent, unlike for other quantum dot growth techniques such as the related droplet epitaxy method or Stranski-Krastanov growth of InAs quantum dots. This review presents a detailed overview of the droplet etching epitaxy growth technique in the molecular beam epitaxy environment, with emphasis on the growth parameters necessary to realize high-quality quantum dots. We systematically cover the three main phases of droplet etching epitaxy - droplet deposition, droplet etching, and nanohole regrowth - and relate experimental results to theories on crystal growth. The review concludes with an introduction to GaAs quantum dot photoluminescence and the extension of droplet etching epitaxy beyond the AlGaAs/GaAs material system.

We develop a kinetic theory of cohesin-driven loop extrusion on a disordered chromatin track with transient barriers. In the stationary state, the mean loop size is shown to obey a universal law determined by the bare processivity and a renormalized obstacle density. Beyond the mean, one-sided extrusion always yields a single-exponential loop-length distribution, whereas two-sided extrusion produces a finite sum of exponential modes and, generically, a peaked distribution. Experimental CTCF-anchored loop statistics exhibit such a peak, thereby providing a direct discriminator of extrusion symmetry. The theory therefore establishes a unified framework for disorder-limited loop extrusion and supports a scenario in which both cohesin arms actively operate in living cells.

The results of a numerical investigation of fluidized beds of spherical particles in a narrow vertical cylindrical pipe, with particular attention to the spontaneous settling along the wall, are reported. Starting from a steady fluidized state, the particles fluctuate because of fluid-particle, particle-particle, and particle-wall interactions. The particles are heavier than the fluid, with diameters d yielding ratios of pipe to particle diameters D/d=4.3 and 4.7. For given ranges of flow velocities and bed sizes, particles settle on the wall, with a decrease in the bed height and particle fluctuations. Either a glass- or crystal-like shell forms along the pipe wall, in qualitative agreement with previous experiments. The polydispersity and the particle-particle friction are varied to test the stability of the particulate shell formation. The shell structure is analyzed by unwrapping it in a plane and locating all particles and their contact points, and we find that it exhibits a hexagonal lattice with a defects density that increases with polydispersity. The shell formation is hindered by polydispersity, and there exists a critical point for polydispersity above which a crystal-like shell is unstable. In a particular case of bidisperse beds, the crystal-like shell only appears when the particle-particle friction is high enough. Finally, we compute the contact forces within particle-particle chains and in particle-wall contacts, which sustain the cylindrical shell, highlighting the dominant role of particle-particle forces.

Understanding mechanisms for the breakdown of thermalization in closed quantum systems is a central problem in quantum many-body physics. We demonstrate strong non-ergodic behavior in the XX model on coupled chains, where domain-wall initial states retain an inhomogeneous magnetization profile for arbitrarily long times. We find that this effect arises due to exponentially many zero modes protected by chiral symmetry. Using an analysis based on the Lanczos algorithm, we identify a localization transition in the thermodynamic limit at a critical coupling between the chains. We further show that antiferromagnetic defects in the initial state and symmetry-breaking perturbations restore slow thermalization, whereas it remains robust for symmetry-conserving perturbations. These results establish that degenerate, symmetry-protected subspaces can give rise to thermodynamically stable non-ergodic dynamics in experimentally accessible quantum systems.

Detachment and fracture are central to many tissue-level processes, but they are challenging to simulate with Voronoi-type models that typically assume a confluent tissue. Here we analyze the finite Voronoi model, a nonconfluent extension of conventional Voronoi models, in which cell boundaries are composed of straight Voronoi edges and circular arcs of fixed radius $\ell$. When the line tension on cell-medium interfaces exceeds the tension on cell-cell contacts, we find that the model exhibits a strong time-step dependence in the fracture timescale of initially intact active clusters: decreasing $Δt$ can unphysically suppress cluster rupture events. We trace this behavior to a divergence of detachment forces in the finite Voronoi model and introduce a simple regularization. Finally, we calibrate the near-detachment mechanics against a deformable polygon model and examine how key physical parameters control the tissue fracture timescale under two different calibration strategies. Our results show that, for studies focused on fracture or intercellular adhesion in nonconfluent monolayers, a physically motivated calibration of near-detachment mechanics in the finite Voronoi model is essential.

{\bf Purpose}: To develop a geometry-governed diffusion framework that explains differential tissue response under FLASH ultra-high dose rate (UHDR) irradiation by explicitly accounting for structural heterogeneity and anomalous transport in biological tissues. {\bf Methods}: We formulate a generalized diffusion--reaction model on fractal substrates to describe molecular transport in heterogeneous media. Tissue architecture is characterized by a fractal (Hausdorff) dimension \(D\), while scale-dependent transport inefficiency and memory effects are captured by a fractional parameter \(θ\). Analytical solutions for radially symmetric geometries are derived and compared with classical normal (Euclidean) diffusion and a Gaussian reference model under identical physical conditions. Transport behavior is quantified through transient probability distributions and steady-state spatial profiles. {\bf Results}: The model reveals systematic suppression of long-range transport and enhanced localization as tissue structural complexity increases. Increasing \(θ\) leads to subdiffusive dynamics, reduced effective diffusion lengths, and persistent non-Gaussian concentration profiles, even in the steady state. While increasing \(D\) alone enhances spatial accessibility, fractional dynamics dominate transport behavior when \(θ>0\), counteracting geometric connectivity. These effects produce a separation between regimes characterized by efficient inter-track overlap and rapid homogenization, and regimes marked by isolated, long-lived reactive domains.

In this paper, we demonstrate how altermagnets can be simulated in the recently released micromagnetic simulation package mumax+. We have added a new magnet class for d-wave altermagnets and demonstrate how mumax+ is able to reproduce the analytical solutions for line profiles of the Néel vector and net magnetization for a Bloch domain wall. Next, we show simulation results of the magnon dispersion relation and its dependence on the anisotropic nature of the exchange interaction. Finally, we study the motion of a Néel skyrmion by applying a spin transfer torque. This new feature was implemented by extending the pre-existing code base for antiferromagnetic simulations. The object-oriented design of mumax+ allows for a correct calculation of the magnetostatic field in multi-sublattice systems, a feature that many other micromagnetic simulators lack.

We investigate the balanced $M=4$, $p=4$ spin-glass model for a one-dimensional long-range proxy for the finite dimensional short-range $p$-spin glass model to examine the nature of the glass transition beyond mean-field theory. We perform large-scale Monte Carlo equilibrated simulations for both fully connected and power-law diluted versions of the model. The critical temperatures extracted from the finite-size scaling (FSS) analysis of spin-glass susceptibility are in good agreement with theoretical predictions for $σ= 0, 0.25$, and 0.55. For these values of the long-range exponent $σ$ (which is the power of the decrease of the interactions between the spins with their separation), one might have expected that mean-field theory would provide a good description of the system. However, the spin-overlap distribution and the value of the $λ$-parameter do not provide numerical evidence for a one-step replica symmetry breaking (1RSB) phase transition. Instead, our results indicate a direct transition from the paramagnetic state to a full replica symmetry broken phase, with a renormalized value of $λ\equiv ω_2/ω_1 < 1$ suggesting a continuous FRSB transition, despite this ratio being equal to 2 at mean-field level. A value of $λ> 1$ is required for the discontinuous 1RSB transition. We argue that strong finite-size effects and closely spaced transition temperatures remove the expected 1RSB transition for the system sizes which we can study. For values of the exponent $σ= 0.85$, which roughly corresponds to a three dimensional system, we find that the renormalized value of $λ$ is again less than 1, with no signs of either the 1RSB transition or the continuous FRSB transition, suggesting that the Kauzmann temperature $T_K$ in three dimensions might be zero and the complete absence of phase transitions in structural glasses.

Many CFTs can be extended to lines of nonlocal CFTs parametrised by the scaling dimension $Δ$ of the fundamental field appearing in the action. $Δ=\frac{d}{2}-ζ$ is set by the exponent of the kinetic term $(-\partial^2)^ζ$, which is nonlocal for noninteger $ζ$. If $Δ$ is tuned to $Δ_\mathrm{local}$, the scaling dimension of the fundamental field in the local CFT, arXiv:1703.05325 showed that we recover the conformal data of that CFT (plus a decoupled sector). One natural question is: how is the local point special on this line of nonlocal CFTs? We prove that these local CFTs lie at the extrema of the (universal part of the) sphere free energy $\tilde{F}(Δ)=\sin(\frac{πd}{2}) \log Z_{S^d}$ of the long-range CFTs: $d\tilde{F}/dΔ|_{Δ=Δ_\mathrm{local}}=0$; and for unitary CFTs they locally maximise it. The simple proof uses the fact that the derivative of $\tilde{F}$ with respect to $Δ$ receives contributions only from the nonlocal terms in the action. The nonlocal terms must be absent in the limit $Δ\to Δ_\mathrm{local}$, and hence the derivative is zero. Demonstrating maximisation then requires a proof of the generalised $F$-theorem in conformal perturbation theory to subleading order. We check our result with the O$(N)$ $φ^4$ and cubic CFTs in the $ε$ expansion and the large-$N$ limit. This result provides a minimal encoding of the scaling dimension of the fundamental field in many CFTs, and also explains the curious derivative structure of the large-$N$ expansion of $Δ_φ$ in the O$(N)$ model. Finally, this nonlocal $F$-extremisation can be viewed as a non-supersymmetric version of the known $c,F,a$-extremisation mechanisms.

Singlet-only always-on gapless exchange (SAGE) spin qubits are an alternative type of exchange-only (EO) qubits that encode a single qubit in the spins of four electrons located in four tunnel-coupled quantum dots. While conventional EO qubits are susceptible to local magnetic field gradients caused by local nuclear environments and $g$-factor variations, the SAGE qubit subspace is inherently protected from magnetic-gradient-induced Pauli errors by virtue of the singlet-only encoding, which is invariant under magnetic field gradients, and the always-on exchange couplings, which provide energetic leakage protection. However, the always-on operation simultaneously increases the qubit's sensitivity to charge noise. Here, starting from a Hubbard model describing the underlying electronic structure of the coupled quantum dots, we characterize the performance of SAGE qubits in the presence of $1/f$ charge noise that induces fluctuations in both the dot chemical potentials and the interdot tunnel couplings. We calculate SAGE idle coherence times and show that realistic CPMG-like pulse sequences can be used to significantly extend SAGE single-qubit coherence times for experimentally relevant charge noise strengths. We likewise study the fidelity of SAGE two-qubit gates in the presence of charge and magnetic noise and again propose a simple refocusing strategy to mitigate the noise, while increased ramp times of the entangling pulse suppress leakage into noncomputational states.

Over the last decade, the development of Josephson devices based on van der Waals (vdW) materials has advanced rapidly, representing a paradigm shift driven by the advent of 2D materials. The diverse vdW materials library, combined with advanced fabrication techniques, enables the integration of materials with vastly disparate properties for scientific exploration. The vdW Josephson junctions (JJs) offer a unique route to explore novel functionalities and associated physics that remain inaccessible in conventional JJs, which have reached an industrial level in terms of fabrication. Beyond material diversity, vdW crystalline materials offer fundamental new control over device symmetries, enabling the realization of Hamiltonians unique to 2D systems. Furthermore, the long relaxation times of myriad excitations in 2D heterostructures open possibilities for creating exquisite quantum sensors, with the 2D material itself acting as an efficient bus for transmitting excitations to the active sensing element. This creative explosion in vdW-based superconducting electronics is rapidly growing, and our review highlights the resulting devices and physics. The confluence of vdW JJs with twistronics and topology has the potential to redefine superconducting quantum technology, enabling applications from quantum computation to ultra-sensitive hybrid sensors. While opportunities abound with vdW JJs, the challenge of scalability must be surmounted for translation into real-world devices. This review synthesizes current developments and offers a roadmap for researchers navigating this burgeoning field.

We study the orientational dynamics of superparamagnetic colloidal dimers that carry both an induced magnetic moment, proportional to the applied field, and an effective permanent moment. In a static, uniform magnetic field, dimers that are permanently fixed together hop between two preferred in-plane angles, developing a bimodal steady-state orientation distribution. When the same field is periodically reversed, we observe a sharp, field-controlled change in the dynamical response from small hopping events with $Δθ\ll π$ to full $Δθ\approx π$ rotations on each field flip. We show that both the static bistability and the switching bifurcation can be rationalised by a magnetic response in the dimer that consists of both a strong induced and weak body-fixed component. This leads to a complex orientational energy/potential landscape, with coupled roll-yaw rotations of the dimer responsible for the bistable dynamics. By combining the misorientation between dimer axis and field, bifurcation field strength and short-time orientational variance, we determine the magnitude and orientation of the net permanent dipole, thereby characterising details of the internal magnetic structure of the particles via microscopy.

We study dense mixtures of passive and active self-aligning disks with isotropic or anisotropic mobility. We find that the passive fraction controls an order-disorder transition that is continuous in the isotropic case and discontinuous in the anisotropic one. A mean-field equation derived from the microscopic heading dynamics captures this dichotomy. Near the transition, both ordered regimes can exhibit multiple metastable oscillating or rotating states, depending on the spatial arrangement of passive particles and lattice defects, but with different transient dynamics: Systems with isotropic mobility visit multiple long-lived attractors during each simulation while systems with anisotropic mobility are trapped by a single attractor. Our results reveal the passive fraction as a physically relevant control parameter in active systems, leading to rich self-organizing dynamics.

The interplay among topology and interactions has been a promising path towards identifying novel phases of condensed matter beyond these predicted by the established classification paradigms. In the present work, we propose such a novel phase of matter by studying the fate of a three-dimensional Berry-dipole semimetal, lying at the topological quantum critical point separating a Hopf insulator from a trivial insulator, in the presence of long-range Coulomb interactions. Utilizing large-$N_f$ analysis and an $ε$-expansion within the renormalization-group scheme, we predict the emergence of a spatially anisotropic non-Fermi liquid with enhanced Berry-dipole moment. The corresponding scaling relations of certain physical observables are derived as functions of the probed energy and temperature scale, and we conclude by providing an observational test for probing the enhanced topological features of the anisotropic non-Fermi liquid.

The dynamics of water in electrolyte solutions exhibits a striking, ion-specific anomaly: the diffusion coefficient of water is enhanced relative to the neat liquid in chaotropic CsI solutions, yet suppressed in kosmotropic NaCl solutions. This phenomenon, long challenging for classical force-field-based molecular dynamics, is studied here using classical molecular dynamics simulations with a many-body machine-learned force field (MLFF) trained within the MACE equivariant graph neural network framework. The force field is trained on energies, forces, and stresses computed at the density functional theory level with the revPBE-D3 exchange--correlation functional, which provides a reliable balance between accuracy and computational efficiency for aqueous systems. Simulations of NaCl and CsI aqueous solutions at ambient conditions over a concentration range of 0.89--3.56 mol/kg reproduce the experimentally observed anomalous diffusion and show a quantitative improvement over previous results obtained with the DeePMD framework, trained on the same theory, particularly for NaCl solutions. This improvement is traced to a stronger Na$^{+}$--water interaction in the first hydration shell and the non-negligible retarding contribution of the second hydration shell of Na$^{+}$. For CsI solutions, the water acceleration is shown to be primarily driven by the anion I$^{-}$, whose diffuse and weakly structured hydration shell facilitates rapid water exchange with the bulk. These results are rationalised through a shell-decomposition analysis of time-dependent water diffusivities and ion--oxygen potentials of mean force providing a coherent microscopic picture of the acceleration--retardation mechanism in the studied aqueous electrolytes.

Spin glass dynamics is a strong function of spatial dimensionality $D$. The lower critical dimension is close to 2.5, so that, in two dimensions, the condensation temperature $T_\text{g}=0$, and only fluctuations are present at finite temperatures. However, by using thin film multilayers, one can explore the dynamics in both $D=3$ and $D=2$ dimensions. Spin glass thin film multilayers transition from $D=3$ dynamics at short to intermediate times to $D = 2$ dynamics at long times. Correlation lengths of CuMn 4.5 nm multilayers at long times are shown to be grow more rapidly in $D=2$ as compared to $D=3$, and for the longest measurement time, experimentally reach equilibrium in qualitative agreement with simulations.

We show that jets of initial data can be approximated up to arbitrary order by finite-gap solutions for classes of so-called BKM systems of PDEs introduced by Bolsinov--Konyaev--Matveev, which include classical PDEs such as KdV, Kaup--Boussinesq and Camassa--Holm. Finite-gap solutions are obtained via a finite-reduction map, defined algebraically, which sends solutions of a Stäckel system to solutions of the BKM PDE. For the classes containing KdV and Kaup--Boussinesq we obtain full jet-surjectivity via a triangular structure, whereas for the class containing Camassa--Holm we establish jet-surjectivity on an open set of initial data over $\mathbb{R}$ and a Zariski-open (dense) set over $\mathbb{C}$.

Self-organized criticality (SOC) is widely proposed as a fundamental mechanism for collective behavior, yet its role in objective-driven, heterogeneous adaptive systems underpinning real complex systems remains less understood. We introduce EvoSK, a minimal evolutionary model in which agents perform memory dependent reinforcement learning on a rugged Sherrington-Kirkpatrick landscape while the population evolves through extremal replacement of the least fit agents. We demonstrate that this coupled dynamics drives the system to a critical state residing on the transition boundary between ergodic and non-ergodic phases. At this boundary, the system exhibits scale-free evolutionary avalanches with a mean-field exponent $τ\approx -1.5$, while simultaneously achieving collective rewards that surpass those of any manually finetuned, non-evolutionary regime. Our results provide a mechanistic link between the statistical physics of ergodicity breaking and the functional optimality of complex adaptive systems, suggesting that the edge of ergodicity breaking acts as a robust attractor for systems adapting on rugged, high-dimensional landscapes.

The introduction of nonlinearities into lattices with topological band structures has led to the discovery of various types of solitons. The Su-Schrieffer-Heeger (SSH) lattice, as the most fundamental topological model, has been extended into the nonlinear regime. In particular, nonlinear edge states and bulk solitons exhibiting intensity humps against a zero background have been extensively studied in nonlinear SSH lattices. In this paper, we systematically investigate dark solitons in nonlinear SSH lattices. These dark solitons maintain a nonzero and constant background, featuring intensity dips either in the bulk of the lattice or at its edges, and residing spectrally in the semi-infinite gap or the middle finite gap. Regardless of the specific type of dark soliton, the intensity dip remains wellpreserved and is not affected by the band structure of the original linear lattice. Although the dark solitons we have identified are generally dynamically unstable across a broad range of parameters, several types exhibit linear stability when the intracell coupling is much larger than the intercell coupling. Our findings may provide valuable insights for the exploration of novel types of solitons in nonlinear topological lattices.

Stochastic models of diffusion are routinely used to study dispersal of populations, including populations of animals, plants, seeds and cells. Advances in imaging and field measurement technologies mean that data are often collected across a range of scales, including count data collected across a series of fixed sampling regions to characterize population-level dispersal, as well as individual trajectory data to examine at the motion of individuals within a diffusive population. In this work we consider a lattice-based random walk model and examine the extent to which model parameters can be determined by collecting count data and/or trajectory data. Our analysis combines agent-based stochastic simulations, mean-field partial differential equation approximations, likelihood-based estimation, identifiability analysis, and model-based prediction. These combined tools reveal that working with count data alone can sometimes lead to challenges involving structural non-identifiability that can be alleviated by collecting trajectory data. Furthermore, these tools allow us to explore how different experimental designs impact inferential precision by comparing how different trajectory data collection protocols affects practical identifiability. Open source implementations of all algorithms used in this work are available on GitHub.

Cell-cell adhesion is widely hypothesised to maintain cohesion within the long streams of follower cells that trail leader subpopulations during collective migration, including in neural crest cell migration, angiogenesis, and cancer cell invasion. Mathematically, non-local advection-diffusion equations provide the canonical continuum framework within which to study such adhesive cell-cell interactions. Here, we study a minimal model of leader-follower migration within this framework, in which leaders migrate at constant velocity while followers are attracted to leaders and to one another over a finite spatial interaction range. Numerical simulations reveal that, although the model can maintain small cohorts of travelling follower cells, the size of these cohorts is limited by the adhesive interaction lengthscale, and is far below what is needed to reproduce the extended streams observed in vivo. This points to a structural limitation of the standard non-local adhesion formulation and highlights the need for the development of extended continuum models capable of sustaining long, coherent migratory streams through purely mass-conserving collective cell movement.

Computational power can be measured by assigning an algebraic structure to a computational device. Here, we convert a small patch of Conway's Game of Life into a transformation semigroup. The conversion captures not only time evolution but also interactive operations. In this way, the cellular automaton becomes directly programmable. Once this measurement is made, we apply hierarchical decompositions to the resulting algebraic object as a way of understanding it. These decompositions are based on a macro/micro-state division inspired by statistical mechanics. However, cellular automata have a large number of global states. Therefore, we focus on partitioning the state space and creating morphic images approximations that can serve as macro-level descriptions. The methods developed here are not limited to cellular automata; they apply more generally to discrete dynamical systems.

Low-dimensional reduction theories, such as the Ott-Antonsen ansatz, have played a crucial role in the study of populations of globally coupled phase oscillators. However, most of these theories are applicable only to models in which the interaction is described by a single harmonic component, limiting their application to more realistic oscillator models. In this paper, by employing the theory of orthogonal polynomials on the unit circle (OPUC), we construct a framework that enables low-dimensional reduction for populations of globally coupled phase oscillators with multiharmonic coupling.

We demonstrate the existence of bright solitons in a repulsively interacting, harmonically trapped quasi-one-dimensional Bose-Einstein condensate described by the Gross-Pitaevskii equation. Using a neural-network quantum state (NNQS) approach, we parametrize the initial wavefunction and optimize it to find solutions that recur after one trap period, effectively balancing repulsion with trap-induced attraction. Aside from the bright solitonic solution, we also report double bright and dark soliton states. Perturbing the initial state with multiplicative phase and amplitude noise confirms that these periodic orbits are orbitally stable. Our results indicate that NNQS provides a powerful framework for uncovering coherent structures in nonlinear wave systems.

Statistical physics can describe the behavior of microbial populations consisting of many heterogeneous individuals. A direct consequence is the existence of phase transitions, where the behavior of a population changes discontinuously upon a small perturbation. While such phase transitions have often been proposed in biology, connecting observed behavior to the underlying physics has remained challenging. We show how phase transitions naturally arise in microbial population dynamics and highlight their connection with genealogies. We rigorously demonstrate the existence of a first-order phase transition in a model of bacterial plasmid engineering and find a strict lower bound on the number of plasmids that can be stably maintained in a population.

Experimental evidence indicates that intracellular chloride concentration regulates the excitation and inhibition (EI) balance, yet the mechanisms by which activity-dependent chloride dynamics drive seizure evolution and stage transitions remain unclear. We present a conductance-based neuronal network in which EI balance emerges from chloride homeostasis via channel-mediated influx and transporter-mediated extrusion. We show that the fraction of inhibitory synaptic conductance contributing to channel-mediated influx acts as a control parameter that organizes seizure dynamics into distinct stages,pre-ictal, ictal-tonic, and ictal-clonic,distinguished by characteristic amplitude and frequency signatures. Decreasing this fraction shortens ictal activity and suppresses seizure initiation, whereas high fraction promotes the emergence of ictal-tonic and ictal-clonic stages and spiral-wave dynamics, rendering seizure dynamics largely insensitive to inhibition. At intermediate values, seizures bypass the ictal-tonic stage and emerge directly as the icta,clonic stage. Moreover, joint variation of fractions with synaptic strengths reveals that recurrent excitation expands the tonic-clonic seizure, while recurrent inhibition prolongs pre-ictal states and suppresses ictal-clonic activity.

Biochemical signalling cascades transduce extracellular stimuli into cellular responses through sequences of discrete, node-to-node activations. While signal fidelity depends critically on local interaction kinetics, the mechanisms governing information propagation in realistic, highly variable kinetic contexts remain poorly understood. In this paper, we develop a mathematical framework for travelling waves in canonical feed-forward pathways governed by nonlinear Michaelis-Menten-type kinetics. For uniform pathways, we characterise the complete steady-state landscape and demonstrate that activation bias (the contribution of the binary states of each node to downstream activation) between connected nodes acts as a key bifurcation parameter dictating wave existence. Extending this framework to heterogeneous networks, we show how parameter gradients and random kinetic variations distort wavefronts and induce heavy fluctuations in propagation speed. To recover predictable signal transmission, we introduce a novel reciprocal-velocity spatial rescaling technique. We demonstrate that this coordinate transformation inherently absorbs local kinetic variations, effectively smoothing wave velocities and preserving wavefront profiles without requiring bespoke parameter tuning or continuous limits. Finally, by testing the framework's limits against extreme parameter variability, we reveal how severe kinetic bottlenecks lead to functional pathway fragmentation, offering a mathematically justified basis for rational model reduction in complex biochemical networks.

Motivation: Statistical analysis of microbial count data derived from 16S rRNA or metagenomics sequencing poses unique challenges due to the sparse, compositional, and high-dimensional nature of the data. While QIIME 2 already provides many tools for data pre-processing and analysis, plugins for statistical regression, classification, and microbial network estimation tailored to compositional count data are relatively scarce. Results: We present q2-classo and q2-gglasso, two novel QIIME 2 plugins that implement penalized regression, classification, and graphical modeling approaches for microbial compositional data. q2-classo enables the prediction of a continuous or binary outcome of interest using compositional microbiome data as predictors. Both sparse log-contrast regression and classification, as well as tree-aggregated log-contrast models are available. q2-gglasso enables the estimation of taxon-taxon association networks through sparse graphical model estimation, such as, e.g., the SPIEC-EASI framework, as well as adaptive and latent graphical models. The latent model can decompose taxon-taxon associations into a sparse direct interaction matrix and a latent (low-rank) matrix which enables robust principal component embedding of a data set. Within the QIIME 2 ecosystem we demonstrate their application on the Atacama soil microbiome dataset, illustrating robust model selection, classification, and microbial network estimation with covariates and latent factors. Availability: The software is freely available under the BSD-3-Clause License. Source code is available at https://github.com/bio-datascience/q2-gglasso and https://github.com/bio-datascience/q2-classo-latest, with installation through QIIME 2 and Docker.

We introduce `Goxpyriment', a new open-source software framework for programming behavioral and cognitive experiments using the Go programming language. The library is designed to address some limitations of existing Python-based experiment tools, particularly the runtime environment complexity that frequently complicates deployment across laboratories. Because Go is a compiled language that can natively embed assets (e.g., graphics, audio files, and stimulus lists), Goxpyriment compiles entire experiments into single, self-contained executable binaries with zero runtime dependencies. This drastically simplifies distribution to collaborators and testing computers. The programming interface, inspired by Expyriment (Krause & Lindemann, 2014), was designed to be human friendly. The library includes an array of visual stimuli (text, shapes, images, Gabor patches, motion clouds, ...) and audio capabilities (WAV playback and tone generation). While developing Goxpyriment, we focused on timing reliability. Input events are timestamped by the operating system at hardware-interrupt time, so reaction times are computed by subtracting two OS-level timestamps rather than relying on continuous polling. Go's garbage collector can be disabled, greatly reducing the probability of unpredictable pauses that could corrupt stimulus timing. Finally, a set of over forty psychology experiments implemented in Goxpyriment are provided that promote not only learning by humans but also improve the ability of modern AI-assisted coding tools to help program experiments. The framework is released under the GNU General Public License v3 and is freely available at https://github.com/chrplr/goxpyriment.

Biological agents navigate complex environments by combining long-term memory of successful actions with short-term suppression of recently visited locations-a capability that remains difficult to replicate in artificial systems, especially under partial observability. Inspired by the complementary timescales of neural and astrocytic dynamics, we introduce a spiking neuron-astrocyte network (SNAN) where spike-timing-dependent plasticity (STDP) reinforces successful action sequences on a distant time scale, while astrocytic calcium transients suppress recently visited states on a short-term time scale, effectively blocking locations already explored. This dual-timescale memory mechanism biases the agent toward unexplored regions, accelerating goal finding without requiring explicit global statistics. We show that in grid-world navigation tasks with extreme partial observability, SNAN reduces median path length by up to sixfold and drastically improves goal completion rates compared to baseline agents. The astrocytic modulation inherently mitigates the exploration-exploitation trade-off as an emergent consequence of local state suppression. This kind of local sensory data modulation can be considered as a new type of working memory referred to as a "Topological-Context Memory". To validate hardware feasibility using neuromorphic approaches, we map STDP to a memristive VTEAM model and implement a subset of the network on a crossbar array, achieving order-of-magnitude gains in speed per area and energy per decision over CPU implementations. Our results establish astrocyte-inspired dual-timescale memory as a scalable, hardware-realizable principle for neuromorphic robotics and edge-AI systems.

Encoding models enable measurement of how our brains represent sensory inputs using electro-and magneto-encephalography (MEEG). Evaluating how closely encoding models reflect the underlying brain functions is a crucial premise for model interpretation and hypothesis testing. However, the ground-truth neural activity is unknown, preventing model evaluation with respect to the target neural signal. Existing evaluation metrics must therefore relate model's predictions to noisy MEEG measurements, where most variance is stimulus-unrelated. Here, I introduce an evaluation framework where model predictions are compared to a ground-truth approximation, obtained by aligning MEEG signals with predictions using canonical correlation analysis and via participant averaging. The resulting metric (CPA-PA) yields single-participant evaluations outperforming conventional scores by 300-1000% on synthetic EEG data and 250% on 34 real MEEG datasets (818 datapoints). These gains reflect increased sensitivity to stimulus-relevant neural activity and reduced dependence on SNR, establishing ground-truth approximation as a robust framework for evaluating encoding models.

Absolute concentration robustness (ACR) means the concentration of certain species stays the same in all the steady states. In this work, we study how conservation laws might effect non-vacuous ACR in reaction networks. The goal is to show whether non-vacuous ACR can be preserved or precluded by adding species that depend on the existing species. We have the following two main results. (i) For networks with conservation laws, we prove a criterion: for a nondegenerate network, augmenting it with one new species that depends on the original species leads to the resulting network having no non-vacuous ACR for any generic choice of rate constants in the new species. (ii) We characterize all non-redundant zero-one networks with dimension of at most two that exhibit non-vacuous ACR for any generic choice of rate constants according to the number of distinct rows in the stoichiometric matrices. An important finding is that if there are at least four distinct rows in the stoichiometric matrix, then the corresponding network has no non-vacuous ACR for any generic choice of rate constants, which implies that many conservation laws prevent non-vacuous ACR in non-redundant zero-one reaction networks.

The spread of infectious disease is strongly influenced by social dynamics. In addition to infection risk, individuals vaccination decisions depend on prevailing social behavior: high infection levels and widespread vaccination can increase vaccine uptake, which in turn suppresses infection. This feedback can generate sustained oscillations in disease prevalence and vaccination behavior. Here, we study two such populations undergoing the same behavioral epidemiological limit cycle and introduce weak coupling between them through social influence. We show that coupling leads to synchronization of disease dynamics between the two groups. Moreover, we find that different payoff sensitivity may lead to synchronization or anti synchronization.

This paper proposes the LiFE-CD algorithm for convergence time analysis of the max-consensus algorithm in multi-agent systems under Bernoulli-distributed link failures. Unlike existing approaches, which either assume ideal communication or provide asymptotic upper bounds on the expected convergence time, LiFE-CD deterministically computes the full probability distribution of the convergence time from network topology and individual link failure probabilities, without simulation. The full probability distribution enables deadline-aware protocol design with specified reliability guarantees. Based on geometrically distributed link delays, the proposed algorithm iteratively reduces the given network topology considering both unicast and broadcast transmissions. LiFE-CD yields exact results for acyclic networks and, for cyclic networks, tight upper bounds on the convergence time via shortest-path spanning tree construction. Numerical results confirm analytical exactness for acyclic networks, validate tightness for cyclic networks, and demonstrate improvement over existing approaches. Our complexity analysis shows reduced computational cost compared to Monte Carlo simulations, while eliminating stochastic variability and enhancing reproducibility. All results extend directly to min-consensus by structural equivalence.

ISAC systems introduce new privacy risks because an unintended sensing node may exploit the shared radio waveform to infer transmitter-related information even when the communication payload remains secure. This paper investigates transmitter privacy, defined as limiting unauthorized inference of transmitter-related information through channel estimation, in a RIS-aided multi-antenna wireless system with a transmitter, a legitimate receiver, a malicious sensor, and a RIS. The malicious sensor is assumed to estimate the transmitter--sensor channel, and the resulting channel state information can then support unauthorized sensing, inference, or related signal processing. To mitigate this threat, we consider a privacy-oriented design in which the transmitter adopts superposition-based signaling with a message signal and transmit-side artificial noise, while the RIS shapes the propagation environment in a privacy-aware manner. The channel-estimation performance at the malicious sensor is first analyzed under imperfect prior knowledge, and both the true and predicted mean-square-error expressions are derived. Based on this analysis, we formulate a joint active--passive beamforming design problem that maximizes the malicious sensor's predicted channel-estimation error subject to a communication quality-of-service constraint, a transmit-power budget, and the unit-modulus constraints of the RIS. The resulting non-convex problem is handled through a numerically efficient alternating-optimization framework based on an augmented Lagrangian reformulation. Numerical results show that RIS-assisted propagation shaping can substantially degrade unauthorized channel estimation relative to the non-RIS case while preserving reliable communication, and further show that the privacy gains also improve a more direct sensing metric, namely the malicious sensor's angle-of-arrival estimation accuracy.

This paper presents a comprehensive link budget analysis for millimeter wave (mm-Wave) and sub-Terahertz (sub-THz) communication systems with primary focus on transmitter (TX) noise propagation, an often overlooked impairment that can dominate in scenarios where path loss is insufficient to suppress TX noise below receiver thermal and atmospheric molecular noise levels. Unlike traditional thermal noise limited analyses, this work demonstrates that TX noise is amplified by component noise figures that degrade significantly with frequency, rising from single digits to more than $15\,\mathrm{dB}$ in the sub-THz range. In the scenarios analyzed, this propagated TX noise reduces the achievable Signal-to-Noise Ratio (SNR) by approximately $15$ to $25\,\mathrm{dB}$ at short distances, creating fundamental SNR ceilings at ranges below about $10\,\mathrm{cm}$. We develop a systematic framework quantifying TX noise dominance conditions as functions of distance, frequency, and component parameters, revealing fundamental performance constraints for short-range next generation wireless systems. Our findings indicate that the TX noise figure should be as low as possible for short-range communication, and both TX noise and atmospheric molecular noise should be considered for medium- and long-range links.

High-precision indoor sensing using monostatic multiple-input multiple-output (MIMO) radar typically relies on increasing the physical aperture size of antennas, leading to high hardware complexity and cost. To overcome this bottleneck, this paper establishes a unified framework for multi-site radar sensing based on equivalent angular resolution, together with a design methodology that uses this metric to optimize distributed Single-Input Single-Output (SISO) configurations. By mapping spatial diversity into the angular domain, the proposed metric enables a direct and physically interpretable comparison with monostatic MIMO beamwidth. The associated methodology provides a principled way to select node placement and geometry to synthesize an effective virtual aperture that suppresses angular glint and multipath. Experiments with commercial 60-GHz radars in cluttered indoor environments validate the superiority of the multi-site SISO configuration over monostatic MIMO, demonstrating a reduction in maximum localization error from 0.58 m to 0.20 m and mean error from 0.35 m to 0.12 m.

Ultra-massive multiple-input multiple-output (UM-MIMO) is a key technology for enabling terahertz (THz) communications in 6G networks, offering high beamforming gain to combat severe path loss. However, the large antenna array expands the near-field region, resulting in a hybrid near- and far-field communication environment. This makes channel estimation significantly more challenging than in conventional networks. To address this issue, we propose a novel attention augmented channel estimator named the fixed-point attention network (FP-ANet), which integrates fixed-point theory with a dual-attention mechanism. By combining a linear and dual-attention residual blocks based non-linear estimator in each iteration, this model-driven approach effectively exploits the sparsity of THz channels in the angular-distance domain, enabling a more precise and physically-grounded channel estimation. Simulation results show that FP-ANet achieves superior channel estimation accuracy compared to state-of-the-art methods while maintaining comparable computational complexity.

Quantitative speed-of-sound (SoS) and attenuation of tissues are closely related to pathology; however, conventional B-mode images are limited to qualitative visualization. Existing ultrasound full-waveform inversion (FWI) methods for quantitative SoS reconstruction are primarily developed under double-sided or ring-shaped arrays, which limits their applicability to widely adopted routine clinical acquisitions. In this work, we develop a frequency-domain, total variation (TV)-regularized FWI framework tailored for single-sided linear ultrasound arrays, which enables quantitative reconstruction of SoS maps using standard clinical probes. To address the severe ill-posedness and computational challenges in this setup, efficient forward modeling, fast gradient evaluation, ADMM-based optimization, and multi-GPU parallelization are integrated into the inversion framework. Numerical experiments in a thyroid cyst imaging scenario demonstrate that the proposed method reconstructs the SoS of both simple (fluid-filled) and solid cysts with improved visual and quantitative performance compared to conventional FWI. Additional 2D and 3D simulations across different target and array apertures further elucidate the capabilities and limitations of single-sided ultrasound FWI.

Low earth orbit (LEO) satellites are a key technology to enable connectivity for rural and remote users. Communication satellites in LEO can provide coverage to much larger areas than terrestrial or aerial systems, while offering improved data rates when compared with geostationary systems. However, a major challenge with LEO satellite communications is the high mobility of the satellite, which results in a rapidly changing communication channel. Due to this, it is challenging to fairly allocate communication resources to multiple users in the system. This work proposes an Adaptive Power Allocation and Scheduling Scheme (APASS) to ensure user fairness in the downlink of a LEO satellite system serving mobile ground users. First, a novel channel and transmission model is introduced to capture the variability in channel statistics due to the satellite's trajectory. Then, a non-convex optimization problem is formulated to maximize the minimum rate across all ground users over a fixed set of time slots. To solve this problem, the proposed APASS dynamically allocates power and schedules transmissions based on predicted future channel gains. Numerical results show that APASS achieves strong performance even with substantial prediction errors, faring close to an upper bound that assumes perfect future channel knowledge. Furthermore, it improves the minimum user rate by a factor of 2.98 compared to equal-power allocation and maintains user fairness with a Jain's fairness index of well above 0.99.

This paper introduces a high-precision indoor positioning and tracking method that utilizes multi-site single-input single-output (SISO) radar systems. We propose a novel velocity synthesis-assisted (VSA) localization algorithm that iteratively refines target position estimates within range bins by fusing radial velocity measurements from multiple radars. This approach ensures enhanced accuracy in both velocity and position estimation. Moreover, the inherent geometric constraints introduced by velocity synthesis enable the proposed algorithm to remain robust under low signal-to-noise ratio (SNR), severe multipath propagation, and large synchronization latency. Notably, our method eliminates the use of multiple-input-multiple-output (MIMO) configurations and stringent phase synchronization requirements, substantially reducing hardware complexity while maintaining high positioning accuracy. We define standardized reference trajectories to facilitate a comprehensive and reproducible performance evaluation. Extensive simulations and experimental validations demonstrate that our multi-site radar systems achieve centimeter-level tracking accuracy for human subjects, outperforming existing methods in complex trajectory tracking.

With the push toward 6G commercialization, Frequency Range 3 (FR3) bands, specifically 7.125-8.4 GHz and 14.8-15.3 GHz, have become focal points for achieving wide-area, high-capacity coverage. However, practical deployment is often limited by the physical aperture constraints of base station antennas. This study conducts comprehensive measurements in Urban Macro (UMa) scenarios using a unified dual-band sounding platform to evaluate channel characteristics and system performance under the strict constraint of "equal physical array aperture." The results indicate that higher frequency bands exhibit increased sparsity in both delay and spatial domains. Regarding coverage, while the 15 GHz band can theoretically accommodate four times the number of antenna elements (128 elements) within the same area to compensate for path loss, empirical data reveals a residual coverage deficit of approximately 3.0 dB at cell edges compared to the 8 GHz baseline. In contrast, the 15 GHz band excels in capacity; the increased element count effectively overcomes channel sparsity, resulting in spectral efficiency (SE) that significantly outperforms the 8 GHz band. Furthermore, the research demonstrates that for a fixed number of elements, system performance remains largely insensitive to specific array topologies (e.g., 1x32, 2x16, or 4x8). Ultimately, FR3 system performance is dictated by the trade-off between propagation characteristics and hardware-enabled gain. These findings provide a theoretical foundation for spatial-domain design and help address engineering challenges for 6G base station implementation

Recovering a source signal from indirect measurements often requires estimating latent parameters, such as wireless channel states or MRI coil sensitivities, that cannot be directly observed. Here, we introduce Physics-Embedded Inverse Learning (PEIL), in which a learned estimator predicts these parameters and a fixed, physics-based inverse operator uses them to reconstruct the signal, so that training requires only the source signal as supervision. In systems where multiple parameter combinations can reconstruct the signal equally well, the estimator exploits this freedom to coordinate parameters that compensate for residual modelling errors rather than match ground-truth parameters. In high-mobility wireless communications, PEIL discovers task-optimal configurations that outperform baselines given access to ground-truth parameters, enabling zero-shot generalisation and over 20-fold reduction in training data relative to supervised baselines. To test whether these properties extend across physical domains, we apply PEIL to parallel MRI, where it discovers physically interpretable coil sensitivity maps without calibration scans, yielding reconstructions grounded purely in acquired measurements. These results demonstrate that non-identifiability, conventionally a liability, becomes a resource when the learning objective targets reconstruction quality rather than parameter accuracy.

This work considers a system where a dual-function radar transmitter (source) performs direct communication with a reader while simultaneously enabling ambient backscatter communication from a tag. The source embeds its message into a coded pulse repeatedly transmitted over a frame, whereas the tag exploits the resulting environmental reverberation (clutter) as an ambient carrier to convey its own message. By leveraging the structure induced by the radar waveforms, we develop two signaling schemes. In the pilot-free scheme, the source and tag messages are conveyed through nonlinear vector modulation; the induced subspace structure enables both joint decoding, where all unknown quantities are simultaneously estimated, and disjoint decoding, where the tag codeword is recovered first, followed by the estimation of the source codeword and the channel vectors. In the pilot-aided scheme, pilot symbols and linearly modulated data symbols are embedded within each frame, enabling both non-iterative decoding based on pilot-derived channel estimates and iterative decoding via alternating channel estimation and data detection. We establish sufficient conditions on the source and tag codebooks that guarantee noiseless identifiability of the involved messages and channels. Finally, performance is evaluated in terms of source/tag error probabilities and channel-estimation accuracy, and the resulting system-level tradeoffs are discussed.

This paper focuses on data-driven fault detection, identification, and recovery (FDIR) for nonlinear control-affine systems under actuator faults. We create a unified framework in the space of probability densities, rather than on individual trajectories, using fault-indexed Perron--Frobenius (PF) operators to predict the evolution of state distributions under different fault profiles. By leveraging the probability-flow representation of the Fokker--Planck equation, we construct deterministic PF operators that reproduce exact stochastic marginals, define forward reachable density families, and establish certifiable 2-Wasserstein bounds on the divergence between fault-driven and nominal density evolutions. These provide quantitative conditions for the detectability and identifiability of various faults. The fault-indexed operators are learned from trajectory data via flow map matching (FMM), and we demonstrate that the observable FMM residual directly bounds the approximation error of the operator in the 2-Wasserstein metric. Additionally, we co-train a contraction certificate that bounds the gap between the learned operator family, the actual fault-driven density flow, and the nominal dynamics. The operator library is then used online for continuous fault parameter fitting over a continuous parameter space to generalize the learned operators to out-of-distribution (OOD) scenarios. To carry out the recovery control, we employ reachable density propagation and Gaussian mixture covariance steering. The proposed framework is validated on a 10-state spacecraft attitude-control system with four reaction wheels.

Erasure qubits are a promising platform for implementing hardware-efficient quantum error correction. Realizing the error-correction advantages of this encoding requires frequent mid-circuit erasure checks that are fast, high-fidelity, and scalable. Here, we realize erasure detection with a hardware-efficient circuit consisting of a single readout resonator dispersively and symmetrically coupled to both transmons of a dual-rail qubit. We use this circuit to demonstrate single-shot erasure detection in 384 ns with minimal impact on the dual-rail logical manifold, achieving a residual error per check of $6.0(2) \times 10^{-4}$, with only $8(3) \times 10^{-5}$ induced dephasing per check, and an erasure error per check of $2.54(1)\times 10^{-2}$. The high degree of matched dispersive readout coupling ($χ$-matching) within the dual-rail qubit code space also allows us to realize a new modality: time-continuous erasure detection performed in parallel with single-qubit gates. Here we achieve a median $7.2 \times 10^{-5}$ error per gate with $< 1 \times 10^{-5}$ error induced by erasure detection. This demonstrates a reduction in erasure detection overhead as well as a crucial ingredient for soft information quantum error correction. Together, these results establish symmetrically coupled dispersive readout as a fast, hardware-efficient, and scalable component for erasure-based quantum error correction using transmon dual-rail qubits.

We compute thermodynamic properties of dilute Bose gases using non-perturbative approximations of the two-particle irreducible (2PI) effective action. It is shown how to systematically renormalise the self-consistent descriptions beyond conventional Gaussian approximations such as Hartree-Fock-Bogoliubov theory. This allows us to determine the condensate depletion from low to high temperatures, including its critical behaviour at the phase transition. While the universal anomalous dimension at criticality is vanishing for Gaussian approximations, we determine its non-zero value at next-to-leading order of a self-consistent expansion in the number of field components.

It is well-known that any two pure quantum states (in the same Hilbert space) can be mapped to any other using unitary transformations. However, previous approaches to this problem required two explicit bases for the Hilbert space, one each for the initial and target states, and thus their complexity necessarily scales with the dimension of the Hilbert space. In this Letter, we show how to utilize novel algebraic methods to construct a closed-form exponential unitary transformation which achieves this in general, using only a single unitary generator. This construction is independent of any bases and agnostic to the dimension of the Hilbert space. We highlight the usefulness of this tool for studying relationships between systems of pure states in quantum information theory, as well in elementary analyses of quantum circuits and unitary operators.

We show that boson correlations from quantum states with a Glauber-Sudarshan representation of their density matrix which provides a well-behaved probability distribution -- including coherent states, thermal states, and all states that can be deemed classical -- are a manifestation of the Simpson paradox: they are spurious correlations from statistical (ensemble) averages over uncorrelated measurements made in varying geometries, due to a process of symmetry-breaking as a confounding factor. Bosonic correlations encoded by the wavefunction appear to be formed in the geometry assumed, which however is not that of the statistical ensemble but varies from realization to realization. This calls to distinguish between quantum and statistical averages and sheds new understandings on the fundamental problems of nonclassicality and quantum advantage.

We review recent progress in string model building in both early and late time cosmology. We describe the main theoretical and phenomenological features of an entire class of inflationary models where inflation is driven by a type IIB Kaehler modulus which enjoys an effective and approximate shift symmetry. We illustrate how reheating can occur via the perturbative decay of the modulus into visible and hidden sector degrees of freedom, paying particular attention to the associated production of axionic dark radiation. We quickly discuss the status of de Sitter vacua versus quintessence model building in string theory, analysing the level of control of these constructions and the main challenges faced by models of dynamical dark energy. We finally present a working model of axion hilltop quintessence in string theory, stressing the importance of initial conditions.

Aziz and Howl argued that a classical gravitational field can generate quantum entanglement through a quantum-field-theoretic channel mediated by virtual matter propagation. However, their claimed channel is more naturally and accurately understood as semiclassical wavepacket motion in an external gravitational field, rather than as a distinctively quantum-field-theoretic entangling effect. Moreover, the result of their perturbative computation is incorrectly magnified: they selected a discontinuous wavefunction with infinite kinetic energy as the initial state and simultaneously treated it as stationary. Once a correct treatment using Gaussian wavepacket is adapted, the resulting effect will be negligibly small.

Engineering large-scale quantum computers which simultaneously provide high-fidelity quantum operations, low memory errors, low crosstalk, and reasonable resource usage remains an outstanding challenge across quantum computing platforms. In trapped ions, progress has largely focused on alkaline-earth and ytterbium ions, whose simple electronic structures facilitate control over their internal state. Here we investigate singly-ionized yttrium ($^{89}\mathrm{Y}^+$), a two-valence-electron ion whose ground-state manifold hosts a nuclear-spin qubit and which also features a variety of low-lying metastable manifolds, for applications in quantum information processing. Because experimental data are limited, we perform high-resolution laser-induced fluorescence spectroscopy to measure the hyperfine structure of several low-lying levels, and carry out comprehensive electronic structure calculations to determine lifetimes, transition matrix elements, and hyperfine coefficients for manifolds addressable with visible, near-visible, or infrared wavelengths. Using these results, we analyze schemes for qubit storage, initialization, readout, leakage mitigation, and single- and two-qubit gates. These results position $^{89}\mathrm{Y}^+$ as a uniquely capable next-generation trapped-ion qubit, combining field-insensitive nuclear-spin or clock-qubit storage with spectrally isolated transitions for operations.

Multifractal Detrended Fluctuation Analysis (MFDFA) is a powerful and widely used technique for characterizing the scaling properties and long-range correlations of complex time series. However, its application often involves significant practical challenges, such as the subjective identification of scaling regions (crossovers) and the disambiguation of the physical origins of multifractality. We introduce MF-toolkit, a high-performance, parallelized Python library designed to address these challenges. It integrates three key innovations: (1) fully automatic crossover detection algorithms (CDV-A and SPIC), which remove operator bias and enhance reproducibility; (2) a built-in implementation of the Iterative Amplitude Adjusted Fourier Transform (IAAFT) for generating surrogate data, enabling the robust identification of the source of multifractality; and (3) a comprehensive suite for generating synthetic time series for rigorous validation. We demonstrate the rigor and utility of MF-toolkit through its application to characterize the multifractal properties of non-stationary noise in gravitational wave (LIGO) data. The MF-toolkit library offers a robust, efficient, and user-friendly tool for advanced time series analysis, facilitating more rigorous and reproducible research across physics and other data-intensive fields.

Quantum key distribution (QKD) has reached a commercially viable stage, with several companies offering hardware systems designed for operational deployment. Evaluating the performance of commercial QKD devices under real-world deployment conditions is essential for users seeking to understand the practical limitations and operational reliability of these systems. In this paper, we present a long-term performance analysis of ID Quantique's Clavis XGR deployed within the Rio Quantum Network, in Brazil. Our study provides a detailed characterization of key operational metrics, such as secret key rate, quantum bit error rate (QBER), visibility, and detection counts, mapping their behavior over extended periods of continuous operation. We analyze the system's stability across two distinct optical links: a 40 km indoor spooled fiber and a 3.5 km outdoor deployed underground fiber. Monitored under both unregulated tropical ambient fluctuations and actively controlled thermal stress, our results demonstrate excellent overall baseline resilience, with the system maintaining visibility above 97% and QBER below 1% on average. These findings provide practical insights into the expected behavior and thermal bottlenecks of commercial QKD systems in field deployments, particularly in tropical climates, helping to inform realistic expectations for operational quantum-safe infrastructures.

Solar-System constraints on a general scalar-tensor theory with generic non-minimal coupling function, non-canonical kinetic function, and scalar potential, are investigated in both the metric and Palatini formalisms. A unified post-Newtonian treatment is developed, yielding analytical expressions for the effective scalar mass, the effective gravitational coupling, and the parametrised post-Newtonian parameters $γ$ and $β$. The results show explicitly how the choice of variational principle affects the weak-field phenomenology. Comparison with Solar-System observations, primarily the Cassini bound on $γ$, indicates that the observational impact of the formalism is strongly model dependent. Generic non-minimally coupled scalar fields may satisfy significantly weaker local bounds in the Palatini case because of stronger Yukawa suppression, whereas in Brans-Dicke gravity the differences are typically small and become appreciable only in restricted regions of parameter space. For the point-particle source considered here, Palatini $f(\hat{R})$ gravity reproduces the general-relativistic exterior post-Newtonian limit, unlike metric $f(R)$ gravity.

Angular correlations in Higgs decays to electroweak gauge bosons, $h \to ZZ^*, WW^*$, provide a powerful probe of both new physics effects and quantum information observables. We present a systematic study of semi-leptonic decays $h \to V V^* \to \ell^+\ell^- q\bar{q}$ and $\ell^\pm ν_\ell q\bar{q}'$, including finite final state fermion masses, NLO QCD, and NLO electroweak corrections. We show that finite final state quark masses can induce effects that go beyond the two-qutrit description in more inclusive regimes, while remaining controllable with suitable kinematic selections. QCD corrections lead to modest percent-level shifts, whereas electroweak corrections can significantly modify the angular structure, particularly in the $h\to ZZ^*$ channels. We assess the impact of these effects on the reconstructed density matrix and entanglement measures, finding that, while they modify the angular observables, semi-leptonic channels retain an effective two-qutrit description.

Commercially-relevant quantum computers will require large numbers of high-performing qubits that can be manufactured, integrated, and controlled at scale. Silicon exchange-only (EO) qubits are a strong candidate modality due to their control-signal simplicity and compatibility with advanced semiconductor manufacturing, but questions remain around the achievability of sufficiently low noise and a scalable control and wiring solution. Here we introduce a quantum processing unit composed of a custom-designed cryogenic CMOS controller, a novel high-density superconducting ribbon cable, and a low-noise EO qubit device. The quantum chip features a three-rail array of 54 exchange-coupled quantum dots, configurable to host up to 18 EO qubits. We integrate and use these components to demonstrate qubit performance for both single-qubit and entangling operations that advances the EO state of the art by an order of magnitude. We further validate this system by implementing a distance-5 repetition code and a quantum error detecting code then make detailed comparisons with simulations. Our approach facilitates a utility-scale quantum computer with manageable operational and capital requirements.

We introduce a method for the selective preparation and detection of quasiparticle wave packets, based on creation operators that generate dressed, localized excitations on top of interacting vacua of (quasi-)one-dimensional quantum lattice theories. This method exploits maximally localized Wannier functions (MLWFs) constructed from quasiparticle bands at intermediate system sizes, enabling the construction of unitary local dressed creation operators. The algorithm allows for species-resolved wave-packet preparation and detection, enabling the separation of known quasiparticle contributions from unknown resonances. We test this approach with matrix product states (MPS) on pure hardcore Hamiltonian QCD on a ladder lattice, detecting scattering outputs and mass resonances.

Quantum error correction is widely believed to be essential for large-scale quantum computation, but the required qubit overhead remains a central challenge. Quantum low-density parity-check codes can substantially reduce this overhead through high-rate encodings, yet finite-size instances with practical logical error rates often achieve encoding rates only around or below $1/10$. Here, building on a recent ultra-high-rate construction by Kasai, we identify new structural conditions on the underlying affine permutation matrices that make encoding rates exceeding $1/2$ compatible with efficient implementation on reconfigurable neutral atom arrays. These conditions define a co-designed family of ultra-high-rate quantum codes that supports efficient syndrome extraction and atom rearrangement under realistic parallel control constraints. Using a hierarchical decoder with high accuracy and good throughput, we study the performance under a circuit-level noise model with $p=0.1\%$, achieving per-logical-per-round error rates of $1.3_{-0.9}^{+3.0} \times 10^{-13}$ with a $[[2304,1156,\leq 14]]$ code and $2.9_{-1.5}^{+3.1} \times 10^{-11}$ with a $[[1152,580,\leq 12]]$ code. These results approach the teraquop regime, highlighting the promise of this code family for practical ultra-high-rate quantum error correction.

Proportional-Integral-Derivative (PID) control is used for automatically regulating a measurable quantity to a desired setpoint. It is widely used in different types of classical control electronics. Here, we show how extending the feedback theory in quantum systems to include the derivative and integral parts influences both the transient and steady-state behavior of the amplitude and squeezing of a mechanical quadrature in an optomechanical system. We show that, in contrast to standard proportional feedback, derivative feedback affects both the conditional and unconditional squeezing. Furthermore, we demonstrate how feedback may be employed to drive a mechanical quadrature to track a desired reference signal. Our findings offer new routes for an improved quantum state control and measurement precision.

Silicon vacancy (VSi) centers in 4H silicon carbide have emerged as a highly promising platform for semiconductor-based quantum technologies, combining excellent spin and optical properties with an industrial-grade, CMOS-compatible material. As these defects are increasingly integrated into practical quantum devices, they inevitably encounter lattice strain. However, while the impact of strain is well documented for other solid-state defects like NV centers in diamond, its specific influence on key VSi spin dynamics such as initialization fidelity and state lifetimes remain largely unexplored. In this work, we address this critical gap by designing fully optical pulse sequences and incorporating the effective spin-3/2 strain Hamiltonian into our analysis. This combined approach allows us to isolate both axial and transverse strain contributions and systematically characterize their effect on the metastable state transition rates. Specifically, we reveal that strain significantly reduces the transition rates from the energetically lowest metastable state to the ground state quartet, leading to decreased photon emission. Supported by first-principles calculations, our findings provide a deeper understanding of VSi spin-strain dynamics, yielding crucial insights for the robust deployment of these centers in realistic, strain-prone environments.

Quantum coherence plays a pivotal role in quantum algorithms. We study the coherence dynamics of the evolved states in Simon's quantum algorithm based on Tsallis relative $α$ entropy and $l_{1,p}$ norm. We prove that the coherences of the first register and the second register both rely on the dimension $N$ of the state spaces of the $n$ qubit systems, and increase with the increase of $N$. We show that the oracle operator $O$ does not change the coherence. Moreover, we study the coherence dynamics in the Simon's quantum algorithm and prove that in overall the coherence is in production when $N>4$ and in depletion when $N<4$.

Turbulent thermal convection governs heat transport in systems ranging from stellar interiors to industrial heat exchangers. Two-dimensional Rayleigh-Bénard convection serves as a paradigm for these flows, reproducing key features such as thin boundary layers, large-scale circulation, and sustained plume dynamics. While Matrix Product State (MPS) methods have demonstrated significant compression of isothermal turbulent fields, their application to buoyancy-driven flows with active thermal coupling has remained unexplored. We apply MPS to two-dimensional Rayleigh-Bénard convection with dynamical simulations up to $\mathrm{Ra} = 10^{10}$. An a priori decomposition of DNS snapshots up to $\mathrm{Ra} = 10^{11}$ shows that the bond dimension $χ$ required to represent the flow fields grows without saturation, in contrast to the plateauing of $χ$ reported for velocity fields in isothermal 2D turbulence. Crucially, however, dynamical simulations solving the governing equations directly in the compressed MPS format at fixed $χ$ show that the $χ$ required to recover statistical observables, such as the Nusselt number, scales significantly more favorably with $\mathrm{Ra}$ than the a priori complexity suggests. At $\mathrm{Ra} = 10^{10}$, a relative error of $1.8\%$ in the mean Nusselt number is achieved with a nearly 9-fold reduction in degrees of freedom, using a $χ$ comparable to that required at $\mathrm{Ra} = 10^{9}$. Spectral analysis confirms the progressive recovery of spatial and temporal scales with increasing $χ$. These findings establish MPS as a scalable tool for simulating thermally driven turbulence, suggesting the method may remain viable for investigations of the ultimate regime at substantially higher $\mathrm{Ra}$.

Quantum protocols require classical signaling, and when classical signals propagate faster than quantum ones, standard rate-loss limits can be surpassed. We introduce an all-photonic measurement-device-independent quantum key distribution protocol that exceeds the single-repeater bound without error correction. When quantum signals travel at two-thirds the classical speed, the key rate scaling approaches $η^{2/5}$. We propose a single-rail, temporally multiplexed architecture that extends twin-field-type protocols to multiple nodes and surpasses their key rate without ideal quantum memories.

Space-based entanglement distribution has the potential to extend the range of quantum communication beyond that achievable through optical fibres that are constrained by exponential losses. Quantum repeaters have been proposed to mitigate the effects of channel losses for both fibre and satellite networks. Although quantum repeaters can improve entanglement distribution efficiency, the rate is constrained by classical communication latency in the entanglement swapping process. Direct dual downlink entangled pair distribution does not suffer such a latency restriction, hence can ``brute force'' the problem of high dual channel loss through increased source rate. Hence, the comparative requirements of direct pair distribution versus quantum repeater satellites are important for the design and deployment of space-based entanglement distribution systems. Here, we consider the simplest case of a single satellite establishing entanglement between two ground stations, comparing the performance of direct dual downlink to that of a space-based quantum repeater for general overpass geometries. We also study the long-term entanglement distribution performance for different ground station pairs and determine altitudinal dependence. Finally, we study the fidelity distribution of a satellite repeater system through Monte Carlo modelling of waiting times and rate statistics, exploring the effect of quantum memory capacity, decoherence rates, and operational policies. These results will inform mission design for future space-borne quantum repeater nodes, as well as requirements on space-based memory platforms.

Nonlinear spectroscopy is a cornerstone of quantum science, providing unique access to multi-point correlations, quantum coherence, and couplings that are invisible to linear methods. However, classical simulation of these phenomena is fundamentally limited by the exponential growth of the Hilbert space, and practical quantum algorithms for the nonlinear regime have remained largely unexplored. Here, we present a unified quantum algorithmic framework for computing $n$-th order nonlinear spectroscopies. By reformulating multi-time responses as a weighted sum of expectation values at finite pump amplitudes via a generalized parameter shift rule, our approach bypasses the costly evaluation of high-order commutators and time-dependent operator expansions. This reformulation enables efficient execution via real-time evolution on current quantum hardware, ensuring inherent noise resilience. We validate the framework on IBM's superconducting quantum processors, successfully obtain higher-order response functions of a 12-qubit XXZ spin-chain. Furthermore, the versatility of our method is demonstrated by resolving quasi-particle excitation spectra in spin-liquids and identifying interaction-induced cross-peaks in atomic systems. Our results establish a practical and scalable pathway for probing complex quantum dynamics on near-term quantum devices, extending the reach of quantum simulation into the nonlinear domain.

Fast Radio Bursts (FRBs) are among the most intriguing phenomena observed in radio astronomy. So far, about 130 FRB signals have been confirmed and characterized by different surveys, and the CHIME telescope has recently reported a new catalog of 4539 bursts. Therefore, these numbers are expected to increase in the coming years. The detection, or lack thereof, of lensed FRB events can be used to probe Primordial Black Holes (PBHs) as a fraction of dark matter. We investigate the potential of three upcoming radio telescopes, LOFAR2.0, FAST Core Array, and BINGO, to test the PBH scenario. We forecast that LOFAR2.0 will constrain $f_{\mathrm{PBH}} < 0.16$ for PBH masses $M_{\rm PBH}>1\,{M_{\odot}}$, while FAST Core Array and BINGO will restrict $f_{\mathrm{PBH}} < 0.39$ for $M_{\rm PBH}>10\,{M_{\odot}}$ and $M_{\rm PBH}>10^{-2}\,{M_{\odot}}$, respectively. Despite the existence of stricter constraints, FRB lensing offers an independent and complementary probe of PBHs in the Universe, which will improve in the future.

The presence of degeneracies plays a crucial role in describing the behavior of non-Hermitian (NH) systems. In these systems, there are two key types of degeneracies: $n$-bolical degeneracies, which are analogous to Hermitian degeneracies, and various forms of exceptional points, each associated with different orders that correspond to sizes of the Jordan blocks. These types of degeneracies may coalesce at the same energy level, forming multi-block degeneracies. To understand how a multi-block degenerate NH system responds to perturbations, one should address how each types of involved degeneracies disperse. In this work, we systematically characterize the asymptotic behavior of all types of multi-block degeneracies in NH systems using a rigorous mathematical formulation. Through a range of examples, we demonstrate that our algebraic approach can facilitate the analysis of NH degeneracies in various settings relevant to experiments.

The emergence of various types of degeneracies plays a crucial role in optimizing and engineering different physical phenomena in non-Hermitian physics. In our work, we focus on the derogatory Exceptional Points (EPs), which are characterized by multiple Jordan blocks corresponding to the same eigenvalue. We demonstrate that, under certain infinitesimal perturbations, a derogatory EP can be converted into an EP of different structure without varying the total order of degeneracy. In particular, such conversion can increase the size of the largest Jordan block and, hence, the sensitivity of the eigenspectrum to parameter variation, which is an important feature for practical applications. Furthermore, by analyzing all possible conversions, we introduce hierarchies of degeneracies of the same order that appear when perturbing non-Hermitian systems. We systematically explore hierarchies in the absence of any symmetry and when pseudo-Hermitian symmetry is present. Our study facilitates engineering various degeneracies of non-Hermitian systems, paving the way to extending the implications of non-Hermitian physics.

Symmetry-based classification of quantum phases of matter is one of the most foundational organizing principles in physics; however, an analogous framework for mixed, decohered quantum states has only begun to emerge. A central new concept is strong-to-weak spontaneous symmetry breaking (SW-SSB), a sharp transition in mixed quantum states that is invisible to any observable linear in the density matrix and that has since been predicted across a broad class of open and monitored quantum systems. It also provides a unifying language for phenomena as disparate as the decodability of topological quantum memories and the emergence of classical hydrodynamics from decohered quantum dynamics. Here we report the first experimental observation of SW-SSB, in dephased single-component fermionic matter imaged by a quantum gas microscope. A quantum-classical estimator built on a machine-learned Gaussian reference state gives direct access to the nonlinear Rényi-1 and Rényi-2 correlators that diagnose SW-SSB, and reveals long-range Rényi order in the dephased Fermi liquid. Adding a commensurate superlattice drives the underlying fermions through a metal-to-insulator transition that, after full dephasing, manifests as a sharp SW-SSB phase transition. Our results uncover the symmetry principle behind information-theoretic transitions in open quantum systems, and extend Landau's symmetry paradigm into the regime of real, decohering quantum devices.

We theoretically study how quantum measurement noise can be engineered in a hybrid cavitymagnomechanical platform for precision force sensing. The proposed configuration consists of a driven optomechanical cavity, with a movable mirror on one side plus a fixed semi-transparent mirror on the other side, coupled to a magnon mode, with an OPA placed inside the cavity. We show that the magnon mediated dynamics reshapes the added-noise spectrum leading to improved sensitivity compared to a conventional optomechanical sensor. In particular, by satisfying the coherent quantum noise cancellation (CQNC) criterion, radiation-pressure back-action can be fully suppressed. In addition, a larger OPA pump gain permits operation beyond the standard quantum limit at substantially reduced laser power, thereby mitigating power-related constraints without sacrificing performance. These combined advantages provide a practical pathway to below-SQL weak force detection and can outperform existing approaches based on squeezing in magnomechanics.

In a double-slit experiment with a bipartite system, the visibility of interference fringes depends on the availability of which-way information. Here, we report the formation of a Bell-like state of photoelectron and residual ion in the multiphoton dissociative ionization of the D$_2$ molecule. Evidence for entanglement is provided by the correlated emission directions of photoelectron and ion, which is observed using a COLTRIMS reaction microscope. In the presence of this correlation, the holographic interference fringes contained in the photoelectron momentum distributions are suppressed, indicating the existence of which-way information. We show that the which-way information is erased, and the interference pattern is restored, when a single ionic state is selected. The experimental observations and conclusions are fully supported by the numerical solution of the electronic-nuclear time-dependent Schrödinger equation. Our work demonstrates that coincidence spectroscopy of ions and electrons is a powerful method for studying fundamental concepts of quantum information science within the context of ultrafast light-matter interactions.

We propose a quantum-resistant quantum teleportation (QRQT) framework protected by post-quantum cryptography (PQC) to secure the classical correction channel, which is vulnerable to quantum adversaries. By applying PQC to the classical control bits, QRQT eliminates the classical attack surface of quantum teleportation. Our analysis reveals that quantum memory is a hidden bottleneck linking physical and computational security: its finite coherence time simultaneously limits communication distance, constrains tolerable PQC overhead, and restricts the adversary attack window. Under realistic parameters (1 ms coherence, fiber-optic propagation), the maximum secure teleportation distance ranges from 191 km (FrodoKEM-1344) to 199 km (Kyber512). We show that the joint classical-quantum attack probability exhibits a non-monotonic, Bell-shaped profile due to the opposing time dependencies of classical cryptanalysis and quantum decoherence, establishing a bounded optimal attack window beyond which adversarial success decays exponentially. We further analyze how leakage of classical correction bits affects teleportation security under four stochastic leakage models: independent exponential, sequential, burst, and correlated leakage, also accounting for amplitude damping on the shared Bell pair. For each scenario, we derive closed-form expressions for the average Holevo quantity and teleportation fidelity as functions of time, providing measurement-independent upper bounds on extractable information and guiding the design of leakage-resilient quantum communication protocols.

Thermodynamics provides a useful interpretation of scalar-tensor gravity, in which the effective imperfect fluid admitted by the nonminimal coupling features a temperature that is associated with the departure from general relativity. However, in this construction, certain thermodynamical quantities are defined with respect to a particular conformal frame. In the present work, we show that the originally proposed effective temperature assigned to nonminimally coupled scalar field theories is not frame invariant, and can thus be arbitrarily tuned by a change of frame. This raises the question of whether temperature can be viewed as an intrinsic property of a scalar-tensor theory rather than a particular representation of it. Working instead with the frame invariant formulation of scalar-tensor gravity, we find that the frame invariant effective fluid is perfect with identically vanishing temperature. The departure from general relativity is then governed not by temperature, but rather by a frame invariant chemical potential, similar to minimal theories. Therefore, general relativity can be interpreted as a state of diffusive equilibrium for any scalar-tensor theory, regardless of whether it is minimal or nonminimal.

Distance-bounding (DB) protocols let a verifier upper-bound a prover's physical distance by timing rapid challenge-response exchanges. Quantum communication promises simpler DB protocols with stronger security guarantees, yet existing quantum distance-bounding (QDB) proposals are analysed in ad-hoc models and, to the best of our knowledge, lack a common game-based treatment of standard fraud attacks. We contribute (i) a reusable security framework for QDB that fixes system and timing assumptions, specifies a quantum-capable adversary model, formalises distance-, mafia-, and terrorist-fraud experiments, and includes a simple i.i.d. depolarizing noise model; and (ii) an application of this framework to a published QDB protocol. For this protocol we characterise the honest per-round acceptance probability under noise and lift it to the multi-round setting, yielding explicit completeness guarantees as a function of the number of fast rounds, the acceptance threshold, and the noise parameter. For active adversaries we bound the per-round success probability of distance-fraud attacks and analyse the best known mafia-fraud strategy, deriving corresponding multi-round soundness bounds. We also show that the protocol is inherently insecure against terrorist-fraud in our model. The framework cleanly separates protocol-independent definitions from protocol-specific analysis and can be used to evaluate existing and future QDB protocols on a common basis.

We point out that the method presented in a recent arXiv article by Jia et al. (arXiv:2512.21848) for constructing local hidden-state models closely follows the framework we developed in N. von Selzam & F. Marquardt (PRX Quantum, 2025) for constructing local hidden-variable models. While Jia et al. cite our work, the extent of the methodological overlap and the degree of textual similarity are not adequately reflected by the attribution given. We document this overlap in detail.

Ground-based gravitational wave detectors are now routinely surveying the dark Universe, finding hundreds of collisions between compact objects such as black holes and neutron stars. However, terrestrial non-Gaussian noise artefacts, commonly known as glitches, reduce the sensitivity to signals and can overlap signals, producing biased astrophysical inferences. We introduce a hierarchical Bayesian model to measure the glitch rate, which improves upon existing trigger-counting methods in its capacity to measure the rate down into the low signal-to-noise regime without contamination from the Gaussian noise background, provided the population is accurately modelled. The methodology builds on standard hierarchical inference, but includes several novel features, including hierarchical inference with quantile compression (HIQC), a generic approximation method for the hierarchical recycled likelihood, and a time-domain rate estimated by fitting basis functions. We validate the methodology using simulated data with injected glitches and then apply it to data from the fourth LIGO-Virgo-KAGRA observing run, demonstrating time-resolved inferences of the glitch rate over a 24 h period. The inferred glitch rate is consistent with estimates from trigger counts, but does not require an arbitrary threshold and provides a more fine-grained view of the temporal behaviour. Finally, we demonstrate how our individual-detector rate estimates can be transformed into a coincident glitch probability and utilise this to validate that the retracted gravitational-wave candidate GW230630_070659 is likely a pair of coincident glitches.

We develop an intrinsic enumerator framework for quantum error correction in unitary representations of symmetry groups. An intrinsic quantum code is a subspace of a representation $V$ of a group $G$, and errors are organized by the decomposition of the conjugation representation on $\mathcal{L}(V)$ into isotypic subspaces. Associated with any orthogonal decomposition of $\mathcal{L}(V)$ we introduce two families of quadratic enumerators, called projector and twirl enumerators, which satisfy positivity, normalization, and Knill--Laflamme type inequalities. When the conjugation representation is multiplicity--free, these enumerators are related by a linear transform that we interpret as an intrinsic MacWilliams identity. For $G=\mathrm{SU}(2)$, we compute this transform explicitly in terms of Wigner $6j$-symbols. Applied to symmetric-power representations, this gives linear programming bounds for permutation-invariant qubit and qudit codes, including extremality results for the four-qubit, seven-qubit, and three-qutrit examples treated here. We also develop the general equivariant theory in the presence of multiplicities, where the enumerators become matrix-valued, the MacWilliams transform becomes block unitary, and the resulting feasibility problem becomes semidefinite; we illustrate this theory in a first non-multiplicity-free $\mathrm{SU}(3)$ example.

The double Schwarzschild solution in the equal mass case is studied in bispherical coordinates. An explicit conformal transformation from cylindrical Weyl coordinates to bispherical coordinates is given in terms of elliptic functions. A multi-domain spectral method for spacetimes in bispherical coordinates is presented to numerically reconstruct this solution.

Gravitational wave (GW) observations of binary black hole (BBH) mergers provide a unique opportunity to probe the nature of spacetime in the strong-field and dynamical regime. We present updated constraints on deviations from the Kerr metric using BBH inspirals from the fourth Gravitational-Wave Transient Catalog (GWTC-4). Building on our previous GWTC-3 analysis, we employ a theory-agnostic framework in which non-Kerr effects of the Johannsen metric are incorporated as parametrized corrections to the GW phase within the post-Newtonian framework. We perform Bayesian parameter estimation on a selected subset of GWTC-4 events to constrain the deformation parameters $α_{13}$ and $ε_3$, yielding significantly tighter bounds compared to earlier results. When varied individually, the deformation parameters are found to be consistent with zero, providing no evidence for departures from the Kerr geometry. Our results reinforce the validity of General Relativity, particularly the Kerr hypothesis, and highlight the progress enabled by GWTC-4.

We study a minimal two-field scalar-tensor completion of Starobinsky inflation motivated by the one-loop effective action of scalar-tensor gravity. The model admits an exact Starobinsky branch, but the relevant question is whether nearby trajectories generate observable multifield effects. We show that a non-trivial class of initial conditions relaxes to an attractor-connected slow-roll branch continuously connected to the Starobinsky solution. We then solve the coupled adiabatic and entropy perturbation equations numerically. On the branch studied here, the entropy mode remains sufficiently suppressed that its sourcing of the curvature perturbation is negligible, while the tensor sector is unchanged. The inflationary observables therefore remain effectively Starobinsky-like, providing a robustness test of Starobinsky inflation against a minimal radiative scalar-tensor deformation.

We present two general methods to implement quantum circuits for the direct measuring of local unitary invariants on quantum computers. We work these out for important three-qubit invariants, and also demonstrate these on the IBM Quantum Platform for important entanglement measures of three qubits.

We study flat Friedmann-Lemaître-Robertson-Walker (FLRW) models with a perfect fluid matter source and a scalar field minimally coupled to matter with power-law-exponential \textquotedblleft hybrid\textquotedblright potential. Using expansion-normalised variables, we formulate the field equations as a constrained three-dimensional dynamical system and determine its equilibrium structure. We show that viable cosmological histories, consisting of a transient matter or radiation era followed by late-time accelerated expansion, arise in restricted regions of parameter space. A central result is that the physically relevant trajectories are confined to an invariant plane, which contains both the transient matter point $\mathcal{B}$ and the accelerated point $\mathcal{C}$. We further show, by centre-manifold analysis, that the accelerated point $\mathcal{C}$ is not a global attractor: it attracts trajectories with $φ>0$ and repels those with $φ<0$. For dust, a standard matter era requires vanishing coupling of the scalar field to matter, while for radiation the interaction term vanishes identically. Finally, we discuss the issue that the qualitative cosmological dynamics may be independent of the precise functional form of the scalar-field potential.

Rydberg atoms, which possess exceptionally large electric dipole moments, offer a promising route for electric field sensing as well as metrology traceable to the International System of Units (SI); however, current research predominantly focuses on the microwave (MW) regime, leaving the quasi-direct current (quasi-DC) and low-frequency bands, ubiquitous in power systems, largely unexplored. In this paper, we present a theoretical investigation into low-frequency electric field detection. To this end, we establish a comprehensive modeling framework incorporating Fisher information (FI) and the Cramér-Rao lower bound (CRLB) to quantify the fundamental precision limits of electromagnetically induced transparency (EIT) readouts. Building upon this framework, we propose a linearized sensing strategy utilizing a DC-biased two-point differential measurement. Numerical validations demonstrate that this approach effectively mitigates the weak-field insensitivity for both DC and AC fields, achieving a CRLB-limited sensitivity bound of approximately $1\times 10^{-4}$ V/m/$\sqrt{\text{Hz}}$. Furthermore, to surpass the single-pass sensitivity limit, we introduce a Fabry-Pérot (FP) cavity-enhanced configuration. This architecture leverages intracavity phase modulation to significantly steepen the transmission slope, boosting the FI by over two orders of magnitude compared to standard free-space configurations. This work provides a rigorous theoretical basis and design guidance for the high-precision quantum monitoring of electromagnetic environments in smart grids.

Flux-tunable superconducting qubits rely on fast flux control pulses to implement two-qubit entangling quantum gates, a key building block for quantum algorithms. However, distortion effects introduced by non-ideal control electronics, parasitic components, and the cryogenic quantum chip response can all degrade the gate fidelity. We present a digital predistortion (DPD) framework for characterizing and then compensating for these distortions using a combination of infinite impulse response (IIR) and finite impulse response (FIR) filters. Experiments on a flux-tunable quantum processing unit (QPU) demonstrate a successful correction of step-response distortions on the flux-control line, with a compensated control signal showing only sub-percent deviations from the ideal target linear behavior. The demonstrated method enables automated rapid calibration of flux control channels for superconducting QPUs.

Grover's algorithm is a cornerstone of quantum algorithms and is strictly optimal in oracle-query complexity. While the full search problem admits no further improvement, one may trade accuracy for speed in the partial search problem, where the task is to identify only the block containing the target item. The best known quantum algorithm for the partial search problem is the Grover-Radhakrishnan-Korepin (GRK) algorithm, whose optimality has long been conjectured but not proved. In this work, we prove the optimality of GRK in the large-block limit. We formulate partial search as a time-optimal control problem and apply the Pontryagin maximum principle to derive the switching-function dynamics, establish the bang-bang structure of regular extremals, and exclude non-optimal switching patterns. As a result, we show that the optimal regular extremal has the global-local-global form, which yields a control-theoretic proof of the asymptotic optimality of the GRK algorithm in oracle-query complexity.

Gravitational-wave events are interpreted in terms of Bayesian posteriors for their source properties inferred under unphysical reference priors. Though these parameter estimates are important intermediate data products for downstream analyses, across the catalog they provide generically biased sourced properties and are therefore unsuitable for direct astrophysical interpretation. Hierarchical parameter estimation is the solution, where joint analysis of the entire catalog of observations not only reduces statistical uncertainties but actually informs the correct prior. Population-informed source properties from there derived are naturally suited to astrophysical interpretation and catalog statistics, such as identification of exceptional events from previous and ongoing observing runs. Using the latest LIGO-Virgo-KAGRA data, we thus demonstrate that population inference is not optional to interpret gravitational-wave observations.

The Hadamard test is a standard quantum primitive for estimating inner products and expectation values, but in data-processing settings its practical utility is often limited by the cost of preparing amplitude-encoded quantum states. In this study, we investigate an angle-encoding variant of the Hadamard test for estimating cosine similarity between normalized real-valued vectors. The proposed method decomposes the similarity computation into elementwise two-qubit Hadamard-test circuits that can, in principle, be executed in parallel, resulting in constant circuit depth with respect to the vector dimension at the expense of a larger qubit footprint and classical post-processing. Because the resulting estimator is approximate, we analyze the induced bias and show that it is non-negative under the approximation used in our derivation. Numerical experiments on random normalized vectors show that, in the tested setting, the estimation error decreases as the vector dimension increases. We further illustrate a possible application to cosine-attention-based Transformer models. These results suggest that the angle-encoding Hadamard test may provide a useful design point for near-term similarity estimation when shallow circuit depth is preferred over compact qubit usage.

Weber's conjecture (1886) governs three aspects of lattice-based cryptography: the solvability of the Principal Ideal Problem, the freeness of modules over rings of integers, and the tightness of worst-case-to-average-case reductions in Ring-LWE (R-LWE) and Module-LWE (MLWE). Existing verifications for $k \ge 9$ rely on Generalized Riemann Hypothesis (GRH). In this paper, we present the first unconditional proof for $k \le 12$. Our method combines the Fukuda-Komatsu computational sieve, inductive structure of the cyclotomic $\mathbb{Z}_2$-tower, and Herbrand's theorem.

Semi-quantum signature (SQS) schemes aim to enable quantum signature functionality in scenarios where only a subset of participants possess full quantum capabilities, thereby improving practical deployability while preserving quantum security advantages. Within this framework, we present a practical SQS protocol based on Bell states. The protocol is designed so that only the signer requires full quantum capability, significantly alleviating the quantum burden on the remaining participants. To strengthen security in semi-quantum environments, we incorporate an improved eavesdropping-detection mechanism that more effectively detects tampering. Compared with many existing schemes, which do not explicitly consider tampering of already generated signatures in their unforgeability analyses, the proposed protocol is designed to remain secure in the presence of such tampering.

Enhanced local-excitation retention in atomic arrays allows to exploit cooperative radiative effects to suppress emission and prolong excited-state lifetimes. We consider an impurity-assisted setting involving a single storage atom being initially excited and study the survival of local excitation under neither write nor retrieval fields. Because the corresponding dynamics can involve multiple interfering collective modes, the survival dynamics cannot determined from the smallest collective decay rate alone. Thus, using a biorthogonal eigenmode decomposition of an effective non-Hermitian Hamiltonian, we show that the survival dynamics are jointly governed by the decay rates of the eigenmodes and their overlaps with the initial excitation. Large oscillations occur when multiple long-lived modes have comparable weights. Accordingly, we introduce a physically motivated spectral surrogate objective that favors both small weighted decay rates and an initial-state weight concentrated on a single subradiant mode. As a proof of principle of this spectral design, we apply the surrogate to constrained atom-position optimization under minimum-distance constraints and obtain nontrivial aperiodic configurations with enhanced local-excitation retention. Our findings unveil spectral design principles for local-excitation retention in impurity-assisted atomic arrays and provide a proof of principle for their inverse design.

We compute the efficiency of the reversible Stirling engine, with and without regeneration, for a broad class of working substances including Van der Waals fluids, quantum ideal gases (Bose and Fermi), Bose-Einstein condensates, thermal conformal field theories (CFTs), and holographic CFTs. Regeneration acts as an internal heat recycling mechanism that enhances efficiency by reducing the net heat exchange with external reservoirs. For regenerative Stirling cycles, a central role is played by the intrinsic heat mismatch between the two isochoric branches, which controls the deviation of the efficiency from the Carnot bound and quantifies the extent to which internally exchanged heat can be perfectly recycled. We identify a general sufficient condition for attaining Carnot efficiency, namely that the fixed-volume heat capacity is independent of the volume, ensuring that the isochoric heat mismatch vanishes. While this condition is satisfied for classical ideal gases and Van der Waals fluids, it is violated for quantum ideal gases and CFT working substances. For thermal CFT states dual to AdS-Schwarzschild and AdS-Reissner-Nordström black holes we obtain exact expressions for the Stirling efficiency. In the fixed-potential ensemble, we show that the Stirling efficiency asymptotes to the Carnot value in the large-potential limit, with a faster approach in the presence of regeneration.

Classical simulation of quantum operations is essential for algorithm design, noise characterisation, and benchmarking of quantum hardware. The most general physically realisable operation can be described by a positive linear map acting on a hermitian operator, representing either a density matrix or an observable. Established simulators vectorise the density matrix on an $n$-qubit Hilbert space and reuse state-vector kernels, storing all $2^{2n}$ elements and forgoing the benefits of hermitian symmetry. In this work, I introduce \emph{Orkan}, a simulation library that uses a tiled memory layout storing only the lower triangle of the hermitian matrix at tile granularity, roughly halving both the memory footprint and the wall time to simulate the evolution of quantum states under generic quantum operations. The implementation treats any hermitian operator uniformly and is agnostic to whether the Schrödinger or Heisenberg picture is used. Dedicated $k$-local conjugation algorithms update all entries of the hermitian matrix in a single pass. Benchmarks against Qiskit Aer, QuEST, and Qulacs show consistent wall-clock speedups of $2$-$4{\times}$ partly attributable to the reduced memory footprint.

A broadband electromagnetic source is important for scientific and technological applications. Quantum vacuum fluctuations, which manifest most prominently in the Casimir effect, provide a fundamentally broadband electromagnetic source. Here we explore a potential consequence of the broadband nature of quantum vacuum fluctuations, by showing that such fluctuations can enable measurement of material permittivity over a broad frequency range. Specifically, we consider the Casimir force in a parallel-plate geometry, with one plate covered by a nanoscopic thin film. Using a machine learning approach, we show that one can infer both the thickness of the film and its permittivity over a broad frequency range, starting from the dependency of the Casimir forces on the spacing between the two plates. Our work highlights the application potential of using vacuum fluctuations as a naturally-existing broadband electromagnetic source for material characterization, and shows that the inverse problem in Casimir force calculation can be solved with machine learning.

We present a new differential mechanical gradiometer for the detection of low-frequency Gravitational Waves. The frequency range is 0.05 to 1 Hz, a frequency gap not covered either by future space-based detectors such as LISA or by ground-based observatories such as Einstein Telescope or Cosmic Explorer. The proposed detection principle is similar to antennas based on torsion pendulums but solves the problem of physical confinement of these antennas by operating vertically and by having a counterweight at one end of each bar and a mass suspended from a long wire at the other. With this configuration, we enlarge the gravitational force acting on the system \textit{without} changing the moment of inertia of the system, so that we move from a signal $Δθ$ of the order of $Δθ= h$, where h is the amplitude of the gravitational wave, to a signal of the order $Δθ= h\frac{L}{D}$, where D is the length of the arm and L is the length of the wire suspending the test mass. This configuration is a further evolution of the recent development of tiltmeters and balances with double suspended arms and interferometric read-out, where the main working principles are already tested. The expected sensitivity will be discussed with respect to the proposed parameters and the present technology.

We study a simple theory based on general relativity, minimally coupled to a constrained scalar triplet and to an auxiliary non-propagating three-form sector. Within a spherically symmetric hedgehog ansatz, the theory admits a continuous exact family of asymptotically flat geometrically regular black holes. For a simple choice of kinetic function, the solutions possess a de Sitter core and approach Schwarzschild with the first correction appearing only at order $r^{-4}$. We analyse their horizon structure, thermodynamics, and main strong-field properties. The black holes carry topological scalar hair and a continuous secondary parameter, but no scalar charge. The regularity established here is geometric: the curvature invariants remain finite, although the matter sector is not completely smooth at the centre.

We demonstrate the use of two-photon excitation for observing the ground state optically detected magnetic resonance (ODMR) of nitrogen-vacancy centers in diamonds at room temperature. An ultrafast femtosecond laser at 1040 nm was used for excitation, while fluorescence signal read out was achieved through a combination of a PMT and a lock-in amplifier. The imaging capability of two-photon excitation fluorescence (2PEF) was utilized to map the distribution of NV centers in a bulk diamond and micro-sized diamonds. For the first time, ODMR traces of the nitrogen-vacancy center are observed with two-photon excitation, providing a promising tool for fast 3D quantum sensing and imaging.

The geometry of quantum states has profound implications in quantum multiparameter estimation. While the Riemannian structure of quantum state space is well understood, the full understanding of the curvature structure of mixed quantum states is still an open problem. Inspired by the Yang-Mills action in non-Abelian gauge theory, we propose a scalar quantifying the Uhlmann curvature and establish its connection to the measurement incompatibility in quantum multiparameter estimation problems. We show that this curvature measure is gauge invariant, reparametrization invariant, and vanishes if and only if the Uhlmann curvature vanishes. We also explicitly calculate the Uhlmann curvature for the joint estimation of phase and phase diffusion as an example.

Non-commutative electrodynamics obtained through the Seiberg-Witten map ceases to have equivalent action-level and equation-level realizations once fixed external currents are introduced, and in the action-level construction associated with the Banerjee current map the canonical location of this source-induced obstruction has remained unclear. Working in the full phase space and treating the current as prescribed and non-dynamical, we apply the Dirac-Bergmann algorithm without imposing current conservation as an external condition. The preservation of the Gauss-type secondary constraint produces a third-stage candidate whose phase-space expression is shown to be algebraically identical, at first order in the non-commutativity parameter and for purely space-space non-commutativity, to the canonical pullback of the divergence of the mapped Euler-Lagrange equations. This identity locates the source-compatibility obstruction directly within the Dirac chain. For generic inhomogeneous sources, the next consistency step feeds this object back into the primary multiplier through a source-dependent kernel, so the chain closes by multiplier fixing rather than by the generic appearance of a quaternary constraint. Reduced-phase-space results, including the gauge generator, Dirac brackets and degree-of-freedom count, are obtained only in a restricted sufficient first-class subcase; no broader claim is made for arbitrary source profiles.

To study generator design for parameterized unitaries in quantum machine learning (QML), we propose an observable-guided generator selection algorithm for $ n $-qubit Pauli-string generator pools. The proposed method selects generators based on two criteria: maintaining large first-order sensitivity in the gradients and suppressing second-order interference in the Hessian matrix. Under a restricted setting with Pauli-string observables and candidate generators, the selection problem can be formulated as a binary optimization problem that favors mutually anti-commuting generators. Numerical experiments on a synthetic dataset with a small-scale five-qubit circuit show that the selected generators yield faster training than random generator selection in our setting, while exhibiting similar expressibility. Furthermore, under additional algebraic assumptions, the proposed criteria admit an interpretation in terms of the $ \mathfrak{g} $-purity of the observable: the first-order sensitivity is proportional to the $ \mathfrak{g} $-purity, whereas the second-order interference, namely the off-diagonal elements of the Hessian matrix, is upper-bounded by it. These results suggest that observable-guided generator selection is a promising direction for improving trainability in restricted QML settings.

Sgr A* exhibits a persistent negative circular polarization (CP) at 230\,GHz, offering a powerful probe of the magnetic field geometry in its accretion flow. Using a stationary semi-analytic radiatively inefficient accretion flow (RIAF) model in Kerr spacetime with polarized radiative transfer, we systematically analyze CP images for six poloidal magnetic field configurations across varying black hole spins, inclinations, and field polarities. We find that CP production is dominated by Faraday conversion in radial, parabolic, quadrupole, and combined geometries, but by intrinsic emission in dipole and vertical fields. The radial and parabolic configurations produce the polarity-invariant net CP, while dipole and vertical fields yield the polarity-sensitive one. As the accretion disk is prograde with respect to the black hole spin, the CP production across all six field geometries is found to be lower at high spin case, while the situation is more complicated in the retrograde case. Moreover, the net CP observed from edge-on views $V_{\rm net} \approx 0$ except for the quadrupole geometry. Comparing with ALMA data, the reversed-field model is excluded at high inclinations and then the magnetic field geometry of Sgr A* is constrained.

Hybrid variational quantum algorithms are promising for solving practical problems, such as combinatorial optimization, quantum chemistry simulation, quantum machine learning, and quantum error correction on noisy quantum computers. However, variational quantum algorithms (derived from randomized hardware-efficient ansatz or adaptive ansatz) become a black box, not trustworthy for model interpretation, and not to mention for application deployment in informing critical decisions. In this paper, we construct the first interpretable quantum regression algorithm, in which the quantum state exactly encodes the classical data table and the variational parameters correspond directly to the regression coefficients, which are real numbers by construction, providing a high degree of model interpretability and minimal cost to optimize due to the right expressiveness. We also exploit the encoded data structure to reduce the gate complexity of computing the regression map. To reduce circuit depth in nonlinear regression, our algorithm can be extended by directly constructing nonlinear features via classical preprocessing, such as independent encoded column vectors. By design, the model performance is determined by the cost function measurement results $\mathcal{C}$ synchronous to the mean squared errors (MSE) for the regression models. We derived the read-out errors induced by one-hot encoding and compact encoding; the required physical qubit resources are exponentially compressed for the compact encoding to be favorable for noisy quantum devices. We also derive the cost function dependent sample complexity $ \in \mathcal{O}\left(σ^{2}(\mathcal{C}) \ln (1/α)/ε^{2}\right)$ under the error budget $ε$ and confidence tolerance $α$.

We introduce an ansatz for force-free electrodynamics in Minkowski spacetime under which the nonlinear system reduces to a semilinear scalar wave equation depending on two spacetime variables. This reduction yields explicit time-dependent solutions, including type-changing configurations with finite energy per unit transverse area and a null kink-type example. For traveling-wave solutions in the magnetically dominated regime, the kernel distribution of the field defines minimal field-sheet foliations.

We introduce a Hybrid Quantum Residual Network (HQRN) and establish an exact functional correspondence between its state evolution and the dynamics of classical networks with residual connections. When inputs are restricted to the computational basis, the HQRN reduces to its classical analog, enabling the direct translation of optimized classical weights into quantum unitary operations, effectively inheriting the landscape benefits of classical optimization. Conversely, when processing general mixed states, the HQRN leverages off-diagonal quantum correlations to resolve features inaccessible to its classical analog. We validate this framework through digit recognition and bipartite entanglement classification. Notably, HQRN achieves high classification accuracy even for adversarial separable states that mimic the marginal measurement statistics of entangled pairs. Our results bridge the gap between classical and quantum residual learning, paving a scalable pathway for deep quantum architectures.

We investigate quantum thermal state preparation algorithms based on system-bath interactions and uncover a surprising phenomenon in the weak-coupling regime. We rigorously prove that, if the system-bath interaction is engineered so that the transition part of the approximate Lindbladian generator satisfies the KMS detailed balance condition, then the unique fixed point of the dynamics can be made arbitrarily close to the Gibbs state in the weak-coupling limit, regardless of the structure of the Lamb shift term. Importantly, this remains true even when the approximate Lindbladian differs substantially from the ideal Davies generator and the Lamb shift term does not commute with the thermal state. Our result shows that the role of the KMS detailed balance condition extends well beyond standard Lindbladian dynamics, serving as a general principle for a broader class of dissipative systems. Furthermore, by combining this with a general perturbation framework, we bound the mixing time of the dynamics and establish an end-to-end complexity of $O(\varepsilon^{-1})$ for Gibbs state preparation. These guarantees apply to any Hamiltonian for which the corresponding KMS-detailed-balance Lindbladian is known to mix rapidly.

The controlled generation of nonclassical light beyond single photons remains a central challenge in quantum optics, due to the difficulty of enhancing multiphoton processes while suppressing lower-order excitations. Here we propose an interference-interaction-engineered scheme for programmable few-photon emission in a cavity-QED system of three atoms coupled to orthogonal cavity modes. By adiabatically eliminating an auxiliary Fabry-Pérot cavity, we generate a tunable cavity-mediated spin-exchange interaction $χ$, which, combined with a controllable geometric phase $φ$, reshapes the many-body dressed-state spectrum. This interplay enables selective addressing of excitation manifolds ($N=1,2,3$), establishing a direct mapping between excitation structure and photon-emission channels. For $φ=0$, constructive interference enhances the spectral anharmonicity of low-excitation manifolds, yielding tunable single- and two-photon emission associated with the $N=1$ and $N=2$ manifolds. In contrast, for $φ=2π/3$, destructive interference suppresses lower-order excitation pathways and activates a resonant three-photon channel originating from the $N=3$ manifold. Importantly, the cavity-mediated interaction $χ$ further enhances spectral separation between manifolds, enabling a substantial improvement in multiphoton purity while maintaining a sizable photon population. We demonstrate a three-order-of-magnitude enhancement in two-photon purity and more than two orders of magnitude improvement in three-photon emission. Our results establish a unified interference-interaction framework in which effective optical nonlinearities can be programmably engineered through phase and interaction, providing a scalable route toward high-purity multiphoton sources and programmable quantum photonic devices.

The generation of high-performance nonclassical light remains a cornerstone of quantum technologies, yet faces a fundamental trade-off between emission purity and brightness. Here, we demonstrate that cavity-mediated many-body spin-exchange interactions provide a route to overcome this constraint by collectively amplifying spectral anharmonicity. In a cavity-coupled atomic array with a programmable relative phase $φ$, the resulting interference-interaction mechanism reshapes the dressed-state manifold and enables deterministic switching between distinct quantum emission regimes. For $φ=0$, constructive interference yields high-purity single-photon emission with antibunching improved by four orders of magnitude while preserving strong photon flux. Conversely, for $φ=π$, destructive interference creates a dark single-photon manifold, resonantly activating two-photon processes to produce bright and pure photon-pair bundles. Our work establishes interference-engineered many-body interactions as a scalable mechanism for on-demand quantum light generation and open a new avenue for harnessing collective many-body physics in quantum photonics.

Group-equivariant quantum models are designed to exploit symmetry and can improve trainability, but it remains unclear how symmetry constraints shape their adversarial robustness. We study this question through a feature-level analysis of equivariant quantum models in a transfer-attack setting. Under equivariance with an invariant readout, predictions depend only on the group-twirled input, which identifies the symmetry-invariant information accessible to the model together with a complementary uninformative subspace. Specializing this framework to a rotationally equivariant quantum model, we derive an explicit characterization of the accessible information in terms of rotation-invariant image statistics distributed across distinct symmetry sectors. Using targeted input transformations, we determine which of these statistics are actually relied upon for classification across several datasets. We find that equivariance alone does not guarantee transfer robustness: even within the restricted invariant feature space, the model can rely on brittle statistics, particularly ring-averaged intensities in the rotationally equivariant model, that remain vulnerable to classical transfer attacks. Guided by this analysis, we show that suppressing the symmetry sector associated with the brittle feature substantially improves robustness. These results establish a systematic mechanism to exploit symmetry-dependent features for adversarial robustness in future quantum machine learning models.

We introduce and experimentally demonstrate an approach for selective mode excitation in non-Hermitian resonant systems using incoherent light. This method eliminates the need for precise phase control that is often required in coherent excitation schemes. Using this technique on a silicon photonic platform with coupled ring resonators, we successfully excite the topological edge state of a non-Hermitian Su-Schrieffer-Heeger (SSH) model. Our work shows that incoherence-assisted excitation is a robust and passive strategy for topological state preparation, which broadens the scope of non-Hermitian topological photonics thereby providing a practical and experimentally viable tool for selective mode excitation.

Quantum systems may contain underlying correlations which are inaccessible to computationally bounded observers. We capture this distinction through a framework that analyses bipartite states only using efficiently implementable quantum channels. This leads to a complexity-constrained max-divergence and a corresponding computational min-entropy. The latter quantity recovers the standard operational meaning of the conditional min-entropy: in the fully quantum case, it quantifies the largest overlap with a maximally entangled state attainable via efficient operations on the conditional subsystem. For classical-quantum states, it further reduces to the optimal guessing probability of a computationally bounded observer with access to side information. Lastly, in the absence of side information, the computational min-entropy simplifies to a computational notion of the operator norm. We then establish strong separations between the information-theoretic and complexity-constrained notions of min-entropy. For pure states, there exist highly entangled families of states with extremal min-entropy whose efficiently accessible entanglement in terms of computational min-entropy is exponentially suppressed. For mixed states, the separation is even sharper: the information-theoretic conditional min-entropy can be highly negative while the complexity-constrained quantity remains nearly maximal. Overall, our results demonstrate that computational constraints can fundamentally limit the quantum correlations that are observable in practice.

We study the Hamiltonian formulation of a gravity model obtained from a Yang--Mills theory for a one-parameter family of (A)dS Lie algebras parametrized by $α$, when the family of algebras is contracted to the Poincaré algebra in the limit $α\to 0$. We derive the canonical structure and first-class constraints and analyze the resulting algebra in the contraction limit. In this limit, the constraints generate the residual Lorentz gauge invariance, and the components of the AdS potential transform as tetrads and Lorentz connection. Finally, we determine the number of physical degrees of freedom, showing that in the non-propagating torsion sector - selected by a Lorentz-covariant gauge condition preserved under dynamical evolution - the theory exhibits only two propagating degrees of freedom.

When a quantum system couples to a scale-invariant environment, what form must its decoherence take? We prove that the answer is unique: under locality, Lorentz invariance, unitarity, and continuous scale invariance, the effect of any such environment is mathematically equivalent to that of an \emph{unparticle bath} -- a scale-invariant continuum of states -- characterized entirely by the scaling dimension $d_{\mathcal{U}}$ of the coupled operator. This is not a modelling choice but a consequence of conformal symmetry. All decoherence and dissipation exponents are fixed by $d_{\mathcal{U}}$ through exact consistency relations, providing falsifiable predictions independent of microscopic details. We validate the framework using multi-channel transport data from the unitary Fermi gas, where two genuinely independent observables yield a consistent $d_{\mathcal{U}} = 7/4$. We further show that quantum Ising criticality, inflationary cosmology, and high-energy astrophysical neutrinos -- spanning more than 25 orders of magnitude in energy -- are unified as specific realizations of the same structure. A decoherence phase transition at $d_{\mathcal{U}} = 5/2$, where quantum coherence is \emph{protected} rather thandestroyed at long times, is a qualitative prediction inaccessible to any memoryless dynamical description.

A key challenge in classical machine learning is to mitigate overparameterization by selecting sparse solutions. We translate this concept to the quantum domain, introducing quantum sparsity as a principle based on minimizing quantum information shared across multiple parties. This allows us to address fundamental issues in quantum data processing and convergence issues such as the barren plateau problem in Variational Quantum Algorithm (VQA). We propose a practical implementation of this principle using the topological Entanglement Entropy (TEE) as a cost function regularizer. A non-negative TEE is associated with states with a sparse structure in a suitable basis, while a negative TEE signals untrainable chaos. The regularizer, therefore, guides the optimization along the critical edge of chaos that separates these regimes. We link the TEE to structural complexity by analyzing quantum states encoding functions of tunable smoothness, deriving a quantum Nyquist-Shannon sampling theorem that bounds the resource requirements and error propagation in VQA. Numerically, our TEE regularizer demonstrates significantly improved convergence and precision for complex data encoding and ground-state search tasks. This work establishes quantum sparsity as a design principle for robust and efficient VQAs.

We introduce the parity-unfolded architecture, a fault-tolerant quantum computing scheme that relies on direct preparation and teleportation of small-angle rotations $ Z^{1/2^{k}}$ rather than approximating them with the conventional (Clifford + $T$) gate set. The architecture is enabled by efficient distillation of gates from an arbitrary level of the Clifford hierarchy, which we refer to as parity unfolding. With it, a state $|Z_k\rangle = Z^{1/2^{k}}|{+}\rangle$ can be prepared fault-tolerantly using $2^{k+3} + O(2^{k/2})$ biased-noise qubits on a planar chip with nearest-neighbour connectivity. For algorithms requiring native $Z^{1/2^{k}}$ gates, such as the Quantum Fourier Transform and phase estimation, the proposed scheme allows to reduce resource overheads for up to $k=7$, i.e., up to $T^{1/32}$. Furthermore, when used for the synthesis of arbitrary small-angle rotations, parity-unfolded distillation of ($T$ + $\sqrt{T}$) reduces the minimum achievable logical error rate by 43% while cutting the resource requirements by 26%, when compared to unfolded distillation of only the $T$ gate.

Quantum search is among the most important algorithms in quantum computing. At its core is quantum amplitude amplification, a technique that achieves a quadratic speedup over classical search by combining two global reflections: the oracle, which marks the target, and the diffusion operator, which reflects about the initial state. We show that this speedup can be preserved when the oracle is the only global operator, with all other operations acting locally on non-overlapping partitions of the search register. We present a recursive construction that, when the initial and target states both decompose as tensor products over these chosen partitions, admits an exact closed-form solution for the algorithm's dynamics. This is enabled by an intriguing degeneracy in the principal angles between successive reflections, which collapse to just two distinct values governed by a single recursively defined angle. Applied to unstructured search, a problem that naturally satisfies the tensor decomposition, the approach retains the $O(\sqrt{N})$ oracle complexity of Grover search when each partition contains at least $\log_2(\log_2 N)$ qubits. On an 18-qubit search problem, partitioning into two stages reduces the non-oracle circuit depth by as much as 51%-96% relative to Grover, requiring up to 9% additional oracle calls. For larger problem sizes this oracle overhead rapidly diminishes, and valuable depth reductions persist when the oracle circuit is substantially deeper than the diffusion operator. More broadly, these results show that a global diffusion operator is not necessary to achieve the quadratic speedup in quantum search, offering a new perspective on this foundational algorithm. Moreover, the scalar reduction at the heart of our analysis inspires and motivates new directions and innovations in quantum algorithm design and evaluation.

Modified gravity theories constitute viable alternatives to the standard cosmological model for explaining the observed late-time accelerated expansion of the Universe. The Effective Field Theory of Dark Energy (EFTofDE) is an efficient framework to describe a wide range of such theories with a limited number of parameters. To robustly constrain them by comparison with clustering and weak lensing data from upcoming large-scale structure surveys, high-resolution cosmological N-body simulations are required to obtain accurate predictions for the matter distribution on non-linear scales. We introduce two new N-body simulation codes for EFTofDE cosmologies: PySCo-EFT, a Python-based particle mesh code, and ECOSMOG-EFT, a RAMSES-based code with adaptive mesh refinement. We consider Horndeski models with a luminal gravitational wave speed. We use iterative solvers and multigrid schemes to solve for the additional scalar field equation in both codes, incorporating the non-linear Vainshtein screening mechanism. We present validation and convergence tests of the codes. We obtain a sub-0.5 percent agreement with linear theory on large scales and a similar agreement between the two codes on non-linear scales. The dominant numerical effects on the matter-power-spectrum boost are mass resolution, finite-volume effects, refinement threshold, and starting redshift, but they are limited to below 2% at the largest wavenumbers (k=10 h/Mpc) for the range of tested values. We investigate the impact of the EFTofDE parameters on the matter-power-spectrum ratios between EFTofDE and $Λ$CDM cases. Depending on the EFTofDE parameters, the screening plays a negligible or dominant role compared to the linearised field equations. Our codes provide tools for generating fast and accurate predictions of the impact of the EFTofDE on the clustering of matter, incorporating non-linear screening.

Quantum algorithms on near-term quantum processors are typically executed using shallow quantum circuits composed of one- and two-qubit gates. However, as circuit depth and gate number increase, gate imperfections and qubit decoherence begin to dominate, limiting algorithmic complexity. An alternative approach is to explore gates involving more than two qubits. In previous work (X. Wu et al., Physical Review X 14, 041030 (2024)), we demonstrated a new superconducting qubit architecture with user-selectable two-qubit interactions via a reconfigurable router, used to connect pairs of qubits. Here, we leverage this novel architecture to realize programmable and efficient multi-qubit operations involving more than two qubits, resulting in faster preparation of multi-qubit entangled states with good fidelities. We also successfully apply model-free reinforcement learning to perform multi-qubit gates, including training a two-qubit controlled-Z gate as well as three-qubit controlled-SWAP and controlled-controlled-phase (Fredkin and Toffoli) gates. Higher $n$th-order gates may also be feasible, using our high-connectivity router design. This could provide a more efficient and higher-fidelity implementation of complex quantum algorithms and a more practical approach to quantum computation.

We present numerically-informed closed-form expressions for the dominant $(\ell,m)=(2,2)$ waveform harmonic of the post-merger emission from mergers of non-spinning binary black holes with comparable masses on highly eccentric orbits. Using 233 non-spinning eccentric simulations from the RIT catalog, we construct time-dependent complex quasinormal mode amplitudes via a Bayesian procedure. We build multivariate polynomial models, represented as functions of the symmetric mass ratio $ν$ and two dynamics parameters evaluated at merger: the mass-rescaled effective energy $\hat{E}_{\mathrm{eff}}^{\mathrm{mrg}}$ and angular momentum $j_{\mathrm{mrg}}$. We further validate the post-merger non-circular waveform model by comparing it against simulations from the SXS catalog. Our models achieve mismatches around $\sim10^{-3}$, including for near-extreme eccentricities. The model can be directly combined with effective-one-body and phenomenological inspiral waveforms to produce accurate inspiral-merger-ringdown waveforms, essential for parameter estimation of both astrophysical and fundamental physics properties of the signals' sources.

We present the first three-dimensional general relativistic magnetohydrodynamic simulation of sustained accretion onto a horizonless singularity in which matter reaches the central object rather than being accumulated outside of it or expelled in outflows. We consider a Joshi-Malafarina-Narayan (JMN-1) spacetime, a well-motivated black hole mimicker that arises from gravitational collapse with anisotropic pressure in general relativity, and adopt a compactness parameter for which the central singularity is null. We find that the system evolves into a sustained magnetically arrested disk state. For parameters appropriate to the low-luminosity active galactic nucleus M87*, we obtain an accretion rate of $\sim(3.0 \pm 0.5)\times 10^{-6} \dot{M}_{\rm Edd}$, in full agreement with estimates based on black hole models and, in particular, comparable to that of our reference Schwarzschild black hole simulation. Synthetic ray-traced images at $230\,{\rm GHz}$, computed using polarized general relativistic radiative transfer, are broadly consistent with the Event Horizon Telescope observations of M87*. We identify a key observational discriminant between a black hole and JMN-1: the presence of detectable brightness inside of the ``observable" shadow of JMN-1. This emission originates very close to the central singularity, in a region that would be hidden behind the event horizon in a black hole spacetime. Although this signature is beyond the reach of current observations, it falls within the projected imaging dynamic range of next-generation radio interferometric instruments, offering a robust test of the black hole paradigm.

In the recent quantum echoes experiment, Google Quantum AI showed that out-of-time-order correlators (OTOCs) for random-circuit time evolution can be measured using a quantum processor more than 10,000x faster than they can be computed to similar accuracy via classical computation. This claim was substantiated by comparison with a variety of state-of-the-art classical simulation methods. One classical simulation method that was not explicitly tested was tensor networks with belief propagation (TNBP). TNBP should be poorly suited to simulating Google's echoes experiment: the states involved are highly entangled, a challenge for tensor network states; and the Willow chip has dense 2D connectivity, a challenge for belief propagation. Here we confirm, via a combination of theoretical scaling arguments and explicit numerical simulation, the intuition that TNBP is unable to simulate the quantum echoes experiment. We show that the OTOC circuits generate enough entanglement that they are largely incompressible, implying that other approaches in which OTOCs are computed by evolving a tensor network state in the Schrödinger picture will also fail. Our results further reinforce the claim that the quantum echoes experiment cannot be reproduced by classical computation.

The ultrastrong coupling regime of cavity photons and quantum materials has emerged as a pathway to modify materials properties, however definitive signatures of ultrastrong coupling remain elusive. Focusing on the quantum photon statistics of light transmitted through a cavity-embedded superconductor, we show that a two-photon Higgs polariton at strong coupling realizes a photonic nonlinearity at the single terahertz photon level. We find that as light-matter coupling increases, the photon statistics show pronounced changes due to the formation of a hybrid photon-matter dark-cavity state with finite photon occupancy, producing testable signatures of ultrastrong coupling. We derive a non-Markovian input output relation and study the cavity-embedded superconductor 2H-NbSe2 as it approaches ultrastrong light-matter coupling. Our results reveal a diagnostic for ultrastrong coupling in the two-photon coincidence statistics that is absent in total counts.

We study conformal quantum mechanics by first considering the perturbative $S$-matrix in various dimensions. The model has two couplings and we study perturbatively the degree of ultraviolet divergences arising in the interplay between the two couplings. We then focus on the inverse square potential in one spatial dimension and compute the beta function to arbitrarily perturbative and non-perturbative orders. This we do in both the bound state sector and scattering sector. We provide explicit, exact and infinite series results of the first few non-perturbative orders.

Stabilizer states admit compact classical descriptions, but many downstream tasks still require their full amplitude vectors. Since the output itself has size $2^n$, the main algorithmic question is whether one can materialize an $n$-qubit stabilizer state vector in optimal $O(2^n)$ time, rather than paying an additional polynomial overhead. We answer this question in the affirmative. Starting from the standard quadratic-form representation of stabilizer states, we give an algorithm that runs in $O(2^n)$ time and $O(2^n)$ space. The idea is to maintain a cached parity word that records all future off-diagonal quadratic phase increments simultaneously. As consequences, we obtain an optimal procedure for materializing a stabilizer state vector from a standard check-matrix description, and an optimal algorithm for expanding a Clifford tableau into its full dense matrix. These results close the asymptotic gap for dense stabilizer and Clifford materialization.

Mechanically isolated defect centers in hexagonal boron nitride are promising coherent quantum emitters, yet spectral instabilities persist, and their spin-related nature remains unclear. Here we investigate a single mechanically isolated quantum emitter in hBN integrated onto a coplanar waveguide. The emitter exhibits exceptionally bright resonant fluorescence with saturation count rates exceeding $10\,\mathrm{Mc/s}$. High-resolution spectroscopy reveals two closely spaced zero-phonon-line transitions originating from the same defect complex. Time-resolved spectroscopy shows that these transitions exhibit markedly different spectral diffusion dynamics, consistent with distinct donor-acceptor-pair-like recombination pathways with different sensitivities to local electrostatic fluctuations. Off-resonant blue illumination redistributes emission between the two transitions and increases the emission duty cycle without significantly modifying the dominant spectral diffusion rates at low temperature, indicating repumping from long-lived shelving states. Magnetic-field-dependent photoluminescence, optically detected magnetic resonance, and pump-probe measurements reveal millisecond-scale relaxation dynamics and magnetic-field-dependent fluorescence contrast, demonstrating spin-dependent population dynamics in the metastable shelving state. These results clarify how charge-driven spectral fluctuations and spin-dependent shelving jointly shape the optical cycling dynamics.

In this article, our goal is to investigate the cosmological dynamics and structure formation in a modified cosmological framework inspired by a generalized mass-to-horizon entropy relation and consistent with the Clausius relation. Invoking the gravity-thermodynamics conjecture leads to alterations in the Friedmann equations as well as Hubble parameter evolution. The effects of the generalized entropy on various cosmographic parameters and on the growth of the linear matter perturbations by constructing perturbed field equations via employing spherical collapse formalism in a flat Friedmann-Lemaître-Robertson-Walker background have been explored. We discuss a novel and well-known diagnostic approach to differentiate various cosmological models vis-à-vis flat and non-flat $Λ$CDM frameworks, and find that the generalized mass-to-horizon entropy-inspired modified cosmology ($n\ne 1$) successfully passes all the litmus tests by falsifying both the flat and non-flat $Λ$CDM paradigms. It is shown that this model also satisfies the requirements for the Universe to achieve thermodynamic equilibrium in the distant future. We also study the halo mass function and cluster number counts in this modified gravity scenario. All the results are compared with the fiducial $Λ$CDM profile, showing that the additional entropic correction influences the expansion history, the growth rate of structures, and the abundance of collapsed halos. We observe that the more massive collapsed structures are less abundant and form at later epochs, which is expected from the hierarchical model of large-scale structure formation.}

Sequential Quantum State Discrimination (SQSD) can be naturally framed as a sequential decision-making problem: at each time step, an agent must decide whether to perform an additional measurement to gather more information or to conclude with an optimal decision based on the current belief. In this paper, we formally cast SQSD into a static-hidden-state Partially Observable Markov Decision Process (POMDP) framework. We demonstrate that this formulation precisely subsumes the conventional minimum-error discrimination (MED) scheme as a special one-step case. Furthermore, we apply a regular grid-based discretization to the continuous belief simplex and approximate the possibly continuous measurement space using a finite library. Then we provide rigorous mathematical bounds on the resulting errors and analyze the computational complexity for both offline planning and online execution. Our analysis confirms that the inherent trade-off between accuracy and complexity, as well as the curse of dimensionality regarding the number of hypotheses, are also prominently observed in the quantum regime. Finally, we provide a working example of binary state discrimination to derive explicit forms of various functions and present numerical simulations for trine state discrimination to visualize the sequential structure of our POMDP-based SQSD.

We present an architecture-level hardware-to-logical-to-decoder execution stack for hybrid continuous-variable and discrete-variable quantum error correction in LiDMaS+. Provider-native records are normalized into a single decoder IO contract and replayed under fixed controls across MWPM, UF, BP, and neural-MWPM. In a Xanadu case study using fixture inputs and sampled public datasets, replay integrity was complete: 108/108 fixture and 4000/4000 real-slice request-response lines, with zero request-parse errors, zero response-parse errors, and zero decoder-name mismatches. Under matched inputs, decoder behavior is clearly regime-dependent. For weighted fixture summaries, average flip count was 1.296 (MWPM), 1.296 (UF), 0.667 (BP), and 1.296 (neural-MWPM). For weighted real-data summaries, average flip count was 0.641 (MWPM), 0.741 (UF), 0.318 (BP), and 0.641 (neural-MWPM); corresponding nonempty-flip rates were 0.490, 0.490, 0.318, and 0.490. Across fixture data, BP reduced weighted correction volume by 48.6\% versus MWPM; across real slices, BP reduced weighted correction volume by 50.4\% versus MWPM and 57.1\% versus UF. Quality controls show the central interpretability tradeoff: BP is intervention-conservative but leaves higher residual burden, while MWPM-family decoders intervene more aggressively and clear more syndrome. Warning-no-syndrome rates remained decoder-invariant and dataset-driven (fixture weighted 0.259; real weighted 0.510), confirming preserved sparsity semantics from hardware input to logical correction. Re-running analysis stages reproduced identical SHA-256 artifacts, enabling deterministic study iteration. These results establish a practical benchmarking foundation for photonic GKP-oriented hardware programs where decoder policy must be selected as a function of operating regime.

Cavity quantum electrodynamics (QED) with quantum emitters coupled to resonators provides a powerful platform for engineering light-matter interactions and exploring collective phenomena. In particular, superradiance, arising from collective quantum interference among emitters, has been explored as a route to ultrastable continuous radiation. However, engineering superradiance in the time domain to realize periodic pulsed sources or frequency combs remains largely unexplored. Here, we investigate the non-equilibrium many-body dynamics of a driven spin ensemble coupled to an on-chip superconducting resonator and uncover a dynamical phase transition from continuous-wave to periodic pulsed superradiant emission. To quantitatively capture the observed dynamical phases, we introduce a driven-dissipative cavity-QED model that elucidates how the periodic pulsed superradiant phase emerges from collective, periodically repeating spin dynamics stabilized by the interplay of coherence growth, disorder, and dissipation. We also find that rare-earth ion spin systems exhibiting both optical and microwave transitions enable phase-synchronized, dual-rail superradiant frequency combs in both the microwave and optical domains. Our results not only open new avenues for dual-rail frequency-comb applications in quantum metrology and information processing, but also establish a fundamental connection between periodic pulsed superradiance and the emergence of a continuous time crystal as a novel nonequilibrium phase in driven open systems.

Predicting heat-related physiological events at the population level is challenging due to the complex interactions among climatic, demographic, and socioeconomic factors, as well as the strong sparsity and seasonality of observational data. In this work, we propose a unified predictive framework that integrates heterogeneous environmental and public-health datasets and evaluates two learning paradigms within a common pipeline: classical machine learning and quantum machine learning. The methodology combines data harmonization, temporal aggregation, feature engineering, and dimensionality reduction to construct a weekly county-level population dataset. On this unified representation, we train both a classical regression baseline and a variational quantum model based on parameterized quantum circuits with angle embedding and data re-uploading. Experimental evaluation on datasets from the United States and Catalonia shows that classical models currently achieve higher predictive accuracy, particularly under conditions of strong class imbalance and sparse targets. Nevertheless, the quantum models demonstrate non-trivial learning capability and capture meaningful predictive structure in several scenarios. These results provide an empirical comparison between classical and quantum learning approaches for population-level physiological prediction and establish a methodological foundation for future hybrid health modeling as quantum hardware continues to evolve.

Hydration status is a key physiological indicator associated with cellular homeostasis, renal function, and overall health. Recent advances in smart sensing environments enable passive monitoring of urinary biomarkers that can provide continuous insight into hydration dynamics. In this work, we investigate predictive modeling approaches for hydration monitoring using biomarker data collected through the Predict Health Toilet (PHT) system. The problem is formulated as a regression task using urinary indicators such as urine specific gravity, conductivity, and volume. We evaluate classical machine learning models and quantum machine learning architectures based on variational quantum circuits. In particular, we introduce a modular Quantum Sequential Model (QSM) designed to construct flexible hybrid quantum classical predictive pipelines. Experimental results compare classical regression models, symmetry-constrained quantum regressors, and QSM architectures. The results provide insights into the potential role of quantum machine learning in digital health monitoring systems and highlight the opportunities and current limitations of near-term quantum computing for physiological data analysis.

In the literature, there are two differing definitions of contextuality: Kochen and Specker's, and Spekkens' (or ``generalised''). However, researchers using one of these definitions rarely consider the other, meaning comparative analysis of these two notions is rare. In this paper, we advance the idea that Kochen-Specker contextuality provides a generalisation of the idea of system being fundamentally nonclassical, while Spekkens' noncontextuality provides a generalisation of the idea of a system being classical. This allows us to reconcile the two approaches, as different stages in a hierarchy of classicality/nonclassicality.

We study the impact of radiation on quantum systems defined by the spins of elementary fermion-antifermion pairs produced at colliders. We present predictions for several processes, showing that energetic final-state radiation can induce decoherence and significantly reduce the entanglement of quantum systems formed by elementary fermion pairs. We investigate the feasibility of observing this effect experimentally in exclusive samples with energetic radiation. A statistically significant signal can be obtained with current data in associated $pp \rightarrow t\bar{t}(g)$ production at the LHC and in $e^+e^- \rightarrow τ^{+}τ^{-}(γ)$ production at Belle 2. Future electron-positron colliders operated at the $Z$ pole or well above the $t\bar{t}$ production threshold will extend these prospects further.

We present a first study of the scattering process $e^+ e^-\toπ^+π^-γ$ beyond next-to-leading order, aimed at providing preliminary insights required for future NNLO predictions for radiative return processes. A complete four-dimensional tensor decomposition of the amplitude is developed, and the associated one-loop polarised amplitudes are evaluated analytically to higher orders in the dimensional regulator, as required for NNLO accuracy. The calculation is complemented by an efficient numerical framework for the evaluation of the resulting five-point Feynman integrals, enabling stable and fast evaluations across the physical production region with evaluation times of a few hundreds of milliseconds. These results are suitable for implementation in Monte Carlo event generators.

We present results from two test beam campaigns that investigate the performance of straw tube detectors as potential candidates for an FCC-ee straw tracker. These studies were carried out at CERN using 150 GeV muon beams. Dedicated algorithms were developed to determine both single tube spatial resolution for the primary coordinate in the $r-φ$ plane and spatial resolution for the secondary coordinate along the tube direction within a straw chamber. Detection efficiency was also evaluated as a function of the extrapolated hit position for each tube. Both datasets showed consistent results for spatial resolutions and efficiency. Our findings will help establish benchmark performance metrics and provide valuable insight for future design, optimization, and construction of straw chambers for high-precision tracking applications.

We present the latest physics results from the FASER experiment at the LHC. Using $pp$ collision data at $\sqrt{s} = 13.6$ TeV collected during LHC Run 3, FASER reports new results on four fronts: a search for dark photons with an improved analysis strategy using $177~fb^{-1}$ of data, yielding world-leading exclusion limits; neutrino cross section measurements and the first search for charm hadron production in neutrino interactions, both using the FASER$ν$ emulsion detector with a 681 kg tungsten target and $9.5~fb^{-1}$ of 2022 data; the first observation of $ν_e$ in the FASER electronic detector at $5.5σ$ using $176.8~fb^{-1}$ of data; and the first double-differential measurement of $ν_μ$ interactions as a function of energy and rapidity with $186~fb^{-1}$ of data.

We construct cubic scalar field theory on $λ$-Minkowski space by combining the Batalin-Vilkovisky formalism with harmonic analysis, and produce two inequivalent noncommutative quantum field theories. The braided theory is based on a braided $L_\infty$-algebra whereby covariance dictates a spectral decomposition into cylindrical Bessel functions that diagonalise the angular Drinfel'd twist; in this theory we find the usual logarithmic ultraviolet divergences and confirm the absence of UV/IR mixing. The standard noncommutative theory is based on a classical $L_\infty$-algebra; in this theory we relate the spectral decompositions into plane wave and cylindrical harmonic eigenmodes of the Klein-Gordan operator, we verify the planar equivalence theorem, and we demonstrate a periodic form of UV/IR mixing in which non-planar correlators are generically ultraviolet finite but become non-analytic on an infinite lattice of exceptional momenta.

The pulsar beaming fraction is a fundamental quantity for connecting the observed pulsar population to the intrinsic Galactic population and for constraining pulsar emission geometry. In this study, we estimate the beaming fraction in each observational band (radio, $γ$-ray, and X-ray) and for each TeV survey (H.E.S.S., HAWC, and LHAASO) using TeV-selected pulsar wind nebulae (PWNe) and unidentified (Unid) TeV sources, assuming that the TeV emission from PWNe is approximately isotropic and that Unid sources are PWNe powered by pulsars whose beams do not intersect our line of sight. Within each survey, the inferred beaming fractions $\sim 0.1-0.3$ are comparable across bands. In contrast, the values differ by more than a factor of two between H.E.S.S. and HAWC/LHAASO. This discrepancy likely reflects survey-dependent selection effects, including differences in angular resolution and energy range, and is also consistent with the possibility that HAWC/LHAASO selected samples preferentially include older pulsars associated with more extended PWNe than those in the H.E.S.S. sample. We further show that the inferred beaming fractions can be reproduced within a unified framework using a time-dependent opening angle, and that this framework remains compatible with the statistical properties of the observed pulsar population.

At long-baseline neutrino experiments, neutral-current (NC) events accumulate in large numbers but are seldom exploited for new physics searches. We demonstrate their potential using non-standard neutrino interactions (NSI) with quarks as a case study. Charged-current (CC) analyses constrain NSI through matter effects on neutrino propagation, which probe almost exclusively the isoscalar combination of up- and down-quark couplings; the orthogonal isovector combination is suppressed by a factor of $\sim$100. Because NSI also modify NC cross sections in a flavor-dependent way, NC events become sensitive to oscillations: the far-to-near detector ratio acquires a dependence on the beam's flavor composition that probes both isoscalar and isovector couplings with comparable weight. Using existing NOvA data and DUNE projections, we derive the first bounded constraints on isovector NSI from a long-baseline experiment and show that combining CC and NC measurements resolves the individual quark couplings, breaking a degeneracy that persists in either analysis alone.

Atmospheric neutrinos provide a unique avenue to probe theories beyond the Standard Model (BSM) over a wide range of energies and path lengths. The theory of non-standard interactions (NSI) of neutrinos is one of the important BSM scenarios, which can modify flavor oscillations of atmospheric neutrinos traveling through the Earth. In this work, we use a high-purity $ν_μ$ charged-current (CC) sample of atmospheric neutrinos from IceCube DeepCore with a livetime of 7.5 years to search for the NSI parameters $\varepsilon_{eμ}$, $\varepsilon_{eτ}$, and $\varepsilon_{ee}-\varepsilon_{μμ}$. The $ν_μ$ CC events mainly come from the $ν_μ$ survival channel having no significant dependence on $δ_{\rm CP}$. Therefore, the constraints on $\varepsilon_{eμ}$ and $\varepsilon_{eτ}$ obtained using this $ν_μ$ CC sample are expected to be free from the $δ_{\rm CP}$-degeneracy. The data sample is found to be in agreement with the standard neutrino interactions. Therefore, we place bounds on these NSI parameters that are consistent with and comparable to existing experimental constraints. These $δ_{\rm CP}$-free constraints from IceCube DeepCore are complementary to those from the long-baseline neutrino oscillation experiments, where the appearance channel depends on $δ_{\rm CP}$.

We propose a non-holomorphic modular $A_4$ model under the assumption of universal couplings. In this framework, a charged lepton mass hierarchy is not created through parameter fine tuning or hierarchical Yukawa couplings, but instead is determined by the modulus $τ$, with certain modular weight assignments of right handed charged leptons. The experimental charged lepton masses are reproduced with high precision for values of $τ$ located near modular fixed points. In neutrino sector the couplings are imposed to be equal in magnitude with different relative phases. By fixing the modulus $τ$ from charged lepton sector, we perform a comprehensive scan over the phase parameters and modular weight assignments of the right handed neutrinos. We find that viable solutions arise only for normal neutrino mass ordering and a unique right handed neutrino modular weight, $k_N=-1$. The model yields strong correlations among mixing angles, the effective neutrinoless double beta decay parameter $m_{ee}$, and the total neutrino mass $\sum m_i$. These results underscore the predictive quality of non-holomorphic modular symmetry with minimal parameter inputs and offer implications for neutrino experiments and cosmological observations that can be tested.

We revisit the constrained log-subtracted two-point Padé (LSTP) ensemble for thermal $\mathcal{N}=4$ supersymmetric Yang--Mills (SYM) thermodynamics in four spacetime dimensions after upgrading the weak-coupling truncation from $\mathcal{O}(λ^2)$ to the exact $\mathcal{O}(λ^{5/2})$ coefficient. We keep the interpolation ansatz unchanged and shift the weak-side matching points to the regime where the new term is numerically significant. The admissible set collapses from $9$ nominal survivors ($3$ distinct curves) to a single distinct curve, the crossover range shrinks to a unique value, and the pointwise band width drops to zero within numerical resolution. The Hermite-Padé (HP) central curve does not coincide with the unique LSTP survivor, so the exact weak-coupling coefficient removes the LSTP scan uncertainty but not the difference between the two routes. The next step is to compute the unknown $\mathcal{O}(λ^{-3})$ strong-coupling coefficient.

The relic abundance of Dark Matter (DM) produced via thermal freeze-in is sensitive to the thermal history during and after cosmic reheating. In minimal models, this opens up the possibility to make predictions for collider observables by combining the requirement to match the DM relic abundance with observations of the Cosmic Microwave Background (CMB). We assess the impact of thermal corrections to the rate of cosmic reheating and the rate of thermal DM production on CMB observables and the relic abundance. We find that such corrections are generally small in the regime where they can be computed by means of finite-temperature field theory. We construct counter-examples where this general rule is violated.

We study projective one-parameter conifold degenerations whose central fiber has finitely many ordinary double points. Existing finite-node theory isolates one rank-one local sector per node on the perverse-sheaf, mixed-Hodge-module, and categorical sides, but does not determine which global extension classes are actually realized by geometry. We show that when the nodes are linked by common cycle geometry or homological relations, the corrected extension is not free nodewise data, but is forced into a smaller relation-controlled subspace. To formalize this, we introduce a cycle-node incidence datum and the associated geometrically realized subspace of the ambient nodewise extension space. Under geometrically admissible and block-adapted incidence hypotheses, we prove that the corrected perverse extension factors through this subspace, with incidence compatibility derived from propagation of local variation data along admissible cycle components, and we show that the same relation law lifts compatibly to the mixed-Hodge-module setting. We then compare this relation law with the resolution and smoothing sides and, in the block-separated cycle family, obtain $R_{\mathrm{res}}=R_{\mathrm{sm}}=R_{\mathrm{ext}}=R_{\mathrm{blk}}$.

Podolsky electrodynamics, a higher-derivative extension of Maxwell's theory characterized by the Podolsky parameter $λ=1/m$, which modifies the photon dispersion relation and regularizes short-distance divergences, is investigated. This framework is then coupled to the Carroll-Field-Jackiw (CFJ) model, in which a Lorentz-violating background four-vector is introduced. Within this extended electrodynamics, the photon propagator is obtained in the combined Podolsky-CFJ framework and subsequently applied to Möller scattering. It is shown that the CFJ contribution can reintroduce the short-distance divergence suppressed by Podolsky's term. In the nonrelativistic limit, both the spatial component--which introduces a preferred direction in space and thus breaks isotropy--and the timelike component--which directly affects the dispersion relation--contribute nontrivially to the interaction potential.

The strong forces between nucleons ($N$=$p$, $n$) are fundamental to the visible universe. The interactions between hyperons (baryons with strange quarks) and nucleons are essential for the intrinsic properties of neutron stars. Whereas the interactions between nucleons ($pp$, $pn$, $nn$) have been extensively studied, the interactions between nucleons and hyperons ($NΛ$, $NΣ$, $NΞ$, $NΩ$, ...) are not well understood, due to the small amount of relevant data, limited by the scarcity of suitable hyperon sources. Here we point out and investigate a new high-quality source: hyperons produced in $pp$ collisions, such as $pp\to pK^+Λ$, $pKΣ$, $pKπΣ$, $pKKΞ$, $pKKKΩ$. At a fixed target experiment using proton beam with known momentum and liquid hydrogen target, $pp\to pK^+Λ$ can be produced copiously. By tagging $p$ and $K^+$, the flux and momentum of the $Λ$ can be determined precisely. By placing an additional target around the primary one, these $Λ$-s serve as an ideal source, enabling an unprecedentedly precise study of $Λ$ interactions with a wide range of targets. Similar methods can be used to obtain high-quality sources of other hyperons, such as $Σ$, $Ξ$ and $Ω$. These novel, high-statistics sources of hyperons with precisely known kinematics present new opportunities for applications in particle and nuclear physics, particularly in understanding the hyperon puzzle of neutron stars. We propose a new high-luminosity experiment with two nested concentric targets, optimized for such measurements. This concept can also be incorporated into existing experiments, such as HADES and CBM at FAIR, as well as proposed experiments, such as H-NS and HHaS at HIAF, by adding a second target without significant modification of the current detectors.

We present a comprehensive study of $Υ(nS)$ ($n$ = 1, 2, 3) production in proton-proton ($pp$) collisions at various LHC energies and rapidity ranges within the framework of leading order non-relativistic quantum chromodynamics (NRQCD) factorization. The transverse momentum ($p_{\rm T}$)-dependent production cross-sections are calculated, incorporating both direct and feed-down contributions. Specifically, feed-down from $Υ(2S)$, $Υ(3S)$, $χ_{bJ}(1P)$, and $χ_{bJ}(2P)$ states to $Υ(1S)$ is included, while $Υ(2S)$ receives contributions from $Υ(3S)$ and $χ_{bJ}(2P)$. No significant feed-down is considered for $Υ(3S)$. The computed cross-sections and cross-section ratios among different $Υ$ states are compared with experimental measurements from ALICE, ATLAS, CMS and LHCb. It is found that the experimental cross-sections and cross-section ratios are well described within the theoretical uncertainties arising due to the choices of the factorization and renormalization scales for $p_{\rm T}$ $>$ 4 GeV and $p_{\rm T}$ $>$ 0 GeV, respectively. Furthermore, the cross-section ratios exhibit a clear saturation behavior beyond $p_{\rm T}$ $\approx$ 40 GeV.

This report summarizes the presentations and discussions during the ISO-BREAK 25 Workshop ``Isospin symmetry violation: kaons and beyond'', which was held at Jan Kochanowski University in Kielce on October 23-25, 2025. We address the current status of the isospin-symmetry breaking discovered by NA61/SHINE in nucleus-nucleus collisions at the CERN SPS, its confirmation by other experiments and studies in \ee and deep inelastic scattering. In addition, we discuss the theoretical status as well as we outline experimental and theoretical priorities towards understanding this currently unexplained phenomenon.

Neutrino masses provide one of the clearest indications of physics beyond the Standard Model. In this brief review, I discuss the main theoretical frameworks developed to account for them, with particular emphasis on scenarios in which neutrinos are Majorana particles. After a short overview of the current landscape of neutrino mass models, I focus on constructions featuring the spontaneous breaking of global lepton number and examine the phenomenological implications of a massless Goldstone boson, the majoron.

Recently, self-dualities based on saddle-point expansions have been proposed as a means to obtain qualitative non-perturbative information in scalar field theories. In this work, we test this proposition quantitatively by studying the phase transition for critical scalar $φ^4$ field theory in 1+1 dimensions using a variational method. We find that saddle-point methods obtain quantitative agreement for the free energy, but differ on the order of 25 percent for the peak location of the correlation length.

We study a boost-invariant, out-of-equilibrium fluid of non-interacting Dirac fermions with a finite canonical spin potential. After solving the Dirac equation in Milne coordinates, we exactly diagonalize the non-equilibrium density operator and compute the partition function and expectation values of relevant observables, including spin polarization, energy density, longitudinal and transverse pressures, spin density, and \emph{spin torque}, i.e. the source of spin non-conservation. We find an analytic expression for the partition function at finite spin potential, and show numerically that thermodynamic relations connecting it to thermodynamic functions hold in the system under consideration. We show that, in a boost-invariant system, both shear-induced polarization and the spin Hall effect are absent, and that a non-vanishing polarization can only arise from a finite spin potential in a free theory. We obtain an analytic expression for the spin polarization as a function of the spin potential in some particular cases, and otherwise compute numerically its exact expectation value at finite spin potential. Our results are discussed in the context of relativistic spin hydrodynamics and quark--gluon plasma phenomenology.

We investigate the properties of neutral and charged mesons in magnetic fields, from weak-field to strong-field regimes. To develop analytic insights, we employ a non-relativistic quark model with a confining potential of the harmonic oscillator type. Short-range correlations, such as Coulomb and color-magnetic interactions, are treated as perturbations. In particular, we focus on the magnetic field dependence of the relative and the center-of-mass motions. The qualitative trends differ significantly between neutral and charged mesons: for neutral mesons, the transverse momenta are conserved and continuous, while charged mesons exhibit quantized transverse dynamics. The Zeeman effects, arising from intrinsic spins and orbital angular momenta, are carefully examined. In particular, for charged mesons with spins $s\ge 1$, we discuss how the zero-point energy in the internal quark motion cancels the Zeeman energy from the orbital angular momentum, ensuring the energetic stability of mesons with high spins. The effectively reduced dimensionality of these mesons in the strong-field limit is also discussed.

We investigate rare special dileptonic decays of $ Λ_b$, $Σ_b$ and $Ξ_b $ baryons in the Standard Model and context of the general Two-Higgs-Doublet Model with Type III. Specifically, we consider the decays $ Λ_b \rightarrow Λ\ell^+ \ell^-$, $Σ_b \rightarrow Σ\ell^+ \ell^-$ and $Ξ_b \rightarrow Ξ\ell^+ \ell^-$, where $\ell$ represents $μ$ or $τ$ lepton. By studying these rare decays, we aim to assess the impact of the Two-Higgs-Doublet Model with Type III on various observables, such as the differential branching ratio, total branching ratio, and lepton forward-backward asymmetries using the decay amplitude and the transition matrix elements in terms of form factors calculated via light cone QCD in full theory. We compare our results to those of the Standard Model, as well as existing lattice QCD predictions and experimental data, to assess the agreement and viability of the Two-Higgs-Doublet Model with Type III. Furthermore, we highlight the potential for experimental investigations of these decay channels in the near future. The soon-to-be updated LHCb and/or Belle II detectors, renowned for their capabilities in studying rare decays, present excellent opportunities for probing the predicted branching ratios.

The nature of the $X(3872)$ and other exotic hadrons has been a subject of extensive investigation. While various theoretical models have been proposed, experimental evidence suggests that the $X(3872)$ may be a mixture state of a hadronic molecule and a $c\bar{c}$ core. In this work, we perform a systematic study of hidden-charm tetraquark candidates $X(3860)$, $X(3872)$, and $Z(3930)$ using a coupled-channel model that incorporates both $c\bar{c}$ states and $D^{(*)}\bar{D}^{(*)}$ hadronic molecular components. The model parameters are fixed to reproduce the masses of the $X(3872)$ and $Z(3930)$, and the resulting framework is used to predict the mass and structure of the $0^{++}$ state associated with the $X(3860)$. Our results support the mixture interpretation of these exotic hadrons, exhibiting strong attractions from the transition potential between $c\bar{c}$ and $D^{(*)}\bar{D}^{(*)}$ components. The molecular component dominates in the $X(3872)$, while the $c\bar{c}$ component plays a more prominent role in the $X(3860)$ and $Z(3930)$.

Jet quenching in heavy-ion collisions probes parton energy loss in the quark--gluon plasma (QGP), but the extracted transport properties may not be universally constrained across centrality, beam energy, and observable class. In this work, we perform an analysis of the compatibility and predictive transferability of Bayesian constraints obtained from a six-parameter JETSCAPE effective energy-loss model across these subsets. The model is calibrated to charged-hadron and inclusive-jet data from ALICE, ATLAS, and CMS in PbPb collisions at $\sqrt{s_{\mathrm{NN}}}=5.02$ and $2.76$ TeV. We find that centrality-dependent posteriors are largely compatible, whereas beam-energy and observable-class splits exhibit moderate shifts within overlapping credible regions, indicating that posterior overlap alone does not guarantee predictive universality. This is further examined by propagating subset posteriors to complementary datasets without refitting, where predictive performance varies across subsets. These results indicate that different observables probe distinct aspects of jet--medium interactions and motivate leading-hadron-selected jet observables to bridge hadron-biased and jet-inclusive constraints.

We investigate the stringy effects on holographic complexity in $(d+1)$-dimensional Gauss-Bonnet gravity using the ``complete volume'' proposal for higher-curvature theories. Our analysis covers unperturbed eternal black holes, as well as the one-sided and two-sided Vaidya spacetimes. The one-sided geometry describes a null shell collapsing into the empty AdS vacuum to form a black hole, while the two-sided geometry represents a null shell injected into an eternal black hole background with arbitrary energy. For unperturbed backgrounds, higher-curvature terms introduce explicit corrections to the standard CV proposal, giving rise to a ``competition effect'' absent in the uncorrected framework. In the dynamical settings, we demonstrate that despite novel jumps in the canonical velocities across the null shell, the complexity growth rate remains universally governed by the conserved momentum, just as in Einstein gravity. Furthermore, our two-sided shock wave analysis reveals that Gauss-Bonnet corrections prolong the critical time, preserving the universal logarithmic dependence for the scrambling time.

The charged current $ν_μ(\barν_μ)$-induced deep inelastic scattering (DIS) from an $^{40}\mathrm{Ar}$ target is studied using a microscopic framework that incorporates nuclear medium effects due to Fermi motion, binding energy, nucleon correlations, mesonic ($π$ and $ρ$) contributions, and nuclear shadowing and antishadowing across the relevant Bjorken-$x$ region. The nuclear structure functions $F_{iA}(x,Q^2)$ $(i=1\text{-}3)$ are evaluated using a relativistic nucleon spectral function ($S_h$) within the local density approximation employing the free nucleon structure functions, $F_{iN}(x,Q^2)$ $(i=1\text{-}3)$. These $F_{iN}(x,Q^2)$ $(i=1\text{-}3)$ are calculated using parton distribution functions (PDFs) from MMHT 2014 parameterization, including higher-order perturbative QCD corrections up to next-to-next-to-leading order (NNLO), along with nonperturbative target mass corrections (TMC). The resulting nuclear structure functions $F_{iA}(x,Q^2)$ $(i=1\text{-}3)$ are subsequently used to compute the differential DIS cross sections for $^{40}Ar$ nucleus. Numerical results are presented for $ν_μ(\barν_μ)$ beam energies $E=4$ GeV and $E=6$ GeV for the differential scattering cross sections $\frac{d^2σ}{dx dy}$ and $\frac{dσ}{dx}$, relevant to ongoing and upcoming liquid-argon neutrino experiments such as DUNE and the Fermilab Short-Baseline Neutrino program.

In this paper, we analyze the double Wick rotation of a rotating BTZ black hole and the entanglement entropy. We derive the transition matrix dual to the double Wick-rotated BTZ black hole, which has the usual shape at an imaginary chemical potential. In the dual gravity side, the double Wick rotated BTZ black hole, which is obtained as a quotient, is equal to a rotating BTZ black hole after the coordinate transformation and the identification of periodicity. The geometric entropy and time-like entanglement entropy are reproduced by the identification. New Lorentzian entanglement growth is defined by the coefficient of linear growth of time-like entanglement entropy.

sPHENIX is a state-of-the-art experiment at the Relativistic Heavy Ion Collider (RHIC), dedicated to the study of heavy-flavor and jet physics. Its precision tracking system, combined with streaming readout, enables heavy-flavor measurements with high-statistics and essentially unbiased data samples. During the 2024 run, sPHENIX was fully commissioned and recorded a sample of 100 billion unbiased $p$+$p$ collisions, together with a minimum-bias Au+Au dataset. The 2025 run further expanded the sPHENIX dataset with high-statistics $p$+$p$, O+O and Au+Au collisions. This extensive $p$+$p$ sample opens the door to heavy-flavor measurements with orders of magnitude more statistics than previously available at RHIC. Notably, there has been no prior measurement of the $Λ_c^+ / D^0$ baseline in $p$+$p$ collisions at RHIC energies. The large sPHENIX dataset now enables the first exploration of key open questions, such as the hadronization mechanism of baryons and the strange-to-light flavor meson ratio.

At finite $N$ the ring of gauge invariant operators is not freely generated. For problems of interest in physics, these rings are Cohen--Macaulay and admit a Hironaka decomposition, in which the full invariant ring is a free module over a polynomial ring generated by the primary invariants. The module basis is given by finitely many secondary invariants. This motivates a physical picture in which the primary invariants are regarded as perturbative degrees of freedom while the secondary invariants are associated with distinguished non-perturbative states or sectors. The purpose of this study is to show that a concrete algebraic version of this picture is visible in simple zero-dimensional matrix integrals.

We study the one-loop correction to the near-extremal quantum entropy of the charged two-dimensional black hole introduced in \cite{McGuigan:1992}. In target space this background can be understood as arising from the dimensional reduction of a three-dimensional solution of the low-energy string effective action. On the other hand, its worldsheet description is provided by the dimensionally reduced $\frac{SL(2, \mathbb{R}) \times U(1)}{U(1)}$ WZW model. Using the latter formulation, we extract the temperature dependence of the one-loop correction to the near-extremal quantum entropy from the corresponding torus partition function. Based on the nearly-$AdS_2$ structure of the near-extremal near-horizon geometry of this black hole, one might naively expect to recover the usual logarithmic correction associated with the universal Schwarzian sector. Remarkably, the final result deviates from this expectation. In the general case, the one-loop correction we obtain is exponentially suppressed in the low-temperature limit. However, upon fine-tuning the microscopic parameters of the theory, specifically the level of the $\mathfrak{sl}_k(2,\mathbb{R)}$ current algebra and the worldsheet coupling constant between $SL(2,\mathbb{R)}$ and the $U(1)$ boson, the correction scales as $\sqrtβ$. In this limit, the target-space partition function is divergent in the extremal regime. We argue that this result provides a worldsheet realization of the black hole/string transition.

Pohl et al. measured the energy difference between the 2P and 2S states of muonic hydrogen and used it to determine a precise value of the proton radius. The result disagreed significantly from values extracted from electronic hydrogen and elastic electron-proton scattering. This discrepancy was exciting because it indicated a breakdown of Coulomb's law. In more technical terms, the discrepancy indicated that a fundamental property of the Standard Model, known as lepton universality, could be violated. This chapter explains the origins, meaning and significance of the puzzle. A resolution, based on very recent experiments, is stated. The proton radius puzzle is no more.

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

We study inelastic dark matter (iDM) in a minimal and ultraviolet-complete framework in which the dark photon mass arises from a dark Higgs mechanism. The spontaneous breaking of a $U(1)_D$ symmetry splits the Dirac fermion into two Majorana states, thus opening new annihilation channels mediated by the dark Higgs. Focusing on sub-GeV dark matter, we assess the sensitivity of DUNE's liquid argon cube at the near detector (ND-LAr) to this scenario. We find that the ND-LAr can probe regions of parameter space consistent with the observed relic abundance due to these new annihilation channels, particularly for large dark photon-to-DM mass ratios where decay-based searches lose sensitivity. Our results highlight the complementarity between cosmological constraints and fixed-target experiments and demonstrate the potential of DUNE's ND-LAr to explore iDM scenarios with extended dark sectors.

Rovibrational transitions in the hydrogen and antihydrogen molecular ions $H_2^+$ and $\overline{H}_2^-$ offer the possibility of testing Lorentz and CPT symmetry to extremely high precision, in principle attaining $O(10^{-17})$. In this paper, the third in a series, we give a comprehensive derivation of the rovibrational spectrum of $H_2^+$ and $\overline{H}_2^-$ in the SME, an effective quantum field theory incorporating Lorentz and CPT violation. New developments described here include a complete analysis of the molecular dynamics from first principles in terms of the spherical tensor representation of the SME couplings, the systematic extension of our previous results to the non-minimal SME, a full description of the quantum number dependence of the rovibrational energy levels in the spherical tensor formalism with both high and low background magnetic fields, and an extended discussion of sidereal and annual variations of transition frequencies arising from both rotations and Lorentz boosts. The resulting sensitivity of the rovibrational spectrum to an extended range of SME couplings, together with the ability to isolate their individual effects using the quantum number dependence of the transition frequencies, enhances the opportunities to detect Lorentz and CPT symmetry breaking through rovibrational spectroscopy of $H_2^+$ and $\overline{H}_2^-$.

We formulate a covariant version of Maxwell-like fracton electrodynamics in six dimensions using a symmetric tensor gauge field with scalar gauge symmetry $δA_{μν}=\partial_μ\partial_νΛ$. This provides a relativistic setting in which the characteristic fractonic restriction on mobility follows directly from gauge invariance and the allowed coupling to matter. We construct the stress--energy tensor and show that its trace has a universal dimension-dependent structure that becomes a total derivative in $d=6$. In the presence of sources, the theory enforces conservation of charge and dipole moment, capturing the immobility of isolated charges and the mobility of dipolar bound states. This structure can also be viewed as a higher-moment form of generalized global symmetry.

The prevalence of matter over antimatter in the observable universe remains one of the most profound puzzles in modern physics, necessitating sources of charge-parity (CP) symmetry violation beyond those incorporated in the Standard Model. While electric dipole moments (EDMs) of elementary particles serve as sensitive probes for such new physics, the hyperon sector has historically remained a largely unexplored territory. This review synthesizes recent advancements in the search for hyperon EDMs, focusing on the novel modular angular analysis of entangled baryon-antibaryon pairs produced in $J/ψ$ decays at electron-positron colliders. We discuss the theoretical formalism connecting the electric dipole form factor to full angular distributions and highlight the recent milestone achieved by the BESIII experiment, which established an improved upper limit on the $Λ$ EDM. This result represents a three-order-of-magnitude improvement over historical measurements.

We analyze the two-dimensional Palatini Gauss-Bonnet theory on an infinite strip (product of a finite interval with the infinite line, corresponding to ``time"). The theory has only boundary degrees of freedom. Its phase space is the cotangent bundle to the group manifold of $SL(2,\mathbf{R})$, subject to a (first-class) constraint quadratic in the momenta. With the simplest choice of boundary Hamiltonian, namely $H = 0$, the theory is shown to describe geodesics on the group manifold of $SL(2,\mathbf{R})$, with a ``mass" determined by the Palatini Gauss-Bonnet coupling constant. Other choices of boundary Hamiltonians compatible with gauge invariance are also possible. The symmetry group contains (left and right) group translations on $SL(2,\mathbf{R})$. These are ``boundary symmetries" from the bulk point of view, one copy acting on one end of the interval, the other copy acting on the other end. Comments on the quantum theory are also given.

We introduce a novel regularization scheme: Systematic Analytic Regularization (SAR). SAR regularizes a theory at the level of the action by analytically continuing the power of the kinetic operator, ensuring that the theory is formally finite before any terms in the Dyson series are evaluated. We demonstrate that SAR fully and self-consistently regularizes $\varphi^4$ and Yukawa theories at NLO.

Recent measurements of Standard Model W and Z boson production performed by the ATLAS and CMS Collaborations at the CERN Large Hadron Collider are reviewed. These include a search for charged lepton flavour violating decays of Z and Z' bosons, a measurement of W-boson angular coefficients and transverse momentum distributions, a triple-differential measurement of Z+jet production, and a study of the jet mass in boosted hadronic W boson decays with an extraction of the W boson mass. Together, these results highlight the role of precision measurements as stringent tests of perturbative quantum chromodynamics and electroweak theory. The large Run 2 dataset enables multi-differential measurements with enhanced sensitivity to parton distribution functions and increasingly precise tests of theoretical predictions.

We propose a manifestly supersymmetric formulation of the Symmetry Topological Field Theory (SuSymTFT) for theories with supersymmetry. The SymTFT is a framework that helps organizing symmetries and anomalies of a QFT. Albeit a lot of activity in the field has been devoted to the construction of the SymTFT for the bosonic symmetry structure, the fermionic case has not been analyzed in detail. Here, we consider the most prominent example of theories exhibiting fermionic symmetries, that is supersymmetric models. These are most naturally formulated on supermanifolds, where the supergeometry approach allows for a manifest organization of the symmetries according to their fermionic grading. We provide the general construction of the SuSymTFT as a super-BF theory living in a (n|m)-dimensional supermanifold and check our proposal in two particular examples, the compact and the chiral super-bosons in two dimensions.

A $ Z_3 $ symmetric 3-Higgs Doublet Model (3HDM) with two inert doublets and one active doublet (which plays the role of the Higgs doublet), the so-called I(2+1)HDM, is studied. We discuss the behaviour of this 3HDM realisation when one allows for a $ Z_3 $ soft-breaking term. In this setup, the lightest $Z_3$ charged neutral scalar can be a Dark Matter (DM) candidate. If the breaking scale is small enough, this model provides a long-lived neutral state with the opposite CP parity to the DM state. This long-lived particle could then be another effective DM candidate when its lifetime is comparable to the age of the universe. In this case, we have studied the properties of the ensuing relic density in the presence of the most recent limits from direct searches for DM. Conversely, the case in which the long-lived particle actually decays into the DM particle and Standard Model particles in a detector at a collider experiment is very attractive from the viewpoint of phenomenology. Under such conditions, we have studied signatures involving missing transverse energy and multiple leptons (and jets) at the International Linear Collider (ILC) in the presence of displaced vertices.

We show that the tree-level spectrum of heavy particles can be directly extracted from the Wilson coefficients of the corresponding effective field theory at low energies. This procedure is exact when the number of resonances is finite, and otherwise approximate. Our results are derived from a new class of analytic dispersion relations that depend nonlinearly on the scattering amplitude and apply to an exceedingly broad class of theories and kinematics.

We study cusped Wilson line operators in the Abelian Higgs model in $ d = 4 - ε$ at large external charges. Using a double-scaling limit $ Q \to \infty $, $ ε\to 0 $ with $ Qε$ fixed, we develop a semiclassical framework that provides analytic control beyond fixed-order perturbation theory. We compute the cusp anomalous dimension for arbitrary charges up to next-to-next-to-leading order in the gauge coupling, while resumming scalar self-interactions to all orders. Our results interpolate between perturbative and large-charge regimes, accessing domains that are invisible in fixed-order perturbation theory. As an application, we provide new predictions for various defect CFT observables, including the Mandelstam-Schwinger-dressed two-point function characterizing the superconducting phase transition.

Recent simulations increasingly resolve the small-scale structure of the circumgalactic medium (CGM), but the dynamical impact of ionising radiation on its cold $10^4$ K component remains poorly understood. We investigate the evolution of cold gas structures exposed to quasars' EUV radiation. We develop an analytical framework to describe the evolution of such clouds, introducing a new threshold that defines when a cloud becomes radiation-shielded. The framework is validated using radiation-hydrodynamic simulations of single static clouds. It predicts three evolutionary paths: (i) an optically thin regime, in which radiation uniformly ionises the cloud; (ii) a radiation-shielded regime, where the cloud remains largely unaffected; and (iii) a rocket-effect regime, in which the propagation of the ionisation front ionises the illuminated side while compressing the opposite side, later accelerating the surviving cold clump. In the latter regime, the cloud's Ly$α$ luminosity can be up to one order of magnitude higher than the optically thin case. Such luminosities are as high as $70\%$ of the values obtained from a fluorescent regime without considering hydrodynamical response. Unless the cloud is shielded, at least $\sim 50$-$60\,\%$ of Ly$α$ emission arises from recombination. Applying this framework to both a ray crossing a population of clouds, and a ray propagating inside a cold stream, we find that the cold CGM around bright quasars ($L_{\mathrm{ν,LL}} \sim 10^{31.6} \, \mathrm{erg\, s^{-1}\, Hz^{-1}}$) is likely fully ionised, whereas the one around faint quasars ($L_{\mathrm{ν,LL}} \sim 10^{28.6} \, \mathrm{erg\, s^{-1}\, Hz^{-1}}$) predominantly experiences a rocket-effect regime. These results imply that the hydrodynamical response of cold CGM structures to quasar radiation must be considered when deriving their physical properties, particularly for faint quasars.

We present the most comprehensive set of empirical optical strong-line metallicity calibrations to date, based on the DEep Spectra of Ionised REgions Database (DESIRED), the largest compilation of HII regions and galaxies with direct electron-temperature determinations assembled to date. We construct a high-quality calibration sample of 2392 spectra$-$1029 extragalactic HII regions, 1296 local star-forming galaxies, and 67 high-redshift ($z > 2$) galaxies$-$drawn from 201 independent literature references and spanning $12+\log({\rm O/H}) \in [6.79, 9.07]$. Physical conditions and chemical abundances are derived homogeneously using up-to-date atomic data. We derive 27 strong-line calibrations covering oxygen-, nitrogen-, sulphur-, argon-, and neon-based line ratios, including 4 previously uncalibrated diagnostics, with reported validity ranges and intrinsic dispersions (typically $\sim0.15-0.35$ dex). For the first time in a systematic calibration framework, all relations are presented for both the homogeneous temperature case ($t^2 = 0$) and a scenario including temperature inhomogeneities ($t^2 > 0$), thereby reconciling abundances from recombination lines (RLs) and collisionally excited lines (CELs) and directly tackling the abundance discrepancy problem. A comparison with previous calibrations shows that the DESIRED relations span the broadest validity intervals while remaining anchored to the empirical data. Crucially, recently proposed JWST-based high-redshift calibrations are consistent with our relations within the intrinsic scatter, demonstrating that the diverse composition of the DESIRED sample naturally encompasses the ionisation conditions found at high redshift. These results indicate that sample diversity, rather than redshift-specific recalibration, is key to reliable abundance determinations across cosmic time.

Understanding supernova (SN) progenitors remains a major challenge in astrophysics, as it involves untangling the complex interplay between stellar physics (e.g., evolution, binarity, explosion) and environments (e.g., metallicity, star formation rate). To address this, we present relative frequencies of core-collapse SNe (CCSNe) as a function of metallicity using two complementary samples: (i) all literature SNe that have associated host galaxy parameters (absolute magnitudes, stellar masses, and/or oxygen abundances); and (ii) SNe classified between 2019 and 2024 with host magnitude information, including distance-limited subsamples within 50 Mpc and 100 Mpc. We found that CCSNe from the literature sample are associated with luminous galaxies, reflecting both the higher stellar content of such systems and selection biases inherent to targeted surveys. In contrast, the distance-limited subsamples provide a less biased view, showing that hydrogen-rich SNe (SNe II) are more commonly found in lower-luminosity galaxies than stripped-envelope SNe (SESNe). Comparisons between the literature sample and distance-limited subsamples indicate that trends derived from global measurements remain consistent. For the SESNe-to-SNe II ratios, we confirm a slight increase with metallicity, reflecting a higher fraction of SESNe in metal-rich environments. Comparison with theoretical predictions shows that models including either binary interactions or rotation can broadly reproduce the observed trends, although degeneracies remain, and no single scenario uniquely explains the data. Overall, our results provide observational constraints on massive-star evolution and highlight the key role of metallicity and binarity in shaping the observed diversity of CCSNe.

We present an analysis of XRISM and NuSTAR data obtained for the nearby low-Eddington active galactic nucleus NGC 7213. Our goal is to examine whether its He-like and H-like iron emission lines can be reproduced by photoionization or collisional ionization processes. Using the broad-band energy coverage of our data (2-60 keV), we first constrained the continuum shape. Then, we focused on the iron-K band in the Resolve spectrum. Gaussian fits to Fe XXV He$α$ and Fe XXVI Ly$α$ lines suggest that they may have different velocity widths: $v_σ=790^{+370}_{-240}$ km s$^{-1}$ for Fe XXV and $v_σ=2610^{+1700}_{-1580}$ km s$^{-1}$ for Fe XXVI. In this case, the He$α$ resonance line (w) and forbidden line (z) have similar intensities of $\approx0.5$-$0.6\times10^{-5}$ ph s$^{-1}$ cm$^{-2}$, while the intercombination lines (x+y) are not significantly detected with upper limits of $\lesssim 0.2\times10^{-5}$ ph s$^{-1}$ cm$^{-2}$. Motivated by the possible difference in the line widths, we tested one- and two-zone photoionized and collisionally ionized models. Our results show that the additional ionized component is not significantly required, and the current data cannot uniquely determine whether photoionization or collisional ionization dominates. Moreover, if the Fe XXV He$α$ complex implies that the weak x+y lines are suppressed relative to the w and z lines, such a structure is difficult to reproduce with either ionization model adopted. Finally, by comparing NGC 7213 with M 81$^\ast$, accreting at a much lower Eddington ratio of $λ_{\rm Edd}\sim 10^{-5}$, we found a decrease in the density of the gas responsible for highly ionized iron emission, which may imply that the density decreases with decreasing $λ_{\rm Edd}$.

The high-frequency gravitational-wave band is often discussed primarily in the context of new physics, but realistic Standard-Model foregrounds remain incompletely characterized. We investigate pulsar polar caps as a physically motivated astrophysical source of high-frequency gravitational waves, generated by repeated discharge cycles in compact near-surface plasma gaps. Our baseline result is population-level: we construct the signal from the Galactic normal-pulsar population rather than from a single especially favorable object. To do so, we calibrate the source dynamics with particle-in-cell simulations performed at real physical scales, with physical pulsar parameters mapped directly onto numerical scales, and then lift the resolved longitudinal discharge to a cap-scale emission model. The gravitational-wave signal is computed in a full Fourier-space framework, retaining finite-source, geometric, and polarization effects explicitly. Within this treatment, the dominant contribution is not the purely electric channel emphasized in some earlier simplified approaches, but a source channel involving the large background magnetic field and discharge-induced transverse fluctuations of magnetic field. Integrating this description over a normal-pulsar population, we find an astrophysical foreground in the MHz-scale high-frequency band that can overlap with and partially obscure the thermal gravitational-wave signal sourced by the plasma of the early Universe. At the same time, the normalization remains sensitive to the modeled assumptions. Although the predicted strain remains far below current experimental sensitivity, pulsar polar caps provide a concrete Standard-Model foreground benchmark in a band often treated as nearly background-free. Alternative source configurations further broaden the plausible signal range around this baseline.

Little Red Dots (LRDs) are some the most intriguing galaxy populations recently identified at z>~4 with JWST. They constitute the most extreme class of a more abundant population of sources with `V-shaped' spectral energy distributions (SEDs) and compact morphologies, which includes also Little Blue Dots (LBDs). Finding brighter analogues to these sources requires surveying sky areas which are significantly larger than those covered with JWST. Euclid deep images are ideally suited for this purpose. We make use of Euclid near-infrared images, complemented by Spitzer Infrared Array Camera (IRAC) data, over 0.75 sq. deg. of the COSMOS field to select a sample of 233 sources with `V-shaped' SEDs at z>4. Out of those, we identify 16 sources with compactness >1sigma above the median of all z>4 galaxies, which we consider robust LRD/LBD candidates in our sample. The stellar masses of these 16 sources are in the range 10^{8.5} - 10^{10.5} Msun, so they are significantly more massive than typical JWST-selected LRDs/LBDs. Interestingly, half of them are about as old as the Universe at their redshifts. In addition, we find that the median photometric properties of the Euclid LRDs/LBDs are similar to those of the so-called Blue Dust-Obscured Galaxies (Blue DOGs). Less than 10% of all our `V-shaped' SED sources, including only one of the Euclid LBDs, correspond to known AGN. The latter mostly constitute a population disjoint to the `V-shaped' SED sources. Spectroscopic follow up of the Euclid LRDs/LBD candidates remains necessary to probe whether they host BLAGN as fainter analogues do and whether constitute a transition phase from these fainter sources to standard AGN.

We examine the dependence of the ellipticity of globular clusters in the Milky Way on their X-ray luminosity using two modern catalogs and combine them with optical and X-ray data from the literature. Kolmogorov-Smirnov tests applied across multiple subsets reveal statistically significant differences in the ellipticity distributions when both $L_{\rm X}$ and optical luminosity are considered. Two X-ray luminosity thresholds, $L_{\rm X}^*(M_{\rm{V}}=-7)=10^{33.05}$ erg/s and $L_{\rm X}^*(M_{\rm{V}}=-7)=10^{32.01}$ erg/s, yield the most reliable distinction. In contrast to earlier findings based solely on optical data, our results demonstrate that globular clusters with the highest X-ray luminosity tend to have higher ellipticity on average.

We introduce an innovative approach employing Cycle Generative Adversarial Networks (Cycle-GANs) to accurately simulate Carbon Monoxide (CO) emissions by learning features identified in thermal dust emission maps from the Planck satellite alongside HI data from HI4PI survey. Our training dataset is complemented by the targets represented by the two rotational transition lines of CO (J:1-0, J:2-1) provided by the Planck satellite. We ensure the robustness of our dataset by focusing on regions with a signal-to-noise ratio (SNR) exceeding 8. The outcomes, assessed utilizing angular power spectra and Minkowski functionals, confirm that our algorithm proficiently achieves the set goals, indicating that the amplitudes of the generated emission accurately reproduce the angular correlations and share the statistical properties of the employed CO targets. We thus aim at improving the current models of CO emission specifically in the high-Galactic latitude areas that have been hardly observed by the most recent surveys, and, in doing so, to address and overcome the limitations affecting current models regions. This research lays the groundwork for creating transformative synthetic simulations, leveraging convolutional neural networks tied to data procured from latest observations.

The purpose of the present work is based on two main observations: the tensions encountered by the standard $Λ$CDM model when confronted to precision small scale cosmological data and the finding that the matter distribution and the expansion of the Universe are axially symmetric roughly in the direction of the CMB dipole. Therefore, we propose, as a model for the inhomogeneous local universe, an axially symmetric Szekeres solution. After describing its main properties, we are left with three metric functions to be fitted to data between the observer and the transition to homogeneity which is an intrinsic feature of Szekeres spacetimes. So as to turn a difficult functional inference problem into a classical parameter estimation problem, we propose to use Chebyshev polynomial expansions, which, as a first step, we truncate after the second order terms. We are thus left with eight constant parameters: six for the metric functions, plus the observer's radial location and the cosmological constant. Here are the proper ingredients needed to implement the data fitting to the model in the future.

NGC 4151 is the brightest Seyfert-1 active galaxy in the pass band of the Resolve calorimeter spectrometer aboard XRISM. It has been observed on 14 occasions, resulting in a total exposure of 893 ks. Herein, we report on an analysis of the time-averaged spectrum. The narrow Fe K$_α$ emission line complex requires contributions from the torus and the optical broad line region (BLR). Models assuming an emissivity index of $q=2$ for these components are statistically preferred over models assuming $q=3$ for a flat disk (where $J\propto r^{-q}$). A smooth shoulder on the red wing of these line components is likely best interpreted as Compton scattering in a medium with bound electrons, potentially signaling the presence of dust at the base of the BLR and in the torus. The data statistically prefer the addition of relativistic reflection from the innermost accretion disk, extending down to a radius of $r = 3.2^{+3.5}_{-2.0}~GM/c^{2}$ and with an inclination of $θ= 29.7^{+0.5}_{-0.4}$ degrees. The Fe K edge at 7.1 keV is best modeled with contributions from multiple charge states, consistent with obscuration due to cool, $kT \simeq 5$ eV collisional gas or photoionized gas. Dust is not evident in the Fe K absorption edge. A spectrum of outflows is clearly revealed, with slow ``warm absorber'' winds spanning Fe XX-XXVI, fast winds primarily seen via Fe XXV and Fe XXVI lines, and ultra-fast outflows (or, UFOs) seen as broad Fe~XXVI lines. The warm absorbers are almost certainly ``failed'' winds that return to the central engine; the data constrain their radius, density, filling factor, and distribution. For the most conservative volume filling factors, the UFOs may not deliver the kinetic feedback needed to halt star formation, on average. However, they may generate galaxy-altering feedback for larger filling factors and/or during certain intervals. (abridged)

Off-nuclear tidal disruption events (TDEs) provide a rare probe of massive black holes (MBHs) outside galactic nuclei. Only a handful are known, including five X-ray-selected candidates and two optically selected events. We present observations of TDE 2025abcr, the second optically selected off-nuclear TDE, discovered at a projected offset of $9.08 \pm 0.02$ kpc from the nucleus of its host galaxy. We analyze X-ray, UV, optical, and infrared (IR) data from Swift, Keck, ZTF, and JWST. Broad H and He emission lines in the optical and IR confirm a TDE-H-He classification. From luminosity scaling relations and MOSFiT modeling, we infer a BH mass of $10^{6}$-$10^{7}\,M_{\odot}$, substantially smaller than the $10^{8.35 \pm 0.41}\,M_{\odot}$ BH inferred for the host-galaxy nucleus. We observe velocity evolution in the N III + He II emission complex, shifting from $-500$ km s$^{-1}$ at day $-7$ to $+1000$ km s$^{-1}$ by day $+29$, which we interpret as radiative transfer effects in an evolving reprocessing layer. The IR SED deviates from a thermal blackbody, with $νL_ν \propto λ^{-2.13 \pm 0.04}$, significantly shallower than the Rayleigh-Jeans slope of $λ^{-3}$. We rule out dust as the source of this IR excess. Two possibilities remain: free-free emission from reprocessing gas, or an unresolved stellar cluster at the TDE location. Reprocessed emission provides a natural explanation for the IR excess but an underlying stellar cluster of mass $\log(M_{*}/M_{\odot}) = 7.57 \pm 0.02$ and age $<$2 Gyr is also consistent with the data. If interpreted as a stellar cluster, the inferred mass suggests a stripped remnant of a satellite galaxy. The wandering MBH most likely originated in a minor merger with a smaller galaxy, although dynamical ejection from the host nucleus cannot yet be ruled out.

We present the results for the galaxy NGC 628 observed with the Planetary Nebulae Spectrograph (PN.S) equipped with the H$α$ arm. With the third PN.S arm, the H$α$ arm, we measure the H$α$ fluxes, in addition to fluxes and line-of-sight velocities (LOSV) of monochromatic spatially unresolved [OIII] 5007Å sources. The narrow band color ([OIII] 5007Å-H$α$) vs m5007 magnitude diagram separates planetary nebulae (PNe) from single compact ionized HII regions and supernovae remnants (SNRs), which also emit in [OIII]5007 Å. The goals are to detect bona-fide PNe in the face-on spiral galaxy NGC 628 (M74) so that we can measure the velocity dispersion of the stars perpendicular to the main plane of the disc. This study validates the empirical selection criteria for PNe with the PN.S in star forming discs. We classified 442 PNe and 251 spatially isolated, unresolved HII regions: the PN.S with the H$α$ arm increased the number of known PNe by a factor 4. We find evidence for two kinematically distinct PN populations in the NGC 628 disc. The kinematically cold PN population dominates the PN luminosity function close to the bright cut-off magnitude, indicating that the PN massive, short-lived progenitors dominate the PNLF bright cut-off in NGC 628. The warmer PN component increasingly dominates at fainter magnitudes. The velocity dispersion orthogonal to the disc plane are σz,cold = 8.8 kms-1 and σz,warm =26.1 kms-1 respectively, over a range of radii 80 to 425 arcsec. These components contribute with the ratio 46% (cold) and 54% (warm). Once the velocity dispersion of the old component is matched with the population's scale height, the decomposition of the rotation curve for NGC 628 leads to a maximal disc, with the rotation of the baryonic component accounting for 78% of the total rotational velocity in NGC 628.

Neutron star low-mass X-ray binaries (LMXBs) are the progenitors of millisecond pulsars. In these systems, old neutron stars (NSs) can be spun up during a long-lasting accretion phase. The discovery of accreting millisecond X-ray pulsars (AMXPs) and transitional millisecond pulsars has provided key observational insights into the connection between millisecond pulsars and LMXBs. In this work, we have investigated both the binary system and the individual NS evolution leading to AMXP properties. We use MESA to analyse the binary evolution of LMXBs, following three distinct evolutionary tracks defined by the AMXP donor types. We find that while the magnetic braking index may affect the mass-transfer history, the initial orbital period is the most influential parameter that shapes the overall binary evolution. We use the mass accretion histories estimated from these binary simulations to study the rotational evolution of NSs employing the model that can account for torque-luminosity relations and the lack of X-ray pulses from most of these systems. With reasonable model parameters, our model results are in agreement with the typical properties of AMXPs. For these AMXP sources from each evolutionary track, we have shown that the model can reproduce the NS and binary properties simultaneously. Finally, we discuss the time-scales of different evolutionary paths, as well as the conditions under which these systems could be detectable at various stages of their evolution.

Open clusters (OCs) in our Galaxy can be found in pairs, possibly forming physical binaries, or in groups. Such objects offer unique insights into the process of star formation. They testify of dynamical interactions at the local and Galactic scales. Making their census as complete as possible is therefore a valuable endeavor. This work aims to identify and characterize new OC pair candidates that were overlooked in previous studies. Two recent comprehensive catalogs were cross-matched in order to identify OCs in the first catalog that are missing in the second one. From this list, counterparts in the second catalog were searched within a three-dimensional distance of 50 pc. Candidate pairs were then selected by applying constraints on the tangential velocity difference. Orbital integration was performed to assess gravitational binding. Similarity in radial velocity and age was evaluated. We identify seven isolated binary cluster candidates, comprising two likely bound systems with stable orbits over 100 Myr, two pairs with a possible common origin but lacking radial velocity confirmation, and three pairs with significant velocity discrepancies suggesting they are unbound or in transitional states. We also identify six cluster group candidates, refining the membership of known complexes such as UBC_672 and NGC_1977, and discovering a new group around FSR_0198. Notably, the UBC_392 group exhibits coherent proper motions but inconsistent radial velocities and large age spreads, indicating it is not gravitationally bound. Additionally, we reconcile 15 clusters with discrepant nomenclature between the two catalogs. Multi-catalog integration combined with kinematic and dynamical validation is essential for establishing a complete census of Galactic cluster pairs. Our findings expand the known binary cluster sample and provide refined targets for future studies.

The age-metallicity relation (AMR) is a fundamental observational constraint for understanding the chemical evolution of the Galaxy. As reliable cosmochronometers, white dwarfs in binary systems with main sequence companions (WD+MS binaries) provide excellent laboratories to study this relation, since both components are expected to be coeval. We construct a sample of widely separated WD+MS binaries using data from the third data release of the Gaia mission in order to investigate the AMR of the Galactic disk. The sample is identified using photometric measurements and parallaxes of both components. White dwarf ages are derived by interpolating their Gaia absolute G magnitudes and BP-RP colours within state-of-the art white dwarf evolutionary sequences. We compile publicly available [Fe/H] abundances for the main sequence companions from the literature and combine them using different statistical approaches to obtain representative metallicity values for each system. We derive the AMR from several sub-samples of WD+MS that use independent measurements of [Fe/H] abundances and consistently find a large dispersion in [Fe/H] at all ages. This behaviour agrees with previous determinations of the AMR based on both WD+MS binaries and samples of isolated stars. Our results reinforce the observational evidence that the AMR in the Galactic disk exhibits substantial intrinsic scatter, likely reflecting the combined effects of multiple mechanisms such as radial migration, inhomogeneous chemical enrichment, and variations in the star formation history.

We investigate the ability of machine learning to infer the virial mass ($M_{\rm vir}$) and the scale radius ($r_{\rm s}$) of galaxy clusters from their observables. Using the Uchuu--UniverseMachine galaxy catalog at $z=0.093$, we generate mock cluster observations that include interlopers, and we encode each cluster as an image representing the two-dimensional joint probability distribution of member galaxies' projected position and line-of-sight velocity. We train two architectures: a baseline convolutional neural network (CNNb) following a previous approach, and an extended model (CNNr) that appends richness as an additional scalar input. We further compare the performance of networks trained on the all cluster sample and on a dynamically relaxed subsample. Across the test ranges $10^{13.7}\leq M_{\rm vir}\leq10^{15.3}$ Msun/h and $10^{1.7}\leq r_{\rm s}\leq10^{2.7}$ kpc/h, all configurations yield nearly unbiased absolute median residuals (within 0.01 dex). For the halo mass, adding richness narrows the residual distribution, reducing the standard deviation from 0.133 to 0.122 dex for the all sample, and from 0.124 to 0.111 dex for the relaxed sample. For the scale radius, restricting the training to relaxed clusters improves the performance more than adding richness. The standard deviation decreases from 0.180 to 0.154 dex for CNNb and from 0.175 to 0.148 dex for CNNr, while the inclusion of richness yields only a modest improvement of 0.005 dex. These results demonstrate that machine learning is a powerful tool to infer the mass and internal mass distribution of clusters, providing a new window for cosmological inferences and understanding galaxy formation processes.

Reactions occurring on the ice-covered surfaces of interstellar dust grains are considered to be among the most important sources of complex species in the interstellar medium. Despite this, molecules such as cyanamide, NH2CN, are largely underpredicted by current astrochemical models suggesting that the network of reactions currently used to describe this species and its tautomer carbodiimide, HNCNH, are incomplete. Here, we performed a theoretical investigation of the reaction of ground state atomic carbon C(3P) with molecular nitrogen N2 in both the gas-phase and on the surface of amorphous solid water (ASW) clusters to examine its potential importance in the formation of NH2CN and HNCNH. We show that the reaction of gas-phase C-atoms with N2 molecules already present on the ASW surface results in the barrierless formation of CNN. Following exothermic hydrogenation reactions, the N-N bond of the C-N-N bearing intermediates is broken allowing the formation of molecules with N-C-N backbones through cyclic intermediates over low barriers. To test the importance of these processes to NH2CN and HNCNH formation, these reactions were included in a three-phase astrochemical model of low-mass protostellar evolution employing a reaction network that was updated to better describe the formation and destruction pathways of related small nitrogen bearing molecules. These simulations demonstrate that the ice surface reaction between C and N2 represents by far the dominant source of NH2CN and HNCNH in protostellar environments and in dense clouds.

We study induced (stimulated) scattering of linearly polarized, strong electromagnetic waves in pair plasmas, which is crucial for understanding the propagation of fast radio bursts (FRBs). Magnetars are the most promising progenitors of FRBs, and FRBs propagate through the magnetar wind and successfully escape before being significantly scattered. We revisit the steady-state solution of linearly polarized electromagnetic waves in pair plasmas with arbitrary amplitude, and demonstrate that the nonlinearity is characterized by the nonlinearity parameter $a_0ω_{pe}/ω_0$ rather than the dimensionless amplitude $a_0$, where $ω_{pe}$ is the electron plasma frequency and $ω_0$ is the wave frequency. We follow the time evolution of the steady-state solution for the linear regime $a_0ω_{pe}/ω_0 \ll 1$ by performing one-dimensional particle-in-cell simulations, and show that the conventional linear analysis of induced scattering assuming $a_0 \ll 1$ is applicable even for $a_0 > 1$ when the Lorentz boost due to the plasma motion in the incident wave is considered. The saturation level is controlled by $a_0ω_0/ω_{pe}$, which corresponds to the ratio of the wave energy to the plasma energy, and the incident wave is hardly scattered for $a_0ω_0/ω_{pe} \gg 1$. We discuss the application of our results to FRBs.

Upcoming cosmological surveys will achieve increasingly precise constraints in cosmological parameter estimation. To guarantee the robustness of cosmological analyses, it is essential to account for and model systematic effects that can bias cosmological constraints, shifting the best fit parameters away from their fiducial values. It is possible to approximately infer the biases that un-modelled systematic effects might introduce in cosmological parameter estimation by means of the Fisher matrix formalism. In this paper, we introduce a new application of this formalism, where by inverting the process, we investigate whether a specific missing or mis-modelled systematic effect can explain away a given tension between two different probes or experiments. We showcase the proposed methodology by examining two representative systematics: galaxy intrinsic alignments and baryonic feedback. As the method is agnostic to the systematic effect and can be applied to a wider range of scenarios, we discuss more possible future applications. While the proposed approach is accurate in the limit of small offsets in the cosmological parameters, where the likelihood can be considered linear in both the cosmological parameters and the systematics, in practice, the region of validity depends on the systematic effect. In general, even beyond this region, the approach still provides a useful test that helps indicate the magnitude and direction of potential biases from systematic effects in data.

A long-term variability study spanning a range of black hole mass systems, from microquasars hosting stellar-mass black holes to active galactic nuclei (AGNs) harboring supermassive black holes, provides new insights into the physics of relativistic jets. In this work, we investigate the optical variability of both jetted and nonjetted AGNs. We apply a stochastic process known as the Damped Random Walk (DRW) to model light curves from the Zwicky Transient Facility (ZTF) DR23. Our results show that the mass-scaled characteristic timescales across the black hole mass exhibit a linear relationship with a slope of 0.35-0.50. This analysis confirms a previously observed correlation between the damping timescales and black hole mass and extends it by incorporating 125 newly identified non-jetted AGNs selected from the Burst Alert Telescope (BAT) AGN catalogue. The derived slope of the relation between the damping timescales and black hole mass aligns with recent theoretical predictions, supporting the presence of a universal accretion mechanism in AGNs across different mass scales. Furthermore, our findings suggest a novel implication: the properties and production mechanisms of relativistic jets may be largely independent of black hole mass.

Bubble N68 in the G35 complex shows clear cloud-cloud collision (CCC) signatures. Its semi-ring-like morphology harbors many significant massive star formation tracers: 6 HII regions, 4 6.7 GHz masers, 5 Midcourse Space Experiment sources, 9 radio peaks, and nearly 10 O/B-type stars. We also identified 163 young stellar objects (45 Class I, 5 Flat, 113 Class II), indicating active star formation toward N68. Our molecular study with CO reveals two distinct molecular clouds (N68a: 47-56 km s$^{-1}$; N68b: 56-64 km s$^{-1}$), with broad bridge features and complementary distributions at their borders, indicating an ongoing CCC. Star formation in N68 is collectively driven by collect-and-collapse (CC), radiation-driven implosion (RDI), and CCC mechanisms. However, compared with the CC and RDI mechanisms, the CCC mechanism does not enhance the star formation efficiency; instead, it tends to trigger the formation of massive stars. N68, along with bubbles N65 and N61, constructs a $\sim100$ pc scale CCC system in the G35 complex.

We present distances to 56 molecular clouds within $10\degr \leq l \leq 20\degr$ and $|b| \leq 5.25\degr$ from the Milky Way Imaging Scroll Painting (MWISP) $^{12}$CO survey, 47 of which are first-time determinations. The molecular clouds were identified using the DBSCAN algorithm, and their distances were measured with the model-calibrated color-distance method using $J-K{_s}$ colors and the distances provided by 2MASS and \textit{Gaia} EDR3. The distances range from $\sim$275 pc to $\sim$2118 pc. We also derived the physical properties of molecular clouds and found a moderate correlation between the dust extinction and the $^{12}$CO integrated intensity.

Although the relationship between molecular gas content and the star formation rate (SFR) has been extensively studied in nearby galaxies, it remains controversial whether the star formation efficiency (SFE) depends on galactic structure. In particular, whether the SFE is suppressed in the bar region compared with other structures, and the physical origin of this suppression, remain poorly understood. In this study, we investigate variations in the SFE and its physical drivers in the bar region of the nearby spiral galaxy M83, using multi-wavelength observations toward the bar and spiral arm regions on a scale of 200 pc observed with ALMA. From the molecular gas surface density derived from $^{12}$CO($J=2-1$) and $^{13}$CO($J=1-0$), the dense molecular gas surface density derived from HCN ($J = 1-0$), and the star formation rate surface density determined from extinction-corrected H$α$, we find that the SFEs in the bar region are roughly a factor of two lower than those in the spiral arm, indicating that the SFE is systematically suppressed in the bar. Moreover, we find that the SFEs of dense gas are lower in the bar than in the arm by a factor of about 0.35. These results suggest that not only the efficiency of converting bulk molecular gas into stars reduced in the bar, but the efficiency of star formation from dense gas is also lower. In addition, the CO line widths are systematically larger in the bar region and exhibit a negative correlation with both the SFE and the dense-gas SFE, consistent with the interpretation that enhanced turbulent motions hinder star formation. Although the analysis is limited to small regions of M83, our results suggest that the suppression of the SFE is related to large-scale dynamical effects on the molecular gas, such as strong shocks induced by cloud-cloud collisions and/or shear, both driven by non-circular motions in the bar.

We present an analysis of the residual velocities from a sample of 105 Galactic planetary nebulae (PNe), the largest done to date on this subject. The analysis has been carried out with long-slit, high dispersion echelle spectra. The data were drawn from the San Pedro Mártir Kinematic Catalogue of Galactic Planetary Nebulae. The residual velocity is identified with turbulence in the plasma and is derived by decomposing the emission line profiles into their structural contributors. Turbulence seems pervasive throughout all the PNe in the sample. We find the values for the residual velocities in the sample to be either transonic or slightly supersonic in the ionized environment. When residual velocities of [N II], [O III] and He II in the same PNe are compared, there is a tendency for the residual velocities of the higher ionized ion to be larger by about 5-10 km s$^{-1}$, indicating that the turbulent structure is larger in the inner zones of the PN. We find in general no clear correlation between the residual velocities and other nebular parameters such as morphology, global expansion velocities, ionization degree and binary cores. The only exception is the case of PNe with H-poor ([WR]-type) central stars, where we confirm previous results that have consistently shown higher residual velocities for this group of PNe as compared to those with H-rich central star atmospheres. Turbulence seems to be a localised, random, dissipative process occurring in the inner sections of the shell and may affect its early evolution.

The orbital observatories PAMELA and AMS-02 have detected a significant excess in the cosmic ray (CR) positron flux at energies above several tens of GeV. The measured values exceed those expected in models of secondary origin of positrons due to inelastic collisions of CR nuclei with the interstellar matter. This excess may be due to the annihilation or decay of hypothetical dark matter particles or, alternatively, to the contribution of the nearest pulsars. The measured positron-to-antiproton flux ratio is virtually independent of energy from 60 to 400 GeV. This can be understood if the observed local spectra of positrons and antiprotons in the range of tens to hundreds of GeV are formed by the same source. Fast-moving pulsars form pulsar wind nebulae with bow shocks (BSPWNe), which accelerate both the positrons and electrons of the pulsar wind and the hadrons and leptons of galactic CRs from the interstellar medium via the Fermi acceleration mechanism in colliding flows. Such a system can produce identical particle spectra regardless of the site of particle injection. The nearest to the Earth millisecond pulsar PSR J0437-4715 forms a pulsar wind nebula (PWN) with a bow shock observable in optical and ultraviolet wavelengths. This BSPWN is a possible candidate for the main near-Earth ``factory'' of antiparticles along with the Geminga PWN. Considering PSR J0437-4715, we provide the Monte Carlo simulations of particle acceleration in its BSPWN and the analytical model of anisotropic diffusion in the local interstellar medium. We show that this pulsar's contribution can explain the observed positron flux from 30 GeV to 1 TeV, and simultaneously the antiproton flux at hundreds of GeV with an almost energy-independent positron-to-antiproton flux ratio using $\sim 25\%$ of the PSR J0437-4715 pulsar wind power to accelerate positrons and electrons and re-accelerate antiprotons.

A major obstacle to improving models of planet formation is understanding how the local environment influences the lifetime of the disks in which they form. The spread in observed disk lifetimes is caused by effects both observational (e.g., target selection, survey sensitivity) and physical (e.g., disk destruction by internal and external photoevaporation); however, the degree to which each plays a role remains poorly constrained. Isolating the impact of external photoevaporation on the disk lifetime benefits from the inclusion of low-mass ($\lesssim0.5$ M$_{\odot}$) YSOs, for which this effect is most predominant. In this work, we measure the inner disk fraction from JHK excess in the ~6000 M$_{\odot}$, ~1 Myr-old star-forming region M17. Using VLT/HAWK-I, we perform a deep photometric survey of an ~8$^{\prime}\times$8$^{\prime}$ field towards the region. The ~4 times greater sensitivity and ~2-3 times higher resolution than previous surveys of M17 reveal 10,339 sources. We select cluster members using the Massive Young Star-Forming Complex Study in Infrared and X-ray (MYStIX) catalog and find a disk fraction of 28$\pm$2%: the first X-ray-selected disk fraction measurement in M17 to include low-mass YSOs, and only the second such measurement in any high-mass star-forming region. After correcting for observational biases, we find no correlation between disk fraction and incident UV flux within M17, likely due to dynamical mixing within the region. However, when compared to other regions of similar age, we find lower disk fractions in regions with higher UV fields, suggesting that external photoevaporation decreases the average disk lifetime.

We present updated hard X-ray polarization measurements of the Crab pulsar and nebula obtained with the balloon-borne polarimeter XL-Calibur in the ~19-64 keV energy range. During the flight, intermittent GPS-failure resulted in poorly constrained timing for ~38% of the Crab dataset. By implementing a new phase-recovery method that reconstructs timing during extended GPS-off intervals, phase tag data is recovered for ~95% of the GPS-off dataset, increasing the precision of the phase-resolved analysis. Phase-information for the data is recovered by using the Crab pulsar, with its 33 ms period, as an external timing source. Using a Markov-Chain Monte-Carlo framework to jointly fit phase offsets and frequency derivatives, sufficient phase accuracy is achieved, across multiple periods without GPS for a phase-resolved analysis. This enables inclusion of nearly the full dataset in the polarization study. The polarization degree of the nebular emission is found to be (27.7${\pm}$4.9)% at a polarization angle of 127.2°${\pm}$5.1° confirming previous XL-Calibur results and remaining aligned with the Crab's spin axis, consistent with synchrotron emission from the inner nebula. Phase-resolved measurements show that the off-pulse and bridge intervals exhibit a strong polarization, while the pulsar peaks, although weakly constrained, remain in agreement with the softer-energy trends of IXPE. These findings reinforce a scenario in which hard X-ray emission arises primarily in the nebular torus and wind regions. The successful recovery of precise phase tagging from GPS-off data demonstrates the capacity to use the pulsar as an external clock even in the case of sparsely populated data.

The most sensitive to-date multimessenger detection of the standing accretion shock instability in real interferometric data is presented, which quantitatively identifies the presence of the SASI in core-collapse supernovae using neutrino and gravitational-wave (GW) signals. In the GW channel, the coherent WaveBurst (cWB) software on its version XP is implemented, among with real LIGO data from the O3 and O4 observing runs. With this, a more accurate estimation of parameters, such as the central frequency and signal duration, is obtained for both sets of data. The SASI identification probability versus false alarm rates is presented in the form of Receiver Operating Characteristic (ROC) curves. For O3, the new study for the combined GW and neutrino detection condition, labeled as $x + y$, shows an identification probability (previous best results from Lin et al. [1]) of 1 (1), 0.90 (0.70) and 0.37 (0.34) at 1, 5 and 10 kpc for a false identification probability of 0.10. On the other hand, using O4 shows that the GW channel by itself is sensitive enough to provide almost perfect identification probability scores, with identification probability values of 1, 0.99 and 0.97 for a false identification probability of 0.01 at 1, 5 and 10 kpc, respectively.

Identification and characterisation of Kilonovae (KNe) can be instrumental in improving our understanding of cosmology and astrophysics. However, their detection poses unique challenges due to rarity and faintness. Upcoming telescopes, with their deep imaging capabilities and wide field-of-views, will provide a unique opportunity to observe these rare and faint transients. The Rubin Legacy Survey of Space and Time (LSST) will generate a deluge of data, making it essential to deploy fast, efficient methods for identifying genuine KNe, especially when they are gravitationally lensed. To address this, we simulate realistic populations of both unlensed and lensed KNe in the six LSST bands. Comparing with the Type Ia Supernovae, we find that the KNe color evolution is more rapid and the two separate out when their colors are compared at two epochs. Since the mergers of compact binaries are probable progenitors of KNe, the KNe properties may be affected by the delay time distribution (DTD) of the mergers, which is dictated by the minimum delay time ($τ$) and power-law slope. For longer $τ$ and shallower slopes, we find an increased rate of detectable KNe in LSST. We generate the first statistically realistic lensed KNe population for different DTDs and find that the rate of detectable lensed KNe increases for DTDs with longer $τ$ for a fixed slope. We further note that an AT2017gfo-like event at a redshift of 0.5~(1.0) needs magnification of at least 5~(44) to be detectable in LSST.

The redshifted 21cm signal from the post-reionization epoch is highly non-Gaussian; thus, higher-order statistics, such as the bispectrum, are required to extract this non-Gaussian information. However, high-signal-to-noise ratio detection of the 21cm auto-bispectrum will be hindered by the presence of residual systematics. Cross-correlating the 21cm signal with galaxies offers a promising path to suppress this uncertainty from residual systematics and potentially increase the signal-to-noise ratio. We present a comprehensive analysis of the HI-galaxy cross-bispectrum using the predictions of theoretical galaxy evolution models defined on large cosmological volumes. Our analysis includes the cross-bispectrum for different triangle sizes and shapes, as well as for different combinations of the HI and galaxy fields. We forecast the detectability of the 21cm-galaxy cross-bispectrum at redshift $z\approx1$ with Euclid-like galaxy survey and SKA-Mid observations in both interferometric and single-dish modes of survey. We find that the 21cm-galaxy cross-bispectrum shows enhanced detectability compared to the 21cm auto-bispectrum for all unique triangles in the presence of instrumental noise for observations in interferometric mode. We forecast a 10$σ$ detection of cross-bispectrum for squeezed-limit triangles and a 100$σ$ detection for all shapes combined for scales $0.2~\text{Mpc}^{-1}\leq k_1 \leq 0.9~\text{Mpc}^{-1}$ with 100 hours of SKA-Mid observations per pointing. However, the detectability of the cross-bispectrum for large scales ($k_1 < 0.1~\text{Mpc}^{-1}$), which is accessible with the single-dish mode of survey, is limited by cosmic variance. Our analysis presents a first step towards an end-to-end analysis pipeline for the future observations of the 21cm-galaxy cross-bispectrum.

Active Galactic Nuclei (AGN) are intertwined with galaxy evolution, injecting energy into the interstellar medium (ISM) that could regulate star formation as a galaxy evolves. However, the phenomena through which we observe AGN are multiphase and multiscale, which can lead to conflicting results for how significantly AGN influence the ISM. We endeavor to characterize the spatial extent and dominant modes of AGN feedback in M51, which hosts a low-luminosity Seyfert nucleus and multi-phase outflow. We identified regions dominated by AGN ionization using an emission line ratio (ELR) function constructed from VENGA integral field spectroscopy. We then investigated how AGN feedback influences the ISM using cloud-scale mapping of dense molecular gas tracers HCN(1-0), HNC(1-0), HCO+(1-0), and N2H+(1-0) provided by SWAN. This combined dataset has a resolution of 180pc, providing a clear demarcation of where AGN feedback dominates the ISM. If we assume that N2H+ is the best tracer of dense, cold gas in SWAN, then AGN-dominated regions defined by the ELR all have greater emission in (1-0) transitions in HCN, HNC, and HCO+ than expected if they traced dense gas alone, implying excitation of these lines from the AGN. The ELR better selects these regions than molecular tracers of AGN activity like HCN/HCO+. The highest ELR values are also associated with optical and molecular shock tracers (HNCO/CO), indicating a potential dense molecular outflow in the nucleus that agrees with the heightened N2H+ emission in this limited region. All tracers of AGN activity point to a "two-stage" feedback scenario, whereby mechanical feedback from the jet-ISM interaction spurs soft X-ray emission that excites molecules such as HCN. Dense gas entrenched in a molecular outflow may also lead to a greater chemical abundance of multiple tracers measured with SWAN, but to a lesser extent than excitation from AGN feedback.

Two time domain surveys, recently funded as part of the Eric and Wendy Schmidt Observatory System; the Argus Array, in the optical, and the Deep Synoptic Array (DSA), in the radio, will transform gamma-ray burst (GRB) science via the serendipitous discovery of hundreds of GRB afterglows per year. In this work, we simulate DSA and Argus observations of GRB afterglows. We find that, of the long-duration GRBs (LGRBs) detected by the Fermi Gamma-ray Burst Monitor, $(24 \pm 2)\%$ will yield afterglow detections with Argus and $(42 \pm 3)\%$ with DSA, corresponding to rate of $47 \pm 4$ and $82 \pm 7$ per year respectively. We also compute rates for both upcoming and proposed GRB monitors; the forthcoming StarBurst Multi-messenger Pioneer, with $62 \pm 5$ detections per year in Argus and $117 \pm 8$ detections per year in DSA and the Moon Burst Energetics All-sky Monitor (MoonBEAM) concept, with $62 \pm 6$ per year in Argus and $105 \pm 10$ per year in DSA. The observatory system will detect also 116$\pm$8 optical and 217$\pm$15 radio afterglows per year, independent of GRB triggers, exceeding the current annual rate with global follow-up. Afterglow counterparts to short-duration GRBs, originating from neutron star mergers, will be detected at $5$-$10$% of the LGRB afterglow rate, which is promising for multi-messenger detections of gravitational wave sources and constraining the neutron star merger rate. The Argus Array, with its second-minute cadence, will detect afterglows before they peak $\sim 18\%$ of the time which will dramatically increase the sample of observed reverse shock and prompt optical emission.