Peer Revenue Returns subject to peer review

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

All of arXiv

Wednesday 2 September

Signal Processing

Bidirectional CFO Separation for Radial Velocity Estimation In 5G NR TDD V2X Links

High-accuracy velocity estimation over orthogonal frequency division multiplexing (OFDM) links is challenging because Doppler shifts and local oscillator (LO) drift contribute the same way to the observed carrier frequency offset (CFO). We propose a bidirectional CFO separation method for a frequency range-1 (FR1) time division duplex (TDD) Uu link between a roadside gNB/TRP and a vehicle user equipment (UE). Downlink (DL) and uplink (UL) CFO estimates are obtained from the physical downlink and uplink shared channel phase tracking reference signals (PT-RS), respectively, their sum isolates the radial Doppler and allow for the extraction of the radial velocity, while their difference recovers the relative LO offset. The estimates are obtained in closed form, without any iterative processing, using only a compact scalar CFO report fed back on the UL, the method can be co-scheduled with an new radio (NR) round trip time (RTT) positioning event to enable simultaneous ranging measurements along velocity estimates. We derive an idealized Cramér-Rao lower bound (CRLB), show that a linear phase slope estimator attains it at high signal to noise ratio, evaluate sensitivity to angle of arrival uncertainty and multipath, and discuss RF chain mismatch. The resulting link level implementation is aligned with NR PDSCH/PUSCH PT-RS procedures and is suitable for network assisted vehicle to everything (V2X) integrated sensing and communication (ISAC).
Read the full abstract

Probability

Aperiodic random walks: uniform estimates, local limits and invariance principles

We prove uniform endpoint estimates, killed local limits, and conditioned-bridge invariance principles for triangular arrays of centered, aperiodic, integer-valued random walks with uniform exponential tails. The endpoint estimates cover all nonnegative heights, while the local limits treat positive diffusive and zero endpoints. The increment law may vary with the scaling parameter.
Read the full abstract

Materials Science

Linker Functionalization and pH Tuning Enhance Solar-Driven Catalytic CO$_2$ Reduction in MOF-5

The wide band gap of metal-organic framework (MOF) 5 constrains its use in photocatalytic carbon dioxide (CO$_2$) reduction despite its high porosity and favorable mass-transport properties. Adopting a state-of-the-art first-principles approach, we systematically investigate the effects of volumetric strain, metal-node substitution, linker functionalization, and pH control as knobs to improve the CO$_2$ photocatalytic ability of MOF-5. Strain and metal-node substitution negligibly affect the gap, whereas linker functionalization narrows it into the visible range via in-gap states while preserving reduction-side alignment at pH = 0. The resulting reduction energetics are strongly sensitive to both linker functionalization and pH. Halogenated and hydroxylated frameworks provide access primarily to HCOOH, CO, and HCHO under alkaline conditions, principally with the Mg and Zn nodes, whereas COOH functionalization offers the broadest thermodynamic accessibility across the full CO$_2$ reduction sequence. NH$_2$ retains thermodynamic feasibility for all target reduction pathways but with larger overpotentials, while NO$_2$ generally yields unfavorable reduction energetics. Crucially, within the COOH series, the choice of the metal node tunes the fundamental gap by over 1 eV with only minor changes in the reduction overpotentials, placing Sr- and Ba-based architectures as the most favorable ones for broad product selectivity with visible-light excitation. Linker functionalization substantially reduces the spatial overlap of the frontier states, promoting photoinduced charge separation. Taken together, these results establish linker functionalization and solution pH as complementary design levers for independently tuning light absorption and CO$_2$-reduction energetics in MOF-5, establishing a rational and viable route for designing efficient MOF-based photocatalysts.
Read the full abstract

All authors Julia Santana-Andreo, Joshua Edzards, Surender Kumar, Caterina Cocchi

Computer Vision

T2LSC-Bench: Benchmarking Localized Semantic Control in Text-to-Image Generation

Recent text-to-image models have become increasingly capable of rendering explicit text, but reliable localized text control requires more than generating the correct string. In applications such as product labeling, signage, and interface design, target text should be rendered within a designated text-bearing region without altering the predefined subject identity or surrounding scene semantics. We refer to violations of this requirement as target-text-associated semantic leakage, in which target-text semantics are expressed through non-textual visual content beyond the designated anchor. Existing visual-text benchmarks primarily evaluate readability, spelling accuracy, and layout, leaving this form of semantic leakage largely unexamined. We introduce T2LSC-Bench, a controlled diagnostic benchmark comprising 50 seed subjects and 1,200 prompt cases per model, yielding 7,160 evaluated images across six models. Its factorized design varies semantic relation, scene openness, prompt mode, and language. A dual-branch protocol combines OCR-VLM text verification with structured VLM semantic judgments to measure Text-at-Anchor Accuracy (TAA), Semantic Subject Preservation (SSP), Semantic Leakage Rate (SLR), and Conditional Semantic Leakage Rate (cSLR). Under stress-test conditions, SLR increases from 1.2% to 18.1% and cSLR from 1.3% to 18.2%, whereas TAA decreases only from 91.4% to 90.9%. Anti-leakage prompting reduces SLR from 16.6% to 8.4% without degrading rendering accuracy. Human validation on 420 images shows strong agreement between automatic and adjudicated annotations. These results show that accurate text rendering does not guarantee local containment of target-text semantics.
Read the full abstract

All authors Yan Wang, Xinyi Hou, Weiguo Lin, Junjun Si, Siwei Ma

Instrumentation and Methods for Astrophysics

Searching for Gravitational Wave Events Counterparts in X-Rays by Tilting the Multiple Shells of Wolter-I Optics

Current X-ray telescopes have a combination of field of view and angular resolution that makes it difficult to quickly localize the electromagnetic counterparts of gravitational wave events — which will become increasingly important when the LIGO and Virgo experiment will work at full sensitivity or when the Einstein Telescope and the Cosmic Explorer observatories will come online in the next decades. We investigate what X-ray optical designs could provide — in a small mission — a field of view wide enough to observe quickly the region of the sky constrained by gravitational wave detectors, and at the same time an angular resolution sufficiently good to localize these sources. We study, using ray tracing simulations, how we can optimize a Wolter-I telescope for this purpose. We find that, by appropriately tilting the shells of a Wolter-I X-ray telescope in different directions, the field of view increases, at the cost of a reduction of the on-axis performance. We draw a possible design made of two identical modules pointing at adjacent regions of the sky.
Read the full abstract

All authors John Rankin, Daniele Spiga, Paolo Conconi, Sergio Campana, Giovanni Pareschi, Stefano Basso, Marta Maria Civitani, Vincenzo Cotroneo

Artificial Intelligence

APEx: Distillation of Agent Procedural Experience for Adaptive Deep Research Question Answering

Deep research agents augment large language models with external tools to answer complex, long-horizon questions through multi-turn reasoning. Learning from prior experience is crucial for continual improvement, yet existing methods either retrieve verbose task-specific traces that burden decision-making, or distill procedural skills that remain decoupled from downstream policy adaptation. We propose APEx, a hierarchical experience utilization framework that organizes interaction history into instance-level trajectory memories and category-level procedural skills, and couples them through a closed-loop architecture of Executor, Distiller, and Planner. The three modules are optimized via a three-stage alternating GRPO training paradigm, enabling reward-guided skill distillation rather than fixed-prompt generation. At test time, distilled skills serve as procedural priors for online Planner adaptation through skill-guided test-time reinforcement learning, allowing ground-truth-free self-improvement with skill-alignment regularization to prevent policy drift. Experiments on 7 benchmarks demonstrate that APEx achieves state-of-the-art performance, surpassing GPT-5.4 by 14.7 points and the strongest memory-augmented baseline by 3.0 points.
Read the full abstract

All authors Jie Ding, Rui Sun, Xinyuan Zhang, Zeyu Zhang, Xin Liu

Robotics

DiffuSearch: How Hybrid Trajectory Planning Benefits from Aligned Objectives in Diffusion and Action Space

In trajectory planning for autonomous driving, hybrid planning architectures are often realized as a collection of disparate modules, each with its own objectives. This lack of a unifying principle can lead to inconsistencies between the initial and refined trajectory, resulting in suboptimal behavior. We address this by introducing DiffuSearch, a novel hybrid planner that uses a unified set of objectives across generation and refinement. Our model encourages all components to follow the same shared driving goals: collision avoidance, drivable area compliance, comfort, and progress. DiffuSearch employs a two-stage architecture. First, a guided diffusion model generates a scene-consistent, joint trajectory prediction, using our driving objectives as differentiable guidance functions to implicitly steer the denoising process. Second, a Monte Carlo Tree Search (MCTS) in a discretized action space performs an explicit, local refinement of this proposal, leveraging the same driving objectives as its reward function. This synergistic design leverages the diffusion model's strength in finding scene-consistent solutions combined with the explainable, constraint-aware refinement of MCTS. Experiments on nuPlan and interPlan reactive closed-loop benchmarks demonstrate that DiffuSearch achieves strong and often state-of-the-art performance, substantially reducing collisions and improving comfort, particularly in complex, interactive scenarios. Our ablation studies indicate that MCTS refinement is the main mechanism behind the gains, while sharing objectives between implicit guidance and explicit search provides further consistent improvements.
Read the full abstract

Computer Vision

Handwriting Trajectory Recovery via Autoregressive Ordered Stroke Instance Prediction

Handwriting trajectory recovery aims to infer the dynamic writing process hidden behind a static handwritten image. Since offline handwriting preserves only the final spatial ink pattern, temporal information such as stroke order, writing direction, and pen-tip motion is lost, making recovery inherently ambiguous. Existing learning-based methods often directly predict the complete character trajectory without explicitly exploiting the stroke-level organization of handwriting. We argue that recovering the writing process should follow the writing process itself. Accordingly, we propose a two-stage framework that first recovers ordered stroke instances and then reconstructs continuous within-stroke motion. The first stage integrates stroke extraction and stroke-order recovery through autoregressive ordered stroke prediction, while direction-related structural cues further support within-stroke trajectory generation. Experiments on Chinese handwriting show that the proposed ordered prediction is more effective than post-hoc stroke ordering. Even without trajectory simplification, our full-point model achieves numerically better results than those reported by all compared baselines, while a controlled analysis shows that trajectory sampling density substantially affects measured recovery performance. Additional experiments demonstrate generalization to unseen Chinese character categories and cross-language extensibility to English and Tamil handwriting.
Read the full abstract

All authors En-Guang Wang, Yan-Ming Zhang, Fei Yin, Cheng-Lin Liu

Multiagent Systems

RideSkill: A Hierarchical Algorithm for Generalized Ride Sharing with LLM-Driven Automatic Evolution

Ride-sharing, which allows multiple passengers with different origin-destination (OD) pairs to share a single vehicle, is a challenging operational problem, as it requires orders with different OD pairs to be efficiently bundled and assigned to vehicles under uncertain and varying scenarios. Although multi-agent reinforcement learning (MARL) solutions have achieved promising performance, they suffer from limited generalization (adapting to different environmental scenarios), low transferability (adapting to different platform objectives), and training difficulties in large-scale systems, such as the curse of dimensionality. Recently, motivated by the scaling of large language models (LLMs), several works have incorporated LLMs into ride-hailing systems, either by employing LLMs directly as decision-making agents or using them for automatic algorithm design. However, none of these approaches support vehicle sharing, which complicates the problem by expanding both the state and action spaces exponentially. Moreover, most of them require frequent LLM calls at inference time, making them infeasible for real-time deployment. To address these issues, we propose RideSkill, a hierarchical method for ride-sharing that leverages LLM-assisted automatic algorithmic design. RideSkill consists of a combiner that assigns appropriate skills to each vehicle from a learned skill repository, enabling adaptive dispatch under varying scenarios and objectives, and a repositioner that sequentially relocates idle vehicles to emerging regions, avoiding conflicts among vehicles. Crucially, the skill repository, combiner, and repositioner are all trained by an LLM-based automatic evolutionary method, eliminating the need for LLM calls during deployment and thus ensuring high real-time performance.
Read the full abstract

All authors Zijian Zhao, Sen Li, Xialiang Tong, Mingxuan Yuan

Combinatorics

Hamiltonicity of graphs of acyclic orientations and acyclic polynomials

We study the graph $\mathcal{AO}(G)$ of acyclic orientations of a graph $G$. Two acyclic orientations are adjacent in this graph if they disagree on the orientation of a single arc. In particular, we focus on the Hamiltonicity of the graphs $\mathcal{AO}(G)$. Using two methods of pattern lacing which generalize the zig-zag method of Brenner, Cardinal, McConville, Merino and Mütze, we characterize which multipaths are $\mathcal{AO}$-Hamiltonian. Moreover, we give a criterion for the gluing of a multipath on a given graph to preserve $\mathcal{AO}$-Hamiltonicity. Building towards an inductive certification of $\mathcal{AO}$-Hamiltonicity via the (open) ear decomposition of 2-connected graphs, we propose three ways of gluing several multipaths to a given graph. In addition, we define the acyclic polynomials to encapsulate both the number of acyclic orientations of a graph and the "parity problem" proposed by Savage, Squire and West: if $-1$ is not a root of the acyclic polynomial of $G$, then $G$ is not $\mathcal{AO}$-Hamiltonian. We explore numerous properties of the acyclic polynomials, proving that they are not instances of the famous Tutte-Whitney polynomials, but that they too exhibit a partial deletion-contraction phenomenon.
Read the full abstract

Software Engineering

From Prompting to Engineering: A Research Agenda for Prompt Engineering in Software Engineering

Prompt engineering is increasingly used across Software Engineering (SE) activities, including requirements analysis, coding, testing, documentation, repository analysis, and planning. Yet prompts and related instruction artifacts are often created and evolved through task-specific and informal practices, with limited support for their systematic evaluation, management, traceability, and governance. To examine how SE can contribute to the maturation of these practices, we organized a structured community discussion at the First International Workshop on Empirical Prompt Engineering for Software Engineering (PROMPT-SE), co-located with EASE 2026. Participants discussed current prompting practices, challenges to their adoption and evaluation, and future directions for integrating prompt engineering into software development. We synthesized these discussions into five areas: prompt artifacts and standardization; evaluation and benchmarking; lifecycle integration; human-AI collaboration and skills; and governance, privacy, and technical debt. Based on these areas, we outline a research agenda to move prompt engineering from predominantly ad hoc interactions toward more systematic, maintainable, evaluable, traceable, and governable SE practices.
Read the full abstract

All authors Vincenzo De Martino, Giovanna Broccia, Fabiano Pecorelli, Jennifer Horkoff, Riccardo Coppola, Antonino Ferraro, Quim Motger, Emma McKenzie, Shahbaz Siddeeq

Computer Vision

SAUF-Net: Structure–Appearance Representation Learning with Uncertainty Feedback for Semi-Supervised Medical Image Segmentation

Semi-supervised learning has shown great potential for reducing annotation costs in medical image segmentation. However, most existing methods mainly exploit unlabeled data through prediction-level consistency, while the reliability of internal feature representations is often overlooked. In medical images, target-related structural cues are easily entangled with unstable appearance variations, which may lead to unreliable pseudo labels and error accumulation during training. To address these issues, we propose SAUF-Net, a Structure–Appearance Representation Learning with Uncertainty Feedback Network for semi-supervised medical image segmentation. SAUF-Net uses the Structure–Appearance Decomposition Module (SADM) to separate bottleneck features into structural and appearance representations. The Disentangled Guidance Module (DGM) injects these representations into the decoding process to enhance structure-aware segmentation. Meanwhile, the Auxiliary Decoder produces branch-specific predictions for reliability estimation and a fused prediction for appearance-swapped consistency. Furthermore, we introduce an Appearance-Swapped Consistency branch to encourage structural representations to remain stable under appearance variations. We also introduce a reliability-map-guided dual-head discriminator with a Validity Head and an Uncertainty Head to provide feature-level uncertainty feedback. Extensive experiments on ISIC-2016 and Kvasir-SEG demonstrate that SAUF-Net outperforms state-of-the-art semi-supervised methods, especially under low-label settings.
Read the full abstract

All authors Qin Lu, Zheyang Jing, Yujie Yang, Jianwang Li, Chen Yi, Shaofeng Jiang

Artificial Intelligence

LLM-as-a-Judge Is Not an Oracle: Why Self-Improving Agents Need Deterministic Guardrails

Self-improving agent pipelines have a problem at their center. An optimizer rewrites prompts to score higher, and the score comes from a judge that is itself an LLM. That judge has the last word on whether the system is getting better, and our position is that it has not earned it. The judge should be demoted from oracle to advisor: its verdict becomes one input among several, and every change is gated instead by a deterministic verification layer the judge cannot override. We reached this position by building the alternative and running it. Over months of running autonomous prompt-optimization loops in production across contract analysis, compliance review, and code quality, we cataloged eleven ways the evaluation signal failed, in four classes: judge bias, harness and metric failures, ground-truth errors, and reward hacking. Agents achieved perfect scores by reading cached answer keys from their environment, a 100% pass rate concealing 68% true capability. A corrupted ground-truth label caused the optimizer to delete correct compliance rules to agree with it. A syntactically broken prompt was promoted as the winner because a silent parser fallback improved the metric. Attempts to fix the judge by rewriting its rubric plateaued; the only reliable gain came from a structural constraint on its output order. In response we describe PROCTOR, a Teacher-Student loop in which a stateful orchestrator holds all tool access, stateless subagents diagnose failures and draft mutations they cannot apply, and a Teacher grades those mutations under five deterministic guardrails: hermetic sandboxes, capability-disjoint roles, acceptance checks that outrank the Teacher, frozen holdouts, and canary cases engineered so that a perfect score is itself evidence of cheating. We report the failures this prevented, and, because the Teacher is itself an LLM judge, the failures it did not.
Read the full abstract

Astrophysics of Galaxies

An updated census of the Vast Polar Structure member galaxies

The Vast Polar Structure (VPOS) is a thin, planar arrangement of dwarf galaxies oriented almost perpendicular to the Milky Way (MW) disc. Since its discovery, the census of VPOS member galaxies has progressively expanded thanks to continuous findings of new dwarfs in the vicinity of the MW and increasingly precise measurements of their phase-space coordinates. In this work, we present the largest census to date of VPOS member galaxies, building on a sample of 73 MW dwarfs for which positional and kinematic data are available in the literature. Based on their orbital poles, we identify 22 VPOS members. We note that the majority of VPOS member galaxies are co-rotating, while only three systems are counter-rotating (Bootes III, Kim 1, and Sculptor). Our revised census increases the known VPOS member population by nearly 50%, introducing Delve 4, Kim 1, Kim 3, Eridanus III, Pictor II, and YMCA-1 as new on-plane dwarfs. From this updated census, we revise the structural parameters of the VPOS, confirming its planar ($c/a =0.27$) and remarkably thin (15 kpc) nature.
Read the full abstract

Artificial Intelligence

Task-Level Natural Language Priors as Learning Signals for Low-Resource LLM Training

Large language models (LLMs) often struggle when low-resource training data are ambiguous or incomplete. Task-level natural-language priors can provide useful guidance in such settings, but existing approaches usually treat these priors as input context rather than as learning signals during training. We study how task-level natural-language priors can supply guidance missing from these data through Prior-Guided Tuning (PGT). Its implementation, Contrastive Prior Steering (CPS), retains the supervised objective and adds positive- and negative-prior-conditioned losses with opposing optimization directions. Across AmbiMath, Diagnostic Anchoring, and Jigsaw, CPS achieves near-perfect exact-match accuracy on the synthetic task, improves accuracy under misleading diagnostic suggestions by 8.2 percentage points, and improves average subgroup Macro F1 on Jigsaw by 10.7 percentage points over plain fine-tuning, averaged over the two evaluated models. On Jigsaw, with one tenth of the experimental training data, CPS slightly exceeds full-data plain fine-tuning. These results show how reusable task-level language can complement limited labeled examples without requiring instance-level rationales or preference pairs.
Read the full abstract

All authors Jian Gao, Xiao Zhang, Xun Zhu, Miao Li, Ji Wu

Neurons and Cognition

Mus siliconus: A Neuro-Musculoskeletal Digital Twin of the Mouse Integrating Neural Dynamics, Biomechanics, and Tactile Sensing

Digital twin technologies could transform neuroscience and biomedicine by creating predictive computational representations of living organisms. However, most animal digital twins model neural circuits, anatomy, or biomechanics separately rather than integrating the processes that generate behavior. We argue that animal digital twins should instead be conceived as embodied dynamical systems that unify neural activity, body mechanics, sensory feedback, and environmental interactions. We propose a neuro-musculoskeletal digital twin of the mouse that combines multimodal anatomical reconstruction from X-ray CT, high-resolution white-light sections, and Scx-GFP imaging with biomechanical simulation, Bonhoeffer–van der Pol neural dynamics, and tactile feedback. This framework forms a closed sensorimotor loop in which behavior emerges through continuous interactions among the nervous system, musculoskeletal system, and environment. The Bonhoeffer–van der Pol model provides a computationally tractable dynamical foundation for large-scale simulation of these interactions. Neuro-musculoskeletal digital twins could provide a convergence point for computational neuroscience, biomechanics, artificial intelligence, and robotics. When coupled with adaptive learning and autonomous experimentation, they may develop from passive simulations into active scientific instruments that generate hypotheses, predict interventions, and guide experiments. Such embodied digital twins could advance the study of biological intelligence and support new adaptive biomedical and robotic systems.
Read the full abstract

Artificial Intelligence

Propose to Learn, Learn to Propose: Evaluability-Aware Assistance under Bounded Rationality

AI assistants often collaborate by proposing candidate edits, plans, or designs that users evaluate before adoption. Existing assistance methods focus on proposal quality or user-goal inference, often assuming that the user can reliably evaluate any proposal, which can fail in practice because of bounded rationality. We study evaluability-aware proposal planning, where proposals serve both as task interventions and as probes for learning latent preferences and evaluation constraints, where the resulting belief updates then guide later proposals. We formalise this setting as ProSE, a hidden-parameter sequential assistance problem, and instantiate it with a KL-regularised bounded-rational binary response model in which acceptance trades off value gain against a distance-dependent evaluability penalty. Analysing the planning consequence of this likelihood reveals that likely accepted proposals and informative probes need not coincide, which explains why planners that only pursue acceptance systematically underperform. We operationalise ProSE with ProSE-Plan, a depth-2 Bayes-adaptive planner that scores proposals by possible responses and response-induced posterior beliefs. In controlled graph simulations, ProSE-Plan improves over evaluability-unaware and myopic baselines when evaluation cost is the bottleneck, and a probe-commit ablation confirms that our approach selects informative proposals that simpler methods miss. Our results thus identify user evaluability as a planning-relevant dimension of AI assistance, complementary to generation quality and preference inference.
Read the full abstract

Machine Learning

Similarity-Aware Personalized Federated Learning in Heterogeneous Environments

Federated Learning (FL) allows decentralized clients to train models collaboratively while preserving data privacy. However, distribution mismatch across clients often leads to poor global generalization and degraded local client-level performance. In such scenarios, some of the clients with their local models trained solely on local data may perform better than the globally learnt model, thus nullifying the benefits of collaborative federated learning. To address this, we propose SAPE-FL (Similarity-Aware Personalized Federated Learning), a novel personalization framework that anchors each client's model to both the global model and a similarity-weighted peer averaged model. By incorporating dynamic, client-specific regularization based on both model similarity and output similarity, SAPE-FL adaptively balances global knowledge transfer and peer collaboration while filtering out dissimilar clients. This dual anchoring mitigates negative transfer and enhances robustness in heterogeneous settings. We theoretically analyze our algorithm establishing its convergence guarantees and empirically show that SAPE-FL outperforms state-of-the-art methods under high statistical heterogeneity and low client data regimes.
Read the full abstract

All authors Arun Kumar A V, Sunil Gupta, Dang Ngyuen, Bao Duong, Dat Phan Trong

Plasma Physics

Instabilities in Cylindrical Geometry Using the Minimalist Approach: Formalism and Rotational Instabilities

The minimalist approach for linear stability analysis is applied to fluids and magnetized ideal plasmas in cylindrical geometry. In this approach, the dispersion relation is obtained by integrating a single first-order differential equation - referred to as the principal equation - subject to appropriate boundary conditions. We first derive the principal equation for a general unperturbed state with radially varying density and pressure, axial and azimuthal components of both the velocity and magnetic field, and a radially directed gravitational field. We then use this formulation to analyze rotating flows with axial magnetic fields, addressing both wall-bounded and interface-driven axisymmetric instabilities. In addition to exact results for selected unperturbed states, we obtain approximate dispersion relations using the WKBJ method in the incompressible and compressible limits. The analysis encompasses centrifugal, magnetorotational, and buoyancy-driven instabilities as special cases, and it clarifies how compressibility modifies their stability properties.
Read the full abstract

Lattice Field Theory

Smeared spectral functions from lattice QCD: opportunities and challenges

A broad class of physical observables can be expressed in terms of smeared spectral functions. Important examples include the hadronic vacuum polarization contribution to the muon $g-2$, hadronic $\tau$ decays, inclusive semileptonic $B$ and $D$ decays, and amplitudes for rare processes such as $B\to K^{(*)}\ell^+\ell^-$. In recent years, a variety of methods have been proposed to reconstruct or constrain such spectral functions from Euclidean correlators and related inputs. However, the inverse problem is intrinsically ill-posed, making it difficult to obtain fully reliable, model-independent results with controlled uncertainties. In this contribution, I summarize recent developments, discuss the opportunities offered by smeared spectral observables, and highlight the main challenges that remain.
Read the full abstract

Distributed Computing

Federated Learning on the American Science Cloud using APPFL

The American Science Cloud (AmSC), established under the Genesis Mission of the U.S. Department of Energy (DOE), aims to integrate DOE high-performance computing systems, experimental facilities, and data resources into a single, coordinated, AI-driven discovery platform. AmSC's early services focus on curated artifacts, such as gated inference access to hosted models, experiment tracking, and function execution across computing facilities. However, what these services lack is a means to train a model across organizational boundaries where data cannot be centralized due to policy, privacy, or scale. This is, by definition, a use case for federated learning (FL) and a growing class of scientific AI. In this paper, we show that this gap can be bridged by deploying the orchestration logic of the Advanced Privacy-Preserving Federated Learning (APPFL) framework as a scalable cloud service on top of the primitives AmSC already provides: project-scoped authentication that supports secure and reliable federation membership, function execution that drives distributed training at each site, experiment tracking that records round-level performance, and finally, the model-hosting and inference infrastructure that can be leveraged to distribute the federated trained models to authorized participants. We argue that offering federated computing as an important AmSC service would unlock privacy-constrained scientific collaborations, enabling public-private partnerships in model building while exercising and enhancing the platform's own federated infrastructure.
Read the full abstract

All authors Zilinghan Li, Abhijit Chunduru, Harinarayan Krishnan, Eric Chagnon, Peter Nugent, Kibaek Kim, Ravi Madduri

Machine Learning

Recursive Value Learning for Long-Horizon Offline Goal-Conditioned RL

Scaling offline goal-conditioned reinforcement learning (GCRL) to long-horizon tasks is difficult because (1) long-range value learning depends on shorter-range estimates that may still be inaccurate, and (2) max-based value backups can amplify overestimation through repeated propagation. We propose DCRL (Divide-and-Conquer RL), which recursively decomposes each trajectory segment into a balanced binary tree and trains the values from leaves to root. Each parent is therefore updated only after its children, using an exact factorization of the observed route rather than selecting among noisy alternatives. Since this objective learns values along demonstrated routes that are not necessarily optimal, DCRL jointly propagates values across trajectories to discover shorter routes. Thanks to the balanced binary tree, DCRL reduces worst-case bootstrap depth from linear to logarithmic, and this shorter dependency structure empirically corresponds to much slower error accumulation. Across diverse goal-reaching tasks, DCRL substantially outperforms prior flat offline GCRL methods, and on the five most challenging long-horizon OGBench tasks, it improves the best prior average score from 55 to 64, surpassing all flat and hierarchical baselines.
Read the full abstract

Artificial Intelligence

PGPO: Potential-Guided Policy Optimization for Multi-Turn Agentic Tasks

Group-based reinforcement learning (RL) has become an effective paradigm for LLM post-training, but in multi-turn agentic tasks with sparse terminal rewards, it often provides coarse credit for intermediate actions. To obtain more fine-grained credit assignment, recent work such as GiGPO introduces step-level advantages for intermediate actions. However, these step-level signals still rely on the final outcome of each individual trajectory. As a result, actions within failed trajectories can remain poorly differentiated, so effective actions can receive the same unfavorable credit as erroneous ones. In this work, we propose Potential-Guided Policy Optimization (PGPO) for multi-turn agentic tasks. PGPO estimates empirical state potentials from anchor-state-group return statistics within each rollout group. It then derives action advantages from potential differences between adjacent states, enabling cross-trajectory credit propagation. This provides finer-grained step-level credit assignment, especially within failed trajectories. Experiments on ALFWorld and WebShop show strong overall performance relative to recent group-based RL methods. Further analysis provides evidence that PGPO yields more informative failure-side credit signals with negligible training overhead.
Read the full abstract

All authors Yuyao Zheng, Haipeng Sun, Junwei Bao, Lemao Liu, Hongfei Jiang, Yang Song, Dejing Dou

Instrumentation and Methods for Astrophysics

Scientific performances of the XGIS instrument on-board THESEUS

The Transient High Energy Sky and Early Universe Surveyor (THESEUS) satellite, currently undergoing a Phase-A study for the M7 ESA mission selection, will mainly rely on the XGIS (X and Gamma Imaging Spectrometer) instrument to discover and precisely localise gamma-ray bursts and other high-energy transients. The XGIS instrument has been designed to provide accurate sky images in a wide field of view (>2 steradians) in the 2-150 keV band, coupled to coverage of nearly half of the sky with good timing and spectroscopic capabilities up to several MeV. This will be achieved thanks to the combination of two identical cameras, based on the coded mask imaging technique, oriented to provide partially overlapping fields of view. We present the capabilities of the XGIS, focussing in particular on the sensitivity and source location accuracy. We discuss the expected scientific advances in the GRB field, estimated through detailed Monte-Carlo simulations which take into account the design of the XGIS imaging system and the properties of the GRB population at high redshifts.
Read the full abstract

All authors E. Arrigoni, S. Mereghetti, R. Campana, C. Labanti, A. Pisapia, P. Calabretto, G. Mattioli, E. Virgilli, L. Amati

Combinatorics

A linear bound for nested cycles without geometric crossings

Cycles $C_1,\ldots,C_k$ in a graph are called nested without geometric crossings if they are pairwise edge-disjoint, $V(C_k)\subseteq\cdots\subseteq V(C_1)$, and each pair of consecutive cycles induces the same cyclic order on the vertices of the inner cycle, up to reversal. Let $f_k(n)$ be the least number of edges that forces such a family in every $n$-vertex graph. Gil Fernández, Kim, Kim and Liu proved that $f_2(n)=O(n)$, answering a question of Erdős, and asked whether $f_k(n)=O_k(n)$ for every fixed $k$. We prove this for all $k$. The proof selects the inner cycles together with a disjoint subgraph that supplies their external neighbours. A reselection argument gives disjoint paths from every inner-cycle vertex to any sufficiently large target set. Sublinear expansion and a rooted clique minor then allow the vertices to be joined in the required cyclic order.
Read the full abstract

Computer Vision

InfraPatch: Cross-Task Targeted Grayscale Patch Attacks on Infrared-Adapted Vision-Language Models

Infrared vision-language models (IR-VLMs) have emerged as a promising paradigm for multimodal perception under low-visibility conditions, yet their robustness to targeted adversarial attacks remains poorly understood. Existing adversarial patch methods mainly study RGB-based models or a single downstream task and do not characterize whether localized perturbations can induce an intended semantic target in IR-VLMs. We propose InfraPatch, a white-box, per-instance framework for targeted digital grayscale patch attacks against IR-VLMs. InfraPatch optimizes a compact single-channel patch within an approximately 5% local-area budget, combines proxy-guided placement with task-adaptive semantic objectives, and induces target behaviors in image classification, image captioning, and binary visual question answering. We evaluate ten infrared-adapted model variants on 300 synthetic infrared-style images generated by applying DiffV2IR to a fixed 30-category COCO subset, using clean-conditioned targeted success criteria. InfraPatch achieves targeted attack success rates from 86.00% to 100% across the ten variants. On CLIP and BLIP-2, proxy location search improves success by 6.67 and 10.33 percentage points over optimized random placement, respectively; LLaVA-1.5 remains saturated near 100% under both settings. Patch-area and objective ablations further expose substantial differences in vulnerability across architectures and task formats. These results show that small grayscale patches can inject chosen target semantics across IR-VLM families under a controlled digital threat model, motivating stronger robustness evaluation for infrared multimodal systems.
Read the full abstract

All authors Chengyin Hu, Dingyi Lu, Jiaju Han, Xiang Chen, Weiwen Shi, Jiahuan Long, Yiwei Wei, Jiujiang Guo

Computers and Society

Fairness-Aware Multimodal Transformer Modeling for Real-Time Student Attention Estimation

Automated student-attention estimation can support learning analytics, but aggregate predictive metrics can conceal demographic disparities. This study evaluates fairness-aware multimodal temporal models on DIPSER, a naturalistic classroom dataset combining facial images, wearable-sensor measurements, attention annotations, and automatically inferred demographic metadata. Three baselines are compared across 10 training seeds: a Visual GRU, a Sensor GRU, and a Residual Fusion Transformer. The multimodal model achieves the best mean test performance (MAE 0.283, RMSE 0.363) and the lowest worst-group error among the evaluated baselines, although its gain over the Visual GRU is modest. Gender- and age-targeted MAE-gap regularization reduces disparities on validation data, but these gains do not consistently transfer to held-out subjects or repeated subject-level splits. On an NVIDIA A100-SXM4-40GB GPU, the warm end-to-end pipeline averages 50.65 ms per prediction window at a one-second stride, while the temporal model itself requires 1.02 ms. The findings show that multimodal fusion can modestly improve prediction and worst-group performance, but validation-level fairness gains should not be assumed to generalize. Robust fairness assessment therefore requires subgroup-aware evaluation, repeated subject-level validation, and larger, better balanced demographic samples.
Read the full abstract

Artificial Intelligence

PhoenixNest-Video: Evidence-Grounded Multimodal Agent Framework for Automated Video Interview Assessment

Interview assessment requires per-criterion judgments grounded in behavioral evidence, yet surging applicant volumes have made human-only evaluation costly and inconsistent, while existing AI approaches yield opaque scores without traceable rationale. We introduce PhoenixNest-Video, an evidence-grounded multimodal agent framework for automated video interview assessment. It builds a semantic video graph as structured working memory, performs rubric-conditioned retrieval with cross-modal verification across visual, audio, and textual streams, and produces per-criterion scores anchored to the candidate's materials. A Scorer trained via Rubrics-based Reinforcement Learning with dual rewards for rubric alignment and score-level differentiation internalizes the discriminative structure of multi-level rubrics. PhoenixNest-Video attains 91.50% grade-level accuracy on VInterview-2025, outperforming substantially larger proprietary models. A compact, rubric-grounded agent therefore scores candidates in closer agreement with an expert panel than direct prompting of much larger models, and exposes the evidence behind each score for human review.
Read the full abstract

All authors Fan Yuxuan, Huang Miaojun, Zhang Haimei, Wu Jingshen, Liu Hao

Astrophysics of Galaxies

Is the Nuclear Star Cluster a Bulge Fossil Fragment?

The Milky Way nuclear star cluster (NSC) is unlike a conventional globular cluster: it contains a broad metallicity distribution and stellar populations formed over an extended interval, while its detailed abundances show a close connection to the inner Galactic bulge. We ask whether the NSC should instead be compared with the proposed "bulge fossil fragments" Terzan 5 and Liller 1, complex stellar systems interpreted as surviving remnants of massive structures involved in early bulge assembly. We place this comparison against the field-bulge metallicity distribution measured by the Blanco DECam Bulge Survey and against high-resolution infrared abundance measurements in the NSC. A broad metallicity distribution by itself is not diagnostic, because the bulge field is intrinsically broad and strongly structured. More discriminating tests are the presence of chemically narrow and age-coherent subpopulations, the relation between [alpha/Fe] and [Fe/H], light-element abundance patterns, spatial segregation, and chemo-dynamical differences among populations. Existing NSC data establish strong chemical links with the bulge but do not support a simple one-to-one analogue of either Terzan 5 or Liller 1. In particular, metal-rich alpha-enhanced stars in the NSC provide an important contrast with the near-solar [alpha/Fe] metal-rich populations of the best-studied fossil-fragment candidates. We argue that the BFF hypothesis is most useful as a falsifiable framework for identifying an early bulge-building component within a composite NSC assembled through in-situ star formation, migration, and cluster infall.
Read the full abstract

Quantum Physics

A proposal for a hybrid free-space optical quantum communication network with hexagonal boron nitride-based single photon sources

Hexagonal boron nitride (hBN) is known as a promising solid-state platform to host room temperature quantum emitters that produce high purity single photons. The bright and spectrally adaptable hBN emitters are space-compatible, making hBN well-suited for satellite-based free-space optical (FSO) quantum key distribution (QKD) at wavelengths where the atmospheric background is naturally suppressed. This paper presents a pathway to develop hBN emitters operating near the Ca-II Fraunhofer line (at 854 nm), enabling daylight FSO operation. Moreover, due to its compatibility with the first telecommunication window at 850 nm, it is possible to interface with optical fibers to bridge the `last mile' in a scenario where multiple end-users connect through a single optical ground station to a QKD satellite. We therefore introduce a concept that combines continuous operation of a quantum network, hybrid links to minimize deployment costs, and high data rates due to the use of realistic single photon sources. The realization would be an important milestone for the development of the quantum internet.
Read the full abstract

All authors Julien Chénedé, Mostafa Abasifard, Tjorben Matthes, Asl{\i} Çakan, Tobias Vogl