Peer Revenue Publish or profit

Papers

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

All of arXiv

Friday 2 October

Materials Science

Optimizing the magnetic hyperthermia efficiency from the structural dependence of dipolar interactions in magnetic nanoparticles assemblies

Abdelhamid Morjane, François Vernay and Vincent Russier

Magnetic hyperthermia, based on the conversion of the magnetic energy of magnetic nanoparticles (MNP) under an alternating external field into heat, presents a great interest, especially because of important applications such as promising advances in cancer treatment. The underlying frequency-dependent response of the MNP assembly under the excitation field depends on the first hand on the individual properties of the MNP such as their size, shape and magnetocrystalline anisotropy and on the other hand on collective properties generated by interactions between MNP. Due to the long-range dipolar component of the latter, the first consequence is the so-called demagnetizing effects, making the response dependent on the external sample shape. This is the reason for studying the properties of MNP in assemblies of anisotropic external shapes such as films, chains, or elongated aggregates. In the present work, using a recently developed Time Quantified Monte Carlo (TQMC) framework for dynamic simulations, we focus more precisely on the importance of the local structure in such anisotropic aggregates to the dipolar interactions contribution of the response to the external field. Here the local structure is understood as both the location of the nanoparticles and the distribution of their easy axes. We show that the latter must be determined in a self-consistent way starting with the equilibrium structure of a ferrofluid modeled as the dipolar hard sphere fluid.
Read the full abstract

Artificial Intelligence

Relevant Evidence Decoding for Audio-Visual Hallucination Mitigation

Hyunjae Ra, Aecheon Jung, Jungin Park et al. (+1 more)

Audio-Visual Large Language Models (AV-LLMs) remain prone to cross-modal hallucinations, where one modality incorrectly affects predictions about another. Although contrastive decoding reduces hallucinations in vision-language models, its direct extension to AV-LLMs overlooks a key challenge: different questions require different perceptual evidence, including audio, video, or their interaction. Notably, we observe that joint audio-visual inference can weaken the prediction even when a model can recover the correct answer from a single informative modality. For example, when asked which instrument is heard, a model may correctly predict violin from the audio alone. Once a video showing a guitar is added, its confidence in violin may drop. In this paper, we introduce Relevant Evidence Decoding (RED), a training-free method that identifies question-relevant evidence and selectively strengthens its contribution. RED uses pointwise mutual information to quantify the predictive support provided by audio and video beyond the question alone. It decomposes their joint contribution into audio, video, and residual interaction components. A question-only inference pass determines the required evidence type, after which the model augments the original audio-visual prediction with the corresponding PMI contribution. Across three audio-visual hallucination benchmarks and three AV-LLMs, RED improves accuracy over standard decoding by up to 7% on CMM, 6.3% on AVHBench, and 3.8% on SVHalluc, with an average relative time to first token of 1.5x standard decoding.
Read the full abstract

All authors Hyunjae Ra, Aecheon Jung, Jungin Park, Sungeun Hong

Artificial Intelligence

Reliable Self-Evolution with Imperfect Proxy Rewards

Kangjun Noh, Soyu Kim and Kyungwoo Song

Large language model (LLM)-based self-evolving search is a promising approach to scientific discovery. However, high-fidelity evaluation of every candidate is prohibitively expensive in some domains. Self-evolving systems in such settings therefore rely on low-cost but imperfect proxy rewards, which may assign high scores to infeasible candidates. These false positives may contaminate both the final output and the feedback used to guide subsequent generations. This motivates statistically calibrated reward intervals for more reliable self-evolving search. We propose Conformal Interval-Driven Self-Evolution (CISE), which constructs candidate-specific reward intervals using conditional conformal inference and iteration-wise online density-ratio estimation. CISE uses conservative interval-based rewards for evolutionary feedback and returns candidates only when all required property intervals lie entirely within their respective feasible regions. We derive fixed-iteration coverage results under explicit assumptions of independence and covariate shift. We evaluate CISE on three self-evolving search tasks in materials science. In our experiments, all candidates returned by CISE are true positives under high-fidelity evaluation, whereas the baselines return more candidates but include false positives. These results highlight the value of a smaller, more precise shortlist when downstream validation budgets are limited. Our repository is available at https://github.com/MLAI-Yonsei/CISE.git.
Read the full abstract

Robotics

From Language Priors to Field Adaptation: Preference Learning for Traversability Estimation

Simon Schwaiger, David Seyser, Alessandro Scherl et al. (+3 more)

Image-based traversability estimation is inherently dependent on the robot platform, deployment domain, and mission preferences, which limits the applicability of purpose-trained models. To facilitate domain adaptation, this work aims to reduce the number of required annotations in the target domain using sample-efficient preference learning. Our method represents traversability through von Mises-Fisher mixture prototypes in a frozen vision-language feature space. Relative natural-language rules provide a commonsense prior, while sparse relative image annotations adapt the prototype directions and utilities to a target domain through computationally and sample-efficient fine-tuning. Experiments on WayFAST demonstrate accuracy competitive with end-to-end trained estimators while enabling sample-efficient image-based adaptation. Qualitative experiments further demonstrate the language prior's zero shot applicability and the fine-tuned estimator's improved dense prediction on semantic maps. Evaluation is complemented via semantic interpretation of learned prototypes by dissecting semantically close natural language prompts. Code and trained estimators available at https://resireg.github.io
Read the full abstract

All authors Simon Schwaiger, David Seyser, Alessandro Scherl, Zlatan Ajanović, Wilfried Wöber, Gerald Steinbauer-Wagner

Logic in Computer Science

An Applicative Multiset Path Order (Extended Version)

Nao Hirokawa, Teppei Saito, Teppei Tanaka et al. (+1 more)

We present a variant of the multiset path order for untyped applicative term rewriting. Compared to existing work, our variant incorporates two distinctive features, dubbed arity assignment and reification, to overcome difficulties in handling partial and variable application.
Read the full abstract

All authors Nao Hirokawa, Teppei Saito, Teppei Tanaka, Wataru Yachi

Artificial Intelligence

CreateScore: Domain-Theory-Informed Bayesian Routing for LLM-Based CV Screening

Rupsa Roy

Large language models (LLMs) can support rubric-based screening of CVs, but applying a high-capability model to every candidate and criterion is costly. We present CreateScore, a domain-theory-informed Bayesian network for criterion-level LLM routing. A hand-specified directed acyclic graph with Dirichlet-multinomial conditional probability tables converts CV evidence into posterior uncertainty; low-uncertainty decisions are resolved by a local 8B model and uncertain ones are escalated to a 120B reference model. The graph is causally motivated, but the system performs standard Bayesian conditioning, not causal inference. The escalation threshold is calibrated on a training fold (target: 70% resolved locally) and then frozen. On 200 synthetic Data Science CVs (139 training and 61 test candidates, five criteria), 77.7% of criterion decisions were resolved locally (237 of 305). Relative to a reference condition in which the 120B model adjudicated every criterion, routed escalation reduced token use by 65.2% and raised exact score agreement from 32.8% (8B alone) to 42.6% (95% CI 31.0-55.1%); at n = 61 the gain was not statistically distinguishable. The uncertainty signal did not, however, identify the decisions on which the 8B model erred: disagreement with the reference was 16.2% among escalated and 19.4% among locally resolved decisions (AUROC 0.47, 95% CI 0.39-0.56), no better than random selection. We also document how an earlier evaluation was invalidated when truncated reasoning-model outputs were silently replaced by local labels, and we recommend safeguards for cascade evaluation. CreateScore is supported as an auditable cost-reduction mechanism, not yet as a targeted error detector, and is not an autonomous hiring system.
Read the full abstract

Human-Computer Interaction

Tracking Human Daily Cognitive Activity from EEG and Biometric Data

Alina Gutoreva and Zhaniya Omar

Understanding human cognitive activity in everyday life remains challenging due to the dynamic, context-dependent, and multimodal nature of cognition. Laboratory-based studies often fail to capture real-world cognitive processes, while single-modality approaches provide only partial insight into cognitive states. Advances in wearable sensing now enable the collection of heterogeneous data streams for a more comprehensive view of daily cognition. This paper presents a multimodal framework for tracking human cognitive activity using electroencephalography (EEG), wearable physiological signals, behavioral context, and self-reported measures. A preliminary pilot study was conducted with observational data from three participants (N = 3) over 280 annotated 10-minute intervals spanning nine activity domains across two weeks. Results reveal consistent temporal patterns, including a discernible mid-day decrease in motivation and energy at 13:00, followed by afternoon recovery. Work and IADLs yielded the highest flow state rates (51% and 50%), while ADLs produced the lowest (15%). Motivation correlated strongly with arousal (r = 0.78) and attention (r = 0.74), whereas perceived stress showed a weaker negative relationship (r = -0.33). A linear regression model predicting motivation from arousal, attention, energy, and stress achieved R2 = 0.76 (MAE = 9.84, RMSE = 12.85). Lag-based analysis indicates that prior energy levels positively predict subsequent motivation, confirming temporal dependencies in cognitive dynamics. These findings demonstrate the feasibility of multimodal cognitive activity analysis in real-world environments and highlight the importance of integrating physiological and behavioral indicators. The proposed framework provides a foundation for future large-scale multimodal systems and applied intelligent solutions.
Read the full abstract

Computation and Language

A Guideline-Augmented Multi-Agent Framework for Schema-as-Code Biomedical Named Entity Recognition

Songtao Li, Shidi Zhang, Jianyuan Yuan et al. (+1 more)

Large language models (LLMs) have shown promising potential for biomedical named entity recognition (BioNER) through instruction following and in-context learning. However, existing LLM-based BioNER methods still face two key limitations. First, retrieved demonstrations and external biomedical knowledge provide limited support for dataset-specific annotation semantics, leaving entity boundaries, type scopes, and annotation conventions ambiguous. Second, free-form generation lacks sufficient structural control, often leading to invalid formats, hallucinated mentions, duplicated entities, and boundary errors. To address these limitations, we propose GAMA, a guideline-augmented multi-agent framework for schema-as-code BioNER. GAMA first induces candidate annotation rules from labeled training instances and verifies them against annotated data to construct reliable dataset-specific guideline memory. Guided by these verified rules, a planning component generates ranked span-type hypotheses with rationales, and a coding component converts them into schema-constrained entity objects. A verification module then checks span grounding, type validity, and structural compliance, and performs dual-loop refinement to correct invalid or low-confidence predictions. Experiments on three widely used BioNER datasets using two different LLMs demonstrate the effectiveness of GAMA, achieving the best performance in four out of six LLM-dataset combinations. Ablation studies further validate the contributions of its key components.
Read the full abstract

All authors Songtao Li, Shidi Zhang, Jianyuan Yuan, Fengyu Zhang

Systems and Control

Profile-Aware Trustworthy Recipe Generation with Planner-Critic Agentic Remediation

Shanhong Liu, Pai Chet Ng and Konstantinos N. Plataniotis

Recipe generation from food images has practical value for intelligent cooking assistance, but traditional one-pass generation often overlooks user-specific safety requirements such as allergies, dietary restrictions, and preparation constraints. We propose PCAR, a Planner-Critic Agentic Remediation framework for trustworthy recipe generation. PCAR separates recipe planning from safety verification: a Planner Agent extracts ingredients and generates recipe drafts conditioned on the user profile, while a Safety Critic Agent audits each draft and provides structured feedback for remediation when violations are detected. This remediation loop enables unsafe recipes to be revised rather than directly returned or discarded. We evaluate PCAR on real food images with 100 benchmark user profiles across four backbone models, including proprietary and locally served open-source models. Results show that PCAR achieves strong safety and generation performance with capable backbone models, while preserving practical recipe quality.
Read the full abstract

Artificial Intelligence

Reasoning with Evidence, Not Merely Rationales: Verifiable Preference Proofs for LLM-Based Recommendation

Yu Hou, Nathaniel Kang, Pengkai Wang et al. (+1 more)

Large language models (LLMs) can infer user preferences from interaction histories and reviews, yet the rationales they generate may not reflect the information actually used for recommendation. A preference claim may be weakly supported by its selected evidence, or may have little effect on the final ranking. We refer to these two failures as the grounding-influence gap. We introduce PROVE-REC, a general framework for verifiable preference reasoning in LLM-based recommendation. Pass A converts the complete pre-target history into a compact preference proof consisting of positive and avoidance claims linked to selected evidence entries. Pass B predicts the next item using only the proof and its selected evidence, preventing the recommender from bypassing the reasoning path. To verify evidence-to-proof grounding, we compare the effect of masking selected evidence with masking a comparable control entry. To verify proof-to-recommendation influence, we remove a preference claim and measure the resulting decrease in the target item's ranking margin. A ranking-preservation objective further retains useful information from the complete history. Comprehensive experiments on wide-ranging real-world datasets demonstrate that PROVE-REC consistently outperforms strong sequential, generative, and LLM-enhanced baselines, with improvements of up to 7.45%. Controlled ablations confirm the effectiveness of the two-pass architecture and verification objectives. Moreover, PROVE-REC produces claims that are more strongly grounded in historical evidence and more influential to recommendation while preserving ranking quality.
Read the full abstract

All authors Yu Hou, Nathaniel Kang, Pengkai Wang, Hua Li

Computer Vision

Post-Training Frontier Text-to-Image Models by Composing Preference and Rubric Rewards

Yuanhao Ban, I-Hung Hsu, Anastasios Angelopoulos et al. (+3 more)

Recent text-to-image generation models have achieved remarkable visual quality, but improving them through post-training remains challenging because no single reward signal captures the full range of human preference. In this work, we develop a simple and effective post-training recipe for open-domain text-to-image generation based on the composition of complementary reward signals. Our reward system consists of two main components: a preference reward, trained on large-scale human preference data using a Bradley-Terry objective to capture overall human aesthetic and perceptual preferences, and rubric-based rewards, which explicitly evaluate prompt faithfulness and other desirable properties while providing safeguards against reward hacking. A key challenge is how to combine these heterogeneous reward signals. We show that a naive weighted average leads to suboptimal optimization behavior, and propose a simple reward composition strategy that more effectively balances preference optimization with rubric satisfaction. In the Arena text-to-image leaderboard (https://arena.ai/), our RL-trained Flux2dev achieves an Elo rating 69 points above the base model, and our post-trained Ideogram-4 surpasses every open-source model on the leaderboard, reaching an Elo of 1223.5. (Claims of state-of-the-art performance are based on the Arena leaderboard snapshot as of September 4, 2026.) Our results suggest that effective rewards for frontier generative-model training require broad coverage of user intent and robustness to exploitation under optimization. To support reproducible research, we release Arena-T2I-Training, a 1K subset of training data that recovers some gains of full-scale training, providing a resource that we hope will facilitate future work on post-training for text-to-image models.
Read the full abstract

All authors Yuanhao Ban, I-Hung Hsu, Anastasios Angelopoulos, Wei-Lin Chiang, Ion Stoica, Cho-Jui Hsieh

High Energy Astrophysical Phenomena

XMM-Newton Resolves Parsec-scale X-ray Jets in the PeVatron Microquasar V4641 Sgr

Naomi Tsuji, Caterina Tresoldi, Kaya Mori et al. (+34 more)

The microquasar V4641 Sagittarii (V4641 Sgr) has recently been established as one of the most powerful Galactic PeV cosmic-ray accelerators (PeVatrons) through the detection of degree-scale gamma-ray emission to 0.8 PeV. Using XMM-Newton, we performed an X-ray survey of its gamma-ray-emitting regions during the quiescent phase. We discovered extended X-ray emission to the north and south of V4641 Sgr, aligned with the radio-jet axis, revealing X-ray jet structures extending over several tens of parsecs for the first time. The northern and southern knot-like features, dubbed n1 and s1, are located approximately 20 pc from the central binary and spatially coincident with the gamma-ray emission. The X-ray emission from each knot, characterized by a photon index of $\Gamma_X = 1.9 \pm 0.1_{\rm stat} \pm 0.5_{\rm syst}$ and an energy flux of $\sim$(2–4)$\times10^{-13}$~erg~cm$^{-2}$~s$^{-1}$ in the 2–10 keV band, is likely non-thermal in origin, indicating synchrotron radiation from accelerated electrons. The morphological coincidence between the X-ray and gamma-ray emission favors a leptonic scenario for the gamma rays, although the X-ray emission is more compact than the gamma-ray-emitting region and the hadronic scenario cannot be excluded. Through X-ray to gamma-ray modeling, we obtain a weak magnetic field strength of 3–5~$\mu$G. We propose that the newly detected X-ray knots represent particle-acceleration sites, such as jet termination or (re-)collimation regions powered by V4641 Sgr, or downstream structures analogous to the knots of the outer jets in SS~433/W50. This discovery suggests that extended X-ray emission along jets would be a key signature, yet previously overlooked, in PeVatron microquasars.
Read the full abstract

All authors Naomi Tsuji, Caterina Tresoldi, Kaya Mori, Hiromasa Suzuki, Matthew Lundy, Dmitry Khangulyan, Samar Safi-Harb, Laura Olivera-Nieto, Felix Aharonian, Shotaro Abe, Zhen Cao, Sabrina Casanova, Maria Chernyakova, Yi-Heng Chi, Stephane Corbel, E. de la Fuente, Hua Feng, Tatsuki Fujiwara, Noa Grollimund, Dezhi Huang, Yoshiyuki Inoue, Ryohei Kanazashi, Shigeo S. Kimura, Marianne Lemoine-Goumard, Jonathan Mackey, Alessio Marino, Samaresh Mondal, Kai Morita, Maxime Parra, Gabriele Ponti, Brian Reville, Megumi Shidatsu, Shunya Takekawa, Michelle Tsirou, Xiaojie Wang, Heshou Zhang, Ping Zhou

Symplectic Geometry

Two degenerate intersections of Clifford tori in $S^2\times S^2$ and $\mathbb{C}P^2$

Wenmin Gong

We construct smooth time-periodic Hamiltonians on the monotone manifolds $S^2\times S^2$ and $\mathbb{C}P^2$ for which the Clifford torus meets its Hamiltonian image in exactly two isolated degenerate points. The product example disproves the degenerate Arnol'd-Givental conjecture.
Read the full abstract

High Energy Physics (Theory)

Ultraviolet dissipation and defect OPE in bottom-up holographic interfaces

Mianqi Wang

We study the radial ultraviolet dissipation quantity $c_{UV}$ defined in arXiv:2601.16888v3 and defect OPE data in bottom-up holographic interfaces. We relate $c_{UV}$ along with the defect primary spectrum and BOPE coefficients in the large defect dimension limit to the optical length $\rho$ in the holographic bulk. In addition, we prove a new holographic bound for conformal interfaces relating entanglement entropy, energy transmission and UV dissipation: \[ c_{LR}\geq \frac{2c_{\mathrm{eff}}}{2+\pi(c/c_{\mathrm{UV}}-1)}. \] This compliments the upper bound of $c_{LR}$ by $c_\text{eff}$ proved in arXiv:2404.01515.
Read the full abstract

Optimization and Control

An Adaptive Symplectic Proximal Point Algorithm: Faster Convergence and Improved Numerical Performance

Yi Zhang

The proximal point algorithm (PPA) stands as a fundamental approach for solving monotone inclusion problems. Notably, several key convex optimization algorithms have been proven to be specific instances of PPA. Given the importance of the PPA, there has been growing interest in developing its accelerated variants. However, for some specific cases, the PPA converges faster than the accelerated PPAs. In this paper, we mainly study an adaptive version of symplectic accelerated PPA, called adaptive symplectic proximal point algorithm (ASPPA). We first prove that the convergence rate of ASPPA with respect to the square norm term is $O(1/k^2)$, which is the same as the convergence rate of some accelerated PPAs. Also, we show that ASPPA exists exponential convergence rate when solving strongly monotone inclusion problem. Our numerical experiments indicate that the performance of ASPPA is very good.
Read the full abstract

High Energy Astrophysical Phenomena

Candidate X-ray Counterparts of Ultra-High-Energy Emission of the Galactic Microquasar V4641 Sgr Revealed by Einstein Probe

Yi-Heng Chi, Shao-Qiang Xi, Ping Zhou et al. (+20 more)

Microquasars have emerged as promising PeVatron candidates capable of accelerating particles to PeV energies, with recent observations revealing ultra-high-energy (UHE) $\gamma$-ray emission from extended structures associated with these systems. Multi-wavelength studies of such extended structures are crucial for understanding the acceleration, transport, and radiation processes of UHE cosmic rays (CRs), yet observational evidence remains limited. Here, we report the discovery of diffuse X-ray structures spatially coincident with the TeV $\gamma$-ray emission regions to the north and south of the microquasar V4641 Sgr, using the Follow-up X-ray Telescope (FXT) onboard the Einstein Probe. We identify X-ray structures located $\sim20$-$40$ pc from V4641 Sgr, extending over several arcminutes. These structures exhibit X-ray fluxes of $\sim1$-$3\times10^{-13}$ erg s$^{-1}$ cm$^{-2}$ and photon indices of $\Gamma\sim2$-3 in 0.5-8 keV per individual knot. Under the one-zone leptonic interpretation, the weak X-ray emission implies a magnetic field of order $1-2 \mu$G. Stronger fields remain possible if only part of the TeV-PeV emission is produced by inverse Compton scattering, with the remaining emission arising from hadronic interactions. The EP-FXT observations support the presence of relativistic electrons on tens-of-parsec scales, while the origin of the UHE $\gamma$-ray emission remains open.
Read the full abstract

All authors Yi-Heng Chi, Shao-Qiang Xi, Ping Zhou, Hua Feng, Ruo-Yu Liu, Songzhan Chen, Naomi Tsuji, Caterina Tresoldi, Gabriele Ponti, Zhen Cao, Yong Chen, Shumei Jia, Chi-Chuan Jin, Chengkui Li, Yuan Liu, Matthew Lundy, Kaya Mori, Haiwu Pan, Lian Tao, Junfeng Wang, Weimin Yuan, Juan Zhang, Haisheng Zhao

Number Theory

Some arithmetical aspects of the two dimensional Jacobian Conjecture

Ronen Peretz

We prove that any counter-example to the two dimensional Jacobian Conjecture with integral coefficients and determinant of its Jacobian matrix equals one, has only paths of asymptotic values that can contain only finitely many points of the planar integral lattice. The second part of our paper gives some nontrivial applications of the theory of certain Theta functions to the two dimensional Jacobian Conjecture over the real field, and over the complex field.
Read the full abstract

Machine Learning

Dirac-Interconnected Neural Elements: Discovering Modularity in Physical Systems Without Reduction

Reiho Li, Razmik Arman Khosrovian, Takaharu Yaguchi et al. (+2 more)

Deep learning has shown remarkable success in the data-driven modeling of dynamical systems. Much of its success is attributed not to the flexibility of neural networks but to inductive biases based on physical prior knowledge, such as energy conservation and symplecticity. However, existing methods do not fully exploit the fact that real-world physical systems are interconnections of components. Some methods require the interconnection to be known a priori, while others assume the system to be reducible to an ordinary differential equation (ODE) and learn only the reduced ODE, discarding the algebraic constraints imposed by the interconnection. Here, we propose Dirac-interconnected neural elements (DINEs), a neural network model that represents a physical system as a differential-algebraic equation (DAE), whose algebraic constraints are given by a Dirac structure in kernel representation. With DINEs, we simultaneously identify from data the interconnection among the components as a Dirac structure and learn the characteristics of the components as neural networks. This allows us to keep the learned subsystems in unreduced form and isolate or compose them to make a new system without retraining. Moreover, DINEs can handle partially observable systems. Experimental results demonstrate these capabilities on physical systems beyond the reach of existing methods.
Read the full abstract

All authors Reiho Li, Razmik Arman Khosrovian, Takaharu Yaguchi, Hiroaki Yoshimura, Takashi Matsubara

Computer Vision

TerraVis: Towards Evaluation of World-Grounded Visual Consistency in Text-to-Image Generation via MLLM Workflows

Shuai Fu, Jing Gu, Jian Zhou et al. (+3 more)

Recent text-to-image models have made substantial progress in photorealism, aesthetics, and text-image alignment. Yet visually appealing images can still violate real-world plausibility, exhibiting malformed object structures, impossible anatomy, physically implausible interactions, or inconsistent spatial relationships. Such failures are not well captured by existing fidelity, aesthetics, preference, or alignment metrics. To address this gap, we introduce TerraVis, a framework for evaluating world-grounded visual consistency in generated images. TerraVis defines a structured taxonomy of world-consistency violations spanning object-, interaction-, and scene-level failures, and employs a multi-stage evaluation framework to identify and quantify them. Given an image, TerraVis first uses an MLLM to assess its eligibility for evaluation, then detects violations across 18 taxonomy-defined types and classifies them as minor or major to derive an overall world-consistency score. Across diverse open-source and proprietary text-to-image models on two widely used benchmarks, TerraVis achieves the strongest correlation with human judgments of world consistency among existing metrics. Our benchmark results further show that models that achieve strong performance on conventional metrics can still exhibit substantial world-consistency failures. These findings highlight world consistency as a complementary evaluation dimension and demonstrate that TerraVis enables systematic quantification, diagnosis, and comparison of such failures. Our code is publicly available at https://github.com/ShyFoo/TerraVis.
Read the full abstract

All authors Shuai Fu, Jing Gu, Jian Zhou, Zicheng Duan, Gengze Zhou, Qi Wu

Plasma Physics

Numerical and diagnostic extensions to the gyrokinetic $\delta f$ code stella

Sven Stroteich, Georgia Acton, Michael Barnes et al. (+2 more)

Understanding turbulence in magnetised plasmas requires robust numerical tools capable of capturing complex free-energy transfer processes across scales. We demonstrate the utility of free energy as a diagnostic for both physical turbulence characteristics and numerical stability in gyrokinetic simulations with the stella code. Applied to the W7-X high-mirror configuration, free-energy diagnostics reveal characteristic power-law behaviour in the turbulent cascade, act as sensitive indicators of numerical instabilities undetected by conventional diagnostics, and expose localised structures in real space not apparent in the electrostatic potential alone. To support these analyses, we present a systematic comparison of explicit time-integration schemes, showing that the optimal choice depends critically on the regime: SSP22 achieves the lowest computational cost in linear ITG simulations, while SSP33 is most efficient in the nonlinear saturated state, with nonlinear simulations requiring two to three orders of magnitude more computational cost than linear cases. These results establish free-energy diagnostics as a valuable tool for interpreting plasma turbulence and provide practical guidance for scheme selection in stella simulations.
Read the full abstract

All authors Sven Stroteich, Georgia Acton, Michael Barnes, Pavlos Xanthopoulos, Ralf Schneider

Machine Learning

Understanding Trajectory Heterogeneity in Federated World Model Learning

Yipan Wei, Zhaokun Yan, Ziming Hong et al. (+2 more)

World models learn state evolution from trajectories, making access to temporal context a central training requirement. Federated learning can use distributed records, while ownership boundaries within a trajectory restrict the examples each client can construct. Our study benchmarks this cross-time setting through hourly action-conditioned clinical prediction on eight MIMIC-IV disease cohorts, comprising 40.87 million transition memberships. We specify severity-based client ownership, patient-separated construction, local history and future-window rules, and paired rollout evaluation from one to 32 hours. A matrix of ten federated algorithms covers 32 disease–partition configurations under five rounds of ten-percent participation. Three findings emerge from existing results and training logs. First, client ownership and participation jointly restrict long-window coverage: only 7.55%–21.36% of pooled-available 32-step windows have a locally complete anchor visited during training, averaged across diseases. Second, finer severity partitions accompany higher FedAvg error in 15 of 16 paired comparisons, while algorithm gains are small and horizon-dependent: FedProx reduces mean error by 0.56%, with no consistent improvement at 32 steps. Third, algorithm labels conceal distinct update behavior, including inactive extrapolation and orders-of-magnitude differences in update scale. Cached-update performance also varies strongly across trajectory partitions under the same benchmark protocol. These results establish temporal access, participation coverage, optimization behavior, and horizon-resolved prediction as complementary dimensions for evaluating federated clinical world models.
Read the full abstract

All authors Yipan Wei, Zhaokun Yan, Ziming Hong, Jiaqi Wu, Lixu Wang

Solar and Stellar Astrophysics

Plato benchmark stars I. The initial benchmark sample

T. Merle, P. F. L. Maxted, O. L. Creevey et al. (+23 more)

The ESA PLATO mission aims to detect and characterise Earth-size exoplanets orbiting bright, late-type stars using high-precision photometry and asteroseismology. Precise stellar parameters of host stars are essential for exoplanet characterisation and model benchmarking. This work defines the PLATO benchmark stars (PBS) catalogue, a comprehensive sample of well-characterised stars designed to support all mission phases. We compiled data from eclipsing and spectroscopic binaries, interferometry, asteroseismology, and large spectroscopic and photometric surveys. Fundamental and atmospheric stellar properties (mass, radius, luminosity, age, effective temperature, surface gravity, density, and limb darkening) were assigned quality levels (B1, B2, B3) reflecting their degree of model dependence. Homogenized ancillary parameters (including Gaia DR3 bias-corrected parallaxes, metallicities, bolometric fluxes, extinction, rotation periods, and X-ray fluxes for M dwarfs) are also provided. The PBS catalogue contains 1090 FGKM main-sequence and subgiant stars, divided into three quality tiers based on model-independent (B1) parameters: gold (15 stars with B1 mass, radius, and luminosity), silver (195 with B1 mass and radius), and bronze (416 with at least one B1 property). The remaining stars feature well-characterised model-dependent (B2) or empirical (B3) properties. About 40% of the sample lies within the PLATO fields of view (285 in LOPN1 and 143 in LOPS2). The PBS sample represents the largest catalogue of benchmark stars assembled to date, providing a key resource for validating automated pipelines, testing evolutionary models, and cross-survey calibration. The catalogue is available online at https://astro.ulb.ac.be/pbs and will be updated regularly.
Read the full abstract

All authors T. Merle, P. F. L. Maxted, O. L. Creevey, U. Heiter, D. B. Palakkatharappi, K. G. Hełminiak, C. C. Aguilera Gómez, M. Bazot, S. Cassisi, N. Ebrahimkutty, R. E. Giribaldi, P. Jofré, F. Kiefer, M. N. Lund, N. Miller, T. Morel, D. Mourard, N. Nardetto, J.-D. do Nascimento, Jr., T. Olander, G. N. Pawar, A. Rebassa-Mansergas, R. Raddi, B. Rojas-Ayala, J. Southworth

Cryptography and Security

Digital Twin-Assisted Mapping of ICS Telemetry to ATT&CK for ICS with Evidence-Driven Dependency Reasoning

Konstantinos E. Kampourakis, Vyron Kampourakis, Vasileios Gkioulos et al. (+1 more)

Reconstructing adversarial behavior from Industrial Control System (ICS) telemetry is difficult because process observations reveal physical changes more directly than the actions that produced them. This paper presents a Digital Twin (DT)-assisted framework that extracts synchronized state changes, converts them into evidence-preserving descriptions, maps them to ATT&CK for ICS through retrieval-augmented Large Language Model (LLM) reasoning, and constructs a typed dependency graph. Evaluation comprises a four-configuration ablation on nine held-out SWaT scenarios containing ten telemetry-evaluable ground-truth episodes, three independent generations of the principal mapping configurations, and external evaluation on BATADAL and WADI. Across the three SWaT generations, DT-enriched mapping produces fewer annotation-relative False Positive (FP) episodes in every run, with a mean per-run reduction of 34.1%. Both configurations obtain a mean recall of 0.633, although recall varies between 0.50 and 0.70 and DT enrichment does not improve F1 in every run. These observations are descriptive: the primary-run paired comparisons do not reach statistical significance, and the mapping advantage does not transfer to either external dataset. An exploratory positive-only dependency benchmark recovers eight of nine documented Co-occurrence relationships with DT context. A separate end-to-end benchmark containing one positive pair and 25 negative controls exposes propagation of mapping errors into unsupported edges. The findings identify both the potential and limitations of DT context for semantic security interpretation, without establishing reliable autonomous attribution, general causal reconstruction, or practical analyst benefit.
Read the full abstract

All authors Konstantinos E. Kampourakis, Vyron Kampourakis, Vasileios Gkioulos, Sokratis Katsikas

Machine Learning

Hyperparameter selection for equation learning with biologically-informed neural networks

William Lavery, Jodie A. Cochrane, John T. Nardini et al. (+1 more)

Biologically-informed neural networks (BINNs) have emerged as a flexible subclass of physics-informed neural networks (PINNs) for learning terms in partial differential equations from data. BINNs are particularly suited for biological systems, where the governing equations are highly nonlinear and only partially known a priori, and where data observations are often sparse, noisy, and incomplete. However, applying BINNs effectively in practice depends critically on hyperparameter selection, which remains a central challenge in equation-learning frameworks. Hyperparameters are often chosen heuristically and only cursorily documented, which limits the reproducibility of results and the transferability of methods. We present a diagnostic workflow for hyperparameter selection that can be used when the ground-truth equations are not known. The workflow is guided by three main questions: (1) Are the benefits of greater network capacity worth the cost? (2) Do more training epochs keep reducing the validation loss? (3) Do the learned terms stop changing as network capacity and training increase? We apply our workflow to synthetic systems of varying complexity with known ground truth, spanning diffusion and growth right-hand side terms and data ranging from 1D+t to 2D+t. We demonstrate that the validation loss generally follows the true error in the learned terms and distil practical rules of thumb for selecting hyperparameters in the BINN architecture. By providing a structured workflow, practical guidelines, and suggested starting values for hyperparameter selection, this work lowers the barrier to BINN-based equation learning.
Read the full abstract

All authors William Lavery, Jodie A. Cochrane, John T. Nardini, Sara Hamis

Machine Learning

SlimKV: Joint Token-Feature KV Cache Compression with Reconstruction-Free Beacon Attention

Zihan Teng, Jiayu Zhao, Wentao Ren et al. (+4 more)

Long-context LLM serving is increasingly bottlenecked by KV-cache memory, especially in resource-constrained scenarios. Among existing KV-cache compression strategies, token-wise methods reduce cached states but risk information loss through eviction or condensation, while feature-wise methods reduce per-token KV dimensions but can require full-dimensional reconstruction to apply positional embedding, limiting decoding speedups. We introduce SlimKV, a question-agnostic joint token-feature KV-cache compression method. SlimKV uses low-rank-aware training to compress long contexts into beacon memory states with latent KV representations, together with layer-adaptive rank allocation. We further uncover a positional asymmetry: removing key-side RoPE affects beacon and raw tokens differently, with much smaller degradation for beacon tokens. Exploiting this asymmetry, SlimKV trains beacon KV projections under a K-RoPE-free constraint and enables latent-space attention during decoding, mitigating reconstruction latency. On LongBench, SlimKV outperforms baselines at 16x/32x compression and remains leading at 4x/8x, where it retains over 96% of the uncompressed model's score. Needle-in-a-Haystack confirms robustness across evidence positions, and efficiency evaluation shows up to 7.34x attention speedup and 3.38x end-to-end decoding speedup over the uncompressed model at 128K length.
Read the full abstract

All authors Zihan Teng, Jiayu Zhao, Wentao Ren, Minhao Fan, Tianrui Ma, Song Chen, Weichen Liu

Software Engineering

GTDD: Generative Test-Driven Development for AI Coding Agents with Adversarial Testing

Masahiro Kato

Test-driven development gives AI coding agents executable requirements for implementing software. Because these agents can adapt their implementations to the examples they observe, passing a predetermined collection of tests can leave substantial parts of the intended behavior unimplemented. We propose Generative Test-Driven Development (GTDD), a formulation of test-driven development in which a separate testing agent generates new inputs after each candidate implementation is fixed, using a human-specified behavioral contract and the feedback from earlier rounds. A trusted evaluator checks these inputs, returns reduced counterexamples to the coding agent, and saves them for regression testing, so development continually confronts failures beyond the initial examples. We characterize the evidence that this process provides through a finite-population analysis of false acceptance under adaptive candidate selection. The resulting bounds quantify how test visibility and repeated feedback affect acceptance, and show that fresh random audits after candidate commitment control false acceptance across development rounds. In a paired experiment on a stateful key-value store, both policies that regenerated tests during development ended with lower mean failure rates than the policy whose tests were generated once by the same language model, and giving the tester the candidate's source produced no detectable additional improvement. Further conditions requesting equal numbers of tests did not isolate any single feature of the policies as the source of this difference. GTDD combines this adaptive development feedback with established regression tests and an independent acceptance rule.
Read the full abstract

Machine Learning

Dynamic Expert Pruning for Multi-Agent Systems

Jabin Koo, Soheil Abbasloo, Sungjae Lee et al. (+1 more)

Mixture-of-Experts (MoE) architectures scale language models efficiently by activating only a few experts per token, but the saving is confined to computation: every expert must stay resident on the accelerator, so memory bounds where these models can be deployed. Expert pruning reduces this footprint, yet existing methods are static — a single mask, calibrated offline, is applied to the model for every subsequent request. This assumption can fail when the workload is heterogeneous, most prominently in multi-agent systems, where one backbone serves many tasks and roles at once: our analysis shows that different tasks and roles recruit different experts, while static methods assign one fixed subset to all of them. We therefore propose Dynamic Expert Pruning (DEP), which rests on a finding we establish here: an agent's system and task prompts are by themselves sufficient to identify the experts that agent and its task require, since that text already describes what the agent will do. A lightweight predictor, trained once on workflow transcripts, turns those prompts into a specialized per-request mask in a single forward pass, with no per-configuration calibration. Across diverse tasks and roles, model scales, and MoE architectures, DEP achieves better overall accuracy than static pruning and merging baselines, and generalizes to workflows unseen in training without retraining. Its margin over those baselines is largest when few experts are retained, suggesting that the role specialization inherent to multi-agent systems permits sparser serving than static pruning allows.
Read the full abstract

All authors Jabin Koo, Soheil Abbasloo, Sungjae Lee, Jungseul Ok

Systems and Control

Memory-Dependent Interval Markov Chain Abstractions of Stochastic Dynamics

Menno van Zutphen, Adrien Banse, Domagoj Herceg et al. (+2 more)

Finite-state interval Markov-chain (IMC) abstractions provide sound verification and performance bounds for continuous-state stochastic systems by enclosing cell-to-cell transition probabilities and costs in intervals. Finite-state Markovian abstractions are generally lossy, as state aggregation often destroys the Markov property. Recently, memory-dependent Markov-chain (MC) abstractions of stochastic systems have been developed to mitigate this by recording recently visited cells. In this paper, we develop the memory-dependent extension of IMC abstractions. Building on the insight that memory narrows the family of state distributions that could occur inside the current cell, we prove that memory tightens the local intervals, at the price of a larger abstraction. We also formulate no-memory IMC abstractions, tightening classic IMC abstraction intervals without enlarging the state space. To characterize spatial refinement relative to memory, we derive two cost-guarantee tightness upper bounds. The bounds depend, respectively, on cell size, and on the worst-case remaining distributional ambiguity after filtering over the remembered past. We then specialize our result to linear-Gaussian systems, for which we derive a closed-form upper bound on this ambiguity. Numerical examples show that the memory-dependent and no-memory constructions can produce tighter expected-cost intervals than comparable partition refined models.
Read the full abstract

All authors Menno van Zutphen, Adrien Banse, Domagoj Herceg, Giannis Delimpaltadakis, Duarte Antunes

Computation and Language

Enhancing Biomedical Named Entity Recognition via Multiple Programming Languages Instruction Tuning and Ensemble Method

Songtao Li, Yijia Zhang, Jianyuan Yuan et al. (+3 more)

Instruction tuning has become a common paradigm for applying large language models (LLMs) to biomedical named entity recognition (BioNER). However, existing instruction-tuning approaches still face two key challenges. First, conventional natural-language instructions typically serialize BioNER annotations as flat textual outputs, providing limited structural constraints for typed entity extraction. Second, high-quality biomedical annotations are limited, and learning from a single serialized output form may restrict structural diversity and reduce model robustness. Although external biomedical knowledge can be introduced to alleviate data scarcity, it often requires costly resource construction. To address these challenges, we propose MITE, a Multiple Programming Languages Instruction Tuning and Ensemble method for BioNER. MITE reformulates BioNER as a structure-to-structure generation task by representing both instructions and entity outputs in code-formatted representations. Specifically, each training instance is transformed into multiple programming-language formats, including Python, C++, and Java, while preserving the same underlying entity semantics. These language-specific representations provide structurally diverse supervision without requiring external biomedical knowledge or additional annotations. During inference, MITE aggregates predictions from different code formats through an entity-level voting strategy, reducing language-specific prediction variance and improving robustness. Experiments on three widely used BioNER datasets demonstrate that MITE achieves competitive performance against representative BERT-based and LLM-based baselines, attaining the best results on two benchmarks. Ablation studies further validate the effectiveness of multi-language representations.
Read the full abstract

All authors Songtao Li, Yijia Zhang, Jianyuan Yuan, Shidi Zhang, Fengyu Zhang, Hongfei Lin

Algebraic Geometry

Extension of solvable coverings across horizontal divisors over the integers

Naganori Yamaguchi

We study the effect of deleting a horizontal divisor from an arithmetic scheme on its étale fundamental group. Assume that the scheme is regular along the divisor, that each prime component contains a normal crossings point of residue characteristic $2$, and that the generic fibers of the boundary components remain geometrically integral after finite étale base change. We prove that the kernel of the induced surjection on fundamental groups is topologically perfect. In particular, the inclusion induces an isomorphism on maximal prosolvable quotients. We apply this result to projective hyperplane complements and to the extension of representations over complete local coefficient rings.
Read the full abstract