Peer Revenue Publish or profit

Papers

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

All of arXiv

Saturday 5 September

Computer Vision

GradeTrap: Authority Cues in Images Shift VLM Judgments Despite Explicit Instructions to Ignore Them

As vision-language models (VLMs) become increasingly capable and are deployed in consequential real-world settings, they must evaluate evidence independently rather than defer uncritically to human authority. We introduce GradeTrap, a controlled evaluation that places two social cues in direct conflict: a student answer, which should attract sycophantic agreement, and a conflicting answer attributed to a peer, teacher, or official answer key, which should attract authority-based deference. Models produce free-form answers while being explicitly instructed to solve independently and ignore all student answers, feedback, and grading marks. We test the models on 60 synthetic real-world trade-off scenarios. Five neutral trials establish a stable model-relative preference, followed by three repetitions of six experimental cues including controls. On the 45-item common intersection across Gemini 3.5 Flash-Lite, GPT-5.6 Luna, and Claude Haiku 4.5, a generic second-answer control yields 5.4% conflicting-answer selection. Relative to that control, pooled within-item changes show no reliable peer-review effect, a 6.9-point teacher-review effect, and a 19.5-point official-key effect. In contrast, a displayed conflicting student answer alone compared to a displayed student reference answer alone only raises selection from 2.2% to 5.2%. Official-key provenance therefore redirects judgements more than a student answer or the generic second-answer control, despite an explicit ignore instruction and an opposing student answer given along with the official key. Effects vary in magnitude across the three models.
Read the full abstract

Numerical Analysis

Crouzeix–Raviart–Marini realisation of computable a priori $L^2$-error bounds on anisotropic meshes

We consider conforming $\Pone$ and lowest-order Crouzeix–Raviart (CR) approximations of the Dirichlet Poisson problem in two and three dimensions. We obtain computable a priori $L^2$-error bounds on anisotropic simplicial meshes without a global $H^2$-regularity assumption. For piecewise constant data, the Marini relation gives an equilibrated lowest-order Raviart–Thomas (RT) flux from the scalar CR solution. The square of the resulting Marini constant is the largest generalised eigenvalue of a problem involving only scalar conforming and CR matrices. Combined with the elementwise Poincaré constant, it controls both the conforming and CR errors. The CR $L^2$ estimate uses positivity of the discrete CR–conforming gap operator rather than a nonconforming Aubin–Nitsche argument. We also derive an exact decomposition of the Marini defect into the CR–conforming energy gap and an explicit geometric term with no aspect-ratio factor. On the algebraically graded L-shaped mesh family considered here, $q>3/2$ yields $\kappa_{M,h}=O(h)$ and computable $O(h^2)$ $L^2$-operator bounds.
Read the full abstract

Quantum Physics

Unraveling-Dependent Metastability in Monitored Quantum Systems and Associative Memories

Metastability in open quantum systems is usually inferred from spectral separation in the Liouvillian, which governs unconditional, ensemble-averaged dynamics. We show that this diagnosis is incomplete at the level of individual quantum trajectories: conditioned realizations of the same unconditional dynamics can bypass, transiently access, or operationally preserve a metastable memory, depending on the monitored channel and the observed record. We demonstrate these mechanisms in a driven-dissipative nonlinear oscillator realizing a quantum associative memory, by comparing spectra, target fidelities, and phase-space distributions. The non-Hermitian dynamics obtained by post-selecting on the absence of detected events supports metastable retrieval, but with a distinct long-time fate: normalization selects the least-decaying mode of the non-Hermitian spectrum rather than the addressed memory branch. In contrast, stochastic jump trajectories can preserve retrieval over extended times when the post-measurement update remains compatible with the coherent memory structure. Thus trajectory-level metastability is not determined by the averaged generator alone; it requires compatibility between the monitored channel, the measurement record, and the metastable manifold.
Read the full abstract

Computer Vision

DriveZero: End-to-End Driving Beyond Human Demonstrations

Most end-to-end autonomous-driving systems learn by imitating human driving logs, leaving their learned behavior constrained by the quality and behavioral coverage of the recorded trajectories. This report presents DriveZero, an end-to-end system that learns driving behavior beyond human demonstrations. It decomposes driving into a perception model and an action model, pretrains each in the regime best suited to it, and combines them into one end-to-end planner. The two models call for different learning recipes: perception must understand the world, and benefits from massive and diverse visual data; action must interact with it, and requires closed-loop feedback. On the action side, we introduce DriveRL, a mixed-agent closed-loop reinforcement-learning framework. It converts real driving logs into interactive worlds, where a privileged teacher policy is trained with PPO through closed-loop rollouts. For the perception model, DriveVFM consolidates multiple frozen vision foundation models, including DINOv3, SigLIP2, SAM and Depth Anything V2, into a single backbone from raw images alone, requiring no task-specific annotations. DriveZero then unifies the two: a camera-only planner that distills the frozen DriveRL teacher through its rolled-out trajectories. The goal-conditioned teacher can moreover be queried under augmented driving intents, yielding diverse, goal-consistent supervision that logged data cannot provide. On nuPlan, DriveRL with value-guided test-time action search achieves a mean score of 93.57 across the Val14, Test14-hard, and Test14-random community splits in both non-reactive and reactive modes, exceeding the Log-Replay expert on all three splits. DriveZero achieves state-of-the-art performance on NAVSIMv1, NAVSIMv2 and the closed-loop HUGSIM benchmark without any human trajectory supervision.
Read the full abstract

All authors Hao He, Chengcheng Hu, Zirun Su, Heng Zhang, Haisong Liu, Jinke Li, Haochen Tian, Zhenwei Shen, Hongyang Li, Zhichao Li, Yunchen Yang, Bochao Huang, Siyu Zhang, Kuangye Chen, Xiongjie Zhang, Wentao Dai, Hengchen Dai, Siyuan Liu, Zehao Huang, Naiyan Wang

Quantum Physics

Propagating sparsely supported states with time-dependent neural quantum states

Neural quantum states (NQSs) have emerged as a powerful ansatz for quantum dynamics. Their high entanglement capacity promises to overcome the entanglement barrier. However, existing time-dependent NQS methods are ill-equipped to propagate sparsely supported initial states due to the inherent mismatch between the supports of such states and their time derivatives. To address this issue, we introduce the interpolation sampling method as a novel form of importance sampling, where samples are drawn from a distribution that interpolates between the wave function and its time derivative. Combined with a global-in-time variational principle, we demonstrate with the example of the two-dimensional transverse-field Ising model that interpolation sampling allows for accurate propagation of extremely peaked states that standard wave function sampling could not tackle. We further improve sampling efficiency by introducing (i) a configuration-time joint sampling scheme where spin configurations and time are both treated as random variables, and (ii) a hybrid strategy for sample proposal that incorporates the knowledge of a Krylov subspace. Our work extends the scope of time-dependent NQS methods to a wider range of physically relevant scenarios, while challenging the prevailing reliance on the Born distribution and its close variants as the default basis for sampling.
Read the full abstract

Machine Learning

Data Quality Rule Generation with LLMs

The validation of data, such as customer and employee data, is an important task in many organizations. Errors in data can have severe consequences. For example, a wrong drug unit in a patient record can lead to life-threatening medication errors, and a missing street number in an address to failed deliveries. Companies often employ rule-based enterprise data quality (DQ) tools, which allow domain experts to specify rules to validate the data over time. While rule-based DQ tools are computationally efficient and provide explainable reports, maintaining a comprehensive rule set manually is challenging, as domain experts often overlook essential rules, especially in complex domains and large data volumes. Hence, closing these gaps remains an open problem in practice. In this paper, we address the challenge of automated DQ rule generation. For this, we formalize a generalizable generate-filter framework and introduce LeDQeR, an LLM-based DQ rule generation approach. First, a large language model (LLM) generates candidates rules from an observed dirty data tuple for a given rule-based DQ tool syntax. Second, we apply four filter techniques that ensure the (i) executability, (ii) correctness, and (iii) generalizability, and avoid (iv) redundancy of the generated rules. An extensive experimental evaluation suggests that LeDQeR is able to produce effective and compact rule sets for various datasets and error types.
Read the full abstract

All authors Anna-Christina Glock, Thomas Hütter, Johannes Fürnkranz, Wolfram Wöß, Christine Dominka-Kiss, Lisa Ehrlinger

Software Engineering

SkillSpec: Intent-Masked Specification Reasoning for Agent Skill Correctness

Autonomous agent systems increasingly depend on reusable skill abstractions for consolidating experiential knowledge and domain expertise. These artifacts typically bundle free-form instructions with heterogeneous resources. However, ensuring their correctness remains challenging. Their failure modes transcend conventional code defects to subtle semantic inconsistencies such as intent conflicts, which manifest as silent failures masked by the underlying model. Moreover, skill correctness must be grounded in intended task boundaries and generalizability. We propose SkillSpec, a Hoare-style framework that formulates skill correctness as a specification reasoning problem. It transforms a heterogeneous skill repository into a unified graph representation that aligns descriptions, instructions and code artifacts. For each node, SkillSpec derives an ExpectSpec from the surrounding declared intent, and infers FactSpecs from encoded behavior under partially disclosed intent. An intent mask regulates access to holistic, lineage, neighborhood, and local views to balance the bias introduced by excessive context against unsupported inference caused by insufficient context. SkillSpec jointly reasons over these views to flag candidate defects, and automatically validates them in an isolated sandbox. On 515 real-world skills from SkillsBench and widely downloaded repositories, SkillSpec identified 763 manually confirmed defects across 239 skills, achieving 61.2% precision. The node-level analysis across multiple model families shows that specification reasoning is consistently reliable for code nodes, whereas plain-text nodes remain a major bottleneck. Most defects arise at the boundaries between declared intent and implementation, demonstrating that explicit specifications provide a practical foundation for skill quality assurance in real-world agent ecosystems.
Read the full abstract

All authors Yizhuo Zhang, Bo Kang, Yi Yang, Zhiyu Duan, Zhouteng Ye, Shunkun Yang

Computer Vision

Image-Scale Robustness and Visual Recognition Performance: A Cross-Architecture Analysis

The sensitivity of visual recognition models to changes in image scale is well established, yet the factors governing this sensitivity across heterogeneous architectures remain unclear. In this work, we investigate whether scale robustness exhibits a common quantitative structure across modern vision models. We evaluate 20 pretrained ImageNet-1K classifiers spanning seven architectural families, including convolutional, mobile, efficient, and Transformer-based architectures. By systematically reducing input image scale, we construct scale-accuracy response curves and define a characteristic scale as a compact measure of the onset of substantial recognition degradation. We then examine the relationship between characteristic scale and baseline recognition accuracy, model parameter count, architectural family, and representation stability. A strong inverse association is observed between baseline accuracy and characteristic scale (Pearson r = -0.890, R^2= 0.792, p < 10^-6). This relationship remains stable under bootstrap resampling, leave-one-architecture-out analysis, and leave-one-family-out analysis. In contrast, parameter count provides negligible additional explanatory power after controlling for baseline accuracy (p = 0.80), while architectural family does not provide significant incremental explanatory power. Furthermore, characteristic scale shows essentially no association with representation stability (r = -0.003, p = 0.991). These results indicate that, across the studied models, scale robustness is strongly organized by baseline recognition performance rather than simply by model size, architectural family, or representation stability. The study provides an empirical framework for characterizing scale robustness across vision architectures and identifies a reproducible accuracy-scale regularity that warrants further theoretical investigation.
Read the full abstract

Classical Analysis and ODEs

A solution to the Erdős Problem #1040

For a compact set $K\subset\mathbb{C}$, let $\vartheta(K)$ be the infimum of the planar areas of the unit lemniscates of all monic polynomials with zeros in $K$, allowing arbitrary degree and repeated zeros. We prove that $\vartheta(K)=0$ whenever $\operatorname{cap}(K)=1$, with no regularity assumption on $K$. The proof uses a centered harmonic polynomial that is positive on all but a set of arbitrarily small area in the polynomial hull of $K$. A Fourier average of exterior harmonic measures realizes this polynomial as the logarithmic potential of a signed measure having bounded density with respect to the equilibrium measure. A positive perturbation and an $L^1$ approximation by empirical measures then produce the required polynomials. This extends the smooth-boundary result of Krishnapur, Lundberg, and Ramachandran to arbitrary compact sets of capacity one. Together with the capacity-greater-than-one theorem of Ghosh and Ramachandran and an elementary argument for unbounded sets, it follows that $\vartheta(F)=0$ for every closed infinite set $F\subset\mathbb{C}$ of transfinite diameter at least one, answering the vanishing question in Erdős Problem 1040.
Read the full abstract

Analysis of PDEs

Stability of Magnetizable Piezoelectric Beam Systems with Two Fractional Infinite Memory Terms

In this paper, we investigate a class of magnetizable piezoelectric beam systems with two fractional infinite memory terms, whose fractional orders are denoted by \(\theta_1, \theta_2 \in [0,1]\). By introducing the Dafermos history variable, we construct an augmented state space and reformulate the system as an abstract evolution equation. The well-posedness of the system is established via semigroup theory. Through a frequency-domain analysis, we characterize the asymptotic behavior of the solutions. We show that the system is exponentially stable if \(\theta_1 = \theta_2 = 1\), and is polynomially stable in all other cases, with the explicit decay rate \(t^{-\frac{1}{2-2\theta_0}}\), where \(\theta_0 = \min\{\theta_1, \theta_2\}\). Furthermore, the optimality of the obtained decay rate is verified.
Read the full abstract

Computation and Language

Generating Adversarial Texts for Machine Translation via GRPO

As machine translation (MT) systems continue to improve, standard benchmarks become less informative for exposing remaining weaknesses. Traditional methods for creating challenging test sets rely on expensive manual creation or curation, while automated approaches struggle to produce sets with the necessary translation difficulty and linguistic diversity. We propose a scalable reinforcement-learning-based approach for rewriting existing source texts into instances that are more difficult to translate for MT systems. We fine-tune a large language model with Group Relative Policy Optimization (GRPO), using reward signals based on translation difficulty together with constraints for semantic similarity, grammaticality, and approximate length preservation. On WMT25, our approach substantially reduces average COMET translation quality from 0.63 to 0.48, while preserving grammaticality and readability, whereas the base model remains at 0.64. Evaluations on the unseen WMT19-WMT24 benchmarks confirm that this behavior generalizes beyond the training data, and human evaluation further shows that the rewrites substantially lower translation quality while incurring a moderate drop in naturalness and only a small change in grammaticality. We release our code to support reproducibility.
Read the full abstract

All authors Florian Zogaj, Jakob Hütteneder, Giovanni De Muri, Federico Villa, Aryan Sood, Vilém Zouhar

Optics

Representation of arbitrary rational functions and polynomials by non-Foster resistor-capacitor-inductor networks

The Bott-Duffin theorem shows that any arbitrary positive real rational function of finite order, representing a passive network, can be represented by a finite number of positive value resistors, capacitors and inductors. It is shown here that if the positivity restriction is lifted, any rational function or polynomial of any order may be represented by a network of (possibly negative) inductors, capacitors and resistors. When there are no repeated roots this is a sum of simple sub-circuits each of which contains no more than four components. When there are repeated roots a more complex circuit is required, here featuring iterated Wheatstone bridges of special form.
Read the full abstract

Robotics

FALCON-S: Fixed-wing ground-effect Aerodynamics Simulator and Flight Control Learning Suite

We present a modular, high-fidelity simulation framework for the development and benchmarking of flight control strategies in fixed-wing aerial robots operating near the ground. Unlike existing simulators that rely on simplified or hover-oriented dynamics, our framework models full 6DoF rigid-body physics, semi-empirical ground-effect aerodynamics, actuator dynamics, sensor noise, and environmental disturbances. This physical realism, combined with modular component design, enables systematic analysis of low-altitude flight behavior under realistic conditions. The simulator supports both CPU and GPU backends via Torch and NVIDIA Warp, enabling high-throughput parallel execution suitable for large-scale reinforcement learning training and optimal control rollouts. A unified interface accommodates a range of controllers (both RL and optical control algorithms) across tasks such as altitude regulation and trajectory tracking. Cross-validation with X-Plane and JSBSim is also supported to facilitate engineering integration and visual fidelity.
Read the full abstract

All authors Matteo El Hariry, Pedro Lima, Andrej Orsula, Matthieu Geist, Miguel Olivares-Mendez

Probability

Càdlàg Solutions to Backward Stochastic Dynamics featuring Oblique Subgradients and driven by Martingale Noise

The present study improves the qualitative analysis of backward stochastic variational dynamics on a general complete filtered probability space, considered in the spirit of Liang, Lyons and Qian (2011). Our primary objective is to overcome a substantial limitation in the study of Bensoussan, Li and Yam (2018), where the boundedness condition imposed on the multivalued subdifferential operator excludes standard obstacle-type constraints and indicator functions of convex sets. We prove the existence and uniqueness of a strong càdlàg solution under the natural assumption that the driving proper lower semicontinuous convex function is merely bounded from below by an affine/quadratic function. Furthermore, we incorporate an oblique reflection governed by a time-dependent, uniformly positive definite symmetric matrix, in the spirit of the pioneering results of Gassous, Răşcanu and Rotenstein (2012, 2015).
Read the full abstract

Human-Computer Interaction

Art2Song: Enhancing Visual Art Appreciation with Contextual Music Generation

Art2Song is a conceptual framework that expresses artworks as sound by separating Non-Visual Context, which is difficult to perceive from the image alone, from Visual Evidence. Visual Evidence, such as objects, colors, and spatial composition, is transformed into Lyrics, while the Contextual Mood derived from the historical and art-historical context in the museum's artwork description is reflected in the background soundtrack. Rather than describing artworks textually, Art2Song aims to explore the possibility of a new mode of art appreciation in which viewers experience hidden stories and emotional context through music. As future interaction directions, we plan an Emotional Layer Blending Slider interface and a structured, traceable song-generation scenario, presenting the possibility that users can explore the relationship between Visual Evidence and Contextual Mood.
Read the full abstract

Materials Science

Momentum-Resolved Electronic Structure for Quasicrystals: Full-Band Spectra and Chern Number

Quasicrystals lack the translational symmetry that underlies Bloch decomposition and Brillouin-zone integration, making full-band momentum-resolved electronic structure difficult to formulate and compute. We develop a systematically convergent reciprocal-space tight-binding framework for a broad class of quasicrystals. The method combines two systematically refinable components: a Fourier-module scattering-channel Hamiltonian that yields local spectral and current-current correlation quantities at each physical momentum, and an expanding hierarchy of pseudo-Brillouin zones that converts the resulting local quantities into bulk thermodynamic observables through an exact local-to-global relation. Applied to the Penrose and Ammann-Beenker models, the framework uncovers full-band momentum-resolved quasibands and a multichannel mechanism for pseudogap formation, both beyond the scope of low-energy effective models. It further resolves Zeeman-driven gap closings and reopenings, quantized Chern plateaus, and the phason invariance of bulk spectral and topological observables. This framework provides a unified reciprocal-space route to full-band spectral and topological properties of quasicrystals.
Read the full abstract

All authors Donglin Yang, Huajie Chen, Dexuan Zhou, Xiaoxu Li

Machine Learning

Minimizing the Effect of Sleep Deprivation in the Forward-Forward Algorithm

This paper addresses the challenge posed by sleep deprivation in the Forward-Forward algorithm, where separating the two passes in this algorithm and imbalancing the data processing in the passes is considered an imitation of the cognitive processes observed in humans suffering from sleep deprivation. Previous research has demonstrated that sleep deprivation in the Forward-Forward algorithm has a catastrophic effect on learning efficacy. To mitigate this issue, we explore several approaches; these include alternative activation, optimized loss function, and threshold tuning. To simulate periodic rest, we reduce the number of positive passes in alternating epochs, creating short break phases. We additionally investigate the potential of caffeine-induced stimulation to enhance performance during sleep-deprived conditions. Experimental evaluations conducted on the MNIST and Fashion-MNIST datasets demonstrate that these modifications improve accuracy under the context of sleep deprivation. For example, a 2%-62% accuracy gain is observed in a severe sleep deprivation setting (16 positive or awake periods and 1 negative or sleep period). The approaches also enhance the resilience of the algorithm and its alignment with the adaptive mechanisms of human cognition.
Read the full abstract

All authors Joy Datta, Puja Saha, Rawhatur Rabbi, Nafiz Imtiaz Rafin, Swakkhar Shatabda, Md. Golam Rabiul Alam, Chad Mourning

2609.06042 · 5 Sep · 14 pages, 6 figures, 7 tables. Accepted at the 28th International Conference on Pattern Recognition (ICPR 2026), Lyon, France, August 17-22, 2026. This is the accepted manuscript; the Version of Record is published by Springer in Lecture Notes in Computer Science and is available online at https://doi.org/10.1007/978-3-032-31933-3_40

Software Engineering

SpecCoder: Specification-Aware Code Generation with Curriculum Dual-Task Reinforcement Learning

Large language models (LLMs) have made substantial progress in code generation but still struggle with challenging programming tasks that require understanding rich natural language requirements. These requirements often specify problem goals, input/output formats, constraints, examples, and edge cases. Overlooking even one may produce executable but functionally incorrect code. Existing training-free methods mainly rely on prompting or agent-based workflows, while training-based methods typically optimize final code outputs. However, existing approaches provide limited supervision for learning the intermediate mapping from raw requirements to structured specifications and for grounding them in concrete implementation behavior. Consequently, models may omit critical constraints, and even when an explicit specification is produced, the implementation may fail to reflect it consistently. Motivated by this gap, we propose SpecCoder, a specification-aware two-stage training framework for code generation. SpecCoder first employs specification-guided SFT to train LLMs to derive structured specification analyses and generate code conditioned on them. It then introduces curriculum dual-task GRPO, which jointly optimizes specification-guided generation and discrimination to encourage stronger correspondence between specifications and code behavior. Experiments on APPS, CodeContests, and xCodeEval demonstrate the effectiveness of specification-aware training, with SpecCoder consistently improving both standalone code generation and agent-based workflows. Additional evaluations on BigCodeBench-Hard and ClassEval, alongside human evaluation and perturbation studies, further validate the role of structured specifications in guiding code generation and discrimination.
Read the full abstract

All authors Yixuan Li, Mingxuan Huang, Jiajing Wang, Weidong Yang, Xinyi Liu, Ben Fei, Lipeng Ma

Mesoscale and Nanoscale Physics

One-dimensional quasicrystals with tensor-network finite-state automata

Quasicrystals occupy a distinctive position between the translational order of crystals and the disordered amorphous matter. Simulating this physics has remained challenging, since quasiperiodic structures lack the translational symmetry exploited for crystals and generally require costly diagonalization of large finite approximants. Quasicrystalline order admits two equivalent descriptions, a cut-and-project scheme from a higher-dimensional periodic crystal, and a discrete set of substitution rules acting on a finite alphabet. We show that the latter, written in a numeration system adapted to the substitution, defines a deterministic finite automaton with output, the digits of a site index are fed, and the automaton returns the letter occupying that site. We further exploit another equivalence to a different construction, the transition matrices are exactly the tensors of a matrix product state, whose bond dimension is the number of automaton states and is independent of system size. This allows efficient representation of extremely large tight-binding Hamiltonians in the tensor-train language, thereby yielding an exact matrix product operator for the quasicrystal Hamiltonian at any system size. We show how this framework works for two families of one-dimensional quasicrystals, the metallic-mean and $k$-bonacci families and we explicitly construct the Fibonacci, silver-mean, and Tribonacci quasicrystals. By leveraging efficient tensor-network compression and the kernel polynomial method, we compute spectral densities for chains with more than $10^9$ sites and directly resolve the hierarchical structure of the spectrum.
Read the full abstract

Dynamical Systems

Integrability for conformally symplectic systems

The goal of this paper is to study the dynamics of conformally symplectic Hamiltonian flows under the light of integrability. As the dynamics of conformally symplectic Hamiltonian flows are dissipative and differ fundamentally from their conservative counterpart we start by proposing several notions of integrability that are better suited to the problem. We will propose two notions of integrability: $C^1$-integrability and Hopf integrability, that depend on the existence of a global attractor and on its shape. Then our main theorem focuses on Tonelli Hamiltonians whose conformally symplectic flows do not have conjugate points. We prove that such flows are automatically Hopf integrable. The proof is geometric and studies the long time evolution of vertical subspaces under the flow. It also makes use of (discounted) weak KAM theory. We also establish several results about the asymptotic Maslov index for integrable conformally symplectic Hamiltonian flows. Finally, we describe some examples to illustrate differences between symplectic and conformally symplectic Hamiltonian flows and to illustrate the pertinence of our definitions of integrability.
Read the full abstract

Quantum Physics

The $I_{3322}$ Bell inequality requires infinite dimensions

The $I_{3322}$ inequality is one of the simplest bipartite Bell inequalities, with three possible questions and two possible answers per party. Yet, despite its simplicity, it has long been conjectured by Pál and Vértesi (Physical Review A, 2010) to possess the following quirk: no finite-dimensional quantum strategy can attain its maximal violation, but an infinite-dimensional strategy can. A Bell correlation, with five questions and three answers per party, that provably possesses the same property has since been discovered by Coladangelo and Stark (Nature Communications, 2020). However, proving that the same property holds for $I_{3322}$, which lives in the simplest Bell scenario in which such a phenomenon could occur, has remained elusive. Here, we provide a proof of this conjecture. An approximate proof, containing all the main ideas, was produced by GPT 5.5 Pro after various rounds of interaction. The presentation of this proof was revised substantially by the author for completeness, correctness, and clarity.
Read the full abstract

Information Retrieval

Visual Analysis of LLM-based Entity Resolution from Scientific Papers

This paper focuses on the visual analytics support for extracting domain-specific entity from extensive scientific literature, a task with inherent limitations using traditional named entity resolution methods. With the advent of large language models (LLMs) such as GPT-4, significant improvements over conventional machine learning approaches have been achieved due to LLM's capability on entity resolution integrate abilities such as understanding multiple types of text. This research introduces a new visual analysis pipeline that integrates these advanced LLMs with versatile visualization and interaction designs to support batch entity resolution. Specifically, we focus on a specific material science field of Metal-Organic Frameworks (MOFs) and a large data collection namely CSD-MOFs. Through collaboration with domain experts in material science, we obtain well-labeled synthesis paragraphs. We propose human-in-the-loop refinement over the entity resolution process using visual analytics techniques, which allows domain experts to interactively integrate insights into LLM intelligence, including error analysis and interpretation of the retrieval-augmented generation (RAG) algorithm. Our evaluation through the case study of example selection for RAG demonstrates that this human-machine collaborative approach improved single-document entity resolution accuracy by approximately 30%.
Read the full abstract

All authors Siyu Wu, Yi Yang, Weize Wu, Ruiming Li, Yuyang Zhang, Ge Wang, Huobin Tan, Zipeng Liu, Lei Shi

Artificial Intelligence

Generator-Independent Runtime Assurance under Partial Observation

Proposal-based controllers—learned policies, language-model planners, and other black-box generators—are increasingly deployed behind runtime verification gates. We ask when the closed-loop safety guarantee decouples from the generator. The prevailing per-candidate certification pattern does not compose: under retry or best-of-$k$ selection a per-candidate false-admission level $\alpha$ can inflate to $1-(1-\alpha)^{k}$. Our main theorem shows that simultaneous setwise soundness—certifying a set of admissible proposals containing no nonviable action—is necessary and sufficient for generator-independent admission soundness, the worst case over all generators of executing a nonviable proposal equalling the probability of setwise failure; together with a design-time certificate and a no-bypass rule it is sufficient for contract safety, with violation bound $\Gamma+\sum_t\varepsilon_t+\eta$ invariant under arbitrary, even adversarial, replacement of the generator. A second theorem bounds every admission mechanism under partial observation: for a fixed probing and admission policy, if two state hypotheses whose information laws lie within total-variation distance $\delta$ require different safe decisions, then $\abar+\beta+\delta\ge1$. A sequential risk ledger makes the guarantee implementable with time-uniform confidence tubes, and shows that deterministic admission computations concentrate all statistical risk in state estimation. Simplex-style runtime assurance and control-barrier-function filtering are recovered as degenerate cases.
Read the full abstract

All authors Guangxi Wan, Yongbo Xie, Yuqi Liu, Qingwei Dong, Qingxin Li, Hongfei Bai, Peng Zeng

Software Engineering

SemVul: Semantic-Enhanced Graph Neural Networks for Code Property Graph-based Vulnerability Detection

Vulnerabilities in source code are often the root cause of cyberattacks worldwide, as attackers exploit weaknesses in software to gain unauthorized access, steal data, or disrupt services. In this study, we evaluated existing research approaches and propose SemVul, a vulnerability detection pipeline that demonstrates better generalization and higher accuracy in learning vulnerable code patterns. We propose a Code Property Graph-based vulnerability-detection approach combined with semantic-level enhancement, enabling the model to capture both the program's structural flow and the semantic meaning of the code. Our approach integrates both node-level and edge-level semantic embeddings using pre-trained code embedding techniques. We systematically evaluate multiple GNN architectures on publicly available benchmark datasets. SemVul is generic with respect to the programming language and supports multiple architectures. By integrating structural and semantic information, the proposed approach improves vulnerability detection performance. Our results show that SemVul outperforms existing approaches and provides better generalization.
Read the full abstract

Fluid Dynamics

Coupled Analytical Model for the Internal Boundary Layer Height, Wall Shear Stress and Mean Velocity Behind a Surface Roughness Transition in Boundary-Layer Flow

A coupled analytical framework is developed to predict the internal boundary layer (IBL) height, wall shear stress and the mean velocity profile downstream of a surface roughness transition in a neutral atmospheric boundary-layer. The new model combines a three-layer analytical velocity formulation with a modified diffusion analogy for the IBL growth rate. The diffusion analogy links the growth rate of the IBL to turbulent diffusion and to mean vertical advection. Our formulation rectifies physical inconsistencies in a previous model that assumed that the turbulent diffusion is controlled only by the upstream surface conditions and that the characteristic streamwise velocity difference caused by streamline displacement is independent of downstream distance from the roughness transition. Key model parameters, namely the turbulent diffusion coefficient and the eddy viscosity augmentation coefficient, are modelled as functions of the upstream-to-downstream aerodynamic roughness length ratio. The coupled model, along with three other analytical models, is tested against twelve datasets covering a wide range of roughness ratios, including wind-tunnel experiments and large-eddy simulations. The new model accurately predicts wall shear stress, mean velocity, and IBL height for all combinations of upstream and downstream roughness values evaluated, for both smooth-to-rough and rough-to-smooth transitions.
Read the full abstract

Functional Analysis

Area operators on Hardy spaces of Dirichlet series II: counterexamples and compactness criteria

We study area operators $\mathbb{A}_{\mu,l}$, $0<l<\infty$, induced by positive Borel measures on the right half-plane and acting on the Hardy spaces of Dirichlet series $\mathscr H^p$, $0<p<\infty$. We first disprove a conjecture proposed by the present authors in an earlier work by constructing a probability measure, valid for all $0<p,l<\infty$, for which the associated area operator is bounded although the measure fails the proposed Carleson conditions. We then characterize the boundedness and compactness of $\mathbb{A}_{\mu,l}$ on both $\mathscr H^p$ and the Hardy space $\mathscr H^p_0$ of Dirichlet series vanishing at $+\infty$, for every $0<p,l<\infty$, via integrability conditions for the associated cone integrals. In particular, the boundedness and compactness are equivalent for these operators. We also give direct proofs of compactness under vanishing Carleson measure and compact $H_{\mathrm i}^p$-Carleson embedding hypotheses. As an application, we establish an estimate for the essential norms of bounded Volterra operators. This leads to a complete characterization of compactness for Volterra operators on $\mathscr{H}^p$ and provides a different proof of compactness for those with Dirichlet series symbols in $\operatorname{VMOA}(\mathbb C_0)$.
Read the full abstract

Dynamical Systems

Joint Point-Distality and structure theory for commuting actions

Let $G$ and $H$ act commutatively and minimally on a compact metrizable space $X$. We prove that if the two subactions are point-distal, equivalently HPI, then the action generated by them is again point-distal. We also construct a common highly proximal extension on which both subactions are strictly HPI. As consequences, transitivity of mixed products and linear iterates upgrades to minimality in the point-distal category. We then compare the canonical structure of the two subactions with that of the joint action. The maximal highly proximal operations coincide, whereas the maximal joint isometric factor over a common factor is the meet of the two subaction-isometric factors. This gives a recursive description of the joint Furstenberg tower. Finally, we construct commuting minimal distal homeomorphisms of $\mathbb T^3$ whose canonical Furstenberg towers are different, showing that the meet formula can be strict.
Read the full abstract

High Energy Astrophysical Phenomena

The accuracy of merger times in hierarchical black hole triples

Orbit-averaged equations are widely used to model the gravitational-wave-driven evolution of compact binaries. Their validity relies on the radiation-reaction timescale being much longer than the orbital period, an assumption that can break down for highly eccentric systems. In this work, we investigate the impact of orbit averaging on merger-time estimates by comparing orbit-averaged and non-orbit-averaged prescriptions. We apply our analysis to two different astrophysically motivated populations of black hole binaries in hierarchical triple systems, where von Zeipel-Lidov-Kozai oscillations can drive the inner binary eccentricities close to unity. We find that merger-time estimates obtained with orbit-averaged equations remain remarkably accurate across a wide range of eccentricities, even when the orbital timescale exceeds the radiation-reaction timescale. Significant discrepancies arise only for the most extreme and compact binaries, simultaneously characterized by very high eccentricities and small semi-latus recta. In this regime, the merger time becomes strongly dependent on the initial orbital phase, with different phases yielding predictions that can differ by several orders of magnitude. We also compare our results with an orbit-averaged merger-time prescription specifically developed for hierarchical triple systems and find substantial differences between this prescription and both the generic orbit-averaged and non-orbit-averaged calculations in the high-eccentricity regime. While these results demonstrate the robustness of orbit-averaged merger-time estimates for most systems, they also highlight the potential importance of non-orbit-averaged effects when merger times are used to infer compact-binary delay-time distributions and formation channels.
Read the full abstract

High Energy Physics (Phenomenology)

Mass Generation for Axion Like Particles

The most widely accepted solution to the Strong CP problem is the QCD axion, whose traditional parameter space is relatively constrained. However, recent studies on exotic axion models have shown that this space can be significantly extended, opening up a wide range of previously unexplored possibilities for model building. I first review the current status of axions and axion-like particles, then focus on the challenges related to the origin of their masses and explore potential connections to other unresolved issues within the Standard Model.
Read the full abstract

Combinatorics

A Conway-Coxeter theorem for decorated frieze patterns

Let $m\geq0$ and put $n=m+3$. We introduce a normalized positive Laurent class of decorated frieze patterns and prove a decorated analogue of the Conway-Coxeter classification theorem. Namely, such friezes of width $m$ are in canonical bijection with weighted triangulations of a convex $n$-gon, whose boundary edges are labeled by independent variables $y_1,\ldots,y_n$ and whose diagonals are labeled by $x_1,\ldots,x_m$. While a weighted Conway-Coxeter propagation algorithm constructs the frieze from a weighted triangulation, a main new ingredient is an explicit nonrecursive Laurent formula expressing each quiddity entry directly from the weighted triangles incident to the corresponding vertex. Conversely, specialization to $1$, Laurent positivity, and a decorated cutting-and-gluing procedure recover the weighted triangulation from the frieze.
Read the full abstract