Information Retrieval
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Monday 31 August
Graph-based Retrieval-Augmented Generation (GraphRAG) supports multi-hop reasoning by organizing corpora into structured graphs. However, graph reachability often captures semantic association rather than evidence support, so a reachable passage may still fail to justify a required cross-passage transition. We propose EvLink, an evidence-linking retriever that preserves passages as retrievable evidence units and builds evidence-supported transitions between them. EvLink constructs two types of reliable links: relation-grounded evidence links justified by explicit source relations, and endpoint-alignment links serving as sourcebounded fallbacks. For retrieval, we introduce a two-stage retrieval strategy. First, bounded breadth-first search over source-grounded evidence links recovers bridge passages missed by similarity-based methods. Then, evidenceneed mining with noisy-OR coverage refinement selects a compact, non-redundant evidence set satisfying distinct question facets. Experiments on three multi-hop and two simple QA benchmarks show EvLink consistently outperforms leading GraphRAG baselines with average gains of 2.4 R@5, 1.9 EM, and 2.4 F1
Machine Learning
Bandit Multiclass PAC Learning: Structural Baselines and a Confidence Direct Sum
How much does high confidence cost when a multiclass learner sees only whether its prediction is correct? We separate two behaviors in realizable bandit PAC learning. An anchored dimension gives a general lower bound that is attained, up to universal constants, by a family of product-block classes. In contrast, for $n\ge16$ divisible by four, an affine multiplexer with bandit DS dimension $2n-1$ has sample complexity $\Theta((n\min\{n,\ln(1/\delta)\}+\ln(1/\delta))/\varepsilon)$ in the standard accuracy and confidence regime. Its lower bound applies to adaptive randomized learners with improper outputs. The obstruction is that evidence collected against one candidate group cannot be reused against the others. The better of two algorithms matches this cost. One finds the correct group through a positive response. The other exhausts the rank of successive groups. The resulting confidence profile rules out a universal additive-confidence bound linear in the bandit DS dimension. Moreover, classes with identical anchored dimension, bandit DS dimension, and pointwise label width can have polynomially different sample complexities. We also sharpen the confidence accounting in ListCascade and show how one positive response replaces the bandit DS dimension by the anchored dimension in a general upper bound.
Read the full abstract
Quantum Physics
Certified exact identification of the I3322 quantum value
We determine the quantum value of the three-setting, two-outcome Bell functional I3322. The value is the unique intersection \(q_\dagger\) of an analytic A-high stable curve of the exact Jacobi characteristic map with a central reverser curve. A computer-assisted proof isolates this intersection in a rational interval of width \(10^{-55}\), beginning 0.2508753845139765356173366109459512989013219739714764755. We prove that the supremum over finite-dimensional tensor-product strategies, the unrestricted spatial value and the commuting-operator value all equal \(q_\dagger\). No finite-dimensional strategy attains the value, whereas a normal infinite-dimensional spatial strategy and a commuting GNS strategy do. The proof combines an exact Bellman/path variational reduction, extraction of a positive bilateral calibrated Jacobi fibre from a maximizing weak-star state, a reflected-fibre central-incidence theorem and interval certificates for the unique stable/central intersection. A separate operator-positive three-remainder decomposition gives the commuting upper bound. The complete verification record is archived at DOI 10.5281/zenodo.22160086.
Read the full abstract
Artificial Intelligence
Fair Like Us? Auditing LLM Alignment in Resource Allocation
Fair allocation of scarce, indivisible resources is an important challenge in many societal problems. While there are several formal theories of fairness, no single definition can always be satisfied. As large language models (LLMs) are increasingly used to support decisions and act as agents, they raise new concerns about distributional justice: their judgments are not directly tied to any specific fairness framework and may violate key normative principles. In this work, we introduce a general method for evaluating fairness reasoning in LLMs. We study first-person fairness judgments across a broad set of models and compare them directly with human responses on matched scenarios and elicitation conditions. We find that LLMs tend to prefer stricter fairness constraints than humans, show more self-interested behavior, are sensitive to how information is framed, and are difficult to align with human judgments using fine-tuning with current datasets.
Read the full abstract
Game Theory
Online House Allocation with Subsidy
House allocation is a fundamental problem in which each agent is assigned exactly one house. While the classical model assumes that all houses are available before the allocation is computed, many practical settings require decisions to be made as houses become available over time. We introduce the online house allocation problem, where houses arrive sequentially and the algorithm must maintain an allocation without knowledge of future arrivals. Unlike online fair division, the one-house-per-agent constraint makes recourse an inherent part of the problem, as accepting a newly arrived house may require reassigning previously allocated houses. We study online house allocation under subsidy-based fairness, where monetary subsidies eliminate envy among agents. We show that envy-freeability can always be maintained online using bounded recourse and that reassignment chains of length linear in the number of agents are unavoidable in the worst case. In contrast, minimizing the total subsidy is fundamentally harder: no deterministic online algorithm against an adaptive adversary, and no randomized online algorithm against a non-adaptive adversary, admits a bounded competitive ratio, even for two agents and four houses. We complement these impossibilities by showing that exact online subsidy minimization is possible whenever there is at most one extra house beyond the number of agents, and that this guarantee is best possible with respect to the number of extra houses. Finally, we develop learning-augmented algorithms that recover the offline optimum under accurate predictions while providing explicit robustness guarantees when predictions are inaccurate.
Read the full abstract
Machine Learning
Predicting Symptoms of Amotivation and Anhedonia among University Students with a Novel Oversampling Method
University students experience disproportionately high rates of common mental health conditions, such as depression, which can impair learning, social functioning, and overall well-being. Within this context, symptoms of amotivation (i.e. loss of motivational drive) and anhedonia (i.e. diminished interest or pleasure) are particularly debilitating, yet they frequently go undetected. Developing new approaches to identify students with prominent amotivation and anhedonia could enable earlier and more targeted intervention. Machine learning (ML) methods have increasingly been used to classify individuals according to symptom severity. However, these ML models often suffer from class imbalance, where the majority of cases fall in the low-symptom group and relatively few in the high-symptom group. This imbalance can reduce model accuracy and bias predictions. To address this, studies commonly employ the popular oversampling strategy SMOTE. However, SMOTE has a notable limitation: it may generate invalid values for nominal variables. In this paper, we introduce a novel and effective oversampling method that addresses this shortcoming. Our approach leverages a predictive model to generate nominal variables, rather than interpolating them. We validate our method on a large-scale GPS location dataset collected from university students and demonstrate that it is significantly better than existing oversampling approaches in predicting elevated symptoms of amotivation and anhedonia.
Read the full abstract
Computation and Language
DP-IPI: A Hybrid Differential Privacy Text Rewriting Mechanism for Indirect Personal Identifiers in Clinical Texts
Despite the strengths of modern anonymization and de-identification techniques, the risk of re-identification remains significant due to the indirect identifiers remaining in texts. To address this problem, recent works have applied text rewriting under Differential Privacy (DP) to prevent data linkage by perturbing texts via noise addition. Such methods privatize all tokens in a text indiscriminately, diminishing text quality and usability in critical domains such as in clinical settings. Focusing on indirect personal identifiers (IPIs), we introduce a utility-preserving DP text rewriting method that only privatizes spans containing IPIs. We show that our method effectively reduces re-identification risks in clinical texts while being producing more coherent and usable output texts, leading to higher privacy-utility trade-offs. In this, we demonstrate the effectiveness of hybrid text privatization, which leverages the promise of DP in an efficient, usable manner.
Read the full abstract
Artificial Intelligence
Sequential knowledge editing breaks a model's ability to tell good evidence from bad, without costing it accuracy
Knowledge editing is evaluated on whether the edited fact changed, whether paraphrases follow, and whether unrelated answers stayed put. A model can pass all three and still lose something none of them measures: the ability to decide, on facts that were never edited, which retrieved documents to believe. We score the log odds a model assigns to its remembered answer against the answer an injected passage asserts, before and after editing, holding the query, the passage and both candidate strings fixed. Our cleanest arm is a conservatively tuned LoRA: after 1,000 sequential edits on Qwen2.5-7B-Instruct it leaves MMLU unchanged to four decimal places, yet the spread of the arbitration quantity across untouched facts falls by 36%. Selective prediction degrades with it. Area under the risk-coverage curve rises by 0.107, against 0.005 for a norm-matched perturbation at the same MMLU, and error on the model's most confident quarter of arbitration decisions goes from 0.217 to 0.342. This is not capability loss. Sweeping random perturbation over five severities, damage bad enough to cut MMLU from 0.6275 to 0.3725 produces less harm (0.088) than MEMIT does at 0.6050 (0.102). The effect holds across three seeds, two model families, two datasets, two probe-disjointness criteria, three prompt templates and paraphrased queries. Layer ablation on saved weight deltas shows it is distributed: no single layer reproduces it, and removing any one recovers about half. Under retrieval with a frozen retriever, accuracy falls from 0.592 to 0.46. A secondary finding may matter more in practice. Three of five model and method pairings we ran collapse to chance MMLU at 1,000 sequential edits under published hyperparameters, while edit success stays at 1.00 and locality reads clean. Sequential-editing evaluations that never measure capability cannot see this.
Read the full abstract
Commutative Algebra
Resultant of an equivariant polynomial system with respect to a direct product of $r$ symmetric groups
In this paper we study the resultant of systems of homogeneous multivariate polynomials which are equivariant under the action of a direct product of symmetric groups. We first treat, in detail, the case of a product of two symmetric groups, and establish a decomposition formula for the resultant of such systems. We then show that this decomposition, together with the underlying combinatorics, extends to an arbitrary (finite) direct product of $r$ symmetric groups. Thanks to these decomposition formulas, we prove that the discriminant of a multivariate homogeneous polynomial invariant under a direct product of $r$ symmetric groups splits into a product of resultants of smaller size that are easier to compute.
Read the full abstract
Representation Theory
An irreducible decomposition of the Weil representation restricted to open compact subgroups
For a $p$-adic field of any residual characteristic, we determine irreducible decompositions of the Weil representation restricted to the maximal open compact subgroups and the Iwahori subgroup by using the Schrödinger model. In particular, these decompositions show that the restrictions are multiplicity free.
Read the full abstract
General Mathematics
Modifications of the Continuous Gromov–Hausdorff Distance
This work studies modifications of the Gromov–Hausdorff distance in which the infimum is taken over pairs of morphisms of a given class $\FF$ — the class of morphisms of a subcategory of the category of metric spaces (the $\FF$-Gromov–Hausdorff distance). We prove that the $\FF$-distance depends Lipschitz-continuously on the class $\FF$ in the Hausdorff metric on morphisms. As a consequence, for second countable $C^k$-manifolds equipped with a metric compatible with the topology, the smooth distance coincides with the continuous one: $\dGH{C_k} = \dGH{C}$. We also introduce the partially continuous and partially locally constant distances (continuity is weakened to continuity on an open subset of the support of full measure) and prove that on the class of metric spaces they coincide with the classical Gromov–Hausdorff distance.
Read the full abstract
Computation and Language
PII-TRACE: A Benchmark for Context-Aware PII Detection in Multi-Turn LLM Conversations
LLM assistants and agentic systems log long multi-turn conversations. AI providers often scan these conversations for Personally Identifiable Information (PII) and mask the PII before storing or processing conversation data. Yet most PII detectors and benchmarks target self-contained records rather than cross-turn evaluation. To evaluate PII detection across turns in multi-turn conversations, we introduce PII-TRACE (Tracing Recurring PII Across Conversational Exchanges), to our knowledge the first PII benchmark to assess whether detectors identify PII in conversational contexts and cover every mention of a recurring identifier across turns. PII-TRACE contains 13,148 synthetic multi-turn dialogues in 13 languages with character-level spans and identifier clusters. Across eleven baselines, including frontier LLMs, no detector achieves full entity-level coverage without substantial false positives on PII-free conversations, and single-pass reading loses a third of the gold characters on long dialogues. To close this gap, we introduce PII-Tracer, a compact 0.6B-parameter detector trained with conversation-level supervision. PII-Tracer attains the highest entity-level coverage of any system we evaluate and also performs strongly on standard single-record benchmarks.
Read the full abstract
Machine Learning
Improving Parameter Utilization by Sharing Neural Experts Across Layers in Transformers
Transformer-based large language models often suffer from inter-layer parameter redundancy, where functional transformations are redundantly learned across network depths. We propose CS-MoE, a novel Transformer architecture featuring cross-layer expert sharing to address this inefficiency. Deviating from the widely used Mixture-of-Experts (MoE) architecture that terminates each Transformer block with layer-isolated experts, CS-MoE combines layer-independent experts with concurrent access to a centralized, globally shared expert pool. This Global Experts Sharing mechanism enables elastic control over token-level parameter activation and computational consumption (FLOPs). Experiments demonstrate that CS-MoE achieves lower perplexity than equal-scale dense Transformers while activating only 55% of parameters. Furthermore, its performance scales monotonically with an increased number of activated experts and approaches MoE counterparts that consume more FLOPs by expanding the shared pool with a fixed FLOPs budget. CS-MoE also establishes a flexible Pareto frontier between computational cost and model capacity, offering an efficient alternative for computation-constrained environments.
Read the full abstract
Medical Physics
Reconstruction of 4D Mitral Regurgitation Hemodynamics from Sparse Planar Data using Deep Operator Networks with Test-Time Adaptation
Quantifying mitral regurgitation severity remains limited by the assumptions of clinical flow convergence methods, while high-fidelity simulation and volumetric velocimetry are too slow for routine use. We investigate whether a learned solution operator can reconstruct transient three-dimensional transvalvular hemodynamics from the sparse observation an in-vitro experiment actually provides: a single planar velocity slice and two boundary pressure traces. A Deep Operator Network is pretrained on an experimentally benchmarked URANS database spanning eleven mitral regurgitation orifice phantoms, learning a mapping from a masked two-component planar velocity snapshot to the surrounding volumetric field, and is subsequently adapted to unseen target cases by fine-tuning on their sparse measurements. Adaptation reliably corrects the flow topology within the supervised plane, reorienting a strongly eccentric jet that the pretrained operator predicts as straight, and yields full-field predictions in minutes rather than the days required by the underlying simulations. Its influence decays sharply with distance from that plane, however: measured against phase-resolved particle image velocimetry, the reconstruction error rises from 24.6% at 2mm to 52.6% at 6mm, and the resulting mismatch between corrected and uncorrected layers degrades physical consistency. Single-plane supervision thus constrains the observed plane far more effectively than the surrounding volume, which we identify as the principal obstacle to coherent 4D reconstruction from sparse planar data.
Read the full abstract
Medical Physics
A novel vortex generator for enhancing bileaflet mechanical heart valve performance proposed through detailed hemodynamic and data-driven dynamic mode decomposition studies
Prosthetic mechanical heart valves (MHVs), particularly the bileaflet types, have been one of the most advanced and commonly implanted heart valves for more than four decades. These bileaflet mechanical heart valves (BMHVs) offer superior durability and improved hemodynamics; however, they have certain drawbacks, including the need for lifelong anticoagulation to prevent blood clot formation and the associated risks of thromboembolism. To overcome these limitations, comprehensive research studies are deployed, among which controlling the overall hemodynamics by means of passive interventions, such as placing (multiple) vortex generators (VGs) on the heart valve leaflets, is one possibility. These VGs, in particular, are known to delay the separation of boundary layers formed around the leaflet surfaces, provided they are precisely positioned and have their chosen design features. Limited research has shown that VGs improve hemodynamics beyond a BHMV when rectangular-type VGs (RVGs) are considered. Specifically, configurations such as co-rotating and counter-rotating are investigated at an incidence angle of $23^{\circ}$; however, their location on the heart valve leaflets and their shape can prove vital, an aspect that has not been explored so far. Suffice it to add, the alteration in hemodynamic performance due to these VGs may often lead to an impact on several critical clinical indices such as the pressure gradients, turbulent intensity, wall shear stress (WSS), blood damage (BD), etc., which typically establish the overall functional efficiency of an MHV.
Read the full abstract
Medical Physics
How does human blood rheology influence arterial and cardiovascular device hemodynamics? A comprehensive review, discussion, and future directions
Human blood is a complex biological suspension whose macroscopic flow behaviour arises from the coupled microscopic dynamics of deformable blood cells, plasma proteins, and evolving cellular microstructures. As a result, blood exhibits a range of nonlinear rheological behaviours, including shear-thinning, viscoplasticity, viscoelasticity, and thixotropy, which strongly influence its physiological and clinical hemodynamics. This review provides a comprehensive assessment of the role of blood rheology in arterial and blood-contacting cardiovascular device flows, with particular emphasis on the extent to which different rheological characteristics alter clinically relevant flow quantities and on whether the conventional Newtonian approximation remains adequate for the analysis. The underlying physical mechanisms responsible for the major rheological behaviours of blood are first discussed, followed by a critical review of the constitutive models commonly employed in hemodynamic analysis, including generalised Newtonian, viscoelastic, and thixo-elasto-viscoplastic (TEVP) ones. The effects of blood rheology on arterial flows involving stenosis, bifurcation, and aneurysm under both steady and pulsatile flow conditions are subsequently reviewed and discussed in terms of several physiologically important parameters, such as pressure drop, flow resistance, wall shear stress, flow separation, and recirculation zone. The influence of blood rheology on cardiovascular devices, including prosthetic heart valves and stents, is also reviewed, particularly regarding flow stagnation, residence time, hemolysis, and thrombosis.
Read the full abstract
Functional Analysis
Grothendieck and $\ell_\infty$-Grothendieck subspaces of $\ell_\infty$
Let $\mathfrak{c}=2^{\aleph_0}$. We prove that $\ell_\infty$ contains $2^{\mathfrak{c}}$ pairwise non-isomorphic non-reflexive Grothendieck subspaces. They may all be chosen to contain the canonical copy of $c_0$ and to have infinite-dimensional reflexive quotient. This is optimal and answers by the authors in 2021. We also show that the relative $\ell_\infty$-Grothendieck property depends on the specified copy of $c_0$ and is not an isomorphic invariant of the subspace. More generally, for every closed $E\subseteq\ell_\infty$ there is an $\ell_\infty$-Grothendieck subspace isomorphic to $\ell_\infty\oplus_\infty E$; taking $E=c_0$ answers a problem by González et al.
Read the full abstract
Computation and Language
ViCo: Visual-oriented Coding with Self-Reflection for Chart Replication
This paper addresses the challenge of generating high-quality academic charts that match the visual standards of human-authored papers. While existing AI agents can produce well-structured text and code, their generated visualizations often lack the stylistic and semantic fidelity of human designs. Advanced coding agents that employ self-reflection mechanisms exhibit poor visual reasoning and limited reflection following, resulting in sparse reward signals that severely undermine their reinforcement learning (RL). We propose ViCo, a training framework for visual-oriented coding that employs iterative reflections to align generated chart images progressively with the reference. We first introduce a self-supervised warm-up stage, which augments Monte Carlo Tree Search with consistency-based pruning to synthesize high-quality reflection trajectories, ensuring that each coding step strictly follows the outcomes of prior reflections. A multi-step RL algorithm is then developed, using counterfactual baselines to estimate advantage for reflection and action steps within each refinement cycle, thereby addressing the reward sparsity. To enable efficient reward in massive training, we propose an automatic, multifaceted evaluation framework that assesses charts' style, layout, and semantic consistency via a hierarchical heterogeneous layout graph structure. Experiments on three public benchmarks demonstrate that ViCo, trained on an 8B model, achieves performance close to proprietary LLMs with adequate reflection capabilities.
Read the full abstract
Robotics
Seeing What the Vehicle Sees: Video-Augmented Virtual Reality for Physical Autonomous Vehicles
Autonomous vehicles are expected to improve road safety and efficiency, but passengers often remain uncertain about what the vehicle perceives and why it acts as it does. Virtual reality (VR) offers a safe and repeatable medium for presenting this information, yet most passenger-facing VR studies rely on fully simulated vehicles or pre-scripted scenarios, so the motion and perception shown to the user do not originate from a physically operating autonomous system. This paper presents a video-augmented VR framework that couples a physical ROS 2 autonomous robot vehicle to a Unity 6 application deployed on a Meta Quest 3S headset. The vehicle state and live onboard camera stream are transmitted over two independent communication channels, allowing the virtual vehicle to mirror the physical robot's motion while the passenger simultaneously views the vehicle's first-person camera feed and its navigation decisions through an in-vehicle dashboard interface. We evaluate the framework over 20 repeated closed-loop navigation trials. The system achieves a mean state-update latency of 29.63 ms, a mean relative route-progress error of 2.28% between the physical and virtual vehicles, and video delivery at 10.006 frames per second with 0.25% frame loss. All monitored navigation decisions were correctly reflected in the VR interface with no missed or incorrect notifications. The results indicate that the framework can support temporally synchronized, semantically consistent, and accurate route-progress representation for immersive observation of physical autonomous-vehicle behavior.
Read the full abstract
Analysis of PDEs
Vanishing boundary geometry and critical pressure compactness for the three-dimensional Navier–Stokes equations
We prove boundary partial regularity for the three-dimensional nonstationary Navier–Stokes equations in bounded domains with fractional Sobolev boundary graphs whose integrability exponent is greater than three. Within the fractional Sobolev graph setting, this lowers the boundary threshold from fifteen quarters to three. We work in Breit's boundary suitable weak-solution framework, with the mixed maximal-regularity pair $(5/3,15/14)$ explicitly included in the solution assumptions. The main difficulty is to recover pressure compactness without imposing higher regularity on the rough boundary coefficients. After a rigid rotation to the tangent plane and parabolic rescaling, we show that each boundary graph becomes small in both Sobolev multiplier classes needed for the local energy inequality and the critical Stokes estimate. Geometry and fluid excess enter a contradiction argument as independent small parameters, so the limiting problem is the standard flat Stokes system. A localized pressure decomposition separates the vanishing nonlinear and coefficient errors from the forced and homogeneous flat pressures. The error pressure converges strongly at the critical mixed integrability pair, whereas higher spatial integrability is used only for the flat problems. These estimates yield a one-step excess decay bound with a constant independent of the contraction radius. A physical-scale iteration, followed by interior–boundary Campanato patching, gives local parabolic Hölder continuity of the velocity at every regular boundary point when the force has integrability exponent greater than five halves. The boundary singular set is relatively closed and has zero parabolic Hausdorff measure in dimension five thirds. The geometric decay determines a uniform starting scale and does not restrict the final regularity exponent.
Read the full abstract
Analysis of PDEs
Linear Corrector Estimates and Corrected Pressure Convergence for a Two-Parameter Hyperbolic Relaxation of the Incompressible Navier–Stokes Equations
We study a two-parameter first-order hyperbolic relaxation approximation of the incompressible Navier–Stokes equations on the two-dimensional torus, under the derivative-level large-perturbation assumptions of Huang, Rohde and Zhang. Building on the auxiliary-system approach used in earlier pressure-convergence results, we introduce a non-autonomous linear corrector that carries the complete initial discrepancy, the linearized convection and the smooth consistency forcing. This corrector describes the full linear response; it is not asserted to be a purely acoustic component. An exact compensated energy identity, including a time integration by parts in the linearized convection term, yields an integrated gradient estimate for the velocity corrector. The two-dimensional Ladyzhenskaya inequality then bounds its quadratic self-interaction by $O(\delta^2)$ in squared space-time $L^2$ norm. Consequently, the nonlinear remainder has weighted energy $O(\delta^2)$, and its pressure and velocity components satisfy the $L^\infty(0,T;L^2)$ bounds $C_T\delta/\sqrt{\epsilon}$ and $C_T\delta$, respectively. In the sufficient parameter window $\delta^2\ll\epsilon\leq\mu_*\delta$, the pressure after subtraction of the linear corrector converges strongly to the incompressible pressure. A mean-zero Fourier-mode construction, transferred to the nonlinear system by the remainder estimate, shows that the uncorrected pressure need not converge even on time intervals bounded away from zero.
Read the full abstract
Analysis of PDEs
A Proof of a Conjecture on a Pólya Functional
Let $\lambda_1(\Omega)$ and $T(\Omega)$ denote the first Dirichlet eigenvalue and torsional rigidity of a bounded convex domain $\Omega\subset\mathbb{R}^2$, and let $M=\max_\Omega u$, where $u$ is the torsion function. We prove the sharp inequalities $\pi^2/24<\lambda_1(\Omega)T(\Omega)/|\Omega|<\pi^2/12$. This proves the two-dimensional case of Conjecture 4.2 proposed by van den Berg, Buttazzo, and Pratelli. Its planar formulation was later restated as Conjecture 1.1 by Bañuelos and Mariano, who proved it for triangles and rectangles. The lower bound follows by combining Payne's strict estimate for $\lambda_1M$ with the sharp torsion-efficiency inequality $T(\Omega)\geq |\Omega|M/3$. For smooth strictly convex domains, we use an Airy stress potential to construct a convex body whose surface-area measure is the torsional first-variation measure. A sharp one-dimensional inequality for directional maximum profiles then yields a stronger geometric containment. For the upper bound, we factor the Pólya functional through the second torsion moment. A level-set torsion–perimeter inequality gives the factor $5/6$, while a sharp weighted one-dimensional estimate gives the factor $\pi^2/10$. Collapsing triangles and elongating rectangles show that both endpoint constants, as well as the efficiency constant $1/3$, are sharp.
Read the full abstract
General Physics
Comment on a proposal for non-relativistic transverse electromagnetic Hamiltonian
We comment on the recent proposal (arXiv:2608.21211v1 (2026)) of a new transverse Hamiltonian for electromagnetic interacting charged particles that ignores both the classical and quantum-mechanical versions of the already known one. 1 Introduction
Digital Libraries
Benchmarking locally hosted language models for journal editorial work on a compact desktop workstation
Journals are beginning to consider language models for manuscript handling, but submitted manuscripts are unpublished, and where policy forbids sending them to an external service the model must run on hardware the journal controls. The capability of locally hosted models on editorial work has not been measured. Here we constructed a benchmark of eight editorial tasks from a journal's Instructions for Authors, from manuscripts carrying defects we seeded and verified independently, and from published reviews of a preprint, and evaluated twenty open-weight models spanning a twenty-five-fold range of weight size on a compact desktop workstation of the kind a laboratory or small editorial office can adopt. The strongest model detected 36 of 40 seeded guideline violations and occupied 81 GB; a 17 GB model detected 33. Across the best configurations tested, we observed no consistent monotonic association between weight size and score: rank correlations were negligible on every task (Spearman |rho| <= 0.19), and within one model family the larger member scored below its smaller sibling. A deterministic checker of regular expressions and arithmetic, using no model, detected 31 of the same violations in a fraction of a second, and the union of its detections with those of the strongest model covered all 40. On the single peer-review case, the best model recovered 6 of 12 points from three published reviews. Prompt structure substantially altered scores within individual models. This level of performance is therefore within reach of a workstation of this class, once the deterministic checks are written.
Read the full abstract
Computer Vision
Lensless Gaze Is Not Private by Default: Auditing Identity Leakage Across Disclosure Surfaces
Lensless near-eye sensing is often described as privacy-friendly because its coded measurements are visually unintelligible. Yet visual unintelligibility reflects human interpretation, not what a learned adversary can recover. We therefore treat identity privacy as a systems property of disclosure surfaces: representations crossing sensing, storage, computation, and output boundaries. We audit a simulated lensless gaze pipeline under a 36-subject known-gallery closed-set identification protocol with a fixed, known PSF; privacy from an unknown or varying optical key is outside our scope. Reported accuracies are empirical attack success rates under matched linear and MLP probes and do not upper-bound stronger adversaries. Simulated lensless measurements yield 96.7% top-1 identification versus 97.7% for matched original eye crops, while an MAE embedding retains 94.3%. Compression alone offers little protection: 8-D PCA and a matched 8-D bottleneck retain 93.2% and 91.8%, whereas separately trained 8-D GSPL bottlenecks yield 77.5% mean recovery across three seeds. A released 128-way gaze token lowers single-frame recovery to 38.1%, while its residual and continuous gaze output expose 62.1% and 72.6%, respectively. Under a source-frame-disjoint tiled protocol, token summaries reach 39.9% at T=25, showing that repeated-output risk depends on representation and aggregation. These rates reflect all subject-correlated information in the evaluated dataset, including acquisition and behavioral cues, rather than isolating intrinsic ocular biometrics. Ordinary least squares residualization against a six-dimensional crop geometry and intensity summary still leaves lensless recovery at 95.1%. Our results show that privacy claims for lensless sensing must be tested at disclosure boundaries rather than inferred from appearance.
Read the full abstract
Computer Vision
AgenticGen: Reward-Guided Agentic Video Generation for Advertising
Advertising video generation is not only a video synthesis task, but also a product-conditioned reasoning problem whose success is measured by online business metrics. Recent video foundation models can generate realistic clips from multimodal conditions, yet they do not optimize how a product should be transformed into an effective advertisement or how future generation should be improved from online business feedback. To close this loop, we propose AgenticGen, a reward-guided agentic framework that decomposes advertising video generation into two trainable reasoning stages, strategy selection and draft generation, thereby exposing optimization targets that online business feedback can supervise. AgenticGen learns a performance-based reward from accumulated online feedback and a complementary rubric-based reward aligned with human quality standards, then uses them to supervise policy optimization. DPO first moves the agentic policies toward online preferences, and GRPO further refines both stages with process and outcome rewards. Offline experiments validate the reward models and successive policy optimization. Online A/B experiments in the TikTok advertising system show that AgenticGen after DPO and GRPO improves CTR by 2.72%, CVR by 2.63%, and Advv by 9.61% over the SFT baseline.
Read the full abstract
Computer Vision
M2LG-DG: A Multi-modal Local-Global Domain Generalization Framework for Cross-site Major Depressive Disorder Classification
Classification models based on resting-state functional magnetic resonance imaging (rs-fMRI) often show lower performance at imaging sites not included during model development, which can limit their use in clinical settings. Domain generalization (DG) addresses this issue by learning representations from source sites that remain effective for unseen target sites. However, existing DG approaches for psychiatric disorder classification commonly rely on a single imaging modality and may not fully account for site-specific acquisition effects on the learned representation space. Subjects scanned at the same site share scanner hardware, acquisition settings, and preprocessing characteristics, which can cause representations to reflect acquisition conditions rather than diagnostic information. In this work, we present M2LG-DG, a source-only multimodal local-global framework for cross-site major depressive disorder (MDD) classification. The framework employs a dual-stream rs-fMRI encoder, where the global pathway models inter-regional dependencies through self-attention and the local pathway performs graph-constrained aggregation over functional connectivity-derived brain graphs. Imaging and non-imaging representations are decomposed into shared and private components and integrated through bidirectional cross-attention with a learned modality gate. A cross-site supervised contrastive objective forms positive pairs from same-class subjects acquired at different source sites, encouraging the fused representation to preserve diagnostic information across acquisition domains. On four held-out REST-meta-MDD sites, M2LG-DG achieves an AUC of 69.48% and exceeds the closest comparison method by 2.18 percentage points. Experiments on the Autism Brain Imaging Data Exchange (ABIDE) dataset further support its applicability to other psychiatric neuroimaging classification tasks.
Read the full abstract
Computer Vision
Integrating Unimodal and Vision-Language Representations in Latent Space for Multi-Label Chest X-Ray Classification
Multi-label chest X-ray classification has attracted considerable attention in recent years, with the effective use of visual representations and clinical semantic knowledge playing an important role. This study proposes a framework that combines unimodal representations from RAD-DINO with vision–language representations from BioViL-T for the classification of 14 labels in the MIMIC-CXR-JPG dataset. The RAD-DINO and BioViL-T embeddings and their combined representation are refined separately in latent space before being normalized and fused across the three branches. In addition to improving classification performance, the study aims to clarify the role of each embedding source and the degree to which they complement one another. Experiments show that RAD-DINO outperforms BioViL-T when used independently, whereas early fusion further improves the results, indicating that the two embedding sources contain complementary information. The best-performing model achieves a mean AUROC of 0.840 and an mAP of 0.467. Ablation analysis shows that hybrid fusion provides consistent and statistically significant improvements over early fusion when each embedding source is refined in latent space, suggesting that fusion effectiveness depends on the quality of the representation supplied by each branch. However, the study has only been evaluated internally on MIMIC-CXR-JPG; its generalizability to data from other healthcare institutions therefore remains to be validated. The source code is available at: https://anonymous.4open.science/r/mimic-report-C210/.
Read the full abstract
Robotics
Robots Influencing Humans to Reveal their Goals during Collaboration and Competition
We propose a unified strategy for fast goal inference in human-robot interaction. The core idea is to drive the human toward Critical Decision Points (CDPs)-states where competing human strategies prescribe different next actions and thus maximally reveal the goal. We formalise CDPs using a goal-conditioned policy divergence measure and incorporate them into a Receding-Horizon Planner that explores future action sequences while optimizing a cost function balancing task progress and information gain. We evaluate this approach in both a collaborative, fully observable cooking task and a competitive, partially observable hide-and-seek game, each in simulation and on real robots. In both scenarios, our method infers human goals more accurately and earlier than baseline strategies.
Read the full abstract
Computer Vision
CrossModalQA: A Cross-modal and Multi-hop Benchmark for Multimodal Retrieval-augmented Generation
Despite the strong capabilities of multimodal large language models (MLLMs), their parametric knowledge remains incomplete and difficult to update, motivating multimodal retrieval-augmented generation (RAG) to ground responses in external text and images. However, existing benchmarks face two major limitations: (i) they typically emphasize single-hop retrieval or reasoning over a small set of provided contexts rather than open-domain evidence discovery; and (ii) they provide fragmented coverage of cross-modal reasoning paths, leaving complex multi-hop and multi-image reasoning underexplored. In this paper, we introduce CrossModalQA, an open-domain benchmark for evaluating multimodal retrieval and reasoning over heterogeneous corpora. CrossModalQA contains 1,863 question-answer pairs constructed from 4,987 Wikipedia articles and 4,431 Wikimedia Commons images. It covers five complementary reasoning paths: vision-to-text, text-to-vision, vision-to-text-to-vision, multi-image intersection, and image-set reasoning. Every question requires retrieving and composing distributed textual and visual evidence, with an average reasoning depth of 3.50 hops. We construct the benchmark through multimodal knowledge graph-guided subgraph sampling and apply rule-based consistency checking and LLM verification to ensure multimodal dependence and traceable evidence. Extensive experiments demonstrate that existing multimodal RAG systems struggle to recover complete evidence chains and can underperform closed-book models when incomplete retrieval introduces distracting context. Further analysis reveals that complete cross-modal retrieval contributes more to answer accuracy than generator scaling, while multi-image retrieval and reasoning remain the primary bottlenecks limiting end-to-end performance.