Peer Revenue Past citations are no guarantee of future returns

Papers

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

All of arXiv

Monday 14 September

Software Engineering

The Future of Safety for SaMD

An artificial organ carries failure consequences on the scale of an aircraft or a reactor, but the software driving it is rarely held to the same standard. Teams building them rely on testing, which only reaches the failure modes someone thought of in advance. In a pump or controller that runs inside a patient for months, the dangerous cases are the ones nobody anticipated. Formal verification closes that gap. Applied to the device's software, it proves the code meets its specification for every execution that specification allows, and where a proof fails, it returns the exact input sequence that breaks it. The same methods already protect rail, aviation, and nuclear control systems, and they extend the IEC 62304 lifecycle that a manufacturer already follows rather than replacing it. In this paper, we explore how to apply formal verification to artificial organs, stage by stage, and what each technique actually guarantees about the device.
Read the full abstract

All authors Rhea Malhotra, Tanya Sharma, Krisha Patel, Satvika Sharma, Heena Purkait, Mehak Nehal Makhija, Aellison Cassimiro, Everett Hildenbrandt, Palina Tolmach, Jaidev Shastri

Machine Learning (Statistics)

Graph Matching Relaxations and Amortization for Supervised Graph Prediction

End-to-end Supervised Graph Prediction (SGP) requires a permutation-invariant loss to compare predicted and target graphs with arbitrary node orderings. Such losses typically involve a costly graph-matching problem. We first study three Optimal Transport relaxations of this problem and show, theoretically and empirically, that the Gromov-Wasserstein (GW) objective is the most suitable for SGP. Then, to avoid solving the resulting inner optimization for every training example, we propose to amortize the graph matching (node alignment) problem. For each training sample, the loss function leverages a transport plan provided by a parametric matcher based on the differentiable Sinkhorn algorithm applied on empirical node distributions. The graph prediction module and the matcher are jointly learned. We showcase the efficiency of this approach on toy and real world SGP problems of increasing complexity including a novel Mass-spectra to Scaffold task that we introduce.
Read the full abstract

All authors Federico Méndez, Paul Krzakala, Gabriel Melo, Charlotte Laclau, Rémi Flamary, Florence d'Alché-Buc

Quantum Physics

Noise robust self testing from genuine local operation shared randomness multipartite nonlocality Tests

Violation of the $N$-partite inequality introduced in Phys. Rev. Lett. 129, 150401 (2022) for genuine multipartite nonlocality under local operations and shared randomness (LOSR) rules out all causal-network models obtained by locally composing arbitrary resources involving at most $N-1$ parties, even when supplemented by shared randomness among all $N$ parties. Here, we demonstrate a device-independent self-test protocol for this. We further establish an analytic noise-robust self-test. For a Bell-score deficit $\epsilon$, we derive explicit $O(\sqrt{\epsilon})$ vector-norm bounds for the state and the corresponding measurement actions.
Read the full abstract

Human-Computer Interaction

Profiling Handwriting-Process Deviations in Developmental Dysgraphia: An Open, Normatively-Referenced Instrument

Online handwriting captured on digitizing tablets yields hundreds of kinematic, temporal, spatial, and dynamic features. These features are rarely organized into interpretable, reusable constructs, and, although standardized tests norm the handwritten product, no openly available instrument places a child's handwriting process relative to a verified-typical reference. We introduce and validate an open measurement instrument for profiling handwriting-process deviations in children with developmental dysgraphia from sentence-level online handwriting. The instrument comprises (i)~a literature-only vocabulary, fixed before any cohort analysis, organizing 136 online-handwriting features into 12 handwriting-process domains, and (ii)~an age- and sex-adjusted normative reference framework, fit on verified-typical children only, that turns those features into a 12-axis deviation profile with per-child bootstrap uncertainty. We establish its measurement properties on the DiaGraMo cohort (N=257 Czech children aged 8–12; 110 verified-typical, 147 with dysgraphia): the vocabulary is structurally coherent, and the reference is calibrated, parsimonious, and leakage-free. On the same cohort, five of twelve domains separate the groups at BH~$q<.05$ (Cliff's~$\delta$ +0.19 to +0.51), on spatial, temporal, and pen-orientation processes, and 96.5% of participant~$\times$~domain scores have a bootstrap CI narrower than one z-unit. No classifier is trained: the instrument reports uncertainty-aware deviation scores rather than a diagnostic label, and its outlier rate is not a diagnostic rate. The vocabulary, the analysis code, and an open reference implementation that scores new participants are all openly released, giving handwriting and dysgraphia researchers a reusable, validated instrument for situating individual children against a normative reference.
Read the full abstract

Information Retrieval

Benchmarking Embedding Models for ESG Data

The use of Environmental, Social, and Governance (ESG) data is fundamental for modern corporate accountability, sustainability reporting, and financial decision-making. Embedding models have emerged as a powerful approach for transforming unstructured ESG text into numerical representations suitable for downstream natural language processing (NLP) tasks. However, their effectiveness in these ESG-specific tasks has not been systematically studied. In this paper, we construct a benchmark dataset specifically tailored to the ESG domain. We benchmark fourteen models, both open-source and closed-source embedding models, comparing their performance with respect to retrieval, and Retrieval-Augmented Generation (RAG). The results demonstrate performance variations across different models, with Qwen3-based models achieving the highest overall performance. This study provides practical insights into which models are better suited for ESG RAG tasks.
Read the full abstract

All authors Motaz Saad, Veronica Cretì, Ivan Gentile, Kianna Kazemi, Antonella Longo

Machine Learning

On the role of the tokenizer in ECG transformer models

Tokenization determines both the physiological content presented to an ECG Transformer and the sequence over which attention operates. We compare eight tokenization strategies across Transformer, Informer, Reformer, and FEDformer on the nine-label CPSC2018 classification task. The input projection and principal backbone capacity are controlled to isolate the effect of token construction. Median-beat and HeartLang tokenization achieve mean macro-AUCs of 0.893 and 0.889 across the four backbones, compared with 0.822 and 0.824 for point-wise and patch-wise tokenization. Pooling the two physiology-aware representations yields an 8.2% relative improvement in macro-AUC. They also reduce mean sequence length from 1,250 to 158 tokens and mean peak training memory from 5.21 to 0.27 GB. The results show that aligning tokens with ECG morphology can improve both predictive performance and memory efficiency without increasing backbone capacity. The source code is available on https://github.com/LeeJarvis996/ecg_tokenizer.
Read the full abstract

All authors Jiawei Li, Fabio Bonassi, Johan Sundström, Thomas B. Schön, Antônio H. Ribeiro

Machine Learning

Single-condition neural solvers encode transferable response spaces for parametric differential equations

Operator learning for parametric partial differential equations (PDEs) typically builds global models over prescribed domains, requiring cross-condition data or costly physics-constrained training. Here we show that the output Jacobian of a neural solution model trained at one condition defines a reusable response space for cross-condition solution variations. We introduce Linearized Subspace Transfer (LST) to exploit this space and recover target solutions by minimizing the target PDE-system residual over response-space coordinates. Because any single response space has finite coverage, Active Transfer Modeling (ATM) uses post-transfer residuals as coverage indicators to selectively acquire response spaces from additional single-condition models. Across six systems, single-condition response spaces supported cross-condition transfer, with enrichment improving accuracy when added spaces expanded representation capacity. Relative to evaluated physics-informed operator baselines, ATM reduced error and offline construction cost, with orders-of-magnitude accuracy gains in representative cases and millisecond-to-second target adaptation. These results establish neural solvers as reusable local parametric models.
Read the full abstract

High Energy Physics (Phenomenology)

Probing Neutrino-Energy Reconstruction with New Superscaling-Based Observables

In this study, we explore the superscaling variable $\psi'$ within the context of neutrino-nucleus charged-current quasielastic interactions. Building on the neutrino-oriented definition of the superscaling variable $\psi'$, we reparametrize it in terms of the neutrino-energy reconstruction bias, thereby establishing a direct conceptual and analytical connection between the superscaling formalism and neutrino-energy reconstruction. This approach allows the hadronic part of the reconstructed $\psi'$ to be expanded around the quasielastic point, yielding new observables that depend only on the outgoing-muon kinematics. At the same time, the resulting distributions remain sensitive to the underlying nuclear dynamics and vary across nuclear models. Among them, a newly identified quantity $L_1$, constructed solely from the outgoing-muon kinematics, provides a nuclear-physics constraint on the neutrino-energy reconstruction bias. This observable largely inherits the scaling properties of $\psi'$ across different neutrino fluxes and nuclear targets. The nuclear-model dependence is investigated using ND280-like $\nu_\mu$ and $\bar{\nu}_\mu$ Monte Carlo samples on carbon, generated with several nuclear models implemented in NEUT, NuWro, and GENIE, while the stability across experimental conditions is tested using additional carbon- and argon-based flux-target configurations.
Read the full abstract

All authors Lounès Amziane, Soniya Samani, Lorenzo Giannessi, Federico Sánchez

Atomic Physics

Second-order Rayleigh-Schrödinger Perturbation Theory for the Grasp2018 Package: Core-valence Correlations

The General Relativistic Atomic Structure package [Grasp2018, C. Froese Fischer, G. Gaigalas, P. Jönsson, J. Bieroń, Comput. Phys. Commun. (2019), DOI: 10.1016/j.cpc.2018.10.032], is based on multiconfiguration Dirac-Hartree-Fock and relativistic configuration interaction (RCI) methods for energy structure calculations. Atomic state function used in the program is built from the set of configuration state functions (CSFs). The valence-valence, core-valence and core-core correlations are explicitly included through expansions over CSFs in RCI. We present a combination of RCI and the stationary second-order Rayleigh-Schrödinger many-body perturbation theory in irreducible tensorial form to account for electron core-valence correlations when an atom or ion has any number of valence electrons. This newly developed method, which offers two ways of use, allows a significant reduction of the CSF space for complex atoms and ions. We also demonstrate how the method and program works for energy structure calculation of Cl III element.
Read the full abstract

Materials Science

Effect of Near-surface Thermal Spikes on Radiation Hardness of Gallium Oxide

Gallium oxide (Ga$_{2}$O$_{3}$) stands out as an extraordinary high radiation-tolerant semiconductor, because its lattice displacements induce polymorph transitions, holding material crystalline instead of leading to amorphization. Meanwhile, under extremely severe irradiation conditions many crystals become amorphous, often starting from surfaces where the translation symmetry breaks. Here, we show that the surface amorphization may prevail over the crystallisation in Ga$_{2}$O$_{3}$, however only if the mass and energy of irradiated ions produce sufficiently dense heat spikes in the immediate vicinity of the free surface. Applying machine-learned molecular dynamics simulations together with experimental broad-beam and focused ion-beam irradiations, we conclude that the presence of the free surface enables asymmetric displacements of Ga and O atoms, leading to a local non-stoichiometry. Consequently, when the affected cascade volume is sufficiently large, this compositional imbalance suppresses recrystallization and promotes amorphization. As such, our results are ready to use for tailoring irradiation conditions to either prevent or induce surface amorphization, depending on the requirements of the intended applications in Ga$_{2}$O$_{3}$ or other compound semiconductors.
Read the full abstract

All authors Tomás Fernández Bouvier, Umutcan Bektas, Alexander Azarov, Ru He, Nico Klingner, René Hübner, Paul Chekhonin, Aleksi Leino, Kai Nordlund, Javier García Fernández, Andrej Kuznetsov, Gregor Hlawacek, Flyura Djurabekova

Statistical Mechanics

Shannon Entropy as an Order Parameter for the Two-Dimensional Confined Coulomb Systems: Exact Balance Law, Topological Charge Sum Rule, and Boundary Saturation

We establish a complete information-theoretic framework for the minimum-energy configurations of the two-dimensional Thomson problem on a hard-wall disk, based on the Shannon entropy of the Voronoi topological charge distribution. Two exact identities are proven and verified to machine precision for all studied systems of size $N=12-10^5$. First, an entropy balance law decomposes the total entropy into bulk, boundary, and mixing contributions without approximation. Second, a topological charge sum rule fixes the total charge at twice the number of boundary particles plus six; together with the disjointness of bulk and boundary coordination sets $-$ justified analytically via the Delaunay empty-circumcircle criterion and elastic disclination splitting $-$ it relates the three- and four-coordinated boundary populations directly to the bulk charge. Applying Euler's theorem to the outermost Delaunay annulus yields an exact shell-mismatch identity: the four-coordinated boundary population equals the adjacent interior shell population, whereas the three-coordinated population equals their step mismatch, directly detecting third-shell nucleation at $N=30$. Furthermore, we prove that the boundary entropy is bounded by $\ln2$ and that its saturation is equivalent to the bulk charge per boundary particle approaching $-1/2$, a regime restricted self-consistently to boundaries of at least $12$ particles. Numerically, we locate bulk nucleation at $N=56$, maximal global disorder at $N=146$, and boundary saturation above $N\sim10^3$, with a direct extension to spherical geometry.
Read the full abstract

2609.15428 · 14 Sep · Article: 34 pages, 4 figures; Supplemental material: 12 pages, 1 table. v3: Added Sec. III.E (Theorem III.4 and Figs. 1, 2) providing the analytical justification of bulk-boundary coordination disjointness and the exact annular shell-mismatch identity; added Remarks II.1, II.2, and III.5; expanded bibliography

Computer Vision

A Conservative OCR-Enabled Workflow for R214 Sodium Screening of South African Packaged Foods

Using food package images to monitor sodium and salt content against South Africa's R214 sodium limits is challenging when screening decisions require product identity, nutrition facts panel evidence, reporting basis, and category-specific thresholds. This study presents a conservative image-based workflow that combines region detection, optical character recognition (OCR), product identity and sodium evidence extraction, R214 category assignment, deterministic threshold comparison, and independent vision language model comparison. The evaluation used 442 packaged food products and 3 929 full package images from a real-world South African food packaging dataset. A YOLO26s small detector generated 4 195 region crops, and strict post-processing produced one sodium evidence row per product. The integrated workflow produced 290 OUTSIDE R214 SCOPE, 139 REVIEW, seven SCREEN-PASS, and six SCREEN-FAIL outcomes. The independent Qwen2.5-VL 7B vision language model workflow produced 387 OUTSIDE R214 SCOPE, 31 REVIEW, twenty SCREEN-PASS, and four SCREEN-FAIL outcomes. The workflows agreed on exact R214 category assignment for 415 of 442 products (93.9%) and on whether the assigned category was within R214 scope for 416 of 442 products (94.1%). Final screening outcome agreement was 307 out of 442 products, or 69.5%. Manual verification on 60 products showed lower strict outcome agreement than regulated status agreement, while all manual INSUFFICIENT DATA cases were kept out of SCREEN-PASS and SCREEN-FAIL by both automated workflows. The findings show that conservative image-based screening can organise package evidence, identify clear cases, and assign uncertain cases to REVIEW rather than forcing SCREEN-PASS or SCREEN-FAIL decisions.
Read the full abstract

All authors Mayimunah Nagayi, Alice Scaria Khan, Tamryn Frank, Rina Swart, Clement Nyirenda

Human-Computer Interaction

Mapping justice as a plural and contested concept in human–computer interaction

Human-computer interaction increasingly studies and promotes justice. Often, this work does not treat justice through a pluralistic lens Justice, however, is not one thing: political philosophy offers several well-established, mutually incompatible conceptions of it. To navigate these meanings, we draw on a map of justice, a philosophical framework identifying six distinct conceptions of justice situated throughout (Western) philosophical thought. We apply this framework to four cases that illustrate different approaches to social justice: the COMPAS system, food democracy, gentrification, and robotics. Our analysis shows that each case already reflects a position on the map, but more importantly, one might adapt the case to other conceptions as well, leading to very different systems and analyses. We discuss how HCI might engage more carefully with such contested concepts, and warn that without pluralistic understanding, academic discourse risks a spiral of silence.
Read the full abstract

Cryptography and Security

An Empirical Security Analysis of Open-Source Software Used in Onboard Satellite Systems

The use of open-source software (OSS) in satellite flight systems is increasing as missions adopt reusable frameworks, shared libraries, and community-maintained components. While this accelerates development, it also introduces software-security risks into systems where patching is costly and failures may affect mission operations. This paper presents an empirical security study of OSS used in onboard satellite systems. We analyze 126 public repositories using a pipeline that combines software bill of materials generation, software composition analysis, static application security testing, infrastructure-as-code analysis, and secret scanning. After rule-based cleaning, onboard-scope filtering, and fingerprint-based deduplication, the pipeline produced a final dataset of 2,827 findings. The results show that security findings are widespread but unevenly distributed. Medium-severity findings account for 49% of the dataset, and 72% are classified as medium severity or higher. A Common Weakness Enumeration (CWE)-based taxonomy assigns all findings to eight weakness families. Memory Safety and Code Quality dominate the dataset, followed by Input Validation and Injection. Most findings occur in project-developed code, accounting for 81.4% of the dataset, while external dependency code remains a relevant source of findings. While these findings do not establish mission-specific exploitability, they provide an empirical characterization of recurring security patterns across the open-source onboard satellite software ecosystem, helping quantify their prevalence and prioritize areas that warrant the greatest security attention.
Read the full abstract

All authors Roee Idan, Tomer Cohen Galor, Asaf Shabtai, Yuval Elovici

Functional Analysis

Absolutely summing Toeplitz operators on weighted Fock spaces

We study absolutely $r$-summing Toeplitz operators $T_f:F_\varphi^p\to F_\varphi^q$ with complex symbols. The weight has a uniformly positive and bounded real Hessian. The main difficulty is that local distance to holomorphic functions does not detect an entire part of the symbol, while the Toeplitz operator does. Using the complete classification of absolutely summing diagonal maps, we characterize absolutely summing Fock–Carleson embeddings. An IDA decomposition, a Rademacher argument, and local scalar recovery then detect the missing analytic part of a complex symbol. We obtain equivalent criteria in terms of the local $L^q$ mean, the complex ball average, and the Berezin transform for every $1\le p,q,r<\infty$. At $q=1$ the criterion is $L^{s_p}$, where $s_p=2p/(3p-2)$ for $p\le2$ and $s_p=1$ for $p>2$. We also obtain a sectorial-symbol criterion without an IDA assumption and prove sharpness by separated models.
Read the full abstract

Information Theory

Free and projective LCD codes

LCD codes over finite commutative chain rings are known to be free. However, if the ring is not indecomposable, there always exists an LCD ideal that is projective but not free. We prove that all two-sided LCD codes are projective. We generalise the characterisation of finite commutative Frobenius rings by means of the size condition of a code and its dual to the noncommutative setting. This result is applied to prove that group codes are LCD if and only if they are generated by a central idempotent.
Read the full abstract

Artificial Intelligence

Empirical Evaluation of Task-Based Permission Scoping Architecture for AI Agents

AI agents are provisioned the same as employee-owned hosts in many enterprise settings with a static credential set fixed at deployment which includes all permissions the employee role might ever need. Role-based access control made this compromise for human principals because scoping access per task was infeasible. For AI agents, the compromise leaves every credential standing exposed whether or not the current task uses them. These permissions can later be utilised by a compromised or misaligned agent. Prior work (Noyan, 2026) defined this as the task-context mismatch, and proposed a three-source permission architecture which includes role-based permission ceilings, a task permission classifier and policy-based prohibitions, together eliminating the exposure preemptively. The work released a 600-prompt labelled dataset to evaluate it. This paper presents that evaluation end to end by implementing the security gate; a fine-tuned RoBERTa-large encoder which matched few-shot trained Claude Haiku 4.5 on classification quality (macro-F1 0.881 against 0.886, precision 0.897 against 0.842, severity-weighted residual risk 0.63 against 1.12). The results show the trusted component does not need to scale with the agent it supervises, and the scalable-oversight margin for this control method is wide. We also propose an attack-surface elimination metric which shows the role ceiling alone closes 27.9% of the severity-weighted surface and adding the task classifier closes 84.4%. The gap displays security advantages of task-granular access control over role-granular, and AI agents are the first principal type for which the task-granular access control is enforceable because their tasks arrive as machine-readable text. The research establishes task-based access control as a measured, potentially deployable mechanism for reducing attack surface in agentic deployments.
Read the full abstract

Optics

A nonlinear hydrodynamic theory of ultrafast laser self-organization

Ultrashort laser pulses can induce order out of chaos. A femtosecond pulse melts a metal surface for about a hundred picoseconds; under photoexcitation with no net polarization axis, the resolidifying film nonetheless organizes into regular nanoscale patterns. Here we derive from first principles the nonlinear hydrodynamics responsible for it and show it accounts for the full phase diagram of morphologies from measurable laser and material parameters. Starting from the Navier-Stokes equations in the lubrication limit, the molten film obeys a damped Kuramoto-Sivashinsky equation that reduces to a Swift-Hohenberg formalism near threshold, with every coefficient fixed by the laser fluence, melt thickness, optical skin depth, and thermocapillary response. The ratio of melt depth to optical skin depth sets the pattern wavelength, while the inter-pulse delay, via post-pulse damping, gates the route from flat surface to ordered pattern to chaos. We validate the phase diagram (hexagonal cavity arrays, bistable mixed states, labyrinths) numerically and confront it with pulse-resolved experiments on two metals, Ni and Fe3Cr. The agreement between theory and experiment quantitatively links the microscopic laser-material response to the macroscopic pattern selected after many pulses. This order is self-limiting: the same corrugation the instability builds scatters the modes that build it, and simulations, with every scale fixed by material parameters, place the coherence breakdown near a roughness of 30 nm after a few tens of pulses, a finite window confirmed by pulse-by-pulse measurements of orientational order on the experimental surfaces; at stronger coupling, disorder does not destroy the pattern but saturates and pins it instead. This yields a closed, predictive description of laser-driven self-organization, from single-pulse melt dynamics to the arrest of nanoscale order.
Read the full abstract

All authors Jean-Philippe Colombier, Emeric Gandon, Quentin Fornasiero, Eduardo Brandao

Signal Processing

Data-Aided Variational Bayesian Inference for CSI Estimation over Doubly-Selective DCO-OTFS MIMO VLC Systems with Affine-Precoded Superimposed Training Sequences

An orthogonal affine-precoded superimposed training sequences (AP-STS)-based architecture is developed for the cyclic prefix (CP)-aided multiple input multiple output (MIMO) direct-current-biased orthogonal time frequency space (DCO-OTFS) visible light communication (VLC) systems relying on arbitrary transmitter-receiver pulse shaping. The data and pilot symbol matrices are affine-precoded (AP) and superimposed in the delay-Doppler (DD)-domain for each transmit light-emitting diode (LED), followed by the development of an end-to-end DD-domain relationship for the input-output symbols. At the receiver for each receiver photodiode (PD), the decoupled pilot and data symbol are extracted by employing orthogonal precoder matrices, which eliminates the mutual interference. Furthermore, a novel pilot-aided (PA) variational Bayesian inference (PA-VBI) technique is conceived for the channel state information (CSI) estimation of MIMO DCO-OTFS VLC systems based on the expectation-maximization (EM) technique. Subsequently, a data-aided (DA) variational Bayesian inference (DA-VBI)-based joint CSI estimation and data detection technique is proposed, which beneficially harnesses the estimated data symbols for improved CSI estimation. Moreover, the Bayesian Cramer-Rao lower bounds (BCRLBs) are also derived for MIMO DCO-OTFS VLC systems. Finally, simulation results demonstrate that the proposed method yields superior performance in terms of normalized mean-square-error (NMSE), pilot overhead, and symbol error-rate (SER).
Read the full abstract

Algebraic Topology

Integral cohomology of classical Artin groups: spherical types $A,C,D$ and affine types $\widetilde B,\widetilde C,\widetilde D$

We determine the additive structure of the integral cohomology groups with trivial coefficients of the Artin groups in the spherical families $A_n,C_n,D_n$ and the affine families $\widetilde B_n,\widetilde C_n,\widetilde D_n$. We first compute the integral cohomology of braid groups by algebraic methods and then use the Salvetti complexes to express the cohomology of the remaining families in terms of these computations. The resulting descriptions apply in every admissible rank. For each family, we establish cohomological stabilization and show that the integral cohomology groups contain no elements of order $p^2$ for any prime $p$. We also give uniform generating series formulas describing the stable integral cohomology in terms of that of the braid groups.
Read the full abstract

Computer Vision

ViCo-SAM3: Vision-Conditioned Alignment for Open-Vocabulary Camouflaged Object Segmentation

Open-vocabulary camouflaged object segmentation (OVCOS) aims to segment unseen camouflaged objects under text guidance. We observe that SAM3 still suffers from a pronounced semantic gap between global textual semantics and fine-grained pixel-level visual cues in OVCOS. Meanwhile, fully fine-tuning the text encoder introduces heavy parameter overhead and risks overfitting to training categories, which compromises open-vocabulary representation flexibility. To address these issues, we propose ViCo-SAM3, a Vision-Conditioned alignment framework designed for OVCOS. Specifically, we introduce vision-conditioned (ViCo) module, which dynamically modulates text embeddings with global visual context, enabling textual representations to adapt to the current image content and thereby effectively bridging the semantic gap between vision and text. Building on this, we further design a vision-conditioned cross-modal binding (ViCoBind) module to enhance cross-modal interaction and semantic alignment between visual and textual representations. Without bells and whistles, ViCo-SAM3 achieves state-of-the-art performance on the OVCamo benchmark and demonstrates strong generalization.
Read the full abstract

All authors Qiangqiang Zhou, Wenjun Tang, Yong Chen, Dandan Zhu, Jiawei Xu

Materials Science

From bending to stretching driven peeling of heterogeneous adhesives

We study theoretically the peeling behavior of adhesives. Adopting a fracture mechanics approach, we derive the equation of motion of the adhesion front propagating at the interface between the adhesive and the substrate from which the peel strength is inferred. The originality of our approach lies in the description of the interplay during peeling between the stretching and the bending modes of deformation of the adhesive that is described as a F{\”o}ppl-Von Karman's thin film. Considering first a straight adhesion front, we retrieve the most salient feature of homogeneous adhesives, namely a peeling angle dependent peel strength driven by bending at large angles and by stretching at low angles. We also derive the shape of the adhesive that can be described using a single bending length scale derived from our model. We then investigate the impact of adhesion heterogeneities. We evidence that the deformations of the adhesion front are governed by a non-local interface elasticity the strength of which decreases with the peeling angle. This phenomenon reflects the transition between a stretching dominated peeling at low angle to a bending driven peeling at large angles that is captured in our model. This transition impacts the stability of adhesive fronts that relaxe more slowly from perturbations and gives rise to a stronger toughening effect in presence of a disorder distribution of adhesion energy at low peeling angles. Overall, this study sheds light on the central role played the elastic deformations of adhesives on their peeling behavior. The proposed framework unfolds the complex interplay between the deformation of adhesives and the peeling driving force that may be leveraged to engineer heterogeneous adhesives with enhanced properties. It also provides rich insights on the mechanisms underlying the emergence of non-local elasticity in interface problems.
Read the full abstract

Cosmology and Nongalactic Astrophysics

Influence of the large-scale structure velocity field on halo mass assembly within cosmic filaments

The cosmic web governs the growth of dark matter halos through anisotropic matter flows, yet the role of local velocity-field dynamics in halo mass assembly remains poorly understood. The velocity field carries non-linear information about matter transport that is not fully captured by density alone, particularly within filaments. We investigate this connection through velocity divergence and vorticity using dark matter halos in the IllustrisTNG TNG50-1-Dark simulation at $z=0$. Cosmic filaments are extracted with the 1-DREAM framework, and we statistically study how halo mass relates to divergence, vorticity, and radial distance from filament spines across three overdensity regimes. The velocity field correlates strongly with the cosmic web, with converging flows tracing filaments and vorticity arising primarily in overdense, non-linear regions. Within filaments, both divergence and vorticity show enhanced kinematic activity at intermediate distances from the spine, with amplitudes increasing with overdensity. We confirm a clear mass segregation, with more massive halos preferentially located near filament centers. Most notably, halos with masses $M \geq 10^{12} M_\odot$ are found almost exclusively in regions of low absolute divergence and low vorticity, while highly dynamical environments are dominated by low-mass halos. We further find that halos residing in kinematically calm regions assembled later and maintain higher recent accretion rates than those in dynamically active environments, linking present-day velocity-field properties to halo growth histories. These results demonstrate that the local kinematic state of the cosmic flow encodes information about halo mass and assembly beyond that contained in density alone, providing a complementary dynamical description of halo growth within the cosmic web.
Read the full abstract

All authors Dannaé Díaz-Axtle, Petra Awad, Divya Rana, Henk Hoekstra

Geometric Topology

The homology cobordism group is not generated by graph homology $3$-spheres

We show the homology cobordism group of homology $3$-spheres is not generated by graph homology $3$-spheres. Our technique is a combination of filtered instanton Floer theory and Gromov's simplicial volume.

Combinatorics

Large Complete Minors from a Cheeger Condition

Let $G$ be a graph on $n$ vertices, and let $e_G(S,V(G)\setminus S)$ be the number of edges with exactly one endpoint in $S$. The Cheeger constant and the restricted Cheeger constant of $G$, where $k\ge1$ is real, are, respectively, \[ h(G)=\min_{\substack{\emptyset\ne S\subseteq V(G)\\|S|\le \frac{n}{2}}} \frac{e_G(S,V(G)\setminus S)}{|S|} \text{ and } h_k(G)=\min_{\substack{\emptyset\ne S\subseteq V(G) |S|\le\min\{k,\frac{n}{2}\}}} \frac{e_G(S,V(G)\setminus S)}{|S|}. \] The contraction clique number $\ccl(G)$ is the largest integer $r$ such that $G$ contains the complete graph $K_r$ as a minor. Krivelevich and Nenadov [Complete minors in graphs without sparse cuts, Int. Math. Res. Not. IMRN 12 (2021) 8996–9015] proved that, for every fixed $\eps>0$ and all sufficiently large $n$ and $d$, if $G$ is a graph on $n$ vertices with maximum degree at most $d$, then $h(G)\ge\eps d$ and $h_{\eps n}(G)\ge(\frac{1}{2}+\eps)d$ imply $\ccl(G)=\Omega_\eps(\sqrt{\frac{nd}{\log d}})$. They asked to determine if one can guarantee the same lower bound on $\ccl(G)$ without the additional condition on $h_{\eps n}(G)$. They showed that this is the case when $d$ is a constant. We answer this question affirmatively. For every $\eps>0$, there are constants $\beta=\beta(\eps)>0$ and $n_0=n_0(\eps)$ such that, whenever $d\ge 3$ is an integer, for every graph $G$ with $n\ge n_0$ vertices and maximum degree at most $d$, if $h(G)\ge\eps d$, then $\ccl(G)\ge\beta\sqrt{\frac{nd}{\log d}}$. The dependence of this lower bound on $n$ and $d$ is best possible up to a constant factor. As a corollary, a lower bound is derived for the contraction clique number of $d$-regular graphs for which the second largest eigenvalue is bounded away from $d$, compared to earlier $\frac{d}{2}$. The proof combines spectral properties of graphs with an analysis of lazy random walks.
Read the full abstract

High Energy Physics (Phenomenology)

A DM candidate indicated at Fermi-LAT and LZ ? Connection with LHC and LC prospects

Recently an analysis of Fermi-LAT data has claimed a significant excess of energetic photons centred around 20 GeV, in the halo of our galaxy but absent in dwarf galaxies. This excess can be interpreted as coming from dark mater (DM) objects with a mass of few 100 GeV which annihilate into hadronic jets containing photons mainly from neutral pions. This observation is independently backed up by a completely independent observation of the LUX-ZEPLIN (LZ) detector of a DM candidate in the same region of mass. To avoid contradiction with upper limits coming from dwarf galaxies, one is led to assume the presence of a spin 2 resonance coupled to these objects, which explains this difference as due to different speed distributions in the two regions. This work recalls that LHC data indicate a tower of Kaluza Klein graviton resonances. This solution is at variance with a SUSY interpretation of DM and therefore does not necessary call for a 1.1 TeV Higgsino, as often proposed. If these KK resonances are coupled to e+e-, as indicated by LHC data, an e+e- collider could observe these DM objects. LHC could confirm our KK graviton interpretation and, eventually, detect the DM particles. This would allow to define de energy needed at a future e+e- collider.
Read the full abstract

Dynamical Systems

On the Hausdorff dimension of the set of non-escaping points in the Julia set

For an entire function $f$ and a non-zero complex number $\lambda$, let $f_\lambda(z)=f(\lambda z)$. We give conditions on $f$ which imply that the Hausdorff dimension of the set of non-escaping points in the Julia set of $f_\lambda$ tends to $1$ as $|\lambda|\to 0$. In fact, we give an upper bound for this dimension in terms of $\lambda$. This generalizes earlier results concerned with the case that $f(z)=\exp z$.
Read the full abstract

Quantum Physics

Exact quantum circuits for a Heisenberg triangle with Dzyaloshinskii-Moriya interaction

We construct short analytical quantum circuits for the exact evolution of a spin triangle with Dzyaloshinskii–Moriya (DM) interaction. We consider pure DM interaction with arbitrary real couplings and two Hamiltonian families with uniform Heisenberg exchange. For pure DM, a change of basis reduces the evolution to two single-qubit rotations. The circuits reproduce the full evolution operator at arbitrary time with a fixed gate count and admit implementations with both triangular and linear qubit connectivity. Experiments on a superconducting IBM quantum processor demonstrate directional quantum-state transfer and reproduce the amplitude and phase of transverse magnetization. In these experiments, the exact circuits reduce distortions relative to circuits based on the second-order Trotter formula. On a 12-spin kagome cluster, exact triangle blocks reduce the gate count at a prescribed accuracy. We also investigate the reduction of coherent errors by adjusting circuit angles and applying echo pulses to neighboring qubits.
Read the full abstract

Analysis of PDEs

Hilbert Expansion of the coupled radiation-Euler in the equilibrium regime

In this paper, we establish the validity of the Hilbert expansion for the coupled radiation-Euler model with non-relativistic source term for general initial data on the torus $\mathbb{T}^{3}$, which leads to the simplified equilibrium-diffusion limit and the initial layer corrections for the radiation intensity coupled with the temperature.
Read the full abstract

High Energy Astrophysical Phenomena

The effect of pressure confinement on the maximum mass of rapidly accreting supermassive stars

We derive an extension of Chandrasekhar's pulsational stability criterion for relativistic stars with a pressure boundary condition at the surface. We then apply the new criterion to hylotropes, a model describing rapidly accreting supermassive stars with a core-envelope structure. We find that confinement typically stabilises stars, delaying the onset of the general relativistic instability when the external pressure at the surface is roughly $\sim10^{-3}$ of the stars' hydrostatic pressure scale. Beyond this value, the critical mass at marginal stability increases rapidly, exceeding the value of $\sim 10^5$ M$_{\odot}$ for helium burning, isolated $n=3$ polytropes by orders of magnitude. We discuss how the required pressure boundary conditions may be produced by the combined ram and magnetic pressure supplied by the accretion flow as it assembles the supermassive star, and provide a fitting formula for the final mass before collapse. As a rough estimate assuming equipartition between magnetic and thermal pressures in a standard $\alpha$-disk, we find that the critical mass doubles at accretion rates of order $\dot M\sim100\,{\rm M_\odot}\,{\rm yr}^{-1}$, again increasing rapidly after this value. Our results show under what conditions, potentially realized in a subset of high redshift halos, non-rotating SMS may collapse into black holes beyond the typical scale of heavy seeds.
Read the full abstract