Peer Revenue Now accepting positions

Papers

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

All of arXiv

Saturday 7 November

General Relativity and Quantum Cosmology

The Energy-Momentum Tensor of a Compact Object in General Relativity to all Multipole Orders

In effective field theory approaches to compact object motion in general relativity, spin and tidal effects are encoded by operators formed from the Riemann tensor and its derivatives. Many such operators contribute degenerately to observables, with a single one often affecting several multipole moments of the body. Using the Mathisson–Papapetrou–Dixon formalism, we construct a natural basis of independent tensors which are each canonically conjugate to precisely one multipole. We find the novel exact distributional formula for the energy-momentum tensor of a compact object in terms of its reduced multipole moments and these tensors.
Read the full abstract

Metric Geometry

Euclidean view-obstruction in dimensions eight to ten and the asymptotic deficit

The Euclidean view-obstruction problem asks for the critical radius of balls centered at the half-integer lattice that meet every positive ray from the origin. Writing $\Delta_n$ for its square, we determine $\Delta_8=8/17$, $\Delta_9=1087/1980$ and $\Delta_{10}=7/11$ by computer-assisted proofs using Kahlon's seven-dimensional theorem.The upper bounds combine full-support subspace averaging, projected lattice coverings and exact finite certificates; dimension ten also requires a finite search through short integer relations. Explicit directions attain all three values, including a ten-dimensional direction previously listed by Gritsenko, Skoruppa and Zagier. In arbitrary dimension, we express their theta-block invariant in geometric normalization and derive the known bounds $c\log(n+1) \le n/12 - \Delta_n \le C\log^3(n+1)$, with repeated coordinates allowed throughout. Thus $\Delta_n \sim n/12$, while the deficit tends to infinity more slowly than every positive power of $n$. The supplementary archive contains the executable verifiers, exact certificates and archived verification records.
Read the full abstract

Machine Learning

EEGFaceSem: An EEG Benchmark with Paired Generative Latents for Semantic Visual Modeling

Electroencephalography (EEG) is widely used to measure brain activity for brain-computer interface applications for its non-invasive nature and high temporal resolution. However, progress in modeling the neural responses to visual semantics has been limited by a reliance on fixed image collections that lack the fine-grained control needed to isolate fine-grained changes in visual features and their relation to perception. In this work, we present EEGFaceSem, a large-scale dataset pairing high-quality EEG with photorealistic face images synthesized from a generative model. We provide the generative latent vector for each stimulus, enabling research into the continuous mapping between a visual latent space and neural representations. Each participant's brain responses were recorded while they attended to specific semantic features (e.g., smiling, hair color) in a rapid serial visual presentation paradigm. The dataset contains EEG signals from 30 participants, totaling 67,200 time-locked EEG recordings. We present benchmarks for single-trial semantic classification under single-subject, cross-subject, and subject-adapted settings, as well as for cross-subject voting that aggregates per-stimulus predictions across participants. We further demonstrate the dataset's unique utility with two complementary applications: deriving semantic directions in the latent space for brain-supervised image editing, and conditioning a generative model on brain-decoded labels for brain-supervised image generation. Our dataset paves the way for advancing brain-computer interfaces (BCI) for semantic-level visual saliency detection and brain-supervised generative modeling. Our dataset and code are openly released at OSF (https://osf.io/2guk3/) and at HuggingFace https://huggingface.co/datasets/yefllower/EEGFaceSem.
Read the full abstract

All authors Jun Ma, Michiel M. Spapé, Keith M. Davis III, Tuukka Ruotsalo

Soft Condensed Matter

A new approach to measuring both the steady flow curve and the planar extensional viscosity of complex fluids using exponential shear

We revisit the protocol known as Exponential Shear (ES), first proposed by Doshi and Dealy in 1986. This is a transient strong flow that can be readily imposed in the laboratory using commercial strain-controlled (separate motor and transducer) rheometers which accurately measure both the (exponentially) time-varying torque and normal force exerted by the sample. Numerous studies over the years have proposed competing definitions of an appropriate material function for reporting data obtained from the exponentially time-varying imposed deformation. We revisit this unresolved challenge and carefully reconsider the exponential material kinematics involved and the proper definition of the corresponding material functions. We show that by understanding the time-varying eigenvalues and eigenvectors of both the shearing kinematics and the evolving state of stress in the sample it is, in fact, possible to use an exponential shear test to quantitatively map out both the full flow curve (of shear viscosity vs. shear rate) and the steady planar extensional viscosity curve as a function of deformation rate. These findings are supported by numerical simulations of a number of nonlinear viscoelastic constitutive models, asymptotic analysis, as well as experiments with several polymeric liquids.
Read the full abstract

All authors Nicholas King, Eugene Pashkovski, Reid Patterson, Gareth H. McKinley

Systems and Control

Real-Time Threat Classification in Multi-Agent Systems via Topology-Coupled Hidden Markov Models

Networked multi-agent systems coordinating in adversarial environments face compound cyber threats that alternate between false-data injection (FDI), denial-of-service (DoS), and stealthy decoy perturbations designed to mimic benign operational anomalies. Because the switching policy is latent and the plant dynamics evolve continuously, deterministic threshold detectors struggle to isolate deception without causing excessive false alarms. This paper presents an online, data-driven framework that couples distributed continuous time unknown input observers (UIOs) with a topology-aware Hidden Markov Model (HMM) for real-time attack detection and classification under unknown adversarial strategies. By decoupling FDI signals from augmented node states, distributed observers deployed across the network generate local augmented residuals that are discretized into emission sequences. The HMM embeds the underlying directed communication graph into the off-diagonal blocks of its transition probability matrix, explicitly capturing inter-agent attack propagation. Both transition and observation probabilities are updated recursively online via an Expectation-Maximization (EM) scheme, adapting the classifier to evolving threat behaviours without prior offline calibration. Numerical simulations on a directed network demonstrate that incorporating graph topology into the latent transition dynamics resolves the severe classification ambiguity between nominal and decoy modes, outperforming decoupled agent-independent decoders while restricting observer deployment to a sparse subset of nodes.
Read the full abstract

Robotics

Passivity-Guaranteed Variable Impedance and Compliance Control via Online Parameter Optimization

Variable impedance control enhances the interaction capability of robot manipulators by adapting impedance parameters online. However, time-varying impedance parameters may inject energy into the system and violate passivity, particularly when tasks demand rapid stiffness changes. Existing methods address this by increasing damping or limiting the stiffness profile, which often degrades the dynamic response. In this work, we propose a quadratic programming-based parameter optimization framework that guarantees passivity by modulating the desired inertia to compensate for energy injected by stiffness variations, while preserving a critically damped system. For the case of diagonal impedance matrices, we optimize the rates of change of inertia and stiffness subject to a passivity constraint formulated as a convex quadratic program. We extend the framework to compliance control, where the desired inertia is coupled to the robot's configuration-dependent inertia matrix. We use an eigendecomposition based on Cholesky factorization to optimize the spectral weights while maintaining passivity. Experiments on the Kinova Gen3 7-degree-of-freedom manipulator validate both methods with human interaction perturbations. Comparisons with an energy tank approach and a nonpassivity-based method demonstrate that the proposed framework achieves less conservative stiffness tracking without requiring an energy reservoir or damping overhead.
Read the full abstract

All authors Huseyin Tugcan Dinc, Ersin Daş, Nikolas Thelenberg, Christian Ott

Combinatorics

On maximum volume polytopes inscribed in the sphere

We combine degree sequence bounds with a supporting tangent plane method to study maximum volume polytopes inscribed in the sphere $\mathbb S^2$. Building on the vertex-star estimates and combinatorial reductions in Wimmer's dissertation, we determine the exact maximum for ten vertices and the unique maximizing combinatorial type for eleven vertices. Moreover, using explicit computational certificates we identify a unique possible maximizing type for polytopes with thirteen and fifteen vertices, and two possible types for sixteen vertices; this approach also recovers Wimmer's fourteen-vertex combinatorial theorem. Finally, we derive a facial inequality which holds for every degree pattern, and prove that the maximum volume for polytopes with thirty-two vertices is $\frac{2}{3}\sqrt{10(5-\sqrt{5})}$, attained precisely by an inscribed realization of the pentakis dodecahedron, which is the convex hull of the regular icosahedron together with its normalized facet centers. Wimmer had already determined this configuration and its optimality within its own combinatorial type. We also obtain a global degree sequence envelope and a universal affine volume bound.
Read the full abstract

Software Engineering

SIExBench: A Benchmark for Automated Repair of Sensitive Information Exposure

Modern software systems routinely process credentials, personal records, and other sensitive information. A sensitive information exposure (SIEx) defect makes such information accessible to an unauthorized actor and can cause serious and often irreversible disclosure. However, prior research has focused largely on detecting such exposures rather than repairing the underlying defects, and existing repository-level issue-resolution benchmarks do not specifically target this defect class. We present SIExBench, to our knowledge the first benchmark dedicated to automated repair of SIEx defects. SIExBench contains 161 execution-verified instances from 64 widely used Python and Java repositories, each with an issue report, buggy repository snapshot, developer patch, executable exposure and regression tests, and defect characterization. We evaluate six repair systems with three LLMs, yielding 18 configurations, and systematically analyze their overall effectiveness, performance across defect characteristics, and failure mechanisms. The strongest configuration resolves only 39.1% of the instances, while 40.4% remain unresolved by all 18 configurations. More importantly, 62.1% of unsuccessful attempts produce a valid patch that still leaves sensitive information exposed, making incomplete exposure removal a central challenge. Systems often generate plausible patches and modify relevant code without fully eliminating the exposure, indicating that effective repair must ensure that all relevant paths to sensitive-information exposure are closed. Overall, SIExBench provides a foundation for systematically evaluating and advancing automated repair of sensitive information exposure defects.
Read the full abstract

All authors Mingming Zhang, Xu Wang, Jian Zhang, Dihao Fan, Xiangxin Meng, Chunming Hu

Differential Geometry

The J-twist of the Dirac operator, spectral Einstein functionals and the noncommutative residue

We study spectral Einstein functionals associated with the $J$-twist of the Dirac operator on almost product Riemannian spin manifolds. On closed even-dimensional manifolds, we compute the corresponding noncommutative residue in terms of the Einstein tensor, the curvature of the connection determined by $D_J^2$, and its zero-order endomorphism. For four-dimensional compact manifolds with boundary, we consider two boundary residue functionals obtained by inserting two covariant derivatives in the first factor of the compositions $\pi^+(D_J^{-2})\circ\pi^+(D_J^{-2})$ and $\pi^+(D_J^{-1})\circ\pi^+(D_J^{-3})$. Under the stated collar and almost product structure assumptions, we calculate their interior and boundary contributions and obtain Dąbrowski–Sitarz–Zalecki type formulas. We also present applications of these results in physics and noncommutative geometry, focusing on spectral actions, spectral distance, and mean-curvature boundary contributions.
Read the full abstract

Computer Vision

Same Measurements, Different Velocity Risk: Complementary Rolling-Shutter Readouts under Clock Uncertainty

Camera rigs with identical measurement budgets can have different velocity estimation rates when exposure timing is uncertain. We establish this distinction for rolling-shutter reconstruction with two calibrated moving anchors, unknown scene points, and independently unknown auxiliary poses and clocks. On the same fixed local physical class near a separated lateral rig, repeated readouts have worst-case squared velocity risk at least of order $\sigma\rho_{\mathrm{eff}}$, whereas complementary readouts have minimax risk of order $\sigma^2$, for a fixed admissible clock displacement $\rho_{\mathrm{eff}}>0$ as the noise standard deviation $\sigma$ tends to zero. A finite physical compensation path gives the lower bound; a nonlinear velocity inverse proves complementary stability. All motion, structure, pose, and clock nuisances remain unknown. We identify mounting configurations in which both designs become regular and quantify the risk floor induced by metric uncertainty. We quantify full-parameter domains with independent 6-ms clocks and test in-domain reconstruction under paired noise and a depth-offset intervention. Raster studies examine image formation and calibration sensitivity.
Read the full abstract

Lattice Field Theory

Constrained field-space estimators for background-field lattice QCD observables

Background-field lattice calculations extract weak-field response coefficients from correlators computed at several field strengths. We treat this extraction as linear estimation in field space and compare fixed cancellation stencils with covariance-optimized, generalized-least-squares estimators of the same coefficient, using archival nucleon double ratios on about 1000 quenched configurations at six quark masses and decuplet correlators for which exact answers follow from the lattice contractions. The fixed $O(B^4)$-$O(B^8)$ stencils are nearly coincident and cost almost nothing in precision. Weight-free null-space tests reject the two-term response model in most nucleon ensembles; whether they reject the three-term model depends on the temporal fits and on the noise of the estimated temporal covariance, which raises the null-contrast error by a factor of 1.7-2.4. With that noise included, covariance optimization gains 1-2% in precision at the five lighter masses (3-30% with two-state time fits, which also lower moments by 3-10%) and nothing on independent data halves, while its weights carry three to four orders of magnitude more leverage on omitted response modes than the fixed stencils and move the moments by up to 1.8 marginal standard errors. The decuplet controls show what such leverage does under misspecification: the first-order responses obey $\mu_{\Delta^+}/\mu_{\Delta^{++}}=1/2$ and $\mu_{\Delta^0}=0$ exactly. The lowest-order estimate misses the first by 10-25 standard errors. The fixed $O(B^4)$-$O(B^8)$ stencils satisfy it within errors in the production windows, whereas the correlated linear fit in $B^2$, with the same constraints as the $O(B^4)$ stencil, misses it by 3-17%; the null-space test rejects that fit's model. The results distinguish statistical precision from robustness to omitted response modes; they are not physical-point magnetic moments.
Read the full abstract

Machine Learning

Universal Approximation by Equivariant Message-Passing Graph Neural Networks with a Bilinear Readout

We prove universal approximation for the bilinear-readout equivariant message-passing network (BREMP), a reference architecture for equivariant graph neural network interatomic potentials. Its layers combine equivariant convolutions with affine maps, and its readout forms invariants from products of the equivariant features. We construct a one-layer feature that determines the local environment up to permutations of atoms of the same type. The construction uses spherical harmonic filters of degree no larger than the largest number of neighbors at a common distance from the center. Iterating the construction recovers the environment reachable in as many hops as there are layers, using the same bound on filter degrees at the layer cutoff. The readout approximates every continuous invariant function of the features uniformly on compact sets. Together these give universal approximation of atomic energies at a prescribed physical range using a shorter layer cutoff. We assume that, throughout the admissible class, message passing reaches all neighbors in each physical environment within a common number of layers. Finally, DPA4, an interatomic potential architecture in production use, inherits this result by approximating the BREMP layers constructed in the proof and realizing the bilinear readout.
Read the full abstract

Computer Vision

WorldFoundry: Towards Unified Infrastructure for World Intelligence

Visual generative models increasingly blur the line between synthesizing observations and simulating environments. Yet visual plausibility alone does not constitute successful world modeling: compelling outputs may lack persistent state, physical dynamics, or causally consistent responses to interactions. We define a generative world model as a predictive dynamical system that maintains a persistent, history-conditioned representation of an environment, updates that representation in response to actions or interventions, and generates temporally coherent trajectories that reflect outcomes under observed and counterfactual interventions. From this definition, we derive an intelligence-oriented framework spanning perceptual grounding, persistent state modeling, controllable transformation, dynamics-aware prediction, and closed-loop interaction. We present WorldFoundry, open-source infrastructure for reproducible inference and evaluation across video, 3D and 4D, embodied action, and interactive world systems. WorldFoundry unifies model and benchmark discovery, runtime and asset preparation, execution orchestration, artifact management, and interactive visualization, while modality-specific adapters preserve native inputs, execution semantics, output representations, and evaluation protocols. Explicit manifests, durable artifact contracts, and evidence-aware scorecards separate execution readiness from evaluative claims, making each claim traceable to its supporting artifacts. The catalog contains 270 model entries, 74 benchmarks, and 62 registered metric entries, with readiness and execution evidence recorded separately. WorldFoundry enables reproducible comparison of persistent, controllable dynamics across heterogeneous systems, advancing the field from perceptual synthesis toward measurable world intelligence.
Read the full abstract

All authors Juanxi Tian, Jiadong Hou, Baiang Li, Xin Lin, Zhengbo Jiao, Weijia Zhang, Yeying Jin, Truong Nguyen, Haitian Zheng, Zhifei Zhang, Jianfei Yang, Ming-Hsuan Yang, Felix Heide, Wei Xiong, Enze Xie, Wanhua Li

Computer Vision

How reliable is an embedding of the seabed? Repeat-look measurement of frozen encoders on synthetic aperture sonar

Pretrained encoders from self-supervised representation learning are increasingly applied to sonar imagery of the seafloor, where they are evaluated with protocols that take one embedding per target. A target patch of seabed, by contrast, is re-imaged from an alternate look on overlapping survey passes, so each embedding is one measurement of that target. Standard protocols cannot say how consistent those measurements are, or how much of their spread belongs to the patch and how much to the acquisition geometry. We treat this as a measurement-reliability problem. Using a synthetic-aperture sonar (SAS) survey, we measure the repeat-look reliability of embeddings from a frozen natural-image encoder (DINOv3 ViT-S/16). Standard single-look metrics, effective rank, clustering and a linear probe for coral, are all stable to the look, and the probe performs well (AUC 0.93). Yet the repeat-look reliability of the same embeddings is poor, with a distance-based intraclass correlation (dbICC) of 0.20 and nearest neighbours and clusters that barely agree across looks. This contradiction resolves as we find the probe learns a stable direction, whose ICC of 0.60 is 3.0 times that of the full embedding. Variance components, calculated from generalizability theory, attribute under a tenth of the within-cell spread to look acquisition geometry. The rest is cell-specific and cannot be separated without repeat passes at matched geometry. A sonar-pretrained encoder (BenthicDINO) and a classical SAS texture descriptor show the same pattern. For this dataset and the tested representations, uses that depend on patch identity fail, such as cross-pass comparison and retrieval, while uses at the level of habitat type (e.g. coral classification) do not. Because the standard metrics did not detect this, we argue that repeat-look reliability should accompany any evaluation of representations of the seafloor.
Read the full abstract

All authors Harold Horsley, Oscar Pizarro, Hayat Rajani, Morten Omholt Alver

Artificial Intelligence

Look Ahead: Search-Driven Reasoning in Embodied AI

Search allows a robotic policy to evaluate candidate futures before committing to an action. Motivated by test-time planning for vision-language action (VLA) models, we study selective search over action chunks in PushT using a frozen Action Chunking with Transformers (ACT) policy. Beam search generates perturbed alternatives around the policy prediction and scores them through rollouts from cloned simulator states. Across 50 evaluation episodes, beam search raises task success from 0% to 18% and increases average cumulative reward from 39.44 to 103.2. We introduce One Step Ahead (OSA), a gate that simulates the nominal policy chunk and invokes full search only when its predicted coverage drop exceeds a threshold. In the reported ablations, OSA preserves or improves success while reducing runtime by approximately 3x relative to full beam search. Ablations over tree depth, branching factor, beam width, and replanning frequency show that larger search budgets do not consistently improve performance. The results support selective test-time search for action-chunking policies; evaluation on language-conditioned VLAs and physical robots remains future work.
Read the full abstract

All authors Ahmed Zeer, Osama Shbib, Muzaffer Kaan Yuce, Himmet Toprak Kesgin, Mehmet Fatih Amasyali

Artificial Intelligence

BlenderFORGE: Framework for Optimizing Reactive 3D-Graphics Editing Ability of MLLMs

Multimodal Large Language Models (MLLMs) have shown promise as interactive agents, yet precise, programmatic 3D graphics editing remains difficult to train because existing systems largely rely on closed-source model APIs and lack deterministic, state-verifiable execution environments. We present BlenderFORGE, a trainable data-environment-optimization framework for open-weight MLLM agents that edit explicit Blender scenes through Blender Python (bpy) code generation. BlenderFORGE targets four fundamental state-verifiable editing task families: object placement, shape-key editing, lighting adjustment, and material modification. It integrates three components: (1) a scalable perturbation-based data generation pipeline built from curated assets and large-scale 3D scene repositories; (2) an isolated Blender Sandbox for deterministic code execution, structured state export, and closed-loop reward computation; and (3) a two-stage training pipeline that uses offline teacher trajectories for cold-start Supervised Fine-Tuning (SFT) and sandbox-computed rewards for multi-turn tool-agentic Reinforcement Learning (AgentLoop-RL). Experiments across the four task families show that the resulting open-weight Qwen3-VL-8B and Qwen3.5-9B BlenderFORGE agents substantially improve over their base models and achieve competitive performance with several zero-shot proprietary MLLM baselines on selected metrics. Additional cross-benchmark evaluations and camera-reframing experiments provide evidence of transfer beyond the original training setting.
Read the full abstract

All authors Zilin Guo, Tianrui Zhang, Yi Rong, Yichen Liu, Yuxin Guo, Jingcheng Ni, Lewei Lu, Zehuan Wu, Dan Xu

Spectral Theory

A new butterfly in the meadow: magnetic lattice graphs with a circulant vertex coupling

We analyze the spectrum of quantum square lattice graph exposed to a homogeneous magnetic field with a rational flux through plaquette, assuming all nontrivial cases of the vertex coupling which is circulant and scale-invariant, with a particular attention to those of them that give rise to a dynamics violating the time reversal invariance. Except for two cases with partially decoupled edges, the remaining twelve split into two spectral classes; the second one offers a hitherto unknown butterfly pattern.
Read the full abstract

Information Theory

New Classes of CCZ-Inequivalent APN Functions with Boomerang Uniformity One in Odd Characteristic: Differential and Boomerang Spectra

Let $q=p^n>3$ be an odd prime power and let $g:\F_q\to\F_q$ be an even perfect nonlinear (PN) function with $g(0)=0$. We construct almost perfect nonlinear (APN) functions $G_{\lambda,M}(x)=g(x)+M(x)+\lambda(x^{q-1}-1)$, where $\lambda\in\F_q^*$, $-\lambda\notin g(\F_q)$, and $M$ is an $\F_p$-linearized permutation polynomial. These one-point modifications of $g+M$ have boomerang uniformity one, the smallest possible value for APN functions in odd characteristic. We give complete differential and boomerang spectrum formulas and obtain closed-form evaluations for selected constructions based on Coulter–Matthews monomials and Dembowski–Ostrom polynomials. For $q>7$, comparison of differential spectra establishes CCZ-inequivalence to all APN monomial families listed in the paper and every APN binomial in the original Ness–Helleseth family or its extension by Zeng et al. We characterize CCZ-equivalence within our construction, showing that any two linearly inequivalent even PN functions over the same field yield CCZ-inequivalent APN functions. We obtain three pairwise CCZ-inequivalent APN functions with boomerang uniformity one over $\F_{p^{15\ell}}$ for every $\ell\ge1$, six over $\F_{p^{120r}}$ for every $r\ge1$, and unbounded numbers of such classes in characteristic three. We further prove that CCZ-equivalence and EA-equivalence coincide for functions at Hamming distance at most $m$ from a PN function, provided that $m(m+1)<q-q/p$. We also show that all our APN functions have the maximal possible algebraic degree $n(p-1)$. Finally, we exhibit both CCZ-equivalent and CCZ-inequivalent APN pairs with boomerang uniformity one at Hamming distance one.
Read the full abstract

Computation and Language

Cascading Circle Loss: Order-Preserving Contrastive Learning for Graded Legal Retrieval

This paper introduces Cascading Circle Loss (CasCL), a contrastive learning framework for legal retrieval tasks with graded relevance. CasCL combines an InfoNCE objective for positive-negative separation with a cascade of ranking boundaries that model ordered relevance levels between queries and legal clauses. The framework also includes a negative-aware self-paced optimization mechanism and a relevance gain weighting strategy to improve ranking quality. Experiments on legal passage retrieval benchmarks show that CasCL consistently outperforms existing methods, including InfoNCE, Circle Loss, MNRL and Margin MSE, across both binary retrieval and graded ranking metrics.
Read the full abstract

Computer Vision

BAPLe++: Cross-Model Transferability and Prompt-Level Defense of Backdoors in Medical Foundation Models

Medical foundation models (Med-FMs) are increasingly adapted to clinical tasks through prompt learning, which tunes a small set of prompt tokens while the backbone remains frozen. Recent work has shown that this efficiency introduces a vulnerability: jointly optimizing learnable prompts and an imperceptible noise trigger during few-shot adaptation implants a backdoor that leaves clean performance intact but steers triggered inputs to an attacker-chosen class. Since prompts are compact and readily shared, this raises two questions: whether a backdoored prompt remains effective on a different Med-FM, and whether it can be repaired without retraining the backbone. We address both in BAPLe++. First, we present the first study of cross-model transferability of prompt-based backdoors, quantified by two proposed metrics, Clean Accuracy Difference (CAD) and Backdoor Accuracy Difference (BAD). Second, we propose DePLe (Defense via Prompt Learning), which uses the infected prompt as a fixed anchor and learns a new prompt on a small clean set that preserves the clean task while diverging from the compromised embedding region. Experiments on four Med-FMs and six chest X-ray and histopathology datasets show that backdoors transfer across models, but asymmetrically and in a dataset-dependent manner. Compared with four training-free input transformations, DePLe reduces backdoor accuracy more consistently while preserving, and in several settings improving, clean accuracy. Our code can be accessed on https://github.com/Ufaqkhan/Baple-Plus-Plus
Read the full abstract

All authors Ufaq Khan, Umair Nawaz, Satyajit Tourani, Awais Rauf, Fahad Shamshad, Asif Hanif, Rao Muhammad Anwer, Muhammad Haris Khan, Mohammad Yaqub

Analysis of PDEs

Local turnpike property for a three-component mean field game system arising from a resource harvesting model

We study a three-component mean field game system arising from a resource harvesting model. It couples Hamilton–Jacobi and Fokker–Planck equations for the harvesters with a reaction-diffusion equation for the resource. The coupling is nonlocal in space and time. For standard second-order mean field game systems, exponential turnpike estimates are known under assumptions such as monotonicity of the coupling. For the present system, Kobeissi, Mazari-Fouquer, and Ruiz-Balet proved algebraic convergence to a stationary solution in a space–time averaged sense, but no exponential turnpike estimate has been obtained. We prove such an estimate near a stationary solution with strictly positive density components. Inspired by the relative entropy method, we add the relative entropy of the resource density to the standard energy for local turnpike estimates. In the time derivative of this modified energy, the nonlocal coupling terms cancel. The two key assumptions are coercivity of a quadratic form determined by the stationary resource density and smallness of the relative density perturbations in $L^\infty$. Under these assumptions, we bound the absolute value of the modified energy by two terms that decay exponentially away from the initial and terminal times. We derive weighted $L^2$ exponential turnpike estimates for the harvester density, the resource density, and the gradient of the value function. We give sufficient conditions for coercivity. For each fixed $T>0$, classical solutions satisfying these assumptions exist for initial and terminal data sufficiently close to the stationary solution. We extend the modified energy estimate to a more general reaction term under the corresponding assumptions.
Read the full abstract

Optimization and Control

Integral Compensation for Discrete-time Bearing-only Target Localization and Circumnavigation

This paper investigates the discrete-time control problem of driving an agent to localize and circumnavigate an unknown target at a prescribed radius using bearing measurements. Discretizing existing continuous-time circumnavigation controllers can introduce steady-state orbit bias, even at sampling rates that preserve stability. To eliminate the sampling-induced orbit bias, we propose an adaptive compensator for a circumnavigation controller formulated directly in discrete time. Driven by the estimated agent-to-target distance, the proposed compensator dynamically learns the geometric correction required to achieve the desired circumnavigation radius. Unlike existing static correction methods, neither the proposed controller nor the compensator update explicitly uses the sampling interval. Exponential convergence is rigorously established using Lyapunov-based analysis. Simulation results illustrate the performance of the proposed algorithms.
Read the full abstract

Dynamical Systems

Directional weak mixing and directional Kronecker for $\Z^2$-actions

Let $T$ be a measure preserving $\Z^2$-action on a Lebesgue probability space $(X,\mu)$ and let $L\subseteq\R^2$ be a $1$-dimensional subspace. For a rational subspace $L$, direction $L$ weak mixing corresponds to weak mixing for the measure preserving transformation $T^{\vv_0}$, where $\vv_0$ is a primitive vector in $\Z^2$ generating $L$. For an irrational direction $L$, we study two different definitions of direction $L$ weak mixing, and show that they are the same: “Dynamical” direction $L$ weak mixing is defined to be the absence of any direction $L$ Kronecker factors. These factors are defined in terms of almost periodicity along a finite width direction $L$ strip in $\Z^2$. “Spectral” direction $L$ weak mixing is defined as the absence of any direction $L^\perp$ line segments $I\subseteq\T^2$ with positive spectral measure. We also give a spectral characterization of the directional Kronecker property defined in \cite{liuxu}, and provide examples of $\Z^2$-actions $K$ that are direction $L$ Kronecker.
Read the full abstract

Machine Learning

ConwayNet: Interpretability, Activation Coherence, and Noise Resilience via a Game-of-Life-Inspired Lateral Inhibition in Artificial Neural Networks

Deep neural networks achieve universal function approximation by stacking interconnected nonlinearities, but as a consequence they are sensitive to noise and resist mechanistic interpretation. In contrast, the local update rules of cellular automata, such as Conway's Game of Life, give rise to not merely interpretable but programmable dynamics. We explore a bridge between these worlds: a lateral inhibition in neural networks based on continuous Game-of-Life rules that derives inhibitory and excitatory responses from the layer input and blends them into the activation map in a state-conditioned way. The mechanism is differentiable and architecture-agnostic, and its regulation is imposed rather than learned: the rule's parameters are fixed and are not optimised, while the rule's values and derivatives shape training throughout the network. Wired into fully connected, convolutional, wide-residual and graph convolutional networks, it gives rise to several properties. Activation maps become coherent across space, depth and seeds, with coarse structure persisting through depth while fine detail is renewed at every layer. The co-activation graph of the hidden units departs from an Erdős–Rényi random graph: it is laid out on the image and its giant component grows in steps, as in a spatial graph. Networks started from nearly identical initialisations, under label noise, disagree on thirty times fewer test images. Saliency maps rest on the object rather than the background, and resilience to pixel noise not seen in training is substantially improved.
Read the full abstract

Optimization and Control

A Bayesian oriented exploration and reinforcement learning in continuous-time mean-variance portfolio optimization

We propose a new reinforcement learning (RL) framework for the continuous-time mean-variance portfolio optimization with unknown market parameters. To better reflect the impact of improved information of the unknown drift on the degree of exploration, we consider a novel randomization based on continuous-time stochastic filtering. Specifically, we construct a continuous-time analog to Thompson Sampling by treating the unknown drift as a random draw from a generalized sampling distribution parameterized by the filter's posterior mean and variance. The new learning objective measures the tradeoff between the wealth performance and the weighted KL divergence between the policy and the true Bayesian posterior, which transforms the classical feedback control into an information guided stochastic policy. As a result, the degree of exploration endogenously depends on the Reward-Biased Maximum Likelihood Estimation that decays with the shrinking filter's variance. This allows us to devise an efficient model-based learning algorithm in the mean-variance portfolio optimization with theoretical guarantee and superior learning performance.
Read the full abstract

Computer Vision

Unified Framework with Convex-Concave Decomposition and Sparse Topological Learning for Skeleton-Based Clinical Assessment

Skeleton-based rehabilitation assessment faces fundamental limitations of fixed topologies which cannot model compensatory patterns, uniform joint treatments overlook clinically important deviations, and lack of geometric understand- ing. We propose Concave-Convex Decomposition Sparse Topological Graph Convolutional Network (CC2-STG), a multi-stream model that combines geo- metric decomposition, sparse feature learning, and adaptive topology. Our framework decomposes joint relationships into complementary convex-concave distance functions to model both local and global movement characteristics, with sparse regularization with proximal optimization selecting clinically relevant geo- metric features, while multi-scale topological analysis captures non-anatomical joint dependencies in pathological movements. Experiments on the KiMoRe and UIPRMD datasets validate the effectiveness of our approach, achieving state-of-the-art performance on clinical pathological movement assessment.
Read the full abstract

All authors Quang-Anh N. D., Duc Pham Minh, Minh-Anh Nguyen, Hien Quang Kha, Manh Hung Ha, Minhhuy Le, Oscal Tzyh-Chiang Chen

Medical Physics

Simultaneous 3D acoustic modulation and imaging of proton-induced nanodroplet vaporization for proton therapy monitoring

Superheated nanodroplets exposed to ionizing radiation vaporize into echogenic microbubbles, enabling ultrasound-based radiation therapy monitoring and proton range verification. However, at body temperature, decafluorobutane nanodroplets are insensitive to primary protons, and vaporize only when traversed by rare heavy secondary particles, limiting the in vivo applicability of this concept. We previously demonstrated that acoustic modulation at pressures well below the Acoustic Droplet Vaporization threshold can transiently increase nanodroplet sensitivity to radiation, enabling vaporization by primary protons. However, as acoustic modulation was applied by a focused single-element transducer, sensitization was restricted to a small region of the irradiated volume, and monitoring required an additional imaging probe. Here, we introduce a compact implementation that combines wide-field acoustic modulation and volumetric imaging on the same large-element matrix array, enabling simultaneous nanodroplet sensitization to protons and vaporization monitoring across the entire irradiated volume. The resulting 3D vaporization maps were fitted to the proton stopping distribution to estimate beam position and width parameters. At body temperature and starting from acoustic pressures as low as 122 kPa, we observed a transition from a response dominated by secondary particles to a response dominated by primary protons. The resulting vaporization peak at the proton range enabled sub-millimeter range estimation and detection of a 1.05 mm range shift with both accuracy and precision below 0.2 mm. These results demonstrate proton range verification and 3D treatment monitoring with acoustically-modulated decafluorobutane nanodroplets and represent an important step towards the envisioned clinical application.
Read the full abstract

All authors Sophie V. Heymans, Bram Carlier, Marcus Ingram, Kenneth Poels, Marc Desmaret, Nicolas Gerard, Alessandro Ramalli, Marc Fournelle, Tom Depuydt, Edmond Sterpin, Jan D'hooge, Koen Van Den Abeele

Cryptography and Security

Mirage: Provable Defense Against LLM KV Cache Timing Side Channels with Topology-Aware Isolation

Large Language Models (LLMs) are increasingly deployed as shared multi-tenant services, relying on advanced optimizations like KV cache and prefix sharing to improve throughput and reduce latency. However, these optimizations introduce a critical vulnerability: attackers can exploit time-to-first-token (TTFT) variations between KV cache hits and misses to infer sensitive prompt tokens via prefix-matching side-channel attacks. Existing defenses indiscriminately restrict cross-tenant cache sharing, which incurs prohibitive performance overhead and negates the fundamental efficiency benefits of the KV cache. We propose Mirage, a topology-aware emission obfuscation framework that provably bounds timing leakage without sacrificing the performance gains of cache sharing. Instead of stalling the GPU or forcing redundant recomputations, Mirage delays token emission at the gateway so the attacker observes a natural cache miss distribution, while the GPU seamlessly continues serving other requests at full speed. Mirage uses a Privacy Odometer that persistently tracks the topological novelty of prefix fan-out expansions to distinguish aggressive probing from benign traffic. Once a prefix exceeds a safety threshold, Mirage engages a two-phase policy that uses a dynamic estimator to pad the TTFT to a calibrated cache-miss baseline. On a 500-sample public corpus across three models, Mirage suppresses the timing side-channel to near the random-guessing floor, while maintaining a significantly lower TTFT than defense baselines. We further show that Mirage withstands a strong adaptive adversary and the same principle carries over to semantic caching layers.
Read the full abstract

Analysis of PDEs

Embeddings for Orlicz potential spaces with applications to mixed local-nonlocal problems

In this paper, we provide further insight into potential spaces within the nonstandard framework of Orlicz functions. More precisely, we establish continuous and compact embeddings of potential spaces in $\mathbb{R}^n$ into both unweighted and weighted Orlicz spaces. Special attention is devoted to the radial setting, for which compactness results are obtained through Strauss- and Ni-type inequalities. As an application, we prove the existence of positive radial solutions to a class of Henon-type problems involving both local and nonlocal contributions in the Orlicz framework.
Read the full abstract

Optics

2D polaritons for Squeezing and Shrinking Light

Two-dimensional (2D) materials have emerged as an extraordinary platform for polaritonics, given their remarkable ability to confine light to deep sub-wavelength scales. However, as material thicknesses reach the nanometer limit, macroscopic electrodynamic models become inadequate for the description of light-matter coupling in 2D materials, leaving a significant gap in our understanding of 2D polaritons. In this work, we develop a comprehensive quantum microscopic framework that describes 2D polaritons through a general, scale-less Hamiltonian. Starting from the quantum properties of a polarizable 2D surface, we derive collective excitations and we identify both transverse electric magnetic modes existing outside the light cone. Our results demonstrate that these 2D systems facilitate a powerful synergy between extreme wavelength shrinking and intrinsic quantum squeezing of electric field quadratures. By establishing this theoretical roadmap, our work provides the essential architecture for controlling non-classical states of light at the nanoscale, opening new avenues for high-precision quantum sensing and the development of next-generation optoelectronic devices
Read the full abstract

All authors Eduardo Bede Barros, Niclas Sven Müller, Stephanie Reich, Christian Thomsen