Mesoscale and Nanoscale Physics
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Friday 11 September
To clarify spin relaxation and precession across the weak-to-strong SOC crossover in chiral conductors, we formulate a density-matrix quantum kinetic theory for a three-dimensional isotropic chiral metal with hedgehog SOC and nonmagnetic impurity scattering. The formulation retains interband coherence, and its collision integral conserves charge, energy, and spin during impurity scattering. We identify three spin modes that evolve continuously from a long-lived D'yakonov–Perel' relaxation mode and two strongly damped precessional modes at weak SOC to one relaxational and two coherent precessional modes at strong SOC. A closed multipole decomposition maps the spin dynamics onto an effective Bloch equation, providing a unified interpretation of relaxation and precession across the crossover. We further derive the ac Edelstein susceptibility and the reciprocal current response to a time-dependent Zeeman field, show that their poles coincide with the spin modes, and verify Onsager reciprocity. The theory thus provides a unified analytic description of spin relaxation, precession, and spin–charge conversion across the weak-to-strong SOC crossover.
Fluid Dynamics
A Variational Formulation of the MHD Induction Equation
We present a variational formulation of the induction equation in magnetohydrodynamics (MHD) based on Gauss's principle of least constraint. The central result is the Euler–Lagrange equation $\mathbf{Z}_B = -\nabla h_m$, where $h_m = \mathbf{A}\cdot\mathbf{B}$ is the magnetic helicity density. This reveals that the magnetic helicity gradient $\nabla h_m$ acts as the constraint force maintaining the solenoidality of the magnetic field, exactly as the pressure gradient $\nabla p$ maintains incompressibility in Taha et al.'s pressure-gradient minimization principle. The magnetic helicity density—which, though gauge-dependent locally, yields a gauge-invariant variational principle—naturally emerges as the Lagrange multiplier enforcing $\nabla\cdot\mathbf{B}=0$. At the solution, the field minimizes the norm of the magnetic helicity gradient $\|\nabla h_m\|^2$. This establishes a structural analogy: magnetic helicity is to the magnetic field as pressure is to velocity. The variational principle connects to Woltjer's theorem, Taylor relaxation, and the Hamiltonian structure of MHD.
Read the full abstract
Astrophysics of Galaxies
The Infrared Glow of Galactic Outskirts: A New Window with NASIM
The outskirts of galaxies retain fossil records of their assembly history and are dominated by old stellar populations shaped by tidal interactions, accretion events, and environmental processes. Observing these low-surface-brightness (LSB) regions in the near-infrared (NIR), particularly in the $K_s$ band, is crucial for tracing evolved stars and minimising the effects of dust. Yet, ground-based NIR studies of faint extended structures remain challenging because of the bright and variable sky background and complex detector systematics. To overcome these limitations, we developed NASIM, the Near-infrared Automated low Surface brightness reduction In Maneage pipeline, a fully automated and reproducible framework designed to recover faint structures in VISTA/VIRCAM data. Using GNU Astronomy Utilities (Gnuastro), NASIM corrects large-scale flat-field patterns and instrumental signatures across the VIRCAM detectors while preserving LSB emission that can be strongly suppressed in conventional NIR reductions. We apply NASIM to $K_s$-band imaging of the Euclid Deep Field South through the KEDFS programme, a key region also targeted by Rubin/LSST and Euclid. The final reduction reaches a surface-brightness limit of $\mu_{K_s} \simeq 27.7$ mag arcsec$^{-2}$, measured at $3\sigma$ over 100 arcsec$^2$, approximately 67 times deeper than 2MASS and 11 times deeper than VHS in surface-brightness sensitivity. Direct comparisons with publicly available VISTA products show that NASIM preserves diffuse emission without substantially compromising compact-source depth. These results open a deep NIR view of faint galaxy outskirts, LSB galaxies, and intracluster light, providing a reproducible ground-based foundation for future studies of the faint Universe.
Read the full abstract
Networking
Support-Aware Telemetry Compression for 5G Positioning via Conditional Conflict Graphs
Geographically separated transmission/reception points (TRPs) report quantized measurements to a Location Management Function (LMF), even when the application requires only a coarse location region. We formulate this task as a distributed zero-error function-computation problem, in which each TRP transmits an index sufficient for the LMF to reproduce the required service decision. Since positioning geometry induces a sparse and nonrectangular support, independently constructed per-terminal characteristic-graph colorings are not necessarily jointly decodable. We introduce a conditional conflict graph that exactly characterizes valid single-terminal updates and develop an alternating codebook construction that preserves global zero-error decodability. In a reproducible three-TRP study with range-equivalent timing measurements and $120$ native bins, all resulting codebooks satisfy an explicit decoder-conflict test. For service-cell sizes up to $100$ m, the achieved ideal rate is $5.38$–$5.44$ bits per epoch per TRP, compared with $6.91$ bits for raw reporting and $6.56$–$6.87$ bits for a globally valid interval-based baseline. A complementary measured six-base-station TDoA study shows that $83.77\%$ of the learned native support recurs on an independent trajectory. For these recurrent tuples, the codebooks preserve the service decision exactly and reduce the ideal rate by $22.8$–$35.6\%$ for $2$–$8$ m service grids. The feasible report-tuple set also provides a single-epoch geometric-consistency check under injected timing bias. Finally, we identify the NRPPa/OpenAirInterface integration points and safeguards required for experimental implementation.
Read the full abstract
Robotics
ARC: Autonomous Robotics Compliance A Three-Layer Governance Architecture for Deployed Autonomous Systems
Proposed governance framework for autonomous robotic systems, introducing a three-layer compliance architecture (ARC) instantiated through model safety validation, cognitive certification benchmarks, and operational authorization standards.
Strongly Correlated Electrons
Spin splitting without symmetry: a nearly compensated ferrimagnet and the origin of altermagnetism
The spin splitting of collinear altermagnets is usually attributed to the crystal symmetry relating the opposite-spin sublattices — a rotation or mirror in place of the translation or inversion of a conventional antiferromagnet. We show that this symmetry organizes the splitting but does not produce it; its origin is the anisotropic arrangement of the magnetic orbitals and their ligands, which also fixes the symmetry itself. Two results establish this. A triclinic ($P1$) \ce{Mn} oxide, a fully compensated ferrimagnet whose two inequivalent sublattices are related by no symmetry, is spin split throughout the Brillouin zone; repositioning the same ligands, atom for atom, to restore an inversion centre relating the two metal sites collapses the splitting to a residual two orders of magnitude smaller, so the splitting follows from the arrangement rather than the symmetry. Symmetry, we show formally, can only ever reverse or forbid a splitting that already exists: it cannot create one between sublattices an arrangement has left electronically identical, nor where the underlying orbitals carry no anisotropy to arrange in the first place. Organizing the analysis through an operation-resolved symmetry hierarchy and reading the crystal classes as a single axis, the enforced nodal planes fall from two to one to zero across the orthorhombic, monoclinic, and triclinic classes while the arrangement-generated splitting persists to the base. The altermagnet and the fully compensated ferrimagnet are thus the two ends of one axis — the same splitting, organized by symmetry at one end and unorganized at the other — and the search for compensated spin-split magnets is a search over orbital arrangements rather than symmetry labels.
Read the full abstract
Other Condensed Matter
Intrinsic excitations and a proposed ground state in an Ammann-Beenker artificial spin ice
Artificial spin ices built on quasiperiodic geometries remain under-explored. Here we investigate an ASI based on the octagonal Ammann-Beenker tiling, whose six vertex types present local environments inaccessible in periodic lattices. Combining Monte Carlo simulations of dipolar-coupled macrospins with micromagnetic calculations of individual vertex energies, we find that vertex types freeze in an order set not by coordination number, but by quasi-degenerate levels at five-island vertices and their local environment. Excitations in the low energy state are correspondingly localised to specific, geometrically determined sites. Using twice-inflated substitution rules as an energetic backbone, we propose an analytical ground state combining a long-range ordered fixed spin network with a sparse set of degenerate regions.
Read the full abstract
Computer Vision
DementiaCare-Bench: A Modality-Validated Video Benchmark
Dementia affects an estimated 57 million people worldwide, and for most families the hardest part of care is not memory loss but the behavioral and psychological symptoms of dementia (BPSD): agitation, wandering, resistance to care, sundowning. Understanding these symptoms requires more than recognizing the behavior itself; it also requires knowing what happened beforehand. The same behavior may call for a different response depending on its trigger. Video-language models (VLMs) could potentially support caregivers, yet no existing benchmark evaluates this capability. To fill this gap, we present DementiaCare-Bench: 56 professionally produced caregiver training videos segmented into 94 clips across nine BPSD categories, with 2023 questions generated by a multi-agent pipeline that grounds every clinical claim in a verbatim transcript span. Each question is then probed under four visual conditions and labelled by the least it requires, so its visual demand is measured rather than assumed. Measurement contradicts intent: we wrote 77.7% of the questions to require ordered frames, and 34.8% do. Across 12 current VLMs the pattern is uniform. The best reach 85% overall, but that average is carried by questions a language model can answer from clinical knowledge alone; accuracy falls by 17 points on average on questions that require the ordered clip, and a leading open model scores at chance on judging whether a caregiver's response was appropriate. A lightweight LoRA fine-tune, DemCare-VLM, moves video dependence from -3.3 to +4.5 points, so what the benchmark exposes can be repaired and not only measured.
Read the full abstract
High Energy Physics (Phenomenology)
NNLL resummation for the production of four top quarks
Four-top production is one of the rarest processes of the Standard Model observable at the Large Hadron Collider, offering sensitivity to the top Yukawa coupling, Higgs width, and beyond-Standard-Model physics. We present precise predictions for the invariant-mass distribution and total cross section using threshold resummation at next-to-next-to-leading logarithmic (NNLL) accuracy, matched to next-to-leading order (NLO) calculations.
Read the full abstract
Robotics
Robust Underwater Grasping of Sloped Objects with a Waterproof Passive Adaptive Gripper
Robust grasping of everyday objects remains challenging for parallel-jaw grippers, particularly when handling sloped or asymmetric items that induce torque-driven rolling and shear slip. These challenges become even more severe in domestic environments such as kitchens, where objects are often wet or submerged, drastically reducing friction between the gripper and the object. To address these issues, we present a waterproof passive adaptive gripper that combines local and global adaptability for high-performance grasping in the submerged environments. The proposed gripper features a fully waterproof design that integrates a passive rotational joint for global self-alignment on sloped surfaces and passive variable-stiffness pads for local surface adaptation. Experiments in both dry and underwater conditions on various cylindrical and conical objects demonstrate superior holding capability and grasp robustness compared to rigid-pad and fixed-joint baselines. The proposed design offers a practical and robust solution that successfully enables stable underwater manipulation of diverse everyday objects, effectively addressing a critical gap in current underwater robotic grasping for daily-life applications.
Read the full abstract
Analysis of PDEs
Regularity of Structurally Stable Cusp Singularities for Two Families of Quasilinear Wave-type Equations
In this paper, we study two families of quasilinear equations: Hunter-Saxton type and Camassa-Hom type equations, with a paramerter $\lambda\in(0,1)$ whose solutions form cusp singularities. When $\lambda=1$, the first system becomes the scalar conservation law. The main result of this paper is to give regularity of two types of structurally stable singularities: Type I on the singular curve, Type II at the point where cusp singularity forms, for some $\lambda\in(0,1)$. When $\lambda\rightarrow 1$, our result indicts the $C^{1/3}$ regularity at the point where singularity forms, which agrees with the regularity of the generic pre-shock solution.
Read the full abstract
Analysis of PDEs
Weak Solutions for the Unregularised Hibler Momentum Equation with Degenerate Coefficients
We study the momentum equation of the decoupled unregularised Hibler sea-ice model with non-negative ice mass and ice strength, both of which may vanish. The ocean drag provides coercivity, while the stress law and mixed boundary conditions are encoded by a convex dissipation functional. We prove existence, uniqueness and stability for the time-discrete problem and reconstruct an admissible stress. A Rothe approximation yields global weak variational solutions for time-dependent coefficients under a one-sided growth condition on the mass. Weak solutions are unique when the ice strength is independent of time. For spatially Lipschitz ice strength, finite dissipation also yields a local measure structure of the weighted deformation and a global bound on the negative part of the weighted divergence.
Read the full abstract
Rings and Algebras
Uniqueness of Rank-Metric Completions of Bratteli Systems
Let $R$ be a unital ring equipped with a Sylvester matrix rank function $\rk$. A harmonic function $\alpha$ on a Bratteli diagram $B$ defines a weighted matrix rank on the associated algebraic direct limit $A(B,R)$. We prove that, if $\alpha$ is extreme and the total weight of blocks of any fixed bounded size tends to zero, then the rank completion of $A(B,R)$ is isomorphic to $\mathcal M_{R,\rk}$, the rank completion of the direct system $\Mat_{2^k}(R)$ with connecting maps $x\mapsto\diag(x,x)$ and normalized ranks $2^{-k}\rk$. The isomorphism preserves the unital $R$-algebra structure and the ranks on all rectangular matrices. The coefficient ring need not be regular, and the specified rank need not be induced from a regular ring. We recover factor-sequence uniqueness and construct corners of every prescribed rank in $(0,1]$ that are isomorphic to $\mathcal M_{R,\rk}$ with their normalized ranks. Examples show that the coefficient rank can affect the isomorphism type and that the completion can be non-regular and non-simple. For complex coefficients, the trace determined by $\alpha$ gives a rank completion of the associated AF $C^*$-algebra canonically isomorphic to the affiliated-operator ring of its GNS closure. The rank completion of the algebraic direct limit can be a proper subring of this ring.
Read the full abstract
Performance
Dissecting GPU Utilization for LLM Inference on Nvidia Hopper
A single SM utilization percentage can make an LLM inference workload look compute-saturated while hiding how much useful work is being done. The problem is not that the counter is wrong, but that it collapses several different mechanisms into one number. This is most severe during decode, where each request contributes only one new token and dense projection GEMMs become small-row matrix multiplications. On Hopper, the bfloat16 GMMA path executes these operations in fixed 64-row matrix fragments, so small-batch decode can fill only a small fraction of each fragment with real token rows. In this paper, we profile vLLM with FlashAttention-3 and cuBLASLt on an H100 NVL across cold prefill, warm prefill, and decode, sweeping sequence length and batch size. We replace the usual single utilization number with eight counter-validated views derived from raw Nsight Compute reports, each pinned to an NCU counter or explicit formula. Together, these views map utilization gaps to concrete mechanisms - fragment fill, occupancy limits, stall signatures, wave quantization, and kernel selection - across four production models and six per-layer kernel roles.
Read the full abstract
Systems and Control
Aerial Non-Stop Trajectories Preserving the Equilibrium of Loads Suspended by Variable-Length Cables
This work studies equilibrium-preserving non-stop trajectories of aerial carriers connected to rigid loads by variable-length cables. Internal-force motions vary the cable directions without changing the load wrench, while cable-length actuation shapes the radial realization of the carrier trajectories. We derive an explicit cable-length law for realizing prescribed carrier-position profiles along a fixed direction, with constant altitude as a relevant specialization. Under this constraint, a carrier stops exactly when its cable direction stops changing. This characterization reveals a distinction between unrestricted and level motion: variable cable lengths enable bounded non-stop motion with two carriers, whereas periodic level non-stop motion is obstructed for two carriers and constructively achieved for three or more under explicit geometric and feasibility conditions. Numerical simulations illustrate the constant-altitude construction.
Read the full abstract
Software Engineering
Intelligent Semantic Matching (ISM) for Video Tutorial Search using Transformer Models
The rise in the number and diversity of available software development video tutorials has enhanced digital learning for developers but also introduced challenges in locating relevant content efficiently. Existing video search methods, including keyword-based approaches and tools like CodeTube and TechTube, rely primarily on retrieval algorithms such as BM25, which fail to capture the semantic nuances and user intentions behind search queries. To address these limitations, we introduce ISM, an approach that uses SBERT to generate semantically rich vectors from video tutorial transcripts to improve the search for programming video tutorials. By segmenting transcripts and implementing a re-ranking process, ISM effectively preserves context and enhances the relevance of search results. Additionally, ISM generates informative video summaries using GPT-4, allowing developers to quickly assess the relevance of video content. To evaluate our approach, we first performed a quantitative study comparing ISM with the baseline TechTube. The results revealed that ISM performs better in both video retrieval and fragment identification, achieving a Hit@5 score of 0.95 and an average F1 score of 0.70 compared to the baseline's 0.58 and 0.52, respectively. We also performed a user study, which revealed that users strongly preferred the semantic matching capabilities and AI-generated summaries of our approach. This work advances the state-of-the-art in programming video tutorial search and summarization by offering more nuanced and user-aligned retrieval and summarization mechanisms.
Read the full abstract
Computer Vision
Input Resolution Matters: Real-Time Object Detection Latency
We model total latency as the convolution of preprocessing, inference, and postprocessing distributions under a simplifying independence approximation, with selected stage parameters expressed as functions of source-image resolution. Under this assumption, the probability density of the total latency is the convolution of the stage-wise densities, and its cumulative distribution function (CDF) provides the distribution of end-to-end detection time. Each stage is modeled by a parametric distribution (e.g., Exponential, Erlang, Normal, Gamma), with parameters expressed as functions of the source-image resolution. Experiments with YOLOv11n on NVIDIA Jetson Orin NX using COCO2017 images across multiple resolutions assess the proposed models against fixed-parameter baselines using Kolmogorov Smirnov, Anderson Darling, and Cramér von Mises statistics. The results indicate that resolution-aware parameterization can improve distributional approximation in the measured setting, particularly for the more flexible Normal and Gamma models, while the quality of fit remains distribution dependent. Our contribution is a theoretically grounded and lightweight formulation for studying resolution-dependent latency distributions in a measured object detection pipeline.
Read the full abstract
Group Theory
Finitely generated groups of diffeomorphisms with prescribed critical regularity in every dimension
Let $n\ge 2$, and let $M$ be either the $n$-dimensional sphere $\mathbb{S}^n$ or the real projective space $\mathbb{RP}^n$. Given $\alpha\ge 1$, we construct a finitely generated group $G_{n,\alpha}$ that embeds into the group of $C^\alpha$-diffeomorphisms of $M$, but does not embed into the group of $C^\beta$-diffeomorphisms of $M$ for any $\beta>\alpha$.
Read the full abstract
Sound
PhaseGAN: High-Fidelity Vocoder via Decoupled Amplitude and GAN-Driven Phase Reconstruction
A vocoder is a pivotal component of modern text-to-speech (TTS) systems. Despite the significant progress of neural network-based vocoders, accurate phase reconstruction remains the main challenge limiting both audio quality and modeling efficiency. We introduce PhaseGAN, a lightweight vocoder that addresses this limitation through a "mel $\rightarrow$ Amplitude $\rightarrow$ Phase" reconstruction pipeline. By reconstructing amplitude and phase spectra via distinct methodologies, the proposed PhaseGAN outperforms state-of-the-art baselines while utilizing fewer model parameters and reduced computational requirements. The compact version generates high-fidelity audio with approximately 500K parameters and 1 GMAC computational load, making it highly suitable for real-time applications on edge devices. In addition, our approach exhibits exceptional musical audio synthesis capabilities despite no training on musical data, illustrating unprecedented cross-domain generalization. See https://github.com/phasegan/phasegan-audio-demo for demos of our work.
Read the full abstract
Computer Vision
PA-CDM: Position-Aware Character Detection Matching for Evaluating Handwritten Mathematical Expression Recognition
Handwritten mathematical expression recognition (HMER) is conventionally scored by exact-match rates and string-similarity metrics that are blind to where an error occurs: two predictions with identical token-error counts receive identical scores whether they misplace a subscript or swap the operands of a fraction. Render-based character detection matching (CDM) aligns glyphs robustly but remains position-blind—on controlled fraction-operand swaps it scores 0.8595 where position-aware scoring yields 0.6253. Tree-edit metrics exhibit a complementary blind spot: rewrites outside the parser's normalization coverage are penalized as structural errors (0.8552 where render-based metrics score 1.0). We propose PA-CDM, a position-aware metric that couples character detection matching with position-forest encoding and divergence-level weighting; StructPerturb v2.0, a frozen benchmark of 1,340 controlled perturbation pairs across 15 type–intensity cells; and a cross-metric consistency protocol combining a sensitivity matrix, a human study, and LLM-judge calibration. In a six-annotator study, PA-CDM attains the highest correlation with human judgments among seven automatic metrics (Spearman rho=0.9535, n=990). A frontier LLM judge correlates slightly higher (rho=0.9613) but is costly, nondeterministic, and API-dependent; PA-CDM approaches it at zero marginal cost with deterministic, diagnosable behavior.
Read the full abstract
Logic
Small masas of the Calkin algebra in the Cohen model
We show that maximal abelian C*-subalgebras (masas) of the Calkin algebra (the algebra of all bounded operators on the separable Hilbert space modulo compact operators) may consistently have their densities strictly less than continuum and we describe many isomorphism types of such masas. Specifically, we prove that after adding any number of Cohen reals to a model of CH the algebra $C(K_{\mathcal A})$ of all complex-valued continuous functions on the Stone space $K_{\mathcal A}$ of a Boolean algebra $\mathcal A$ of cardinality $\omega_1$ is $*$-isomorphic to a masa of the Calkin algebra if and only if $\mathcal A$ does not admit a countably generated ultrafilter. Moreover, for every such Boolean algebra we obtain $\omega_2$ pairwise unitarily non-equivalent such masas, none of which has a commutative lift. We also show in ZFC that if a C*-algebra of the form $C(K)$ for any compact Hausdorff $K$ is $*$-isomorphic to a masa of the Calkin algebra, then no point of $K$ may have character smaller than $\mathfrak p$. Therefore, consistently, there may not be any masa of the Calkin algebra of density less than continuum.
Read the full abstract
Computation and Language
LLM-Enhanced Dual-Branch Learning for Large-Scale Multi-Label Text Classification
Large-scale multi-label text classification assigns a small subset of relevant labels to each document from a vocabulary containing thousands or tens of thousands of candidate labels. Although pretrained language models have improved semantic text representations, most representation-based approaches center their prediction pipelines on a primary encoder or combine auxiliary features within a single ranker. The complementarity between heterogeneous language models therefore remains insufficiently explored. We propose DualMLC, a dual-branch framework that processes the same document through an autoregressive decoder-only language model and a bidirectional encoder. Each branch maintains its own representation pathway and independently estimates relevance scores over the shared label space. DualMLC combines the two score vectors through late logit fusion, allowing shared evidence to reinforce relevant labels and branch-specific evidence to compensate for limitations in the other branch's representation. DualMLC achieves state-of-the-art results on three widely used large-scale multi-label text classification benchmarks. Ablation results further confirm that integrating the heterogeneous predictors produces stronger rankings than either branch alone. The source code is publicly available at https://github.com/huiyegit/DualMLC.
Read the full abstract
High Energy Physics (Phenomenology)
What has the LHC told us about the electroweakino sector of the Minimal Supersymmetric Standard Model?
We perform global fits of the electroweak sector of the Minimal Supersymmetric Standard Model (MSSM) using a comprehensive set of LEP searches, 34 Run 2 LHC searches, and 63 Run 2 LHC measurements. Scanning the bino, wino and Higgsino mass parameters, and the ratio of the Higgs vacuum expectation values, we find that for a light, bino $\tilde{\chi}_{1}^0$, the mass of the next-to-lightest neutralino must be $m_{\tilde{\chi}_{2}^0} \gtrsim 760$ GeV. While MSSM electroweakinos can explain individual excesses observed by ATLAS and CMS in searches targeting compressed spectra, we find no scenarios that fit these excesses simultaneously. When we add a light gravitino, neutralinos are further excluded up to about 1 TeV, though this depends on their composition; Higgsino-dominated $\tilde{\chi}_{1}^0$ requires only $m_{\tilde{\chi}_{1}^0} \approx m_{\tilde{\chi}_{2}^0} \gtrsim 650$ GeV. Lastly, the newer LHC searches and measurements exclude a low-mass region that was preferred in a previous study. This is the most complete summary of collider constraints on the electroweakino sector of the MSSM performed to date.
Read the full abstract
Computation and Language
Parameter-Efficient Retrievers for Polish and European Languages
Dense retrieval systems increasingly rely on multi-billion-parameter language models, whose memory and computational requirements make large-scale indexing, frequent corpus updates, and low-latency serving costly. We present a three-stage training pipeline for developing compact and efficient retrievers that remain competitive with substantially larger models. The pipeline combines cross-lingual alignment, relational knowledge distillation, and contrastive fine-tuning. It requires no original ground-truth relevance labels, relying exclusively on supervision generated by strong embedding models and rerankers utilised as teachers. Using this pipeline, we develop PolDense and EuroDense, both supporting contexts of up to 8,192 tokens. PolDense is a family of six Polish retrievers ranging from 17M to 1B parameters. EuroDense is a 435M-parameter retriever supporting nine European languages. We conduct an extensive evaluation covering 41 Polish and 150 multilingual retrieval tasks. The results demonstrate strong quality-efficiency trade-offs. PolDense-1B outperforms the evaluated retrievers with up to 9B parameters, while the PolDense family forms the Pareto frontier across model sizes. Among the evaluated models below 1B parameters, EuroDense ranks first in both task-averaged and language-averaged performance and leads in seven of nine languages. We release all models publicly.
Read the full abstract
Numerical Analysis
Structure-preserving quasi-interpolation for vector-valued function with multiple physical constraints
We develop a unified theory of vectorial quasi-interpolation that exactly preserves intrinsic physical structures characterized by systems of linear constant-coefficient differential operators, including divergence-free and curl-free constraints as special cases. The key ingredient is a family of matrix-valued kernels constructed from an orthogonal projector in the frequency domain, whose columns analytically satisfy the prescribed constraints. The resulting quasi-interpolant is a weighted combination of kernel translates with sampled function values as coefficients, and therefore requires no constrained optimization. Simultaneous error estimates for the approximand and its derivatives are established within a bias-variance framework. For numerical vector-field decomposition, we introduce a generalized Helmholtz–Hodge decomposition that splits a smooth vector field into two components with distinct physical structures, and develop corresponding structure-preserving quasi-interpolation schemes and error estimates. Explicit kernels are derived for several representative constraints, including steady-state acoustic constraints, combined divergence-curl constraints, coupled divergence-free constraints, and second-order Saint-Venant compatibility conditions. Numerical experiments confirm the theoretical error estimates and exact structure preservation, demonstrate accurate identification of sources and sinks, and show three-dimensional reconstruction of generalized Helmholtz–Hodge components for general smooth vector fields without prior constraint assumptions.
Read the full abstract
Cryptography and Security
Shuffling is Not Enough: Breaking Permutation-Based Model Confidentiality in Hybrid FHE Inference
Hybrid fully homomorphic encryption~(FHE) inference improves the practicality of private inference by letting the server evaluate linear layers homomorphically while the client decrypts and applies nonlinearities. Recent schemes attempt to protect model confidentiality by returning noisy, output-permuted responses and appealing to shuffle-model differential privacy~(DP). We show that this protection fails in the correctness regime required by hybrid FHE systems. For a $d$-input linear layer, $d+1$ admissible queries suffice for exact recovery of a permutation-invariant layer summary, hence for perfect model distinguishability. We further show that input DP is orthogonal to model confidentiality and that the local-DP premise required for shuffle amplification cannot hold under correctness-bounded noise. We recover all linear layers of a \safhire{}-style ResNet-20 end-to-end from TFHE transcripts with zero error, using $d+1$ queries per layer for a total of $5{,}712$ direct queries. Under the same query model, we also confirm exact per-layer recovery on pretrained ImageNet-scale CNNs and ViT-B/16. The leaked spectra enable fingerprinting, lineage attribution, and improved logit-based extraction, while suppressing them destroys inference utility.
Read the full abstract
Human-Computer Interaction
NeuroClick: Preserving Surgeon Autonomy through Hands-Free Earable Tooth-Click Control in Neurosurgery
Neurosurgeons frequently interact with operating room (OR) technologies while sterility and occupied hands constrain control. We introduce earables as a direct, hands-free control platform for neurosurgery using tooth-click input. Formative OR observations and interviews with 10 domain experts grounded the design. Using OpenEarable 2.0 data from 12 participants, we developed a real-time recognition pipeline whose classifier achieved a median macro F1-score of 98.6% under leave-one-subject-out cross-validation. We evaluated the technique with 20 neurosurgeons during a simulated resection task in a neurosurgical OR. Participants reported few focus shifts and rated Earable favorably for workflow integration and perceived safety. Autonomous microscope control was rated significantly higher with Earable than Delegation, whereas Delegation enabled faster task completion under continuous assistant availability. Workload, usability, and task errors showed no significant differences. Preferences depended on training, reliability, context, and assistant availability. Earables thus add a direct, hands-free option for controlling selected functions alongside established workflows.
Read the full abstract
Cryptography and Security
Forging Tree-Ring: Reproducing and Instrumenting Black-Box Semantic Watermark Forgery
Semantic watermarking schemes such as Tree-Ring hide a detectable pattern in the initial noise latent of a diffusion model. Recent work shows these watermarks are not only removable but forgeable: an attacker who never sees the watermarking key can still produce images the genuine detector accepts. We reproduce the Reprompt forgery attack of Müller et al. against Tree-Ring on Stable Diffusion XL, using the authors' released code, on free-tier dual T4 GPUs with 14.6 GB of usable memory per device, substantially less per-GPU memory than the A40 hardware used in the original study. The attack reproduces. Over six trials of three arms we detect genuine images 6/6, clean images 0/6, and forged images 5/6, at 325-332 s per attack. Three further results came out of running it under constraint. The released detector computes a non-central $\chi^2$ statistic and hands back only its CDF, so we recovered the discarded statistic; our recovery matches the released detector exactly, and two natural scores built from it separate the forged arm from the clean null at AUC 0.861 and 0.972 on the same eighteen observations. Running SDXL in half precision requires patching the pipeline's direct autoencoder calls, and a controlled probe confirms the patched path leaves the detector statistic unchanged. Finally, we report a prediction we made from reading the detector source that our measurements then contradicted. The notebook, the pinned fork and every measurement artifact are released with the paper.
Read the full abstract
Dynamical Systems
Abundance of typical horseshoes for the random standard map
We introduce a notion of typical horseshoe, consisting of a pair of rectangles admitting Markov returns at times of positive lower density, with controlled hyperbolic geometry and orbits that shadow typical trajectories. We prove the abundance of typical horseshoes for the random standard map and establish several statistical properties of the system, including exponential mixing and large deviation estimates for the associated projective and two-point processes.
Read the full abstract
High Energy Astrophysical Phenomena
The influence of free-free absorption on the radio spectrum of Particle-Accelerating Colliding-Wind Binaries
The study of massive stars in binary and higher-multiplicity systems that participate in particle acceleration is a key topic at the crossroads of massive star physics, shock physics, and galactic cosmic ray astrophysics. From an observational perspective, radio measurements are our pri- mary tool for identifying these systems through their synchrotron radio emission. Out of the 54 such systems known to date, all but two have been discovered via their non-thermal radio emission. However, identifying these systems is not straightforward. The main challenge lies in free-free absorption, which can obscure the synchrotron signature and hinder detection. This paper summarizes recent developments in our understanding of these systems, with a particular focus on the strong observational bias introduced by free-free absorption. This bias significantly limits our ability to determine the true fraction of particle accelerators among colliding-wind binaries.