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

Computation and Language

To Each Language Its Tokenizer: Modular Tokenizers for Efficient Multilingual LLMs

Multilingual Large Language Models (LLMs) traditionally rely on a single vocabulary shared by all supported languages, which can lead to uneven compression across them. Moreover, their large embedding and output matrices increase memory usage and slow inference, notably for small-scale models. It is also wasteful as models are often used for only a subset of languages. To address these issues, we introduce a modular framework for multilingual model training. First, we propose methods to learn large modular BPE and Unigram tokenizers that enable extraction of subtokenizers tailored to any language subset. These subtokenizers achieve compression on par with monolingual tokenizers and improve cross-lingual fairness. Second, we design a pretraining strategy that samples subtokenizers to form batches, restricting predictions to the relevant vocabulary subset and allowing efficient training despite a large vocabulary. This supports efficient inference with any combination of language-specific vocabularies. Therefore, it reduces memory usage and speeds up inference in models without sacrificing performance.
Read the full abstract

All authors Franck Signe, Hippolyte Pilchen, François Yvon, Édouard Grave

Machine Learning

Beyond Noise: Understanding and Overcoming Temperature Effects in Analog DNN Inference

The energy efficiency of analog computing makes it one of the most promising candidates for deploying resource-intensive machine learning workloads on constrained platforms such as mobile and embedded devices. However, analog accelerators are inherently susceptible to noise and non-idealities arising from physical component variations, whose behavior is further sensitive to environmental factors. These effects can significantly degrade inference accuracy. In this work, we conduct a comprehensive experimental study on a representative example of analog hardware to investigate the impact of temperature. We first characterize the behavior of stochastic and systematic non-idealities across a range of operating temperatures. Following this, we compare a set of simulation-based and hardware-based mitigation strategies aimed at improving robustness against temperature-induced performance degradation. Our results suggest that temperature-induced degradation is driven primarily by systematic non-idealities rather than stochastic noise alone. Noise-aware training improves robustness, while hardware-in-the-loop training and temperature-aware calibration provide the strongest accuracy retention across varying thermal conditions.
Read the full abstract

All authors Niklas Summ, Xiao Wang, Hendrik Borras, Bernhard Klein, Holger Fröning

Information Theory

Spatially Coupled MacKay-Neal Codes Achieve Capacity on BMS Channels at Fixed Degrees

We establish two degree-dependent results for spatially coupled MacKay-Neal codes on binary-input memoryless symmetric channels. The ensembles use uniform random smoothing and shorten both variable types outside the active chain. For $r=g=3$, every integer $\ell\geq4$, and each channel of capacity greater than $R=r/\ell$, there is a sequence of code realizations whose actual transmitted rate tends to $R$ and whose average sum-product bit error tends to zero. For $r=g=2$ and each $\ell\in\{3,4,5\}$, an interval of binary symmetric channels has capacity greater than $R$ but retains positive transmitted-bit error under terminated density evolution as chain length grows relative to coupling width. Both results follow from the signs of the same density-valued potential at uncoupled fixed points. The degree-three proof combines analytic bounds for $\ell\geq33$ with exact interval certificates for $4\leq\ell\leq32$, followed by threshold saturation and a projection argument for the actual rate. The degree-two proof analytically constructs a fixed point with negative potential and controls both boundary contributions. We relate the potential and the degree-two bifurcation condition to earlier statistical-mechanical predictions. The finite certificates and verification software are available in a versioned supplement.
Read the full abstract

Information Theory

Infinite families of 3-designs from linear and nonlinear codes

The connection between coding theory and combinatorial $t$-designs is an important research topic at the intersection of coding theory and combinatorics. Let $q=p^m$, where $p$ is an odd prime and $m\geq 2$. In this paper, we investigate a class of linear codes $\mathcal{C}$ over $\mathbb{F}_{q^2}$ and their connection with combinatorial $3$-designs. By analyzing the relevant structural properties of $\mathcal{C}$ and $\mathcal{C}^{\perp}$, we show that the supports of the codewords of every nonzero weight in $\mathcal{C}$ and {the supports of the codewords of weight $4$ in $\mathcal{C}^{\perp}$} form $3$-designs. We further investigate a class of nonlinear codes $\mathcal{C}_2$ associated with $\mathcal{C}$, and prove that the supports of the codewords of every nonzero Hamming weight in $\mathcal{C}_2$ also form $3$-designs. These results identify further classes of linear and nonlinear codes whose codewords support combinatorial $3$-designs. In particular, the nonlinear case provides additional examples of codes supporting $3$-designs, a topic for which relatively few results are currently available. As applications, we construct from $\mathcal{C}^{\perp}$ an entanglement-assisted quantum error-correcting code with parameters $[[q+1,q-3,4;4]]_q$. We also prove that $\mathcal{C}$ is an all-symbol locally repairable code with locality $3$. Furthermore, we show that the code $\mathcal{C}$ {meets the Singleton-type bound for locally repairable codes} and hence is optimal in some cases.
Read the full abstract

All authors Shiyan Xiong, Xiaoqiang Wang, Dabin Zheng, Qinqin Ji

Computer Vision

Assessing nnU-Net Generalization across Brain Tumor Populations in BraTS-GoAT 2026

BraTS-GoAT evaluates tumor segmentation across heterogeneous populations. We trained a conventional 3D nnU-Net on 1,351 labeled cases using five-fold cross-validation and 1,000 epochs per fold. The final predictor averaged all folds and applied test-time mirroring. On pooled official validation, global DSC values were 0.7805, 0.8288, and 0.8854 for enhancing tumor (ET), tumor core (TC), and whole tumor (WT). Under matched fold-0 inference, mean regional Dice decreased from 0.9058 on source out-of-fold (OOF) cases to 0.8310 on pooled validation (difference–0.0747). Mirroring gave small single-fold gains but no clear ensemble benefit; a residual-encoder alternative reached 0.8282 mean Dice. In labeled OOF predictions, failure cases had substantially smaller reference ET volumes; after adjustment for ET and WT volume, lower Dice remained associated with more disconnected ET components and a smaller fraction of ET contained in the largest component.
Read the full abstract

All authors Tristan Kirscher, Vivian Metzger, Philippe Meyer, Xavier Coubez

Cryptography and Security

Automating Attack Graph Construction for Agentic Pentesting. Towards Neuro-Symbolic Vulnerability Hunting

Logic attack graphs grounded in scanner output provide explicit and auditable attack path reasoning LLM-based agents lack. Integrating symbolic frameworks such as MulVAL to contemporary security workflows or agentic pipelines, however, requires translating scanner evidence to initial facts, and creating domain-specific rules. We present a semi-automated pipeline that addresses this interoperability problem and depict its feasibility in a web-security case study. Our pipeline parses findings from Trivy, Semgrep, and Nmap into MulVAL predicates and uses an LLM-assisted process to construct domain-specific Datalog rules linking scanner-detectable evidence to attack techniques. MulVAL/XSB then performs symbolic inference to generate structured attack paths. We evaluate the attack-graph construction infrastructure on 54 web Capture-the-Flag tasks from CyBench within an agentic pipeline (Hybrid Reasoner); we do not evaluate the performance of the downstream agent. Every task produced at least one goal-reaching graph, and we achieve mean ground-truth vulnerability coverage of 53.7%, with 51.9% achieving full coverage; mean noise-path rate was 83.9%. With median end-to-end time of 24.9 s (MulVAL reasoning: 2.7 s) the pipeline is feasible and runtime-practical for agentic workflows, but predicate coverage, rule coverage, and path precision remain limiting factors. Next steps include semantic rule validation and agent-level comparison for graph-guided pentesting.
Read the full abstract

Computation and Language

Psychosis involves a deficit of information compression in connected speech

Large language models (LLMs) with human-like performance on linguistic tasks have transformed the study of language in neurodiverse conditions. LLMs provide representations of linguistic input in the form of high-dimensional vectors (embeddings), and next-token predictions computed from these embeddings. Previous crosslinguistic evidence suggests a complexity reduction in the form of both lower intrinsic dimensionality (ID) of LLM representations and higher mean surprisal (prediction error) in psychosis. We hypothesized that these metrics reflect a general deficit of information compression in psychosis, linked to grammatical organization as what enables predictions in language.We operationalized surprisal difference as the difference between surprisal as estimated from word frequency and surprisal as based on a contextual LM, which is sensitive to grammatical organization over and above lexical concepts. Using a dataset of 144 Turkish speakers, including 106 patients with schizophrenia-spectrum disorders (SSD) - 56 with chronic schizophrenia (SZH), 33 with first-episode psychosis (FEP), and 17 with schizoaffective disorder (SZA) - and 38 healthy controls. We report: (1) Surprisal difference is attenuated in all clinical groups relative to controls, independently of word count; (2) Compressibility (intrinsic dimension) is reduced in SZH and FEP; (3) Syntactic complexity and compressibility both predict surprisal difference. These results, further refining an alteration in the geometry of the semantic space in psychosis as previously attested, suggest a broader deficit in information compression in this disorder, with a mechanistic underpinning in the operations of grammar.
Read the full abstract

All authors Samuele Vallisa, Claudio Palominos, Rui He, Emre Bora, Burcu Verim, Cemal Demirlek, Berna Yalincetin, Philipp Homan, Wolfram Hinzen

Machine Learning

End-to-End Verifiable and Robust Federated Learning

Federated learning enables multiple parties to train a shared model without centralizing raw data with the help of an aggregator, but introduces integrity risks once participants or infrastructure are not fully trustworthy. Two requirements are particularly important: robustness to poisoned or Byzantine client updates, and verifiability of the aggregator so that clients or third parties can audit the reported aggregation without learning individual updates. Existing work has largely treated these goals separately, and efficient public verifiability for robust, outlier-excluding aggregation remains limited. We present a verifiable federated learning protocol that makes a robust aggregation pipeline publicly auditable. Our design combines cryptographic commitments with non-interactive zero-knowledge proofs to certify both (i) cosine-similarity-based outlier exclusion and (ii) aggregation over the selected set, without revealing individual client updates to verifiers. In experiments under representative poisoning attacks, our method maintains high accuracy, with an average accuracy loss below 4% across the evaluated configurations, while keeping verification overhead practical: proof artifacts can be generated and verified within minutes at the scale studied. In summary, our results show that robust outlier exclusion and public verifiability can be jointly achieved in a federated learning setting.
Read the full abstract

All authors Doryan Lesaignoux, Enrique Mármol Campos, Gabriele Spini, José L. Hernández-Ramos, Stephan Krenn

Numerical Analysis

Strongly enstrophy-stable integrators for the incompressible Navier-Stokes equations

We propose a mixed finite element discretisation for the incompressible Navier-Stokes equations that preserves the evolution laws of both energy and enstrophy, in a stronger sense than previous discretisations. In two dimensions, the evolution law for enstrophy only permits dissipation for thermodynamically isolated systems, leading to a Reynolds-number-independent bound on the velocity gradient that naturally stabilises the scheme, even on severely under-resolved meshes. In three dimensions, the scheme preserves both dissipation and the generation of enstrophy through vortex stretching. We enforce these evolution laws by systematically introducing auxiliary variables into the discretisation. While conforming implementations of these schemes require discrete Stokes complexes with enhanced regularity, we introduce both (i) equivalent reparametrisations and (ii) penalty formulations that require only the typical curl- and div-conforming spaces from the standard discrete de Rham complex. The scheme handles different types of boundary conditions and curved domains. The robust stabilisation properties of the proposed scheme are demonstrated through numerical simulations of a shear flow, a spherical vortex, and flow past an obstacle. We observe numerically that preserving the discrete evolution of enstrophy in this way has a strong stabilising effect on the numerical solution, especially in two dimensions.
Read the full abstract

Signal Processing

Predictability Measures for Power Quality Time Series in Medium-Term Forecasting

Medium-term forecasting of Power Quality (PQ) parameters, on horizons of weeks to about one year, supports proactive maintenance and the early detection of limit exceedances in transmission network monitoring. Its practical value, however, depends on knowing in advance which time series can be forecast reliably at all. This article addresses that question in two stages, based on 2,807 weekly time series of PQ parameters from a long-term measurement campaign at 66 sites in the German transmission system, covering the 110 kV, 220 kV, and 380 kV levels. First, eight forecasting models are benchmarked. The STL-ARIMA hybrid achieves the highest accuracy, with an average sMAPE of 17.63% and a lower error than the seasonal naive benchmark for 72% of the time series, while accuracy varies substantially across PQ parameters and measurement sites. Second, model-free features computed from the training set alone are evaluated as measures of intrinsic predictability. SVD entropy and the Crest Factor correlate strongly with the realized forecast accuracy of all eight models and are used in a logistic regression to estimate the probability of a poor forecast before any model is applied. The resulting measures allow operators to separate time series suitable for automated forecasting from those requiring manual review.
Read the full abstract

All authors Peter Feistel, Max Domagk, Jan Meyer, Marco Lindner

Analysis of PDEs

The combined limits of large bulk viscosity and small angular viscosity for the 2D compressible micropolar fluid equations

This paper is concerned with the combined limits of large bulk viscosity and small angular viscosity for the two-dimensional compressible micropolar fluid equations. We first establish the global well-posedness of strong solutions to the Cauchy problem with large bulk viscosity, when the initial velocity is well-prepared in the sense that $\sqrt\nu \|{\rm div} u_0\|_{L^2}$ is uniformly bounded. The global a priori estimates obtained are uniform with respect to both the bulk viscosity $\nu$ and the angular viscosity $\varepsilon$. Based on these uniform estimates, we then justify the combined limits as $\nu\to\infty$ and $\varepsilon\to0$, and show that the solutions of the compressible micropolar system converge to those of the viscous and non-diffusive incompressible micropolar system. The convergence rates for the incompressible limit and the vanishing angular viscosity limit are also derived. In particular, the convergence rate for the incompressible limit is shown to be of order $\nu^{-1/4}$ in general, and can be improved to $\nu^{-1/2}$ under an additional decay condition on the initial density. Compared with previous works, the vanishing angular viscosity limit is established without the extra technical assumption $\lim_{\varepsilon\to0+}\zeta/\varepsilon^\alpha<\infty$ for some $1/2\leq \alpha<\infty$, which extends the relevant results from the incompressible case to the compressible setting. As a by-product, the global regularity of large solutions for the 2D compressible micropolar equations with/without angular viscosity is obtained, provided the bulk viscosity is large enough.
Read the full abstract

Artificial Intelligence

Beyond Safe Answers: Segment-Aware Listwise Alignment for Reasoning Safety in Large Reasoning Models

Large Reasoning Models (LRMs) pose a dual-surface safety challenge: both intermediate reasoning traces and final answers can contain harmful content. Existing alignment methods often operate at the whole-response level, allowing unsafe reasoning to be masked by a safe-looking final answer. We propose Segment-aware Listwise Target DPO (SaLT-DPO), which addresses this gap through three mechanisms: (1) segment-aware listwise alignment that decomposes responses into reasoning and answer segments, independently scores each segment's safety, and aligns length-normalized segment rewards with soft target distributions over multiple candidates; (2) joint safety coherence regularization that applies a weakest-link principle to promote safety consistency across both segments; and (3) utility anchoring on benign prompts to mitigate over-refusal and reasoning degradation. Experiments on three LRMs show that SaLT-DPO consistently reduces unsafe rates for both reasoning and answer segments while mitigating degradation in benign compliance and preserving general reasoning performance. Ablation studies demonstrate the complementary contributions of its components.
Read the full abstract

All authors JungMin Yun, Junehyoung Kwon, Hayeong Ryu, Byeonggeuk Lim, Hoejoon Kwon, YoungBin Kim

Cryptography and Security

Misleading the Planner through Deceptive Resumes: Registration-Time Injection in Centralized Multi-Agent Systems

A centralized LLM-based multi-agent system (MAS) extends its functionality by registering new worker agents, whose descriptions are read by the planner to decide how a task is decomposed, which worker executes each subtask, and what each subtask requires. Third-party descriptions are authored outside the system but trusted by the planner, creating a registration-time injection channel. The payload is planted before any user instruction arrives, targets the planner and propagates through the generated plan to benign workers, taking effect even when the crafted worker is never assigned a subtask or invoked. We define four worker-description fields: functionality, input specification, output specification, and usage constraints. Among 32,000 descriptions from three public agent marketplaces, most omit input specifications and usage constraints, while at least 23.35% contain content outside these fields. We construct eight description-manipulation attack strategies targeting task decomposition, capability grounding, and subtask specification, and evaluate them on GAIA. In the most severe cases, a single manipulated description reduces task success from 84.31% to 37.25%, or increases token consumption or execution time by over 111%, while the user objective remains unchanged and workers faithfully execute the resulting plan. These effects persist across two MAS implementations, six planner LLMs, four LLM evaluators, and the real-world descriptions from three marketplaces. We further propose DescGuard, a registration-time defense that retains only worker-scoped interface information before descriptions reach the planner. DescGuard restores the targeted planning metrics and downstream performance toward their baseline levels without modifying worker implementations, the planner, or the orchestration logic, and composes with existing isolation, permission-control, and runtime mechanisms.
Read the full abstract

All authors Zhaofeng Yu, Haokai Ma, Dongyang Zhan, Hongli Zhang, Han Fang, Ee-Chien Chang

Combinatorics

Coherent presentations for stylic monoids via row rewriting

We study stylic monoids from the viewpoint of higher-dimensional rewriting. A coherent presentation records not only the defining relations of a monoid, but also the relations among them. Using the $N$-tableau model, we construct a finite coherent presentation for every stylic monoid of finite rank. We start from the known finite convergent row presentation. By adjoining one generating confluence for each critical branching, Squier's coherence theorem yields a finite coherent extension. We then simplify the resulted coherent presentation by two successive homotopical reductions adapted to the row combinatorics of $N$-tableaux. Finally, coherent Tietze transformations and a further homotopical reduction transfer the construction from the auxiliary row generators to the standard letter generators. This gives an effectively computable finite homotopy basis for the standard letter presentation. The construction is illustrated in detail in rank $2$.
Read the full abstract

Classical Analysis and ODEs

Spherical maximal operators on rank one Riemannian symmetric spaces of noncompact type

The spherical maximal operator of order $\mu$ was introduced by El Kohen in the setting of real hyperbolic spaces, where its $L^p$-boundedness properties were studied. In this paper, we extend this notion to rank-one Riemannian symmetric spaces of noncompact type. We establish sufficient conditions on the parameter $\mu$ for the corresponding spherical maximal operator to be bounded on $L^p$ for $1<p\leq\infty$. We also obtain a necessary condition for $L^p$-boundedness when $\mu$ is real and $1<p<\infty$. In particular, for $1<p\leq2$, the necessary condition agrees with the sufficient condition, yielding the sharp range of real $\mu$ in this regime.
Read the full abstract

High Energy Physics (Phenomenology)

Rotational effects on the chiral crossover transition in QCD matter

We analyze the impact of rotation on the chiral crossover transition of QCD matter within the framework of a hadron resonance gas model. By extending the conventional formulation of the renormalized chiral condensate to a rotating hadronic medium, we find that rotation significantly modifies the chiral condensate and systematically suppresses the pseudocritical temperature as the angular velocity increases. The pseudocritical line in the $T-\omega$ plane is examined, and the rotational dependence of the pseudocritical temperature is quantified through leading- and next-to-leading-order rotational curvature coefficients. We further investigate the combined effects of angular velocity and baryon chemical potential and find that their interplay leads to a stronger suppression of the pseudocritical temperature. Additionally, the spatial dependence of the chiral crossover transition is examined through the radial variation of the pseudocritical temperature. We find that the pseudocritical temperature decreases with increasing radial distance, indicating that the chiral crossover sets in at progressively lower temperatures as one goes away from the rotation axis.
Read the full abstract

All authors Nandita Padhan, Kshitish Kumar Pradhan, Arghya Chatterjee, Raghunath Sahoo

High Energy Physics (Theory)

A fluid dual to large $D$ membrane paradigm at first subleading order

The large $D$ membrane paradigm establishes a duality between the dynamics of black holes and the evolution of a codimension-one timelike membrane in a non-gravitational background. In this work, we formulate the relativistic fluid dynamics dual to an uncharged black hole in asymptotically flat spacetime at the first subleading order in the $1/D$ expansion. Due to the absence of a timelike asymptotic boundary, unlike in AdS/CFT fluid-gravity duality, we construct an effective fluid intrinsically on the dynamical membrane, whose equations of motion are exactly equivalent to the subleading order membrane equations. By performing this fluid-dynamical analysis in the Landau frame, we show that the system behaves as a fluid influenced by an effective background force. Out-of-equilibrium viscous effects emerge naturally, allowing us to extract the fluid transport coefficients directly from the bulk viscous pressure and shear stress tensor. Furthermore, the fluid exhibits a negative pressure, capturing the intrinsic surface tension of the $(D-1)$-dimensional membrane worldvolume.
Read the full abstract

Computation and Language

Authorship attribution and aesthetic evaluation of AI poetry: a case study with Haiku

This paper investigates the generation and human evaluation of Japanese haiku by contemporary Large Language Models (LLMs), focusing on authorship perception and aesthetic judgment within a constrained poetic form. Using a few-shot prompting strategy, Japanese haiku were generated across a heterogeneous set of large language models, including open- and closed-source systems, medium-scale and large-scale architectures, models with native or adapted Japanese support, and multilingual proprietary models. These AI-generated haiku were combined with human-written ones and presented in a questionnaire distributed to students at Japanese universities in Tokyo. The survey assessed whether respondents could distinguish between AI-generated and human-written haiku and which cues informed their judgments. Recognition accuracy varied across models. GPT-5, Gemini 2.5, and StableLM-7B performed at approximately chance level (approx 0.50), whereas LLM-JP, Gemma-2B, and LLaMA-2 showed moderate detectability (approx 0.59-0.67). However, recognition was strongly item-dependent. Ratings of fluency, coherence, poeticness, and related aesthetic dimensions predicted perceived humanness but not correct classification, indicating an attribution bias linked to aesthetic evaluation and revealing a dissociation between aesthetic evaluation and true authorship detection. The extended analysis additionally examines generation-constraint adherence, participant-level characteristics, and exploratory LLM-based evaluations of haiku authorship. Overall, the findings suggest that as LLMs improve, surface-level creative plausibility may reduce reliable human discrimination within constrained poetic settings.
Read the full abstract

All authors Livia Oddi, Simone Scardapane, Toru Sugimoto, Donatella Genovese

Computational Geometry

Perfectly Guarding Straits: Exact Algorithms for Weak Visibility Polygons

The Art Gallery Problem (AGP) asks for the fewest guards that see all of a simple polygon. It is $\exists\mathbb{R}$-complete, hence NP-hard. We show that for a particular class of polygons, confining guards to a single edge makes AGP exactly and efficiently solvable. We call this the Strait Guarding Problem (SGP). Its input is a weak visibility polygon (WVP): a simple polygon where every point is seen from some point of one fixed edge, the base. SGP places the fewest guards on the base that jointly see the whole polygon. First, a structural fact: guards on the base edge that cover the boundary already cover the entire interior, turning a two-dimensional covering problem into a one-dimensional one. Our main result is the Witness-Guard Algorithm, which solves SGP exactly in $O((n + \mathrm{OPT} \cdot \rho)(\log n + \log \mathrm{OPT}))$ time, where $\rho$ is the number of reflex vertices in the WVP and OPT is the minimum number of guards. It is output-sensitive and certifies optimality by a witness set of size OPT derived from its output. We also study the guarding-the-vertex version and prove a tight $\Theta(n \log n)$ bound, with the lower bound following from Sorting. As a corollary of SGP, we obtain two results for altitude terrain guarding (ATG), a special case that SGP generalizes. We give a linear-time perfect-guarding algorithm, improving the previous $O(n^2 \log n)$ bound of Daescu, Friedrichs, Malik, Polishchuk and Schmidt. We also resolve their problem on the minimum guarding altitude, in $O(nk + k^2 \log k)$ time, improving on the $O(k^2 \lambda_{k-1}(n) \log n)$ bound of Kang, Kim and Ahn.
Read the full abstract

Robotics

StereoPatch: Patch-Aligned RGB-Depth Fusion for Spatial Perception in Robot Manipulation

Recent advances in robot imitation learning have produced visuomotor policies that predict actions directly from visual observations. Yet visually similar scenes can require different actions as target position, object height, or contact geometry changes. Pretrained RGB features may map these geometrically distinct states to similar policy inputs, while simply adding depth requires the policy to learn RGB-depth correspondence from the same limited demonstrations used to learn control. We introduce StereoPatch, a patch-aligned RGB-depth representation that binds registered metric geometry directly to the RGB patches used for action prediction. On a shared 2-D patch grid, asymmetric cross-attention incorporates depth information into the corresponding RGB features before action decoding. The resulting StereoPatch Tokens provide a geometry-aware visual representation that can condition general visuomotor policies without changing their underlying learning objectives. Across six real-robot tasks, StereoPatch achieves higher closed-loop success than appearance-only, geometry-only, raw RGB-D, and late-fusion baselines. Additional experiments across three simulation suites evaluate compatibility across visuomotor policy architectures, spatial generalization, and operating limits. Results suggest that resolving control-relevant geometric ambiguity benefits from aligning depth directly with the visual features used for action prediction, rather than supplying it as an independent modality. Project page: https://aus.bot/research/stereopatch/.
Read the full abstract

All authors Yanan Zhou, Zhaoyan Qian, James Zhao, Weiming Zhi

Signal Processing

Contact-Limited Throughput of a Buoyless Acoustic-to-LEO Gateway With Anticipatory Preparation

A buoyless acoustic-to-LEO gateway with predictable contacts is analyzed. Advanceable cross-medium preparation competes with acoustic collection, while residual RF acquisition consumes contact time. Exact fluid and packet service, a fluid-optimal fixed preparation lead, tandem-queue stability, and one-contact reliability are derived. An explicit service-bias identity shows why treating acquisition as advanceable can select an insufficient lead. For lognormal preparation, conditions are characterized under which variability improves mean service while degrading reliability. Monte Carlo and packet-queue simulations corroborate the analysis with synthetic parameters; in a diagnostic case, accounting for residual RF acquisition raises the actual sustainable rate from 8.776 to 9.079 kbit/s.
Read the full abstract

Differential Geometry

Existence of least-area discs

We prove that every closed Lipschitz curve in Euclidean space extends to a Lipschitz disc of least area, with Lipschitz constant controlled by that of the boundary curve. Our proof gives a new approach to the classical Douglas–Radó solution of the Plateau problem but accommodates arbitrary self-intersections and avoids Sobolev maps. The argument extends to metric spaces with curvature bounded above in the sense of Alexandrov.
Read the full abstract

Probability

Canonical analytic realizations of hyperbolic determinantal processes

Krishnapur asked whether the invariant hyperbolic determinantal point processes on the disk admit a random analytic zero-set interpretation at noninteger parameters. We construct such a realization for every positive real parameter as the full compact-open limit in distribution of normalized finite Blaschke products. The zeros determine the modulus and normalized analytic shape, leaving one independent uniform phase. We prove exact Möbius covariance and classify all realizations with this covariance and square-integrable logarithmic modulus at the origin: they are precisely independent positive random multiples of the canonical function. Within this covariant class, matching the canonical logarithmic mean and variance uniquely determines the canonical function law. The family is weakly continuous in the parameter and agrees at positive integers with determinants of matrix-valued Gaussian power series. An explicit Barnes $G$-function Mellin transform determines the basepoint normalization.
Read the full abstract

All authors Xiang Fang, Feng Guo, Shengzhao Hou, Qi Zhou

High Energy Physics (Phenomenology)

Complex-energy spectrum and dissociation of heavy quarkonia in an anisotropic collisional quark-gluon plasma

We solve the Schrödinger equation for heavy quarkonia using the full complex potential of a momentum-anisotropic, collisional quark-gluon plasma. The real and imaginary parts of the complex eigenvalue give the in-medium binding energy and thermal width, while the corresponding eigenfunction determines the spatial distribution of the state. The noncentral anisotropic interaction is treated by expanding the potential in Legendre multipoles and solving the resulting coupled radial equations at fixed magnetic quantum number. The collision rate is tied to the running coupling, with the pinch singularity of the imaginary potential used to restrict the allowed parameter domain. Along the resulting pinch-free trajectories, the dissociation temperature of the $1S$ charmonium and bottomonium states increases with anisotropy. The widths obtained from the full complex eigenvalue differ appreciably from first-order estimates based on the imaginary potential, while the corresponding $1S$ eigenfunctions remain dominated by their leading $S$-wave component. We compare the thermal widths with available lattice-QCD results and use them to illustrate a static survival probability.
Read the full abstract

Computation and Language

Evaluating Losslessness in Speculative Decoding Under Finite-Precision Inference

Lossless speculative decoding is typically defined at the algorithmic level: a speculative procedure proposes multiple tokens and a verification procedure is designed to preserve the output trajectory of an autoregressive reference model exactly. In practical neural inference, however, this guarantee is implemented using finite-precision floating-point computations, and discrete token selection can amplify small numerical differences into divergent generation trajectories. We investigate this distinction using Orthrus, a hybrid autoregressive-diffusion architecture that performs self-drafting and self-verification within a frozen autoregressive backbone, as a representative case study. Across 1,190 prompts from 12 domains, exact trajectory matching under BF16 occurs for only 45% of the authors' checkpoint generations and 43% of those from our independently trained model. The probability of matching is strongly associated with the response-conditional perplexity of the autoregressive reference, indicating that trajectory divergence is not uniform across inputs. Despite these divergences, Orthrus does not exhibit systematic degradation on the evaluated downstream tasks. In contrast, FP32 inference yields exact trajectory matching on all evaluated prompts. These results demonstrate a gap between algorithmic losslessness and its implementation under finite-precision arithmetic, and motivate evaluating lossless speculative decoding at the level of exact generation trajectories as well as downstream task performance.
Read the full abstract

Computer Vision

SyntheticDoc: A Large Synthetic Dataset for Document Unwarping and Illumination Correction

Deep learning models have become the standard tool for document rectification and illumination correction, yet their performance is fundamentally bound by their training data. For nearly a decade, the community has heavily relied on Doc3D, a pioneering but increasingly limited document unwarping dataset in terms of scale and quality. To address this bottleneck, we introduce SyntheticDoc, a massive, high-quality dataset designed to push the boundaries of document unwarping. SyntheticDoc is composed of 1,000,000 high-resolution procedurally generated training samples, alongside extensive validation and test sets. Each sample is paired with rich, pixel-perfect annotations, including UV maps, normal maps, albedo and shading. To ensure physical accuracy and photorealism, the paper geometries are generated via a physics-based simulator and rendered using a path tracer. To demonstrate the benefit of our dataset, we train a simple baseline model on SyntheticDoc and report on its performance in comparison to state-of-the-art methods on both document unwarping and illumination correction tasks. Our dataset is available at https://igl.ethz.ch/projects/SyntheticDoc/ and the code used to generate it at https://github.com/tanguymagne/SyntheticDoc .
Read the full abstract

Combinatorics

Maximum number of spanning trees in bipartite graphs with a given diameter

The number of spanning trees is a classical graph invariant and an important measure of network reliability, as it counts the minimal connected spanning substructures that can maintain communication in a network. Let $\mathcal{B}(n,d)$ be the set of connected bipartite graphs of order $n$ and diameter $d$. Motivated by reliability design problems for bipartite network models with fixed order and diameter, this paper determines all graphs with the maximum number of spanning trees in $\mathcal{B}(n,d)$. The result gives an extremal characterization of bipartite network topologies with the largest number of connected spanning backbones under prescribed order and diameter constraints, and provides a structural reference for the design of reliable bipartite networks.
Read the full abstract

Analysis of PDEs

Time derivative estimates for normalized parabolic $p$-Laplace equations

We establish the local boundedness of time derivatives of bounded viscosity solutions to the normalized parabolic $p$-Laplace equation for every $p\in(1,\infty)$. The proof combines a logarithmic Bernstein argument with quantitative stability to derive a recursive estimate between different regularizations. For $p=1$ and $n\geq2$, we construct a bounded viscosity solution which is not Hölder continuous in time with any positive exponent.
Read the full abstract

Dynamical Systems

Uniform bounds for bubbles of neutral quadratic polynomials

Given a quadratic polynomial with an irrationally indifferent fixed point of bounded type, a bubble is an iterated preimage of its Siegel disk. Similar to pseudo-Siegel disks, one constructs pseudo-bubbles from bubbles by filling in parabolic fjords. We introduce the near-degenerate regime for the $\alpha'$-Dynamics controlling the geometry of these pseudo-bubbles in deep scale. Consequently, we obtain uniform bounds on the size and regularity of pseudo-bubbles independent of the rotation number. In [DLL], jointly with Lyubich, this result serves as a necessary ingredient for establishing uniform butterfly bounds and ultimately the combinatorial rigidity of the full attractor of neutral renormalization.
Read the full abstract

General Topology

Homogeneous linearly ordered spaces

Every compact subspace $K$ of a homogeneous generalized ordered (GO) space $X$ satisfies $\chi(K)\leq\omega_1$ and $|K|\leq 2^{\omega_1}$. If $\chi(K)>\omega$ for some such $K$, then $\chi(X)=\omega_1$ and $\pi\chi(X)=\omega$. We construct a homogeneous $\sigma$-compact linearly ordered topological space (LOTS) $\mathbf{H}$ with a compact subspace $K$ such that $\chi(x,K)=\omega_1$ for every $x\in K$ and $c(K)=w(K)=|K|=2^{\omega_1}$, where $c(K)$ denotes cellularity. Thus both bounds are attained by the same compact space. Every semitopological group that is a GO space is hereditarily paracompact and has metrizable compact subspaces. If it is not a $P$-space, it is also first countable and submetrizable. A semitopological group that is a LOTS, or a topological group that is a GO space, is either metrizable or is a $P$-space. We also show that in an order-homogeneous LOTS every compact subset is first countable.
Read the full abstract