Peer Revenue Now accepting positions

Papers

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

All of arXiv

Sunday 8 November

Artificial Intelligence

Verify Before You Fetch: Reliable KV Cache Compression for LLM Inference

Long-context applications, long-lived agents, and multi-agent systems require large language models (LLMs) to process and retain extensive information. In these settings, the KV cache can exceed GPU high-bandwidth memory (HBM) capacity, necessitating offloading to host memory, where CPU-to-GPU transfers can dominate inference latency. While KV cache compression can reduce data movement, query-agnostic compression that guarantees accurate attention for arbitrary queries is extremely challenging, if not impossible. We introduce VerKV, a hybrid approach that provides $(\epsilon, \delta)$ guarantees for approximating attention on any given query. VerKV maintains a compact set of query-agnostic pivotal tokens on the GPU, together with a token-level sketch of the full KV cache, while retaining the complete cache in CPU memory. We introduce a novel stochastic sampling procedure, PoissonComp, to select the pivotal tokens. At inference time, our proposed KVFetch algorithm uses the pivotal tokens and the sketch to estimate the relative error of the resulting attention computation. If the estimate satisfies a user-specified $(\epsilon, \delta)$ tolerance, inference proceeds without accessing host memory. Otherwise, VerKV performs query-aware retrieval to fetch additional tokens from the CPU, refining the attention computation to meet the required tolerance. Thus, VerKV enables attention with bounded approximation error while minimizing CPU-to-GPU transfers. We show that PoissonComp outperforms state-of-the-art approaches to KV cache compression. Combined with KVFetch, it achieves near-dense performance (90-100% of dense quality at 32K context length, averaged over RULER and LOFT), using Qwen3-4B, Qwen3.5-9B and Ministral-3-8B, while reducing the GPU memory footprint by 8-32x and keeping CPU-to-GPU memory fetches within 2-23% of the cache per generated token, depending on the model.
Read the full abstract

All authors Devanshu Jha, Gaurav Kumar, Agniva Chowdhury, Kumar Krishna Agrawal, Aditya Desai

Machine Learning

PROLA: Principal-Orthogonal Low-rank Adaptation for Predictive Spatiotemporal Weather Downscaling

High-resolution weather prediction is important for resolving local atmospheric patterns missed by coarse global forecasts. Large pretrained weather backbones have improved global forecasting, but adapting them to produce future high-resolution fields from recent coarse states remains challenging and costly. We formulate this setting as Predictive Spatiotemporal Weather Downscaling (PSWD), a task that jointly refines spatial and temporal resolution while predicting future high-resolution trajectories. To this end, we propose a framework that decomposes prediction into spatial and temporal stages using a shared pretrained backbone, fixed numerical scaffolds, and lightweight residual heads. For backbone adaptation, we introduce PROLA (PRincipal-Orthogonal Low-rank Adaptation), which allocates a fixed low-rank budget between pretrained weights' principal directions and their orthogonal complement. PROLA further rescales rank-1 optimizer updates using gradient signals, improving adaptation without increasing trainable rank. Experiments in the PSWD setting show that PROLA outperforms representative low-rank adaptation baselines. Further experiments across diverse datasets and tasks support the broader applicability of PROLA, showing that its benefits extend beyond the primary weather downscaling setting.
Read the full abstract

All authors Minseo Yoon, Minseong Bae, Sojin Lee, Hyunwoo J. Kim

Neural and Evolutionary Computing

Understanding Spike Encoding across IoT Sensing Modalities: An Empirical Study

Spiking neural networks (SNNs) offer a path to energy-efficient sensing on Internet of Things (IoT) devices. The input encoder shapes that opportunity by changing the cues available for recognition and the work required to process them. We study these choices across wireless, inertial, surface electromyographic, and tactile sensing. For wireless activity recognition, interleaving reduces estimated energy for arithmetic in the network layers, latency from the prepared input window to returned scores, and energy on the graphics processing unit and system-on-chip (GPU/SoC) rail, compared with repeating the full window on Jetson AGX Orin. With the same input size at each SNN time step, repeating a compact input can yield higher mean development accuracy than interleaving. The inertial experiments show that higher accuracy can come at higher device cost. Direct input can also be faster and more accurate than an event encoder while using less energy on the measured rail, even when its estimated energy for arithmetic in the network layers is higher. Level crossings improve grasp recognition on a later day for previously observed users. In tactile slip detection, level crossings shift the balance between missed slips and false alarms, and checkpoint selection can reverse which input offers higher slip recall. We use these findings to outline how edge sensing systems can profile encoders against the signal structure, the errors that matter to the task, and the cost of the full inference pipeline.
Read the full abstract

All authors Yuxuan Hou, Di Yu, Linshan Jiang, Jin Zhang

Digital Libraries

ReGKG-2025: A GKG-Compatible Reconstruction of the GDELT Global Knowledge Graph for the Missing Period of 14 June to 2 July 2025

The June 2025 outage in the Global Database of Events, Language, and Tone (GDELT) collection pipeline left a gap in the Global Knowledge Graph (GKG), interrupting article-level news metadata used in downstream research. ReGKG-2025 is an archive-derived sample of 281,620 reconstructed article records covering all 1,665 missing 15-minute batches. Articles were recovered from Common Crawl and the Wayback Machine using source quotas estimated from surrounding native GKG data. Text was screened and normalized, non-English bodies were translated, and flagged records were manually reviewed. Eligible articles were converted to the 27-column GKG 2.1 layout. The versioned package includes record-level provenance, archive replay and WARC locators, source summaries, source-by-day calibration weights, and the complete manual decision journal. The sample reaches about 14% of modeled native volume and represents sources accounting for 84% of selected reference record mass. In an intact-week test, 41% of recovered articles matched native GKG records; field agreement on matched pairs is close to the processor baseline (tone MAE 1.69). The dataset supports retrospective content analysis and source-level aggregates; its tone and theme distributions differ from native GKG, and it does not support exhaustive counts. Data: https://doi.org/10.5281/zenodo.23964265.
Read the full abstract

Fluid Dynamics

Flash evaporation Riemann Problem: Thermodynamic Relaxation and Oscillation-Free Real Fluid Finite Volume Method

The Flash evaporation Riemann problem (FeRP) formulated the wave structures of flash evaporation in the vapor-liquid homogeneous equilibrium limit. However, physical flash evaporation involves finite-rate interphase heat and mass transfer, inducing the wave structure to deviate from the equilibrium state. To solve such a non-equilibrium Riemann problem, this paper first derives the governing equation of mechanical equilibrium evolution in a relaxation system, based on which an oscillation-free Homogeneous Relaxation Real Fluid (HRRF) finite volume method is developed. In addition to the Euler equations with mass relaxation, the HRRF method evolves advection equations for the internal energy–pressure affine parameters and the vapor volume fraction, in order to eliminate numerical pressure oscillations inherent in conservative methods while preserving phase-transition-induced pressure changes, and consistently degenerate into a completely pressure-equilibrium method in supercritical flows. The HRRF method is then employed to comprehensively evaluate how thermodynamic relaxation affects the FeRP solution. Results indicate that the non-classical expansion shock of the FeRP loses its traveling-wave characteristics with increasing relaxation time, continuously degenerating into a self-similar rarefaction wave. Compared with the equilibrium limit, an increased mass transfer relaxation time smooths the expansion branch of the RP solution, lowers the intermediate pressure and velocity, and induces a pseudo-splitting of the rarefaction wave as well as a pseudo-advance of the phase transition onset. Meanwhile, the heat transfer relaxation time primarily dictates the temperature difference in the vapor-liquid mixture, thereby modulating the liquid superheat and in turn influencing the mass transfer rate.
Read the full abstract

All authors Haotong Bai, Ping Yi, Yixin Yang, Guoyan Zhao, Wenjia Xie, Mingbo Sun

Differential Geometry

Ricci pseudo symmetric Anisotropic Fluid Spacetime admitting Hyperbolic solitons with Ricci vector field

This work focuses on an anisotropic fluid spacetime with these unique forms of the stress-energy tensor associated with radial pressure and a transverse pressure p. In an anisotropic fluid spacetime , we mani- fest curvatures symmetry such as Ricci pseudo-symmetric and prove that Ricci pseudo-symmetric anisotropic fluid spacetime is generalized quasi- Einstein spacetime. We deduce the Einstein gravitational field equation of Ricci pseudo-symmetric anisotropic fluid spacetime. In the framework of pseudo-Riemannian geometry, specifically, we study the quasi-conformal Ricci pseudo-symmetric anisotropic fluid spacetime, which exhibit the dark matter era. Moreover, we discuss the hyperbolic Ricci soliton and Hyperbolic Yamabe soliton on Ricci pseudo-symmetric anisotropic fluid spacetime with a time-like Ricci vector field. Finally, we turn up the char- acteristics of hyperbolic Ricci soliton and hyperbolic Yamabe soliton on Ricci pseudo-symmetric anisotropic fluid spacetime with a time-like Ricci vector field.
Read the full abstract

Machine Learning

Shape-Adaptive Subspace-Constrained Quadratic Matrix Factorization

Quadratic manifold fitting provides a flexible framework for modeling nonlinear low-dimensional structures embedded in high-dimensional spaces. However, existing methods typically employ prescribed noise models or fixed loss functions, limiting their adaptability to varying noise characteristics. To address this limitation, we propose an alternating maximum likelihood estimation method for subspace-constrained quadratic matrix factorization, termed AMLE-SQMF, which jointly estimates the quadratic manifold and the underlying noise distribution. Specifically, we model the residuals using a generalized Gaussian distribution (GGD), whose shape parameter controls the tail behavior and determines the associated fitting loss. By exploiting the closed-form maximum likelihood estimator of the scale parameter, we eliminate the scale variable and derive a profile negative log-likelihood. Based on this formulation, we develop an alternating optimization algorithm that iteratively updates the manifold parameters and estimates the noise shape parameter. This enables adaptive loss selection within a unified likelihood-based framework, without requiring prior specification of the noise characteristics. Experiments on synthetic and real-world datasets demonstrate the effectiveness of AMLE-SQMF in quadratic manifold fitting and its robustness to diverse noise distributions.
Read the full abstract

Numerical Analysis

A Discrete de Rham method for linear and nonlinear fourth-order plate models

We develop and analyze a Discrete de Rham (DDR) method for the Kirchhoff–Love plate problem on general polygonal meshes and demonstrate its extension to geometrically nonlinear plates. The construction uses local degrees of freedom and polynomial reconstructions, avoiding the implementation difficulties associated with globally $C^1$-conforming approximation spaces. A new deflection reconstruction obtained from the discrete Hessian permits a uniform treatment of the load term at all polynomial degrees. The use of this reconstruction improves the convergence rates of previous DDR methods and enables the formulation of a lowest-order scheme with only vertex unknowns. We provide a stability and convergence analysis of the method and validate the theoretical results through a comprehensive set of numerical tests. The method is then extended to the coupled von Kármán equations, for which we demonstrate convergence numerically. We also study the flat-branch spectrum of a compressed perforated plate and its small-amplitude forced nonlinear response near the first critical load.
Read the full abstract

All authors Daniele A. Di Pietro, Jérôme Droniou, Arax Leroy, Vito Patierno

Statistical Methodology

Causal discernment and the direction of clinical research

A clinical trial can establish an intervention's effects while leaving their causal pathways unresolved. We call causal discernment the practice of identifying explanations that fit the evidence but imply different decisions or research directions, and seeking studies that distinguish them. METHOD: We combine causal graphs and prose around four questions: delivery of the intended target change, routes of benefit and harm, stability of causal relations across regimes, and whether unresolved pathways could change the decision. With quantitative assumptions, graphs support plausible reasoning by showing which findings could distinguish explanations even when effects are not identified. Prose specifies clinical meaning, effect size and timing, and supporting evidence. We use conceptual analysis, hypothetical calculations, CAST and lipid-lowering therapy. SUBSTANTIVE RESULTS: Concordant coronary associations for HMGCR and PCSK9 perturbations support LDL mediation over target-specific explanations predicting unequal effects per unit of LDL reduction, but leave a shared receptor pathway possible. Diabetes associations leave LDL lowering, receptor-mediated uptake and weight gain as possible explanations. An illustrative causal record specifies a genetic comparison and assumptions under which it could distinguish them. CONCLUSION: The causal record formalises the reasoning behind the choice of the next study by making competing explanations, evidence and assumptions explicit and stating what each design could resolve. Prepared before a design or recommendation is adopted, it enables scrutiny of that reasoning and interpretation of results in relation to the stated explanations. Its added value over ordinary rationale review remains to be assessed.
Read the full abstract

Materials Science

sdos: spectral densities of large sparse Hermitian and generalized eigenproblems

sdos is an open Python package that computes the density of states (DOS) and related spectral quantities of very large sparse Hermitian matrices A, and of generalized problems A psi = lambda B psi, from Chebyshev moments estimated with random vectors. The density is reconstructed by the kernel polynomial method (KPM, the default) or by maximum entropy (Maxent); the same interface gives band energies, Fermi levels, local and projected densities, harmonic thermodynamics, exact HOMO-LUMO gaps and eigenpairs in a spectral window. Interfaces cover electronic (empirical tight-binding, DFTB, SIESTA) and vibrational (matrix-free LAMMPS Hessians) problems. I checked most ingredients against dense diagonalization or an independent code; the exceptions are listed in Sec. V. On a workstation (4 threads, 19 GB) the DOS of an 884,736-atom amorphous silicon model (a 6^3 tiling of a 4096-atom cell; 3.5 million orbitals; the orthogonal empirical tight-binding model of Kwon et al.) takes 71 s (M=100, R=2, no exact moments; 383 s with the default exact moments), and a matrix-free 2,097,152-atom crystal 10 min in 1.8 GB. A 17^3 replica of a 216-atom amorphous cell (1,061,208 atoms) agrees with exact Bloch spectra: electronic DOS in 229 s, matrix-free ACE vibrational DOS in 2.2 h on one GPU. Shift-invert, incomplete-Cholesky setup, exact moments of high order and Maxent at high moment counts do not scale. The underlying methods are due to Silver, Roder, Kress, Voter, Drabold, Sankey and collaborators; the Introduction states what is added.
Read the full abstract

Metric Geometry

Fractal solutions to the two-dimensional floating body problem

In this paper, the first construction of an entire class of two-dimensional fractal bodies that float in water in equilibrium is explicitly given. Their boundaries are described by planar fractal Zindler curves which are constructed using fractal projective hedgehogs. The areas of these bodies are seen to be well-defined and their expression is explicitly computed. To this end, a result concerning the continuity of the Lebesgue measure with respect to the Hausdorff metric for compact sets is given. Furthermore, the Hausdorff dimension $w$ of these fractal curves is shown to coincide with the Hausdorff dimension of the graph of a Weierstrass function associated to the corresponding fractal projective hedgehog. The fractal Zindler curves are proved to have halving chords of constant length that split both the area and the $w$-dimensional Hausdorff measure of the curve into exact halves.
Read the full abstract

Cryptography and Security

SkillCODER: Towards Auditing and Attribution of Copyright Infringement in LLM Agent Skills

Agent skills package reusable workflows and task-specific procedures into deployable natural-language artifacts for LLM agents. Their scale and readability make them valuable but vulnerable to unauthorized redistribution and repackaging. Existing prompt copyright auditing methods struggle with long skills and introduce token-level perturbations that compromise readability. We propose SkillCODER for copyright auditing and buyer attribution in a black-box, content-only setting. During embedding, SkillCODER parses a skill into a section-aware intermediate representation, SkillIR, to identify structural anchors. It distributes conditional instructions across these anchors to form an Artifact-Gated Capsule, designed to activate under verification queries and remain dormant otherwise. Controlled-Vocabulary Error-Correcting Codes encode buyer-specific fingerprints through controlled lexical choices within the capsule. During verification, the defender issues paired positive and negative probes, extracts capsule evidence, and recovers ownership and buyer identity through error-correcting decoding. Across three domains, six models, and three agent frameworks, SkillCODER achieves 99.2% owner verification and 99.3% buyer Top-1 attribution accuracy while preserving task fidelity, substantially outperforming prompt-level baselines. Embedded signals remain robust under paraphrase, compression, clause deletion, and reorganization. Under five adaptive attacks, including a method-aware LLM sanitizer and multi-buyer collusion, ownership evidence remains reliably detectable while buyer attribution degrades gracefully.
Read the full abstract

All authors Enhao Huang, Chunshu Xia, Yiming Li, Yuchen Yang, Bingrun Yang, Zhan Qin, Dacheng Tao, Kui Ren

Statistical Methodology

Sharp bounds for posterior causal attribution given complete factual records

Retrospective causal attribution evaluates whether altering an earlier event would have prevented an observed outcome. Observational data determine marginal potential-response probabilities at fixed inputs while leaving cross-world joint distributions unspecified. Conditional on a complete factual record in a finite Markovian structural causal model, independence across mechanisms is preserved and each unobserved conditional response distribution is restricted to a capped probability simplex. Every collection of conditional distributions satisfying these local bounds is jointly attainable within a single compatible structural causal model. Optimizing the counterfactual expectation over the resulting product domain yields sharp attribution bounds for multistage interventions and mediated effects, and compatible response-vector polytopes extend these bounds to structural response restrictions. Exact algorithms for causal chains and general directed acyclic graphs, consistent plug-in endpoint estimators, and finite-sample and asymptotic confidence intervals are developed and illustrated using synthetic reactor experiments and health survey data.
Read the full abstract

Computation and Language

ScrapeBench: Evaluating Legal Compliance of AI Agents in Website Scraping

AI agents deployed on the open web are becoming more capable of performing actions that may breach contractual provisions and violate computer security laws. In this paper, we introduce ScrapeBench, the first agentic benchmark for empirically evaluating the legal compliance of AI agents with website scraping policies, namely terms of service (ToS) and robots$.$txt directives. Across widely used closed-weight and open-weight AI agents, we find that most agents exhibit low compliance rates (below 5%) when instructed to scrape websites in our sample of 1,626 popular websites that explicitly prohibit scraping. In addition, we observe that even where AI agents actively check the applicable ToS and robots$.$txt, they continue to exhibit low compliance rates. Conversely, in our sample of 269 websites that do not prohibit scraping, we observe that agents sometimes refuse to scrape despite it being lawful, such as Claude Opus 4.6 exhibiting an over-refusal rate of 98.5%. Our human baseline experiments (n = 180 participants) suggest that humans exhibit compliance rates comparable to AI agents but over-refuse at much higher rates.
Read the full abstract

Cryptography and Security

Least-Privilege Auditing of Agent Tool Calls with Intent Graphs and Reasoning-Guided Escalation

LLM agents that call tools are exposed to indirect prompt injection: instructions hidden in tool outputs can steer an agent into calls the user never asked for. We present a runtime audit that derives an executable intent graph from the user's objective and the deployment's tool catalog with one LLM call per task, and checks every tool call against it with a deterministic loop that makes no LLM call (median 5.7 microseconds per call). Two further mechanisms recover utility. Lookups of the user's own state bypass the audit, because the harm of an attack lies in its write or outward step. The calls that remain flagged are escalated to a reasoning guardian, in the spirit of tool-integrated reasoning models such as START: a reasoning model that sees only the objective, the plan, the call history without tool outputs, and the flagged call, and can verify facts with a restricted Python-expression tool. A current agent showed no baseline attack success, so end-to-end validation remains open.
Read the full abstract

Computer Vision

TransArt3D: From Two States to Articulated 3D Reconstruction with Video Transition Priors

Reconstructing interactable articulated assets from a few images is key to robotics and simulation, yet highly underconstrained. Existing optimization-based methods fit an articulated asset under a static 3D prior that scores each state independently, so per-state reconstruction errors leak into the estimated moving part and joint as false motion. Video diffusion models capture how objects move between states, but only implicitly in 2D. To this end, we present TransArt3D, an optimization-based framework that reconstructs an articulated 3D asset from two endpoint images by translating this video transition prior into explicit part structure and joint kinematics. It first lifts frames generated between the endpoints with a transition-coupled 3D consensus that applies the static 3D prior jointly across frames, stabilizing unchanging regions and initializing the parts and joint. A rectified-flow kinematic refinement then lets the video prior supervise a 3D motion field over the whole transition, while part assignment and a shared joint constrain the field throughout optimization, keeping per-state errors out of the articulation. Finally, a motion-aware canonical appearance fusion uses the recovered articulation to map both endpoint images onto one canonical surface, texturing regions that only one state reveals. On PartNet-Mobility, TransArt3D outperforms methods given the same two endpoints in geometry and joint accuracy, and surpasses methods given six observed states in articulated-state geometry, joint localization and appearance.
Read the full abstract

All authors Chenhao Guan, Liying Yang, Xinxing Yu, Hao Mo, Tianshun Han, Renjun Gao, Yanyan Liang

Distributed Computing

NCCL-BOT: Closing the Loop on NCCL's Cost Model with Online Bandits

NCCL picks an algorithm-protocol pair for every collective from an analytic cost model (AUTO) that never sees a measured latency. The model is a strong prior but a brittle final decision: priced on forced-arm runs, its choice is exact on most contexts and badly wrong in a few. We built NCCL-BOT, a tuner plugin that closes the loop: a per-context bandit, warm-started from the model, learns from measured latencies. We replay a forced-arm H100 sweep, tuned at 4-8 nodes and scored at 16-32, and price the model's choice on the same forced arms; that sweep has no stock-NCCL run. The bandit cuts the contexts where the model's choice loses more than 10% from 19 to 1 (p99 103% to 9.9%, mean 6.1% to 1.6%; p=0.047 over nine scenarios after Holm correction). The prior and continued learning both matter: cold-started bandits, and committing after one probe of each arm, lose more on the mean in every scenario. Supervised selectors fit once on the same data and tuned with the same budget match the mean, and their heavier tail is not significant (7 against 1 contexts above 10%, p=0.06); at seen scales they match or beat online selection. With hyperparameters frozen from H100, UCB1 cuts AUTO's p99 from 60% to 23% and from 56% to 15% on two GB200 fabrics, but its contexts above 10% go from 16 to 8 on one and from 10 to 13 on the other, and scenario-level tests do not separate the two. Live, on two latency-emulated node sets of one H100 system with one run per policy, the learners repair most of a gross AUTO error on one (selection loss 31% to 6-8% for deterministic Greedy and UCB1; average all-reduce bandwidth 1.18-1.24x stock) but average 0.94-0.99x stock on the other, and on both lose to stock below 1 MiB (on the first through plugin overhead); on H100 the cost model is exact there offline, so small sizes can be left to it. We do not show a net end-to-end win.
Read the full abstract

All authors Hanan Shteingart, Amedeo Sapio, Chen Avin, Devendar Bureddy, Manjunath Gorentla Venkata, Srivatsan Krishnan, Shie Mannor, Sreeram Potluri, Gil Shabat, Zsolt-Alon Wertheimer

Soft Condensed Matter

A Rule That's Made to be Broken? Reframing the Arrhenius Law for Ionic Transport

The Arrhenius law is ubiquitous in the physical sciences. It describes how an observable property scales with temperature, $T$, as $\exp(\pm E_A/k_BT)$ for a constant $E_A$, the 'activation energy'. Fitting to the Arrhenius law is a common procedure for both experimental and computational data. However, the physical significance of any 'activation energy' calculated is unclear. Here we focus on self-diffusion coefficients for potential solid electrolytes, and by identifying the individual 'diffusive steps' within the system, show how a physically significant activation energy can be extracted. There are also many examples of non-Arrhenius behaviour including `super-Arrhenius' curvature, where the effective activation energy appears to increase with decreasing temperature. We demonstrate how it is still possible to extract the underlying Arrhenius law, by quantifying self-correlations such as back-and-forth motion. The methodology described here is material-agnostic and can be utilised to analyse the molecular dynamics trajectories of both ordered and disordered materials.
Read the full abstract

Probability

Phase Transitions for Random Least Common Multiples with Memory

We study the least common multiple of a random subset of $\{1,\ldots,n\}$ when the selection follows a correlated Bernoulli process with memory. The copying rule preserves the marginal retention probabilities and creates a genealogical forest. At fixed density, the logarithmic least common multiple exhibits a fluctuation phase transition at the memory parameter $a=1/2$. The boundary limits depend on how genealogical families are distributed among the divisibility classes carrying the leading arithmetic weight. Our proofs combine the von Mangoldt representation with a reduction to large primes. At fixed density, an exact expansion separates an innovation martingale from a memory term. An asymptotic projection onto the retained count identifies the contribution of memory. Marked point-process convergence and estimates for rare genealogical families give the boundary limits.
Read the full abstract

Algebraic Geometry

All-genus mirror symmetry for the type-$D$ minimal resolution Calabi-Yau threefold

Let $Y$ be the product of a minimal resolution of a binary dihedral surface singularity with an equivariant line, with volume-preserving scalar action. We identify its stable equivariant descendant Gromov–Witten theory and its free energies in genus at least two with the sign-sector topological recursion of the degenerate type-$D$ relativistic Toda spectral curve. The $R-$matrix calibration is determined at a partial large-radius limit, where the trivalent exceptional curve retains a nonzero Kähler parameter, while the remaining parameters vanish. The limiting theory on the mirror side is a sum of twisted point theories and the local $(-2,0)$ curve theory. Moreover, we identify the equivariant K-group supported on the surface zero section with an integral relative homology group of the mirror. Explicit cycles for the dual tautological bundles realize their Gamma central charges and all stable Gamma-framed descendants.
Read the full abstract

Computation and Language

EASE-KV: Diagnosing the Gap Between Memory Availability and Effective Use

Reusable key-value (KV) memory avoids repeatedly prefilling retrieved text, but stored and injected memories do not necessarily become usable evidence. We present EASE-KV, a controlled framework for understanding this gap under fixed candidates and a frozen language model. EASE-KV separates three interfaces: whether required relations are preserved during encoding, whether preserved memories receive appropriate influence during allocation, and whether outcome supervision can identify useful allocations. Matched experiments show that causal encoding improves compositional accuracy by 25.0 percentage points over block-diagonal encoding, while the tested inference-time bridging methods fail to recover the lost contextual information. With unchanged stored states, diagnostic evidence-guided allocation improves LongMemEval accuracy from 31.2% to 38.4%, demonstrating that memory influence can be controlled without re-encoding. However, controllability does not guarantee learnability: binary correctness provides informative reward contrast for 53.0% of sampled query groups, compared with 73.6% under the evaluated shaped-reward setting. These findings show that effective KV memory requires more than retrieval and injection: the required relations must be encoded, preserved memories must receive appropriate influence, and supervision must distinguish better allocations from worse ones.
Read the full abstract

All authors Di Lu, Zechen Li, Xuanyu Yue, Jiabo Fan, Hao Xue, Flora D. Salim

Number Theory

New families of maximal curves not covered by the Hermitian curve and their cyclic automorphism groups

For every prime power $q>2$, integer $n\ge2$ and $1\le r\le 2n-1$, we construct an $\mathbb F_{q^{2n}}$-maximal curve $T_{q,n,r}$ with Kummer equation \[ z^{q^n+1}=\eta^{-R}y(1-y^{q-1})^R, \qquad \frac{q^r-1}{q-1},\] where $\eta\in\mathbb F_{q^{2n}}^*$ satisfies $\eta^{q^n}=-\eta$. For fixed $q,n$, these are cyclic covers of the projective line over a single field, indexed by the integer $r$. Writing $\delta=\gcd(q^n+1,R)$, we prove that the genus is $(q-1)(q^n+1-\delta)/2$ and that $T_{q,n,r}$ is covered by the Hermitian curve $\mathcal H_{q^n}$ over an algebraic closure if and only if $\min(r,2n-r)$ divides $n$. For $1<r<n$ and $\delta=1$, the full automorphism group is cyclic of order $(q-1)(q^n+1)$, and distinct values of $r$ give curves that are nonisomorphic over an algebraic closure despite having the same genus and full automorphism group. We also give explicit equations and genera for all their Galois subfields. When $n\ge 4$ is a power of two, this produces families over a fixed field whose members are not covered by $\mathcal H_{q^n}$.
Read the full abstract

Analysis of PDEs

Regularity and Asymptotic Radial Symmetry of Solutions to Mixed Local-Nonlocal Parabolic Equations

We study the regularity and the long-time asymptotics of the solution to the following mixed local-nonlocal parabolic problem \[ \left\{ \begin{alignedat}{2} \partial_t u - \Delta u + (-\Delta)^s u &= f \qquad &&\text{in }\Omega\times(0, T), u&=0 \qquad &&\text{in }\bigl(\mathbb{R}^{n}\setminus\Omega\bigr)\times[0, T), u(\,\cdot\,,0)&=u_{0} \qquad &&\text{in }\Omega, \end{alignedat} \right. \] where $\Omega\subset\mathbb R^n$ is a bounded domain, $s\in(0,1)$ and $(-\Delta)^s$ denotes the fractional Laplacian. We first develop the well-posedness theory for the problem on finite time intervals ($0<T<+\infty$). Existence and uniqueness of weak solutions are established in the natural energy space, followed by interior $H^{2,1}$-regularity and global $L^\infty$-bounds. Then, under the restriction $s\in(0,1/2)$ and $0<\alpha\le 1-2s$, we prove that weak solutions become classical ones, obtaining $C^{2+\alpha,1+\alpha/2}$-regularity up to the boundary via the method of continuity and a perturbation argument. The accompanying a priori estimate guarantees that these classical solutions can be extended globally in time, thereby allowing us to consider the parabolic problem on the whole half-line ($T=+\infty$). Finally, for the special case $\Omega=B_1(0)$ and $f=0$, we employ the moving-plane method together with the previously established boundary regularity to show that every positive classical solution is attracted, as $t\to +\infty$, by a steady state that is radially symmetric and strictly decreasing.
Read the full abstract

Robotics

Single-Radar Bilateral 3D Near-Field Imaging for Mobile Robots

Mobile robots operating in homes need reliable 3D perception of nearby furniture and surrounding structures, yet recovering this information on both sides of the robot's path with a single forward-mounted millimeter-wave radar remains challenging because of near-field wavefront curvature, viewpoint-dependent target visibility, and limited elevation diversity. This paper presents a motion-assisted framework for reconstructing structures on both sides of a robot's trajectory. Our near-field model relates scene reflectivity to stepped-frequency multiple-input multiple-output (MIMO) observations across frequencies and robot positions using spherical paths from transmitter (Tx) to scatterer to receiver (Rx). Bilateral Tx combining provides wideband directional weighting, while sliding-aperture back-projection coherently focuses the measurements using near-field propagation distances. Top-K power fusion combines local images with different visibility. The reconstruction provides lateral-forward support for multi-layer height extraction, followed by guarded adaptive height refinement using the original MIMO channels. Experiments in controlled, office/study, and corridor environments produce mean lateral-forward localization errors of 4.65, 5.63, and 5.98 cm, respectively, with peak-to-background ratios (PBRs) exceeding 13 dB. The framework recovers bilateral 3D scattering structure from one forward pass using a single radar, without relying on learned representations of the scene.
Read the full abstract

All authors Yingqi Wang, Tingwei Chen, Zhongqin Wang, Min Xu

Databases

Samyama-SDG: A Declarative GraphQL @count and Regex/Ontology Synthetic Data Compiler with $O(1)$-Memory Referential Integrity and Causal Ground Truth

Synthetic data generators face a divide between declarative relational mockets and domain causal simulators. Declarative generators specify entities, regexes, and foreign keys via schemas, but buffer parent primary-key pools in RAM ($O(N)$ memory) and lack ontology bindings or causal ground truth. Conversely, domain simulators emit realistic streams and ground-truth labels, but require thousands of lines of procedural code per domain. We present Samyama-SDG (sdg-rs), a Rust synthetic data compiler unifying declarative schema compilation with ontology-bound causal simulation. At Layer 1, Samyama-SDG compiles Cardinality-Extended GraphQL schemas (@count, @pk, @ontology, @regex, @dist, @fk, @after) and an Attribute Data Dictionary into a topologically sorted DAG. Combining stateless counter-based seekable PRNGs (SplitMix64 over $(\text{seed}, \text{entity}, \text{row}, \text{field})$) with index-bijective primary keys $f_{\text{PK}}(s, E, i)$ lets child tables resolve uniform or Zipfian foreign keys and multi-hop @after constraints in $O(1)$ memory without buffering parent tables. At Layer 2, causal processes layer fault propagation and machine-scorable ground truth (ground_truth.jsonl) over the same seekable PRNG and samyama-ontologies binder. Across three schemas, Samyama-SDG generates $1{,}000{,}000$ records in $128.79\text{ ms}$–$172.43\text{ ms}$ ($5.80\text{M}$–$7.76\text{M}$ rows/s) and $5{,}000{,}000$ records in $676.12\text{ ms}$–$808.56\text{ ms}$ at flat $31.4\text{ MiB}$–$35.1\text{ MiB}$ memory, with $0$ conformance errors across $1{,}150{,}000$ foreign keys and $900{,}000$ temporal constraints.
Read the full abstract

2611.11295 · 8 Nov · 8 pages, 3 figures, 4 tables. v2 adds an end-to-end differentially private fit-generate-audit example on a public table (Section 4) and positioning on orthogonal differential-privacy marginal fitting via the post-processing theorem and updates companion citations. Code: https://github.com/samyama-ai/samyama-synthetic-data-generator (tag v2.1.0)

Group Theory

Property $R_\infty$ for Higman$\unicode{x2013}$Thompson Groups

We prove that every Higman$\unicode{x2013}$Thompson group $G_{n,r}$, with $n\ge2$ and $r\ge1$, has property $R_\infty$. In particular, every automorphism of Thompson's group $V = G_{2,1}$ has infinitely many twisted conjugacy classes. These results answer questions of Olukoya and of Lins de Araujo, de Oliveira-Tosti and Santos Rego. We also prove that a homeomorphism under which no eventually periodic sequence has a finite orbit has a locally finite centralizer in $G_{n,r}$.
Read the full abstract

Software Engineering

AWS Serverless Infrastructure from WebAssembly Components

The infrastructure definitions and the code for serverless applications are often only loosely connected. This yields a large error class: the referenced infrastructure components might not exist. It is also hard to keep the application and its permissions in sync. To address this problem we present an infrastructure from code solution where we provision cloud infrastructure from Wasm Interface Type (WIT) - the language defining the interface of WebAssembly components. We limit ourselves to the basic example of an AWS Lambda function accessing S3 buckets. We build two interpreters for the WIT language: one that derives the definition of S3 buckets as infrastructure as code and least-privilege IAM permissions, and a generic runtime that runs the component as a Lambda function and provides access to the S3 buckets as capabilities. We formulate consistency properties that ensure the existence of the S3 buckets and correct permissions of the Lambda, and provide evidence that our interpreters satisfy them. We show how selected failures for deployment of such cloud components can be prevented using our technique and how a baseline setup fails at this task. Together with the strict sandboxing of the WebAssembly runtime we achieve an instance of capability-based security for access to AWS infrastructure as a further result.
Read the full abstract

High Energy Physics (Phenomenology)

Rapid, High-Precision Numerical Evaluation of Multiloop Feynman Diagrams

We numerically evaluate multiloop Feynman diagrams by using Sinc function approximations to transform the integrals into rapidly convergent, multidimensional sums. We compare the efficiency of this approach to Monte Carlo evaluations of electron g-2 terms. We show that the two strategies reach parity at precisions of around 1% for the full sets of three and four loop diagrams. Beyond this level Sinc function methods are dramatically faster - computing 3-loop integrals to six digit precision takes minutes whereas Monte Carlo timings map to years. We show that the method can evaluate individual 5-loop g-2 integrals and discuss extrapolation techniques that would further improve performance. We note that the method appears to be applicable to more general diagrams.
Read the full abstract

Distributed Computing

Multi-Node Transformers for Object Detection and Captioning Across Distributed Edge Networks

This paper presents a unified transformer-based framework for object detection and dense region captioning across distributed edge networks. The architecture uses a shared visual encoder followed by detection, region extraction, and autoregressive language-decoding stages, allowing localization and region description to reuse the same image features. This design removes the second image-encoding path required by decoupled detection and captioning pipelines. To support execution on resource-limited devices, a workload-allocation formulation assigns model stages and input streams across connected edge nodes while accounting for execution time, memory capacity, processor utilization, power limits, and intermediate tensor transfer cost. The framework is evaluated on a six-device edge testbed comprising four NVIDIA Jetson Orin Nano and Orin NX inference nodes and two Raspberry Pi 5 auxiliary CPU nodes, with TensorRT FP16 and INT8 execution on the NVIDIA devices. On a Jetson Orin Nano, the unified INT8 model achieves 38.1 FPS with 3.5 GB peak memory use and 11.8 W measured power. Under four concurrent streams, the proposed allocation achieves 35.2 FPS and 28.4 ms mean end-to-end latency with an internode transfer rate of 6.8 MB/s.
Read the full abstract

Analysis of PDEs

Critical radial Green potential heat flow: A tension-controlled residue and a two-bubble example

We compare two heat flows for the same corotational map of a disk into a warped product of two spheres, with the same fixed boundary values. The ordinary harmonic map heat flow develops a singularity in finite time. When the domain metric is instead determined by the critically normalized Green potential of the current energy distribution, the solution exists for all time. Two harmonic spheres of energy $4\pi$ form at separated scales, in addition to a nonconstant harmonic root. The energy and the image diameter of the necks tend to zero without restriction to a sequence of times. The main estimate controls the difference between the inner source mass and its critical value by the weighted tension. This is necessary because that difference is integrated over a neck of unbounded logarithmic length. Along a sequence of times the Green metric on the middle neck is asymptotically cylindrical after a spatially constant normalization.
Read the full abstract