Peer Revenue Now accepting positions

Papers

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

All of arXiv

Friday 6 November

Probability

Mesoscopic eigenvalue statistics for random regular graphs: a case study

We study mesoscopic fluctuations of linear eigenvalue statistics for random $d$-regular graphs on $N$ vertices with fixed $d\ge 3$. Revisiting the approach developed by Dumitriu-Johnson-Pal-Paquette (2013) and Johnson (2015), we establish a mesoscopic central limit theorem in the bulk and identify its universal limiting variance. Our result applies to test functions with smooth compactly supported Fourier transforms and to a specific range of spectral scales $(\log N)^{-1}\ll\eta\ll 1$. In particular, this case study allows the mesoscopic scale to tend to zero arbitrarily slowly.
Read the full abstract

Digital Libraries

Replication Certification at Publication: Checkable by Anyone, Anywhere, in Seconds, Forever

Authors increasingly deposit replication datasets—the data and code behind publications—in archives, and journals increasingly require them to, under progressively onerous rules. But a deposit alone proves little, so authors keep paying: months for the journal's reanalyses, years answering readers' questions, and rancor as failed replications become accusations because no record establishes what actually ran. We introduce a replication certificate: a cryptographically signed record of the deposited files and logged output. It replaces the journal's reanalyses with evidence anyone can check in seconds, which no one, including the archive, can backdate, and no one can fake without risking detection. Certification asks little of anyone and benefits everyone: The author deposits, then reruns the analysis once under a random seed the archive computes from the deposit's hash and a blockchain block; the archive signs, timestamps (in that blockchain), and stores the record—confidential data included—without rerunning anything; the journal merely checks a website for the certificate; and anyone can check or replay any step decades later.
Read the full abstract

Risk Management

Joint modelling of financial cost and carbon footprint in motor insurance claims: a BINAR(1) decision framework for circular auto parts adoption

Motor insurance claims management increasingly faces a dual objective: controlling claim costs while reducing greenhouse-gas emissions associated with vehicle repair decisions. This paper proposes a bivariate, discrete-time modelling framework that links the dynamics of financial claims and repair-related carbon footprint inspired by a BINAR(1) dependence structure, allowing for temporal persistence, cross-dependence and a positive/negative correlation between the two dimensions. Using externally sourced emission factors and cost information for replacement parts, we operationalise the carbon footprint of claims under two repair pathways: new parts versus circular-economy reused parts (PIEC). We then conduct scenario analyses in which the acceptance rate of reused parts varies, and we quantify the joint implications for expected monetary outlays and CO 2 -equivalent emissions. Results show that higher PIEC acceptance can simultaneously reduce expected emissions and average claim cost, and the magnitude of the joint benefit depends on the dependence structure and persistence captured by the BINAR(1) process. The proposed framework provides a transparent, reproducible tool for insurers to evaluate repair-policy levers and to align claims operations with decarbonisation targets while retaining an explicit stochastic risk perspective.
Read the full abstract

Software Engineering

On the Limits of Execution Traces for Repairing Assertion Failures in LLM-Generated Code

Iterative repair returns a failed program's error to a large language model (LLM) and asks it to fix it. After iterative repair, the remaining failures are mostly assertion failures: with four Qwen2.5 models, they make up 79% of residual failures on BigCodeBench, all on HumanEval and 93% on LiveCodeBench. We argue, through an information view of the feedback, that an assertion failure says little about where the fault lies. We propose Breakpoint Debugging (BPD), which returns the failed program annotated inline with its variables' runtime values. We take the tasks that iterative repair failed to fix with five repair attempts and compare two ways of spending five further generations on them: BPD, and resampling. On BigCodeBench, BPD repairs 26.7% of these tasks against 22.4%, and its advantage grows with model size. On HumanEval, BPD outperforms resampling but only slightly, and on LiveCodeBench, fresh sampling repairs more (6.5% against 4.9%). Furthermore, BPD reduces unchanged-code resubmissions from 28% to 8% on BigCodeBench. Using only the first failing test's trace, it outperforms tracing all failing tests while using fewer tokens than both all-trace BPD and plain iterative repair. On LiveCodeBench, BPD gives no gain even on the public tests, so a visible failing input is not enough for a trace to help. These results match the information view: a trace helps only when the test exposes a failing input, and the fault is an implementation slip; otherwise, a repair loop should resample rather than trace.
Read the full abstract

Solar and Stellar Astrophysics

Chromospherically Active Stars in the MeerKAT Galactic Plane Survey

Stellar magnetic activity has long been established to occur in subsets of late-type main-sequence stars and manifests itself in the form of chromospheric and coronal activity. This activity is driven by magnetic fields generated by dynamo mechanisms associated with stellar rotation and convection, which channel and dissipate magnetic energy to the outer atmospheric layers. Optical spectroscopic follow-up was carried out for 99 radio-X-ray coronally active stars of G-, K-, and M-spectral types in the SARAO MeerKAT Galactic plane survey point sources. There are 46 stars that exhibited chromospheric activity. Specifically, 39 stars show Ca II H&K emission lines, whereas 18 show Ca II IRT, 11 show both H&K and IRT lines. Of the 39 stars displaying H&K emission, 37 show activity levels greater than the Solar minimum activity. Their X-ray to bolometric luminosity ratios are within the coronal thresholds of > -4.5. In addition, we explored optical photometry to investigate the rotation information and found that these stars show short rotation periods; in comparison with activity levels, the activity levels dampen as the rotation periods increase. We also analysed optical light curves for a subset of stars, and they show flaring behaviour at the same epoch as enhanced chromospheric activity, indicating the flares and chromospheric emission are manifestations of the same underlying magnetic activity process.
Read the full abstract

All authors Okwudili D. Egbo, David A. H. Buckley, Paul J. Groot, Orapeleng Mogawana, Anke van Dyk, Yusuke Tampo, Simon de Wet

Robotics

TPS-A*: Path Planning for Any-Shape Vehicles with Certified Collision Checking in Trajectory Parameter Space

Global planners for ground vehicles usually approximate the vehicle by a disk, trading safety against conservativeness, while exact-footprint checking pays for every candidate motion, pose, and footprint edge. We present TPS-A*, a lattice planner whose edges are kinematically feasible trajectories from parameterized trajectory generators, built on certified collision checking in trajectory parameter space. A grid precomputed per footprint returns, in one pass over the local obstacles, a certified free distance along every candidate trajectory, so the per-node collision cost is independent of footprint complexity and trajectory count. We prove that this grid never overestimates the free distance of the continuously swept, possibly non-convex footprint, and state the planner's guarantees: collision-free paths and bounded suboptimality relative to its lattice abstraction. Goals with or without heading are reached exactly by Reeds-Shepp shots certified as trajectories of the same generators. An exact independent test finds no collision in its paths. On 300 cluttered worlds, with the true footprint and the same goal pose for all planners, it solves every query with a collision-free goal pose in a median of three milliseconds on an AMD Ryzen 7 2700X, on par with hybrid A*, 45% of whose paths overlap obstacles, and nearly sixty times faster than a state-lattice planner, the only other collision-free one; with position goals, eighty-nine times faster than hybrid A*. It matches hybrid A* in automated parking and on narrow doorways, docks a non-convex fork footprint that no convex model admits, and drove a real robot across a university campus among people.
Read the full abstract

All authors Jose Luis Blanco-Claraco, Shravan S. Rai, Fernando Cañadas-Aránega, José Luis Torres-Moreno, José Carlos Moreno-Úbeda

High Energy Physics (Theory)

Carrollian minimal massive gravity and the fate of third-way consistency

We construct the Carrollian limit of minimal massive gravity. The limit is singular for the algebra and regular for the Chern-Simons-like data. The one divergent bracket points along the null direction of the contracted bilinear form. The Lagrangian is therefore finite while the brackets are not. The contracted curvature still obeys a cubic closure condition, forced on shell rather than identically. The third-way coupling itself, however, is removable. For generic parameters an invertible redefinition of the auxiliary field and the boost connection takes it to Carrollian topologically massive gravity up to a boundary term. The same redefinition fails before the limit. We solve for an anti-de Sitter vacuum and a Carrollian black hole, whose rotation parameter leaves the frame and survives only in the boost connection. The degenerate vector flavor metric obstructs the standard charge formula, which we replace by a canonical one. The boundary charges, obtained in two independent ways, carry no trace of the deformation. A Dirac analysis gives nine first-class constraints and no local degrees of freedom on any background. The massive graviton is removed. The asymptotic symmetry algebra is bms_3, with central charges carried by the Einstein and Lorentz-Chern-Simons terms. The black hole has two sets of thermodynamic data, each with its own first law. One obeys the Bekenstein-Hawking area law, the other keeps the Chern-Simons entropy.
Read the full abstract

General Economics

The Network Adaptive Market Hypothesis: Detection, Attribution, and Network-Mediated Efficiency

We propose the Network Adaptive Market Hypothesis (NAMH), a novel extension of the adaptive markets framework that conceptualizes market efficiency as a network-dependent equilibrium emerging from long-memory predictability and cross-market information transmission. Using G20 equity indices and five commodity futures over 2001-2026, we identify Japan as the central information hub, while the United States primarily acts as an information broadcaster. Contrary to conventional expectations, systemic crises do not exhibit higher network spectral radii, challenging the assumption that crisis-related fragility necessarily manifests through increased network amplification. Instead, crisis-related spectral contributions are dominated by Australia and the United States. Importantly, variance-decomposition connectedness is more closely associated with network modularity than with spectral radius, with crisis-period increases driven primarily by cross-cluster spillovers. Notably, efficiency-based market groupings explain network connectedness only marginally compared with geography and asset class, while the middle efficiency tier consistently exhibits the greatest idiosyncratic behaviour.
Read the full abstract

Distributed Computing

RPEG: Training Reasoning Language Models Scalably and Cost-Effectively

Reinforcement learning (RL) is now central to training Reasoning Language Models (RLMs), but RLM pipelines are difficult to scale and costly to run because each RL step couples multiple large models, heterogeneous stages, cross-step state, and complex asynchronous dependencies. Existing systems typically optimize only parts of this design space. We address this complexity by jointly optimizing the pipeline structure as well as its execution using a novel data-centric graph intermediate representation called RLM Pipeline Execution Graph (RPEG). RPEG exposes structural choices such as bounded asynchrony and actor-critic sharing together with execution optimizations such as stage fusion, fine-grained scheduling, and multi-dimensional model parallelism. It is coupled with a comprehensive analytical performance model for RLM training, jointly capturing compute, communication, memory, and hierarchical rack-, node-, and GPU-level effects. Across systems with up to 256 GH200 GPUs, our distributed runtime engine implementing RPEG improves performance by up to 2x over ReaL and reduces step latency by up to 2.48x over verl, while increasing hardware utilization by up to 30%. Overall, RPEG provides a graph-driven foundation for next-generation training pipelines for frontier reasoning LLMs.
Read the full abstract

All authors Maciej Besta, Leonard Schmidt, Pierre Pang, Patrik Okanovic, Robert Gerstenberger, Torsten Hoefler

Number Theory

Bounded Ratios of Best Approximations: Sharp Bounds and Spectra of Diophantine Exponents

We solve Problems 1, 2, and 3 posed by Marnat, Moshchevitin, and Schleischitz (arXiv:2505.15964). We consider totally irrational real n-by-m matrices and two conditions on their sequences of best approximations: (B) bounded successive height ratios and (C) bounded successive error ratios. Under (C), we prove that the ordinary Diophantine exponent is at most m and, when both dimensions are at least two, that the uniform exponent is at most m^2/(m+1). Combining these bounds with known results, we determine the individual spectra of both exponents under (C), with and without the additional assumption (B). When both dimensions are at least two, imposing (B) changes neither spectrum. A key ingredient is a block lemma for consecutive best approximations lying in a fixed rational subspace.
Read the full abstract

Networking

Computation-Efficient Aerial-Marine Integrated Networks for Search and Rescue via Cooperative HAPS, UAVs, and MASSs

In this paper, we propose an aerial-marine integrated network architecture tailored for maritime search and rescue (MSAR) operations, where a high altitude platform station (HAPS) and maritime autonomous surface ships (MASSs) are equipped with edge servers, providing computing services for surveillance uncrewed aerial vehicles (UAVs). These UAVs generate substantial volumes of computation-intensive and delay sensitive tasks, some of which can be strategically offloaded to both MASSs and HAPSs for distributed processing. We formulate the joint task offloading and resource allocation as a mixed-integer nonlinear program (MINLP) to minimize the system computation overhead (CO), quantified by a weighted sum of task completion time and energy consumption. Given the stochastic nature of task arrivals from UAVs and the dynamic characteristics of marine channel conditions, we employ a two-stage optimization to decompose the problem into four tractable subproblems, i.e., 1) edge server selection, 2) transmission power allocation, 3) edge computing resource allocation, and 4) local computation resource allocation. In stage I, we optimize the edge server selection via many-to-one matching. In stage II, we determine the transmission power of each UAV with quasi-convex optimization and solve the edge computing and local computing resource allocation subproblems via a projected gradient descent method and convex optimization, respectively. Simulation results validate that the proposed scheme can achieve superior system performance over benchmark schemes.
Read the full abstract

Computation and Language

Representational drift in a code model: behaviour settles, which neurons carry it does not

In neuroscience, representational drift is a change in which neurons carry a representation while behaviour stays the same; we measure its analogue in OpenCoder-1.5B during pretraining, for 169 Python constructs over two trillion tokens at a constant learning rate, with a noise unit on every change, at checkpoints as close as 1,000 training steps. For 16 of the 169, those whose grammar fixes a next token, the loss at those positions reaches its floor within the first 4% of training, while the model's reliance on the neurons carrying them grows to the end: removing them costs 0.06, 0.25 and 0.40 nats more than removing a random set, at 4 billion, 84 billion and 2 trillion tokens. Which neurons carry a construct never settles: for all 169 the carrier set is still changing at the last checkpoint, and an abrupt settling at any date from 84 billion to 1.5 trillion tokens, or a gradual one before 500 billion, would have been detected in 99% of injected cases. The set measured at 84 billion tokens, removed at the end, cannot be separated from a random set, so the function is carried by whichever neurons are current. Up close the change is a persisting pool and a changing sample: a neuron that leaves a construct's set returns at three times the rate of a matched neuron, and no set reassembles. The hypothesis we fixed in advance, that the carriers settle, is refuted; we discuss candidate causes and name the experiment that would separate them.
Read the full abstract

Machine Learning

Copies Are Not Confirmations: How News Duplication Fools LLM Judges and Continual Learners

A language model that keeps learning from the news sees the same claim many times, but many reports are copies of one source: wire stories, content shared within a media group, syndication and press releases. We ask whether LLM judges and continual learners treat such copies as independent confirmations. In 200,000 English articles from CC-News, a near-duplicate detector identifies copy links to another outlet for 18.9% of the articles, a detector-based lower bound. In a controlled experiment with claims about invented people, three identical reprints instead of one report raised the lies accepted by Llama 3.2 from 0.06 to 0.71 and by phi4 from 0.21 to 0.90; Gemini 2.5 Flash already believed a single report, and Gemini 2.5 Pro followed provenance information most consistently. In a simulated continual-learning stream, showing the judge one report per detected copy chain cut the copied lies accepted by 0.10-0.24 across four judges, at a mean cost of at most 4.2 points of accepted truths. Text similarity links none of the copies reworded by three LLMs. An entity-novelty test, the share of a report's named entities absent from an earlier look-alike, separates these rewrites from 100 author-labelled independent reports of the same event (AUC 0.98-0.99), statistically indistinguishable from TF-IDF and the best tested embedding model, but recognises only 17% of natural same-source derivatives. With both detectors run on 15,084 real articles carrying inserted claims, one report per chain cut the lies accepted by phi4 from 0.57-0.67 to 0.09 (copies), 0.14 (rewrites) and 0.34 (rewrites with invented details), and by Llama 3.2 from 0.56-0.62 to 0.07, 0.09 and 0.25, at a cost of 2.4 and 10 points of accepted truths. Counting reports or chains without a judge is unreliable, and colluding sources that do not copy each other remain an open problem.
Read the full abstract

Materials Science

Pressure-induced symmetry breaking and polarity in the van der Waals magnet CrSBr

The van der Waals magnet CrSBr has emerged as a fascinating platform for exploring coupled structural, magnetic, and electronic phenomena. Yet, the evolution of its crystal symmetry and electronic states under pressure and temperature remains unresolved. To elucidate this we employ low-temperature single-crystal X-ray diffraction, Raman scattering and photoluminescence measurements under hydrostatic pressures of up to 12 GPa, complemented by first-principles calculations.We observe a cascade of pressure-induced structural transitions from the orthorhombic $Pmmn$ to monoclinic $P2_1/m$ phase at $\approx$ 1.5 GPa, followed by the emergence of a polar orthorhombic $Pmm$2 phase $\approx$ 6.5 GPa. The change in symmetry is accompanied by phonon anomalies and a direct-to-indirect bandgap crossover. Our results demonstrate pressure as a knob to engineer structure, symmetry and electronic properties in CrSBr in a coupled manner.
Read the full abstract

All authors Amit Pawbake, Sitaram Ramakrishnan, Kseniia Mosina, Aljoscha Soll, Zdenek Sofer, Pierre Rodiere, Rolf Heid, Gaston Garbarino, Pierre Bouvier, Clement Faugeras

Numerical Analysis

Boundary integral methods and shape reconstruction for periodic fluid–solid interfaces

This paper studies direct scattering by periodic fluid–solid interfaces and interface reconstruction from multifrequency near-field data. Using Helmholtz decomposition and quasi-periodic single-layer potentials, we formulate a coupled boundary integral system with the standard stress traction, establish its equivalence to the original fluid–solid transmission problem under suitable assumptions, and construct a corresponding Nyström discretization. For near-field observations at finitely many measurement points, we establish local Fréchet differentiability of the observation map with a quadratic remainder estimate. In the smooth setting, the derivative transmission problem and its boundary integral formulation are used to compute the observation Jacobian required for interface reconstruction. The interface is reconstructed by regularized Gauss–Newton iteration in a Fourier space that is progressively enriched as the frequency increases. Numerical experiments assess direct-solver accuracy and reconstruction performance for different interfaces, material configurations, and noise levels.
Read the full abstract

All authors Yongxin Liu, Junliang Lv, Ning Wang, Yi Wang

Information Retrieval

Structure and Lineage Evidence for Vietnamese Statutory Article Retrieval on DRiLL

We study Vietnamese statutory Article retrieval on DRiLL using features for provision identity and local form, same-law support, and citation and lineage evidence. Across five random seeds, selectors sharing candidates, listwise Qwen evidence, and learner settings improve official $F_2$ from 71.77% to 72.91%, with the largest descriptive gain on two-Article questions. Added features account for 18.20% of the proposed selector's total decision-tree split gain, measured as training-objective reduction, mainly from same-law and citation evidence. Cases reveal base provisions and amendments, companion recovery, distractor removal, and failures. These findings motivate research on other datasets, representations, and learners without establishing individual feature effects or requiring LightGBM, hand-crafted scores, or explicit graphs.
Read the full abstract

Sound

SAS: Scene-Annotated AudioSet Strong for Joint Scene-Event Analysis

Context is central to human auditory perception, as listeners interpret sounds within their acoustic environment to resolve ambiguity. However, large-scale datasets that jointly capture sound events and acoustic scenes remain limited. We address this by augmenting AudioSet Strong with clip-level scene annotations under two taxonomies for joint sceneevent modeling. A corpus-scale analysis shows that scene identity is better captured by events distinctive to a scene than by those that merely occur frequently. Reweighting event co-occurrences toward scene-specific associations produces representations that better align with human-derived taxonomies, while also exposing scenes where the acoustic and semantic views diverge. This geometric divergence predicts classification errors, while supervised validation confirms that the proposed scene annotations provide meaningful complementary information for developing scalable context-aware machine listening systems.
Read the full abstract

All authors Michel Olvera, Paraskevas Stamatiadis, Romain Serizel, Slim Essid, Gaël Richard

Earth and Planetary Astrophysics

New optimisations for exact, scalable 2D Gaussian process inference, with applications to astronomical time-series

Astronomical datasets often consist of a grid of datapoints in two or more dimensions, such as images, spectroscopic time-series, and imaging time-series. Gaussian processes (GPs) are a powerful tool for dealing with correlated noise in these data, arising from both astrophysical and instrumental systematics. Despite this, popular GP optimisations such as celerite are largely restricted to 1D datasets. Extensions to 2D data typically come with limited flexibility or poor scaling. As a motivating example, this limitation is significant for transiting exoplanet spectroscopy with JWST, as current analyses typically assume each spectroscopic light curve is statistically independent, despite many potential sources of wavelength-correlation, including detector readout noise and stellar granulation. This introduces a potential bias in atmospheric retrievals which is typically ignored. To solve this problem, we present new, exact GP optimisations implemented in the JAX-based package luas, which extend 1D GP optimisations to multidimensional data using Kronecker product algebra. These GP optimisations are highly flexible, permit efficient analysis of 2D datasets containing millions of datapoints, and in many cases have a cheaper cost than transit light curve modelling. These optimisations are key to pushing the limits of exoplanet spectroscopy to smaller targets around bright host stars, as we increasingly hit instrumental and astrophysical noise floors. We suggest further applications to the analysis of ground-based high resolution spectroscopic time-series, extreme precision radial velocities, and high-contrast imaging, drawing on the connection that certain 2D GP models represent a probabilistic generalisation of PCA. Applications may even extend to machine learning and geostatistics, as we present novel optimisations for the Linear Mixing Model and the Linear Model of Coregionalisation.
Read the full abstract

Artificial Intelligence

Safe Reinforcement Learning via Probabilistic Option Shielding

Shielding is a safe reinforcement learning technique that provides safety guarantees while learning, by blocking unsafe actions. Traditional shielding relies on a complete model of the environment to provide long-term guarantees. Probabilistic logic shielding (PLS) lifts the requirement of having such model, but only considers safety in the short term, typically looking only one or few steps ahead. We propose Probabilistic Option Shielding (POS), which extends PLS to the option level, recovering long-horizon safety without reintroducing the need for a complete environment model. POS learns a model-free critic directly from one-step PLS signals, to estimate the probability that an entire option will complete safely under uncertainty. We evaluate POS on safe navigation tasks in the LogiCity environment, and show that it achieves considerably safer training while matching the task performance of standard step-level PLS and other safety related reinforcement learning baselines.
Read the full abstract

All authors Gian Luca Ascari, Isidoro Tamassia, Ying Jiao, Luc De Raedt, Giuseppe Marra

Computation and Language

Calibrated concept profiles: making the reading of a model's neurons a measurement

Interpretability has largely moved from reading a model's neurons directly to reading a second model trained on the first: a probe, a sparse autoencoder, a crosscoder. The move answered real faults – mixed meanings, thresholds without a unit, readings that did not transfer from one study to the next – but gave up what a direct reading alone provides: nothing trained in between, the neurons themselves named, and a reading that can be repeated and compared. Here the direct reading is made a measurement. A neuron's activation varies with the prompt, so its reading is a distribution over a population of stimuli. We declare that population: every factor around a target construct is drawn at its rate in a reference corpus and logged. From the declared distribution we generate a battery of prompts, and draw it twice. From the battery we compute the calibrated concept profile, one number per neuron, stated in units of the neuron's own spread on ordinary input; the two draws give every number a noise unit. Before the profile is read, a fixed set of checks licenses the battery on the model, and the smallest change the profile can resolve across training checkpoints is measured. Realised for Python on OpenCoder-1.5B and followed across checkpoints spanning two trillion training tokens, the profile gives what the alternatives do not report: a unit on every number, a chance level for every comparison, and which neurons carry a construct at every checkpoint. The battery, the code and the data behind every result are released.
Read the full abstract

Computer Vision

InstantHMR : Real-Time Human Mesh Recovery on Your Phone

Real-time 3D human mesh recovery from a single image on mobile devices is a challenging problem with many applications, including robotics, biomechanics, AR/VR, and exergames. Current state-of-the-art models are either too slow to run efficiently on mobile devices or sacrifice accuracy for speed. We present InstantHMR, a family of open-source single-image regressors suited for mobile inference, based on the Momentum Human Rig (MHR) parametric model and trained exclusively on publicly available annotations from the SAM 3D Body dataset. Our approach achieves remarkable performance on a Samsung Galaxy S23 NPU, recovering a 6,890-vertex body mesh in 5.5 ms per person, and 0.98 ms on an NVIDIA RTX 4070 GPU. This speed surpasses all existing methods we evaluated. On mobile, it is 12x faster than the single mobile-compatible method that slightly outperforms ours in accuracy, while achieving competitive accuracy with 40.0 mm PA-MPJPE on 3DPW, 48.8 mm on EMDB and 41.7 mm on RICH. To ensure fair comparisons, we re-evaluate 11 state-of-the-art methods under a unified protocol aligned with the closest MHR-based competitors, including Meta's official MHR-to-SMPL conversion, and report their reproduction results. Additionally, we measure latency across eight different targets, ranging from desktop TensorRT to two phone NPUs. Code and models are available at https://github.com/Naturalpad-studio/InstantHMR.
Read the full abstract

Artificial Intelligence

Large Language Model Consensus and Corpus-Adaptive Topic Modeling for Classifying and Mapping Patent Innovation

Patent analytics increasingly relies on transformer representations and large language models (LLMs), yet linking conceptual classification, topic formation and interpretation requires explicit decisions about model roles and the evidence retained at each stage. In this design-oriented study, we propose and demonstrate a patent analytics pipeline that links these tasks and keeps each decision open to audit. Three LLMs separately classify each patent by novelty (radical or incremental) and innovation type (product, process, service). Radical and innovation-type labels each require two supporting votes, with Incremental used as the fallback novelty label. Topics are discovered by clustering PatentSBERTa embeddings, with parameter ranges scaled to corpus size. A patent-specific stopword list is applied only when topics are represented with class-based term frequency–inverse document frequency (c-TF-IDF), and a second embedding model supports inter-topic similarity, hierarchical refinement and visualization. An LLM then labels each topic from its ranked keywords and representative patents. We demonstrate the pipeline on 30,964 public-sector-related patents retrieved from Lens.org. Across five category corpora ranging from 591 to 29,215 patents, the pipeline produced 74 category-specific topic instances with a macro-average document coverage of 98.1% and topic diversity of 0.900–1.000, and fewer than 1% of classification outputs returned UNKNOWN. Because every model vote and the evidence behind each topic label are retained, analysts can audit and revise the outputs. The demonstration establishes the feasibility and traceability of the pipeline; testing its labels against expert coding and benchmarking it against alternative pipelines are identified as the next steps. The pipeline offers a reusable basis for LLM-assisted patent analysis in technology intelligence.
Read the full abstract

High Energy Physics (Phenomenology)

Restoration of rotational symmetry in light-front dynamics

Solving the bound state problem in the framework of the light-front dynamics (with the state vector defined in the light-front plane t+z=0), one finds that after truncation - representing approximately the state vector by a finite number of the Fock components, the binding energies become to depend on projection of the angular momentum on the z-axes. This is so called violation of rotational symmetry in light-front dynamics. It is shown that the physical solution is the superposition of these states with some well defined coefficients. For a test, we also solve the Bethe-Salpeter equation and projecting its solution on the light-front, we find the wave functions which do not suffer from violation of the rotation invariance. Comparison with them of solution in the form of superposition shows that not only binding energies but also the wave functions themselves are described by the superpositions of the split states fairly well.
Read the full abstract

Number Theory

Elliptic units, derived Heegner classes and the p-adic Birch and Swinnerton-Dyer quotient of CM elliptic curves of rank two

Let $E\colon y^2=x^3-Dx$ have rank two and let $p\equiv1\pmod4$ be a prime of good reduction. We study the conjecture that the $p$-adic quotient $q_p$, the leading coefficient of the cyclotomic $p$-adic $L$-function divided by the Euler factor and the $p$-adic regulator, equals the complex quotient $q_\infty=L''(E,1)/(2\Omega_E\mathrm{Reg}_\infty)$, and that the latter is rational. In rank one the analogous statement is a theorem of Perrin-Riou and Rubin, because a Heegner point computes both sides; in rank two there is no such point. We show that $q_p$ is the constant of the quadratic term of the two-variable Katz measure; this is how the conjecture arises, and it computes $q_p$ modulo any power of $p$ from critical Hecke $L$-values. We then show that $q_p$ is the coordinate of two global classes: Rubin's rank-two elliptic-unit class, whose position in $E(\mathbb{Q})\otimes\mathbb{Q}_p$ is $q_p$ times an explicit vector, and the derived Heegner class over an auxiliary imaginary quadratic field $F$, for which we prove an exact identity. As consequences, whenever such a field $F$ satisfying our hypotheses exists, $q_p$ is a square in $\mathbb{Q}_p$ times the product of the Tamagawa numbers of $E$, so that the unit $q_p/A_E$ of the $p$-adic Birch and Swinnerton-Dyer formula is a square when Sha is finite; if moreover the derived Heegner class has non-zero formal logarithm, the conjecture of Agboola and Castella for $E/F$ holds up to a $p$-adic unit, and with a theorem of Burungale, Castella and Skinner the anticyclotomic main conjecture holds integrally; and our conjecture is equivalent to the analytic form of theirs, with the products of anticyclotomic heights read sesquilinearly, which differs by a sign from its literal formulation. The conjecture is verified modulo $p^3$ for 5520 pairs $(E,p)$ with conductors up to $3.9\cdot10^{12}$, including curves with analytic Sha of order 4 and 9.
Read the full abstract

Computer Vision

Bootstrap Your Own AV-Proxies: Adaptive Contrastive and Prototype Learning for Audio-Visual Segmentation

Audio-Visual Segmentation (AVS) aims to accurately segment sounding objects in video frames by leveraging audio-visual correspondence cues. However, it remains challenging due to the intrinsic semantic incompleteness within a single modality and the semantic gap between audio and visual representations. Traditional feature-fusion decoding approaches struggle to suppress fusion noise effectively, while recent methods that incorporate data-dependent priors often increase the complexity of modeling audio-visual correlations, leading to poor cross-domain generalization. To address these issues, we propose a novel adaptive contrastive and prototype learning framework, BYOAVP, for AVS. Specifically, we design a Self-Supervised Audio Enhancement (SSAE) module that introduces contrastive learning to adaptively align audio representations with gradient-blocked visual embeddings, thus narrowing the semantic gap between modalities. Furthermore, a Dynamic Prototype Constraint (DPC) module is developed to refine pixel-wise category perception via momentum-based prototype updating, while enhancing localization of sounding regions through cross-modal interaction. Extensive experiments demonstrate that our method achieves SOTA performance across two AVS benchmarks and six sub-tasks, exhibiting strong robustness and generalization ability.
Read the full abstract

All authors Junbo Zhang, Hang Su, Zhaofan Li, Hang Dong, Chao Sun

Functional Analysis

Entropy and frame measures for self-similar measures

For $0<\rho<1$ and a set of $N\ge2$ distinct real digits $\mathscr D=\{d_0,\ldots,d_{N-1}\}\subset\mathbb{R}$, let \[ \mu_{\rho,\mathscr D}=\frac1N\sum_{i=0}^{N-1}\mu_{\rho,\mathscr D}\circ S_i^{-1}, \qquad S_i(x)=\rho(x+d_i). \] Write $X=\sum_{j\ge1}\rho^j d_{\omega_j}$, where the symbols $\omega_j$ are independent and uniform on $\{0,\ldots,N-1\}$. We prove that $\mu_{\rho,\mathscr D}$ admits no frame measure if the conditional entropy $H(\omega_1\mid X)>0$. In particular, this holds for all $\rho>1/N$, without any absolute-continuity assumption or arithmetic restriction on the digits or the contraction. At $\rho=1/N$, for commensurable digits, a frame measure exists if and only if $\mu_{\rho,\mathscr D}$ is normalized Lebesgue measure on its attractor. For $\mathscr D=\{0,\ldots,N-1\}$ and $\rho<1/N$, we also prove nonexistence whenever $\rho^{-m}=a/b$ for some $m\in\N$, with $a/b$ in lowest terms, $b\mid N$ and $N/b\nmid a$, extending the integer-power obstruction. The extension compares mask quotients using finite families of target pole classes in place of one target class.
Read the full abstract

Machine Learning

RoleWalk: training-free structural embeddings for directed graphs that transfer across graphs

Structural role embeddings describe a node by its position in a graph - hub, bridge, periphery, member of a dense core - rather than by who its neighbours are. Existing methods either learn coordinates specific to one graph (struc2vec, Role2Vec, RESD) or read scales and normalisations from the whole graph (GraphWave, Digraphwave), so the same neighbourhood can receive different coordinates in two graphs; few read edge directions. We present RoleWalk, a training-free embedding that follows a short lazy random walk from each node and records, at each step, two distributions: that of the transition probabilities, through its characteristic function at a fixed grid of frequencies, and that of the walker's distance to its starting node. On directed graphs, walks that go forward then backward add who shares a node's targets or sources. The embedding depends only on a node's neighbourhood within a fixed radius, so its coordinates mean the same thing in every graph. We show that, without curvature, it is the kernel mean embedding of the walk's footprint, that it separates graphs that the Weisfeiler-Leman test cannot, and that pruning small probabilities comes with a per-node error certificate. Empirically, RoleWalk is competitive with the best role embeddings on five classification benchmarks at a fraction of their cost, has the highest scores on real directed roles in Wikipedia adminship, exceeds the published accuracy of Digraphwave at every noise level of its own Enron network-alignment benchmark, transfers centrality predictors to graphs four times larger better than existing inductive embeddings, hand-made features and a trained GIN, and embeds a million nodes in 20 s to 3 min on one CPU core. The code is open source (BSD license): https://github.com/kerighan/rolewalk
Read the full abstract

Systems and Control

Controlled Excitation for Identification of Nonlinear Systems with Saturated Outputs

We consider nonlinear systems whose measured output is the product of an unknown matrix and a known saturation function of the unmeasured state. Generating excitation that allows the matrix to be recovered from the output is challenging when the dynamics are largely unknown. We exploit saturation to construct this excitation. Under a geometric accessibility condition, we derive explicit open-loop controls that drive the state into regions where the saturation takes prescribed values. For componentwise saturations, this condition holds generically with as few as two independent control directions. The controls require no state measurement and only partial knowledge of the dynamics, with no information on the structure of its nonlinear part. The resulting regressor is known in advance, and the unknown matrix is recovered exactly by a closed-form regression. Under stronger structural assumptions and with sufficiently many independent outputs, an output feedback tracks prescribed reference trajectories within the unsaturated region to arbitrary accuracy and permits exact state recovery. Suitably chosen references then also enable exact identification of linearly parameterized (nonlinear) dynamics. We analyze robustness to measurement noise and illustrate the method numerically in the presence of both measurement and dynamical noise.
Read the full abstract

All authors Adel Malik Annabi, Jean-Baptiste Pomet, Dario Prandi, Ludovic Sacchelli

Optimization and Control

Real-Time Nonlinear Mixed-Integer Model Predictive Control With Horizon-Tail Relaxations

Model predictive control (MPC) of nonlinear systems with discrete decisions leads to mixed-integer nonlinear programs (MINLPs) whose combinatorial complexity often prevents real-time deployment. In this paper, we propose a decomposition-based MPC method that reorders and reduces computations to shorten the delay between receiving a state measurement and applying the control input. Before the measurement arrives, we solve a continuous relaxation of the MINLP at a predicted state and construct a mixed-integer quadratic program (MIQP). Once the measured state is available, we solve the MIQP that reconstructs the discrete controls and simultaneously performs one sequential quadratic programming step toward the corresponding nonlinear solution. To reduce the combinatorial complexity in the horizon length, we either fix the integer controls or relax their integrality over the prediction horizon tail, while retaining an integer-feasible first control input. We derive error bounds relating the nonlinear objective values obtained with the horizon-tail approximations to the optimal value of the original mixed-integer problem. In a data center cooling example, the proposed method reduces mean feedback computation time by 94% relative to a state-of-the-art MINLP method with a total closed-loop cost difference of less than 1%.
Read the full abstract

Functional Analysis

Recurrent and rigid multipliers on commutative Banach algebras. The Fourier algebra

We study recurrence, rigidity and uniform rigidity of multipliers on semisimple commutative Banach algebras. For a power-bounded multiplier $T$ on a Tauberian, self-adjoint algebra $\mathfrak{A}$, we show that $T$ satisfies each of these properties if and only if the multiplication operator on $C_0(\Delta(\mathfrak{A}))$ induced by the Gelfand transform of $T$ does. One consequence of our analysis is that, if $\Delta(\mathfrak{A})$ is $\sigma$-compact, then recurrence implies rigidity for such multipliers. We examine in detail the multipliers of the Fourier algebra $A(G)$. In particular, we show that recurrence of a multiplier of $A(G)$ is equivalent to recurrence of the corresponding multiplier on $C_0(G)$ when the connected components of $G$ are open. For multipliers that are not power bounded, a flexible construction based on chains of compact subgroups of a locally compact abelian group $G$ produces multipliers of $A(G)$ that are rigid but not uniformly rigid, and multipliers that are recurrent but not rigid, such that in each case the corresponding multiplier on $C_0(G)$ is uniformly rigid. Via the Fourier-Stieltjes transform, these results describe the recurrence and rigidity properties of convolution operators on $L^1(G)$.
Read the full abstract