Peer Revenue Now accepting positions

Papers

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

All of arXiv

Monday 9 November

Computation and Language

Adaptive Entropy-Sparing for Efficient Reasoning

Large Reasoning Models (LRMs) produce explicit chains of thought to improve transparency, but they frequently overthink, generating long yet inefficient reasoning chains that inflate computational costs and risk introducing errors. While numerous reinforcement learning methods struggle to penalize verbosity to achieve conciseness, they typically apply the penalty uniformly across lengthy sequences, which leads to credit confounding, that is, the failure to distinguish essential reasoning steps from superfluous ones. In this paper, we study a measurable proxy for this confounding: high-entropy tokens that often coincide with surface transition points where reasoning re-enters verification. Based on this insight, we propose Adaptive Entropy-Sparing (AES), which selectively spares high-entropy tokens that occur before an efficient in-group reference, while penalizing overlength high-entropy suffix tokens associated with superfluous branching. Our AES thereby navigates the suppression-exploration trade-off, reducing overthinking while preserving average reasoning accuracy. We also present an entropy-length model that motivates why local uncertainty can serve as a proxy for future reasoning cost. Across eight reasoning benchmarks, AES reduces reasoning length by up to 52.9% while increasing mean accuracy by up to 3.8% over the base models, consistently outperforming existing efficiency-oriented methods. Our code is available at https://github.com/MobiusDai/AES.
Read the full abstract

All authors Cong Jiang, Xiaofeng Zhang, Tom Ko, Zheng Zhang

Representation Theory

Invariant operations on vector fields on the line are Lie polynomials

We prove that all $W_1$-equivariant multilinear operations on the Witt Lie algebra $W_1$ are Lie polynomials. We study this question in terms of Wronskian Lie brackets and Novikov products, which are abstractions of the Lie bracket $[a,b]:=ab'-ba'$ and the nonassociative product $a\circ b=ab'$ one can define on a commutative differential algebra. In terms of these notions, our main result asserts that the operad of Wronskian Lie algebras is isomorphic to the operad of equivariant multilinear operations on~$W_1$; we also prove a parallel result for the operad of Novikov algebras, characterizing it as the operad of affine equivariant multilinear operations on $W_1$. Two immediate consequences of our work are a characterization of Lie polynomials in free Novikov algebras and a computation of the degree zero homology of the operadic twisting of the Novikov operad. Finally, while it is known that over a field of zero characteristic, a Novikov algebra embeds into its differential envelope, we show that the same is not necessarily true for Wronskian Lie algebras.
Read the full abstract

Data Structures and Algorithms

How Far Must a Test Explore? Certifying Composable Read-Write Interfaces from Their Own Random Edits

Random tests of stateful code check laws, like reading back a written value, at states that random writes reach. The laws hold at every state exactly when the code behaves like a record with hidden data, so a pass is taken as evidence. This can fail: the laws can hold at every sample while the code is far from any record. The laws are local, but behaving like a record is global. Self-testing and stability theorems bridge this gap when the query distribution is controlled by the tester or preserved by the tested maps, but here it is generated by the tested writes, which need not be invertible. We ask how far a test must explore. An interface has $m$ binary variables, each with a read and two writes; in a record, reads return their bits and writes set only their bits. Samples are the ends of geometric runs of mean $\ell$ of uniformly random writes from a start distribution; a tester adds its own write chains of depth $d$. The distance is the least, over records with hidden data, of the largest probability that a read or write differs from it at a sample. From one start state, law checks at $O(m^2\log m/\epsilon)$ samples certify, at fixed confidence, any threshold $\epsilon$ after geometric runs of every positive mean, while runs of fixed positive length can pass at distance $1/2$. For the worst start distribution, take $\epsilon=m^{-\beta}$ with fixed $0<\beta\le1$. For $d\le cm$ with fixed $c<1/2$, a polynomial budget of samples and writes exists exactly when $\ell=\Omega(m)$, where law checks suffice. With no runs, the least depth for this budget is $(1/2+o(1))m$. Counting samples alone, in the worst case at $\epsilon=1/(32m)$ and $d\le m/16$, the critical mean for a polynomial budget is $\Theta(m\log(m/d)/\log m)$, by exhaustive search. So one start state needs only law checks, and the worst case at these thresholds needs exploration linear in $m$ if writes are counted.
Read the full abstract

All authors Jiaxuan Guo, Xingyu Ren, Jingxin Yang, Youran Sun, Haizhao Yang

Artificial Intelligence

Disagreement-Aware Distribution Matching Distillation for Few-Step Diffusion Models

Distribution Matching Distillation~(DMD) enables strong few-step diffusion generation by matching the student distribution to a pretrained teacher. Existing improvements to DMD often augment its supervision or modify its optimization machinery. We ask a complementary question: can DMD diagnose its own updates using signals already produced during training? DMD inherently computes the discrepancy between a pretrained teacher and a fake model tracking the student distribution, but conventionally uses it primarily to determine the distribution matching direction. We find that the magnitude of this teacher–fake disagreement provides additional diagnostic information. Across extensive experiments, large disagreement in the high-noise regime consistently coincides with substantially higher-energy DMD updates. This reveals a dual role of disagreement: it determines not only where the student should move, but also provides an intrinsic signal of how aggressively the update should be applied. Based on this observation, we introduce Disagreement-Aware Distribution Matching Distillation (DA-DMD), which activates only in the high-noise regime and continuously modulates the update strength according to teacher–fake disagreement. DA-DMD reuses quantities already available during DMD training and requires no additional model, evaluator, or inference-time modification. We show that it exactly recovers vanilla DMD outside the selected region, reduces update energy only through selected samples, and exhibits redescending influence under extreme disagreement. Across multiple diffusion backbones, DA-DMD consistently improves few-step generation at the same inference cost. On SD3.5-L, DA-DMD improves HPSv2 from $28.23$ to $30.93$, while achieving a $64.4\%$ average win rate over DMD2.
Read the full abstract

All authors Zikai Zhou, Zhengpeng Xie, Shitong Shao, Lichen Bai, Tianhao Zhang, Bojun Cheng, Zeke Xie

Computation and Language

ShopMind: A Mental World Model for Personalized and Realistic E-commerce User Simulation

Large language models are widely used in e-commerce user simulation, providing a foundation for understanding users and predicting human decisions. However, existing simulators still struggle to achieve both personalization and behavioral realism. Most methods predict the next action directly from interaction histories and the current context, without accounting for the evolving mental world that underlies user behavior. Theory of Mind suggests that behavior is driven by mental states, including beliefs, uncertainties, and intentions. Users may interpret the same information in distinct ways, leading to divergent behaviors, while an individual's mental state evolves through interaction and shapes subsequent actions. Ignoring this process prevents simulators from faithfully capturing individual differences and realistic behavior patterns, and undermines its effectiveness in downstream applications. To address this, we propose ShopMind, a Mental World Model for e-commerce user simulation that captures how consumers interpret shopping information and how their mental states evolve, better reflecting individual users and real-world behavior patterns. We further use behavior-mental state pairs synthesized by ShopMind to train a Mental State Reconstruction (MSR) model, which infers consumers' latent mental states from observable behavior. Experiments show that ShopMind improves personalization by 18.0% and behavioral realism by 19.6% over the strongest alternatives. Moreover, MSR improves performance on consumption value measurement and CTR prediction, showing the utility of ShopMind for understanding and predicting real user behavior.
Read the full abstract

All authors Peixuan Hou, Bin Chen, Li He, Jian Xu, Xiuli Ma, Guojie Song

Artificial Intelligence

LEGION: Benchmarking Language Agents on Sparse-Signal Investigation Under Adversarial Noise

Autonomous language-model agents are increasingly deployed on discovery-driven tasks where signal is sparse and most retrieved information is irrelevant. Current benchmarks fail to evaluate this regime due to dense signals, cooperative environments, and ground truth typically being only a few tool calls away. We introduce Legion, a benchmark framework coupling a stochastic enterprise-telemetry simulator (Legion-Sim) with a tiered library of 74 expertly-crafted adversarial scenarios (Legion-Bench) and an evaluation protocol that scores investigation efficiency alongside detection accuracy. Scenarios are injected into freshly simulated companies, allowing independent control over confounder density and event volume. We test five agent architectures on two simulated companies and find that while recall is high (0.82-0.88) precision lags behind (0.61–0.72). They fail to find all of the indicators of compromise of the attack or mistake genuine indicators with attack related ones, which is critical in the context of cybersecurity investigation. Failures are consistent across reasoning tiers and persist across confounder density and event volume. We release Legion, the scenario library, performance on common agent architectures, and pipeline for reproducibility and extension.
Read the full abstract

All authors Samuel Joseph Amouyal, Ofek Glick, Yaniv Menasherov, Eyal Fisher, Ohav Barbi

General Relativity and Quantum Cosmology

Quasinormal Modes of Black holes in the generalized Schwarzschild-de Sitter solution of $f(R)$ gravity

We analyze scalar, vector, and {axial} tensor perturbations in a Schwarzschild-type static spherically symmetric solution of modified $f(R)$ gravity. This solution differs from the Schwarzschild metric of general relativity through the presence of a linearly growing term and a logarithmic correction, and it depends on four integration constants. Using the higher-order Wentzel-Kramers-Brillouin (WKB) approximation and an analytical method, we calculate the quasinormal modes (QNMs) for all three types of perturbations. Our results show that, for appropriately chosen parameter values, the QNMs closely match those of the Schwarzschild black hole. In particular, the model parameter $A$ produces a nonlinear increase in both the oscillation frequency and damping rate of ringdown gravitational waves. A similar effect arises for the parameter $\eta$ when $A$ is small, while for larger $A$ values the effects are opposite, where the scalar perturbations display a distinct deviation compared to the vector and {axial} tensor cases. Contour plots of QNMs in the parameter space support these findings. Finally, time-domain profiles are obtained to illustrate the dynamical evolution of the gravitational wave signal.
Read the full abstract

Robotics

Distilling Instructions into Weights: Closed-Form Edits Fix a Robot Policy's Misreading

Generalist robot policies often fail at one step of a task they can otherwise perform. We ask whether such a failure can be repaired by editing the policy's weights directly, without gradients, training data, or a change to the instruction. Told to put the cube then the banana in the bowl, a zero-shot $\pi_{0.5}$ vision-language-action policy in RoboLab releases the cube before it reaches the bowl. Replaying 21 of these decisions under rewordings shows that it binds "in the bowl" to the wrong object. Closed-form instruction distillation copies the per-layer effect of an explicit rewording into the language model's MLP weights. On 128 held-out seeds, success rises from 27 to 106 episodes, with 81 fixed and 2 broken ($p=7\times10^{-22}$), while an update of the same size in random directions leaves success unchanged. The edit also lifts two other tasks without detectable harm on the remaining 13, and a variant that stores 0.16% of the parameter count keeps most of its effect. Placed in the action expert or action head instead, the same distillation helps less or not at all. Our released tool reproduced the fix end to end on new held-out seeds, from 8/64 to 36/64. Edits aimed at perception, a missing grasp, contact physics, or the gripper threshold did not help, and of seven blind Claude Opus editors, only the one given the misread instruction produced an edit that passed held-out scoring. These results suggest that weight edits work best when a known-good input already exists to distil. Because reruns with unchanged weights flip about a third of paired outcomes, evaluations should pair seeds, rerun the unedited policy, and confirm gains on held-out seeds.
Read the full abstract

Computer Vision

Developing a Street View Imagery Processing Pipeline for Building-Level Vacant House Assessment

Street view imagery (SVI) is widely used to audit urban environments remotely. Automated assessments usually assume that each image shows the target building clearly. In practice, an image may lack task-relevant evidence or show a neighboring house. If this goes unchecked, a model's score cannot show whether it judged the correct building. We present a reproducible SVI processing pipeline that makes image suitability and target identity explicit and checkable. We apply it to vacant house assessment in South Korea, using 10,311 images of 1,609 detached-house parcels. Task-suitability filtering adapts its criteria to image composition and administrative area. This keeps coverage even across areas. While keeping as many images as a single uniform filter, our filtering left 101 more parcels with a usable image. Target-house alignment identifies the parcel's house from camera metadata and parcel boundaries. It needs no building footprints or additional training. Alignment matched human labels for 85.6% of images, compared with 79.8% for the best simple rule. In vacancy classification, filtering and alignment raised AUROC from 0.684 to 0.719, but marking a different house in the same image still gave 0.709. Grad-CAM showed that the model often looked at areas outside the target house. Image suitability, target identification, and model evidence are therefore separate problems, and scores alone do not resolve them. Recording which images were used and which house was selected lets reviewers verify each assessment and can support human virtual audits.
Read the full abstract

All authors Dongjoon Park, Sunjae Lee, Jung Woo Kim, Bumjoon Kang

Machine Learning

When Answer Decodability Is Not a Decision: Separating Final-Answer and Imminent-Emission Signals in Chain-of-Thought

Answer probes applied to chain-of-thought hidden states are often read as evidence that a language model has already decided its final response. This reading overlooks a competing signal, the candidate the model is about to mention. We introduce emission-controlled dual-target probing (ECDP), which fits the same low-rank reader family separately to the imminent mention and to the final answer at identical positions, with target-specific nulls and thresholds fixed before any fit. On correct DecisionDepth (DD) rollouts in which a distractor is mentioned before the final answer, Qwen3.5-4B decodes the next mention at .966 (null .234) but the final answer only at .221 (null .186), an emission gap of .696 [.648, .744]; Llama-3.1-8B gives .575 [.514, .632]. This pattern is not universal. In a pre-specified evaluation of Qwen3-8B with thinking enabled on held-out MMLU questions, using a full-prefix attention reader over every thinking position, both targets are decodable and the final answer is read far more strongly (mention-minus-answer lift -.398 [-.459, -.336] at layer 22 over 467 questions). Model, task, positions, and reader change together between these settings, and an attempted matched comparison on DD produced no valid statistics because the imminent-mention probe did not train reliably, so we do not identify what reverses the order. On DD, planted-signal controls show that the reader recovers an injected final-answer signal at one tenth of full dose. Answer-candidate decodability is therefore target-relative and protocol-dependent: a strong answer readout does not by itself establish when a model commits, and a probe that fails to train is not evidence that the information is absent. ECDP is a practical control for distinguishing imminent text emission from evidence of a settled decision.
Read the full abstract

Computation and Language

A Large-Scale Dataset for Cross-Lingual Multi-Jurisdiction Patent Drafting from Patent Cooperation Treaty Applications

Securing patent protection across countries requires drafting claims separately for each national patent office, in the relevant language and according to its legal and stylistic conventions. Patent claims can therefore differ across languages and jurisdictions because of translation, target-office drafting practices, and prosecution-stage changes. We formulate cross-lingual patent drafting as generating plausible candidate claim sets for a specified target language, jurisdiction, and prosecution stage from a source claim set in another language. The task models jurisdiction- and stage-specific transformations beyond content-preserving translation. We construct a large-scale Japanese-English dataset by linking Japan Patent Office (JPO) and United States Patent and Trademark Office (USPTO) records through 884,179 Patent Cooperation Treaty (PCT)-derived patent families. Across six released configurations, these families yield 4,067,168 directional claim-set instances; four core configurations form the main benchmark. Commercial machine translation and zero- and few-shot prompting capture part of the task, while supervised adaptation provides the strongest overall baseline, showing that task-specific patterns can be learned from linked cross-jurisdictional claim sets.
Read the full abstract

All authors Trung Phuong Le, Hirofumi Nonaka, Koji Marusaki, Katsuhito Sudoh, Seiya Kawano

Computer Vision

Training Vision Transformers to Focus: Emergence of Spatial Attention Head Specialization

Self-supervised learning in Vision Transformer models, such as DINO, enables robust visual perception without requiring labeled data by leveraging large-scale image datasets. However, these models lack a focusing mechanism during training, resulting in perceptual representations that are often ambiguous or overly similar. In this work, we introduce two visual focusing loss functions designed to establish correspondence between the model's attention and specific regions within images. Specifically, we leverage the multi-head attention mechanism within the model to selectively steer the model's focus, leading to the emergence of spatial attention head specialization, characterized by diverse and spatially focused perceptual capabilities without requiring annotations. Through both qualitative and quantitative evaluations, we demonstrate that our method substantially increases the spatial selectivity and diversity of attention heads, improving both interpretability and performance on dense prediction tasks such as zero-shot image segmentation.
Read the full abstract

High Energy Physics (Theory)

Critical Slowing Down and the Crossover to Quantum Scrambling in Holographic CFTs

We investigate critical relaxation dynamics and their connection to quantum information scrambling in the conformal field theory dual to a Reissner-Nordström anti-de Sitter (RN-AdS) black hole. Employing the holographic dictionary, we formulate the thermodynamics in terms of boundary CFT variables and study the stochastic evolution of the reduced horizon radius using Langevin and Fokker-Planck formalisms. We demonstrate pronounced critical slowing down in the dual CFT, characterized by an enhancement of the autocorrelation time and stochastic fluctuations, accompanied by a suppression of the lowest nonzero Fokker-Planck eigenvalue. The autocorrelation time exhibits a universal power-law divergence, $\tau\sim|\epsilon|^{-\Delta}$, with $\Delta\simeq 2/3$, consistent with the analytical prediction and previous bulk gravitational results. Going beyond the conventional single-parameter analysis, we incorporate variations of the CFT central charge and identify a combined scaling variable involving temperature and central charge, demonstrating that the leading relaxation dynamics retain the same critical exponent along different trajectories in the two-dimensional parameter space. Finally, we explore the interplay between thermodynamic relaxation and quantum information scrambling by comparing the relaxation timescale with the fast-scrambling timescale. The comparison reveals a crossover between thermodynamic relaxation and quantum scrambling for which we derive the corresponding crossover scale analytically and find that, in the large-central-charge limit, it behaves as $\hat{\epsilon}\sim(C\ln C)^{-3/2}$. We also verify this numerically using the full nonlinear free-energy landscape. Our results connect critical dynamical scaling, finite-central-charge effects, and quantum scrambling within the holographic CFT framework.
Read the full abstract

Databases

OmniSynth: A Meta-Evaluation Framework for Benchmarking the Quality of Missingness-Inclusive Synthetic Health Data

Synthetic data are an emerging solution to the reproducibility crisis in machine learning for health, driven by data privacy restrictions surrounding electronic health records. Yet no consensus exists on how to evaluate synthetic health data quality, existing metrics functionally overlook missingness despite its prevalence in clinical data, and benchmarking frameworks have never been empirically validated. We introduce OmniSynth, a modular Python benchmarking suite that evaluates synthetic tabular data across fidelity, missingness similarity, utility, and privacy - the first framework to explicitly evaluate missingness as a distinct dimension of synthetic data fidelity. OmniSynth further proposes a meta-evaluation procedure that stress-tests evaluation performance under controlled degradation scenarios, providing a systematic characterization of how an evaluation framework behaves across datasets, sample sizes, and variable types. Applying this procedure to OmniSynth in two health datasets (Pima Indians Diabetes; MIMIC-IV-ED), we found that scores responded monotonically to degradation, three of four novel missingness metrics captured induced missing data patterns, and individual metric contributions to the scoring system were empirically justified.
Read the full abstract

All authors Annabel Zhu, Océane Doremus, Ariel Guerra-Adames, Marta Avalos-Fernandez, Emmanuel Lagarde, Cédric Gil-Jardiné, Vianney Jouhet

Machine Learning

How Many Candidates Does a One-Step World Model Need? A Random-Codebook Account of Nearest-of-m Training on Frozen DINO Tokens

One-step stochastic world models avoid the mode averaging of deterministic feature predictors without the tens of network evaluations per step that diffusion and flow matching require. The simplest such objective is nearest-of-m matching (conditional implicit maximum likelihood estimation, IMLE): draw m candidate futures and pull only the one closest to the observed outcome. Recent work conjectures that a small m suffices for feature-space world models and that a strictly proper energy score should replace the hard minimum. We test both claims with one-step world models on frozen DINOv2/v3 patch tokens, using benchmarks whose futures can be enumerated. Under mode-like uncertainty, nearest-of-m with m=16 covers the enumerated futures better than 50-step flow matching (at about 1/30 of its latency), one-step flow generators and an energy-score generator at matched parameter counts, and its rollouts stay on the token manifold where flow rollouts drift. One quantity governs the choice of m: the nearest-of-m objective is the expected distortion of a random codebook of m draws from the generator. This view yields three predictions, recorded before the corresponding runs: (i) memorization of the realized outcome in contexts that never repeat, whose strength is governed by the codebook's miss distortion I(m) rather than its miss probability, and which an energy-score generator does not escape by raising m; (ii) a flattening of mode probabilities with m, matching a known optimum within 0.01; and (iii) under-dispersion on continuous high-dimensional outcomes, which bounds the method on real robot video. Per-step matching on self-rollouts cuts long-horizon error by a third on a stochastic Push-T variant; an ablation separates its manifold and coverage benefits. Planning success does not improve despite accurate risk estimates, a negative result we report with its diagnosis.
Read the full abstract

Information Retrieval

Multi-Modal Agricultural Document Understanding at the DAGRI Task

Agricultural documents, such as prefectural technical reports and crop calendars, are critical knowledge resources. However, their complex table layouts, significant variations in document length, and dense multi-modal elements pose substantial challenges for automated understanding. To address these issues, the NTCIR-19 DAGRI (Data Analytics for aGRicultural Information) task introduced three subtasks focusing on different dimensions of document intelligence. In the Table Information Extraction (TableIE) task, we developed a multi-stage pipeline combining structure-preserving parsing via MinerU, region-adaptive segmentation for long documents, and a dual-modal (vision-text) extraction strategy for complex work schedules. This approach achieved first place with an exact match score of 0.6566. For Multi-Table Multi-Hop Question Answering (MTMH-QA) over interrelated JSON tables, we adopted an agentic search framework utilizing DCI-Agent to iteratively explore directories via bash commands, reinforced by a cross-model validation mechanism. This yielded a revised accuracy of 0.8444, initially ranked first. In the Figure and Table Question Answering (FiT-QA) task, which requires comprehending single large composite images, we implemented a dual-modal VLM pipeline enhanced by deterministic answer normalization, securing the highest BLEU score. Together, our system effectively addresses regional structural variations, dynamic cross-file evidence discovery, and multi-modal alignment, providing a robust framework for complex agricultural document understanding.
Read the full abstract

All authors Zhefan Wang, Da Li, Keping Bi, Nanqing Dong

Robotics

Design of a hybrid tendon-linkage finger with coupled and adaptive motion

Dexterous robotic fingers require coordinated joint motion and adaptability to object contact, while independently actuating multiple joints increases the complexity of the transmission mechanism. This paper presents a hybrid tendon–linkage robotic finger with four mechanical degrees of freedom driven by three actuators. The MCP abduction/adduction joint is driven through a bevel gear, while MCP and PIP flexion/extension are driven by separate bidirectional tendon transmissions. The DIP joint is not independently actuated but is coupled to the PIP joint through a four-bar linkage with a torsional spring and mechanical stopper. The tendon routing also introduces mechanical coupling between the MCP and PIP joints. Kinematic analysis was conducted to describe the coupled and adaptive motions, and experiments were performed to evaluate the joint motion characteristics. ArUco-based measurements showed that the coupled motion closely matched the theoretical prediction, whereas the adaptive motion was partially limited by interference between linkage components. The fabricated finger demonstrated different contact postures with cylindrical, cuboid, and irregularly shaped objects, and the mean peak fingertip force was 4.56 N. The proposed mechanism provides both coupled motion and contact-dependent adaptation without requiring an additional actuator for the DIP joint.
Read the full abstract

All authors Hyojae Kang, Giup Seo, Jae-Young Lee, Yeong Jae Park, Jongwoo Park, Hyunmin Do, So-Hee Park, Dongil Park, Jeongdo Ahn

High Energy Astrophysical Phenomena

The shape and velocity of black hole coronas

The first highly significant detection of the polarization of the X-rays from the Black Hole X-ray Binary Cygnus X-1 revealed surprisingly strong polarization parallel to its radio jet. The observation is now commonly interpreted as resulting from the Comptonization of some lower frequency seed photons in a laterally extended corona. The emitting plasma is likely moving along the jet axis with a sizable fraction beta of the speed of light, amplifying the polarization degree. In this letter we address the question of how laterally extended the corona actually needs to be to account for the observations, and we calculate the required outflow velocities of velocity-stratified coronas as a function of the stationary-frame aspect ratio h=H/2R of cylindrical coronas of height H and radius R. We find that models with h lesser equal 0.8 and beta=0.5-0.9 at the top of the corona can give both the observed photon index of Gamma=1.6 and the observed 4% polarization parallel to the symmetry axis. We do not find suitable solutions for larger h. Given the constraints on h, we discuss the likely size of the corona, emphasizing that compact coronas close to the black hole are preferred as they can tap more efficiently into the gravitational energy and the rotational energy of the black hole, and are less exposed to their own emission reprocessed in the optically thick portion of the accretion flow.
Read the full abstract

Earth and Planetary Astrophysics

Ejecta fate and deposition in the Didymos-Dimorphos system after the DART impact

Kinetic impact is a promising technique for asteroid deflection in planetary defense. The NASA Double Asteroid Redirection Test (DART) spacecraft collided with Dimorphos, the satellite of the binary asteroid (65803) Didymos, at 6.1449 km s^{-1} in 2022. Observations revealed changes in both the orbit of Dimorphos around Didymos and the heliocentric orbit of the Didymos-Dimorphos system. The ESA Hera spacecraft will arrive at the system at the end of 2026 to investigate the post-impact surfaces. Here, we present an analytical model, based on Maxwell's Z model, to predict ejecta behavior following the DART impact. The model provides the relationships among ejection velocity, ejection angle, and ejecta mass, allowing the momentum enhancement factor beta to be estimated for both Dimorphos and the Didymos-Dimorphos system. We show that the observed beta can be reproduced when 0.3-1% of the impact kinetic energy is converted into ejecta kinetic energy and the effective strength of the target is less than 100 Pa. The escaping mass is estimated to be 0.3-3.0% of the mass of Dimorphos. The globally averaged and minimum ejecta thicknesses on Dimorphos are estimated to be 2.0 m and 0.5 m, respectively, indicating that the entire surface was resurfaced by ejecta deposition, although boulders larger than 1 m were only partially buried. The model also predicts the momentum partition parameter kappa_p to be as low as 0.6, suggesting that the mass and the bulk density of Dimorphos may be revised upward. These predictions provide quantitative constraints for interpreting forthcoming Hera observations.
Read the full abstract

High Energy Physics (Phenomenology)

Interplay of Scalar and Pseudo-Scalar Mesonic Channels in $\Lambda\bar{\Lambda}$ Spin Correlations

The spin correlation of $\Lambda\bar{\Lambda}$ hyperon pairs produced in heavy-ion collisions is investigated through scalar and pseudo-scalar mesonic channels that mediate the creation of $s\bar{s}$ pairs. Because a $J=0$ meson forces $L=S$ while parity requires $(-1)^{L+1}=+/-1$, the two vertices project the pair onto distinct total-spin states: a scalar ($0^+$) vertex selects the $L=S=1$ configuration with a spin weight of $+1$, whereas a pseudo-scalar ($0^-$) vertex selects the $L=S=0$ configuration with a weight of $-3$; these weights are the traces of the correlation tensor, corresponding to $P_{s\bar{s}}=+1/3$ and $-1$, respectively. Implementing this framework within the three-flavor Nambu-Jona-Lasinio model at the Random-Phase-Approximation level, we compute the scalar and pseudo-scalar spectral weights. Our results show that the pseudo-scalar channel dominates, resulting in a negative net correlation fraction throughout. And the spin correlation also depends on the opening angle: strongest spin correlation at around $\theta=0$ while weakest at $\theta=\pi$. Furthermore, we show that the collision-energy dependence of the momentum-integrated spin correlation along the freeze-out line is sensitive to the trajectory's path relative to the chiral phase boundary. Trajectories passing near the crossover region exhibit a non-monotonic structure, providing a sensitive probe for the location of the chiral phase transition.
Read the full abstract

Machine Learning

When Does Physics Beat Data? Data-Efficient Traffic State Estimation with Amortized PINNs

Traffic state estimation (TSE) reconstructs traffic density at unmonitored locations from a sparse set of sensors. Data-driven models work well when sensors are dense but degrade sharply as coverage thins. We show that imposing the Lighthill-Whitham-Richards conservation law as a physics-informed prior makes TSE more data-efficient. On four real corridors (NGSIM, I-80, PeMS04, PeMS08), physics-informed estimators generalise to held-out sensors where data-driven models, including classical and deep kriging, overfit the observed ones. On the two shock-rich corridors the effect is a clear crossover: as more sensors are held out, data-driven models collapse while the physics-informed estimator degrades gracefully, leading the no-physics network by about 26-38% at 85-90% held out and the strongest classical kriging by about 10-21% at 90%. A matched smoothness prior does not reproduce the gain, and an extended Kalman filter on the same LWR model is outperformed on the shock corridors, so the advantage comes from how the conservation law is imposed. Finally, one loss-conditional PINN removes the per-corridor loss-weight search, matching a 20-point search at about 5x lower cost.
Read the full abstract

Representation Theory

Presentations of inverse Artin monoids of type $D_n$ via cluster mutations

Let $D_n^\varepsilon$ denote the Dynkin diagram of type $D_{n+1}$ with the endpoint $\varepsilon$ of its long branch frozen, and let $\mathcal{M}(D_n^\varepsilon)$ be the mutation class of any orientation of this diagram under mutations at mutable vertices. For each $\Gamma\in\mathcal{M}(D_n^\varepsilon)$, we define an inverse monoid $M_\Gamma$ by generators and relations determined by $\Gamma$. We prove that $M_\Gamma\cong M_{\mu_k(\Gamma)}$ for every mutable vertex $k$. Consequently, every quiver in this mutation class yields a presentation of the inverse Artin monoid of type $D_n$. Our argument establishes mutation invariance for the presentations of inverse Artin monoids of both types $A_n^\varepsilon$ and $D_n^\varepsilon$.
Read the full abstract

Popular Physics

Work needed to lift a mass in Schwarzschild spacetime

We determine the work needed to lift a mass from $r_2$ to $r_1$ in Schwarzschild spacetime in three ways. First we use a winch and a light rope. Then we use a photon collision. Finally we let the mass decay into photons and reconstruct it at the higher radius. All three methods give the same local work at $r_1$, $W_1=mc^2[1-N(r_2)/N(r_1)]$, where $N(r)=\sqrt{1-r_s/r}$. Thus, independently of $r_1$, the work tends to $mc^2$ as $r_2$ approaches the horizon. The rope example also gives a useful general-relativistic version of a simple machine: a rope in equilibrium need not have equal local forces at its two ends. The redshift factor plays the role of the lever arm. For a real rope the energy conditions prevent the tension from decreasing with height, so this lever advantage is confined to the idealization. The examples are intended as teaching material for undergraduate relativity courses.
Read the full abstract

Fluid Dynamics

Velocity gradient tensor reconstruction from tracer orientations via Jeffery dynamics

The velocity gradient tensor is a local quantity, but methods that obtain it by spatial differencing of the velocities of spherical tracers amplify measurement noise. We propose a method that reconstructs the velocity gradient tensor from the orientations of non-spherical tracers such as flakes and rods. Based on the Jeffery-type equation, we show that, for incompressible flow, all components of the velocity gradient tensor are uniquely determined from the orientations of four tracers and their time derivatives in the neighbourhood of a reference point (orientation-based method). The method uses neither tracer positions nor spatial differencing. It therefore requires orientation measurements only near the reference point, without requiring measurements throughout a three-dimensional region, and measurement noise is amplified only through temporal differencing. Consequently, the minimum achievable error scales as $\sigma^{1/2}$ with the noise level $\sigma$, compared with $\sigma^{1/3}$ for the positional PTV-plus-differencing approach (position-based method). Numerical tests further show that the vorticity is reconstructed more accurately than by the position-based method, whereas the strain rate is reconstructed with comparable accuracy. Finally, applying the method to spherical Couette flow, we find that the reconstruction fails where the neighbouring tracers align, which renders the inversion ill-conditioned, and that this degeneracy appears as the characteristic light-and-dark pattern familiar from flow visualization.
Read the full abstract

Machine Learning

Tropical Gaussian Anticoncentration: Optimal Instance-Dependent Bounds for Online Learning in Extensive-Form Games

We study the minimax regret of full-information online decision making in extensive-form games. For prediction with expert advice, the optimal regret is $\Theta(\sqrt{T\log K})$, where $K$ is the number of experts, or equivalently, the number of normal-form actions. For treeplex strategy spaces, existing efficient algorithms, including KOMWU and online mirror descent with the weight-one dilated entropy regularizer, achieve regret $\mathcal{O}(\sqrt{T\log|\mathcal{V}|})$, where $\mathcal{V}$ is the set of reduced normal-form strategies. We prove that this logarithmic dependence is minimax optimal in the worst-case full-information reward model. The main difficulty is that pure plans in a treeplex are not independent experts: their rewards are correlated through the recursive structure of decisions and observations. To capture this correlation, we introduce tropical Gaussian formulae, recursive Gaussian expressions built from maxima and normalized sums which preserve the variance. The index count of such a formula matches the number of reduced normal-form continuation plans in the corresponding subtree. Our main analytic result shows that every log-balanced tropical Gaussian formula with index count $N$ has Gaussian width $\Omega(\sqrt{\log N})$. While a direct approach struggles to recover this lower bound, our proof uses a spherical entropy profile to measure scale-dependent separation among the Gaussian directions, and then applies Turán's theorem and Sudakov minoration to obtain the width lower bound. Applying this result to treeplexes gives an oblivious hard distribution with valid transition outcome and Rademacher terminal rewards, establishing a regret lower bound of $\Omega(\sqrt{T\log|\mathcal{V}|})$. This matches known upper bounds up to universal constants and shows that the DilEnt/KOMWU dependence on the reduced normal-form complexity is unavoidable.
Read the full abstract

All authors Ashkan Soleymani, Zhiyuan Fan, Lillian J. Ratliff, Patrick Jaillet, Gabriele Farina

Distributed Computing

Accelerating Direct Sparse Solvers with Graph Computing Accelerators

Solving large sparse linear systems by Gaussian elimination is a challenging computational task due to the introduction of non-zero elements, or fill-in. Practitioners use heuristics such as the nested dissection reordering algorithm to try to reduce fill-in. In practice, reordering codes do not achieve the same parallel scaling as factorization, nor can they exploit AVX instructions or GPUs. Thus, reordering can be the primary runtime bottleneck in large-scale computer-aided engineering applications. We consider the use of a novel accelerator, the UpDown system, to accelerate the METIS multilevel nested dissection reordering algorithm. The UpDown system is designed for a broad class of graph computing and irregular algorithms. We report results for small-scale UpDown systems that will be available commercially in 2027, using validated simulation and FPGA emulation platforms. Our findings indicate that small UpDown systems can speed up METIS by two-orders of magnitude versus a single CPU. Compared to ParMETIS on hundreds of MPI ranks, we anticipate the same small UpDown systems offer speedups of 4 - 7, providing the potential to significantly speed finite element simulations
Read the full abstract

All authors Jiya Su, François-Henry Rouet, Jeff Beisheim, Robert Lucas, Yuqing Wang, Alexander Fell, Chris Mccarthy, David F. Gleich, Andrew A Chien

Computer Vision

Beyond Head Importance: Collective Computation in Video Transformers

Attention heads in video Transformers vary greatly in functional importance. Understanding their contributions requires examining not only which heads are important, but also how they participate in collective computation within the network. We study this collective computation from two perspectives: the downstream use of head outputs and the composition of interactions across layers. Ablation analyses of four video Transformers show that high-importance heads identified on a single task also support other video understanding tasks and help preserve the representational structure that distinguishes different videos. Representational responses to head ablation correlate weakly with head importance after the attention module, but more strongly after the MLP in the same block. Adapting this downstream processing through MLP-only fine-tuning further improves the use of lower-importance head outputs. Meanwhile, through composition across layers, sets of heads with shorter spatiotemporal attention distances support stronger perturbation responses at distant tokens than sets with longer distances. Their joint ablation causes even greater performance loss than ablating equally sized sets of high-importance heads, revealing collective contributions not fully captured by individual importance rankings. These findings highlight collective computation as a key perspective for understanding attention head function: functional contributions depend on both how head outputs are used downstream and how local interactions compose across layers.
Read the full abstract

Statistics Theory

An iterated-logarithm price of anytime validity for log-optimal e-processes

Log-optimal e-variables at fixed horizons, also called numéraires, have been conjectured to “almost” form an e-process for exponential-family null hypotheses. A related open question is whether an e-process can attain the optimal e-power at every horizon up to a bounded loss. Given a final horizon, we study the minimax regret of e-processes relative to the numéraires at all times up to it. For the Gaussian location null and the composite alternative of Gaussian distributions with a smaller variance and an unknown location, we show that the minimax regret in the worst case over the location grows like the iterated logarithm of the final horizon. A single e-process attains this rate at every horizon. This growth answers the question negatively. It also refutes the conjecture if “almost” means a regret bounded uniformly over horizons. For this Gaussian location problem, we thus exhibit the frontier precisely: the minimax regret is bounded if and only if the alternatives have at least the variance of the null.
Read the full abstract

Computer Vision

Where Does Proposal Selection Fail? Diagnosing Frozen Vision-Language Queries over Detector Proposals

In modular grounding a frozen detector proposes regions and a vision-language model (VLM) selects one, at lower cost than letting the VLM decode a box, but it is unclear where such systems fail. We hold perception fixed—every trained selector chooses among the same frozen proposals, which contain the referent for at least $99.8\%$ of RefCOCO/+/g queries—and vary the selection interface one factor at a time for two VLMs of one family (Qwen2.5-VL-7B, Qwen3-VL-8B). On RefCOCOg, what the selector reads matters more than how it is built: the VLM's image-conditioned query encodes the referent's location and beats a text-only query by $0.15$–$0.18$ with every head tested (by less for InternVL3-8B, whose vision encoder differs; one head tested). The remaining errors are rankings among a few top candidates; a cross-attention selector supervised with every valid box reaches the VLM's coordinate decoding on RefCOCOg, and with Qwen2.5-VL on RefCOCO/+ when also trained on their data. Beyond that distribution it does not: on Ref-L4's Objects365 images, mostly of categories outside training, every trained head falls to $0.19$–$0.28$ while Qwen decoding keeps about $0.8$ (InternVL3's drops too), neither a wider detector prompt nor a class-agnostic proposer closes the gap, and on public reasoning benchmarks Ref-Adv-s and RefBench-PRO full-budget heads trail decoding by $0.19$–$0.37$ although proposals cover most referents. A trained verifier reduces to an existence probe on the VLM's query state and does not beat the VLM's yes/no answer on public no-target benchmarks.
Read the full abstract

Statistical Methodology

Distribution-Free Symmetry Detection in Multi-Reference Alignment at Low SNR

Symmetry can reveal the structure of a signal, but detecting it becomes difficult when observations are noisy and randomly transformed. We study how to infer symmetry directly from multi-reference alignment data with additive isotropic Gaussian noise of known variance, without aligning the observations or knowing how the transformations were sampled. For cyclic symmetry, we identify a simple statistical signature that yields tests with exact finite-sample error guarantees. We demonstrate their performance even at very low signal-to-noise ratios. Dihedral symmetry presents a deeper challenge: we show that, for some signals, invariant moments below a sharp order threshold cannot distinguish reflectional from purely rotational symmetry. For generic signals, however, this distinction emerges much earlier. We use these insights to develop tests for dihedral symmetry with asymptotic error control under explicit nondegeneracy conditions. Together, the results clarify what we can learn about symmetry from noisy, unaligned data and suggest a path toward symmetry inference in cryo-EM.
Read the full abstract