Robotics
VAST: V2X/Dynamic Map-Aware Autonomous Driving Systems Validation Toolchain
Shunsuke Ito and Takuya Azumi
Cooperative autonomous driving in the IoT-to-Edge-to-Cloud continuum requires system-level validation across vehicles, infrastructure sensors, edge-side Dynamic Map services, and in-vehicle autonomous-driving stacks. This paper presents VAST, a V2X/Dynamic Map-aware validation toolchain that connects Scenic, Scenario Simulator v2, AWSIM, Autoware, and SIM-LDM. VAST does not introduce a new search
algorithm; instead, it addresses interoperability challenges, including Lanelet2-to-Scenic mapping, ROS 2-based co-simulation through SS2, Dynamic Map object injection into Autoware, and collection of TTC, PET, collision, timeout, and performance measurements. In occluded-intersection scenarios, Lanelet2-compatible constrained sampling increases the edge-case discovery rate from 40.0% to 80.0% and reduces the average time per discovered edge case from 259.7 s to 110.4 s. Under the same generated scenario distribution, Dynamic Map availability reduces the collision rate from 78.0% to 40.0% and increases non-collision outcomes from 22.0% to 60.0%, with statistically significant TTC/PET shifts. A throughput study with 1-16 NPCs shows that sampling remains below 0.1 s, whereas AWSIM/Autoware execution and restart overhead dominate runtime. These results position VAST as a practical validation infrastructure for cooperative autonomous-driving CPSs.
Read the full abstract ▾
Artificial Intelligence
FINSKILLOPS: A Self-Evolving Multi-Agent System for SEC Filing QA
Yanzhang Ma, Zhenghan Tai, Hanwei Wu et al. (+25 more)
Financial QA systems are typically improved before deployment through better
retrieval, prompting, or
agent coordination, leaving their reliability behavior fixed thereafter. In practice, new SEC-filing questions repeatedly expose heterogeneous errors in period, entity, evidence use, and calculation. Existing self-improvement methods can turn failures into new behaviors, but offer limited control over where a correction should apply or which previously correct answers it may break. We therefore frame post-deployment improvement as controlled behavioral maintenance: recurring failures should become scoped skill patches, and each patch should earn deployment with- out introducing regressions. We instantiate this view in FINSKILLOPS, a multi-
agent system for SEC filing QA. FINSKILLOPS derives reusable skills from evidence-grounded, typed failure diagnoses and governs them through targeted validation, protected-case regression checks, negative controls, and versioned replacement or retirement. Across six financial QA
benchmarks, a single frozen skill registry achieves the highest verdict-weighted correctness and reference consistency among the evaluated systems. Evolved skills raise correctness from 3.70 to 4.55 on our enhanced
benchmark. In a separate 12-round operational study, only six of 33 proposed skills are promoted, while the monitoring non-correct rate falls from 20.0% to 12.5%. These results establish controlled skill scope, admission, and lifecycle management as the foundation for reliable self-improvement.
Read the full abstract ▾
All authors Yanzhang Ma, Zhenghan Tai, Hanwei Wu, Sizhe Guan, Jianliang Lei, Hailin He, Chaolong Jiang, Jijun Chi, Tung Sum Thomas Kwok, Bohuai Xiao, Jingrui Tian, Xinlu Wu, Xingao Zhan, Peng Lu, Muzhi Li, Yihong Wu, Liheng Ma, Sicheng Lyu, Tianshuo Yan, Junhao Zhu, Yaqian Xu, Lei Ding, Yufei Cui, Ziquan Liu, Boyu Han, Hengli Liu, Ling Zhou, Xinyu Wang
Mathematical Physics
Contact Large Deviations of Stochastic Vector Bundles
D.Y. Zhong
Large deviation theory lacks a
geometric foundation for its rate functions and fluctuation
symmetries. This paper develops a contact large deviation theory on
stochastic vector bundles, in which the rate function, the scaled cumulant generating function, and a Gallavotti–Cohen-type fluctuation duality all follow from the contact 1-form. The constraint function acts as a generalized Lagrangian; the least constraint principle yields the dynamics, and the contact potential is built order by order from the master equation, producing a coupled Hamilton–Jacobi–transport system governed by the invariant density, drift, and fluctuation tensor. The contact path measure satisfies a large deviation principle with rate function given by the constraint action; the scaled cumulant generating function obeys a stationary
eigenvalue equation with a Donsker–Varadhan variational characterization. The
entropy production rate, the fluctuation–dissipation combination $e=\tfrac12 g^T Ag-\sigma$, is the physical observable; the time-reversal involution $J:(t,y,\phi)\mapsto(t,y,-\phi-\nabla\ln\rho)$ with reversed drift $v^{\mathrm{rev}}=-v+Ag$ yields a Gallavotti–Cohen-type duality $\lambda_{\mathrm{fwd}}(q)=\lambda_{\mathrm{rev}}(q+1)+\lambda_{\mathrm{fwd}}(-1)$. The classical one-dimensional GC
symmetry is recovered in the reversible case $Ag=0$.
Read the full abstract ▾
Superconductivity
Crystallographic imperfections and exotic superconductivity of UBe$_{13}$
Andreas Leithe-Jasper, Markus König, Yusei Shimizu et al. (+16 more)
The
symmetry of the superconducting gap is related to the
symmetry of the
crystal structure. In unconventional
superconductors, big changes of the critical temperature, critical field or even the gap structure can happen even for small perturbations of the
lattice. In this letter, we use microstructuring to study aluminium-free
crystals of UBe$_{13}$ which are expected to be closer to "perfect" material than previously studied aluminium-grown single
crystals. We compare the effect of minuscule imperfections on the value of the critical temperature and critical field, which, in the case of UBe$_{13}$, has drastic effects, supporting its unconventional nature. We
conjecture that this likely arises from the oxidation state of uranium, which is sensitive to its
crystal environment. Our findings suggests that uranium-based materials provide not only an excellent reservoir of new unconventional phenomena, but also a way to identify the gap
symmetry of unconventional
superconductors.
Read the full abstract ▾
All authors Andreas Leithe-Jasper, Markus König, Yusei Shimizu, Konstantin Semeniuk, Primož Koželj, Mitja Krnel, Paul Simon, Wilder Carrillo-Cabrera, Nazar Zaremba, Marcel Naumann, Ulrich Burkhardt, Thomas Doert, Yurii Prots, Alfred Amon, Elena Gati, Matthias Vojta, Yuri Grin, Elena Hassinger, Eteri Svanidze
Game Theory
A Logarithmic Regret Bound for Optimistic Hedge in General-Sum Games
Junsoo Ha
Can simple no-regret dynamics attain smaller regret in self-play than against arbitrary adversaries? In $n$-player general-sum games, Daskalakis et al. 2021 proved an $O(n\log d_i\log^4 T)$ individual regret bound for Optimistic Hedge, which improves upon the classical $O(\sqrt T)$
adversarial regret bound. In this work, we show that Optimistic Hedge with a constant step size can further achieve $O(\sqrt n\log d_i\log T)$ individual external regret under expected loss-vector feedback. The time-averaged play consequently enjoys a coarse correlated
equilibrium gap $O(\sqrt n\log d\log T/T)$, where $d=\max_i d_i$. The improvement comes from a larger admissible step size $\eta=\Theta(1/(\sqrt n\log T))$. Our analysis proves factorial bounds on high-order differences of probability-weighted pairwise loss gaps, then applies finite-difference interpolation in a fixed Euclidean norm. These estimates sharpen the analysis of Daskalakis et al. 2021 and yield a logarithmic regret bound.
Read the full abstract ▾
Quantum Physics
Linear combination of Schrödingerization for quantum linear systems with optimal matrix-query complexity
Yin Yang, Yue Yu and Long Zhang
Quantum linear systems
algorithms (QLSAs) aim to solve linear systems $A\bb{x}=\bb{b}$ exponentially faster than classical methods under certain conditions. In this work, we develop
quantum algorithms for solving linear algebraic equations from an ODE-based perspective. Inspired by the linear combination of
Hamiltonian simulation (LCHS) representation in the
Fourier approach \cite{Childs2017QLSA}, we express the solution $\bb{x}$ as a linear combination of solutions to a system of linear convection equations, which become Schrödinger-type equations with
unitary evolutions in the
Fourier domain. We refer to this representation as LC-Schrödingerization. Based on this result, we construct an LCHS-based
quantum algorithm with two LCHS instances: one for time-marching and one for numerical integration. The key construction uses the derivative of a Gaussian-smoothed hat function and recovers the solution over a fixed auxiliary interval. This permits a truncation time independent of the target accuracy and avoids the loss in success probability from selecting a single grid point. Periodization and explicit
Fourier coefficients provide the corresponding projection error bounds. Under the stated oracle assumptions and given a constant-factor estimate of the solution norm, direct simulation of the select operators and block preconditioning achieve the optimal
matrix-query complexity $\mathcal{O}(\kappa_A\log\frac1\varepsilon)$ without using variable-time amplitude amplification (VTAA). The same upper bound holds for queries to the right-hand-side preparation oracle.
Read the full abstract ▾
High Energy Physics (Phenomenology)
Neutralino Dark Matter in SU(5) after H.E.S.S. and Direct-Detection Searches
M. Adeel Ajaib and Fariha Nasir
We examine the impact of gamma-ray line and direct-detection searches on neutralino dark matter in supersymmetric SU(5) with non-universal gaugino masses. Under the assumption that neutralinos constitute all
halo dark matter, a nearly pure wino
benchmark at 1 TeV is essentially excluded by the H.E.S.S. line limits for both cusped and cored density profiles. Higgsino and wino-Higgsino
benchmarks are independently excluded by direct detection. In contrast, a bino
benchmark near thermal saturation combines suppressed present-day line emission with scattering below the displayed experimental limits. Mixed bino-wino solutions illustrate the complementarity of the two searches and the dependence of the line interpretation on the
Galactic halo profile. We relate these results to the GUT-scale gaugino hierarchy and the SU(5) scalar and
Higgs boundary conditions. Thermally underabundant solutions require additional production to realize the full-density assumption. Comparing the calculated signals with published electroweakino predictions, we identify the spectrum-dependent corrections and
collider tests needed to determine the surviving parameter space.
Read the full abstract ▾
Machine Learning
Conservation Buys Stability and Factoring Buys Counterfactuals in Physical World Models
Yufeng Wang, Parivesh Priye, Lu Wei et al. (+1 more)
A learned simulator can reproduce its training conditions accurately yet fail in two distinct ways once those conditions change. Over long rollouts, small errors accumulate until the trajectory drifts away from physically plausible behavior; under an intervention on a physical parameter, the model may continue to follow the law seen during training rather than the intervened one. We show that these two failures require different structural remedies. Evolving a learned energy with a
symplectic integrator preserves the geometry of the conservative dynamics and keeps rollouts bounded and physically meaningful for up to $100\times$ the training horizon, while equal-capacity predictors, an energy-regularized predictor, and a tuned neural ODE diverge. By contrast, encoding the physical coupling through an explicit linear factorization enables the model to follow a never-seen sign of that coupling, whereas an unrestricted parameterization remains locked to the training law. Crucially, the two mechanisms are separable: removing the structure responsible for long-horizon stability leaves counterfactual transfer intact, while removing the factorized coupling destroys counterfactual transfer without eliminating stability. This double dissociation, established with matched controls that remove or replace one structural component at a time, persists beyond the headline three-body system and remains visible when the physical state must be inferred from pixels rather than provided directly. The result is a concrete design principle for physical world models: long-horizon stability and changed-law generalization arise from distinct structural commitments, and each can be imposed deliberately without requiring the other.
Read the full abstract ▾
All authors Yufeng Wang, Parivesh Priye, Lu Wei, Haibin Ling
High Energy Physics (Phenomenology)
Clarifying the puzzling mass shift of the $\psi(4160)$ via a reanalysis of $R$-value data with unquenched charmonium spectroscopy
Tian-Cai Peng and Xiang Liu
The long-standing upward shift of the extracted $\psi(4160)$ mass, from about $4.16$~GeV to $4.19$~GeV in later analyses, remains a puzzling issue in charmonium spectroscopy. In our previous study, this problem was investigated through the $B^+\to K^+\mu^+\mu^-$ process within an unquenched charmonium framework, where the lower-mass $\psi(4160)$ assignment was found to be compatible with the data. Here we revisit the BESII $R$-value data, which played an important role in the historical extraction of the higher $\psi(4160)$ mass, and provide an independent examination. In contrast to the conventional quenched picture with $\psi(4040)$, $\psi(4160)$, and $\psi(4415)$, the unquenched vector-charmonium spectrum contains six states: $\psi(4040)$, $\psi(4160)$, $\psi(4220)$, $\psi(4380)$, $\psi(4415)$, and $\psi(4500)$. Including these states together with the near-threshold $\psi(3770)$, we find that the BESII $R$-value line shape can be well reproduced over the full energy range while retaining the lower-mass $\psi(4160)$ assignment. The enhancement around $4.19$~GeV then arises from the coherent interplay among the nearby $\psi(4040)$, $\psi(4160)$, and $\psi(4220)$ amplitudes, rather than requiring an upward shift of the $\psi(4160)$ mass itself. The additional higher states also naturally describe the line-shape structure in the $4.4$~GeV region. We further show that the seven-resonance coherent amplitude contains six complex zeros, yielding $2^6=64$ mathematically equivalent solutions with identical line shapes but substantially different di-electron widths and relative phases. Comparing these solutions with available experimental information and representative unquenched charmonium predictions, we provide a qualitative assessment of their phenomenological consistency and highlight several solutions that appear more compatible with present information.
Read the full abstract ▾
Astrophysics of Galaxies
On the Probability Distribution and Null-hypothesis Testing of Cross-correlation for Light Curves in Active Galactic Nuclei
Yan-Rong Li and Jian-Min Wang
Cross-correlation is crucial in studies of multiwavelength flux variability in active
galactic nuclei (AGNs), especially for reverberation mapping analysis, where interpolated cross-correlation function is widely used to measure time lags between light curves. While time-lag uncertainties can be estimated via the flux randomization and random subset selection method, an appropriate framework for assessing cross-correlation significance remains lacking in the literature. Here we attempt to fill this gap by leveraging the well-established property from
stochastic time series theory, namely that the probability distribution of cross-correlation coefficients for independent
stochastic light curves asymptotically approaches a normal distribution. Its variance can be analytically estimated using the auto-correlation functions of the light curves. We employ Monte Carlo simulations to validate this property for irregularly sampled, red-noise AGN light curves, and then propose a fast procedure to perform null-hypothesis testing for the cross-correlation of AGN light curves. We also present exemplary applications to AGN reverberation mapping data. This procedure requires prior determination of the auto-correlation functions of light curves, which can be obtained via model fitting. The long-standing issue regarding unbiasedly recovering the auto-correlation function remains unresolved when light-curve duration is comparable to the typical variation timescale, warranting further future investigation.
Read the full abstract ▾
Artificial Intelligence
When2Think: Learning When and How Much to Reason
Jaejun Shim, HyunJin Kim, Young Jin Kim et al. (+1 more)
Large
Reasoning Models (LRMs) often overthink easy problems and underthink hard ones, leading to inefficient computation allocation. Existing methods regulate generated computation or select between direct answering and explicit
reasoning, but do not jointly control whether}to reason and how much computation to allocate within
reasoning. We call the resulting difficulty-dependent loss in accuracy under computation reduction the efficiency tax. We propose When2Think, an RLVR-based post-training framework for instance-adaptive computation allocation. Its core mechanism, Instance-level Difficulty-Aware Control (IDAC), uses cached reference statistics of success and
token cost to modulate a correctness-gated efficiency bonus based on generated
token count. Importance sampling supports exploration of Think and NoThink, while Batch-Wise Standardization constructs standardized advantages for critic-free optimization. The framework requires neither a learned reward model nor a learned critic, and offline reference caching avoids online reference-model queries during policy updates. On AIME24, When2Think improves Pass@3 by 10.0 percentage points while reducing
token usage by 27.9% relative to the backbone.
Read the full abstract ▾
All authors Jaejun Shim, HyunJin Kim, Young Jin Kim, JinYeong Bak
Machine Learning
CoRe: Coherence and Relational Alignment for Multivariate Time Series Forecasting
Xiaoyu Lin, Huiran Duan, Yining Liu et al. (+3 more)
Direct
forecasting has become a standard paradigm for multivariate time-series
forecasting because it predicts the full future horizon in a single pass. However, its training objective is often still decomposed into pointwise errors such as MSE. Such objectives provide stable supervision, but they do not explicitly preserve the structure of the future trajectory: temporal coherence within each variable and relational consistency across variables can both be weakened. We propose CoRe, a model-agnostic learning objective for direct multivariate
forecasting. CoRe replaces pointwise supervision with two output-space constraints: a frequency coherence loss that aligns predicted and target spectra, and a low-rank relational graph loss that matches sampled pairwise differences in a target-derived PCA subspace. The resulting objective introduces no trainable parameters and can be applied to existing
forecasting backbones by changing only the loss. Experiments on standard
benchmarks show that CoRe improves strong baselines, compares favorably with recent
forecasting objectives, and remains effective across different backbones,
datasets, and hyperparameter settings overall consistently.
Read the full abstract ▾
All authors Xiaoyu Lin, Huiran Duan, Yining Liu, Zhixiang Wu, Chu Lin, Lin Lu
Computer Vision
Beyond Patch Removal: Persistent Adversarial Effects in Vision-Language-Action Policies
Enhao Wu, Fusen Guo, Yuxin Cao et al. (+3 more)
Adversarial patches to
Vision-
Language-Action (VLA) policies can cause both immediate action corruption and persistent state effects that remain after the patch is removed. Existing evaluations largely focus on continuous attacks and do not separate these two effects. We introduce a state-restoration protocol that removes the patch at matched action-chunk boundaries and measures subsequent recoverability under the same remaining step budget. Clean, random-patch, deviation-matched, and fixed-direction controls distinguish
adversarial effects from occlusion, action-error magnitude, and directional persistence. We also evaluate a recovery adapter trained on attack-induced states under controlled intervention latency. On OpenVLA-OFT with EDPA attacks, only 36.2% of LIBERO-Long episodes remain recoverable after five chunks, compared with 89.9% and 87.0% for the deviation-matched and fixed-direction controls. Similar persistent effects are observed on autoregressive OpenVLA. The recovery adapter improves recovery from 7.7% to 47.4% at one-chunk latency, but its benefit decreases substantially with delayed intervention. These results show that
adversarial effects can persist after patch removal and that timely intervention is critical for recovery.
Read the full abstract ▾
All authors Enhao Wu, Fusen Guo, Yuxin Cao, Ziyang Lyu, Lin Li, Wei Song
Functional Analysis
Surjective Fischer–Muszély maps on positive cones of JB-algebras
Osamu Hatori and Shiho Oi
We prove that every surjective Fischer–Muszély map between positive cones of arbitrary JB-
algebras is additive and positively homogeneous, and extends uniquely to a bounded positive linear surjection. In particular, this gives a complete affirmative answer to Molnár's problem for arbitrary $C^*$-
algebras. The
theorem includes nonunital and exceptional JB-
algebras and assumes neither injectivity nor continuity. The proof connects the norm-valued functional equation to positive-cone
rigidity through translation invariance of an induced pseudometric. For a surjective map satisfying the Fischer–Muszély identity with uniform error $\varepsilon$, we also construct a unique bounded positive linear map $L$ at uniform distance at most $3\varepsilon$. The set $L(A_+)$ is dense in $B_+$, although $L$ need not be surjective, even when the original map is continuous and bijective. For bijective FM maps, we obtain weighted Jordan representations on the biduals. We also extend the characterization of surjective norm-sum preservers to JB-
algebras and characterize norm arithmetic-midpoint and harmonic-mean identities on positive invertible cones of unital JB-
algebras.
Read the full abstract ▾
Fluid Dynamics
Experimental study of the impact dynamics of polymeric hollow droplets
Mohammad Mahdi Nasiri, Mohammad Reza Daneshvar Garmroodi, Damian Vadillo et al. (+1 more)
The impact dynamics of hollow droplets, while influential in applications such as coating and spraying, remain less explored than their dense counterparts. In particular, the impact dynamics of viscoelastic hollow droplets have yet to be fully explored. This study presents an experimental investigation into the impact of hollow Newtonian (water) and viscoelastic (polymeric solution) droplets on a solid surface at different impact velocities and
polymer concentrations. We demonstrate two hallmark features of hollow droplet flattening: the formation of a central counter-jet and the final deposition, both associated with the entrapped air bubble. For Newtonian droplets, the counter-jet exhibits rapid growth and breakup due to capillary instabilities. Introducing
polymer additives fundamentally alters this behavior: viscoelasticity affects the counter-jet's height and velocity, delays bubble rupture, and inhibits droplet detachment. Crucially, we observe the emergence of beads-on-a-string structures during filament thinning, a signature of the competition between elastic and capillary forces. By systematically varying the
polymer concentration and impact velocity, we identify the conditions under which three distinct outcomes occur: deposition, partial deposition, and detachment. Our results show how inertia, viscosity, capillarity, and elasticity together govern the splashing morphology of hollow non-Newtonian droplets.
Read the full abstract ▾
All authors Mohammad Mahdi Nasiri, Mohammad Reza Daneshvar Garmroodi, Damian Vadillo, Moussa Tembely
Robotics
Towards High-DoF Dexterous Manipulation through VLA Post-Training
Junlei Zhu, Shenzhe Yao, Chaogui Huang et al. (+6 more)
Imitation-learned
vision–
language–action (VLA) foundation models acquire broad manipulation capabilities by scaling
robot data across tasks and embodiments, but reliable deployment on a specific downstream task and hardware platform still requires post-training. Dexterous hands make this adaptation particularly difficult: their broad behavioural repertoire and high degree of freedom create a large and structured action space. Three obstacles are central: open-source VLAs do not natively provide an action interface for high-DoF hands; gesture mismatch during human-gated DAgger takeover creates command discontinuities and contaminates corrective trajectories; and reinforcement learning in the raw joint space is sample-inefficient. We present a unified four-step post-training pipeline comprising a learned temporal hand-action codec, supervised fine-tuning, DAgger, and real-world residual reinforcement learning. The codec adapts a pretrained VLA to absolute dexterous-hand commands. Buffered rollback, pose
alignment, and smooth command blending enable continuous, task-relevant DAgger corrections, while latent residual RL confines exploration to coordinated hand motions captured by the codec. We evaluate the pipeline on five diverse real-world tasks spanning bimanual transfer, in-hand reorientation, and tool use. Within the reported post-training budgets, the resulting policies achieve 100% success on every evaluated task over 20 trials per task. These results provide a practical path for adapting VLA foundation models to reliable real-world dexterous manipulation.
Read the full abstract ▾
All authors Junlei Zhu, Shenzhe Yao, Chaogui Huang, Wenkai Zhu, Jingwei Peng, Guanqi He, Soren Schwertfeger, Jiahao Chen, Yide Liu
Robotics
Runtime Safety Filtering for Two-Terminal Hazards in Robotic Battery Recycling
Yuxin Cao, Wei Song, Xianglin Yang et al. (+4 more)
Runtime safety filters for learned manipulation policies typically define unsafe states as unions of object-wise keep-out regions. This representation can be unnecessarily restrictive for hazards that depend on a joint spatial relation, such as battery recycling, where a conductive payload can short a charged
cell only when it approaches both terminals simultaneously. We study runtime filtering for this two-terminal hazard in LIBERO using frozen OpenVLA policies. We factor a runtime filter into three design choices: the predicate structure, its
geometric margin, and the fallback action applied when a commanded action is rejected. We compare a conjunctive predicate, a conventional two-site keep-out, and a composite of the two. For each predicate, we vary its margin to obtain a frontier between task success and residual hazard. We then compare four fallback strategies at matched operating points: holding, retreat, sampled search, and a continuous-action barrier projection. Across three workcells, the three predicate families trace nearly identical safety–utility frontiers once each is evaluated over its own margin. In contrast, the fallback strategy has a substantially larger effect: holding reduces task success by up to 0.302 relative to retreat without reducing hazard, while both minimally invasive fallbacks leave substantially more residual hazard. This ordering transfers to a second policy and task suite, while retreat-based filtering remains effective under standing errors in the clearances available to the filter, although correlated error in the estimated payload size is more damaging than larger independent errors in terminal position. These results show that, for proximity-defined manipulation hazards, margin selection and fallback strategy can matter more than predicate structure in determining the safety–utility trade-off of a runtime filter.
Read the full abstract ▾
All authors Yuxin Cao, Wei Song, Xianglin Yang, Fusen Guo, Lin Li, Xiao Cheng, Jin Song Dong
Computer Vision
VideoResearcher: Self-Improving Tool Design for Long-Video Understanding
Dingqiang Ye, Dongdi Zhao, Kaishen Wang et al. (+11 more)
Video agents have made substantial progress in long-
video understanding. Yet effective
video-
agent systems require costly, time-consuming manual design and trial and error. Current self-improvement methods either refine low-impact
prompts, recombine predefined micro-tools, or struggle with convergence in harness optimization. To bridge this gap, we target high-impact
video-tool with VideoResearcher, a training-free multi-
agent framework that autonomously designs, tests, and refines tools for
video understanding, like a human researcher. VideoResearcher operates through dual Solving and Evolving loops: it analyzes tool-use trajectories to identify capability gaps, coordinates specialized
agents to develop and validate executable tools, and reuses evolved tools to strengthen evidence acquisition in subsequent
video reasoning. Through iterative tool refinement and validation, it progressively strengthens evidence acquisition without updating model parameters. VideoResearcher achieves state-of-the-art performance among self-improving
agents and approaches the human-designed upper bound, demonstrating a training-free paradigm for long-
video understanding that expands
agent capabilities through autonomous tool development while reducing costly manual engineering.
Read the full abstract ▾
All authors Dingqiang Ye, Dongdi Zhao, Kaishen Wang, Qingqiao Hu, Jingchen Sun, Yijun Liang, Yuqi Jia, Yiqiao Huang, Yunjie Tian, Jiaxing Zhang, Chuanyang Jin, Ke Zhang, Vishal M. Patel, Di Fu
Mathematical Physics
A c-transform optimal transport method for high-contrast freeform reflector design
Gang Bao, Yixuan Zhang and Jiaqi Yu
Design of high-contrast freeform reflectors is challenging, as the presence of zero-intensity regions leads to degeneracy in the associated Monge-Ampere-type equation. A common remedy is to add a positive artificial background to the target intensity, which improves the regularity of the equation. However, this regularization inevitably reduces the achievable illumination contrast and introduces nonzero intensity into regions that are intended to remain dark. We propose a fast c-transform method based on optimal transport duality for high-contrast freeform reflector design in the far field without artificial background regularization. The method integrates repeated discrete c-transforms into a measure-based dual optimization scheme to maintain the c-concavity of the reflector potential throughout the iteration. This preserves the admissibility of the iterates and enables stable convergence for targets with zero-intensity regions. To efficiently compute the discrete c-transforms, we develop a localized search
algorithm that exploits their contact structure and prove its exactness and linear complexity. Numerical experiments demonstrate that the proposed method accurately realizes high-contrast illumination targets with zero background.
Read the full abstract ▾
Computer Vision
Towards Active Cross-View Object Geo-Localization
Shunyu Yao, Xiaohan Zhang, Zhuoran Yang et al. (+5 more)
Cross-view object geo-localization (CVOGL) typically assumes a fixed query image, overlooking the ability of mobile
agents to actively acquire more informative observations. To address this limitation, we introduce Active Cross-View Object Geo-Localization (ActiveGeo), where an
agent sequentially selects new viewpoints and determines when to stop, aiming to improve localization with minimal observations. We further propose ActiveMoPT, an ActiveGeo framework with three-stage training. First, Multi-View
Prompt-Preserving Adaptation enables the model to aggregate multiple query views while reusing the initial
prompt. Second, Trajectory-Guided Policy Initialization uses supervised
agent trajectories to learn viewpoint selection and initial stopping behavior. Third, Cost-Aware Policy Refinement employs GRPO with a gain-cost reward to jointly optimize localization accuracy and observation efficiency. We also construct ActiveGeo-858, a zero-shot test set containing 858 scenes and 1,716 target annotations. Experiments show that ActiveMoPT achieves state-of-the-art performance on MoP-
UAV using only 1.45 query views on average, and substantially outperforms previous CVOGL approaches under zero-shot evaluation on ActiveGeo-858.
Read the full abstract ▾
All authors Shunyu Yao, Xiaohan Zhang, Zhuoran Yang, Haoqi Lai, Qi Ming, Xiaoxi Hu, Hui-Liang Shen, Si-Yuan Cao
Robotics
ReShoot: Generative Visual Domain Randomization of Recorded Robot Demonstrations for Visuomotor Policy Learning
Chiyoung Kim, Min Sung Choi, Jinho Ju et al. (+5 more)
Imitation-learned
robot policies are frequently overfit to the visual conditions present in their training demonstrations. Consequently, variations in object color or background appearance often induce substantial performance degradation. A common mitigation strategy is to acquire additional demonstrations in each novel visual context; however, this approach is resource-intensive, requiring repeated access to a
robot, a controlled environment, and human operation for every appearance condition to be covered. We introduce ReShoot, a framework that synthesizes visual diversity by re-rendering previously recorded demonstrations under altered appearances, thereby shifting the burden from data collection to generation. A
vision-
language model captions the scene, edits a targeted attribute (e.g., background, object color, or material), and an edge-conditioned
video generator re-renders both camera views to match. The instruction is updated accordingly. The action sequence and proprioceptive trajectory are copied verbatim without relabeling, so each generated episode retains the recorded action and proprioceptive labels. On LIBERO, a policy trained on an equal mixture of recorded and re-rendered demonstrations matches the performance of recorded-only training (96.5% vs. 96.9%). Moreover, the mixed training set improves robustness to scene perturbations on LIBERO-Plus (85.5% vs. 82.3%). Across two physical robotic platforms, deploying ReShoot with 43 and 100 pre-collected demonstrations increased the success rate on recolored objects from 0.0% to 42.9% and 47.5%, respectively, while maintaining performance under the original recorded appearance.
Read the full abstract ▾
All authors Chiyoung Kim, Min Sung Choi, Jinho Ju, Chanhoe Gu, Donghwan Hwang, Wonseok Choi, Woongsun Jeon, Minhyeok Lee
General Economics
Screening Out the Needy: The Effects of SNAP Work Requirements
Lexin Cai, Hyewon Kim and Pauline Leung
We examine the effectiveness of work requirements as a screening device in the Supplemental Nutrition Assistance Program (SNAP). Work requirements for "able-bodied adults without dependents" were suspended after the Great Recession and gradually reinstated across counties and states in the 2010s. Using linked administrative SNAP and employment data from five states and a triple-differences design, we find that work requirements reduce SNAP participation by seven percent without increasing labor supply and disproportionately screen out low-income individuals. We develop a welfare framework to interpret these results and find that the social costs of work requirements exceed budget savings.
Read the full abstract ▾
Robotics
EmbodiedMind: Adaptive Data Curation and Prefix-Tree Reinforcement Learning for Efficient Embodied Intelligence
Feifan Wang, Zongbing Zhang, Yu Zhang et al. (+8 more)
Training
embodied foundation models typically requires massive-scale
datasets and extensive computational resources, yet often suffers from three critical limitations: (1) inefficient sample utilization due to low-informative samples; (2) imbalanced gradient contributions across heterogeneous tasks; and (3) severe credit assignment problem in long-horizon planning, where trajectory-level rewards indiscriminately penalize all
tokens. To address these issues, we propose an efficient training paradigm that achieves state-of-the-art average performance through strategic data selection and hierarchical policy optimization. Our approach consists of three synergistic stages. First, Rejection Sampling-based Fine-Tuning (RSFT) filters out low-informative samples to establish robust behavioral priors while preventing distributional collapse. Second, Iterative Rejection GRPO (IR-GRPO) employs task-specific queues stratified by difficulty to keep
datasets balanced across reinforcement learning iterations, coupled with a hybrid reward mechanism for precise cross-task feedback. Third, to enhance long-horizon task planning, we introduce Trie-GRPO, a novel reinforcement learning
algorithm based on action prefix trees, which enables step-level advantage estimation. This resolves the credit assignment problem by isolating intermediate correct decisions from downstream errors, while effectively balancing exploration efficiency and depth compared to conventional search trees. As a result, EmbodiedMind achieves a state-of-the-art average performance of 70.02% across 18
benchmarks, and significantly outperforms other
embodied foundation models in long-horizon task planning accuracy. Our project will be released for reproducibility.
Read the full abstract ▾
All authors Feifan Wang, Zongbing Zhang, Yu Zhang, Lingfeng Wang, Yurui Zhu, Jin Deng, Mingliang Zhang, Zhengguang Gao, Yongcheng Wang, Jin Xu, Ri Yang
Instrumentation and Detectors
Proposing, never assigning: fair allocation of control-room shifts in experimental particle physics collaborations from ranked wishes under institutional quotas
Kamal Benslama
Experimental particle physics collaborations staff their detectors around the clock with members drawn from dozens of institutions. The established practice is a credit system with self-sign-up: institutions owe a quota of credit, and people take the seats they want when booking opens. This paper describes a method in which seats are instead allocated from ranked wishes under institutional quotas, the result is issued as offers that each person accepts or declines, and an independent checker recomputes every rule before anything is published: the platform proposes, never assigns, and proves that it kept the rules. The credit and quota model is inherited from current practice. Around that core the paper adds a quota rule that keeps indivisible blocks fair to institutions whose share of a period is smaller than one block, and four rules that make
fairness visible beyond the count: shares of owl, weekend and holiday blocks, seats shown in the laboratory's clock and the institution's own, reachable pools for institutions that cannot travel, and booking order by balance. On a synthetic collaboration of 48 institutions and 256 eligible members, the method allocates 70 to 72 of 72 seats per period in two rounds, with a mean rank satisfaction of 84 to 95% and no rule broken; after a year of running every institution of four members or more is within 13% of its cumulative share, and the smallest institutions, which one block overshoots by arithmetic, are brought within the same range by pooling in pairs. The method is stated completely enough to be reimplemented; a reference implementation exists and is available to collaborations under license. It applies as it stands to any scientific collaboration that staffs a facility in blocks of shifts under institutional quotas.
Read the full abstract ▾
Performance
PrefixBench-H100: Characterizing Prefix Reuse and Time-to-First-Token in H100 LLM Serving
Omkar Shewale, Deepak Kumar and Divakar Kumar Yadav
Repeated
prompt prefixes are increasingly common in LLM serving workloads, appearing in system
prompts, templated
retrieval-augmented generation pipelines,
agent frameworks, and multi-turn conversations. Modern inference runtimes such as vLLM and TensorRT-LLM provide mechanisms for reusing previously computed KV-cache state across requests, yet it remains unclear when prefix reuse materially improves serving performance on contemporary accelerators and when its benefits are limited by scheduling, cache granularity, concurrency, or
memory pressure. This paper presents PrefixBench-H100, a reproducible
benchmark and measurement framework for characterizing prefix reuse on a single NVIDIA H100. PrefixBench-H100 combines controlled synthetic traces with chat-style and
retrieval-style workloads, and evaluates two widely used LLM serving runtimes under matched workload conditions. The
benchmark varies shared-prefix length, suffix diversity, request arrival pattern, concurrency, output length, and cache configuration, while collecting time-to-first-
token, inter-
token latency, end-to-end latency, throughput, cache-hit statistics, GPU
memory usage, and selected profiling traces. The goal of PrefixBench-H100 is not to introduce a new caching
algorithm, but to expose the practical operating envelope of prefix reuse for H100-class LLM serving. The study identifies the regime where prefix reuse provides substantial first-
token latency reductions and the regime where cache pressure erodes them, while showing that cache effectiveness itself is largely insensitive to concurrency and output length; the cross-runtime differences that remain arise above the cache, in the scheduling layer.
Read the full abstract ▾
Information Retrieval
Self-Evolving Search Index
Sangam Lee, Wonjae Lee, Sunghwan Kim et al. (+5 more)
Information
retrieval is increasingly important as LLM
agents tackle complex tasks involving diverse information needs. Because
retrieval relies on an index that represents each document through index keys,
retrieval quality depends heavily on how effectively these keys expose the knowledge contained in each document. However, effective index representations vary across
retrieval environments, making it difficult for any fixed optimization strategy to perform consistently. Yet evolving an index to its
retrieval environment remains largely human-driven, requiring humans to diagnose
retrieval failures, refine the optimization strategy, and reprocess the index accordingly. We propose SELF-INDEX, a framework that enables an index to self-evolve without human intervention. Its Optimizer autonomously diagnoses
retrieval shortfalls, selectively revises the responsible index keys, and validates each revision before updating the index. Beyond reacting to observed
retrieval demands, SELF-INDEX proactively explores additional demands through a Query Simulator, allowing the index to evolve beyond the queries already available for optimization. Across diverse corpora and retrievers, SELF-INDEX consistently improves
retrieval performance while outperforming existing index optimization methods. We further show that these benefits extend to downstream applications, improving the effectiveness and efficiency of search
agents and helping
agent memory systems retrieve useful past interactions.
Read the full abstract ▾
All authors Sangam Lee, Wonjae Lee, Sunghwan Kim, Deogyong Kim, Jaehoon Kim, Daye Nam, SeongKu Kang, Dongha Lee
Combinatorics
Maximum and Minimum Spectral Radii in an Exceptional Family for Edge-Disjoint Spanning Trees
Xuanyu Cao and Chunxiang Wang
For a connected graph $G$, let $\tau(G)$ denote the maximum number of pairwise edge-disjoint spanning trees, and let $\rho(G)$ be its adjacency spectral radius. For integers $n\ge1$, $k\ge2$, and $k\le\delta\le2k-1$, let $\mathcal{G}_{n,\delta}$ be the class of connected $n$-vertex graphs with minimum degree $\delta$, and let $\mathcal{L}_{\mathcal{H}}^{2}(n,k,\delta)\subseteq\mathcal{G}_{n,\delta}$ be the exceptional family introduced by Chang, Li, and Zhang. For sufficiently large $n$, the sharp adjacency-spectral threshold for $\tau(G)\ge k$ is determined by maximizing $\rho(G)$ over this family. Set $h=\delta-k$. For each fixed admissible pair $(k,\delta)$ and all sufficiently large $n$, we determine the maximum and minimum adjacency spectral radii in $\mathcal{L}_{\mathcal{H}}^{2}(n,k,\delta)$. In the core–placement representation, let $M$ be the missing-edge graph of the bounded core. For $h\ge2$, the unique maximizer, up to isomorphism, satisfies $M\cong K_{1,h}\cup(h+1)K_1$, with the exceptional edges nested on the large-clique side. Hence the matching configuration proposed in
Conjecture~2 of Chang–Li–Zhang is not extremal. For every $h\ge1$, the unique minimizer satisfies $M\cong hK_2\cup2K_1$, with the $2h$ exceptional edges having distinct endpoints on the large-clique side. We also settle the cases $h=0,1$ and all equality cases. The proof combines an exact core–placement parametrization with a uniform Schur-complement resolvent expansion. The first candidate-dependent coefficient is affine in $\sum_{x\in V(M)}d_M(x)^2$, while the first placement-sensitive coefficient is a squared-load functional. Equitable quotient
matrices yield the two extremal radii, and the maximizing graph gives the sharp global adjacency-spectral threshold.
Read the full abstract ▾
Artificial Intelligence
Replan, Repair, or Edit? A Unified Empirical Evaluation of Travel Agents for Itinerary Revision under Resource Disruptions
Xiaofei Yuan, Yan Zhang, Shaobo Qiao et al. (+7 more)
Travel-planning
agents generate itineraries that may become infeasible after acceptance because of flight cancellations, hotel unavailability, or attraction closures. Revising these itineraries involves full replanning, classical plan repair, and LLM-based travel-
agent revision, whose differing task formulations and evaluation protocols hinder comparison. We conduct a systematic empirical study using two TREK-derived
benchmark sets: 500 single-disruption cases, including feasible and infeasible instances, and 200 feasible simultaneous compound-disruption cases. We compare LLM-Z3 full replanning, IPyHOPPER hierarchical repair, and an iTIMO local-revision adapter across effectiveness, plan stability, and computational cost. LLM-Z3 with Gemini achieved the highest observed compound-disruption success. IPyHOPPER nearly matched that configuration's single-disruption overall success, while preserving substantially more of the accepted itinerary on successful repairs. Successful hierarchical and local repairs made fewer edits and retained more accepted commitments than full replanning. Computational profiles differed: IPyHOPPER used no LLM inference, the evaluated LLM-Z3 adapter used compact one-call inference, and the iTIMO adapter consumed substantially more
tokens. The study provides practical guidelines for balancing feasibility recovery, commitment preservation, and computational cost within evaluated settings.
Read the full abstract ▾
All authors Xiaofei Yuan, Yan Zhang, Shaobo Qiao, Huangleshuai He, Leyan Ni, Mingchen Ju, Lujia Yang, Sijia Xu, Yifu Tang, Zhengyi Yang
Solar and Stellar Astrophysics
The Contribution of Starspots to the Lithium Spread in NGC 2264
Yan Luo, Gregory J. Herczeg, Min Fang et al. (+2 more)
Star clusters with ages of a few Myr have a previously unexplained dispersion in the lithium equivalent widths. In this work, we show that much of the lithium spread of the young cluster NGC 2264 can be attributed to the observational presence of starspots. We demonstrate that the equivalent width of Li I 6708 Angstrom changes with observed spot coverage, using a sample of 15 Weak-Lined T Tauri
Stars (WTTSs) with multi-epoch high-resolution ESPaDOnS spectra and photometry. To quantify the contribution of starspots, we construct an empirical two-component starspot model. When differences in visible spot coverage are taken into account, 68% of the WTTSs fall within the model-predicted EW(Li) dispersion for a range from 0-90% visible spot coverage. We find no correlation between the age spread or
gravity and Li I equivalent width in NGC 2264. Our results suggest that invoking starspots can explain much of the apparent lithium spread in NGC 2264 and reduces but does not fully eliminate the need to invoke rotation-driven lithium depletion or other non-standard explanations at such young ages. Future simultaneous measurements of lithium and beryllium in young
stars may provide an independent test of this scenario.
Read the full abstract ▾
All authors Yan Luo, Gregory J. Herczeg, Min Fang, Qinghui Sun, Xiaoting Fu
High Energy Physics (Phenomenology)
Strange Two-Photon Exchange in Time-Like Electromagnetic Form Factors
Xu Cao, Yong-Hui Lin and Hai-Qing Zhou
While two-
photon exchange (TPE) effects in space-like nucleon electromagnetic form factors have been extensively studied for over two decades, their manifestations in the time-like region have yet to be definitively established. We propose the isospin-violating decay $J/\psi \to \bar{\Sigma}^{0}\Lambda$ and its charge-conjugate process as a sensitive probe of TPE contributions to time-like hyperon electromagnetic form factors. This channel provides a clean laboratory for isolating TPE effects through sensitive and robust polarization and
spin-correlation observables, which are particularly enhanced by the $\Lambda$-$\Sigma^0$ mass splitting. Combining a dispersion-relation analysis of the one-
photon-exchange contribution with explicit loop calculations of the TPE amplitudes, we perform a quantitative study of TPE effects in the threshold region. Our results demonstrate that the considered observables can provide unambiguous signatures of TPE contributions and present predictions for their size together with the associated theoretical uncertainties.
Read the full abstract ▾