Peer Revenue Past citations are no guarantee of future returns

Papers

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

All of arXiv

Wednesday 30 September

Artificial Intelligence

Learning Steganography Is Easy, Learning Steganographic Reasoning Is Hard

Julian Schulz, Lukas Fülle and Rieke Fruengel

Chain-of-thought monitoring as an approach for AI oversight and control is threatened by the possibility of steganographic reasoning, where LLMs conceal their reasoning inside innocuous-looking text. Two neighbouring capabilities, steganographic messaging (passing a concealed message) and encoded reasoning (reasoning in an illegible but unconcealed format), have already been shown to emerge under training pressures that occur in real pipelines, such as reinforcement learning against monitors. This suggests that steganographic reasoning too might arise as an unintended side effect of training. Here, we compare how easily models learn steganographic reasoning and these two neighbouring capabilities across three elicitation methods: reinforcement learning, in-context learning, and supervised fine-tuning (SFT). For most tasks, models learn steganographic reasoning only under SFT, while they learn steganographic messaging and encoded reasoning under all three elicitation methods. Even under SFT, steganographic reasoning requires at least twice as much training as messaging, and for several model-task combinations it is not learned at all. However, on a cover task that makes hiding information especially convenient, steganographic reasoning can be successfully learned under all three elicitation methods. Steganographic reasoning is thus much harder than steganographic messaging and encoded reasoning, and learning the latter two does not imply learning the former. Yet it lies within reach: an easy version is learned under every elicitation method, when the cover task is convenient for hiding information.
Read the full abstract

Machine Learning

Fast Regularized Policy Mirror Descent with One-Step TD Updates

Qipei Chen, Wenye Li, Yule Sun et al. (+1 more)

Policy mirror descent (PMD) enjoys fast convergence in regularized Markov decision processes (MDPs), but existing guarantees often rely on exact or increasingly accurate policy evaluation. We analyze PMD coupled with a persistent critic advanced by one temporal-difference (TD) update. For finite discounted MDPs, we establish global linear convergence in value for exact coordinate-wise Bellman updates, with any positive constant actor stepsize and arbitrary finite critic initialization. The proof combines a resolvent-based auxiliary distribution with a decaying Bellman-violation correction and a potential weighted by inverse coordinate weights. We then study stochastic TD-PMD with general strongly convex mirror maps under a single off-policy Markov trajectory. With suitably chosen constant stepsizes and a finite-batch TD update, the method achieves an expected value gap of $\epsilon$ after $\widetilde{O}(1/((1-\gamma)^5 \widetilde{\sigma}_b \epsilon))$ transitions. The stochastic analysis relies on the trajectory-wise Lipschitz continuity of the regularizer, derived from uniform bounds on vertex Bregman divergences, together with a visitation-weighted resolvent estimate for signed critic-error propagation that yields an inverse-linear dependence on behavior coverage $\widetilde{\sigma}_b$. In contrast to many prior guarantees for regularized policy optimization, our sample-complexity guarantee holds without trajectory resets, generative-model access, or nested policy-evaluation loops. Numerical results are consistent with the theoretical convergence analysis.
Read the full abstract

All authors Qipei Chen, Wenye Li, Yule Sun, Ke Wei

Computer Vision

Spherical Interpolation for Backward-Compatible Multimodal Representations

Simone Ricci, Niccolò Biondi and Federico Pernici

Contrastive vision-language models map visual and textual representations into a shared normalized embedding space, making cosine similarity the natural metric for cross-modal retrieval. A practical challenge arises during model upgrades: independently trained models generally produce incompatible representation spaces, so replacing a deployed model typically requires recomputing embeddings for the entire gallery, which is prohibitively expensive at scale. Orthogonal post-hoc alignment can partially mitigate this problem by mapping new-model queries into the old-model gallery space. However, because independently trained models can differ in fine-grained representation structure, the orthogonal alignment remains approximate, leaving a residual angular discrepancy between the old-model query and the aligned new-model query. We study whether interpolation along the spherical geodesic between these two normalized query representations can improve retrieval without re-indexing the gallery. We characterize when this path contains an interior query direction closer to an idealized retrieval-optimal direction than either endpoint, and connect this characterization to Recall@$K$ through a local margin-based certification result. Experiments across multiple benchmarks and model families show that post-alignment spherical interpolation improves over orthogonal alignment alone, recovering backward-compatibility in most evaluated settings. Consistent with our geometric characterization, per-query oracle analysis shows that retrieval-favorable interior points occur frequently in practice. Code is available at https://github.com/miccunifi/SLERP_backward_compatibility .
Read the full abstract

Quantum Physics

A self-tallying quantum anonymous voting protocol for multiple-selection elections

Guang-Bao Xu, Meng-Na Li, Yu-Guang Yang et al. (+1 more)

Self-tallying quantum anonymous voting (SQAV) has drawn considerable attention since it was proposed. However, constrained by design challenges, all existing protocols only accommodate single-selection voting and cannot support multi-selection voting. In this paper, we propose a SQAV protocol with multi-selection voting functionality. In our protocol, $nm$ $n$-particle entangled states are equally divided into $n$ groups, and the $n$ particles in each entangled state are delivered to $n$ voters, one particle per voter. Each voter obtains $n$ voting vectors by sequentially measuring all his (or her) particles that belong to $n$ different group, and then encodes his or her voting information into the vector corresponding to the secret index. Due to entanglement correlation, each voter can verify whether his or her ballot is correct through computation, while any participant can obtain the vote count for each candidate through computation. Our protocol satisfies self-tallying, nonreusability, verifiability, and fairness.
Read the full abstract

All authors Guang-Bao Xu, Meng-Na Li, Yu-Guang Yang, Dong-Huan Jiang

Quantum Physics

All-Microwave Multiqubit Gates

Guanqi Wang, Yao Song, Peng-Bo Li et al. (+1 more)

Implementing high-fidelity multiqubit gates is critical for reducing circuit depth in near-term quantum processors and fault-tolerant architectures. However, realizing multiqubit interactions and entanglement remains a critical challenge that limits the implementation of multiqubit gates. Here we propose an all-microwave scheme to realize single-step multiqubit gates incorporating n control and m target qubits, tailored for frequency-tunable superconducting transmon networks. By leveraging cross-resonance (CR) drives, this approach induces effective two-body ZX interactions that are significantly stronger than those achieved in conventional resonant regimes, circumventing the need for tunable couplers. The system's effective Hamiltonian is analytically derived using both quad frame rotation and rigorous block diagonalization. These combined theoretical methods facilitate the identification of optimal parameter regimes that simultaneously enhance desired target couplings and suppress parasitic higher-order terms, such as ZXX interactions. Through numerical simulations and gradient-based pulse-shape optimization, we demonstrate three-qubit gates achieving fidelities exceeding 99.9% within a 60 ns gate duration. Furthermore, we evaluate the fundamental scalability of this framework by extending the optimization to a five-qubit CXXXX architecture, confirming the physical viability of the multi-target driving scheme. Because these multiqubit operations naturally emulate stabilizer measurements, this architecture provides a hardware-efficient strategy for reducing the circuit depth of stabilizer-type operations by replacing sequential two-qubit gate decompositions with single-step multiqubit gates.
Read the full abstract

All authors Guanqi Wang, Yao Song, Peng-Bo Li, Xiu-Hao Deng

Machine Learning

RainAtlas: A Multi-Continental Dataset for Precipitation Downscaling

Pierre-Louis Lemaire, Luca Schmidt, Wietze Suijker et al. (+2 more)

Extreme rainfall events are increasing in intensity and frequency as climate change accelerates. While kilometer-scale precipitation forecasts are critical for supporting local decision-making, the limited availability of high-resolution precipitation observations hinders their accuracy, especially in under-resourced regions. Machine learning models are widely used to downscale precipitation data to km-scale, but their application to unseen geographies presents challenges. First, processing raw high-resolution precipitation datasets across regions requires significant engineering and domain expertise. Second, generalization across regions remains difficult. To help overcome these barriers, we release RainAtlas, a large-scale, ML-ready and multi-continental dataset for precipitation downscaling. Covering three continents, RainAtlas harmonizes heterogeneous hourly km-scale observations to a common 2-km grid. Each regional partition contains around 210,000 aligned low- and high-resolution precipitation pairs, respectively from ERA5 reanalysis and direct observations. We benchmark state-of-the-art ML-based downscaling models across RainAtlas using a wide range of metrics. Our evaluation reveals substantial variance in out-of-domain generalization depending on the training regions. This underscores the need for cross-regional, multi-source km-scale evaluation, establishing RainAtlas as a well-positioned benchmark for precipitation downscaling research.
Read the full abstract

All authors Pierre-Louis Lemaire, Luca Schmidt, Wietze Suijker, Alex Hernandez-Garcia, David Rolnick

Computer Vision

P-SRM: Selective Recovery of Rejected Predictions in Visual Tracking

Youbin He and Siwei Wang

Many visual tracking methods use rejection mechanisms to suppress unreliable predictions. However, these mechanisms can also reject correctly localized candidates, leaving useful information unused. We investigate how to identify and recover these candidates while preserving native accepted outputs and candidate coordinates. To this end, we propose P-SRM (Post-rejection Selective Recovery Method), which combines spatial responses, past accepted states, and native decision margins to reassess candidates and selectively restore reliable predictions. We evaluate P-SRM on six trackers and four datasets spanning category-specific, point, and generic object tracking. Across all nine configurations, P-SRM improves rejected-candidate ranking and overall tracking performance. These results show that post-rejection verification can identify and recover useful predictions discarded by native rejection, demonstrating the value of reusing rejected information. Project repository: https://github.com/PalestyHR/P-SRM.
Read the full abstract

Quantum Physics

Dynamic distributed quantum sensing of radio-frequency fields via time-bin entanglement

Vedansh Nehra, Richard Birrittella, Benjamin Malia et al. (+6 more)

We propose a novel framework for discrete-variable (DV) distributed quantum sensing, wherein traditional polarization-based probes are replaced with time-bin entangled qubits. We show that the intrinsic temporal structure of time-bin Bell states can be utilized as a tunable, built-in two-time differential sampling reference for sensing radio-frequency (RF) fields. Unlike existing protocols based on polarization Bell states, which are limited to static measurements, our proposed architecture enables the coherent mapping of dynamic RF phase signals onto static quantum optical phases. By matching the time-bin separation to the RF or intermediate frequency (IF) half-period (antipodal sampling), we theoretically demonstrate that our protocol provides a 6 dB sensitivity enhancement over one-sample static polarization encodings while preserving the 3 dB per-pair entanglement advantage over the separable standard quantum limit benchmark. The performance enhancement is validated with Monte Carlo simulations. This approach provides a solution for RF sensing in DV photonic quantum networks, offering a clear path toward quantum-enhanced and frequency-agile distributed sensor arrays and broader utilization of time-domain quantum probe encodings.
Read the full abstract

All authors Vedansh Nehra, Richard Birrittella, Benjamin Malia, Nicholas J. Barton, Christopher Tison, James Schneeloch, David Hucul, Benjamin Kyle, Erin Sheridan

Genomics

MerKurio: sequence extraction and annotation based on matching k-mers

Lukas Schönmann, Heinz Himmelbauer and Juliane C. Dohm

K-mers, short subsequences of length k, play an important role in computational analyses using biological sequence data. A critical downstream processing step involves getting back to the source sequences of selected k-mers for further analysis or validation. Here, we present MerKurio, a high-performance command line tool written in Rust with two main functionalities: 1) extracting sequence records from FASTA/FASTQ files using k-mers and 2) annotating or filtering aligned sequences in SAM/BAM files with k-mer tags. The tool automatically selects the pattern matching algorithm via empirically derived rules based on query characteristics and supports paired-end reads, compressed input files, and reverse complement searching. Benchmark comparisons demonstrate that MerKurio outperforms existing tools in terms of speed while providing additional utility including detailed matching statistics and comprehensive file format support. MerKurio is a fast and user-friendly tool for sequence record extraction and tagging of aligned sequences based on k-mers. MerKurio is freely available at https://github.com/lschoenm/MerKurio.
Read the full abstract

Machine Learning (Statistics)

Estimation of the Label-Noise Transition Matrix with Performance Guarantees via Selective Classification

Xabier de Juan, Santiago Mazuelas, Yilun Zhu et al. (+1 more)

Modern machine learning depends heavily on massive datasets, but obtaining high-quality annotations at scale is often expensive. As a result, learning from noisily-labeled data has become common, making accurate estimation of the label-noise transition matrix crucial. However, existing transition matrix estimators rely on the fragile estimation of class-posteriors and do not provide finite-sample performance guarantees. In this work, we propose a novel methodology to estimate the transition matrix based on one-sided selective classification. This approach bypasses class-posterior estimation, provides finite-sample performance guarantees, and leverages flexible learning methods for binary classification. Moreover, we introduce effective algorithms to implement the proposed methodology and provide their refined finite-sample performance bounds.
Read the full abstract

All authors Xabier de Juan, Santiago Mazuelas, Yilun Zhu, Clayton Scott

Information Retrieval

KUAISHOU Explorer LLM-Rec Challenge 2026: Reasoning Generative Recommendation

Jiangxia Cao, Hao Peng, Wenlong Xu et al. (+97 more)

Generative recommendation, has been attracted a surge of attentions in industrial and academic research community, towards to build more smart system to build next-generation recommender. Under the significant developing wave of large language model, our team have been developed Semantic ID based OneRec/OneRec-V2. These models have been widely deployed in production and demonstrate the scaling potential of the autoregressive next-item prediction paradigm for industrial recommender systems. Building on the success of OneRec, we further explored a series of models, including OneRec-Think, OpenOneRec, and OneReason, that connect item Semantic IDs with natural language in a unified representation space and seek to unlock the potential of natural-language chain-of-thought (CoT) reasoning for recommendation. However, our preliminary works found that introducing reasoning CoT does not always improve the recommendation performance. To address this issue, OneReason strengthens the semantic alignment between items and language, introduces structured template-based supervision for interest reasoning, and applies advanced reinforcement learning techniques to make reasoning more beneficial to recommendation. As a frontier topic to building recommendation foundation models, we believe this topic has significant research value and hope to encourage more researchers to explore it together. To this end, together with the SIGIR 2026 community, we organized the KUAISHOU Explorer LLM-Rec Challenge 2026: Reasoning Generative Recommendation.
Read the full abstract

All authors Jiangxia Cao, Hao Peng, Wenlong Xu, Jiaxin Deng, Zhixin Ling, Xingmei Wang, Kun Shang, Can Tang, Zhihuai Cai, Jun Du, Fang Su, Xiaojuan Liu, Yiling Li, Chenglong Yu, Chongling Rao, Haixuan Gao, Haitao Xu, Jian Liang, Ruiming Tang, Chenglong Chu, Guohong Mu, Honghui Bao, Hui Wang, Jialong Chen, Jiao Ou, Muhao Wei, Peng Zhang, Renpu Liu, Ruochen Yang, Shugui Liu, Xinqi Jin, Yan Sun, Yifan Wang, Yingzhi He, Yufei Ye, Yusen Huo, Tingkuo Wang, Jihong Zhang, Lanxi Zhu, Pengyuan Liu, Zhipeng Yi, Luankang Zhang, Hang Lv, Xuyang Zhi, Tianyu Li, Bintao Wu, Chuang Ou, Siyue Su, Ziyuan Wang, Yuliang Sun, Baiyan Che, Feiyang Xu, Shiwen Zhang, Shiteng Cao, Chongcong Jiang, Yuan Fang, Xiangwu Yang, Hao Deng, Zijian Du, Pengxun Wang, Xiaoming Wang, Shun Qin, Yingqi Song, Tianyi Li, Naixiao Peng, Chenyu Zhou, Qiliang Jiang, Quan Zheng, Cheng Jin, Siying Zeng, Hongjia Xu, Junwu Hu, Teng Fu, Zhengkang Mei, Haijun Yu, Kai Li, Shengyang Zhou, Zhijia Wei, Siyi Xiong, Bo Liu, Zichun Guo, Zhubin Han, Jinpeng Fu, Bingqian Liu, Yuyi Wang, Yu Liu, Qinghai Tan, Ruijie Zhou, Zhuohang Li, Zhijia Zhong, Xiangnan He, Jirong Wen, Min Zhang, Wenwu Ou, Peng Jiang, Han Li, Kaiqiao Zhan, Yanan Niu, Lantao Hu, Kun Gai

Computation and Language

Synthetic Pre-pretraining Survives Scale, but Not as a Grammatical Prior

Atsuki Yamaguchi, Tatsuro Inaba, Joel Niklaus et al. (+3 more)

Pre-pretraining (PPT) on synthetic non-natural language data improves token efficiency during language model pre-training (PT). Prior work attributes this gain to a grammatical prior, i.e., a structural inductive bias learned during PPT that transfers to natural language grammar. However, PPT has only been tested on models of at most 1B parameters and PT budgets below 2B tokens on predominantly web text. It is unknown whether PPT is effective at larger scales and under more realistic PT data mixtures that combine diverse sources (e.g., code and math). We therefore present a comprehensive study on PPT spanning five PPT tasks, four PT data mixtures, four parameter scales (500M to 7B), and PT budgets of up to 100B tokens. Our results demonstrate that the downstream performance and token efficiency gains of PPT persist at scale, e.g., saving at least 21B PT tokens at the 3B scale. However, in contrast to prior work, we find no consistent evidence that these gains stem from a grammatical prior. Downstream performance does not consistently align with grammatical acceptability across model sizes. Instead, we find that downstream gains arise from PPT tasks that improve long-range retrieval. Finally, PPT performance gains are robust to how PT data mixtures are composed and diminish only when web text is absent. Overall, PPT is a low-cost addition to PT, and future PPT task design should target long-range retrieval rather than natural language grammar.
Read the full abstract

All authors Atsuki Yamaguchi, Tatsuro Inaba, Joel Niklaus, Michal Štefánik, Aline Villavicencio, Nikolaos Aletras

Fluid Dynamics

Is Vorticity Amplification Essential for Anomalous Dissipation and Intermittency in Turbulence?

Alessandro Chiarini

The distinct dynamical roles of vorticity amplification (VA) and vortex tilting (VT) in three-dimensional homogeneous isotropic turbulence are investigated using direct numerical simulations of a modified Navier–Stokes system. By selectively suppressing VA while retaining VT, we demonstrate that gradient amplification by VA is strictly required to sustain the classical forward energy cascade. Progressively suppressing VA weakens small-scale velocity gradients, heavily depletes extreme fluctuations, and entirely eliminates the classical energy dissipative anomaly, causing the normalized energy dissipation to decay as $Re_\lambda^{-1}$. In the limit of complete VA suppression, enstrophy emerges as the relevant inviscid invariant. Scale-by-scale budget analyses confirm that the dynamics transition to a purely forward enstrophy cascade, characterized by an anomalous enstrophy dissipation and a robust $E(k) \sim k^{-3}$ intermediate energy spectrum. We verify that these asymptotic scaling limits—including a $Re_\lambda^{-1/2}$ finite-Reynolds-number correction for the enstrophy anomalous dissipation—are universal properties of the VA-suppressed dynamics, independent of whether the large-scale forcing is helical or non-helical. Remarkably, despite the smoothing of the velocity field and the elimination of the energy dissipative anomaly, anomalous structural scaling and a broad multifractal spectrum persist. These results reveal a fundamental mechanistic separation: while VA is responsible for amplifying intense localized fluctuations, geometric reorganization by VT alone is mechanically sufficient to sustain multifractal intermittency.
Read the full abstract

Classical Analysis and ODEs

The Pearl ensemble and the logarithmic energy on the Sphere

Bianca Gariboldi and Giacomo Gigante

We define a family of random sets of points on the sphere $\mathbb S^2$, the Pearl ensemble, depending on several parameters. The expected value of the logarithmic energy can be computed exactly up to terms of order $O(\sqrt{N}\log N)$, with $N$ the number of points. For properly chosen values of the parameters, the coefficient of the linear term in the expansion of the logarithmic energy can be taken as close as desired to the value $(1-\log 3)/2\sim -0.0493061\ldots$. Among the explicit constructions currently known to us for which the logarithmic energy expansion has been rigorously determined up to the linear term, the Pearl ensemble yields the smallest coefficient.
Read the full abstract

Logic

Continuous graph homomorphisms of higher dimensional abelian group actions

Ruijun Wang

For every fixed integer $d\geq2$, we prove that the finite graphs receiving a continuous homomorphism from the standard Schreier graph $F(2^{\mathbb Z^d})$ form a $\Sigma^0_1$-complete set. This extends a theorem of Gao, Jackson, Krohne and Seward when $d=2$. For the positive part of the reduction, we show that if a graph $H$ satisfies a certain property then there is a continuous graph homomorphism from $F(2^{\mathbb Z^d})$ to $H$, in particular, the complete graph $K_4$ satisfies this property. This extends a theorem of Gao and Jackson. We also prove that $\big (\mathcal H_d\big )$ is a strictly decreasing sequence.
Read the full abstract

Quantum Physics

Improved Quantum Random Self-Reduction for Linear Problems

Vahid R. Asadi, Shuichi Hirahara and Nobutaka Shimizu

We study quantum random self-reductions for linear problems over finite fields. Let $M\in\mathbb{F}^{n\times n}$ be an arbitrary matrix, and let $\mathcal{O}$ be an oracle that agrees with the linear map $x\mapsto Mx$ on an $\varepsilon$-fraction of inputs $x\sim\mathbb{F}^n$. Given coherent access to $\mathcal{O}$ and coherent entry access to $M$, we give a uniform quantum reduction that computes $Mx$ on any prescribed input $x$ with probability at least $2/3$ in time $\widetilde{O}(nT^{1/3})$, for $n\le T\le n^{3/2}$ and constant field size and $\varepsilon$, where $T$ is the cost of one coherent query to $\mathcal{O}$. In particular, when $T=\widetilde{O}(n)$, the reduction runs in time $\widetilde{O}(n^{4/3})$, improving the $\widetilde{O}(n^{3/2}+T)$ reduction of Asadi, Golovnev, Gur, Shinkar, and Subramanian (SODA 2024). Our reduction uses the Bogolyubov–Ruzsa subspace guaranteed by additive combinatorics, but it avoids learning this subspace explicitly, which was computationally expensive for the previous reduction; in particular, it does not recover a basis for its orthogonal complement. The main technical step is to decompose the inputs into sparse pieces and find a vector that lies outside the Bogolyubov–Ruzsa subspace via a quantum search based on amplitude amplification. This yields a tunable tradeoff between the cost of querying the average-case oracle and the cost of verifying matrix-vector products.
Read the full abstract

Robotics

Toward Real-Time VLAs: Stage-Aware Two-Step Flow Denoising and System-Level Evaluation

Di Wu, Rongtian Shen, Ping Liu et al. (+8 more)

Vision-language-action (VLA) models face a timing gap between low-rate inference and high-rate robot execution. We characterize this gap through end-to-end latency measurements of model inference and the robot execution chain. Repeated Flow Matching denoising contributes substantially to inference cost, while robot-side delays mainly arise from perception acquisition, communication scheduling, and physical response. Analysis of the velocity field shows relatively stable magnitude and direction in early integration, followed by stronger directional correction near the terminal steps. Based on this stage heterogeneity, we propose two-stage non-uniform denoising, reducing the number of steps from 10 to 2 and model-inference time from 61.557 ms to 21.956 ms. We also develop a distributed real-time VLA framework with independent inference, action-publication, and robot-control rates, modular observation acquisition, and action-provenance logging. Using {\pi}0.5 as the baseline, we evaluate six real-time execution methods on a long-horizon physical garment-folding task. Legato performs best overall among training-based methods, while Temporal Smoothing leads among training-free methods; both perform strongly in task success, completion time, action continuity, and acceleration smoothness. Combining two-step denoising with representative execution methods substantially reduces inference cost with a small reduction in task performance. These results motivate joint optimization of model-inference efficiency and robot-system timing.
Read the full abstract

All authors Di Wu, Rongtian Shen, Ping Liu, Yan Shen, Zhenhan Yin, Shun Zuo, Xuhua Chen, He Zheng, Lingfeng Zhang, Jianglin Zhang, Tao Zhang

Quantum Physics

Learning Random Quantum Circuits and the Emergence of Pseudorandomness

Srinivasan Arunachalam, Qizhao Huang and Makrand Sinha

We give an efficient algorithm for learning $k$-dimensional brickwork random quantum circuits using only copies of the output state obtained by applying $U$ to the all-zero input. For a depth-$d$ circuit on $n$ sites with random $2\ell$-qubit gates, the algorithm learns the original circuit $U$ with high probability in $\text{poly}(n,2^{\ell d})$ time for every constant dimensional lattice. In particular, the algorithm is polynomial time as long as $\ell d = O(\log n)$. In one dimension, this reaches the natural boundary suggested by pseudorandomness: pseudorandom states require $\ell d=\omega(\log n)$, and structured cryptographic constructions suggest that this scale may be achievable from above. In higher dimensions, it remains plausible that the same $\ell d=\omega(\log n)$ scale remains the threshold for ancilla-free pseudorandomness, and our results help clarify the conditions under which pseudorandomness can arise in this setting. The main idea behind our algorithm is a local correlation criterion that identifies gates in the final layer without learning their entire backward light cones, avoiding a bottleneck in the previous approaches. A key technical ingredient is a Carbery-Wright type anticoncentration inequality for low-degree polynomials of Haar random unitaries whose small-ball exponent is independent of the matrix dimension. The dimension-independent exponent is crucial for handling gates of growing locality. This anticoncentration result may also be of independent interest.
Read the full abstract

Robotics

Learning from Runtime Feedback through Failure-Bank Self-Evolution for Vision-Language-Action Models

Mingyue Cui, Zheyuan Liu, Yihan Zhu et al. (+2 more)

Vision-language-action (VLA) models generalize broadly across robotic manipulation tasks, but complex environments require balancing task success with unintended contact. Runtime shields can correct individual actions, but they leave the underlying policy unchanged, so repeated disagreements may create a persistent policy-shield mismatch that blocks task progress. To address this challenge, we introduce FailBank, a four-stage self-evolving framework that converts runtime feedback into persistent policy improvement. During collection, a fixed CBF-based safety module serves as an observe-only teacher, producing counterfactual corrections while the policy remains in control. Outcome-aware admission then converts useful proposals into corrective targets and retains successful uncorrected actions as quiet anchors for guarded LoRA updates. We evaluate FailBank on the VLA-Arena benchmark across two difficulty levels and two VLA backbones. Compared with the base policies, FailBank improves the joint success-cost operating point. Across the two backbones, FailBank improves task success rate by 8.5 and 6.9 percentage points, while reducing policy-induced cumulative cost by 35.6% and 23.8%, respectively. Compared with runtime shielding, FailBank raises task success rate by 25.4 and 9.5 percentage points, while maintaining comparable policy-induced cumulative cost. These results show that runtime feedback can serve as persistent policy supervision rather than only as a temporary action constraint.
Read the full abstract

All authors Mingyue Cui, Zheyuan Liu, Yihan Zhu, Zheyuan Zhang, Meng Jiang

Operating Systems

Capture the lifecycle: KV Cache management in ReAct Agents with KVTether

Kaihua Fu, Yukun Zhou, Chaokun Chang et al. (+6 more)

Efficient serving of long-context reasoning-and-acting (ReAct) agents relies on KV cache reuse to reduce large language model (LLM) prefill latency and monetary cost. However, a semantic gap exists between agent harnesses and the underlying serving stack. Through context mutation, tool execution, and subagent coordination, context messages may become actively engaged, permanently discarded, and temporarily unused, while the serving stack only observes accesses to the corresponding KV cache. This lifecycle blindness prevents recency-only policies such as LRU from reclaiming dead KV promptly and from preserving older KV that will be reused sooner than newer entries. We present KVTether, a lifecycle-aware KV cache management framework for ReAct agents. By tracing semantic primitives embedded in agent harnesses, KVTether captures runtime lifecycle semantics during highly dynamic execution. KVTether then translates message-level semantics into KV-level lifecycle states and uses these states to drive state-prioritized cache management without exposing physical complexities to agent harnesses. After reclaiming dead KV, KVTether preferentially preserves live-but-idle KV that is waiting for reuse, reducing premature eviction before reuse. Across agent benchmarks and production workloads, KVTether reduces end-to-end request latency by up to 26.3% and 17.4% relative to LMCache and MORI, respectively, and lowers estimated task cost by 40.0% and 33.2% on average.
Read the full abstract

All authors Kaihua Fu, Yukun Zhou, Chaokun Chang, Yinghao Yu, Luping Wang, Guodong Yang, Jiuchen Shi, Quan Chen, Wei Wang

Machine Learning

Should I stay or should I show? Learning to selectively disclose information

Carlotta Giacchetta, Alessando Bogani, Cesare Barbera et al. (+4 more)

In many high-stakes settings, human decision-makers can acquire support information before making a decision. However, acquiring information is costly, and disclosure may fail to improve human decisions or may even impair them. We tackle this problem by studying selective disclosure, i.e., the problem of learning when to reveal support information to a human decision-maker under a budget constraint. We first show that the optimal policy is a threshold rule on the Value of Information (VoI), i.e., the expected reduction in human decision risk induced by disclosure. Since VoI is unknown in practice, we estimate the regime-specific human risks and bound the possible degradation of the resulting plug-in policy relative to lack of disclosure, as well as its regret relative to the optimal policy. Experiments on benchmark datasets show that selective disclosure outperforms both no disclosure and full disclosure, regardless of whether the support information is beneficial or harmful. Two user studies show that human-AI team performance can improve when disclosure is led by our learned policy and not human-selected, although this advantage varies across tasks. A counterfactual benchmark, which replaces participants' predictions with a machine-learning prediction when disclosure occurs, suggests that these differences might depend on lower adherence to advice when the information is automatically provided rather than self-requested.
Read the full abstract

All authors Carlotta Giacchetta, Alessando Bogani, Cesare Barbera, Giovanni De Toni, Michele Caprio, Andrea Pugnana, Andrea Passerini

Materials Science

Phonon-mediated closing of topological Floquet gaps in graphene: Non-phenomenological analysis

Royi Ledermann, Rave Hanoch Saadon, Adam Herling et al. (+1 more)

Floquet band engineering has been intensively studied for its potential to control material properties via laser driving. In particular, Floquet topological states have been measured on the surface of Bi$_2$Se$_3$. Nonetheless, the original prediction of Floquet topological bands in graphene remains unobserved, with works only measuring non-topological Floquet-Bloch states or indirect features. Here, we theoretically explore graphene irradiated by circular lasers with non-phenomenological electron-phonon (e-ph) coupling. We show that e-ph interactions substantially broaden Floquet bands due to graphene's large zero-point motion, as expected from phenomenological treatments. However, we further find that larger displacement $\Gamma$ optical phonons reduce Floquet gaps by about half, even in absence of instrumentation broadening, and can also induce additional features like flat bands. These phonons "enter the gap" and blur photoemission signatures. Including non-$\Gamma$ phonons closes the gap and greatly reduces visibility. When these modes couple to reasonably expected instrumentation broadening, the effect is significantly exacerbated. Our results shed light on the missing Floquet topological gaps and also lead to possible mitigation strategies: (i) Pre-pumping coherent phonons to counteract blurring, or (ii), transition to a Dirac system with more favorable phonon statistics. Moreover, our real-space analysis proposes that Floquet topological physics is alive in localized regions, such that properties of interest like transport should be accessible. We expect this work to impact experimental analysis and lead to set-ups where graphene Floquet topology might finally be directly observed.
Read the full abstract

All authors Royi Ledermann, Rave Hanoch Saadon, Adam Herling, Ofer Neufeld

Machine Learning

Beyond Accuracy: Prefix-Invariant Realizations of Low-Precision Fast Matrix Multiplication

Shuxiao Xie, Shuyang Xie, Yuan Cao et al. (+3 more)

Fast matrix multiplication saves multiplications through exact cancellation, but rounding sums that mix token rows can leave contributions from later tokens in earlier language model outputs. This threatens prefix invariance, which multiple-choice likelihood scoring relies on: a scored likelihood must depend only on its allowed prefix. On Qwen2.5-14B-Instruct, two fast FP8 realizations repaired to ordinary-looking accuracy still change the answers chosen by likelihood on 5.83% and 10.00% of 240 OpenBookQA items when only the text after the allowed prefix is replaced with the bf16 model's own greedy continuation. Both row-local controls, the bf16 model and a deployed FP8 matrix multiplication kernel, change none. Accuracy thus does not certify prefix invariance, and the stability criteria we analyze cannot tell realizations apart: across all 512 sign variants of two-level Strassen they stay constant while teacher-forced perplexities span a 772.4$\times$ range on the same model. We therefore construct certified realizations of two-level Strassen on bounded integer codes that quantize token rows independently, then mix and cancel exactly before rescaling, using 49 block multiplications instead of 64. Our certificate guarantees bitwise equality to a prescribed row-local classical int8 operator at the same quantization specification, so every certified realization inherits its prefix invariance. Certification thus turns realization choice into a pure cost decision: which certified realization runs can no longer change a single scored likelihood.
Read the full abstract

All authors Shuxiao Xie, Shuyang Xie, Yuan Cao, Dezhi Ran, Wei Yang, Tao Xie

Algebraic Geometry

Potential isotriviality of isocrystals and proper covers

Adrian Langer and Lei Zhang

We study convergent and overconvergent isocrystals that become trivial after pullback along a proper surjective morphism. On a geometrically unibranch variety over an algebraically closed field, every such object is already trivialized by a finite étale cover. Over a perfect field, this shows that a proper cover giving geometric triviality can be replaced by a finite étale cover over the ground field, and implies finiteness of the geometric monodromy group. For proper geometrically unibranch varieties, finite geometric monodromy is also sufficient. In the overconvergent case, a dominant morphism can replace the given proper cover. A nodal curve shows why the geometrically unibranch hypothesis is needed.
Read the full abstract

Quantum Physics

Sharp Quantum Entropy Mixing Rates and the Operator Layer Cake Theorem

Alexander Stottmeister

At what rate does the von Neumann entropy of an ensemble of quantum states change under Hamiltonian evolution of its constituents? Bravyi proposed the small incremental mixing conjecture controlling the mixing rate of a binary ensemble $\{(1\!-\!p,\rho_1),(p,\rho_2)\}$ by $c\,\|{H}\|h_{2}(p)$ (with the binary entropy $h_{2}$). The proof of Bravyi's conjecture was subsequently reduced to the matrix inequality $\|[A,\log B]\|_{1}\!\leq\! c\,h_{2}(p)$ for positive trace-class operators $A\!\leq\! B$ with $\text{Tr} A\!=\!p$ and $\text{Tr} B\!=\!1$ on separable Hilbert spaces by Mariën, Audenaert, Van Acoleyen, and Verstraete, and they conjectured the optimal constant to be $c\!=\!1$. Here, I prove the latter conjecture using an exact integral representation of the commutator $[A,\log B]$ obtained from the operator layer cake theorem, due to Cheng and Liu. The result gives a sharp dimension-independent limit on how rapidly unitary evolution of one component can change the entropy of a binary quantum ensemble. It follows that mixing rates satisfy small incremental mixing with the optimal constant, and entangling rates of bipartite Hamiltonians are bounded by $(2\log d\!+\!1)\|H\|$. Moreover, the result corrects a conjecture by Lieb and Vershynina for mixing rates of general ensembles.
Read the full abstract

Machine Learning

Backward-State Policy Is Part of the Learning Algorithm

Shuxiao Xie, Shuyang Xie, Dezhi Ran et al. (+2 more)

Low-precision training rounds tensors that the backward pass reads again, often for several gradients; each use can read the forward's rounded value, the original, or a new random rounding. This backward-state policy looks like a memory and precision detail, settled by copy accuracy and final loss. We argue that it is part of the learning algorithm, and that neither check shows whether it is right. Copy accuracy does not decide the outcome: in three pairs of 390M runs with an emulated FP8 backward, training fails when attention's backward reuses the forward's rounded output and succeeds with a new rounding from the same distribution. Even the most accurate copy, the original itself, can be wrong by our reference: the gradient of the forward pass as it actually ran, with gradients passed through rounding unchanged. For example, a normalization output stored in low precision feeds two gradients: the gain's gradient needs the original, but the next layer's weight gradient needs the rounded value that layer multiplied. Final loss, the other check, does not rule out the error of reading the original for both: it persists in models trained with such a store, while planned loss comparisons stay within a margin fixed in advance. We therefore derive from this reference which value each use must read, or which substitute gives the same gradient on average with the forward held fixed, and check these per-use requirements on single operators, without training. In three tests using PyTorch and Transformer Engine, the requirements predicted beforehand whether reuse changes what the backward computes on average relative to an independent copy, and every prediction held. Backward-state policy is thus part of the learning algorithm: it should be specified and checked use by use, not settled by copy accuracy and final loss.
Read the full abstract

All authors Shuxiao Xie, Shuyang Xie, Dezhi Ran, Wei Yang, Tao Xie

Classical Analysis and ODEs

On Entropy and Orlicz endpoint estimates and the dual Muckenhoupt-Wheeden estimate

Israel P. Rivera-Ríos

In this paper the incomparability of certain Rahm entropy maximal functions introduced by Rahm and Orlicz maximal functions that arise in quantitative endpoint estimates for Calderón-Zygmund operators is established. Also the dual Muckenhoupt-Wheeden two-weight estimates introduced by Lerner, Ombrosi and Pérez are revisited, improving the possible choices for the maximal function in the left hand side of the inequality, via a transference principle from two-weight endpoint inequalities.
Read the full abstract

Mesoscale and Nanoscale Physics

Magnetization relaxation of interacting chains of nanomagnets

Denis Ledue, Renaud Patte, Francois Vernay et al. (+1 more)

We investigate the magnetization dynamics crossover from single-particle to collective behavior in a one-dimensional chain of dipolar-coupled nanomagnets with uniaxial anisotropy. Using both an intermediate-to-high damping (IHD) analytical approach based on Langer's theory and time-quantified Monte Carlo (TQMC) simulations, we derive and validate semi-analytical expressions for the relaxation rate and the magnetization relaxation curves. Our main results include: (i) a closed-form expression for the relaxation rate accounting for (weak) dipolar interactions, (ii) a two-exponential semi-analytical formula for the magnetization dynamics $m(t)$ of an interacting chain, and (iii) a systematic comparison with TQMC simulations, showing good agreement for a wide range of parameters. The analysis reveals a field-controlled crossover from uniform (macrospin-like) reversal to edge-nucleation propagation, driven by the spatial inhomogeneity of dipolar stabilization. The derived expressions provide a computationally efficient framework for predicting the relaxation behavior of dipolar-coupled nanomagnetic assemblies.
Read the full abstract

All authors Denis Ledue, Renaud Patte, Francois Vernay, Hamid Kachkachi

Machine Learning

A Comprehensive Benchmark of Source-Free Universal Domain Adaptation on Time Series Representations

Romain Mussard, Fannia Pacheco, Maxime Berar et al. (+2 more)

Source-Free Universal Domain Adaptation (SF-UniDA) extends Universal Domain Adaptation by removing access to source data at adaptation time while still handling label-set mismatches between domains. Despite growing interest in this setting for image data, no benchmark exists for time series, which are more challenging. We present the first SF-UniDA benchmark on time series. In addition, we provide the first study of pretrained foundation models as feature extractors for time series domain adaptation. In this context, we identify a critical and previously underexplored limitation of all existing SF-UniDA methods: the inference threshold for unknown-sample rejection is highly sensitive. We address this by proposing a plug-in auto-thresholding module that can be integrated into any SF-UniDA method. Experiments on three well-known time series datasets confirm the suitability of this module. They also highlight that foundation models do not systematically outperform classical backbones and that SF-UniDA tailored for time series is yet to be developed.
Read the full abstract

All authors Romain Mussard, Fannia Pacheco, Maxime Berar, Paul Honeine, Gilles Gasso

Populations and Evolution

A Stage-Structured Deterministic Model of Fall Armyworm Infestation on Maize Farming

Donald Okoth Ojwang, Mamadou Pathe Ly and Shaibu Osman

Fall Armyworm (FAW) poses a serious threat to maize production in many regions due to its aggressive feeding habits and rapid development cycle. In this study, we developed and analyzed a stage-structured mathematical model to evaluate the impact of different FAW larval instars on maize dynamics during the vegetative and reproductive stages. Analytical results indicate that the two models have unique and positively bounded solutions for all time $t\geq 0$ and admit four equilibrium points: the trivial, non-trivial, maize extinction and coexistence equilibria. The behavior of the model was studied using stability analysis to find conditions under which FAW dies out or continues to spread. Furthermore, sensitivity analysis and numerical simulations were conducted to examine how key parameters affect FAW population dynamics and maize. Numerical simulations of the model in both stages indicate that there is high destruction of maize plants in both vegetative and reproductive stages of maize production due to increased egg production and larval population density. The extensive damage caused by large populations of eggs, larvae and adult moths motivated an extension of the two models to include intervention measures such as traditional methods like handpicking and chemical pesticides. Numerical results indicate that these control strategies significantly suppressed the FAW population with a resultant increase in the maize plant population towards its maximum capacity during the vegetative and reproductive stages, respectively.
Read the full abstract