Peer Revenue Now accepting positions

Papers

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

All of arXiv

Sunday 4 October

Machine Learning

ASCENT: Online Test-Time Training of Long-Horizon Agents via Self-Distillation of Verified Experience

Haodong Lu and Dong Gong

A large language model (LLM) agent solves long-horizon tasks through many reasoning-action turns, with one verification signal at termination. Deployed agents face streams of related tasks, making their trajectories a natural resource for improvement. In-context adaptation agents store reflections, memories, or skills as text, so reuse depends on retrieving the right experience and on a frozen policy executing it. We study Online Agentic Test-Time Training (OaTTT), which trains the LLM's weights on its own execution trajectories during deployment. The agent executes each task once, in one pass over the stream, and the executed trajectory with its verification result is the only learning signal for weight updates that persist across tasks. Directly imitating or reinforcing the generated tokens of this single attempt destabilizes the policy. We introduce ASCENT (Agentic Self-distillation for Cross-task EvolutioN at Test-time), which instead self-distills verified experience. A stable version of the LLM, its frozen initial copy, receives the verified trajectory as privileged information and predicts next-token distributions along it with this hindsight. Distilling them into persistent LoRA fast weights updates the agent for later tasks, without an external reference solution or stronger teacher. By further removing invalid-action turns, ASCENT distills enhanced privileged experience for more efficient execution. We characterize its population target and the limits of sparse outcome selection. Across ALFWorld, WebShop, and AppWorld at varied model scales, ASCENT improves task success and interaction efficiency as experience accumulates, outperforms online adaptation methods, and transfers to held-out scenes, showing that an agent can consolidate verified experience into its weights without a separate training phase or memory retrieval. Project page: https://artificer-ai-lab.github.io/ASCENT
Read the full abstract

Functional Analysis

Incoherent filter banks

William J. Brinkley, Matthew Fickus, John Jasper et al. (+1 more)

Equi-isoclinic (EI) and equichordal (EC) tight fusion frames (TFFs) are types of minimally coherent (maximally orthogonal) subspace packings that arise in several applications, such as compressed sensing. They seem challenging to construct, and the problem of characterizing their existence remains mostly open. In this paper, we characterize those special complex EITFFs and ECTFFs that arise from isometries whose cross-Gram matrices commute, as is the case with filter banks. In particular, we show that every filter bank EITFF is equivalent to a direct sum of equiangular tight frames (ETFs). More generally, we show that every filter bank ECTFF is analogously equivalent to a novel generalization of ETFs that we dub stratified unit norm tight frames (SUNTFs). We further establish a general theory of SUNTFs, providing various explicit constructions of them, and some necessary conditions on their existence. All of our techniques are explicit, yielding direct constructions of incoherent filter banks.
Read the full abstract

All authors William J. Brinkley, Matthew Fickus, John Jasper, Dustin G. Mixon

Computers and Society

From Access to Realized Affordances: University Students' Generative AI Engagement across Linguistic and Sociotechnical Contexts

Ming Li, Qin Xie, Ariunaa Enkhtur et al. (+2 more)

Generative artificial intelligence (GenAI) is increasingly embedded in university students' academic work, yet student engagement is often examined through adoption, frequency of use, or general perceptions, with less attention to how it is shaped by linguistic and sociotechnical conditions. This comparative qualitative study examines how university students access, incorporate, and evaluate GenAI across three contrasting higher education settings in China, Japan, and Mongolia. Semi-structured interviews were conducted with 42 undergraduates from one university in each setting, and the data were analyzed through qualitative interpretation supported by co-occurrence network and correspondence analyses using KH Coder. Across the three samples, GenAI was incorporated into academic practices spanning exploration, production, and refinement. However, access pathways differed across the three contexts. Participants in China navigated between domestic and global GenAIs, those in Japan predominantly used global GenAIs through Japanese, and those in Mongolia more frequently shifted to English when Mongolian outputs were perceived as less satisfactory. During their use of GenAI, students evaluated its accuracy, privacy implications, potential for dependence, and effects on critical thinking. The findings suggest that technological access alone does not ensure equivalent educational usability across linguistic and sociotechnical contexts. Building on the empirical findings and the analytical perspectives, an empirically informed integrative framework is developed in which student GenAI engagement is understood as situated, mediated, enacted, and evaluated. The framework conceptualizes realized affordances as educational possibilities that become practically usable through the interplay of sociotechnical conditions, linguistic resources, academic practices, and evaluative judgments.
Read the full abstract

All authors Ming Li, Qin Xie, Ariunaa Enkhtur, Lilan Chen, Fei Cheng

Artificial Intelligence

MESH-Harness: Self-Improving Agent Harnesses via Bandit-Guided Compositional Evolution

Zhiwei Shang, Yu Huo, Mingrong Gong et al. (+6 more)

An agent harness is the code that organizes context, maintains state, and coordinates tool calls for a language model. We study how to improve the harness under a limited evaluation budget while keeping model weights fixed. Our method, MESH-Harness, organizes each harness into functional modules with explicit role-specific interfaces, allowing alternative implementations of each module to be substituted and recombined. It uses shared module representations and full-covariance LinUCB to score candidate combinations based on predicted performance and exploration value. Mixed-start coordinate ascent selects complete configurations for evaluation without enumerating the combinatorial space. Validation traces then guide local code edits, and the resulting candidates are incorporated into fixed-capacity role-specific pools for subsequent recombination. On text tasks, retrieval-augmented mathematical reasoning, code generation, and interactive scientific tasks, MESH-Harness outperforms Meta-Harness by 5.70, 7.01, 2.00, and 5.00 points, respectively, under matched candidate-evaluation budgets. Iterative harness optimization improves MESH-Harness by 5.63-7.79 points over its first-round configurations. For the reported configurations, aggregate test-time cost is 44.2% lower than that of Meta-Harness, while total cost including search is 14.6% lower. These results show that combining module-level design reuse with feedback-driven compositional search can systematically improve agent harnesses while keeping overall optimization cost under control.
Read the full abstract

All authors Zhiwei Shang, Yu Huo, Mingrong Gong, An Yan, Zikun Qu, Junhao Dong, Bryan Kian Hsiang Low, Chenglin Wu, Zhongxiang Dai

Representation Theory

Calculating double-leaf coefficients in the Hecke category

Colton Sandvik

Given a morphism of Abe-Bott-Samelson bimodules, we construct an algorithm for computing its double-leaves coefficients. The key ingredient is a modification of Bruhat upper-triangularity which allows us to use forward substitution to solve for the double-leaf coefficients. We provide a complete implementation of the algorithm in the Sage computer algebra system. We use this to compute the Zamolodchikov 3-color relation in type $H_3$ for balanced realizations. We further simplify the 3-color relation using thick Soergel calculus and express it as a difference of two thick Elias–Williamson diagrams.
Read the full abstract

Computer Vision

Revisiting Ground-Truth Synthesis from High-Speed Video: Exact Validity Conditions and an Audited Consumer Capture Corpus

Abdullah Al Shafi and Sumaiya Rahim Suma

Motion-deblurring datasets are commonly synthesised by averaging $N$ consecutive high-frame-rate frames and labelling the result with the central frame. We show that this label is unbiased for every capture timing only when the window is odd and the frames' sample durations are equal. An even window shifts every label by a fixed fraction of the blur length, even under perfect timing. Unequal durations are subtler: their mean misalignment is zero, so dataset statistics cannot reveal them, yet when the offset cannot be read from the blur they convolve the supervision rather than adding noise to it. Auditing 51 smartphone clips recorded at a nominal 240 fps, we find 19 captured near 176 fps, a behaviour recorded only in the container's timing tables. In a controlled test the parity choice costs a fitted linear deblurring filter far more than these timing irregularities do, and interpolation labels, unlike blur labels, can be repaired with the true frame times. We provide a tool that checks both conditions without decoding, and will release the clips and their timing tables.
Read the full abstract

Machine Learning

Do Neural Networks Learn Structure-Preserving Maps? A Case Study in Latent-to-Hilbert Embeddings

Muhammad Adnan Shahzad

We ask whether a neural network can learn a structure-preserving map from a compressed latent space to a Hilbert-space representation. Using an 8-dimensional autoencoder bottleneck on MNIST and $n$-qubit product-state targets from PCA-based angle encoding, we report four findings. Although the target angles are generated by a nonlinear sigmoid transformation of the latent projections, the resulting mapping is well approximated by a linear function over the observed latent distribution: linear regression from $z$ to the true target angles achieves $R^{2} = 0.98$, while regression to the MLP's recovered angles achieves $R^{2} = 0.91$. The learned map's primary direction is strongly aligned with the target-induced direction, with cosine similarity $0.989$, while remaining nearly orthogonal to the input's principal direction, with cosine similarity $0.002$. The map is genuinely rank-4: removing any singular direction degrades inner-product preservation by $2.5$–$5.9\times$ despite a singular-value spectrum with two dominant and two small values. The learned subspace does not coincide with the PCA basis used to construct the target, and different random seeds recover the same primary direction but diverge in higher ranks. Finally, kernel ridge regression with an RBF kernel outperforms a tuned MLP (IP error $0.0144$ vs.\ $0.0197$), suggesting that for approximately linear structure-preserving mappings, classical kernel methods may be a simpler and more effective alternative.
Read the full abstract

Machine Learning

FlexCast: Adaptive Weather Forecasting from Arbitrary Field Sets

Yuang Zhang, Chen Hui, Weisi Lin et al. (+4 more)

Most deep learning weather models assign a fixed set of variables and pressure levels to predefined channels, limiting transfer across atmospheric field configurations. This dependence on a fixed field set limits the transferability of trained models across atmospheric field configurations. We propose FlexCast, a field-adaptive weather forecasting model that uses a single set of parameters to produce identity-aligned forecasts for variable-cardinality subsets drawn from a 69-field ERA5 registry. Specifically, a metadata-conditioned adapter the first encodes variable identity, pressure level, and field type and combines them with spatial features. Then, shared rank-16 projec?tions are modulated by metadata-dependent gates to produce field?specific features, while masked set fusion aggregates the available fields into a fixed-width representation. Subsequently, a multiscale U-Transformer processes the fused atmospheric features, while an identity-aware query decoder produces forecasts for the requested fields. Finally, FlexCast learns a standardized six-hour increment and applies it recursively to generate forecasts at longer lead times. Experiments on the 2020 ERA5 test set demonstrate that FlexCast operates across varying field configurations. Compatible cross-field context is associated with lower forecast errors, whereas mismatched context increases them.
Read the full abstract

All authors Yuang Zhang, Chen Hui, Weisi Lin, Haiqi Zhu, Xiulai Wang, Sun-Yuan Kung, Feng Jiang

Artificial Intelligence

Readable Before Actionable: Causal Tracing of Indirect Prompt Injection

Zhe Yu, Wenpeng Xing, Xingxing Yang et al. (+1 more)

Indirect prompt injection causes LLM agents to follow commands embedded in external data. A probe may distinguish instructions from data without identifying a state edit that changes the next action. We study this gap through counterfactual role probes, component-wise activation patching, and separate interventions on AgentDojo trajectories. Role decoding survives changes in content and format. In controlled Qwen tests, it precedes strong tool-choice effects from patches along an independently estimated role direction. On AgentDojo, directions estimated from hijacked and resisted training trajectories reduce attack success at pre-action and injected-span positions, but have little effect at random positions. In longer Qwen trajectories, single-position edits become less effective at later layers; span-wide and repeated edits reduce attack success on the same evaluation set. Removing the learned channel subspace preserves role decoding, yet effective intervention directions transfer poorly across the tested channels. These findings distinguish a readable role signal from an effective behavioral intervention: depth matters in controlled tool choice, while position and context also matter in attack trajectories.
Read the full abstract

All authors Zhe Yu, Wenpeng Xing, Xingxing Yang, Meng Han

High Energy Physics (Experiment)

Precision measurement of ocean tides using cosmic-ray muons

Jiwon Seo, Jongseok Chung, Chang Hyon Ha et al. (+8 more)

Cosmic-ray muons provide a natural probe of the integrated mass along their trajectories. Here we show that cosmic-ray muons can monitor ocean-tide-induced variations in seawater overburden within an underwater tunnel. The observed modulation closely tracks independent tide-gauge measurements, yielding an attenuation response coefficient of $k=43.2\pm1.0$~counts$\rm /h/m$ and resolving meter-scale variations in seawater overburden. A spatial muography scan validates the overburden model, confirming that the seawater contribution is essential to reproduce the observed attenuation. Unlike conventional tide gauges that measure sea level locally, muography measures the integrated overburden above the detector. These results demonstrate that muography can quantitatively resolve small, time-dependent variations in environmental overburden, providing a new approach to monitoring dynamic natural systems.
Read the full abstract

All authors Jiwon Seo, Jongseok Chung, Chang Hyon Ha, TongZhou Huang, Jinyoung Kim, Sooa Kim, Hani Kimku, Byoung-cheol Koh, Minsu Kwak, Sihyun Lee, Yujin Lee

Computer Vision

BossouChimpanzee: Long-term Chimpanzee Video Dataset

Daniel Schofield, Susana Carvalho, Vladimir Iashin et al. (+10 more)

We describe the BossouChimpanzee video dataset, a unique long-term visual record of wild chimpanzees at an outdoor laboratory for field experiments in Bossou, Guinea, spanning three decades (1988-2018) and comprising over 1,200 hours of continuous video recordings collected through collaborative fieldwork and research. In this paper, we outline the history and scientific contributions of the experimental paradigm and video archive, provide key statistics and details on the structure of the main video dataset, and release an initial ~74h snapshot, BossouChimpanzee70h, covering 23 identified individuals focused on chimpanzee individual and action recognition, ahead of the full video resource. This dataset represents a valuable resource for cognitive and behavioural research in ethology and a rich benchmark for training and evaluating machine learning models on audiovisual data from the wild.
Read the full abstract

All authors Daniel Schofield, Susana Carvalho, Vladimir Iashin, Andrew Zisserman, Max Bain, Arsha Nagrani, David Ng, Claudia Sousa, Boniface Zogbila, Jules Doré, Dora Biro, Misato Hayashi, Tetsuro Matsuzawa

Machine Learning

Erased, Rerouted, or Rescaled? Post-Training and the Causal Quotient of a Language Model's Belief State

Weihan Li, Tianshi Zheng, Junhao Wu et al. (+1 more)

What happens to information a pretrained model already encodes when post-training no longer rewards using it? The common language of representation compression conflates three fates: information may be erased, rerouted away from the decision while still represented, or rescaled to occupy less variance while still represented and used. We make these fates identifiable in models whose pretraining recovers Bayesian belief states. A reward that reads only a coarse function of the hidden state defines an exact reward-null kernel. The kernel lets us separately measure whether the information remains recoverable, whether decisions causally depend on it, and how much activation variance it occupies. Theory says what is protected: KL-anchored reinforcement learning preserves the reference policy's log-odds among equally rewarded outputs, supervised and unanchored objectives carry no such constraint, and spectral compression implies neither erasure nor loss of use. In controlled worlds, post-training mostly reroutes or rescales reward-null information and leaves it decodable. Without an anchor decisions can stop using it although the representation survives, and with one they keep using it. Erasure appears only under prolonged weight decay, for distinctions that neither reward nor next-token prediction can see. Open language models show the same dissociation: in-context belief geometry stays decodable under late-layer spectral compression, and within-class behavior depends on the anchor. Post-training thus selects a causal quotient of the pretrained belief state: the reward defines decision-equivalence, the anchor and the state update protect part of what it ignores, and optimization decides whether the rest is erased, rerouted, or rescaled.
Read the full abstract

All authors Weihan Li, Tianshi Zheng, Junhao Wu, Xinlei Chen

Machine Learning

Smoothed Gradient Method for Nonconvex Federated Stochastic Bilevel Optimization

Xinwen Zhang, Peiran Yu, Zhaosong Lu et al. (+1 more)

In recent years, federated stochastic bilevel optimization has attracted increasing attention due to its wide range of applications in machine learning. To reduce the computational overhead associated with second-order Hessian and Jacobian matrices, several first-order methods have been proposed. However, existing methods typically impose restrictive assumptions on the lower-level function, suffer from a strong dependence on the condition number in their convergence rates, and require different learning-rate scales for variables across the upper- and lower-level problems, limiting their practical applicability and complicating hyperparameter tuning. To address these challenges, we propose a stochastic doubly smoothed gradient method for nonconvex federated stochastic bilevel optimization problems, which decouples the learning rates of upper- and lower-level variables and does not require a strongly-convex lower-level loss function. We establish rigorous theoretical guarantees for the proposed algorithm, demonstrating an improved convergence rate of $O(\kappa^{15/2}/\epsilon^5)$ and a communication complexity of $O(\kappa^{4}/\epsilon^3)$, where $\kappa$ denotes the condition number and $\epsilon$ represents the solution accuracy. Notably, these bounds exhibit significantly better dependence on the condition number $\kappa$ than those of existing methods. Extensive experiments validate the effectiveness of our algorithm.
Read the full abstract

All authors Xinwen Zhang, Peiran Yu, Zhaosong Lu, Hongchang Gao

Computer Vision

Mobile-4DGS: Unified Static-Dynamic Real-time Mobile Gaussian Splatting

Xiaobiao Du, Zhen Fang, Tianqing Zhu et al. (+2 more)

Recent advances in 3D Gaussian Splatting (3DGS) have achieved remarkable performance in novel view synthesis, yet deploying both static and dynamic Gaussian representations on resource-constrained mobile devices remains challenging due to heavy storage, redundant primitives, and costly per-frame computation. We present Mobile-4DGS, a unified lightweight framework for high-fidelity real-time static and dynamic Gaussian rendering on mobile platforms. For compact appearance modeling, we introduce a Monte Carlo Specular Energy Aggregator that compresses high-order radiance residuals into the first-order Spherical Harmonics (SH), together with an Attribute-Conditioned SH Enhancement module whose predicted offsets are pre-baked before inference. We further propose a Multi-View Alpha-Based Densification and Pruning strategy to suppress redundant primitives while maintaining multi-view consistency. For dynamic scenes, we develop a compact explicit 4D representation by constructing second-order Gaussian motion, learnable temporal support, and a binary static-dynamic partition, enabling continuous-time modeling without runtime deformation networks. Based on this partition, a Depth-Order Certificate selectively reuses previously committed depth orders to reduce re-projection, sorting, merging, and index-buffer updates during playback. Extensive experiments on static and dynamic scenes demonstrate that Mobile-4DGS substantially reduces storage and rendering overhead while maintaining competitive visual quality, enabling real-time 3D and 4D Gaussian Splatting on mobile devices. \textcolor{magenta}{\href{https://xiaobiaodu.github.io/mobile-4dgs-project/}{Code has been released: https://xiaobiaodu.github.io/mobile-4dgs-project/}}.
Read the full abstract

All authors Xiaobiao Du, Zhen Fang, Tianqing Zhu, Richard Hartley, Xin Yu

Quantum Physics

Fast Random Compilers for Time-Dependent and Time-Independent Lindbladian Simulation

Youngjin Seo, Dhrumil Patel, Hyukjoon Kwon et al. (+1 more)

Randomized product formulas provide simple and flexible methods for quantum simulation. While qDRIFT-type randomized simulation has been extended to Lindbladian dynamics, higher-order randomized compilers have primarily been developed for Hamiltonian simulation. In this work, we develop first- and second-order randomized sampling algorithms for Lindbladian simulation in both the time-independent and time-dependent settings. Building on first-order qDRIFT-type Lindbladian simulation, we introduce a qSWIFT-type second-order correction for time-independent Lindbladians with a finite local decomposition. For $\lambda t \geq 1$, where $\lambda$ is the sum of the Hamiltonian coefficients and dissipative rates, the resulting approximation achieves normalized diamond-norm error at most $\varepsilon$ using $r=O(\lambda^2 t^2/\sqrt{\varepsilon})$ time slices, improving the $\varepsilon^{-1}$ precision dependence of the first-order qDRIFT-type approximation. We then develop first- and second-order continuous-time sampling algorithms for general time-dependent Lindbladians, with the second-order construction likewise achieving an $\varepsilon^{-1/2}$ dependence on the target precision in the number of time slices. We further extend the time-dependent construction to a linear-combination input model over local Hamiltonian and dissipative generators, allowing the randomized sampler to be formulated in terms of local evolutions rather than the full time-dependent Lindbladian. Because the second-order corrected maps need not be CPTP, we use them to estimate observable expectation values rather than to deterministically prepare output states.
Read the full abstract

All authors Youngjin Seo, Dhrumil Patel, Hyukjoon Kwon, Mark M. Wilde

Earth and Planetary Astrophysics

Irradiated Atmospheres VII. Effect of Mixing Energy Flux on Atmosphere-Interior Coupling

Wei Zhong, Zhen-Tai Zhang, Mei Ting Mak et al. (+5 more)

Highly irradiated, underdense rocky planets such as puffy Venuses may host strong coupling between their atmospheres and magma-ocean interiors. Previous studies have mainly treated this coupling through radiative transfer and chemical equilibrium, leaving the role of vertical mixing unclear. Here we quantify how energy transport associated with vertical mixing modifies the thermal structure, volatile chemistry, and emission spectra of puffy Venuses. Atmospheric circulation and gravity-wave breaking are possible dynamical sources of such transport. We show that this energy transport provides an additional greenhouse-like heating pathway by warming the deep atmosphere, thereby increasing the pressure and temperature at the magma-ocean surface. The resulting thermal changes also reshape the atmospheric composition, especially the abundances of \ce{CO} and \ce{H2O}, and change the partitioning of C-, H-, and O-bearing volatiles between the atmosphere and the magma ocean. Moreover, these changes reshape the mid-infrared planet–star flux ratio, especially at low equilibrium temperatures and in higher-mass planets. Under high C/O, it slightly increases the flux near $3.5$–$4.0\,\mu$m, but more broadly reduces atmospheric transparency and suppresses the flux. This may affect radius estimates for puffy Venuses over a limited temperature range.
Read the full abstract

All authors Wei Zhong, Zhen-Tai Zhang, Mei Ting Mak, Xiang-Xing Chen, Bo Ma, Xian-Yu Tan, Cong Yu, Dong-dong Ni

Analysis of PDEs

Existence of a separating norm for the semilinear heat equation

Jin Takahashi

We consider the initial value problem of the semilinear heat equation in the Fujita supercritical range for which global-in-time solutions may exist. The main result shows the existence of a separating norm $\|\cdot\|_X$ in the sense that, for nonnegative initial data $u_0\in L^\infty(\mathbf{R}^n)$, the smallness of $\|u_0\|_X$ implies global-in-time existence, whereas the largeness of $\|u_0\|_X$ implies finite-time blow-up. The separating norm is unique up to equivalence on nonnegative data.
Read the full abstract

Information Retrieval

Quality-Aware Cross-Model Computation Reuse

Jin Cheng, Xiangxiang Dai, Maoli Liu et al. (+3 more)

An intermediate result computed by one model can be reused by other models to perform their tasks. Existing work mainly focuses on practical execution, leaving a theoretical gap in optimizing reuse decisions. This optimization faces two challenges: quality uncertainty, because the effect of reuse on task quality is uncertain across models, and coupled scheduling, because tasks need to share the cost of preparing reusable results. These challenges compound each other: quality must be learned online, but the shared preparation structure makes scheduling NP-hard even with known quality, breaking the key assumption in existing methods. We formulate cross-model computation reuse as an online decision problem and develop the Quality-Aware Reuse Scheduling (QARS) algorithm to address it. For quality uncertainty, QARS learns task-dependent reuse quality from selected, possibly delayed feedback and uses optimistic estimates to guide decisions. For coupled scheduling, it jointly chooses which results to prepare and which tasks should use them, adapting scheduling accuracy to the remaining quality uncertainty. For the considered problem, our analysis separates learning and optimization error in regret and quantifies the tradeoff between scheduling accuracy and computation. Completing the quality-aware stopping rule yields $\widetilde O(\sqrt{T})$ regret while preserving feasibility. Experiments demonstrate the effectiveness of QARS in optimizing cross-model reuse, reducing the combined cost of computation and quality loss by up to 18.0%, and mean regret by 63.9% over the strongest scheduling baseline.
Read the full abstract

All authors Jin Cheng, Xiangxiang Dai, Maoli Liu, Ziyi Han, Zhuohua Li, John C.S. Lui

Artificial Intelligence

Fusion is the New Mutation: Bandit-Guided Evolution on Workflow Graphs

Zhiwei Shang, Jiahang Sun, Mingrong Gong et al. (+9 more)

Automated agentic workflow optimization relies on costly evaluations, making it essential to allocate a limited evaluation budget effectively. Multi-parent fusion can reuse designs from previously discovered workflows, but identifying promising parent combinations requires learning from limited fusion feedback. We introduce DAGO (Directed Acyclic Graph Optimization), a contextual-bandit-guided framework that learns which parent workflows to fuse under a limited evaluation budget. DAGO formulates each candidate parent combination as an arm, represented by pretrained embeddings of its constituent workflows' code and prompts. A diagonal LinUCB policy learns a shared reward model across arms and balances exploitation of arms with high predicted offspring quality against uncertainty-driven exploration. After an arm is selected, an LLM generates a child workflow through summary-guided fusion, and the child's validation score serves as the reward for updating the bandit. A shared directed acyclic graph maintains discovered workflows and their multi-parent lineage, providing an expanding pool of parents for subsequent arm proposals. Across six benchmarks covering mathematical reasoning, code generation, and question answering, DAGO achieves the highest macro-average score among the evaluated baselines. Under matched validation-evaluation budgets, it improves over AFlow from 80.3 to 81.7 while reducing aggregate search expenditure by 11.2%. Ablation studies show that LinUCB-guided arm selection outperforms both random selection and its exploration-free variant, supporting the value of feedback-driven selection and exploration-exploitation balance.
Read the full abstract

All authors Zhiwei Shang, Jiahang Sun, Mingrong Gong, Mingze Kong, Zikun Qu, Pingchen Lu, Junhao Dong, Zhipiao Liu, Hongwei Yang, Guoqing Xie, Yao Shu, Zhongxiang Dai

Classical Analysis and ODEs

Null-additive sets and packing measures

Ondřej Zindulka

A separable metric space has strict measure zero if its packing measure induced by any gauge is zero. We provide a few characterizations of strict measure zero, look at cartesian products and prove that in every Polish locally compact group every set of strict measure zero is null-additive and that in the Cantor set, Euclidean spaces and tori the two notions are equivalent.
Read the full abstract

Computation and Language

Red-TTT: Test-Time Training for Automated Jailbreaking Large Language Models

Tongyan Hu, Hao Li, Xiaogeng Liu et al. (+8 more)

Large language models remain vulnerable to jailbreaks, and automated red teaming is the standard way to find jailbreaks in large language models at scale. Current methods either draw more samples at test time through search, rewriting, and tree expansion, or train a stronger attacker offline with reinforcement learning. Both share a limitation: once an attack on a specific target behavior begins, the attacker's weights are frozen. Any signal it gathers about the behavior stays in its context window and is discarded afterward. The attacker never adapts its proposal distribution mid-attack, so success depends almost entirely on the sampling budget, and under a budget affordable at scale, many behaviors remain unbroken. We propose Red-TTT, which updates the attacker's parameters during the attack on each behavior. At each round, the attacker samples a group of candidates, scores them against the victim's replies, and takes a policy-gradient step before drawing the next group, so what it discovers about the current victim is consolidated into weights rather than accumulated as context. We also adapt the training objective to red teaming, where success is judged by the single best sample rather than the average. Red-TTT requires only sampling access to the victim and integrates into existing attack pipelines with no other changes. Against the Best-of-N baseline, Red-TTT raises attack success rate from 55.9% to 72.4% on average at a budget of 120 samples, improving over the baseline in every configuration and cracking many behaviors previous method cannot. The code is available at https://github.com/SaFo-Lab/Red-TTT
Read the full abstract

All authors Tongyan Hu, Hao Li, Xiaogeng Liu, Ruida Wang, Zhengyu Liu, Shuyao Xu, Ning Zhang, Ziyang Li, Yinzhi Cao, Bryan Hooi, Chaowei Xiao

Artificial Intelligence

When Agent Context Goes Stale: Incoherence in Volatile Agent Context

Yingying Liu, Junzhou Fang and Chenxiong Qian

Modern agents increasingly ground their reasoning in observations returned by tools, such as file contents read from a workspace. However, the data sources underlying these observations may later be modified by users, other agents, or external tools, while the model retains only the stale content in its context window. Existing agent runtimes provide little support for notifying the model that a previously observed fact has become stale, causing agents to reuse outdated observations and make incorrect claims about the current workspace state. We propose Concord, a context coherence framework that maintains the consistency between tool observation in agent context and the mutable sources from which they were derived. Concord links each observation to its source, detects source changes, and uses configurable handling policies to update, annotate, or suppress stale context before reuse. Concord is applicable across different agent runtimes and external resources, and can be easily extended to new runtime-resource settings. We implement Concord as a general framework, and instantiate a concrete use case to assess its effectiveness. We construct ConcordBench, where previously observed file contents become stale after subsequent edits. Across three evaluated frontier models, Concord produces answers consistent with the restored workspace state in all evaluated cases under these constructed conditions, matching the oracle on recover count for this benchmark, while using 46.4% fewer tokens than the strongest non-oracle baseline.
Read the full abstract

Quantum Physics

Quantum n-coloring is undecidable for every n $\ge$ 3

Christian Bo Kidmose-Frederiksen and Alfred Leth Nielsen

The quantum $n$-coloring problem, given graph $G$, asks whether there exists a perfect quantum strategy for the $n$-coloring game of $G$. Previously it was known that quantum $n$-coloring is undecidable only for $n=3$. We extend this to all $n\geq 3$ with one elementary reduction. Letting $G'=(G\square K_3)\lor K_{n-3}$, we show that $G$ is quantum $3$-colorable if and only if $G'$ is quantum $n$-colorable.
Read the full abstract

Machine Learning

Fast Convergence through Distributed Augmentation for Class-Imbalanced Federated Learning

Arathi Nair M, J. Harshan and Anwitaman Datta

In federated learning, mitigating class imbalance is essential to improve minority-class performance. A common approach to address this problem is to augment minority-class samples to achieve local class balance. Existing approaches treat augmentation as a heuristic and do not establish how the amount of augmentation influences the convergence of federated learning, leading to excessive augmentation and increased training time. To address this limitation, we first establish the relationship between augmentation and the convergence behavior of federated learning. Leveraging this insight, we propose DAFL, a distributed augmentation framework that determines the minimum augmentation required for each client-class pair by jointly minimizing augmentation and training time while constraining global class imbalance, thereby improving minority-class F1-score. Experimental results demonstrate that DAFL consistently improves minority-class F1-score while substantially reducing training time, particularly under severe global class imbalance and high label proportion imbalance.
Read the full abstract

Computation and Language

When Verifiable Counts Depend on Wording: Auditing Wording Robustness in Instruction Following

Qishi Zhan, Seoyeon Jang, Zihan Dong et al. (+3 more)

Verifiable instruction-following benchmarks often express each constraint through one fixed template. We test whether scores remain stable when the operational requirement is unchanged but its wording varies. We introduce WISE, a matched evaluation suite and reporting protocol instantiated on exact word count, keyword inclusion exactly once, and an inclusive 8–12 word range. Across 100 matched tasks, up to thirteen models from seven providers, and repeated generations scored over the complete visible output, wording alone produces substantial compliance shifts. In an avoidance-family panel, five avoidance and exclusion forms fall below the positive baseline, while constructional controls also shift compliance substantially: in the nine-model control panel, compliance is 54.9% for the original positive form, 48.2% for a longer positive form, 36.7% when the target appears later, and 33.8% for AVOID1. A strict JSON-structure probe shows wording sensitivity beyond counting, with a different direction of effect. Effect sizes, failure directions, weakest forms, and model rankings vary across realizations. Under the most disruptive exclusion form, the top-ranked model changes and 24.1% of strictly ordered model pairs reverse. Human validation further shows that unanimous agreement on an exact-count interpretation can coexist with substantially different model behavior. WISE supplements conventional scores with mean and worst-form compliance, wording gaps, failure profiles, and ranking stability.
Read the full abstract

All authors Qishi Zhan, Seoyeon Jang, Zihan Dong, Minxuan Hu, Ziheng Chen, Tonghui Qu

Software Engineering

Mind the Gaps: From Failure Attribution to Closed-Form Repair of Code Language Models

Jian Gu, Hongyu Zhang, Chunyang Chen et al. (+1 more)

Code language models must be maintained like the software around them: when a library evolves, a model keeps writing the interface that it saw during training. Repairing the model itself lets one correction reach all downstream uses. Existing repair methods attribute a failure to neurons, select the highest-ranked ones, and apply a generic update. This pipeline assumes that the attributed neurons are the ones to patch and that a generic update fits every failure, and neither assumption has been examined. We examine both on executable API evolution tasks in Python and Rust with three code models and identify two gaps. The targeting gap separates the neurons that failure attribution targets from the neurons that can carry a patch: their top sets have a Jaccard overlap of only 0.15 to 0.20. The tailoring gap separates a generic update from a patch built for the failure: the patches that different carrier neurons need are nearly orthogonal. To address both gaps, we propose ASTRA. It targets neurons by contrastive semantics, an attribution that scores a neuron by its contribution to the logit contrast between the target token and the produced token. It then tailors the patch by solving one small linear system in closed form, which corrects all failing tokens of a sample jointly and needs neither an optimizer nor a backward pass. Contrastive semantics selects significantly better carrier neurons than gradient-based attributions in 3 of 6 settings and comparable ones in the others. On average, ASTRA reaches 66.7 percent Pass@1, against 48.4 percent for the best of AlphaEdit, STAR and low-rank adaptation, and repairs a sample in 2.7 seconds. It is the best method in all 6 settings and for every type of API change, and this advantage persists under an unseen phrasing of the test prompt. Its side effects on unrelated code are small on the large models and larger on the small one.
Read the full abstract

All authors Jian Gu, Hongyu Zhang, Chunyang Chen, Aldeida Aleti

Differential Geometry

An improved result on Chern conjecture

Qing-Ming Cheng, Fengjiang Li, Guoxin Wei et al. (+1 more)

Let $X: M^n\to \mathbb S^{n+1}(1)$ be a closed minimal hypersurface with constant scalar curvature. Chern conjecture says that a closed minimal hypersurface with constant scalar curvature, $S\geq 2n$ if $S>n$, where $S$ is the squared norm of the second fundamental form. As a partial result, Peng-Terng \cite{pt1, pt2}, Yang-Cheng \cite{yc1, yc2, yc3} and Suh-Yang \cite{sy} obtained that if $S>n$, then $S>n+ \dfrac{3}7n$. Recently, Chen \cite{c} has proved $S>n+\frac{26}{59}n$ if $S>n$. Since resolving Chern conjecture is a hard problem, as a midway for resolving Chern conjecture, one wants to prove $S>n+\frac12 n$ if $S>n$. In this paper, we give a positive solution for this midway problem. We should remark that our topological condition is complete which is weaker than closed in Chen \cite{c}.
Read the full abstract

All authors Qing-Ming Cheng, Fengjiang Li, Guoxin Wei, Jiabin Yin

Computation and Language

Inductive Claims Extraction at Scale

Sandrine Chausson and Björn Ross

A large part of political discourse on social media is built and expressed at a level of claims: i.e. declarative, typically single-clause statements, which convey a particular interpretation of reality and can range from factual to evaluative. Moreover, rather than occurring randomly, claims coalesce, recur in patterns, and come to be associated with different world views. When paired with structural computational tools such as Social Network Analysis, claims can be a powerful unit of analysis to study political phenomena such as echo chambers or polarisation. In this paper, we present a pipeline that uses a large language model (LLM) to inductively extract and catalogue claims from large social media corpora, and apply it to two different Twitter datasets: one relating to the 2020 US presidential election and the other to the 2022 FIFA World Cup. We comprehensively evaluate the approach by measuring the pipeline's recall and precision against manually annotated samples, run ablation studies isolating the contribution of its various components, and perform a qualitative error analysis. We discuss the value of the approach in the context of Computational Social Science research, and illustrate its capabilities by presenting the claims catalogue obtained from each dataset.
Read the full abstract

Computer Vision

Answer with Evidence: Consistency-Aware Grounded Visual Question Answering for Roadside Traffic Scenes

Runwei Guan, Rongsheng Hu, Shangshu Chen et al. (+10 more)

Roadside traffic reasoning requires every free-form textual claim to be backed by visual evidence. Existing grounded multimodal large language models (MLLMs) frequently exhibit say-point mismatch, in which the textual answer contradicts the bounding boxes the model localizes. Evaluation metrics that score answers and boxes separately leave this failure unpenalized. We trace the mismatch to the conventional answer-then-ground factorization, which commits to a numerical claim before any object is enumerated. To measure it, we build RoadSceneVQA-G, a benchmark of 34.7K question-answer pairs in which every free-form answer is linked to the set of boxes that witnesses it, and we propose the Answer-Grounding Consistency (AGC) evaluation suite. To address it, we introduce Enumerate-then-Answer (EtA), which reverses the generation order so that answer-evidence agreement becomes a property of the output structure, and Enumeration-Consistent Policy Optimization (ECPO), a reinforcement learning stage that uses the union of multiple rollouts as a recall teacher without ground-truth boxes. EtA raises say-point consistency from 26.6% to 93.7% and grounding F1 from 52.2% to 73.0%, and ECPO further increases F1 to 75.6% without per-box supervision. On gRefCOCO, the same framework outperforms the strongest compared method, indicating that it transfers beyond traffic scenes. The project is available at https://github.com/GuanRunwei/RoadSceneVQA-G.
Read the full abstract

All authors Runwei Guan, Rongsheng Hu, Shangshu Chen, Ningwei Ouyang, Shaofeng Liang, Heyi Lin, Jinjing Zhu, Yang Shi, Dongming Wu, Jianfeng Dong, Daizong Liu, Henghui Ding, Hui Xiong