Peer Revenue Play credits. Real papers.

Papers

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

Computer Science

cs Watch Computer Science

Monday 14 September

Computer Vision

LynnReal-Omni: Native multi-modal Video Generation for Agentic Visual Workflows

Xiaofeng Mao, Peijia Lin, Shaohao Rui et al. (+3 more)

Video diffusion models are stochastic and hard to control: precise content often requires repeated sampling without guaranteed success, and long-horizon scenes drift in appearance, interactions, and temporal coherence. Agentic visual creation provides explicit references, editable 3D scenes, or executable game states for stable control, but does not by itself guarantee high object or character fidelity. Combining the two can enable stable, high-quality generation. To realize this combination, we present LynnReal-Omni, a native multimodal video generation framework built on a 32B shared multimodal diffusion transformer that unifies text-to-video, image-conditioned generation, reference-guided generation, structural control, editing, degraded video restoration, and long-video generation. It accepts heterogeneous visual inputs, including appearance references, editable 3D renders, and game recordings, allowing agents to compose visual conditions within a unified model. We also train a dedicated 27B Flash shared multimodal diffusion transformer for real-time rendering. We build a systematic data pipeline for video cleaning, subject association, multimodal annotation, and aligned control construction, yielding a curated corpus of multi-shot audiovisual segments, and introduce MSAVP, a 100-prompt, 20-metric evaluation design that separates instruction following, generating plausibility, visual quality, temporal behavior, and audio coordination. LynnReal-Omni-Flash further reduces inference cost through model and decoding acceleration, including a lightweight VAE decoder; on one H100, warm generation and decoding of a 22-frame 540p video take 843 ms with LynnReal-Omni and 377 ms with Flash. These results provide a foundation for real-time streaming video generation, making LynnReal-Omni a unified, controllable, and efficient basis for agentic visual creation.
Read the full abstract

All authors Xiaofeng Mao, Peijia Lin, Shaohao Rui, Yibo Zhang, Haibin Wan, Weijie Ma

Robotics

DuctAM: A Duct-Assisted Quadrotor-Based Aerial Manipulator Enabling High-Force Push-and-Pull Interactions

Yi Wang, Rui Jin, Xinhang Xu et al. (+4 more)

Uncrewed Aerial Manipulators (UAMs) extend the capabilities of Uncrewed Aerial Vehicles (UAVs) from perception to physical interaction. Among various aerial interactions, push-and-pull operations are fundamental manipulation primitives that require sustained horizontal forces while maintaining stable flight. In this paper, we propose DuctAM, a compact aerial manipulation platform that enhances horizontal force capability for push-and-pull interactions using two ducted fans integrated along the quadrotor interaction axis. An attitude-force decoupled control scheme enables controllable horizontal forces without requiring large attitude changes. Extensive real-world experiments are conducted to validate the DuctAM. Figure-eight trajectory tracking experiments demonstrate stable flight and accurate motion control in both quad and duct modes. Force-measurement experiments quantify the decoupled longitudinal force capability of DuctAM. Finally, representative push-and-pull interaction tasks, including cart pushing, door closing, and drawer opening, verify the practical effectiveness of DuctAM. The results show that DuctAM achieves significantly improved horizontal interaction force capability while maintaining stable flight compared with conventional UAVs.
Read the full abstract

All authors Yi Wang, Rui Jin, Xinhang Xu, Haotian Jin, Ruiyang Liu, Yizhuo Yang, Lihua Xie

Artificial Intelligence

LongAgent: History-Guided Agentic Search for Longitudinal Outcome Prediction

Siyao Wang, Florian Guitton, Shuojie Fu et al. (+3 more)

Extracting informative representations from longitudinal data that can predict future outcomes remains a critical challenge in medicine. Medical datasets are inherently heterogeneous, consisting of a large number of variables collected from different sources, sampled with different temporal spacings, and representing different aspects of human health status. This requires identifying those variables with predictive value, processing longitudinal information, and integrating multiple variables for outcome prediction. Here, we propose a novel agent-based approach, LongAgent, that can autonomously search over combinations of variable sets, temporal windows and longitudinal aggregation functions, and identify candidates with promising predictive performance. LongAgent utilises a history memory of previous searches and numerical evidence to guide subsequent exploration. On synthetic data, LongAgent achieves a mean prediction RMSE of 1.7376 and improves over the strongest non-agent baseline by 0.0151 (95% CI: [0.0045,0.0260]; p=0.0273). On a real clinical dataset, it performs comparably to the best baseline.
Read the full abstract

All authors Siyao Wang, Florian Guitton, Shuojie Fu, Guanyu Tao, Kai Sun, Wenjia Bai

Machine Learning

Task-Directed Residual AddUNet:Perfect-Reconstruction Routing for Full-Rate Representations

Vikram R. Lakkavalli

This paper establishes a perfect-reconstruction (PR) interpretation of AddUNet and its full-rate realization, and introduces a Residual Full-Rate PR architecture for task-directed representation learning. The survivor–skip structure of a constrained additive U-Net is shown to be exactly equivalent to a critically sampled multirate PR filter bank. The full-rate formulation removes the complementary-subband restrictions of the critically sampled system while preserving PR. A Residual Full-Rate PR architecture is then proposed to progressively route task-irrelevant, nuisance, or redundant structure away from the task-facing survivor while retaining the routed information explicitly. Exact reconstruction is guaranteed for arbitrary shape-compatible linear or nonlinear routing operators, without requiring invertibility, a matched synthesis bank, reconstruction loss, or learned decoder. The resulting architecture decouples representation design from reconstruction design: conservation is structural, while learning is devoted to task-directed routing. The same formulation identifies an identity-shortcut ResNet with its residual output retained as a full-rate PR system. Experiments verify exact single-channel routing of linearly separable factors to machine precision. On TIMIT, the proposed front-end improves test PER from $28.60\pm2.09\%$ to $25.76\pm0.41\%$ with the recognizer and training protocol held fixed, while maintaining exact reconstruction. Speaker probing further shows that structural conservation does not itself imply task-specific invariance.
Read the full abstract

Computation and Language

K-Bench: a clinically calibrated benchmark for evaluating large language models in high-risk mental health conversations

Laura M. Vowels, Matthew J. Vowels, Shivali Sharma et al. (+9 more)

People increasingly use large language models (LLMs) for mental health support, yet their safety in evolving, high-risk conversations remains poorly characterised. We developed K-Bench, a clinician-calibrated, protected benchmark evaluating 125 model configurations representing 33 base models from 14 providers across a fixed cohort of 200 multi-turn vignettes involving suicide, self-harm, domestic violence, substance misuse, and no-risk presentations. Synthetic patient conversations showed substantial distributional overlap with real human-AI conversations. A frozen GPT-4o judge achieved 94.2% exact agreement with clinician consensus across 6,751 eligible item comparisons from 151 clinician-rated transcripts. Leading models combined strong supportive conversation with combined-risk scores above 95, whereas risk exploration exposed substantial variation among lower-performing configurations. Therapeutic prompting produced configuration-specific gains concentrated among weaker models, while elevated reasoning produced no average improvement. K-Bench combines broader clinical coverage and configuration-scale comparison with a continuously updated public leaderboard whose operational test materials are protected from direct optimisation. The leaderboard is available at www.k-bench.ai.
Read the full abstract

All authors Laura M. Vowels, Matthew J. Vowels, Shivali Sharma, Apoorv Jha, Rehnuma Choudhury, Wasseem El Sarraj, Rachel Francois-Walcott, Aruba Hussain, Sarah Ingram, Angela Loulopoulou, Adva Segal, Elena Volkova

Computation and Language

Learning to Coach for Experiential Learning

Guanheng Chen, Tianzhu Ye, Li Dong et al. (+3 more)

Language models can learn from experience, but raw solution trajectories are often too long and noisy to provide effective guidance. In this work, we propose Learning to Coach (L2C), a framework that trains a dedicated LLM-as-a-Coach to extract actionable experiential knowledge from an actor model's previous trajectory. The actor remains frozen, while the LLM-as-a-Coach is trained to maximize a reward given by the correctness of the actor's guided response. We study two such rewards: a same-instance reward, which improves subsequent responses on the original problem, and a cross-instance reward, which elicits knowledge that transfers to other instances. Across mathematical reasoning and interactive text-games, L2C consistently outperforms self-refinement and an untrained LLM-as-a-Coach. Running experiential learning for more iterations further improves accuracy and uses additional inference compute more effectively than enlarging the actor's decoding budget. The trained LLM-as-a-Coach also transfers to out-of-distribution tasks and adapts its guidance to the specific actor it coaches.
Read the full abstract

All authors Guanheng Chen, Tianzhu Ye, Li Dong, Xun Wu, Shaohan Huang, Furu Wei

Discrete Mathematics · Cross-listed from math.CO

Neither simpliciality nor mutation connectivity: conjectures of Las Vergnas and Cordovil-Las Vergnas fail

Qiyuan Gu and Kolja Knauer

We construct a simple rank-$7$ oriented matroid on $24$ elements with no simplicial tope, disproving the Las Vergnas simplex conjecture from 1980. We then show that the mutation graph on uniform oriented matroids is disconnected for infinitely many values of rank $r$ and ground-set size $n$, disproving the Cordovil–Las Vergnas conjecture from 1988.
Read the full abstract

Computation and Language

Before You Poll with LLMs: A Deliberative Diagnostic Framework

Ahmed Wali and Hassaan Tayyab

Can LLMs reason through new information like humans, or do they merely retrieve cached opinions? This is critical for silicon sampling, where LLM personas simulate public opinion at scale. Current evaluations test only whether personas hold the right opinions – a static snapshot. But opinion research increasingly depends on dynamic fidelity: whether personas update beliefs in response to new arguments, as humans do during deliberation. No existing benchmark tests this. We introduce the Deliberative Polling Diagnostic Framework, which compares human and LLM belief shifts after identical informational interventions. Grounded in deliberative polling, it surfaces failures invisible to static evaluation: models that produce plausible partisan opinions can still misrepresent how those opinions change. Applying the framework to five frontier models using data from America in One Room (526 personas, 72 questions), we find that every model fails, each in a unique manner. GPT-5.1 exhibits reversal: its personas become more hostile toward the opposing party after balanced information, while humans become less so. This reversal is selective (80% on outgroup vs. 26% on policy questions) and symmetric across partisan identities. Gemini 2.0 Flash, Claude Sonnet 4.5, and Llama 3.3 70B exhibit overshoot, shifting correctly but at 5-7x human magnitude. DeepSeek V3 exhibits rigidity with near-zero change. Targeted ablations reveal that policy content triggers these failures and that they are identity-specific: GPT-5.1 reverses on outgroup questions but overshoots on ingroup; Gemini shows the inverse. We term this signature self-sycophancy: conformity to the model's internal stereotype of the persona rather than reasoning from the information provided. Our framework offers a concrete protocol: run the deliberative diagnostic before trusting LLM personas to mimic revised beliefs.
Read the full abstract

Data Structures and Algorithms

Scalable Triangle Counting: The Threshold Algorithm

Asaf Etgar, Anna Gilbert, Quanquan C. Liu et al. (+1 more)

We study one-pass triangle counting on random-order edge streams. We present a remarkably simple algorithm—read edges from the stream until $Q$ triangles are observed in the prefix, then output $Q\,(m/S)^3$ where $S$ is the stopping length—and prove that, when the maximum number of triangles incident to any edge satisfies $\eta \le T^{2/3}$, this is a $(1\pm\varepsilon)$-approximation of $T$ with probability $1-\delta$ using $O(\varepsilon^{-2}\log(1/\delta)\, m/T^{1/3})$ memory. Crucially, the algorithm does not need any a priori estimate of $T$, in sharp contrast with state-of-the-art sampling-rate based algorithms (McGregor and Vorotnikova, PODS 2020; Tsourakakis et al., KDD 2009). It also does not need a prescribed memory budget: the stopping rule self-selects the prefix length and can return an estimate before reading the entire stream. The proof rests on a Schudy–Sviridenko concentration argument for an independent-edge-sampling estimator, coupled to the without-replacement prefix produced by the algorithm. On six real temporal streams, the algorithm's stopping prefix follows the predicted cube-root scaling and achieves at most $6\%$ error at a $10\%$ prefix, without using $T$. At a fixed stored-edge budget, variance-reduced reservoir samplers are often more accurate, but only after reading the entire stream. On a separate, much larger, $1.8\times10^9$-edge graph, the threshold algorithm reads $0.46\%$ of the stream and returns $3.8\%$ error, while the strongest reservoir baselines do not finish a pass within the wall-clock cap.
Read the full abstract

All authors Asaf Etgar, Anna Gilbert, Quanquan C. Liu, Andrew McGregor

Networking

Proportional-Fair Resource Allocation and Dual-Threshold Early-Exit Inference for Secure Cooperative Multi-Layer Edge Intelligence

Thai T. Vu, John Le, Tu N. Nguyen et al. (+3 more)

This paper proposes FREDI (Fair Resource Allocation for Edge Dual-Threshold Inference), a secure wireless edge-intelligence framework for event-triggered inference in a cooperative user equipment (UE)–edge server (ES)–cloud system. Each UE performs early-exit convolutional neural network (CNN) screening using dual confidence thresholds, while critical events are securely offloaded to an edge server for detailed classification. We formulate a proportionally-fair utility maximization problem that jointly optimizes UE–ES association, wireless and processing resources, and confidence thresholds. FREDI decomposes the problem into proportional-fair resource allocation and dual-threshold inference optimization. We prove that the detected-critical event set is set-monotone non-increasing in both thresholds, and exploit the finite empirical confidence domain for exact threshold optimization. An empirical resource–utility response envelope yields a computable global suboptimality bound and a sufficient condition for global optimality. By pre-eliminating infeasible UE–ES pairs and exactly projecting out bandwidth and transmit-power variables, the resource-allocation subproblem is reduced to a mixed-integer exponential-cone program solvable to the certified global optimality within a prescribed gap. Numerical results with early-exit MobileNetV2 and ShuffleNetV2 demonstrate near-perfect UE fairness with aggregate utility close to a Sum-Utility benchmark, reveal security-induced resource fragmentation, and demonstrate the Stage-A scalability from 6 to 144 UEs with median solving time below 0.1~s in the tested configurations.
Read the full abstract

All authors Thai T. Vu, John Le, Tu N. Nguyen, Jun Shen, Quang Vinh Duong, Ha Nguyen

Cryptography and Security · Cross-listed from quant-ph

Instantiating Microcrypt: Obstacles and opportunities via tailored state certification

Jose Carrasco, Jens Eisert, Soumik Ghosh et al. (+3 more)

Recent work has introduced the Hamiltonian phase state (HPS) assumptions, which postulate that Hamiltonian phase states can be used to instantiate pseudorandom and one-way state generators. Additionally, it has been conjectured that these assumptions can be true, even if one-way functions do not exist. This is exciting, because if true, then the HPS assumptions provide a route to the instantiation of Microcrypt. In this work we falsify this conjecture, by proving that if the HPS assumptions are true, then one-way functions exist. While this removes the possibility of instantiating genuine Microcrypt cryptography with Hamiltonian phase states, it shows that the HPS assumptions provide novel inherently quantum assumptions for the construction of classical cryptography. Technically we achieve this via a method for the construction of one-way puzzles from one-way state generators and tailored "measure first, ask later" state certification protocols. This generalizes prior constructions of one-way puzzles from one-way state generators via classical shadows and allows us to relate properties of the one-way puzzle to properties of the state certification protocol used in the construction. Specifically, if the state certification protocol admits efficient classical post-processing then one obtains an efficiently verifiable one-way puzzle, and if the state certification protocol can be efficiently classically simulated in a certain sense, then one obtains a classical one-way puzzle, which implies one-way functions. The latter observation allows us to prove that the HPS assumptions imply one-way functions, by exploiting properties of state certification protocols for phase states. The former observation provides a new toolbox for the construction of efficiently verifiable one-way puzzles by exploiting tailored state certification protocols for pseudorandom and one-way state generators.
Read the full abstract

All authors Jose Carrasco, Jens Eisert, Soumik Ghosh, Dominik Hangleiter, Nicky Kai Hong Li, Ryan Sweke

Game Theory

The Existence of Maximin Share Allocations is $\Delta_2^P$-complete

Sushmita Gupta and Sanjay Seetharaman

The maximin share (MMS) guarantee has become one of the central fairness benchmarks for allocating indivisible items. Since Kurokawa, Procaccia, and Wang [EC'14, JACM'18] showed that exact MMS allocations need not exist, a substantial literature has developed around the existence and computation of approximate MMS allocations, including the recent work of Heidari, Kaviani, Seddighin, and Shahrezaei [SODA'26]. In contrast, a fundamental complexity question posed more than a decade ago by Bouveret and Lemaître [JAAMAS'16] has remained unresolved: how hard is it to decide whether an exact MMS allocation exists? For additive valuations, Lonc and Truszczynski [JAIR'20] established membership in the complexity class $\Delta_2^P$ (also known as $P^{NP}$), but no hardness result was known. For the more general class of 2-additive valuations, Bouveret and Lemaître established NP-hardness, leaving a substantial gap to the $\Delta_2^P$ upper bound. Moreover, the complexity of MMS existence for additive and $k$-additive valuations was posed as open questions. We resolve this longstanding question by proving that deciding the existence of an MMS allocation for additive goods is $\Delta_2^P$-complete, even for five agents. Additionally, we prove weak coNP-hardness for three agents, thereby establishing a precise dichotomy with the known existence guarantee for two agents, and strong coNP-hardness when the number of agents is unrestricted. Furthermore, our strong hardness construction yields an inverse-polynomial gap in the optimal MMS approximation ratio, ruling out an FPTAS for approximating this ratio unless P=NP. Finally, we show that all these results for goods extend to the chores setting through a polynomial-time transformation that preserves MMS existence.
Read the full abstract

Robotics

Uncertainty-Guided Sparse Refinement for Action Chunking Transformer Policies

Chenyang Wang, Yuntian Wang, Xiaoxiong Yang et al. (+3 more)

Learning chunk-based visuomotor policies for long-horizon robot manipulation remains challenging. Recent action-chunking methods have shown promising performance by predicting temporally extended action sequences. However, their failures are often dominated by prediction errors at a small number of critical timesteps rather than uniformly poor predictions across the entire action chunk, making uniform refinement inefficient and insufficiently targeted. To address this bottleneck, we propose Uncertainty-Guided Refinement (UGR), a sparse refinement framework for chunk-based visuomotor policies. Specifically, UGR follows a coarse-to-refine design: it first predicts a full action chunk, estimates per-step temporal uncertainty from the coarse hidden states, and applies residual correction only to the most uncertain timesteps selected by a binary mask. The uncertainty branch is decoupled from the coarse action predictor, enabling clean attribution of the refinement gains to uncertainty-guided correction rather than additional predictor capacity. Extensive experiments on five dual-arm manipulation tasks from the RoboTwin benchmark show that UGR achieves the best success rate on four tasks, improves over the ACT baseline by up to 13% absolute, and outperforms both full-chunk and position-agnostic block refinement in ablation studies.
Read the full abstract

All authors Chenyang Wang, Yuntian Wang, Xiaoxiong Yang, Dingde Jiang, Siao Liu, Yang Liu

Machine Learning

Per-Matrix Optimality Is Not Enough: Three-Level Optimization for Low-Rank LLM Compression

Huicheng Zhang, Xiyao Feng, Ze-Tong Li et al. (+6 more)

Per-matrix singular value decomposition (SVD) truncation is Eckart-Young optimal in the whitened Frobenius norm, but errors from independently compressed matrices compound through the block's nonlinear forward pass. Inspired in part by hierarchical variational optimization in quantum many-body methods, we introduce a three-level chain that widens optimization scope from individual matrices to Transformer blocks to the full model: whitened SVD~(L1), block-level joint optimization~(L2), and end-to-end language-modeling loss refinement~(L3), all from 256 calibration sequences, with no instruction or recovery data. On LLaMA-7B at 60% compression, the chain reduces WikiText-2 perplexity from 42.1 to 19.1 to 11.4. The block-level stage acts as a regularizer: skipping it worsens Penn Treebank (PTB) perplexity by 24 points, a gap that additional end-to-end training did not close in our experiments. Perplexity gains hold across 20-80% compression, five architectures up to 13B parameters, and both in-distribution and out-of-distribution benchmarks, though the cross-architecture rows use architecture-specific configurations and the ratio sweep was not run under one common protocol. With more calibration data, skipping the block-level stage becomes competitive, revealing an offline compute–data trade-off. We therefore claim improvements only in perplexity and compression fidelity; downstream accuracy remains well below the dense model.
Read the full abstract

All authors Huicheng Zhang, Xiyao Feng, Ze-Tong Li, Chengkai Zhu, Xiao Shi, Xiwei Pan, Jinguo Liu, Ge Bai, Xin Wang

Data Structures and Algorithms

Consistency-Robustness Tradeoffs for Online Bipartite Allocation with Multiple Stages

Alexander Lindermayr, Nicole Megow and Lauren Paul

We study learning-augmented online bipartite allocation with multiple stages. In the $k$-stage vertex-weighted fractional bipartite matching problem, demand vertices arrive in $k$ stages, and the algorithm receives possibly inaccurate predictions of the allocation in each stage. While tight consistency-robustness tradeoffs were known for the two-stage case, no nontrivial tradeoff was known for an arbitrary number of stages. Our main result is the first consistency-robustness tradeoff for $k$-stage vertex-weighted fractional bipartite matching with predictions, for every $k\ge2$. Let $R_k=1-(1-1/k)^k$. For every $R\in[0,R_k]$, our algorithm is $R$-robust and $C_k(R)$-consistent, where $C_k(R)=k(1-R)^{1/k}+R-(k-1)$. This simultaneously recovers the known tight two-stage tradeoff and the optimal prediction-free $k$-stage competitive guarantee $R_k = C_k(R_k)$, while strictly dominating the natural randomized coin-flip baseline between these endpoints. We also present an algorithm for the classical online setting, where demands arrive one by one and the number of demands is unknown in advance. It has a consistency ratio of at least $C_\infty(R)=1+R+\ln(1-R)$ for a given robustness $R\in[0,1-1/e]$, improving the best previously known tradeoff for this problem. Finally, we extend the framework to fractional AdWords and fractional predictions. Our algorithms are based on stage-wise convex programs with carefully calibrated vertex-dependent penalties. The penalties maintain a dynamic safety reserve for each supply vertex, balancing protection against adversarial future arrivals with the ability to exploit the predicted allocation.
Read the full abstract

Computational Engineering

Block-modular grey-box identification of a short-stroke seawater-pump power take-off for wave energy: loss decomposition and multi-regime validation

Mahdi Tayyebati, Amir Hamza Siddiqui, Mian Masoud et al. (+2 more)

Physically interpretable power take-off models support wave-energy development at every stage. For the seawater-pump class, few have been identified on a real machine and validated at the system level. This paper presents a block-modular grey-box model of a short-stroke seawater pump, identified from bench data alone. The pump decomposes into chamber, relief valve, piston seal and tip check valve, each structured by component physics, and the few parameters are identified block by block from terminal pressure, displacement, force and discharged-mass measurements, with a closed-cycle mass balance that decouples the degenerate seal and valve losses and closes to within 1% of the inflow. Over the sinusoidal campaign, the pressure-energised piston seal carries 84% of the lost mass and 97% of the lost energy, and the pump delivers about four-fifths of its input as high-pressure flow. Generalisation is demonstrated rather than assumed. With parameters fixed on ramps and sinusoids, the model reproduces 360 s sea-state records in free run to about 10% of range, captures 80% of several thousand peaks within 5 bar, transfers to a re-sprung valve on cracking strokes once its crack pressure is identified, follows thousand-cycle endurance records that drift below 0.1 bar, and reproduces the force and cycle-mean absorbed power, 0.1 to 5.7 kW, to a 4% median error. A practical-identifiability and uncertainty analysis confirms the terminal data resolve the retained parameters; thirty-seven repeated records set the experimental floor. The result is a compact, physically labelled model for design, energy assessment and model-based control of seawater-pump power take-offs.
Read the full abstract

All authors Mahdi Tayyebati, Amir Hamza Siddiqui, Mian Masoud, Kristian Glejbøl, Christian Berggreen

Computer Vision

TRACE: Two-Stage Detector-Response Estimation With Angular Cosine Expansion for Ring Artifact Correction in Photon-Counting CT

Jigang Duan, Heran Wang, Ligen Shi et al. (+3 more)

Detector response nonuniformity introduces systematic projection errors and ring artifacts in photon-counting detector computed tomography (PCD-CT). In measured PCD-CT data, residual stripe amplitudes vary slowly with projection angle, which fixed-bias models cannot adequately capture. We propose TRACE, a two-stage unsupervised sinogram decomposition method for estimating and correcting these response-related errors. TRACE represents stripes as a fixed bias plus low-order discrete cosine transform (DCT) components, using a small number of coefficients to describe angular variations at each detector element. A learnable analysis–synthesis architecture represents the ideal projections, while two-stage optimization separates them from fixed and then dynamic stripes. An angular-gradient soft orthogonality constraint suppresses correlated variations within the shared DCT gradient subspace, reducing the leakage of object structures into the artifact estimate. All parameters are optimized directly on the measured sinogram without paired training data. Experiments on measured QRM mouse phantom and porcine trotter data show that TRACE suppresses ring artifacts and improves image uniformity while preserving edge sharpness, soft-tissue texture, and trabecular detail.
Read the full abstract

All authors Jigang Duan, Heran Wang, Ligen Shi, Zheng Sun, Ping Yang, Xing Zhao

Computer Vision

Integrating Multi-view Multi-light Surface Reconstruction into Cultural Heritage Workflows

Baptiste Brument, Robin Bruneau, Benjamin Coupry et al. (+6 more)

Cultural heritage documentation increasingly relies on image-based 3D surface reconstruction, with photogrammetry software making such workflows accessible to archaeologists, conservators, and heritage technicians. These tools have been successful for conventional multi-view acquisition, but they do not routinely exploit richer multi-view, multi-light data, despite its potential for improving fine-scale surface reconstruction. This limitation is particularly relevant in heritage contexts, where controlled-light acquisition devices such as RTI domes are already used to capture illumination-varying image sets. The challenge is therefore to connect these existing acquisition practices with recent computer vision methods in a form that can be used within operational heritage workflows. In this work, we address this need by integrating state-of-the-art components from computer vision for multi-view, multi-light surface reconstruction into Meshroom, an open-source photogrammetry framework. Rather than proposing a new reconstruction algorithm, our contribution is to assemble and expose existing advanced methods, namely a complete photometric stereo ecosystem (calibrated, self-calibrated and universal), automatic object masking, and multi-view normal-and-reflectance integration, within a usable heritage-oriented workflow. The proposed system thus provides an intermediate software layer between computer vision research code and practical cultural heritage applications, making recent techniques easier to use and evaluate.
Read the full abstract

All authors Baptiste Brument, Robin Bruneau, Benjamin Coupry, Vincent Demoulin, Jean Mélou, Antoine Laurent, Fabien Castan, Jean-Denis Durou, Lilian Calvet

Numerical Analysis

Superconvergence of Centered Finite Difference Approximations

Mario J. Bencomo, Joseph Igot and Emma Maltes

The finite difference (FD) method is commonly used to approximate derivatives of smooth functions, with accuracy typically determined by stencil size and derivative order. However, certain centered stencils exhibit unexpectedly higher accuracy, a phenomenon known as superconvergence, which has been observed in practice but lacks rigorous explanation. We present a mathematical framework for superconvergence in centered FD approximations based on Taylor expansions of the truncation error and the resulting linear system for the FD coefficients. By analyzing symmetry properties of these coefficients and their interaction with the parity of the derivative order, we identify conditions under which higher-order error terms cancel. We show that superconvergence occurs for odd-order derivatives with even centered stencils and for even-order derivatives with odd centered stencils, while no superconvergence occurs for even derivatives with even centered stencils. Numerical experiments in MATLAB confirm the predicted convergence rates.
Read the full abstract

Computation and Language

CiteGuard-RAG: A Validation-Centered AI System for Evidence-Grounded Question Answering

Sumit Barua, Guan Hong, Halil Dursunoglu et al. (+2 more)

Retrieval-augmented generation (RAG) can improve access to complex information; however, retrieving evidence alone does not ensure that answers are grounded, citation-valid, or appropriately refused. This paper introduces CiteGuard-RAG, a validation-centered AI system for evidence-grounded question answering. The system integrates hybrid semantic-lexical retrieval, citation-constrained generation, sentence-level grounding validation, and single-pass regeneration. Validation is used at runtime to determine whether a candidate answer should be accepted, refused, or regenerated before final delivery. CiteGuard-RAG is evaluated on 400 questions across a controlled housing-law dataset, PrivacyQA, and CUAD. In the controlled evaluation, it achieves 99.1% retrieval accuracy, 98.3% grounded-answer accuracy, and 98.3% citation validity, with no validation-detected hallucinations. Ablation results show that grounded-answer accuracy drops sharply when validation is removed, even when retrieval accuracy remains unchanged. External evaluation shows that while citation validity remains strong, evidence utilization, span alignment, and refusal calibration become harder under domain shift. These findings indicate that trustworthy RAG systems require explicit validation between retrieval and final answer delivery. CiteGuard-RAG provides a practical architecture for linking retrieval, generation, citation checking, abstention, and regeneration in high-stakes information access.
Read the full abstract

All authors Sumit Barua, Guan Hong, Halil Dursunoglu, Charles Rodgers, Alvis Fong

Machine Learning

Sharp Regret Bounds and a Task-Covariance Correction for Spectral Representation Learning

Dier Tang, Jing Yee Tan and Guangyue Han

Spectral contrastive learning keeps the $k$ directions along which two views of the data depend most strongly, while downstream tasks may need other directions. Existing guarantees for spectral features assume that the distribution of tasks has no preferred direction. We show that this isotropy condition is the wrong criterion. With linear probing, the task distribution affects the average probing gain only through the matrix $B\Lambda B^\top$, where $B$ records the dependence between views and $\Lambda$ is the task covariance: optimal features span its top-$k$ eigenspace, whereas spectral features span that of $BB^\top$. Transfer therefore depends on alignment, and strongly uneven task preferences cost nothing when they favour the spectral directions. Otherwise the regret of spectral features is at most linear in the anisotropy of $\Lambda$. No representation chosen without task information improves this rate when the leading spectrum is flat, and a spectral gap makes it quadratic. With labelled tasks, the same matrix gives a correction: Fisher discriminant directions for a frozen encoder, or a one-term change to the spectral contrastive loss. We bound the regret of the correction by the error in estimating $B\Lambda B^\top$, and compare this error with the misalignment of spectral features to decide when to correct. The analysis extends to finite samples, continuous variables and noisy views. Correction reduces synthetic held-out regret from $0.858$ to $0.003$ with $2000$ labelled tasks, and the decision rule gives the lowest regret among the tested rules at every bank size. On CIFAR-100, spectral features of frozen encoders keep most of the task signal on unseen classes, and where a bank of other classes would lower accuracy, the rule keeps them in every case.
Read the full abstract

Cryptography and Security

First Galileo SAS Authenticated Time Solution

Aleix Galan-Figueras, Ignacio Fernandez-Hernandez, Wim De Wilde et al. (+5 more)

Spoofing attacks against civilian GNSS receivers have grown more common, especially near conflict zones where they now disrupt civil aviation, maritime operations, and critical infrastructure on a daily basis. Spoofing is possible because legacy civil GNSS signals are largely predictable in both their navigation data and ranging codes, allowing an attacker to forge a signal that imposes a false position and time on an unsuspecting receiver. Cryptographic authentication schemes such as Galileo's Open Service Navigation Message Authentication (OSNMA) mitigate this threat by verifying the authenticity of the navigation data. The ranging code itself, however, remains unprotected. To close this gap, Galileo is introducing a Signal Authentication Service (SAS) in the E6-C signal, which directly authenticates ranging measurements. SAS is currently transmitted by only two satellites in an elliptical orbital plane, of which at most one is visible at a time, meaning a full position solution is not yet possible; however, a georeferenced receiver can still obtain an authenticated time solution. This paper presents, to the authors' knowledge, for the first time, a timing solution computed from an authenticated civil GNSS signal. We develop a snapshot software receiver implementing a simplified version of the Galileo SAS protocol to compute the receiver clock bias from an authenticated pseudorange, using radio-frequency data recorded with an engineering prototype software-defined radio receiver from Septentrio. We evaluate the resulting timing solution using recordings from both SAS-capable satellites collected at different locations, demonstrating the feasibility of authenticated timing ahead of full SAS operational deployment.
Read the full abstract

All authors Aleix Galan-Figueras, Ignacio Fernandez-Hernandez, Wim De Wilde, Rafael Terris-Gallego, Gonzalo Seco-Granados, Cillian O'Driscoll, Sibren De Bast, Sofie Pollin

Numerical Analysis

Regularization and compression of integral operator kernels with tensor paraproducts

Oluwadamilola Fasina and Ronald R. Coifman

We present a new perspective on integral operator kernels given by a nonlinear function of the distance. This perspective is useful since it permits one to quasilinearize the nonlinear transformations acting on the tensor Haar expansion of the distance, d(x,y). This leads to a new representation comprised of a principal and residual term with desirable features; namely, enhanced local regularity of the principal component (and in certain situations rapid off-diagonal decay leading to O(N) storage) and a residual expansion of improved approximation as a consequence of enhanced global regularity. Numerical experiments on the potential kernel are included
Read the full abstract

Artificial Intelligence

AlgoEvo: Self-Evolving Agentic Search for Automated Algorithm Discovery

Junhao Qiu, Qinglong Hu, Ji Cheng et al. (+3 more)

Large language models have advanced automated algorithm discovery by synthesizing executable code, but existing frameworks trap them in rigid search pipelines with pre-defined control flows. This limitation restricts adaptive reasoning, blocks cross-paradigm transfer, and overlooks richer execution feedback. To bridge this gap, we introduce an end-to-end framework, AlgoEvo, a unified agentic architecture that transforms automated algorithm discovery into an interactive, knowledge-accumulating process. An autonomous agent dynamically inspects, diagnoses, and edits code based on runtime feedback. A design skill hub decouples paradigm-specific knowledge from the core discovery engine, allowing a unified workflow to seamlessly handle single-heuristic, multi-objective, and multi-component design. Meanwhile, a hierarchical experience bank organizes search trajectories into a task-level tree to guide exploration and consolidates cross-task patterns into reusable skills. Across six representative benchmark tasks, AlgoEvo reaches state-of-the-art performance with as little as 7% of the evaluation budget and reduced token consumption, demonstrating strong intra-task accumulation, cross-task transfer, and the ability to reproduce or exceed the strongest existing methods through flexible skill activation.
Read the full abstract

All authors Junhao Qiu, Qinglong Hu, Ji Cheng, Xialiang Tong, Liyong Lin, Qingfu Zhang

Systems and Control

Hydrogen-Diesel Dual-Fuel Engine Operation Using Real-Time GRU-Based Nonlinear Model Predictive Control

Alexander Winkler, Vasu Sharma, Julian Bedei et al. (+4 more)

Hydrogen-diesel dual-fuel (H2DF) combustion reduces combustion-out CO2 emissions but exhibits nonlinear cycle-to-cycle dynamics at high hydrogen energy shares (HES). This work evaluates nonlinear model predictive control (NMPC) with a gated recurrent-unit deep neural network dynamics model for transient H2DF control. Trained on 99,800 engine cycles, the model predicts indicated mean effective pressure, nitrogen oxides (NOx), particulate matter (PM), and maximum pressure-rise rate. Single-cylinder Cummins 4.5 L experiments follow an unseen 4,900-engine-cycle trajectory. Compared with production diesel-only control, NMPC improves load-tracking mean absolute error by 27.8% and reduces mean PM by 61.1%, while mean engine-out NOx increases by 105.1% without exhaust-gas recirculation. Mean and peak HES reach 39.7% and 55.1%; a high-hydrogen setting achieves 77.8% peak HES without constraint violations. Robust to feedback noise and model-plant mismatch and executing in 3 to 7 ms per engine cycle on low-cost embedded hardware, learned-dynamics NMPC enables practical, real-time, constraint-aware transient H2DF control with substantial diesel substitution.
Read the full abstract

All authors Alexander Winkler, Vasu Sharma, Julian Bedei, Edward Sperling, Charles Robert Koch, David Gordon, Jakob Andert

Artificial Intelligence

Atria Dawn: The Dawn of Agentic Superintelligence

Honglin Guo, Tao Gui, Kun Cai et al. (+140 more)

As AI agents become participants in the development of their successors, they reshape both the production of intelligence and the role of human researchers. We introduce Atria Dawn Preview, a foundation agentic language model designed for scientific research and engineering workflows, with the goal of expanding the frontier of agent productivity in the real world. This model is trained via a Verifiable Experience Pipeline that connects tool-mediated interactions to executable environments and externally verified outcomes. Across 16 benchmarks spanning real-world research, engineering, and digital work, Atria Dawn Preview is competitive with frontier agents and achieves the highest reported score on five of them. Beyond standalone performance, we examine the real research-and-development process behind this model as a case study of human–AI collaboration, analyzing 769 task records from 56 participants together with agent logs. When asked to evaluate completed tasks under comparable conditions, participants rated about one-third of completed AI-assisted tasks as infeasible without AI. More strikingly, agents frequently propose methods and implement revisions, while humans retain most final decisions and guide exploration through judgment and feedback. These observations indicate a shift from task-level execution to project-level partnership, with human effort concentrating on what is worth pursuing and how evidence should guide research. Progress toward more autonomous AI research must therefore advance both the capacity for discovery and the capacity for meaningful human oversight, preserving accountable human authority over the risks and direction of continued development.
Read the full abstract

All authors Honglin Guo, Tao Gui, Kun Cai, Haodong Chen, Yicheng Chen, Guanting Dong, Qiming Ge, Yuyang Hu, Zixian Huang, Jiajie Jin, Alexander Lam, Yining Li, Jiahang Lin, Yanjiang Liu, Xinyu Lu, Haijun Lv, Zerun Ma, Junlin Shang, Qisheng Su, Guoqiang Wang, Rui Wang, Zhecan Wang, Hao Xiang, Xinchen Xie, Shuhao Xing, Xiaoyu Xing, Wanghan Xu, Xinyu Yang, Yajie Yang, Chengfeng Zhao, Haoran Zhao, Penghao Zhao, Ruojun Zhou, Yunhua Zhou, Dongsheng Zhu, Yicheng Zou, Qiye Cai, Xinmeng Che, Jiabei Chen, Jiahao Chen, Jiayi Chen, Yujia Chen, Lizhi Cui, Youheng Dai, Xin Deng, Yi Dong, Shihan Dou, Chenya Gu, Xu Guo, Ding Han, Feiyang Hao, Haotan He, Jie Hou, Binze Hu, Zijian Hu, Junhao Huang, Huicheng Jiang, Jiazhen Jiang, Shufan Jiang, Jiahao Kuang, Bowen Lai, Bo Li, Jiaqiang Li, Peng Li, Qilong Li, Zhuoqun Li, Jiaxiang Liu, Shuainan Liu, Tong Liu, Yi Liu, Zhonghang Lu, Jianwen Luo, Yanyi Luo, Huijie Lv, Ningsheng Ma, Houcheng Min, Chengjun Pan, Qiyuan Peng, Xiaoxuan Peng, Jianmin Qian, Jiantao Qiu, Wanying Ren, Huayu Sha, Jifei Shan, Zixin Shang, Bing Shao, Zhuohui Sheng, Jiayang Shi, Yang Shu, Aierpanjiang Simayi, Sirui Song, Yuxiao Song, Zhe Sun, Zhichao Sun, Wenzhe Tan, Wenhui Tian, Zhongbo Tian, Hanchen Wang, Pengbo Wang, Rui Wang, Yiding Wang, Yuhui Wang, Zhiheng Xi, Caijun Xu, Chao Xu, Yongfeng Xu, Xiaolei Yang, Zhixiong Yang, Qian Yao, Shihong Yi, Yuankai Ying, Jia Yu, Dingbo Yuan, Hao Yuan, Junjie Yuan, Bo Zhang, Caixian Zhang, Qiuyinzhe Zhang, Jiyuan Zhao, Ying Zhao, Pujun Zheng, Xiaoxue Zhong, Xiaohao Zhou, Xinyu Zhou, Guanru Zhu, Yulun Zhu, Yaojie Lu, Tao Ji, Hongyu Lin, Yutao Zhu, Pengfei Cao, Guoxiu He, Xianpei Han, Ben He, Zhicheng Dou, Kang Liu, Qi Zhang, Le Sun, Jun Zhao, Ji-Rong Wen, Xuanjing Huang, Yu-Gang Jiang, Bowen Zhou

Computational Complexity

NP-hardness of ideal lattice problems

Daniel E. Martin

We establish the worst-case hardness of several ideal lattice problems (including SVP and CVP) in the $\ell_2$ norm by providing a dimension-preserving, deterministic polynomial time reduction from their generic lattice versions. The reduction constructs an ideal lattice in the canonical embedding of a number field that approximates some input lattice up to scaling and orthogonal transformation. The integers defining the ideal and the ambient number ring, in particular its discriminant, are all polynomial in bit length relative to the generic input lattice. Furthermore, the ideal is invertible, the ring is monogenic, and the number field is totally real. If the number ring is also required to be a full ring of integers, the reduction conjecturally succeeds in bounded-error quantum polynomial time.
Read the full abstract

Computer Vision

VC-Attention: Value Smoothing and Softmax Casting for Low-bit Attention

Xingyang Li, Dongyun Zou, Shining Zhang et al. (+8 more)

Diffusion Transformers deliver state-of-the-art video generation, but their long spatiotemporal sequences make attention the dominant deployment cost, and a deployable low-bit kernel must be accurate and fast. Accuracy is limited by outliers: a block's quantization scale is set by its largest entries, leaving typical entries confined to a narrow range of representable values. Prior work smooths queries and keys, but value outliers follow no fixed channel or spatiotemporal structure and remain the dominant source of output error. Speed is limited by softmax: low-bit Tensor Cores accelerate only the two matrix multiplications, so the high-precision exponential between them becomes the longest pipeline stage on datacenter GPUs. We propose VC-Attention, a training-free low-bit attention framework that addresses both by pairing Value smoothing with a fused probability Cast. V-Smooth reorders value tokens by lightweight online clustering, so the tokens in a hardware block quantize well together. It quantizes only the residual after subtracting the block mean, and restores that mean from the row sum the online softmax already maintains. ExpCast-FP8 maps log-domain scores directly to E4M3 probability codes with one fused multiply-add, eliminating the FP32 exponential and the format conversion. We implement VC-Attention for B200, B300, H200, RTX PRO 6000, and RTX 5090. Across Wan2.2, LongCat-Video, HunyuanVideo-1.5, and MiniMax-H3, VC-Attention improves fidelity over low-bit baselines, speeds up the attention kernel over BF16 FlashAttention-4 by 1.46-1.59x on datacenter Blackwell and Hopper and by 2.3-3.6x on workstation cards, and generates a clip 1.13-1.19x and 1.36-1.70x faster end to end.
Read the full abstract

All authors Xingyang Li, Dongyun Zou, Shining Zhang, Jiacheng Chen, Haocheng Xi, Lvmin Zhang, Jun-Yan Zhu, Song Han, Zhekai Zhang, Yujun Lin, Muyang Li

Performance

Accelerating Transfer-Learning-Based Autotuning with Predictive LLVM IR Performance Ranking

Md Arafat Hossain, Thomas Randall, Akash Dutta et al. (+3 more)

As the complexity of High Performance Computing (HPC) ecosys- tems continually increases, achieving optimal performance becomes a challenge. Traditional performance autotuning techniques pro- vide promising means to navigate this complexity, these techniques remain computationally intensive and require many evaluations to find optimal configurations. This work proposes an autotuning framework that designs a machine learning-based ensemble LLVM Intermediate Representa- tion (IR) ranker, Neural Configuration Scorer (NCS). NCS ranks the performance of IRs sampled by a transfer-learning-based autotuner, improving the efficiency of the tuning process by reducing tuning overheads and circumventing subpar evaluations. By leveraging knowledge from related tasks, we are able to effectively exploit the transfer relationship to access high-performing configurations in fewer samples than traditional techniques that rely upon itera- tive refinement. Our framework can achieve similar performance improvements as state-of-the-art autotuning techniques with up to 61.67% fewer evaluations, averaging 27.85% fewer evaluations across various HPC benchmarks.
Read the full abstract

All authors Md Arafat Hossain, Thomas Randall, Akash Dutta, Xingfu Wu, Rong Ge, Ali Jannesari

Systems and Control · Cross-listed from math.OC

Revisiting Proximal Bundle Methods: Improved Rates under Hölder Smoothness

Feng-Yi Liao and Yang Zheng

Proximal bundle methods (PBMs) are classical algorithms for nonsmooth convex optimization. Existing analyses of the classical PBM couple the null steps with the descent test. This coupling obscures how the bundle updates approximate the proximal subproblem. In this work, we consider composite objectives $F=f+h$ and view each null-step cycle as an inner bundle subroutine, called $\mathtt{ProxBundle}$. We analyze $\mathtt{ProxBundle}$ independently of any stopping criterion under general bundle model conditions and show that it automatically adapts to Hölder smoothness. Combining this inner-loop analysis with the descent-step analysis yields sharper complexity bounds for the classical PBM. For any fixed proximal parameter, its overall complexity is $\mathcal O\big(\epsilon^{-\frac{3-\nu}{1+\nu}}\big)$ for $\nu\in[0,1)$ and $\mathcal O(\epsilon^{-1})$ for $\nu=1$, where $\nu$ is the Hölder smoothness exponent. Choosing the proximal parameter proportional to $\epsilon$ improves the rate to $\mathcal O\big(\epsilon^{-\frac{2}{1+\nu}}\big)$ for $\nu \in [0,1)$. These are the first guarantees under Hölder smoothness with $\nu\in(0,1)$ for the classical descent test. We further introduce an absolute model-error test. The resulting PBM variant admits a clean inexact proximal-point analysis, and for any fixed proximal parameter, achieves the same complexity $\mathcal O\big(\epsilon^{-\frac{2}{1+\nu}}\big)$. Overall, our analysis separates the roles of the descent and null steps and gives a modular understanding of PBMs across different~tests.
Read the full abstract