代码大模型 / AI 编程

1510.06108 2026-06-04 eess.SY cs.SY 50%

A Counter-Example Guided Framework for Robust Synthesis of Switched Systems Using Control Certificates

一种用于切换系统鲁棒综合的反例引导框架使用控制证书

Hadi Ravanbakhsh, Sriram Sankaranarayanan

专题命中代码评测：program synthesis(abstract)

AI总结本文提出通过生成

Comments The paper has been withdrawn by the author. The main reason is incorrect proof of Theorem 4 and some incorrect results in the evaluation parts due to some bugs in the code

2603.13384 2026-06-04 cs.SE cs.AI 81%

VulnAgent-R2: Evidence-Calibrated Multi-Agent Auditing for Repository-Level Vulnerability Detection

VulnAgent-R2: 证据校准的多智能体审计用于仓库级漏洞检测

Renwei Meng, Haoyi Wu, Jingming Wang

机构 * stu.ahu.edu.cn（安徽大学）

专题命中仓库级理解：repository(title,abstract)；分类 cs.SE、cs.AI

AI总结提出VulnAgent-R2，一个预算感知的多智能体审计框架，通过反事实证据重加权、构建感知验证计划合成和成本风险帕累托调度器，在仓库级漏洞检测中提升F1和AUROC，并降低在线令牌消耗。

Comments 13 pages, 4 figures

2606.04903 2026-06-04 cs.LO cs.AI cs.MA cs.PL 62%

Provably Auditable and Safe LLM Agents from Human-Authored Ontologies

基于人类编写本体的可审计且安全的LLM智能体

Aaron Sterling

机构 * Thistleseeds

专题命中仓库级理解：repository(abstract)；分类 cs.AI、cs.PL

AI总结提出Agentic Redux架构，通过类型化λ演算证明其在适当领域上的执行语义正确且决策可审计，并引入本体优先的智能体设计方法。

2606.04986 2026-06-04 cs.CV 50%

Food-R1: A Unified Multi-Task Food Vision-Language Model with Reinforcement Learning

Food-R1: 一种基于强化学习的统一多任务食品视觉语言模型

Yu Zhu, Yongkang Li, Wenjie Zhu, Haoyi Jiang, Wenyu Liu, Wei Yang, Bin Li, Xinggang Wang

机构 * Huazhong University of Science and Technology（华中科技大学）

专题命中仓库级理解：repository(abstract)

AI总结针对现有食品视觉语言模型依赖监督微调导致推理和泛化能力受限以及营养标注稀缺的问题，提出包含链式思维标注的大规模基准CalorieBench-80K和基于强化微调（GRPO）的统一多任务食品视觉语言模型Food-R1，在食品相关任务上持续超越强基线。

1902.01328 2026-06-04 eess.SY cs.SY 50%

Optimization-based Feedback Manipulation Through an Array of Ultrasonic Transducers

基于超声换能器阵列的优化反馈操控

Josef Matouš, Adam Kollarčík, Martin Gurtner, Tomáš Michálek, Zdeněk Hurák

专题命中仓库级理解：repository(abstract)

AI总结本文提出了一种新型实验室实验平台，用于通过声压非接触平面操控（定位）毫米级物体，物体要么漂浮在水面上，要么在固体表面滚动。通过8x8换能器阵列实时形变声压场，利用平方电压驱动，通过周期性更新相位差实现位置反馈控制，采用数值优化确定电压相位差，该平台可作为低成本测试平台用于多执行器非接触操控算法的开发。

Comments final version before submission to IFAC Mechatronics 2019 conference

1903.10620 2026-06-04 math.OC cs.CR cs.SY eess.SY 50%

An Optimal Graph-Search Method for Secure State Estimation

一种用于安全状态估计的最优图搜索方法

Xusheng Luo, Miroslav Pajic, Michael M. Zavlanos

专题命中仓库级理解：repository(abstract)

AI总结本文提出了一种新的最优图搜索算法，用于在大规模线性时不变系统中正确识别恶意攻击并安全估计状态，同时优于现有算法在最优性和执行时间方面。

Comments 16 pages, 10 figures

1810.00615 2026-06-04 math.NA cs.NA 50%

A Note on Parallel Preconditioning for All-At-Once Evolutionary PDEs

关于所有一次演化PDEs并行预条件的注记

Anthony Goddard, Andrew Wathen

专题命中仓库级理解：repository(abstract)

AI总结本文研究提出了一种并行预条件方法，用于解决所有一次演化PDEs的预条件问题，并通过Neumann级数方法扩展了预条件器，实现了高效的并行执行。

Comments 13 pages, 3 figures

1506.05957 2026-06-04 math.NA cs.NA physics.comp-ph 50%

Inexact Krylov iterations and relaxation strategies with fast-multipole boundary element method

不精确的Krylov迭代与松弛策略与快速多极边界元方法

Tingyu Wang, Simon K. Layton, Lorena A. Barba

专题命中仓库级理解：repository(abstract)

AI总结本文提出基于快速多极边界元方法的不精确Krylov迭代与松弛策略，通过降低多项式阶数提升计算效率，对拉普拉斯和斯托克斯问题实现2.1至4.0倍的加速。

Comments 21 pages, 20 figures. Second version submitted for peer review on March 2016, with all results re-computed and revised author list. Rejected in October 2016. Currently undergoing revision for a third submission. See progress of open revision in https://github.com/barbagroup/inexact-gmres

1602.07033 2026-06-04 math.AP cs.NA math.NA q-bio.PE 50%

A numerical framework for computing steady states of size-structured population models and their stability

计算大小结构化种群模型稳态及其稳定性的数值框架

Inom Mirzaev, David M. Bortz

专题命中仓库级理解：repository(abstract)

AI总结本文提出一种计算一般演化方程稳态近似解的数值框架，用于研究稳态的存在性和稳定性，并通过线性Sinko-Streifer模型和非线性种群平衡方程验证其有效性。

1409.1465 2026-06-04 math.NA cs.NA 50%

Multilinear PageRank

多线性PageRank

David F. Gleich, Lek-Heng Lim, Yongyang Yu

专题命中仓库级理解：repository(abstract)

AI总结本文将PageRank扩展到高阶马尔可夫链，提出多线性PageRank作为可计算的近似方法，基于空间随机漫步模型，探讨了其收敛性理论及非收敛案例。

1801.03833 2026-06-04 cs.LO cs.SY eess.SY 50%

Experiments in Verification of Linear Model Predictive Control: Automatic Generation and Formal Verification of an Interior Point Method Algorithm

线性模型预测控制的验证实验：内点法算法的自动生成与形式验证

Guillaume Davy, Eric Féron, Pierre-Loïc Garoche, Didier Henrion

专题命中程序分析与验证：code generation(abstract)

AI总结本文研究了线性模型预测控制中内点法算法的自动生成与形式验证，通过代码专门化阶段生成额外的注释来形式化算法的意图规范，并利用演绎方法自动证明这些断言的有效性，同时通过SMT求解器验证整个证明过程。

Journal ref 22nd International Conference on Logic for Programming Artificial Intelligence and Reasoning (LPAR-22), Nov 2018, Awassa, Ethiopia. https://easychair.org/smart-program/LPAR-22/

2606.03482 2026-06-04 math.NT cs.IT math.CO math.IT math.MG 50%

Majorization and Gaussian-Mass Maximality for Construction-A Lattices from Binary Self-Dual Codes

从二元自对偶码构造的Construction-A格子的优超性与高斯质量极大性

Scott Duke Kominers

专题命中其他AI编程：code model(abstract)

AI总结本文证明，对于由二元自对偶码生成的所有幺模Construction-A格子，整数格在给定秩的整格子中最大化高斯质量，通过将theta级数不等式简化为码的优超性陈述，并利用凸序和Jensen不等式。

Comments 8 pages; v2: fixed formatting typo in metadata

1. 代码评测 7 篇

A Counter-Example Guided Framework for Robust Synthesis of Switched Systems Using Control Certificates

2. 仓库级理解 9 篇

VulnAgent-R2: Evidence-Calibrated Multi-Agent Auditing for Repository-Level Vulnerability Detection

Provably Auditable and Safe LLM Agents from Human-Authored Ontologies

Food-R1: A Unified Multi-Task Food Vision-Language Model with Reinforcement Learning

Optimization-based Feedback Manipulation Through an Array of Ultrasonic Transducers

An Optimal Graph-Search Method for Secure State Estimation

A Note on Parallel Preconditioning for All-At-Once Evolutionary PDEs

Inexact Krylov iterations and relaxation strategies with fast-multipole boundary element method

A numerical framework for computing steady states of size-structured population models and their stability

Multilinear PageRank

3. 程序分析与验证 1 篇

Experiments in Verification of Linear Model Predictive Control: Automatic Generation and Formal Verification of an Interior Point Method Algorithm

4. 其他AI编程 1 篇

Majorization and Gaussian-Mass Maximality for Construction-A Lattices from Binary Self-Dual Codes