WorldSolver: Can LLM Agents Simulate the Physical Dynamics via Solver Generation?
作者: Siru Jiang, Yongzhe Lyu, Shuo Lu, Yubin Wang, Yuxiang Zhang, Yue Liao, Bin Wang, Jian Liang, Tieniu Tan
分类: cs.AI
发布日期: 2026-10-06
🔗 代码/项目: GITHUB
💡 一句话要点
提出WorldSolver以解决物理动态模拟问题
🎯 匹配领域: 支柱九:具身大模型 (Embodied Foundation Models)
关键词: 物理模拟 大型语言模型 求解器生成 智能体评估 计算机图形学 动态系统
📋 核心要点
- 现有的LLM智能体在生成物理动态求解器方面能力不足,难以满足复杂物理现象的模拟需求。
- 本文提出WorldSolver基准,旨在通过168个物理现象模拟任务,评估LLM智能体生成求解器的能力。
- 实验结果显示,尽管GPT-5.6-Sol和Claude-Opus-5表现较好,但整体得分仍低,表明该领域仍需进一步研究与发展。
📝 摘要(中文)
基于大型语言模型(LLM)的智能体在科学和工程问题解决中取得了显著进展,物理模拟作为一种复杂现象的再现测试平台,具有重要的应用价值。本文提出WorldSolver,一个包含168个来自61篇经典计算机图形学论文的物理现象模拟任务的基准,涵盖7个物理领域。每个任务提供固定的模拟环境,智能体需完成求解器的实现。我们从执行检查、视觉逼真度和物理合理性三个维度对智能体进行评估,实验结果表明,生成可执行求解器本身就很困难,而满足视觉和物理正确性更是挑战。GPT-5.6-Sol和Claude-Opus-5的表现相对较好,但总体得分仅为48.7%和46.7%。
🔬 方法详解
问题定义:本文旨在解决LLM智能体在生成物理动态求解器方面的不足,现有方法在复杂物理现象模拟中面临挑战,尤其是在执行和视觉、物理正确性方面的要求。
核心思路:通过构建WorldSolver基准,提供168个物理现象的模拟任务,允许智能体在固定环境中实现求解器,从而评估其生成能力和准确性。
技术框架:WorldSolver的整体架构包括任务生成模块、求解器实现模块和评估模块。任务生成模块提供固定的模拟环境,求解器实现模块由智能体完成,评估模块则从执行、视觉和物理三个维度进行评分。
关键创新:WorldSolver的主要创新在于将物理现象的模拟任务标准化,提供了一个系统化的评估框架,使得不同LLM智能体的性能可以进行有效比较。
关键设计:任务设计中包含固定的代码框架,智能体需在此基础上实现求解器。评估指标包括执行检查、视觉逼真度和物理合理性,确保生成的求解器不仅可执行,还能在视觉和物理上符合预期。
🖼️ 关键图片
📊 实验亮点
实验结果显示,GPT-5.6-Sol和Claude-Opus-5在生成可执行求解器方面表现相对较好,分别获得48.7%和46.7%的总体得分,表明在物理动态模拟中仍存在较大的改进空间。
🎯 应用场景
该研究的潜在应用领域包括机器人、游戏开发和影视特效等,能够推动智能体在复杂物理环境中的应用能力。通过提高LLM智能体的求解器生成能力,未来可能实现更高效的物理模拟,促进相关领域的技术进步。
📄 摘要(原文)
LLM-based agents are increasingly advancing scientific and engineering problem solving, with physics simulation emerging as a challenging yet practical testbed for reproducing complex physical phenomena with application in embodied AI, games and films. As the workhorse of such simulation, a solver computes how the state of a dynamic system evolves over time. Building such solvers requires physical understanding to identify appropriate models, mathematical reasoning to formulate the underlying dynamics, and software engineering to implement them as executable code, yet this capability of LLM agents remains underexplored. To this end, we introduce WorldSolver, a benchmark of 168 simulation tasks derived from physical phenomena in 61 classic computer graphics papers, spanning 7 physical domains. Each task contains a code scaffold that provides a fixed simulation environment for the scene, with the solver implementation left for the agent to complete. Specifically, we evaluate them along three dimensions: Execution Checks for successful execution, Visual Fidelity for reproducing the intended dynamic behavior in the rendered simulation, and Physical Plausibility for physics-grounded verification of the generated dynamics. Experiments on frontier agents reveal that producing executable solvers is difficult itself, and satisfying visual and physical correctness is even harder. GPT-5.6-Sol and Claude-Opus-5 perform comparatively better than the other evaluated agents, yet achieve overall scores of only 48.7% and 46.7%, respectively. WorldSolver is an early step toward agentic solver generation, and we hope it helps drive progress toward agents that can faithfully simulate the dynamic physical world. Code is available at https://github.com/sirujiang/WorldSolver.