Perceptual Self-Reflection in Agentic Physics Simulation Code Generation
This addresses the 'oracle gap' in physics simulation code generation for engineering workflows and data generation pipelines, though it is incremental as it builds on existing multi-agent and vision-language model approaches.
The paper tackles the problem of generating physics simulation code from natural language descriptions by introducing a multi-agent framework with perceptual self-reflection, which validates rendered animation frames using a vision-language model instead of code inspection, achieving substantial improvements over single-shot baselines with robust pipeline stability at approximately $0.20 per animation.
We present a multi-agent framework for generating physics simulation code from natural language descriptions, featuring a novel perceptual self-reflection mechanism for validation. The system employs four specialized agents: a natural language interpreter that converts user requests into physics-based descriptions; a technical requirements generator that produces scaled simulation parameters; a physics code generator with automated self-correction; and a physics validator that implements perceptual self-reflection. The key innovation is perceptual validation, which analyzes rendered animation frames using a vision-capable language model rather than inspecting code structure directly. This approach addresses the ``oracle gap'' where syntactically correct code produces physically incorrect behavior--a limitation that conventional testing cannot detect. We evaluate the system across seven domains including classical mechanics, fluid dynamics, thermodynamics, electromagnetics, wave physics, reaction-diffusion systems, and non-physics data visualization. The perceptual self-reflection architecture demonstrates substantial improvement over single-shot generation baselines, with the majority of tested scenarios achieving target physics accuracy thresholds. The system exhibits robust pipeline stability with consistent code self-correction capability, operating at approximately \$0.20 per animation. These results validate our hypothesis that feeding visual simulation outputs back to a vision-language model for iterative refinement significantly outperforms single-shot code generation for physics simulation tasks and highlights the potential of agentic AI to support engineering workflows and physics data generation pipelines.