Perceptual Self-Reflection in Agentic Physics Simulation Code Generation

TL;DR

This paper introduces a multi-agent system with perceptual self-reflection for generating accurate physics simulation code from natural language, significantly improving over traditional methods by analyzing rendered animations with vision-language models.

Contribution

The novel perceptual self-reflection mechanism enables iterative validation and correction of physics simulations using visual analysis, addressing the oracle gap in code correctness.

Findings

Substantial improvement in physics accuracy across seven domains.

Robust code self-correction and pipeline stability.

Cost-effective simulation generation at approximately $0.20 per animation.

Abstract

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.