TopOptAgents uses six LLM agents to self-refine topology optimization

Researchers from Seoul National University have introduced TopOptAgents, a multi-agent framework that leverages six LLM-based agents to fully automate topology optimization—a design method that distributes material to meet objectives and constraints. The agents collaborate in iterative self-refinement cycles covering problem formulation, validation, code generation and execution, and quality assessment. This enables error correction and progressive improvement of both the optimization setup and the resulting design, addressing decisions that previously required domain expertise.

TopOptAgents was tested on optimization problems with varying levels of literature coverage and numerical characteristics. The framework showed the most benefit for problem classes where pretrained language models have limited prior exposure, such as formulations with sparse open-source implementations. In these cases, TopOptAgents reliably produced converged designs where a single state-of-the-art LLM struggled, suggesting that self-refinement broadens the range of topology optimization problems that LLM-based automation can reliably address.