LLM-in-the-Loop Topology Optimization for Autonomous Microstructural Material Design

  • Ogawa, Shun (The University of Tokyo)
  • Suzuki, Katsuyuki (The Univerisity of Tokyo)

Please login to view abstract download link

This study proposes a framework in which agents based on large language models (LLMs) autonomously derive optimal material structures. Topology optimization is employed for exploring material structures. The agents treat topology optimization as a tool and consistently automate the entire workflow, including the formulation of optimization problems, generation of required programs, extension of libraries, execution of optimization calculations, and evaluation of the results. This enables the formulation of design requirements expressed in natural language, which has been difficult in conventional approaches, and allows the automation of the optimization design process traditionally performed by human designers.