Automated Stellarator Design Using Open-Source Magnet Codes and AI Optimization

Recent advancements in stellarator design have enabled the development of fully automated workflows that integrate key open-source tools to conceptualize entire devices from a few input parameters. This presentation will demonstrate how tools such as STELLOPT, REGCOIL, FOCUS, and VMEC are employed in a comprehensive workflow to optimize stellarator designs. Beginning with a high-level parameter input—such as periodicity, major radius, and radial build—these workflows automate the generation of magnetic coil configurations, evaluate plasma equilibria, and produce detailed CAD models for further analysis. By leveraging these tools in combination with AI-based surrogate modeling and multi-objective optimization, it is now possible to achieve rapid exploration of design spaces, balancing physics performance, engineering constraints, and cost. Poincaré plot generation, magnetic field evaluations, and coil optimization are integrated to ensure precise reproduction of desired plasma configurations while adhering to practical engineering requirements. Case examples will highlight how these workflows reduce design iteration cycles and computational costs, paving the way for faster and more scalable stellarator development. This automated approach democratizes access to stellarator design tools, enabling both research institutions and private fusion companies to accelerate progress toward commercially viable fusion reactors.