Driftless Star: Transforming Stellarator Design for Fusion Energy

PI: Rogerio Jorge, assistant professor of Physics

Co-PI:

Misha Khodak, assistant professor of Computer Sciences

Co-Investigators:

Qin Li, professor of Mathematics

Adelle Wright, assistant professor of Nuclear Engineering & Engineering Physics

Stephen Wright, professor of Computer Sciences

Kyle Cranmer, professor of Physics and Data Science Institute Director

Description: One of the most promising devices for attaining safe, consistent, and carbon-free energy from nuclear fusion is the stellarator, which confines hot plasma using intricately shaped magnetic fields. While public and private investment in stellarators is growing, today, no tool can take a proposed design and reliably predict how well it works. Instead, different performance aspects are evaluated by a sequence of experts using disconnected software, which slows discovery, reproducibility, and scalability.

Driftless Star, an integrated stellarator workflow, will enable robust verification and AI-driven acceleration. Using this pipeline, the team will deploy modern optimization and machine learning tools to design stellarators under production constraints. The impact of the project will be accelerated via a public competition benchmark in which scientists worldwide can test and compare approaches, bringing together multiple research communities around the grand challenge of practical fusion and positioning UW–Madison as a leader in stellarator design.