Driftless Star is an open-source workflow that computes transport-consistent plasma profiles for a given stellarator boundary and plasma conditions in a unified, reproducible pipeline. It takes in a plasma pressure profile and boundary shape for the magnetic confinement and returns the generated power output. Along the way, it determines the nested flux surfaces needed to achieve magnetic equilibrium, calculates the heat and particle fluxes between the nested surfaces, and uses the fluxes to evolve the pressure profiles which can be used to repeat the loop until the pressure and equilibrium is consistent with each other.
The pipeline also enables AI-accelerated fusion device design by allowing neural surrogates to replace costly physics stages and large language model agents to orchestrate the chain, make judgment calls between stages, and autonomously search for improved geometries.
Raheem Hashmani
PhD student at UW–Madison
UW–Madison • DSI • AMS • CERN
Ph.D. student interested on novel applications of deep learning in high energy physics and astrophysics.