Prof. Andrew Christlieb. Using Machine Learning to Solve First Principle Models in Plasma Physics

Electrical power generation from fusion plasmas is one of the grand technological challenges that, when met, will lead to sustainable, carbon-free processes. Fusion reactors are tremendously complex and so computer simulations are needed to help optimize their designs. Developing computer programs to simulate these complex systems is a daunting task due to the large dynamic range in spatial scale (from microns to many meters) and time scale (from picoseconds to minutes) that must be addressed. The computational representation of fusion plasma reactors also requires kinetics approaches where, in principle, the trajectory of each electron and ion in the plasma are individually computed. In this video, Prof. Andrew Christlieb of Michigan State University describes new computational methods (from fundamental physics based to artificial intelligence) that are being developed to simulate fusion reactor systems, and bring sustainable electricity from fusion a step close.