Speaker: Michael Churchill, Princeton Plasma Physics Laboratory
Abstract: Digital modelling of the physics and engineering of next-step fusion devices will become increasingly important for their successful design and operation. A classical divide exists between modeling relying on pure simulation and pure experimental scaling laws (the so-called “sim2real” gap). High-fidelity modeling can help to close this gap, but often does not fulfill the speed needed for certain workflows such as design optimization and control room analysis. I will present several efforts and techniques including AI/ML and advanced optimization being pursued towards building out faithful digital twins, based on a range of simulations covering differing physics and levels of fidelity. Examples will range from stellarator design optimization with the StellFoundry SciDAC collaboration, to fast simulation-based inference with experimental diagnostics.