Computational Methods in Fusion Science and Engineering
Friday, September 14, 2018
10AM – 11AM
The ambition of building net energy-producing fusion reactors remains one of the great unfulfilled challenges of modern science and engineering. One reason for this is that the computational plasma (e.g. fully ionized gas) physics and material science (i.e. plasma-material interaction) that govern the behavior of these machines is extremely complicated, requiring expensive numerical phase-space solutions that span multiple spatial and temporal scales. Because of this, techniques are being developed and integrated into capabilities for reducing the simulation complexity of these systems, while simultaneously preserving essential features of interest that inform prediction. To address some of these challenges, we will discuss techniques in uncertainty quantification, multifidelity modeling, structure preservation, and machine learning that we are currently applying to this field.
Craig Michoski is a Research Scientist at the Center for Predictive Engineering and Computational Science within the Institute of Computational Engineering and Sciences at UT-Austin. He received his Ph.D. in Chemistry and Mathematics from UT in 2009. He joined ICES as a Postdoctoral Researcher shortly after and has been at UT since that time. His research interests include method development, optimization, machine learning, geometric methods, and uncertainty quantification.