Title: Advances and opportunities for data-driven design and analysis of experiments at Sandia’s Z Pulsed Power Facility

Abstract: Sandia’s Z machine is the world’s largest pulsed power facility, capable of compressing electrical current in space and time to deliver over 20 million amperes of peak current with O(100ns) rise time to a variety of targets for high- energy density science experiments. Due to the extreme environments created, relatively rare nature of experiments, and costly multi-physics simulation tools used for design, it is critical that data-driven methods are increasingly being incorporated in the full experiment lifecycle. In this seminar, I will cover sufficient background on the Z machine and the Magnetized Liner Inertial Fusion (MagLIF) concept to introduce several challenges for problems ranging from design to analysis of experiments. I will present several exemplars for the MagLIF platform that utilize machine learning and statistical methods to address many of these challenges. This will include work from a recently concluded laboratory directed research and development project in collaboration with Professor Roshan Joseph at the Georgia Institute of Technology H. Milton Stewart School of Industrial and Systems Engineering. I will conclude by highlighting additional areas of interest for further growth of data science and means for collaboration within the Pulsed Power center at Sandia. SAND2023-10723A

Bio: Dr. William Lewis is a physicist and applied data scientist in the radiation & ICF target design group in Sandia’s Pulsed Power Sciences Center. He received a BS in Physics and Mathematics in 2012 from the University of Arkansas, where he conducted research in theoretical laser physics. He went on to obtain an MS in 2015 working in the field of experimental nano-spectroscopy and a Ph.D. in 2018 studying transport theory of strongly interacting cold quantum gases, both from the University of Colorado Boulder. William joined Sandia as a postdoctoral researcher in June of 2019. During his tenure as a postdoc, he applied a variety of data-driven methods including Bayesian statistics and machine-learning to significantly improve understanding of the conditions produced in pulsed power inertial confinement fusion (ICF) experiments. William converted to permanent staff in December 2021. As a staff member, he has continued to develop and regularly publish methods allowing for large-scale surveys of experimental and simulation based ICF data.