Date: Oct 2, 2024, Wednesday

Time: 11:30-12:45pm ET
Venue: hybrid, in-person Monarch Hall 1113B,
 
Title: Data-driven Reduced Order Modeling for Complex Physical Systems
Speaker: Xuping Xie (ODU)
 
 
Abstract: Many complex physics applications and engineering design processes often require models that capture the predictive power of first-principles simulations yet are computationally less demanding by many orders of magnitude. Reduced order modeling (ROM) provides an efficient solution, striking a balance between high-fidelity simulations and accurate surrogate models. Artificial Intelligence (AI), promises a revolution in how physics and engineering can be bridged for authentic predictive control and design of engineering systems with ROM. Our work focuses on developing efficient ROM techniques, combined mathematical principles, and scientific machine learning (SciML) methods, to enable predictive design and control in complex systems such as fluids and plasma physics. In this talk, I will introduce contemporary ROM approaches for nonlinear systems in fluids and plasma physics. I will discuss the data-driven ROM closure strategies for turbulent flows, and non-intrusive deep reduced physics model with autoencoder for fusion disruption mitigation.
 
Bio: Dr. Xie is an Assistant Professor in the Department of Mathematics & Statistics at ODU and a guest scientist at Los Alamos National Laboratory (LANL). Before joining ODU, he was a postdoctoral researcher at LANL in the Applied Mathematics and Plasma Physics (T-5) Group. He received a Ph.D. in Mathematics from Virginia Tech. His research focuses on scientific machine learning and AI for physical sciences, with an emphasis on data-driven discovery of latent dynamics and applications in plasma physics and fluid dynamics.