- This event has passed.
Penn AI Seminar Series: Geometric and Physics Stucture Preservation in Scientific Machine Learning
September 2, 2025 at 12:00 PM - 1:00 PM
Geometric and Physics Stucture Preservation in Scientific Machine Learning
Nat Trask, Associate Professor in Mechanical Engineering and Applied Mechanics at Penn Engineering, will share about his work constructing real-time digital twins built upon a data-driven finite element exterior calculus; constructing auto-regressive integrators with guaranteed long-term stability independent of rollout length; and constructing data-driven particle models built upon metriplectic bracket theory, which preserve emergent statistical mechanics.
Lunch will be served.
Registration is required. Sign up here.
Nat Trask
Associate Professor, Mechanical Engineering and Applied Mechanics
Nat Trask is an associate professor in the department of mechanical engineering at UPenn, holding cross-listings in Applied Mathematics and Computational Science and Material Science Engineering as well as a joint faculty appointment at Sandia National Laboratories. He came to Penn in 2023 after spending 8 years at Sandia as a Principal Member of Technical Staff, developing scientific machine learning and advanced PDE discretization tools for multiscale/multiphysics problems spanning fusion power, fluid-structure driven fracture, semiconductor physics, shock physics, combustion, and multiphase flow. His research focuses on incorporating ideas from geometric dynamics and the finite element exterior calculus to design machine learning architectures which provide notions of trust and physical realizability. He is a recipient of the DOE Early Career award and the NSF-MSPRF award, and currently serves as deputy director for SEA-CROGS, an ASCR MMICCs center for digital twins of climate systems and embedded systems.