On this Invited Discuss from SC19, Karen Willcox from the College of Texas presents: Predictive Knowledge Science for Bodily Techniques: From Mannequin Discount to Scientific Machine Studying.
Reaching predictive knowledge science for bodily methods requires a synergistic mixture of knowledge and physics-based fashions, in addition to a crucial must quantify uncertainties. For a lot of frontier science and engineering problem issues, a purely data-focused perspective will fall brief — these issues are characterised by complicated multi-scale multi-physics dynamics, high-dimensional unsure parameters that can’t be noticed immediately, and a must challenge predictions that transcend the precise circumstances the place knowledge could also be accessible. Studying from knowledge by means of the lens of fashions is a option to convey construction to an in any other case intractable downside: it’s a option to respect bodily constraints, to embed area data, to convey interpretability to outcomes, and to endow the ensuing predictions with quantified uncertainties. This discuss highlights how physics-based fashions and knowledge collectively unlock predictive modeling approaches by means of two examples: first, constructing a Digital Twin for structural well being monitoring of an unmanned aerial car, and second, studying low-dimensional fashions to hurry up computational simulations for design of next-generation rocket engines.
At SC Invited Talks, you’ll hear how translational analysis and applied sciences and their purposes handle a number of the most complicated challenges of our time. These talks will present insights with a broad context and from a longer-term perspective.
Karen E. Willcox is Director of the Oden Institute for Computational Engineering and Sciences, and a Professor of Aerospace Engineering and Engineering Mechanics on the College of Texas at Austin. She holds the W. A. “Tex” Moncrief, Jr. Chair in Simulation-Based mostly Engineering and Sciences and the Peter O’Donnell, Jr. Centennial Chair in Computing Techniques. Earlier than becoming a member of the Oden Institute in 2018, she spent 17 years as a professor on the Massachusetts Institute of Know-how, the place she served because the founding Co-Director of the MIT Heart for Computational Engineering and the Affiliate Head of the MIT Division of Aeronautics and Astronautics. Previous to becoming a member of the MIT school, she labored at Boeing Phantom Works with the Blended-Wing-Physique plane design group. Her analysis has produced scalable computational strategies for design of next-generation engineered methods, with a specific deal with mannequin discount as a option to study principled approximations from knowledge and on multi-fidelity formulations to leverage a number of sources of unsure info. She is a Fellow of the Society for Industrial and Utilized Arithmetic (SIAM) and Fellow of the American Institute of Aeronautics and Astronautics (AIAA).
See our full protection of SC19
Take a look at our insideHPC Occasions Calendar