Overview and details of the sessions of this conference. Please select a date or location to show only sessions at that day or location. Please select a single session for detailed view (with abstracts and downloads if available).
PL3: "Nonlinear manifold approximations for reduced-order modeling of nonlinear systems" Karen E. Willcox (UT Austin)
Time:
Tuesday, 19/Mar/2024:
12:00pm - 1:00pm
Session Chair: Heike Faßbender
Location:G26/H1
Lecture Hall 1 in Building 26; size: 572
Presentations
Enforcing physics structure in scientific machine learning: The role of projection-based reduced-order modeling
K. E. Willcox
University of Texas at Austin
Surrogate modeling plays a critical role in achieving design, control and uncertainty quantification for complex systems. It is also a key enabling technology for predictive digital twins. This talk discusses recent work that combines classical ideas from projection-based reduced-order modeling with a data-driven scientific machine learning perspective. The result is a scalable, non-intrusive, physics-informed approach to surrogate modeling. Rather than learn a generic approximation with weak enforcement of the physics, as in other machine learning approaches, we learn low-dimensional operators of a dynamical system whose structure is defined -- through the lens of projection -- by the physical problem being modeled. The talk will highlight the importance of embedding this physics structure, especially for challenging problems in engineering where training data are sparse and expensive to acquire.