• Home
• Program
• Authors centre
• Important dates
• Social programme & conference dinner
• Committees
• Sponsors, & Exhibitors
• Invited speakers
• Registration
• Practical information
• Contact
• Privacy policy

14:30   Zuiderduinzaal: Keynote: Karen Willcox, Oden Institute for Computational Engineering and Sciences, UT Austin 14:30 45 mins Towards efficient multifidelity modeling for engineering design under uncertainty: From model reduction to scientific machine learning Karen Willcox Abstract: While physics-based simulation has revolutionized engineering design, the computational cost of high-fidelity models means that uncertainty quantification remains a prohibitively expensive task. This is especially true in early-stage design where an engineer wishes to explore many hundreds or thousands of design options over a high dimensional space of uncertain parameters. This talk will describe our recent advances in data-driven model reduction to derive low-dimensional approximations of an underlying high-fidelity model. Our approaches blend the classical perspective of physics-based model reduction together with the flexibility of data-driven machine learning. The approaches are demonstrated for nonlinear systems of partial differential equations arising in various aerospace engineering design applications.