DCAMM seminar - Non-intrusive reduced order models using Gaussian processes

A DCAMM seminar will be presented by

Jan S. Hesthaven

Professor of Mathematics
Chair of Computational Mathematics and Simulation Science
 EPFL, Lausanne, Switzerland



The development of reduced order models for complex applications, offering the promise for rapid and accurate evaluation of the output 
of complex models under parameterized variation, remains a very active research area. Applications are found in problems which require many evaluations, sampled over a potentially large parameter space, found in optimization, control, uncertainty quantification and applications where near real-time response is required.


However, many challenges remain to secure the flexibility, robustness, and efficiency needed for general large-scale applications, in particular for nonlinear and/or time-dependent problems.


After a brief introduction to projection based reduced order models, we discuss the use of data driven Gaussian process regression to enable the development of fast and accurate nonintrusive models for complex problems, including techniques for greedy regression/active learning and error estimation. We illustrate the performance by examples taken from nonlinear mechanics and fluid dynamics.


To enable the modelling of more complex problems, we discuss the development of hybrid element based reduced order model with local nonlinear elements, allowing for the rapid and fast evaluation of the response of large and complex structures with applications in design, uncertainty quantification, and risk assessment.


Time permitting we extend the discussion to the multi-fidelity case where different models are combined through co-kriging and the use of data driven techniques to enhance the predictive accuracy of the reduced models.


This work has been done with in collaboration with M. Guo (EPFL, CH), Z. Zhang (EPFL, CH), M. Kast (ETHZ, CH).

Danish pastry, coffee and tea will e served 15 minutes before the seminar starts.

All interested persons are invited


Mon 18 Nov 19
15:00 - 16:00



Building 101 - meeting room 2 (Ly101-R3.143)
Technical University of Denmark
12 JULY 2020