Post by Jörg Fehr

Running in the morning near the Eiffel Tower, and later speaking at the Collège de France — not a bad way to spend a day in Paris. I had the pleasure to present our work at the international symposium “Mathematical and Applied Aspects of Complexity Reduction Methods”, organised by @Yvon Maday at the Collège de France. My talk was titled: “From Latent Space Representations to Practical Surrogate Models for Structural Dynamical Systems” I presented some of our recent work on combining model order reduction, system identification, and machine learning to obtain efficient and still physically meaningful surrogate models for engineering applications. I presented results from several collaborations with Jonas Kneifl, Johannes Rettberg, J. Nathan Kutz, Paolo Conti, Julius Herb, Manfred Bischoff, Bernard Haasdonk  and others — all contributing in different ways to the goal of bringing these methods closer to practical CAE workflows. For me as an engineer, it was especially enlightening to discuss with researchers who work on the mathematical foundations and proofs behind the methods we use. In engineering, we often need to make decisions based on simulations: to design better, faster, safer, and smarter technical systems. Reliable complexity reduction methods, with controlled accuracy and error estimates, are an important part of making such decisions trustworthy. It was also great to see a strong SimTech - Cluster of Excellence | University of Stuttgart presence from the Universität Stuttgart with Benjamin Stamm, Bernard Haasdonk, and myself contributing to the program. Many thanks to Yvon Maday for organising this wonderful symposium and for creating such an open and inspiring atmosphere for exchange between mathematics, scientific computing, machine learning, and engineering applications. #SimTech #ModelOrderReduction #ReducedOrderModeling #CAE #Simulation #MachineLearning #ScientificComputing #Engineering #UniversityOfStuttgart #CollègedeFrance