Bio: Charles Poussot-Vassal is Researcher Director with ONERA, the French aerospace lab (Toulouse, France) and co-funder of MOR Digital Systems (http://mordigitalsystems.fr/). His research lies in the wide field of (mostly linear) dynamical model (rational) approximation, control design and analysis. A major part of his activities is dedicated to the development and implementation of numerical scheme and to apply them to a large range of problems including multiple industrial applications. He was born in Grenoble, France, in 1982. In 2005, he completed his Engineering degree and M.Sc. in control and embedded systems from Grenoble INP-ESISAR and Lund University of Technology. In 2008, he completed his Ph.D. degree in control systems theory, within the GIPSA-lab. He obtained his French habilitation, in control and applied mathematics, in 2019 from Toulouse INP. At the beginning of 2009, he worked as a research assistant with the Politecnico di Milano. From mid-2009, he joined ONERA, as a full-time researcher. From July 2020, he is co-funder of MOR Digital Systems, a start-up providing solutions in dynamical model approximation, identification and analysis.
Speech Title: The Loewner Framework for Parametric Systems: Taming the Curse of Dimensionality, for Real...
Abstract: In this keynote we address the problem constructing (reduced complexity) multi-parametric dynamical models, directly from data collected on an experimental set-up or a complex simulator. To do so, we invoke the (interpolatory) Loewner framework, extended to linear parametric systems with an arbitrary number "n" of parameters. The first innovation established, is the construction of data-based realizations for any number of parameters. Second and equally importantly, we show how to alleviate the computational burden, by avoiding the explicit construction of large-scale "n"-dimensional Loewner matrices of size N x N, through a recursive approach. The latter drastically reduces the complexity from O(N^3) to about O(N^1.4), thus taming the curse of dimensionality and making the solution scalable to very large data sets and highly parametric set-up, as faced by engineers. The keynote focuses on the numerical and experimental aspects, with multiple applications used in both the research and industrial contextes, highlighting the versatility and importance of the result in the wide domains of control engineering and (parametric) mechatronics systems.
