Antrittskolloquium von Dr. Feliks Nüske: Data-driven Modeling of Dynamical Systems - Analysis and Model Reduction
Antrittskolloquium von Dr. Feliks Nüske, Forschungsgruppe Daten-basierte Modellierung komplexer physikalischer Systeme, Max-Planck-Institut Magdeburg
- Datum: 12.05.2022
- Uhrzeit: 14:00 - 15:00
- Vortragende(r): Dr. Feliks Nüske
- Leiter der Forschungsgruppe Daten-basierte Modellierung komplexer physikalischer Systeme (DMP) am Max-Planck-Institut Magdeburg
- Ort: Max-Planck-Institut Magdeburg
- Raum: Hybride Veranstaltung: Online und im Großen Seminarraum "Prigogine"
- Kontakt: sek-csc@mpi-magdeburg.mpg.de

Das Kolloquium findet hybrid statt.
40 Plätze im Großen Seminarraum "Prigogine". Es besteht Maskenpflicht im Raum.
Meeting-ID: 994 1575 0761
Kenncode: 887953
Abstract
Data-driven Modeling of Dynamical Systems - Analysis and Model Reduction
Computer simulations of complex dynamical systems are ubiquitous in the natural and engineering sciences. However, efficiently generating sufficient simulation data in order to make reliable predictions for real-world applications - such as molecular systems at atomistic or quantum resolution - remains an open problem. In this context, leveraging the capabilities of modern machine learning techniques has attracted significant attention in recent years.
In this talk, Feliks Nüske will introduce the Koopman operator framework for data-driven modeling of dynamical systems. He will provide an overview of relevant algorithms including illustrative examples. Furthermore, he will explain in more detail how the resulting models can be used to extract slow modes of the dynamical system, as well as reduced models. Finally, he will provide an overview of open research questions for the coming years.