Springe direkt zu Inhalt

MSDI3: Learning governing equations of dynamical systems from data

Mar 02, 2021 | 11:00 AM - 04:30 PM

Chairs: Stefan Klus, Christof Schuette, Wei Zhang

Over the last years, various data-based methods have been developed for learning governing equations of dynamical systems in diverse areas of science and engineering. In particular, the sparse identification of nonlinear dynamical systems (SINDY) method uses sparsity regression to find the fewest terms in the governing equation that match the data, which has drawn considerable attentions and has been successfully applied to nonlinear dynamical systems in mechanics, fluid, biology, etc.

Besides, methods based on Koopman operator, tensor or machine learning techniques have also demonstrated their ability in learning or inferring nonlinear complex dynamics from data. This minisymposium aims to bring together experts from different fields in order to discuss recent advances in this direction, including both new computational algorithms and their successful applications in different areas.

Find abstracts for all talks here or linked individually below.


Alex Mauroy (University of Namur, Belgium):

Linear identification of nonlinear ODEs and PDEs via Koopman operator theory


Zoltan Tuza (Imperial College London, GB)

ODE composer - a versatile software tool to compose ODE models from time-series data


Jan-Hendrik Niemann (ZIB, Germany):

Data-driven model reduction of agent-based systems using the Koopman generator

12:30 Open Discussion
13:00 Lunch Break

Patrick Gelß (FU Berlin, Germany)

Tensor- and kernel-based recovery of dynamical systems

Alex Goeßmann (TU Berlin, Germany)

Recovery of tensor-structured functions: Scalable approaches in high dimensions

Hayden Schaeffer (Carnegie Mellon University, USA)

Function Approximation via Sparse Random Features

Open Discussion