Koopman Theory for Partial Differential Equations

J. Nathan Kutz,Joshua L. Proctor,Steven L. Brunton

DOI: https://doi.org/10.48550/arXiv.1607.07076

2016-07-25

Abstract:We consider the application of Koopman theory to nonlinear partial differential equations. We demonstrate that the observables chosen for constructing the Koopman operator are critical for enabling an accurate approximation to the nonlinear dynamics. If such observables can be found, then the dynamic mode decomposition algorithm can be enacted to compute a finite-dimensional approximation of the Koopman operator, including its eigenfunctions, eigenvalues and Koopman modes. Judiciously chosen observables lead to physically interpretable spatio-temporal features of the complex system under consideration and provide a connection to manifold learning methods. We demonstrate the impact of observable selection, including kernel methods, and construction of the Koopman operator on two canonical, nonlinear PDEs: Burgers' equation and the nonlinear Schrödinger equation. These examples serve to highlight the most pressing and critical challenge of Koopman theory: a principled way to select appropriate observables.

Pattern Formation and Solitons

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to deal with nonlinear partial differential equations (PDEs) through Koopman theory, especially how to select appropriate observables to achieve an accurate approximation of nonlinear dynamical systems. Specifically, the authors explore how to use the Dynamic Mode Decomposition (DMD) algorithm to calculate the finite - dimensional Koopman operator, including its eigenfunctions, eigenvalues and Koopman modes. The paper emphasizes the importance of choosing appropriate observables, because this can not only enable an accurate approximation of the nonlinear behavior of dynamical systems, but also provide a physical explanation of the spatio - temporal characteristics of complex systems and establish a connection with manifold learning methods.

To show the influence of observable selection and the method of constructing the Koopman operator, two classical nonlinear PDE models, namely the Burgers equation and the nonlinear Schrödinger equation, are used as examples in the paper. These examples aim to highlight one of the most urgent and critical challenges faced by Koopman theory: how to select appropriate observables in a principled manner. Through these specific case analyses, the authors hope to provide new perspectives and methods for the dynamical modeling of nonlinear PDEs.