Linear quadratic control of nonlinear systems with Koopman operator learning and the Nyström method

Abstract

In this paper, we study how the Koopman operator framework can be combined with kernel methods to effectively control nonlinear dynamical systems. While kernel methods have typically large computational requirements, we show how random subspaces (Nyström approximation) can be used to achieve huge computational savings while preserving accuracy. Our main technical contribution is deriving theoretical guarantees on the effect of the Nyström approximation. More precisely, we study the linear quadratic regulator problem, showing that the approximated Riccati operator converges at the rate $m^{-1/2}$, and the regulator objective, for the associated solution of the optimal control problem, converges at the rate $m^{-1}$, where $m$ is the random subspace size. Theoretical findings are complemented by numerical experiments corroborating our results.

References in corpus (16)

• Linear predictors for nonlinear dynamical systems: Koopman operator meets model predictive control
• Kernel Mean Embedding of Distributions: A Review and Beyond
• Learning Decentralized Controllers for Robot Swarms with Graph Neural Networks
• Koopman operator-based model reduction for switched-system control of PDEs
• Eigendecompositions of Transfer Operators in Reproducing Kernel Hilbert Spaces
• Recursive Sampling for the Nyström Method
• Kernel-based approximation of the Koopman generator and Schrödinger operator
• Koopman spectra in reproducing kernel Hilbert spaces
• Certainty Equivalence is Efficient for Linear Quadratic Control
• Scalable Extended Dynamic Mode Decomposition using Random Kernel Approximation
• Deep Koopman Learning of Nonlinear Time-Varying Systems
• Efficient Approximation of Molecular Kinetics using Random Fourier Features
• Learning Dynamical Systems via Koopman Operator Regression in Reproducing Kernel Hilbert Spaces
• Koopman Kernel Regression
• Estimating Koopman operators with sketching to provably learn large scale dynamical systems
• Sketch In, Sketch Out: Accelerating both Learning and Inference for Structured Prediction with Kernels

Cited by in corpus (1)

• An overview of Koopman-based control: From error bounds to closed-loop guarantees