L. Grüne. Curse-of-dimensionality-free Lyapunov functions via ISS small-gain & neural networks

Talk at the Online Seminar on Input-to-State Stability and its Applications

Speaker:
Lars Grüne (University of Bayreuth)

Title:
Curse-of-dimensionality-free computation of (control-)Lyapunov functions via ISS small-gain techniques and neural networks

Abstract:
The literature reports a large number of computational methods for Lyapunov and control Lyapunov functions. For most of these approaches the computational effort grows exponentially in the state space dimension, making them infeasible even for moderately high-dimensional problems. In this talk we show that this so-called curse of dimensionality can be overcome for Lyapunov functions with a suitable separable structure, as they are provided by ISS small-gain theory.

The talk is related to the following papers:

Date:
Wednesday, 31st May 2023