Inferring stability properties of chaotic systems on autoencoders' latent spaces

  • 2024-10-23 17:25:36
  • Elise Özalp, Luca Magri
  • 0

Abstract

The data-driven learning of solutions of partial differential equations canbe based on a divide-and-conquer strategy. First, the high dimensional data iscompressed to a latent space with an autoencoder; and, second, the temporaldynamics are inferred on the latent space with a form of recurrent neuralnetwork. In chaotic systems and turbulence, convolutional autoencoders and echostate networks (CAE-ESN) successfully forecast the dynamics, but little isknown about whether the stability properties can also be inferred. We show thatthe CAE-ESN model infers the invariant stability properties and the geometry ofthe tangent space in the low-dimensional manifold (i.e. the latent space)through Lyapunov exponents and covariant Lyapunov vectors. This work opens upnew opportunities for inferring the stability of high-dimensional chaoticsystems in latent spaces.

 

Quick Read (beta)

loading the full paper ...