Learning Koopman eigenfunctions of stochastic diffusions with optimal importance sampling and ISOKANN

DOI10.1063/5.0140764arXiv2301.00065OpenAlexW4390789263WikidataQ130103850 ScholiaQ130103850MaRDI QIDQ6188560FDOQ6188560

Authors: A. Sikorski, Enric Ribera Borrell, Marcus Weber 

Publication date: 7 February 2024

Published in: Journal of Mathematical Physics (Search for Journal in Brave)

Abstract: For stochastic diffusion processes the dominant eigenfunctions of the corresponding Koopman operator contain important information about the slow-scale dynamics, that is, about the location and frequency of rare events. In this article, we reformulate the eigenproblem in terms of $chi$-functions in the ISOKANN framework and discuss how optimal control and importance sampling allows for zero variance sampling of these functions. We provide a new formulation of the ISOKANN algorithm allowing for a proof of convergence and incorporate the optimal control result to obtain an adaptive iterative algorithm alternating between importance sampling and $chi$-function approximation. We demonstrate the usage of our proposed method in experiments increasing the approximation accuracy by several orders of magnitude.

Full work available at URL: https://arxiv.org/abs/2301.00065

Recommendations

• A Koopman framework for rare event simulation in stochastic differential equations
• A data-driven approximation of the koopman operator: extending dynamic mode decomposition
• Data-driven approximation of the Koopman generator: model reduction, system identification, and control
• Eigendecompositions of transfer operators in reproducing kernel Hilbert spaces
• Koopman operator, geometry, and learning of dynamical systems

Mathematics Subject Classification ID

Molecular physics (81V55) Computational molecular dynamics in statistical mechanics (82M37)

Cites Work

• Metastability in reversible diffusion processes. I: Sharp asymptotics for capacities and exit times
• The Monte Carlo computation error of transition probabilities
• On the numerical approximation of the Perron-Frobenius and Koopman operator
• Fuzzy spectral clustering by PCCA+: application to Markov state models and data classification
• Importance sampling in path space for diffusion processes with slow-fast variables
• Solving high-dimensional Hamilton-Jacobi-Bellman PDEs using neural networks: perspectives from the theory of controlled diffusions and measures on path space
• A Koopman framework for rare event simulation in stochastic differential equations

Cited In (1)

• Cascades of scales: applications and mathematical methodologies

This page was built for publication: Learning Koopman eigenfunctions of stochastic diffusions with optimal importance sampling and ISOKANN

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6188560)