Data-Adaptive Learning of Dynamical Systems by Matching Transfer Operators and Invariant Measures

DOI10.48550/arXiv.2607.00391MaRDI QIDQ7270499zbMATH OpenarXivFDO

Authors Jonah Botvinick-Greenhouse, Yunan Yang, Yinong Huang

Publication date 1 July 2026

Published in arXiv

Tldr We learn dynamical system models by approximating the Perron--Frobenius transition operator with an Ulam matrix adapted from data. This allows us to train via gradient descent on a loss that matches invariant measures or transport between regions of state space, improving robustness against noise and sparse sampling.

Full work available at URL https://doi.org/10.48550/arXiv.2607.00391

Mathematics Subject Classification ID

Time series analysis of dynamical systems (37M10) Computational methods for ergodic theory (approximation of invariant measures, computation of Lyapunov exponents, entropy, etc.) (37M25) Optimal transportation (49Q22) System identification (93B30)

arXiv classification

math.DSmath.NA

Cites work

• An adaptive subdivision technique for the approximation of attractors and invariant measures
• Data-driven Computation of Molecular Reaction Coordinates
• Discovering governing equations from data by sparse identification of nonlinear dynamical systems
• Estimating physical invariant measures and space averages of dynamical systems indicators
• Finite approximation for the Frobenius-Perron operator. A solution to Ulam's conjecture
• Finite approximations of Frobenius-Perron operators. A solution of Ulam's conjecture to multi-dimensional transformations
• Invariant Measures for Data-Driven Dynamical System Identification: Analysis and Application
• Invariant Measures in Time-Delay Coordinates for Unique Dynamical System Identification
• Learning dynamics on invariant measures using PDE-constrained optimization
• Monte Carlo and quasi-Monte Carlo methods
• Neural Ordinary Differential Equations
• On the numerical approximation of the Perron-Frobenius and Koopman operator
• Optimal partition choice for invariant measure approximation for one-dimensional maps
• Optimal Transport for Parameter Identification of Chaotic Dynamics via Invariant Measures
• PageRank beyond the web
• Stability of temporal statistics in Transition Path Theory with sparse data
• Stabilized neural ordinary differential equations for long-time forecasting of dynamical systems

This page was built for publication: Data-Adaptive Learning of Dynamical Systems by Matching Transfer Operators and Invariant Measures

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7270499)