Learning Closure of Dynamical Systems with Kernel Ridge Regression
Abstract
We develop a closure modeling framework for identifying missing components of dynamical systems using Kernel Ridge Regression (KRR). The framework addresses two classes of closure problems: difference-equation closures arising in ODE and PDE settings, and algebraic closures arising from moment closure in kinetic equations. For the first class, we derive an error bound in an ODE setting that quantifies contributions from time integration, approximation of unresolved scales, and interpolation required to couple unresolved-scale effects to the resolved solver. Numerical experiments on the Lorenz-63 system and the Kuramoto-Sivashinsky equation demonstrate accurate long-horizon predictions and substantial improvements over an LSTM-based closure model. For the second class, we consider moment closure for a one-dimensional kinetic equation by modeling discrepancies between kinetic and macroscopic fluxes as a function of the resolved macroscopic variables. We compare global KRR models based on PCA coordinates with spatially local models. While the global model performs well for unimodal initial conditions, its accuracy deteriorates for bimodal initial conditions. Spatially local models with appropriate modeling inputs improve robustness and achieve higher predictive accuracy.
Related
- Efficient Closure Learning in Dynamical Systems: A Comparison of Data-Assimilated and Embedded Approaches
- FluxDisco: Symbolic Regression for Stoichiometric Dynamical Systems via Monte Carlo Graph Search
- Koopman operator learning for predictive control via Khatri-Rao kernel regression
- Nonparametric In-Context Learning and Inference via Measure-Indexed Kernel Ridge Regression
- Learning between the peaks: sharp asymptotics for kernel ridge regression under power-law anisotropy
BibTeX
@misc{habbershaw2026learning,
title = {{Learning Closure of Dynamical Systems with Kernel Ridge Regression}},
author = {Evan Habbershaw and John Harlim and Senwei Liang},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.02564},
eprint = {2610.02564},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.02564},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Learning Closure of Dynamical Systems with Kernel Ridge Regression
AU - Evan Habbershaw
AU - John Harlim
AU - Senwei Liang
PY - 2026
DA - 2026/10/01
PB - arXiv
DO - 10.48550/arxiv.2610.02564
UR - https://arxiv.org/abs/2610.02564
AB - We develop a closure modeling framework for identifying missing components of dynamical systems using Kernel Ridge Regression (KRR). The framework addresses two classes of closure problems: difference-equation closures arising in ODE and PDE settings, and algebraic closures arising from moment closure in kinetic equations. For the first class, we derive an error bound in an ODE setting that quantifies contributions from time integration, approximation of unresolved scales, and interpolation required to couple unresolved-scale effects to the resolved solver. Numerical experiments on the Lorenz-63 system and the Kuramoto-Sivashinsky equation demonstrate accurate long-horizon predictions and substantial improvements over an LSTM-based closure model. For the second class, we consider moment closure for a one-dimensional kinetic equation by modeling discrepancies between kinetic and macroscopic fluxes as a function of the resolved macroscopic variables. We compare global KRR models based on PCA coordinates with spatially local models. While the global model performs well for unimodal initial conditions, its accuracy deteriorates for bimodal initial conditions. Spatially local models with appropriate modeling inputs improve robustness and achieve higher predictive accuracy.
ER -