Learning Coarse Spaces for Two-Grid Methods from Smoothed Errors with Applications to High-Contrast Porous Media Flow
Abstract
Efficient two-grid solvers for high-contrast parametric elliptic problems require coarse spaces that capture error components poorly reduced by local smoothing. As these components vary with the problem parameters, repeatedly constructing parameter-adaptive coarse spaces can be computationally expensive. We propose a coarse-space learning method based on finite-step errors generated by additive Schwarz smoothing. For each training parameter, proper orthogonal decomposition (POD) of smoothed-error snapshots in the discrete energy inner product defines a reference coarse space. A continuous-coordinate neural generator is then trained using a basis-invariant, whitened subspace loss, enabling the prediction of parameter-adaptive coarse spaces on unseen grids. We establish discrete perturbation estimates relating coarse-space discrepancies to the associated nonsymmetric two-grid operators and derive conditional finite-step Krylov residual-polynomial bounds without assuming stationary energy contraction. Numerical experiments on high-contrast interface problems, moving-interface Darcy flow, and nonlinear Forchheimer flow demonstrate effective solver performance across varying parameters and grid resolutions. On the finest static test grid, with approximately 264 times as many degrees of freedom as the training grid, the learned spaces achieve FGMRES iteration counts close to those of parameter-specific POD references and comparable to GenEO spaces, while substantially reducing repeated coarse-space construction costs.
Related
- A Smoothed Discrepancy Principle for Random Feature Methods and Neural Networks
- Composite and Stochastic Smoothed Decentralized Bregman Proximal-Gradient Methods
- Nonmonotone Coderivative-Based Newton Methods for Nonsmooth Optimization with Machine Learning Applications
- Oracle-Efficient and Parameter-Free Agnostic Smoothed Online Learning
- Advanced Statistical Learning: Limit Theorems for Nonparametric Conditional U-Statistics Smoothed by Asymmetric Kernels Under Missing-at-Random Sampling
BibTeX
@misc{bao2026learning,
title = {{Learning Coarse Spaces for Two-Grid Methods from Smoothed Errors with Applications to High-Contrast Porous Media Flow}},
author = {Chenglong Bao and Chen Cui and Ruohua Li},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.05837},
eprint = {2611.05837},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.05837},
}
RIS
TY - UNPB
TI - Learning Coarse Spaces for Two-Grid Methods from Smoothed Errors with Applications to High-Contrast Porous Media Flow
AU - Chenglong Bao
AU - Chen Cui
AU - Ruohua Li
PY - 2026
DA - 2026/11/04
PB - arXiv
DO - 10.48550/arxiv.2611.05837
UR - https://arxiv.org/abs/2611.05837
AB - Efficient two-grid solvers for high-contrast parametric elliptic problems require coarse spaces that capture error components poorly reduced by local smoothing. As these components vary with the problem parameters, repeatedly constructing parameter-adaptive coarse spaces can be computationally expensive. We propose a coarse-space learning method based on finite-step errors generated by additive Schwarz smoothing. For each training parameter, proper orthogonal decomposition (POD) of smoothed-error snapshots in the discrete energy inner product defines a reference coarse space. A continuous-coordinate neural generator is then trained using a basis-invariant, whitened subspace loss, enabling the prediction of parameter-adaptive coarse spaces on unseen grids. We establish discrete perturbation estimates relating coarse-space discrepancies to the associated nonsymmetric two-grid operators and derive conditional finite-step Krylov residual-polynomial bounds without assuming stationary energy contraction. Numerical experiments on high-contrast interface problems, moving-interface Darcy flow, and nonlinear Forchheimer flow demonstrate effective solver performance across varying parameters and grid resolutions. On the finest static test grid, with approximately 264 times as many degrees of freedom as the training grid, the learned spaces achieve FGMRES iteration counts close to those of parameter-specific POD references and comparable to GenEO spaces, while substantially reducing repeated coarse-space construction costs.
ER -