Gaussian-process surrogate indicators for residual-based adaptive GMsFEM

DOI10.48550/arXiv.2609.26843MaRDI QIDQ7353517arXivFDO

Authors Eric T. Chung, Yiran Wang, Siqing Liu, Eric Chung

Publication date 22 September 2026

Published in arXiv

Tldr We introduce a non-intrusive Gaussian-process surrogate for residual-based adaptive GMsFEM indicators to speed up repeated-query settings. The operational predictor is the GP posterior mean, which provides algebraically equivalent kernel ridge regression estimator without changing multiscale solve or basis enrichment, and outperforms offline adaptivity in controlled held-out tests with 2-3x faster online evaluation excluding data generation and training.

Full work available at URL https://doi.org/10.48550/arXiv.2609.26843

arXiv classification

math.NAstat.ML

Cites work

• A locally conservative multiscale method for stochastic highly heterogeneous flow
• A multiscale finite element method for elliptic problems in composite materials and porous media
• A unifying view of sparse approximate Gaussian process regression
• Adaptive multiscale extended finite element method (MS-XFEM) for the simulation of multiple fractures propagation in geological formations
• Adaptive multiscale model reduction with generalized multiscale finite element methods
• AMS-Net: Adaptive Multiscale Sparse Neural Network with Interpretable Basis Expansion for Multiphase Flow Problems
• An adaptive GMsFEM for high-contrast flow problems
• An adaptive space and time method in partially explicit splitting scheme for multiscale flow problems
• Deterministic error bounds for kernel-based learning techniques under bounded noise
• Gaussian Processes for Machine Learning
• Generalized multiscale finite element methods (GMsFEM)
• Goal-oriented adaptivity for GMsFEM
• Meshfree generalized multiscale finite element method
• Mode decomposition methods for flows in high-contrast porous media. Global-local approach
• Multiscale finite element methods for high-contrast problems using local spectral basis functions
• Multiscale finite element methods for nonlinear problems and their applications
• Multiscale model reduction. Multiscale finite element methods and their generalizations
• Operator learning with Gaussian processes
• Proper Orthogonal Decomposition Surrogate Models for Nonlinear Dynamical Systems: Error Estimates and Suboptimal Control
• Residual-driven online generalized multiscale finite element methods
• Scientific Machine Learning through Physics-Informed Neural Networks: Where we are and What's next
• Solving and Learning Nonlinear PDEs with Gaussian Processes

This page was built for publication: Gaussian-process surrogate indicators for residual-based adaptive GMsFEM

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7353517)