Learning optimal multigrid smoothers via neural networks

Abstract

Multigrid methods are one of the most efficient techniques for solving linear systems arising from Partial Differential Equations (PDEs) and graph Laplacians from machine learning applications. One of the key components of multigrid is smoothing, which aims at reducing high-frequency errors on each grid level. However, finding optimal smoothing algorithms is problem-dependent and can impose challenges for many problems. In this paper, we propose an efficient adaptive framework for learning optimized smoothers from operator stencils in the form of convolutional neural networks (CNNs). The CNNs are trained on small-scale problems from a given type of PDEs based on a supervised loss function derived from multigrid convergence theories, and can be applied to large-scale problems of the same class of PDEs. Numerical results on anisotropic rotated Laplacian problems demonstrate improved convergence rates and solution time compared with classical hand-crafted relaxation methods.

References in corpus (10)

• DGM: A deep learning algorithm for solving partial differential equations
• Solving high-dimensional partial differential equations using deep learning
• A unified deep artificial neural network approach to partial differential equations in complex geometries
• MgNet: A Unified Framework of Multigrid and Convolutional Neural Network
• Multi-level Residual Networks from Dynamical Systems View
• Learning Neural PDE Solvers with Convergence Guarantees
• Learning to Control PDEs with Differentiable Physics
• Learning to Optimize Multigrid PDE Solvers
• Deep Multigrid: learning prolongation and restriction matrices
• Optimizing Geometric Multigrid Methods with Evolutionary Computation