1: \begin{abstract}

2: Multigrid methods are one of the most efficient techniques for solving large sparse linear systems arising from Partial Differential Equations (PDEs) 

3: and graph Laplacians from machine learning applications. One of the key components of multigrid is smoothing, 

4: which aims at reducing high-frequency errors on each grid level. 

5: However, finding optimal smoothing algorithms is problem-dependent and can impose challenges for many problems. 

6: In this paper, we propose an efficient adaptive framework for learning optimized smoothers from operator stencils

7: in the form of convolutional neural networks (CNNs).

8: 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,

9: and can be applied to large-scale problems of the same class of PDEs.

10: Numerical results on anisotropic rotated Laplacian problems and variable coefficient diffusion problems

11: demonstrate improved convergence rates and solution time compared with classical hand-crafted relaxation methods.

12: \end{abstract}

13: