1: \begin{abstract}

2: We study the combination of proximal gradient descent with multigrid for solving a class of possibly nonsmooth strongly convex optimization problems.

3: We propose a multigrid proximal gradient method called MGProx, which accelerates the proximal gradient method by multigrid, based on using hierarchical information of the optimization problem.

4: MGProx applies a newly introduced adaptive restriction operator to simplify the Minkowski sum of subdifferentials of the nondifferentiable objective function across different levels.

5: We provide a theoretical characterization of MGProx.

6: First we show that the MGProx update operator exhibits a fixed-point property.

7: Next, we show that the coarse correction is a descent direction for the fine variable in the general nonsmooth case.

8: Lastly, under some assumptions we provide the convergence rate for the algorithm.

9: In the numerical tests, we show that MGProx has a faster convergence speed than several competing methods on nonsmooth convex optimization problems such as the Elastic Obstacle Problem.

10: \end{abstract}

11: