Nonsymmetric multigrid preconditioning for conjugate gradient methods

Abstract

We numerically analyze the possibility of turning off post-smoothing (relaxation) in geometric multigrid when used as a preconditioner in conjugate gradient linear and eigenvalue solvers for the 3D Laplacian. The geometric Semicoarsening Multigrid (SMG) method is provided by the hypre parallel software package. We solve linear systems using two variants (standard and flexible) of the preconditioned conjugate gradient (PCG) and preconditioned steepest descent (PSD) methods. The eigenvalue problems are solved using the locally optimal block preconditioned conjugate gradient (LOBPCG) method available in hypre through BLOPEX software. We observe that turning off the post-smoothing in SMG dramatically slows down the standard PCG-SMG. For flexible PCG and LOBPCG, our numerical results show that post-smoothing can be avoided, resulting in overall acceleration, due to the high costs of smoothing and relatively insignificant decrease in convergence speed. We numerically demonstrate for linear systems that PSD-SMG and flexible PCG-SMG converge similarly if SMG post-smoothing is off. We experimentally show that the effect of acceleration is independent of memory interconnection. A theoretical justification is provided.

7 pages

References in corpus (1)

• Convergence theory for preconditioned eigenvalue solvers in a nutshell

Cited by in corpus (7)

• Olympus: a benchmarking framework for noisy optimization and experiment planning
• Convergence theory for preconditioned eigenvalue solvers in a nutshell
• Superfast CUR Matrix Algorithms, Their Pre-Processing and Extensions
• Differentiable Implicit Layers
• Iterative solution to the biharmonic equation in mixed form discretized by the Hybrid High-Order method
• A fault-tolerant domain decomposition method based on space-filling curves
• A dimension-oblivious domain decomposition method based on space-filling curves