Matrix-free multigrid block-preconditioners for higher order discontinuous Galerkin discretisations

• Verfasst von: Bastian, Peter [VerfasserIn]
• Verfasst von: Müller, Eike Hermann [VerfasserIn]
• Verfasst von: Müthing, Steffen [VerfasserIn]
• Verfasst von: Piatkowski, Stephan-Marian [VerfasserIn]
• Titel: Matrix-free multigrid block-preconditioners for higher order discontinuous Galerkin discretisations
• Verf.angabe: Peter Bastian, Eike Hermann Müller, Steffen Müthing, Marian Piatkowski
• E-Jahr: 2019
• Jahr: 4 June 2019
• Umfang: 23 S.
• Fussnoten: Gesehen am 27.08.2019
• Titel Quelle: Enthalten in: Journal of computational physics
• Ort Quelle: Amsterdam : Elsevier, 1961
• Jahr Quelle: 2019
• Band/Heft Quelle: 394(2019), Seite 417-439
• ISSN Quelle: 1090-2716
• DOI: doi:10.1016/j.jcp.2019.06.001
• Datenträger: Online-Ressource
• Sprache: eng
• Sach-SW: Discontinuous Galerkin
• Sach-SW: DUNE
• Sach-SW: Elliptic PDE
• Sach-SW: Matrix-free methods
• Sach-SW: Multigrid
• Sach-SW: Preconditioners
• K10plus-PPN: 1672251664

Abstract

Efficient and suitably preconditioned iterative solvers for elliptic partial differential equations (PDEs) of the convection-diffusion type are used in all fields of science and engineering, including for example computational fluid dynamics, nuclear reactor simulations and combustion models. To achieve optimal performance, solvers have to exhibit high arithmetic intensity and need to exploit every form of parallelism available in modern manycore CPUs. This includes both distributed- or shared memory parallelisation between processors and vectorisation on individual cores. The computationally most expensive components of the solver are the repeatedapplications of the linear operator and the preconditioner. For discretisations based on higher-order Discontinuous Galerkin methods, sum-factorisation results in a dramatic reduction of the computational complexity of the operator application while, at the same time, the matrix-free implementation can run at a significant fraction of the theoretical peak floating point performance. Multigrid methods for high order methods often rely on block-smoothers to reduce high-frequency error components within one grid cell. Traditionally, this requires the assembly and expensive dense matrix solve in each grid cell, which counteracts any improvements achieved in the fast matrix-free operator application. To overcome this issue, we present a new matrix-free implementation of block-smoothers. Inverting the block matrices iteratively avoids storage and factorisation of the matrix and makes it is possible to harness the full power of the CPU. We implemented a hybrid multigrid algorithm with matrix-free block-smoothers in the high order Discontinuous Galerkin (DG) space combined with a low order coarse grid correction using algebraic multigrid where only low order components are explicitly assembled. The effectiveness of this approach is demonstrated by solving a set of representative elliptic PDEs of increasing complexity, including a convection dominated problem and the stationary SPE10 benchmark.