Multigrid methods for the Stokes problem on GPU systems

This paper presents a matrix-free multigrid method for solving the Stokes problem, discretized using Hdiv-conforming discontinuous Galerkin methods. Our method operates directly on both the velocity and pressure spaces, eliminating the need for a global Schur complement approximation. We employ a mu...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Cui, Cu (VerfasserIn) , Kanschat, Guido (VerfasserIn)
Dokumenttyp: Article (Journal)
Sprache:Englisch
Veröffentlicht: 30 August 2025
In: Computers & fluids
Year: 2025, Jahrgang: 299, Pages: 1-11
ISSN:1879-0747
DOI:10.1016/j.compfluid.2025.106703
Online-Zugang:Verlag, kostenfrei, Volltext: https://doi.org/10.1016/j.compfluid.2025.106703
Verlag, kostenfrei, Volltext: https://www.sciencedirect.com/science/article/pii/S004579302500163X
Volltext
Verfasserangaben:Cu Cui, Guido Kanschat
Beschreibung
Zusammenfassung:This paper presents a matrix-free multigrid method for solving the Stokes problem, discretized using Hdiv-conforming discontinuous Galerkin methods. Our method operates directly on both the velocity and pressure spaces, eliminating the need for a global Schur complement approximation. We employ a multiplicative Schwarz smoother with vertex-patch subdomains and the Schur complement method combined with the fast diagonalization for the efficient evaluation of the local solvers. By leveraging the tensor product structure of Raviart-Thomas elements and an optimized, conflict-free shared memory access pattern, the matrix-free operator evaluation demonstrates excellent performance, reaching over one billion degrees of freedom per second on a single NVIDIA A100 GPU. Numerical results indicate efficiency comparable to that of the three-dimensional Poisson problem.
Beschreibung:Online verfügbar: 12. Juni 2025, Artikelversion: 17. Juni 2025
Gesehen am 27.10.2025
Beschreibung:Online Resource
ISSN:1879-0747
DOI:10.1016/j.compfluid.2025.106703