Robust MultigridPreconditioning for
Parameter-Dependent Problems I:
The Stokes-type Case

Joachim Schoberl
Johannes Kepler University
Institut für Mathematik
A-4040 Linz, Altenbergerstrasse 69, Austria

Abstract

Parameter dependent problems can be discretized by selective reduced integration methods to achieve parameter independent discretization errors. The convergence rate of standard multigrid solvers applied to the primal linear system deteriorates, if the parameter becomes small. In this paper, we construct multigrid components leading to parameter independentrates. We need a robust base iteration as smoother and uniformly continuous grid transfer operations. The suggested multigrid preconditioner is applied to problems from linear elasticity.


Contributed November 27, 1997.