Parallel sparse approximate preconditioners for the solution of large dense linear systems
Patricia González, Tomás F. Pena, Jose Carlos Cabaleiro · 2000
The aim of this work is the research into strategies to speed up the convergence of iterative methods applied to dense linear systems, as those systems of equations arising from BEM (Boundary Element Method) problems. Different procedures for the choice of a sparsity pattern for the approximate inverse preconditioner are developed. The construction of the preconditioners, as well as the solution method, were developed for a distributed memory multicomputer. Some experiments and numerical results are presented in this paper.