Author(s) :
Matthias Bollhöfer
,
Youcef Saad
The paper is published :
SIAM J. Matrix Anal. Appl., 23(3):692-705, 2001.
MSC 2000
- 65F05 Direct methods for linear systems and matrix inversion
-
65F10 Iterative methods for linear systems
-
65F50 Sparse matrices
Abstract :
In an earlier paper we presented a few results which established
some strong relations between incomplete factorization methods
and factored approximate inverse-type methods.
In this paper some of these relations are exploited
to develop new techniques for stabilizing factorized
approximate inverse preconditioners using pivoting. This method
yields stable preconditioners in many
cases and can provide successful preconditioners in
many situations when the underlying system is highly indefinite.
Numerical examples illustrate the effectiveness of this approach
Keywords :
sparse matrices, ILU, sparse approximate inverse, AINV, pivoting
Notes :
In SIMAX a revised version appeared under the title "A Factored Approximate Inverse Preconditioner With Pivoting"