Factorized sparse approximate inverse preconditionings. IV: Simple approaches to rising efficiency.(English)Zbl 0983.65062
[For part III see J. Math. Sci., New York 101, No. 4, 3237-3254 (2000;Zbl 0961.65042).]
Summary: This paper continues the theoretical and numerical study of the so-called factorized sparse approximate inverse (FSAI) preconditionings of symmetric positive-definite matrices and considers two new approaches to improving them. The first one is based on the a posteriori sparsification of an already constructed FSAI preconditioner, whereas the second one amounts to applying another FSAI preconditioning to an already preconditioned matrix. Numerical results for sample FE problems from structural mechanics are presented.
Summary: This paper continues the theoretical and numerical study of the so-called factorized sparse approximate inverse (FSAI) preconditionings of symmetric positive-definite matrices and considers two new approaches to improving them. The first one is based on the a posteriori sparsification of an already constructed FSAI preconditioner, whereas the second one amounts to applying another FSAI preconditioning to an already preconditioned matrix. Numerical results for sample FE problems from structural mechanics are presented.
MSC:
| 65F35 | Numerical computation of matrix norms, conditioning, scaling |
| 65F10 | Iterative numerical methods for linear systems |
| 65F50 | Computational methods for sparse matrices |
| 74S05 | Finite element methods applied to problems in solid mechanics |
Keywords:
sparse approximate inverses;preconditioning;symmetric positive-definite matrices;numerical resultsCitations:
Zbl 0961.65042Cite
Full Text:DOI
References:
| [1] | and . An MPI implementation of the SPAI preconditioner on the T3E. Technical report LBNL-40794, Lawrence Berkeley National Laboratory, Berkeley, CA, 1997. |
| [2] | Benzi, SIAM J. Sci. Comput. 17 pp 1135– (1996) ·Zbl 0856.65019 ·doi:10.1137/S1064827594271421 |
| [3] | Benzi, SIAM J. Sci. Comput. 19 pp 968– (1998) ·Zbl 0930.65027 ·doi:10.1137/S1064827595294691 |
| [4] | Benzi, Appl. Numer. Math. |
| [5] | Benzi, BIT 38 pp 234– (1998) ·Zbl 0909.65027 ·doi:10.1007/BF02512364 |
| [6] | and . Orderings for factorized sparse approximate inverse preconditioners. Technical report LA-UR-98-2175, Los Alamos National Laboratory, 1998. |
| [7] | and . Ordering, anisotropy and factored sparse approximate inverses. Preprint, Department of Computer Science, University of Waterloo, Waterloo, 1998. |
| [8] | A priori sparsity patterns for parallel sparse approximate inverse preconditioners. Manuscript, 1998. |
| [9] | An efficient parallel preconditioner for the conjugate gradient algorithm. Technical report HDL-TR-97-175, O’Reilly Institute, Hitachi Dublin Laboratory, 1997. |
| [10] | Manuscript, 1998. |
| [11] | Grote, SIAM J. Sci. Comput. 18 pp 838– (1997) ·Zbl 0872.65031 ·doi:10.1137/S1064827594276552 |
| [12] | Hong, Linear Algebra Appl. 172 pp 27– (1992) ·Zbl 0768.15012 ·doi:10.1016/0024-3795(92)90016-4 |
| [13] | Huckle, Appl. Numer. Math. (1999) |
| [14] | Kaporin, Numer. Lin. Alg. Appl. 1 pp 179– (1994) ·Zbl 0837.65027 ·doi:10.1002/nla.1680010208 |
| [15] | Kolotilina, SIAM J. Matrix Anal. Appl. 14 pp 45– (1993) ·Zbl 0767.65037 ·doi:10.1137/0614004 |
| [16] | Kolotilina, Int. J. High Speed Comput. 7 pp 191– (1995) ·doi:10.1142/S0129053395000117 |
| [17] | and . Incomplete block factorizations as preconditioners for sparse SPD matrices. In: Recent Advances in Iterative Methods, pages 119-133. Vol. 66 of The IMA Volumes in Mathematics and its Applications, 1994. ·Zbl 0803.65058 ·doi:10.1007/978-1-4613-9353-5_8 |
| [18] | Iterative Methods for Sparse Linear Systems. PWS Publ. Co., Boston, 1996. |
| [19] | Yeremin, Zap. Nauchn. Semin. POMI 248 pp 17– (1998) |
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.
