On the role of orthogonality in the GMRES method
In the paper we deal with some computational aspects of the Generalized minimal residual method (GMRES) for solving systems of linear algebraic equations. The key question of the paper is the importance of the orthogonality of computed vectors and its influence on the rate of convergence, numerical stability and accuracy of different implementations of the method. Practical impact on the efficiency in the parallel computer environment is considered.
Unable to display preview. Download preview PDF.
- 1.R. Barret, M. Berry, T. Chan, J. Demmel, J. Donato, J. Dongarra, V. Eijkhout, R. Pozo, C. Romine and H. van der Vorst. Templates for the Solution of Linear Systems. SIAM, Philadelphia, 1994.Google Scholar
- 2.M. Benzi, M. Tůma: A sparse approximate inverse preconditioner for nonsymmetric linear systems. to appear in SIAM J. Sci. Comput.Google Scholar
- 3.T. Davis: Sparse matrix collection. NA Digest, Volume 94, Issue 42, October 1994.Google Scholar
- 4.J. Drkošová, A. Greenbaum, M. Rozložník, Z. Strakoš, Numerical Stability of the GMRES Method, BIT 3, pp. 309–330, 1995Google Scholar
- 5.R.W. Freund, G.H. Golub, N.M. Nachtigal, Iterative Solution of Linear Systems, Acta Numerica 1, pp. 1–44, 1992Google Scholar
- 6.A. Greenbaum. M. Rozložník, Z. Strakoš, Numerical Behaviour of the Modified Gram Schmidt GMRES Method, (submitted to BIT)Google Scholar
- 7.Y. Saad, M.H. Schultz, GMRES: A Generalized Minimal Residual Algorithm for Solving Nonsymmetric Linear Systems, SIAM J. Sci. Stat. Comput. 7, pp. 856–869, 1986Google Scholar
- 8.H. F. Walker, Implementation of the GMRES method using Householder transformations, SIAM J. Sci. Stat. Comput. 9, 1 (1988), pp. 152–163.Google Scholar