Abstract
Krylov methods are, since their introduction in the 1980s, the most heavily used methods to solve the two problems Ax = b and Ax = λx, x ≠ 0 where the matrix A is very large. However, the understanding of their numerical behaviour is far from satisfactory. We propose a radically new viewpoint for this longstanding enigma, which shows mathematically that the Krylov-type method works best when it is most ill-conditioned.
CERFACS Technical Report TR/PA/00/40
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Chatelin, F.: Spectral approximation of linear operators. Academic Press (1983).
Chaitin-Chatelin, F.: Comprendre les méthodes de Krylov en précision finie: le programme du Groupe Qualitative Computing au CERFACS. CERFACS report TR/PA/00/11 (2000).
Chaitin-Chatelin, F., Frayssé, V.: Lectures on finite precision computations. SIAM (1996).
Chaitin-Chatelin, F., Gratton, S., Traviesas, E.: Sensibilité des méthodes de Krylov au vecteur de départ. Work in progress. 189
Chaitin-Chatelin, F., Harrabi, A., Ilahi, A.: About Hölder condition number and the stratification diagram for defective eigenvalues. To appear in IMACS J. on Math. of Comp. CERFACS report TR/PA/99/19 (1999).
Chaitin-Chatelin, F., Toumazou, V., Traviesas, E.: Accuracy assessment for eigencomputations: variety of backward errors and pseudospectra. Lin. Alg Appl. 309 (2000) 73–83.
Saad, Y.: Numerical methods for large eigenvalue problems. Algorithms and architectures for advanced scientific computing. Manchester University Press (1992).
Saad, Y.: Iterative methods for sparse linear systems. PWS, Minnesota (1995).
Traviesas, E.: Sur le déploiement du champ spectral d’une matrice, Ph.D. thesis, University Toulouse I and CERFACS (2000).
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2001 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Chaitin-Chatelin, F., Traviesas, E., Plantié, L. (2001). Understanding Krylov Methods in Finite Precision. In: Vulkov, L., Yalamov, P., Waśniewski, J. (eds) Numerical Analysis and Its Applications. NAA 2000. Lecture Notes in Computer Science, vol 1988. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-45262-1_23
Download citation
DOI: https://doi.org/10.1007/3-540-45262-1_23
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-41814-6
Online ISBN: 978-3-540-45262-1
eBook Packages: Springer Book Archive