Search

The thesis is about the incomplete Cholesky factorization and its va- riants, which are important for preconditioning a system with symmetric and positive definite matrix. Our main focus is on solving these systems, which arise in many technical applications and natural sciences, using preconditioned Con- jugate Gradients. Besides many other ways we can apply Cholesky factorization approximately, incompletely. In this thesis we study existence of the incomplete Cholesky factorization and we evaluate behaviour and potential of different vari- ants of the generic algorithm. 1

[EN] . In this paper we improve the BIF algorithm which computes simultaneously the LU factors (direct factors) of a given matrix and their inverses (inverse factors). This algorithm was introduced in [R. Bru, J. Mar´ın, J. Mas, and M. T˚uma, SIAM J. Sci. Comput., 30 (2008), pp. 2302– 2318]. The improvements are based on a deeper understanding of the inverse Sherman–Morrison (ISM) decomposition, and they provide a new insight into the BIF decomposition. In particular, it is shown that a slight algorithmic reformulation of the basic algorithm implies that the direct and inverse factors numerically influence each other even without any dropping for incompleteness. Algorithmically, the nonsymmetric version of the improved BIF algorithm is formulated. Numerical experiments show very high robustness of the incomplete implementation of the algorithm used for preconditioning nonsymmetric linear systems, Received by the editors January 26, 2009; accepted for publication (in revised form) by V. Simoncini June 1, 2010; published electronically August 12, 2010. This work was supported by Spanish grant MTM 2007-64477, by project IAA100300802 of the Grant Agency of the Academy of Sciences of the Czech Republic, and partially also by the International Collaboration Support M100300902 of AS CR.

In this paper we survey our work on preconditioners based on the Inverse Sherman-Morrison factorization. The most important theoretical results are also summarized and some numerical conclusions are provided., This work was supported by the Spanish Ministerio de Economia y Competitividad under grant MTM2014-58159-P and by the project 13-06684S of the Grant Agency of the Czech Republic.