Convergence Behavior of Some Iterative Solution Methods

摘要

Among the most competitive methods for solving large sparse linear systems of equations are conjugate gradients, the generalized minimum residual method and conjugate gradients-squared. The paper presents an overview of these methods and focuses in particular on their convergence behavior. Understanding the convergence behavior, in view of (spectral) properties of the given matrix, may be of help in applying these methods successfully and in selecting or modifying suitable preconditioning matrices. (Copyright (c) 1989 by Faculty of Technical Mathematics and Informatics, Delft, The Netherlands.)