5. Iterative Methods for Linear Systems with Matrix Structure

Raymond H. Chan, Michael K. NgSociety 1999 · DOI · OpenAlex

Abstract

5.1 INTRODUCTION So far in the book, only direct methods for the solution of linear systems of equations have been discussed. These methods involve computing a factorization for the coefficient matrix A, say, an LDU or a QR factorization, and then solving for the unknown vector x by solving triangular systems of equations. There is another class of methods, known as iterative methods, where the solution x is approximated by successive iterates starting from an initial guess . A major algorithm in this regard is the so-called conjugate gradient (CG) method. However, its convergence rate, as we shall demonstrate in Sec. 5.2, is dependent on the clustering of the eigenvalues of the coefficient matrix A; the more clustered the eigenvalues are the faster the convergence of to x.
BibTeX
@inproceedings{chan1999iterative,
  author = {Raymond H. Chan and Michael K. Ng},
  title = {{5. Iterative Methods for Linear Systems with Matrix Structure}},
  booktitle = {Society},
  year = {1999},
  doi = {10.1137/1.9781611971354.ch5},
}

Loading related content...

Figure