- About this Journal ·
- Abstracting and Indexing ·
- Advance Access ·
- Aims and Scope ·
- Annual Issues ·
- Article Processing Charges ·
- Articles in Press ·
- Author Guidelines ·
- Bibliographic Information ·
- Citations to this Journal ·
- Contact Information ·
- Editorial Board ·
- Editorial Workflow ·
- Free eTOC Alerts ·
- Publication Ethics ·
- Reviewers Acknowledgment ·
- Submit a Manuscript ·
- Subscription Information ·
- Table of Contents
Abstract and Applied Analysis
Volume 2014 (2014), Article ID 237808, 11 pages
Some Generalizations and Modifications of Iterative Methods for Solving Large Sparse Symmetric Indefinite Linear Systems
1Department of Mathematics, National Kaohsiung Normal University, Kaohsiung 824, Taiwan
2Institute for Computational Engineering and Sciences, The University of Texas at Austin, Austin, TX 78712, USA
Received 27 November 2013; Revised 10 January 2014; Accepted 4 February 2014; Published 3 April 2014
Academic Editor: Chi-Ming Chen
Copyright © 2014 Yu-Chien Li et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
We discuss a variety of iterative methods that are based on the Arnoldi process for solving large sparse symmetric indefinite linear systems. We describe the SYMMLQ and SYMMQR methods, as well as generalizations and modifications of them. Then, we cover the Lanczos/MSYMMLQ and Lanczos/MSYMMQR methods, which arise from a double linear system. We present pseudocodes for these algorithms.
The authors dedicate this paper to the memory of Professor David M. Young, Jr., for his pioneering research, inspirational teaching, and exceptional life
Frequently, when computing numerical solutions of partial differential equations, one needs to solve systems of very large sparse linear algebraic equations of the form where is an matrix, is an vector, and one seeks a numerical solution vector or a good approximation of it. Particularly for large linear systems arising from partial differential equations in three dimensions, well-known direct methods, such as Gaussian elimination, may become prohibitively expensive in terms of both computer storage and computer time. On the other hand, a variety of iterative methods may avoid these difficulties.
For linear systems involving symmetric positive definite (SPD) matrices, the conjugate gradient (CG) method (and variations of it) may work well. On the other hand, when solving linear systems, where the coefficient matrix is symmetric indefinite, the choice of a suitable iterative method is not at all clear. On the other hand, the SYMMLQ and MINRES methods have been shown to be useful in certain situations (see Paige and Saunders ). For nonsymmetric systems, Saad and Schultz  generalized the MINRES method to obtain the GMRES method.
In Section 2, we review the Arnoldi process. In Sections 3 and 4, we describe the SYMMLQ and SYMMQR methods. Then we can generalize them, in Section 5, and we outline the modified SYMMLQ method, in Section 6. Next, in Section 7, we discuss applying the MSYMMLQ and MSYMMQR methods applied to a double linear system. Finally, we present pseudocodes in Sections 8–11.
2. Arnoldi Process
We begin with a review of the Arnoldi process.
Theorem 1. Suppose that is an symmetric matrix. One can generate orthonormal vectors using this short-term recurrence where Here, one assumes that and , for all . Then the following properties hold, for (, ):
Proof. If we let , then the subspace
is equivalent to the Krylov subspace
We obtain since .
From Theorem 1, in matrix form, it follows that where
Example 2. We illustrate Theorem 1 for the case .
3. SYMMLQ Method
We choose , such that . Hence, we have
Imposing the Galerkin condition , we obtain We obtain because Instead of solving for directly from the triangular linear system (15), Paige and Saunders  factorize the matrix into a lower triangular matrix with bandwidth three (resulting in the SYMMLQ method). Also, we have where is an orthogonal matrix, and where . Since , we have Letting then Next letting we have Defining we have where We let where From (21) and (28), we have . Since we have If , then is nonsingular. We can find by solving
4. SYMMQR Method
We choose such that . Hence, we have Imposing the Galerkin condition , as before, we obtain Since we have Instead of solving for directly from the triangular system (35), Paige and Saunders  factorized the matrix into a lower triangular matrix with bandwidth three.
We can use a different factorization of to obtain a slightly different method, which is called the SYMMQR method. We multiply the matrix by an orthogonal matrix on the left-hand side instead of the right-hand side. We have where We obtain the matrix , where with being the Givens rotation. Letting be the solution of then we have which satisfies the Galerkin condition , where . We note that is not always nonzero and, thus, might be singular. We assume that is nonsingular and then we define where We have
For the next iterate , we need to solve where Applying the Givens rotation to both sides of (45), we have where .
To eliminate , we compute the th Given rotation by By multiplying times and times , we have where Let We define . Since , then and is nonsingular. We can solve for from We discuss the case later.
Consider solving the least square problem involving minimizing , where We have Hence, the solution from minimizes and .
Let where We have Since we obtain We note that is the estimated solution vector satisfying the Galerkin condition, while with minimizing .
5. Generalized SYMMLQ and SYMMQR Methods
Now, we generalize the SYMMLQ and SYMMQR methods.
Theorem 3. Suppose that is an symmetric positive definite (SPD) matrix and is an symmetric matrix. One can generate orthonormal vectors using this short-term recurrence where Then the following properties hold, for (, ):
Proof. We obtain Since .
As before, we let Moreover, we have where
As before, we let Imposing the Galerkin condition again, we have We obtain because Since is symmetric, we can apply the same techniques as in the SYMMLQ method. Also, if , the method reduces to the SYMMQR method.
6. Modified SYMMLQ Method
Next, we outline the modified SYMMLQ method.
Theorem 4. Suppose that is an symmetric (not necessary positive definite) matrix and is an symmetric matrix. One can generate orthonormal vectors using this short-term recurrence where Then the following properties hold, for : and, for (, ),
In addition, we have Imposing the Galerkin condition, , as we did before, we obtain In other words, we use We obtain because HereWe note that is symmetric, for :
7. Lanczos/MSYMMLQ Method
Next, we consider this double linear system: We obtain the block symmetric matrices , , and , where
For example, the modified SYMMLQ method and the modified SYMMQR method can be applied to the double linear system (86). This leads us to the LAN/MSYMMLQ method and the LAN/MSYMMQR method. The pseudocodes for these methods are given in the following sections. For additional details, see Li . See the books by Golub and Van Loan  and Saad , as well as the papers by Lanczos  and Kincaid et al. , among others.