Convergence on Gauss-Seidel iterative methods for linear systems with general H-matrices

Electronic Journal of Linear Algebra, Dec 2015

It is well known that as a famous type of iterative methods in numerical linear algebra, Gauss-Seidel iterative methods are convergent for linear systems with strictly or irreducibly diagonally dominant matrices, invertible H−matrices (generalized strictly diagonally dominant matrices) and Hermitian positive definite matrices. But, the same is not necessarily true for linear systems with non-strictly diagonally dominant matrices and general H−matrices. This paper firstly proposes some necessary and sufficient conditions for convergence on Gauss-Seidel iterative methods to establish several new theoretical results on linear systems with nonstrictly diagonally dominant matrices and general H−matrices. Then, the convergence results on preconditioned Gauss-Seidel (PGS) iterative methods for general H−matrices are presented. Finally, some numerical examples are given to demonstrate the results obtained in this paper.

A PDF file should load here. If you do not see its contents the file may be temporarily unavailable at the journal website or you do not have a PDF plug-in installed and enabled in your browser.

Alternatively, you can download the file locally and open with any standalone PDF reader:

https://repository.uwyo.edu/cgi/viewcontent.cgi?article=1972&context=ela

Convergence on Gauss-Seidel iterative methods for linear systems with general H-matrices

Volume 1081-3810 Convergence on Gauss-Seidel iterative methods for linear systems with general H-matrices Cheng-yi Zhang 0 1 2 Dan Ye 0 1 2 Cong-Lei Zhong 0 1 2 0 Xi'an Polytechnic University 1 Henan University of Science and Technology 2 School of Science, Xi'an Polytechnic University SHUANGHUA SHUANGHUA Xi'an Polytechnic University Follow this and additional works at: https://repository.uwyo.edu/ela Part of the Numerical Analysis and Computation Commons Recommended Citation - http://math.technion.ac.il/iic/ela CONVERGENCE ON GAUSS-SEIDEL ITERATIVE METHODS FOR LINEAR SYSTEMS WITH GENERAL H −MATRICES∗ CHENG-YI ZHANG† , DAN YE‡ , CONG-LEI ZHONG§ , AND SHUANGHUA LUO¶ AMS subject classifications. 15A15, 15F10. 1. Introduction. In this paper, we consider the solution methods for the system of n linear equations (1.1) Ax = b, where A = (aij ) ∈ Cn×n and is nonsingular, b, x ∈ Cn and x unknown. Let us recall the standard decomposition of the coefficient matrix A ∈ Cn×n, A = DA − LA − UA, where DA = diag(a11, a22, . . . , ann) is a diagonal matrix, LA and UA are strictly lower and strictly upper triangular matrices, respectively. If aii 6= 0 for all i ∈ hni = ∗Received by the editors on October 13, 2014. Accepted for publication on September 12, 2015. Handling Editor: James G. Nagy. †Department of Mathematics and Mechanics of School of Science, Xi’an Polytechnic University, Xi’an, Shaanxi, 710048, P.R. China (). Supported by the Scientific Research Foundation (BS1014), the Education Reform Foundation of Xi’an Polytechnic University (2012JG40), and National Natural Science Foundations of China (11201362 and 11271297). ‡School of Science, Xi’an Polytechnic University, Xi’an, Shaanxi, 710048, P.R. China. §School of Mathematics and Statistics, Henan University of Science and Technology, Luoyang, Henan, 417003, P.R. China. ¶Department of Mathematics and Mechanics of School of Science, Xi’an Polytechnic University, Xi’an, Shaanxi, 710048, P.R. China. Supported by the Science Foundation of the Education Department of Shaanxi Province of China (14JK1305). {1, 2, . . . , n}, the Jacobi iteration matrix associated with the coefficient matrix A is the forward, backward and symmetric Gauss-Seidel (FGS-, BGS- and SGS-) iteration matrices associated with the coefficient matrix A are C.-Y. Zhang, D. Ye, C.-L. Zhong, and S. Luo HJ = DA−1(LA + UA); HF GS = (DA − LA)−1UA, HBGS = (DA − UA)−1LA, HSGS = = HBGS HF GS (DA − UA)−1LA(DA − LA)−1UA, 844 (1.2) (1.3) and (1.4) (1.5) respectively. Then, the Jacobi, FGS, BGS and SGS iterative method can be denoted the following iterative scheme: x(i+1) = Hx(i) + f, i = 0, 1, 2, . . . where H denotes iteration matrices HJ , HF GS , HBGS and HSGS, respectively, correspondingly, f is equal to DA−1b, (DA − LA)−1b, (DA − UA)−1b and (DA − UA)−1DA(DA − LA)−1b, respectively. It is well-known that (1.5) converges for any given x(0) if and only if ρ(H) < 1 (see [ 11 ]), where ρ(H) denotes the spectral radius of the iteration matrix H. Thus, to establish the convergence results of iterative scheme (1.5), we mainly study the spectral radius of the iteration matrix in the iterative scheme (1.5). As is well known in some classical textbooks and monographs, see [ 11 ], Jacobi and Gauss-Seidel iterative methods for linear systems with Hermitian positive definite matrices, strictly or irreducibly diagonally dominant matrices and invertible H−matrices (generalized strictly diagonally dominant matrices) are convergent. Recently, the class of strictly or irreducibly diagonally dominant matrices and invertible H−matrices has been extended to encompass a wider set, known as the set of general H−matrices. In a recent paper, Ref. [ 2, 3, 4 ], a partition of the n × n general H−matrix set, Hn, into three mutually exclusive classes was obtained: The Invertible class, HnI , where the comparison matrices of all general H−matrices are nonsingular, the Singular class, HnS , formed only by singular H−matrices, and the Mixed class, HnM , in which singular and nonsingular H−matrices coexist. Lately, Zhang in [ 16 ] proposed some necessary and sufficient conditions for convergence on Jacobi iterative methods for linear systems with general H−matrices. 845 A problem has to be proposed, i.e., whether Gauss-Seidel iterative methods for linear systems with nonstrictly diagonally dominant matrices and general H−matrices are convergent or not. Let us investigate the following examples. Example 1.1. Assume that either A or B is the coefficient matrix of linear  2 1 1   2 −1 −1  system (1.1), where A =  −1 2 1  and B =  1 2 −1 . It is verified −1 −1 2 1 1 2 that both A and B are nonstrictly diagonally dominant and nonsingular. Direct computations yield that ρ(HFAGS ) = ρ(HBBGS ) = 1, while ρ(HBAGS ) = ρ(HFBGS ) = 0.3536 < 1 and ρ(HSAGS ) = ρ(HSBGS ) = 0.5797 < 1. This shows that BGS and SGS iterative methods for the matrix A are convergent, while the same is not FGS iterative method for A. However, (...truncated)


This is a preview of a remote PDF: https://repository.uwyo.edu/cgi/viewcontent.cgi?article=1972&context=ela

Cheng-yi Zhang, Dan Ye, Cong-Lei Zhong, SHUANGHUA SHUANGHUA. Convergence on Gauss-Seidel iterative methods for linear systems with general H-matrices, Electronic Journal of Linear Algebra, 2015, Volume 30, Issue 1,