On Inner Iterations of Jacobi-Davidson Type Methods for Large SVD Computations
On Inner Iterations of Jacobi-Davidson Type Methods for Large SVD Computations
复制标题
大型 SVD 计算的 Jacobi-Davidson 型方法的内迭代
DOI：
10.1137/18m1192019
复制
发表时间：
2017-11
影响因子：
3.1
通讯作者：
Jia Zhongxiao
中科院分区：
文献类型：
--
作者：
Huang Jinzhi;Jia Zhongxiao
We make a convergence analysis of the standard, harmonic, refined and refined harmonic extraction versions of Jacobi-Davidson SVD (JDSVD) type methods for computing an interior singular triplet of a large matrix $A$. At every iteration of these methods, a correction equation, i.e., inner linear system, is solved approximately by using iterative methods, which leads to four inexact JDSVD type methods, as opposed to the exact methods where inner linear systems are solved exactly. Accuracy of inner iterations critically affects the convergence and overall efficiency of the inexact JDSVD methods. A central problem is how accurately the correction equations should be solved so as to ensure that each of the inexact JDSVD methods can mimic its exact counterpart well, that is, they use almost the same outer iterations to achieve the convergence. In this paper, we prove that each inexact JDSVD method behaves like its exact counterpart if all the inner linear systems are solved with $low\ or\ modest$ accuracy during outer iterations. Based on the theory, we propose practical stopping criteria for inner iterations. Numerical experiments confirm our theory.
登录
查看更多内容
DOI：
10.1137/1.9781611970937
发表时间：
1997
期刊：
--
影响因子：
--
作者：
A. Greenbaum
通讯作者：
A. Greenbaum
DOI：
10.1137/080732110
发表时间：
2009-03
期刊：
SIAM J. Matrix Anal. Appl.
影响因子：
--
作者：
M. Hochstenbach;Y. Notay
通讯作者：
M. Hochstenbach;Y. Notay
影响因子：
14.3
作者：
G. Stewart
通讯作者：
G. Stewart
DOI：
10.1137/s1064827597327309
发表时间：
1999-12
期刊：
SIAM J. Sci. Comput.
影响因子：
--
作者：
H. Simon;H. Zha
通讯作者：
H. Simon;H. Zha
DOI：
10.1360/04za0020
发表时间：
2004
期刊：
Science in China Series A: Mathematics
影响因子：
--
作者：
Zhongxiao Jia
通讯作者：
Zhongxiao Jia