1: \begin{abstract}

2: We make a convergence analysis of the harmonic and refined

3: harmonic extraction versions of Jacobi-Davidson SVD (JDSVD) type methods

4: for computing one or more interior singular triplets of a large matrix $A$.

5: At each outer iteration of these methods, a correction equation,

6: i.e., inner linear system, is solved

7: approximately by using iterative methods, which leads to two inexact JDSVD type

8: methods, as opposed to the exact methods where correction equations

9: are solved exactly. Accuracy of inner iterations critically affects

10: the convergence and overall efficiency of the inexact

11: JDSVD methods. A central problem is how accurately

12: the correction equations should be solved so as to ensure

13: that both of the inexact JDSVD methods can mimic their exact counterparts

14: well, that is, they use almost the same outer iterations to achieve

15: the convergence. In this paper, similar to the available results on

16: the JD type methods for

17: large matrix eigenvalue problems, we prove that each inexact JDSVD method

18: behaves like its exact counterpart if all the correction equations are solved

19: with $low\ or\ modest$ accuracy during outer iterations.

20: Based on the theory, we propose practical stopping criteria for inner iterations.

21: Numerical experiments confirm our theory and the effectiveness of the inexact

22: algorithms.

23: \end{abstract}