1: \begin{abstract}

2: 

3: We shall derive and propose several efficient 

4: overlapping domain decomposition methods for solving some typical

5: linear inverse problems, including the identification of the flux,

6: the source strength and the initial temperature in

7: second order elliptic and parabolic systems. The methods are iterative, and

8: computationally very efficient: only local forward and adjoint problems

9: need to be solved in each subdomain, and the local minimizations have

10: explicit solutions.

11: %The ill-posed inverse problems are formulated into the most frequently used and

12: %reliable output least-squares formulation with Tikhonov

13: %regularization.

14: Numerical experiments are provided to demonstrate the

15: robustness and efficiency of the methods, in particular, the convergences

16: seem nearly optimal, i.e., they do not deteriorate or deteriorate

17: only slightly when the mesh size reduces.

18: \end{abstract}