1: \begin{abstract}

2: \noindent

3: The Douglas--Rachford algorithm is a classical and

4: very successful 

5: splitting method for 

6: finding the zeros of the sums of  

7: monotone operators. When the underlying operators are normal cone

8: operators,

9: the algorithm solves a convex feasibility problem.

10: In this paper, we provide a detailed study of the Douglas--Rachford 

11: iterates and the corresponding {shadow sequence} when the 

12: sets are affine subspaces that do not necessarily intersect.

13: We prove strong convergence of the shadows to the nearest

14: generalized solution.

15: Our results extend recent work from the consistent to the

16: inconsistent case. 

17: Various examples are provided to illustrates the results. 

18: \end{abstract}

19: