1: \begin{abstract} \noindent

2: The Douglas--Rachford algorithm is a classical and very

3: successful method for solving optimization and feasibility

4: problems. In this paper, we provide novel conditions sufficient for

5: finite convergence in the context of convex feasibility problems. 

6: Our analysis builds upon, and considerably extends, pioneering

7: work by Spingarn. Specifically, we obtain finite convergence in

8: the presence of Slater's condition in the affine-polyhedral and

9: in a hyperplanar-epigraphical case. Various examples illustrate

10: our results. Numerical experiments demonstrate the

11: competitiveness of the Douglas--Rachford algorithm for solving linear equations with a

12: positivity constraint when compared to the method of alternating

13: projections and the method of reflection-projection. 

14: \end{abstract}

15: