1: \begin{abstract}

2: 

3: Parallel and cyclic projection algorithms are proposed for

4: minimizing the sum of a finite family of convex functions over the intersection

5: of a finite family of closed convex subsets of a Hilbert space. These algorithms are of predictor-corrector type, with each main iteration consisting of an inner cycle of

6: subgradient descent process followed by a projection step. We prove the convergence of these methods to an optimal

7: solution of the composite minimization problem under investigation upon assuming

8: boundedness of the gradients at the iterates of the local functions and

9: the stepsizes being chosen appropriately, in the finite-dimensional setting. We also discuss generalizations and  limitations of the proposed algorithms and our techniques.

10: \end{abstract}

11: