A primal-dual algorithm for distributed optimization

Consider a set of N agents who cooperate to solve the problem infx equation where the convex cost functions (f n , g n ) are local to the agent n. It is assumed that the functions f n are differentiable and have Lipschitz gradients. In this paper, a primal-dual algorithm for distributively solving this problem is proposed. This algorithm is an instance of a primal-dual algorithm separately introduced by Vu and Condat.