Analysis of the Alternating Direction Method of Multipliers for Nonconvex Problems

This work investigates the theoretical performance of alternating-direction method multipliers (ADMM) as it applies to nonconvex optimization problems, and in particular, problems with constraint sets. The alternating direction is an that has largely been analyzed for convex problems. ultimate goal assess what kind convergence properties case, this end, contributions are twofold. First, analyzes local optimal solution ADMM subproblems, which contrasts much analysis requires global solutions subproblems. Such a consideration important practical implementations. Second, established still satisfies result. concludes some more detailed discussion how relates previous work.