摘  要：We present several recently developed Augmented –Lagrangian -based splitting methods for minimizing a sum of separable functions with linear constraints. Under the assumption that several functions involving in the objective function are strongly convex, we show the global convergence of the direct extension of the classic alternating direction method of multipliers (ADMM), along with rate of convergence results. When all the functions are only convex, we introduce some recent methods, both parallel and successive in solving the subproblems. We show the convergence of the methods when there is a simple “correction” step and demonstrate their numerical efficiency by some preliminary numerical results.

             欢迎各位老师与广大研究生参加！