Speaker: Prof. Jianlin Jiang
Time: 15:00-16:00, 4, April, 2018
Room: 1st Lecture Room, Jiulonghu Campus
Title:A splitting method based on alternating direction method and proximal point algorithm for three-block separable convex programming
Abstract: The alternating direction method of multipliers (ADMM) and proximal point algorithm (PPA) are effectively applied to solving separable convex programming problems with linearly constrains. In this paper, we propose a splitting method for three-block separable convex programming by adding an additional proximal term to some subproblems and changing the iterate order of multiplier of the classical ADMM. Moreover, the splitting method is ready to be accelerated by an over-relaxation strategy. We prove the global convergence and establish the O(1/t) convergence rate for the method. The proposed method inherits the advantages of ADMM and PPA and some satisfactory numerical results for several concrete application problems are reported to verify its efficiency
