Parallel alternating direction method of multipliers

作     者：Yan, Jiaqi Guo, Fanghong Wen, Changyun Li, Guoqi 

作者机构：Nanyang Technol Univ Singapore Singapore Zhejiang Univ Technol Hangzhou Zhejiang Peoples R China Tsinghua Univ Beijing Peoples R China 

出 版 物：《INFORMATION SCIENCES》 (信息科学)

年 卷 期：2020年第507卷

页      面：185-196页

核心收录：

学科分类：12[管理学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：National Natural Science Foundation of China Zhejiang Provincial Natural Science Foundation of China [LQ19F030008] 

主　　题：Distributed optimization ADMM Parallel algorithm 

摘      要：In this paper, we consider the distributed optimization problem, where the objective function is the sum of local cost functions. To solve this problem, a new parallel Alternating Direction Method of Multipliers (ADMM) algorithm is developed, which guarantees that the agents cooperatively reach an optimal agreement. Different from most of the existing ADMM approaches, our algorithm allows all the agents to update their local variables simultaneously in a parallel manner. It is theoretically proved that the local solutions of all the agents could reach a consensus, and converge to the optimal solution asymptotically with the rate of O(1/k). Numerical examples are finally provided to validate the effectiveness of the proposed method. (C) 2019 Elsevier Inc. All rights reserved.