ADD-OPT: Accelerated Distributed Directed Optimization

被引：133

作者：

Xi, Chenguang ^{[1
]}

Xin, Ran ^{[1
]}

Khan, Usman A. ^{[1
]}

机构：

[1] Tufts Univ, Dept Elect & Comp Engn, Medford, MA 02155 USA

来源：

IEEE TRANSACTIONS ON AUTOMATIC CONTROL | 2018年 / 63卷 / 05期

关键词：

DEXTRA; directed graph; distributed optimization; linear convergence; DECENTRALIZED CONSENSUS OPTIMIZATION; CONVEX-OPTIMIZATION; SUBGRADIENT METHODS; ALGORITHM; NETWORKS; DIGRAPHS; GRAPHS; COMMUNICATION; COMPUTATION; SPECTRUM;

D O I：

10.1109/TAC.2017.2737582

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

In this paper, we consider distributed optimization problems where the goal is to minimize a sum of objective functions over a multiagent network. We focus on the case when the interagent communication is described by a strongly connected, directed graph. The proposed algorithm, Accelerated Distributed Directed OPTimization (ADD-OPT), achieves the best known convergence rate for this class of problems, O(mu(k)), 0 < mu < 1, given strongly convex, objective functions with globally Lipschitz-continuous gradients, where k is the number of iterations. Moreover, ADD-OPT supports a wider and more realistic range of step sizes in contrast to existing work. In particular, we show that ADD-OPT converges for arbitrarily small (positive) step sizes. Simulations further illustrate our results.

引用

页码：1329 / 1339

页数：11