Discrete-Time Zero-Gradient-Sum Algorithm for Distributed Optimization over Directed Networks
Xinyi Zhao, Weifeng Gao, Jin Xie · 2021
This paper proposes a discrete-time Zero-Gradient-Sum (ZGS) algorithm over directed networks. It is used to solve the distributed convex optimization problem, which is involved in distributed learning algorithms over directed networks under the background of big data. The proposed ZGS algorithm converges to the globally optimal solution and achieves exponential convergence, whose validity is proved by both theory and numerical simulations.