Two timescale extra for smooth non-convex distributed optimization problems

作     者：Peng, Zeyu Farokhi, Farhad Pu, Ye 

作者机构：Department of Electrical and Electronic Engineering University of Melbourne Australia 

出 版 物：《arXiv》 (arXiv)

年 卷 期：2024年

核心收录：

主　　题：Convex optimization 

摘      要：We propose Two-timescale EXTRA (TT-EXTRA), extending the well-known EXact firsT-ordeR Algorithm (EXTRA) by incorporating two stepsizes, for distributed non-convex optimization over multi-agent networks. Due to the two-timescale strategy, we are able to construct a suitable Lyapunov function and establish the sub-linear convergence to consensual first-order stationary points. Additionally, we introduce a sequential parameter selection method and the numerical results support the theoretical guarantees. © 2024, CC BY-NC-SA.