A Modified Approach to Distributed Bregman ADMM for a Class of Nonconvex Consensus Problems
Zhonghui Xue, Qianfeng Ma, Yazheng Dang · Journal of Mathematics · 2025
This article presents a refined iteration of the distributed Bregman alternating direction method of multipliers (ADMM) tailored to tackle nonconvex consensus issues, especially those with multiple blocks. The reliability of this novel approach is established through demonstrating its robust convergence under specific conditions. These conditions entail the requirement that the potential function satisfies the Kurdyka–Łojasiewicz property and that the penalty parameter exceeds a predefined constant. Initial numerical trials have shown encouraging outcomes, suggesting notable efficiency enhancements in the refined distributed Bregman ADMM algorithm.