English

A Linearly Convergent Robust Compressed Push-Pull Method for Decentralized Optimization

Optimization and Control 2023-03-14 v1

Abstract

In the modern paradigm of multi-agent networks, communication has become one of the main bottlenecks for decentralized optimization, where a large number of agents are involved in minimizing the average of the local cost functions. In this paper, we propose a robust compressed push-pull algorithm (RCPP) that combines gradient tracking with communication compression. In particular, RCPP is compatible with a much more general class of compression operators that allow both relative and absolute compression errors. We show that RCPP achieves linear convergence rate for smooth objective functions satisfying the Polyak-{\L}ojasiewicz condition over general directed networks. Numerical examples verify the theoretical findings and demonstrate the efficiency, flexibility, and robustness of the proposed algorithm.

Keywords

Cite

@article{arxiv.2303.07091,
  title  = {A Linearly Convergent Robust Compressed Push-Pull Method for Decentralized Optimization},
  author = {Yiwei Liao and Zhuorui Li and Shi Pu},
  journal= {arXiv preprint arXiv:2303.07091},
  year   = {2023}
}
R2 v1 2026-06-28T09:14:03.882Z