Consensus-based Distributed Optimization for Multi-agent Systems over Multiplex Networks

arXiv (Cornell University)(2023)

引用 0|浏览0
暂无评分
摘要
Multilayer networks refer to systems composed of several interacting networks. This approach provides a generalized framework for exploring real-world and engineering systems in more depth and with greater accuracy than the traditional single-layer network scheme. Despite the wide set of results in distributed optimization in single-layer networks, there is a lack of developments in multilayer systems. In this paper, we propose two algorithms for distributed optimization problems using the supra-Laplacian matrix of a multiplex network and its diffusion dynamics. The algorithms are a distributed primal-dual saddle-point algorithm and a distributed gradient descent algorithm. We establish a set of saddle-point equations based on the relation between consensus and diffusion dynamics via the graph Laplacian and impose an analogous restriction to the traditional Laplacian in our multiplex distributed optimization problem to obtain the multiplex supra-Laplacian matrix. We extend the distributed gradient descent algorithm for multiplex networks using this supra-Laplacian matrix. We propose several theoretical results to analyze the convergence of both algorithms, demonstrating that they converge to the unique optimizer. We validate the proposed algorithms using numerical examples. We also explore the existence of a phase transition in the consensus time due to the interlayer diffusion constant in the numerical examples. Finally, a coordinated dispatch for interdependent infrastructure networks (energy-gas) is presented to illustrate the application of the proposed framework to real engineering problems.
更多
查看译文
关键词
distributed optimization,networks,consensus-based,multi-agent
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要