Distributed Optimization of Finite Condition Number for Laplacian Matrix in Multi-Agent Systems

摘要

This paper addresses the distributed optimization of the finite condition number of the Laplacian matrix in multi-agent systems. The finite condition number, defined as the ratio of the largest to the second smallest eigenvalue of the Laplacian matrix, plays an important role in determining the convergence rate and performance of consensus algorithms, especially in discrete-time implementations. We propose a fully distributed algorithm by regulating the node weights. The approach leverages max consensus, distributed power iteration, and consensus-based normalization for eigenvalue and eigenvector estimation, requiring only local communication and computation. Simulation results demonstrate that the proposed method achieves performance comparable to centralized LMI-based optimization, significantly improving consensus speed and multi-agent system performance. The framework can be extended to edge weight optimization and the scenarios with non-simple eigenvalues, highlighting its scalability and practical applicability for large-scale networked systems.