Submodular Optimization for Consensus Networks with Noise-Corrupted\n Leaders

2017/12/21 by Erika Mackin, Mackin, Erika, Stacy Patterson +1
Biochemistry, Genetics and Molecular Biology · Computer Science · Mathematics · #Distributed Control Multi-Agent Systems #FOS: Electrical engineering #FOS: Mathematics #Gene Regulatory Network Analysis #Markov Chains and Monte Carlo Methods #Optimization and Control (math.OC) #Systems and Control (eess.SY) #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.1712.08212

openalex publication_date 2017/12/21 · openalex created_date 2022/08/15 · openalex updated_date 2026/07/28

Abstract

We consider the leader selection problem in a network with consensus dynamics\nwhere both leader and follower agents are subject to stochastic external\ndisturbances. The performance of the system is quantified by the total\nsteady-state variance of the node states, and the goal is to identify the set\nof leaders that minimizes this variance. We first show that this performance\nmeasure can be expressed as a submodular set function over the nodes in the\nnetwork. We then use this result to analyze the performance of two greedy,\npolynomial-time algorithms for leader selection, showing that the leader sets\nproduced by the greedy algorithms are within provable bounds of optimal.\n

Related

• A Resistance Distance-Based Approach for Optimal Leader Selection in Noisy Consensus Networks
• Minimizing Convergence Error in Multi-Agent Systems via Leader\n Selection: A Supermodular Optimization Approach
• Optimal k-Leader Selection for Coherence and Convergence Rate in One-Dimensional Networks
• Joint Centrality Distinguishes Optimal Leaders in Noisy Networks
• Leader Selection in Multi-Agent Networks with Switching Topologies via Submodular Optimization