2018/10/29 by Marie Maros, Maros, Marie, Joakim Jaldén +1
Computer Science · #Distributed Control Multi-Agent Systems #FOS: Mathematics #Optimization and Control (math.OC) #Stochastic Gradient Optimization Techniques
paper · pdf · doi:10.48550/arxiv.1810.12240
openalex publication_date 2018/10/29 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
In this paper we consider distributed convex optimization over time-varying\nundirected graphs. We propose a linearized version of primarily averaged\nnetwork dual ascent (PANDA) while requiring less computational costs. The\nproposed method, economic primarily averaged network dual ascent (Eco-PANDA),\nprovably converges at R-linear rate to the optimal point given that the agents'\nobjective functions are strongly convex and have Lipschitz continuous\ngradients. Therefore, the method is competitive, in terms of type of rate, with\nboth DIGing and PANDA. The proposed method halves the communication costs of\nmethods like DIGing while still converging R-linearly and having the same per\niterate complexity.\n