Distributed Prediction‐Correction Algorithm for Convex Optimization With Coupled Constraints

Abstract

ABSTRACT In this article, we propose an optimization algorithm based on a distributed predictive‐correction framework, called DPCA, for solving coupled constrained convex optimization problems in undirected connected multi‐agent networks. Based on the augmented Lagrangian, the algorithm which combines the generalized proximal point algorithm and the idea of the unified prediction‐correction framework can achieve global optimization through local communication and computation in the case of incomplete information sharing. We demonstrate that DPCA attains a convergence rate of regarding both optimality and feasibility. Finally, the simulation results show that DPCA has better convergence performance and smaller convergence error than other algorithms.