Fully Distributed Continuous-Time Algorithm for Nonconvex Optimization over Unbalanced Directed Networks

TL;DR

This paper presents a novel fully distributed continuous-time algorithm for nonconvex optimization over unbalanced directed networks, achieving convergence without requiring convexity or global network information.

Contribution

It introduces a new algorithm based on topology balancing and adaptive control that guarantees convergence without prior knowledge of network parameters or convexity.

Findings

Algorithm converges asymptotically to the optimal solution.

Removes dependence on the smallest strong convexity constant.

Validated with two illustrative examples.

Abstract

This paper investigates the distributed continuous-time nonconvex optimization problem over unbalanced directed networks. The objective is to cooperatively drive all the agent states to an optimal solution that minimizes the sum of the local cost functions. Based on the topology balancing technique and adaptive control approach, a novel fully distributed algorithm is developed for each agent with neither prior global information concerning network connectivity nor convexity of local cost functions. By viewing the proposed algorithm as a perturbed system, its input-to-state stability with a vanishing perturbation is first established, and asymptotic convergence of the decision variables toward the optimal solution is then proved under the relaxed condition. A key feature of the algorithm design is that it removes the dependence on the smallest strong convexity constant of local cost functions, and the left eigenvector corresponding to the zero eigenvalue of the Laplacian matrix of unbalanced directed topologies. The effectiveness of the proposed fully distributed algorithm is illustrated with two examples.