A Bi-level Globalization Strategy for Non-convex Consensus ADMM and ALADIN

TL;DR

This paper introduces a bi-level globalization strategy that simplifies and strengthens the global convergence analysis of non-convex consensus optimization methods like C-ADMM and C-ALADIN, with broad applicability.

Contribution

It proposes a novel bi-level globalization approach that guarantees global convergence with mild assumptions, simplifying proofs for non-convex consensus optimization algorithms.

Findings

C-ADMM achieves global convergence in non-convex cases.

C-ALADIN globally converges to local optima, complementing existing local convergence results.

The strategy requires only mild assumptions, making it practical for real-world applications.

Abstract

In this paper, we formally analyze global convergence in the realm of distributed consensus optimization. Current solutions have explored such analysis, particularly focusing on consensus alternating direction method of multipliers (CADMM), including convex and non-convex cases. While such efforts on non-convexity offer elegant theory guaranteeing global convergence, they entail strong assumptions and complicated proof techniques that are increasingly pose challenges when adopted to real-world applications. To resolve such tension, we propose a novel bi-level globalization strategy that not only guarantees global convergence but also provides succinct proofs, all while requiring mild assumptions. We begin by adopting such a strategy to perform global convergence analysis for the non-convex cases in C-ADMM. Then, we employ our proposed strategy in consensus augmented Lagrangian based alternating direction inexact Newton method (C-ALADIN), a more recent and generalization of C-ADMM. Surprisingly, our analysis shows that C-ALADIN globally converges to local optimizer, complementary to the prior work on C-ALADIN, which had primarily focused on analyzing local convergence for non-convex cases.