Fast and stable nonconvex constrained distributed optimization: the ELLADA algorithm

Distributed optimization using multiple computing agents in a localized and coordinated manner is promising approach for solving large-scale problems, e.g., those arising model predictive control (MPC) of plants. However, distributed algorithm that computationally efficient, globally convergent, amenable to nonconvex constraints remains an open problem. In this paper, we combine three important modifications the classical alternating direction method multipliers optimization. Specifically, (1) extra-layer architecture adopted accommodate nonconvexity handle inequality constraints, (2) equality-constrained nonlinear programming (NLP) problems are allowed be solved approximately, (3) modified Anderson acceleration employed reducing number iterations. Theoretical convergence proposed algorithm, named ELLADA, established its numerical performance demonstrated on NLP benchmark Its application MPC also described illustrated through process system.