Distributed Optimization and Control with ALADIN

Authors

B. Houska, Y. Jiang

Reference

In T. Faulwasser, M.A. Müller, and K. Worthmann (Editors),
Recent Advances in Model Predictive Control:
Theory, Algorithms, and Applications, pages 135-163, Springer, 2021.

Abstract

This chapter aims to give a concise overview of distributed optimization and control algorithms based on the Augmented Lagrangian based Alternating Direction Inexact Newton (ALADIN) method. Here, our goal is to provide a tutorial-style introduction to this relatively new distributed optimization algorithm. In contrast to other existing algorithms, which are often tailored for convex optimization problems, ALADIN is particularly suited for solving non-convex optimization problems. Moreover, another principal advantage of ALADIN is that it can achieve a super-linear or even quadratic convergence rate if suitable Hessian approximations are used.

Download

Distributed Optimization and Control with ALADIN Chapter

Bibtex

@INBOOK{Houska2021,
author = {Houska, B. and Jiang, Y.},
chapter = {Distributed Optimization and Control with {ALADIN}},
booktitle ={Recent Advances in Model Predictive Control: Theory, Algorithms, and Applications},
editors = {Faulwasser, T. and M"uller, M.A. and Worthmann, K.},
publisher = {Springer},
pages = {135–163},
year = {2021}
}