The Barycenter Method for Direct Optimization: an Overview

Abstract

A randomized version of the recently developed barycenter method for derivative--free optimization has desirable properties of a gradient search. We developed a complex version to avoid evaluations at high--gradient points. The method, which is also applicable to non--convex and to non--smooth functions, is parallelizable in a natural way and shown to be robust under noisy measurements. The goal of this paper is to present an overview of the method, whose properties make it particularly useful in control applications.