Multiobjective Nelder-Mead algorithm using a mesh-map of weighted sums
Patrick Charles Nadeau · cIRcle (University of British Columbia) · 2020
An algorithm for finding approximations to Pareto fronts in the optimization of multiple objective functions is presented. A mesh of weighted sums of the multiple objective functions serves as a model to approximate the Pareto front. The Nelder-Mead algorithm then solves these individual weighted sums without the use of derivatives. This multiobjective Nelder-Mead algorithm was found to be competitive with current algorithms on convex problems.