H. Trautmann

Hot off the Press: An Evolutionary Approach for the Computation of ϵ-Locally Optimal Solutions for Multi-Objective Multimodal Optimization

Here we briefly summarize the main findings of the above-mentioned paper by Hernández et al., 2024 [3]. In this paper, we address the computation of finite-size approximations of the set of ϵ-locally optimal solutions of a multi-objective optimization problem (MOP), a problem relevant in multi-objective multimodal optimization (MMMO). We propose a bounded archiver, ArchiveUpdateLQ,ϵB, the algorithm LQ,ϵMOEA, which directly uses this archiver in selection, and a hybrid with a multi-objective continuation method for improved accuracy when gradient information is available. Numerical results demonstrate the benefits of the proposed methods.