EOS: a Parallel, Self-Adaptive, Multi-Population Evolutionary Algorithm for Constrained Global Optimization

This paper presents the main characteristics of the evolutionary optimization\ncode named EOS, Evolutionary Optimization at Sapienza, and its successful\napplication to challenging, real-world space trajectory optimization problems.\nEOS is a global optimization algorithm for constrained and unconstrained\nproblems of real-valued variables. It implements a number of improvements to\nthe well-known Differential Evolution (DE) algorithm, namely, a self-adaptation\nof the control parameters, an epidemic mechanism, a clustering technique, an\n$\\varepsilon$-constrained method to deal with nonlinear constraints, and a\nsynchronous island-model to handle multiple populations in parallel. The\nresults reported prove that EOSis capable of achieving increased performance\ncompared to state-of-the-art single-population self-adaptive DE algorithms when\napplied to high-dimensional or highly-constrained space trajectory optimization\nproblems.\n