Asynchronous Differential Evolution with Strategy Adaptation for Global Numerical Optimization

Abstract

A new variant of Differential Evolution (DE) algorithm called A-SaDE algorithm is proposed. A-SaDE algorithm is based on two DE variants, Self-adaptive DE (SaDE) and Asynchronous DE (ADE) algorithms. SaDE algorithm is one of the well-known DE algorithms and shows powerful optimization performance by automatically tuning the mutation strategies as well as the control parameters. The asynchronous scheme is recently proposed, and it helps to find better solutions by increasing the greediness. We incorporated these two algorithms called A-SaDE algorithm and tested it with 13 scalable benchmark problems. The experimental results confirm that A-SaDE algorithm outperforms original SaDE algorithm, especially for unimodal functions.