Abstract
This chapter proposes a novel algorithm for handling high dimensionality in large scale problems. The proposed algorithm, here indicated with Differential Evolution for Large Scale problems (DELS) is a Differential Evolution (DE) based Memetic Algorithm with self-adaptive control parameters and automatic population size reduction, which employs within its framework a variation operator local search. The local search algorithm is applied to the scale factor in order to generate high quality solutions. Due to its structure, the computational cost of the local search is independent on the number of dimensions characterizing the problem and thus is a suitable component for large scale problems. The proposed algorithm has been compared with a standard DE and two other modern DE based metaheuristics for a varied set of test problems. Numerical results show that the DELS is an efficient and robust algorithm for highly multivariate optimization, and the employment of the local search to the scale factor is beneficial in order to detect solutions with a high quality, convergence speed and algorithmic robustness.