Sunshine: A novel random search for continuous global optimization

Abstract

Random search algorithms are widely used in many ill-structured global optimization problems. This wide application is due to random search algorithms' capability to model and solve continuous, discrete, or hybrid problems. Moreover, the researchers discuss that the random search algorithms yield a proper solution in terms of fitness and time consumed. However, these algorithms lack guarantee of achieving the global optimum. Regarding the discussed researches, this paper considers the most critical shortcoming of studied algorithms as getting trapped in local optimums. Focusing on continuous global optimization problems, a novel algorithm is proposed. This algorithm, called "SUNSHINE", fulfills the aforementioned shortcoming. Besides, the other advantages of SUNSHINE, including efficient time complexity, robustness, and low sensitivity of accurate adjustment of parameters, are illustrated through a comprehensive case study. Moreover, the paper discusses the capability of SUNSHINE in parallel implementation.