A self-switching base vector selection mechanism for differential mutation of differential evolution algorithm

The Evolutionary Computing (EC) field of Computer Science has a pool of potential optimization algorithms. They are collectively termed as Evolutionary Algorithms (EAs). The Differential Evolution (DE) algorithm is added recently to this pool. It is known for its simplicity and applicability. DE differs from other EAs by its Differential Mutation logic. There exist many strategies for this mutation logic. This paper proposes a new mutation strategy which employs a self-switching base vector selection mechanism. This self-switching mechanism uses the diversity in the population as a measure to select the base vector either randomly or the best candidate in the population. It is found from the results of the experiments carried out that the proposed mechanism works better on a standard set of benchmarking functions solving unimodal optimization problems.