Global optimization by an improved differential evolutionary algorithm (Q2371478)

The following minimization problem is considered: \[ \text{Minimize }f(x)\text{ subject to }x\in S\subseteq\mathbb{R}^n, \] where \(f(x)\) may be nonlinear, non-convex and non-differentiable. A global minimizer \(x^*\in S\) satisfying \(f(x^*)\leq f(x)\) for all \(x\in S\) should be found. The paper focuses on the case of a box constraint and non-differentiable objective function. In order to improve the performance of the differential evolutionary algorithm (DE-algorithm) a new hybrid evolutionary algorithm is proposed. The new algorithm employs a pattern search method to speed up the local exploiting; and the differential operators used in the DE-algorithm help to jump to a better point. A good computational experience with solving numerical test examples with 13 high-dimensional objective functions using the proposed method is reported in the concluding part of the paper.