Multidimensional Global Optimization Method Using Numerically Calculated Derivatives

In the present paper, an efficient method for solving complex computationally consuming multiextremal global optimization problems is considered. To organize the adaptive global search, both the objective function and its first partial derivatives are supposed to satisfy the Lipschitz condition. The proposed approach deals with numerical estimates of the derivatives, which makes it possible to essentially reduce the amount of required computations with a nested dimension reduction. The results of the performed experiments prove the prospects of the developed approach.