Abstract:
The paper considers the parallel global optimization algorithm and discusses its implementation with the Intel oneAPI toolkit. We suppose that the objective function is given as a black-box and satisfies the Lipschitz condition. The parallel algorithm presented in the paper uses the scheme of dimensionality reduction employing the Peano curve, which continuously maps an interval of the real axis onto a hypercube. The Intel oneAPI tools, that allows one to write the same code for both the central processor and the graphics accelerator, were used for implementation of the parallel global optimization algorithm. The results of numerical experiments obtained by solving a series of time-consuming multiextremal optimization problems are presented.