Computational Mathematics

Use of the parallel characteristical algorithms for solving multivariate problems of global optimization

K. A. Barkalov

N.I. Lobachevsky State University of Nizhni Novgorod (Nizhni Novgorod, Russian Federation)

Abstract: In this paper the problems of multidimensional multiextremal optimization and multilevel scheme of dimension reduction are considered. The proposed scheme allows to reduce solution of multidimensional problems to solution of a number of subproblems with less dimension, which can be solved in parallel. The multilevel scheme combines the ideas of Peano-type space filling curves and nested optimization. To solve the reduces subproblems the parallel characteristical algorithm is used. Results of numerical experiments confirm convergence and speedup of the parallel algorithm.

Keywords: global optimization, multiextremal functions, dimension reduction, characteristical algorithms, parallel algorithms.

UDC: 519.853.4

Received: 11.08.2014

DOI: 10.14529/cmse140409