ON EFFICIENCY OF ADAPTIVE SEARCH IN OPTIMIZATION OVER SMALL IMPLICITLY DEFINED FEASIBLE REGION

Abstract

An optimization problem related to optimal design of processes in oil industry is considered. The problem is difficult because of small, implicitly defined feasible region. In such a case it is difficult not only to const-ruct a rational algorithm for search for minimum, but also to construct an algorithm for search for feasible points. Seve-ral algorithms for finding feasible points are proposed as well as several algorithms for optimization in a region approximated using points scattered in the region. A set of test functions is constructed to model the considered industrial optimization problems which normally are not suitable for testing of algorithms because of computational intensity. Testing results are presented, and conclusions about algorithms efficiency are drawn.