ADAPTIVE APPROXIMATION MODELS IN OPTIMIZATION

The paper proposes a method for optimization of functions of several v ariables that substantially reduces the number of objective function e valuations compared to traditional methods. The method is based on the property of iterative refinement of approximation models of the optim and function in approximation domains that contract to the extremum po int. It does not require subjective specification of the starting poin t, step length, or other parameters of the search procedure. The metho d is designed for efficient optimization of unimodal functions of seve ral (not more than 10-15) variables and can be applied to find the glo bal extremum of polymodal functions and also for optimization of scala rized forms of vector objective functions.