An Approach to Minimize the Number of Function Evaluations in
Global Optimization

Tomi Haanpää

A new simple mathematical model is created for search algorithms
of global optimization to make the search more effective with
respect to the number of function evaluations. If a new solution
candidate is close to a vector for which the function value is far
from the so-far known best one, then there is no need to evaluate
the candidate. To detect promising candidates, a radial basis
function is adapted to available information of the objective
function without a need to solve any optimization subproblems.
Thus, the idea is to avoid increasing computational cost.