Alexis I. Pospelov, Fedor V. Gubarev, Alexey M. Nazarenko
11th World Congress on Computational Mechanics (WCCM XI)
In many engineering applications phenomena to be studied are so complicated, that we have to submit to strict limitations on amount of physical and/or computer experiments. In this case one says that model under consideration is expensive to evaluate. Most practically relevant models of this sort do not provide any additional (i.e. analytical)information apart from plain model responses. Usually, these “black boxes” describe quite complex landscapes with many local minima.