摘要：AbstractIn computer experiment framework, the choice in a design of experiments is an important concern. When there is no information about the black box function to be approximated, an exploratory design is usually taken. Two well-known exploratory criteria are minimax and maximin. In a hypercube the standard strategy consists of taking a maximin design within a class of Latin hypercube designs. However, in a non hypercubic context, it does not make sense to use the Latin hypercube sampling. A theoretical justification to maximin and minimax criteria with regard to the kernel interpolation is first provided. Then, a simulated annealing algorithm is proposed in order to find a maximin design in any bounded connected Domain.
