The rank transformation applied to a multivariate method of global optimization

Rank transformation has been used successfully in a recently proposed nonparametric global optimization method. The nonparametric method determines the location of its next guess based on the rank-transformed objective function evaluations rather than on the actual function values themselves. Here, application of the rank transformation to the multivariate method of global optimization is shown to significantly reduce the number of function evaluations needed for convergence within a specified tolerance