The Procedure of Kriging Variance Estimation Based on Semi parametric Bootstrapping in Deterministic Simulation

摘要

A new procedure, semi parametric bootstrapping, is proposed to estimate kriging variance that is not biased in deterministic simulation. There are two new algorithms generated from this procedure. They are direct bootstrapping and non-direct bootstrapping algorithms. To test these procedures in order to work well, a multi-modal test function is used with varied input dimensions (which enable high dimensions). The treatment for the generic estimator of the two algorithms and an increase of the estimator based on plug-in kriging prediction are examined in two things. First, it is based on the increase of bootstrap sample and second there are many points observed. Based on the results of the simulation, the general correlation among the three estimators is obtained.