Sample and Computationally Efficient Stochastic Kriging in High Dimensions

High-dimensional Simulation Metamodeling Stochastic kriging has been widely employed for simulation metamodeling to predict the response surface of complex models. However, its use is limited cases where design space low-dimensional because sample complexity (i.e., number points required produce an accurate prediction) grows exponentially in dimensionality space. The large size results both a prohibitive cost running model and severe computational challenge due need invert covariance matrices. To address this long-standing challenge, Liang Ding Xiaowei Zhang, their recent paper “Sample Computationally Efficient Kriging High Dimensions”, develop novel methodology — based on tensor Markov kernels sparse grid experimental designs that dramatically alleviates curse dimensionality. proposed theoretical guarantees shows outstanding performance numerical problems as high 16,675 dimensions.