Singular value decomposition as an equation solver in co-kriging matrices

Abstract

One of the most significant elements in solving the co-kriging equations is the matrix solver. In this paper, the singular value decomposition (SVD) as an equation solver is proposed to solve the co-kriging matrices. Given that other equation solvers have various drawbacks, the SVD presents an alternative for solving the cokriging matrices. The SVD is briefly discussed, and its performance is compared with the banded Gaussian elimination that is most frequently used in co-kriging matrices by means of case studies. In spite of the increase in the memory requirement, the SVD yields better results.