EcoSta 2018: Registration
View Submission - EcoSta2018
A0780
Title: Numerical computation of the higher order central moments of the multivariate normal distribution Authors:  Fumiyasu Komaki - The University of Tokyo (Japan) [presenting]
Abstract: The higher order central moments of the multivariate normal distribution naturally appear in a problem of Bayesian prediction. Although a general formula (Isserlis' theorem) for the central moments of the multivariate normal distribution is widely known, it is not suitable for numerical evaluation of them. We investigate a method to evaluate the higher order central moments of the multivariate normal distribution by using MCMCMC. A class of discrete distributions closely related to the higher order moments is introduced. Applications of the method to Bayesian prediction are discussed.