GENERATING RANDOM DEVIATES FROM MULTIVARIATE PEARSON DISTRIBUTIONS

A general method is presented for generating multivariate Pearson random deviates. Based on joint product moments, the technique employs the nested-conditional factorization approach for multivariate densities. For multivariate Pearson distributions, coefficients of differential equations corresponding to a univariate marginal density, a bivariate density, a trivariate density, and so on, can be determined by solving linear equation systems that are constructed on the basis of joint moments to fourth order. These sets of coefficients are utilized to determine conditional moments and appropriate univariate Pearson distributions from which individual deviates are generated using existing techniques. The method can be used in an approximate sense for distributions not of the Pearson class. © 1990.