MR Let (X1,X2,⋯,Xn) be a vector of chance variables with a nonsingular multivariate normal distribution. The problem is to evaluate P(X1>a1,⋯,Xn>an). The author obtains a reduction formula for this probability, involving integrals of partial derivatives of the probability with respect to the elements of the covariance matrix of (X1,⋯,Xn). For n=3 and n=4, the reduction formula enables the author to express the probability as a finite sum of single integrals of tabulated functions. These integrals have to be evaluated by numerical quadrature, but for certain cases simple approximations to them are given.
Description
The case of n=2 of the basic identity appears in Pearson 1901, http://archive.org/details/philtrans01501516
%0 Journal Article
%1 plackett1954reduction
%A Plackett, R. L.
%D 1954
%J Biometrika
%K Gaussian_processes integrals multivariate_Gaussian numerical_methods useful_identity
%P 351--360
%T A reduction formula for normal multivariate integrals
%U http://www.jstor.org/stable/2332716
%V 41
%X MR Let (X1,X2,⋯,Xn) be a vector of chance variables with a nonsingular multivariate normal distribution. The problem is to evaluate P(X1>a1,⋯,Xn>an). The author obtains a reduction formula for this probability, involving integrals of partial derivatives of the probability with respect to the elements of the covariance matrix of (X1,⋯,Xn). For n=3 and n=4, the reduction formula enables the author to express the probability as a finite sum of single integrals of tabulated functions. These integrals have to be evaluated by numerical quadrature, but for certain cases simple approximations to them are given.
@article{plackett1954reduction,
abstract = {[MR] Let (X1,X2,⋯,Xn) be a vector of chance variables with a nonsingular multivariate normal distribution. The problem is to evaluate P(X1>a1,⋯,Xn>an). The author obtains a reduction formula for this probability, involving integrals of partial derivatives of the probability with respect to the elements of the covariance matrix of (X1,⋯,Xn). For n=3 and n=4, the reduction formula enables the author to express the probability as a finite sum of single integrals of tabulated functions. These integrals have to be evaluated by numerical quadrature, but for certain cases simple approximations to them are given. },
added-at = {2012-10-18T06:54:51.000+0200},
author = {Plackett, R. L.},
biburl = {https://www.bibsonomy.org/bibtex/216aeb1bde598e2e51a73f8acd03edbb0/peter.ralph},
description = {The case of n=2 of the basic identity appears in Pearson 1901, http://archive.org/details/philtrans01501516},
fjournal = {Biometrika},
interhash = {8ed6e6d313cd674ab4b63d818328fe93},
intrahash = {16aeb1bde598e2e51a73f8acd03edbb0},
issn = {0006-3444},
journal = {Biometrika},
keywords = {Gaussian_processes integrals multivariate_Gaussian numerical_methods useful_identity},
mrclass = {60.0X},
mrnumber = {0065047 (16,377c)},
mrreviewer = {L. Weiss},
pages = {351--360},
timestamp = {2015-02-20T02:54:32.000+0100},
title = {A reduction formula for normal multivariate integrals},
url = {http://www.jstor.org/stable/2332716},
volume = 41,
year = 1954
}