Mathematica 9 is now available
Student Support Forum
-----
Student Support Forum: 'Multivariate normal : large number of variables' topicStudent Support Forum > General > Archives > "Multivariate normal : large number of variables"

Help | Reply To Topic
Author Comment/Response
Myeong Ae Kang
10/28/10 2:34pm

Hello.

Can I evaluate numerically the CDF of large number of variables with small number of different correlation coefficients(but, all nonzero)?
like more than 10000 random variables but less than 10 different correlation coefficients.

I tried NIntegrate[MultinormalDistribution[mean vector, Sigma matrix],integration range] with much less than 10000 number of RV's. But, what I got is "out of memory" message.

I just want to know whether I have to apply some approximations (which I have already) to reduce the number of variables, or there's a way to calculate it with Mathematica directly.

URL: ,
Help | Reply To Topic