[R] [FORGED] Simulating multivariate gamma

[Rolf Turner]

On 25/07/15 07:59, gmoyeyemi wrote:

> Hi, I'm having problem simulating multivariate gamma. Is there
> anyone to assist in simulating multivariate gamma. I know that for
> multivariate normal, we can use; mvrnorm (n, means, sigma)

For starters you have to specify exactly what you mean by "multivariate 
gamma"; there is not a *unique* multivariate gamma distribution.

A little bit of googling on "multivariate gamma" will lead you to some 
useful material.  In particular there is a reference to the copula 
package which apparently provides a means of simulating a multivariate
distribution with gamma marginals.

It is not however clear to me how one specifies the desired correlation 
structure, or what the limitations on such a structure are.  I find the 
documentation to be opaque.

Perhaps someone with knowledge and insight on this issue will chime in.

cheers,

Rolf Turner

-- 
Technical Editor ANZJS
Department of Statistics
University of Auckland
Phone: +64-9-373-7599 ext. 88276