the Dirichlet distribution

In the course, since we are still introducing some concepts of dependent distributions, we will talk about the Dirichlet distribution, which is a distribution over the simplex of https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri11.gif. Let https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri01.gif denote the Gamma distribution with density (on https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri03.gif)

https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri02.gif

Let https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri04.gif denote independent https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri05.gif random variables, with https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri06.gif. Then https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri07.gif where

https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri08.gif

has a Dirichlet distribution with parameter

https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri09.gif

Note that https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri10.gif has a distribution in the simplex of https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri11.gif,

https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri40.gif

and has density

https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri12.gif

We will write https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri13.gif.

The density for different values of https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri20.gif can be visualized below, e.g. https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri21.gif, with some kind of symmetry,
https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/dirichlet222.gif
or https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri22.gif and https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri23.gif, below
https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/dirichlet522.gif
and finally, below, https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri24.gif


Note that marginal distributions are also Dirichlet, in the sense that if

https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri13.gif

then

https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri14.gif

if https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri15.gif, and if https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri16.gif, then https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri17.gif‘s have Beta distributions,

https://f.hypotheses.org/wp-content/blogs.dir/253/files/2017/07/diri18.gif

See Devroye (1986) section XI.4, or Frigyik, Kapila & Gupta (2010) .This distribution might also be called multivariate Beta distribution. In R, this function can be used as follows

> library(MCMCpack)
> alpha=c(2,2,5)
> x=seq(0,1,by=.05)
> vx=rep(x,length(x))
> vy=rep(x,each=length(x))
> vz=1-x-vy
> V=cbind(vx,vy,vz)
> D=ddirichlet(V, alpha)
> persp(x,x,matrix(D,length(x),length(x))

(to plot the density, as figures above). Note that we will come back on that distribution later on so-called Liouville copulas (see also Gupta & Richards (1986)).