Rev Language Reference

dnScaledDirichlet - Scaled Dirichlet Distribution

Scaled Dirichlet probability distribution on a simplex.

Usage

dnScaledDirichlet(RealPos[] alpha, RealPos[] beta)

Arguments

alpha : RealPos[] (pass by const reference)
The concentration parameter.
beta : RealPos[] (pass by const reference)
The rate parameter.

Domain Type

Simplex

Details

The scaled Dirichlet probability distribution is the generalization of the dirichlet distribution. A random variable from a scaled Dirichlet distribution is a simplex, i.e., a vector of probabilities that sum to 1. If b[1]=b[2]=...=b[n], then the scaledDirichlet(alpha,beta) collapses to the Dirichlet with the same alphas.

Example

# lets get a draw from a Dirichlet distribution
a <- [1,1,1,1]   # we could also use rep(1,4)
b <- [1,2,3,4]   # if these are all equal, the scaled Dirichlet is equivilent to the Dirichlet(a)x ~ dnScaledDirichlet(a,b)
x
# let check if b really sums to 1
sum(x)

See Also

Edit this page