Rev Language Reference

`dnLog` - Log-scaled distribution

Log-scales a given distribution.

Usage

dnLog(Distribution__Real logDistribution)

Arguments

logDistribution :	Distribution__Real (pass by const reference)
	The distribution in log-space.

Domain Type

RealPos

Details

If X ~ dist then exp(X) ~ dnLog(dist) This provides a way to construct distributions like dnLognormal and dnLoguniform directly from the underlying distribution in log-space. It can therefore express distributions that are not directly implemented. The distribution `dist` can be either univariate (dnNormal) or multivariate (dnMultivariateNormal).

Example

x ~ dnLog(dnNormal(0,1))          # Draw from the log-Normal distribution
x ~ dnNormal(0,1) |> dnLog()      # Expressed using pipes.
x ~ dnLognormal(0,1)              # This is equivalent.
y ~ dnNormal(0,1)
x := exp(y)                       # This is also equivalent.

x ~ dnLog(dnGamma(2,3))           # There is no equivalent for this.
x ~ dnIID(10,dnLog(dnGamma(2,3))) # Draw 10 log-Gamma(2,3) random variables.

mu = [1.0, 2.0, 3.0, 4.0]
Sigma ~ dnWishart(df=4, kappa=2, dim=4)
x ~ dnLog(dnMultivariateNormal(mu,Sigma))