 Log-uniform distribution

For $0<A<B<\infty$, this density makes $\log X$ uniform between $\log A$ and $\log B$. It is a proper finite-interval version of the scale-invariant prior. The chosen logarithm base changes only the endpoint coordinates.