Closed rezayas closed 7 years ago
I can't do Weibull mean/stdev, because those are (a) complicated, and (b) require inverse gamma functions.
I can't do Dirichlet; that's a multidimensional distribution, and in the K=2 case it's just beta.
Multinomial is out too.
Let's find a way to also implement multi-dimensional distributions. We need them!
That's done. Ideally, we could do Dirichlet-multinomial as well, but that's a bit tricky on the pdf front. On the simple generation front, it's a lot easier; just pick a random from Dirichlet and feed to multinomial.
Continuous distributions
Discrete distributions