Parametrisation of generalised gamma distribution amenable to Stan

[tillahoffmann]

@Richie_M

I may just also cite this Stan forum URL as your simulation result is gamma distribution specific. For your metric n_eff / n, did you use a large n, close to the magnitude of a n_eff at least 1000 for well-mixed chains)?

I have to admit that I don’t quite remember, but I’ll have a look whether I can find the notebook with the experiments I ran; they’re on an older computer that I’ll have to start up. Fit a beta distribution that captures correlation in the fitted parameters? may also be relevant and might have other pointers (although I’ve only skimmed it).