I am trying to reimplement the ‘ordered_logistic’ regression in Julia’s Turing probabilistic programming language, but am obtaining no variation in the samples. I suspect it is due to the ordered nature of the cutpoints parameter. Can someone please point me at the implementation of the ‘ordered’ type for Stan, so that I can mimic it in Julia? I have searched through stan-dev GitHub repository, but couldn’t find what I am looking for.
Thank you in advance,