Just wondering if there are plans for the inclusion of
multivariate noncentral hypergeometric distributions in Distributions.jl? In particular, I need to make use of Wallenius’ version. If it’s not on the cards anytime soon and those in the know (not me) think it would be a relatively straightforward modification of the univariate case, then I’ll happily have a crack at it and open a pr.
It think it would be best to ask in an issue at
That said, if you make a PR I don’t see why anyone would object to merging it; but even if it does not end up being merged you can just put it in its own package.
I wasn’t sure of the etiquette around filing a feature request as an issue before making any attempt to contribute.
Hello, did you end up with your own implementation or did you make a PR on Distributions.jl ?
Neither! This fell by the wayside for now, as other projects took priority. In the next week or two, I will most likely whip up a quick and dirty function to directly update the log likelihood in Turing. Then hopefully refine it (and add sampling function etc) and make a PR.
Thanks. For info, it has been added on SciPy in the latest version (1.6)…