Andreas Noack
Andreas Noack
Yes. We'd have to open some PRs against those packages to fix them but I don't think it should too much work.
Dahua factored these out without much discussion and announced it in https://groups.google.com/forum/#!topic/julia-stats/nOObNwKM1b8. As you can see from the thread, I also thought it was a good idea but less than...
Such functionality is actively being worked on right now so hopefully it'll be availiable within a couple of months.
I'm not completely sure what to do with the issue mentioned in https://github.com/JuliaStats/StatsFuns.jl/pull/148#discussion_r1030585347
We should probably have a general derivative rule for `cdf` defined somewhere. @devmotion any thoughts?
My point is that the derivative of `x -> cdf(..., x)` is readily available. However, supporting all the partial derivates of `betacdf` will require more work.
I think many of these were relaxed at a time when we discussed the proper location in the type tree for dual numbers. Initially, they were not `Real` so to...
I've transfered this issue to StatsFuns since the issue is in https://github.com/JuliaStats/StatsFuns.jl/blob/5e866baa763ff0f5d279185c1923e1cd87ea5699/src/distrs/fdist.jl#L21-L29 where we make use of the relationship between the F and Beta distributions. However, this relationship becomes numerically...
Great. You can do it in one or two PRs. Either is fine. Adding a comment is almost always a good idea.
@bjarthur Do you have a use case where it would be convenient to handle the `NaN`s instead of the exception?