David Widmann

Results 1463 comments of David Widmann

I think it would be good to make these improvements right away, without the delay and discussions around a breaking release.

Maybe PosteriorStats.jl provides what you're looking for? https://julia.arviz.org/PosteriorStats/stable/api/

Possibly related: https://github.com/TuringLang/TuringTutorials/issues/420

> We're somehow defining a likelihood based on the mean of the SDE simulations We don't base it on the mean of the SDE simulations, do we? I don't remember...

AFAICT the test errors on Julia 1.0 are caused by the way that Pkg resolves and installs test dependencies on such old Julia versions. When the package is built, the...

I don't expect the AbstractFFTs design to change any time soon since nothing has happened in the last four years (Keno opened the issue in AbstractFFTs linked above in August...

I don't think there is any other way to detect whether a AbstractFFTs backend is available than calling the FFT methods and checking whether a `MethodError` is thrown. To improve...

> the UntypedVarInfo vi has a value for x that is internally stored as a Vector{Real} Could this be changed? In my opinion, matrix multiplication shouldn't promote containers of `Real`s...

> There are checks in the functions for the domains. The point of NaNMath for me (and others I guess) is that `sqrt`, `pow` etc. don't error. That's not guaranteed...