Mohamed Tarek

Yes that's correct. But use a type parameter for the `d` field.

Oh sorry it should be `b.orig.d`.

@bdeonovic it would be nice if you could write a blog post or something about this if you can. This is an important side of Turing, custom distributions and bijectors,...

Well, `inv(b::Inverse)` is equivalent to `b.orig` so may be use the first if you don't want to access internal field names. The bijector type `Inverse` just stores the original bijector...

Thanks for your kind words :)

Btw I am almost done with a more "proper" fix for this stochastic support issue but it does add a fair bit of complexity to the internals. So we still...

Hmm I think we can empty the cache automatically after sampling.

Hi Chad, I thought it's about time we integrate Soss and Turing! So I think the easiest path forward is to: 1. Define a function that converts a Soss model...

So after https://github.com/TuringLang/Bijectors.jl/pull/99 is merged (soon), you can get `link`, `invlink` and `logpdf_with_trans` defined automatically if you define `logpdf` and `bijector` for your distribution. Subtyping Distribution is important yes unfortunately....

>What if some values are discrete? Bijectors handles discrete variables no problem using the `Identity` bijector.