David Widmann

I think a test of the interface does not require AD tests (https://github.com/JuliaMath/AbstractFFTs.jl/pull/76#issuecomment-1229551356)? I assumed one would only want to check whatever is required by the interface, i.e., basically what's...

The p-Wasserstein distance between two univariate histograms `H` and `H'` (ie. sum of Diracs) with the same number `n` of components is just `1/n ||H - H'||_p^p` (see e.g. eq...

It seems this optimization problem can be solved analytically at least for `p = 2`. Using Langrange multipliers it seems (better check it carefully, I just did a quick sketch)...

The diagnostics in MCMCChains already work with `AbstractArray`, so they do not require the `MCMCChains.Chains` data structure and should already be generally usable. However, I guess it would be nice...

I assumed it's best to just try and see how lightweight or heavy the diagnostic part of MCMCChains is: https://github.com/TuringLang/MCMCChains.jl/issues/266#issuecomment-843001213

Can't you use the same approach as in https://github.com/FluxML/Functors.jl/issues/6#issuecomment-744895426?

Yes, my rough idea was similar. However, my main idea was to work with untransformed values by default, i.e. to implement basically only the type definition you proposed and ```julia...

> Aside from this, there is also [JuliaMath/FFTW.jl#182](https://github.com/JuliaMath/FFTW.jl/issues/182) , so that ScaledPlans miss the `region` subfield, which could be mitigated by a function like I opened #65 a while ago...

PythonPlot is the PythonCall version of PyPlot: https://github.com/stevengj/PythonPlot.jl I've never used it, so I can't comment on the differences.

Can we release a new version that is compatible with AbstractMCMC 0.5?