Chad Scherrer

I guess I could overload `mcmcsample`?

Thanks @devmotion , I was hoping to allow a convergence criteria as a stopping condition, so this is great. There does seem to be an assumption that everything the user...

I can't return them as `state`, that would make them unavailable since `save!!` doesn't include `state` as an argument. I think everything needs to be in `sample`, then I can...

For MeasureTheory, we need an instance for `rand` and one for `logdensity`. I think `rand` can just return a `Sampled`. I'm not sure yet about `logdensity`. Also, after making this...

https://github.com/thautwarm/JuliaVariables.jl/pull/4

Hi @tk3369 , I have a start on this here: https://github.com/JuliaArrays/StructArrays.jl/pull/160 I think the approach will work, just a question of integrating into what's there. Possible it could need a...

Oops, I had missed the solution above. I started working on this after reading https://github.com/JuliaArrays/StructArrays.jl/issues/126. @piever could you please close that if it's no longer the case? I'm not sure...

Thanks @piever , In the example above, an array is allocated, and then (I assume) garbage collected after it's used to construct the more efficient StructArray. There's a fair amount...

Great, thank you! For a while it seemed StructArrays might not have good support for this, so I started working on https://github.com/cscherrer/NestedTuples.jl. My use case requires deep nesting, so that...

Good point. Some extra compile time is ok if you can make it up in performance of the compiled code, but there's certainly a tradeoff to keep in mind. I...