David Widmann

> It should be good to merge if the tests added are satisfactory. I improved the tests. They were a bit redundant and even a bit worse than some existing...

IMO the current behaviour of the `Quantity` constructor is a bug. I just ran into an issue caused by this in a project: In this project, values and units are...

I think it's a bug because the `Quantity` constructor is "lying" about its return type. If I call `Quantity(x)` I expect that the return value is of type `Quantity`, which...

Just remembered https://github.com/TuringLang/AbstractMCMC.jl/pull/124. I still think this should be fixed in AdvancedHMC rather than AbstractMCMC - we can't add all downstream options to AbstractMCMC and it means that custom implementations...

https://github.com/TuringLang/AbstractMCMC.jl/pull/117 will allow samplers to specify behaviour of warm-up steps. AdvancedHMC would specialize `step_warmup`, possibly depending on a warm-up algorithm instance stored in the sampler.

This will fix the error I ran into, but the more general problem still exists: Since AdvancedHMC adds special keyword arguments to `sample`, `sample` and the `step` and `steps` iterators/transducers...

I don't have a strong opinion on the name `Kernel` vs `AbstractKernel` - there is no consensus in Julia (not even in Base - `AbstractArray`, `Real` and `Number` are all...

Relevant parts of the docstrings: ```julia help?> isequal search: isequal issetequal isequal(x, y) Similar to ==, except for the treatment of floating point numbers and of missing values. [...] Implementation...

> this could allow even more structured matrix types to be defined for specific kernel types and returned in the future Yes, I imagine this would be useful also in...

I have the same feeling as @sethaxen: > I considered using one of the many existing array types in the ecosystem for lazily representing an array, but defining a novel...