Michael Abbott

FWIW, I think calling these "keys" in my package was a mistake, as `Base.keys` already has a meaning. (And because these are auxiliary info, if you are using them to...

Yes I experimented a bit. In fact AxisKeys.jl is one experiment really, at making its wrapper commute with that of NamedDims.jl -- it runs the tests in both orders. But...

Thanks for the note. This might be tricky to fix, will think a bit. The other type inference issue which needs a look is this -- ideally it would realise...

This came up in https://discourse.julialang.org/t/fast-4d-argmax/58566 One possibility would be that, when the macro sees a finaliser (such as `abs2`), it always removes the reduction indices from consideration by the tiling...

The second of these seems to be doing what's intended. Since there isn't a unique solution, it won't guess how much of the range of `x`'s index is taken by...

Fixed by #481, I think: ``` julia> println(a[1].partials == d_false_zero.partials) true ``` > the entry is technically zero That's the question. Opinions vary but the most mathematical literature seems to...

Xref https://github.com/JuliaDiff/ChainRules.jl/issues/544 , about numbers, because they were slow. And https://github.com/JuliaDiff/ChainRules.jl/pull/412 for those 3,4-arg `*` cases which go to `mul!`. Unlike real numbers, the order matters for vectors & matrices....

I think the stack overflow is telling you it's not supported. The method is (I presume) meant to send `besseli(0, 3)` to `besseli(0, 3.0)`, but instead catches dual numbers for...

Note to self: https://github.com/JuliaDiff/ReverseDiff.jl/pull/180 should make this possible now

Would still be neat to build this in as a convenience function. If it took a modified DerivativeConfig then it could save the allocation of `Dual.(x,v)`, and handle `f!`, etc,...