David Widmann
David Widmann
I wonder if any of the implementations could benefit from RealDot.jl? It contains optimized implementations of `real(dot(x, y))`.
Not sure why you pinged me but this seems unrelated to this PR and not specific to complex inputs? It would be good to keep the PR minimal and just...
> DistributionsAD could just go away completely ForwardDiff is not the main blocker, it's Tracker and ReverseDiff. There are only very few definitions for dual numbers remaining: https://github.com/TuringLang/DistributionsAD.jl/blob/master/src/forwarddiff.jl
I think the definition of `Float64` in this PR is reasonable for operator overloading AD (and might be the right thing to do). However, I am worried that `Int` (and...
I just learnt some days ago that Tracker contains an even more general form of the approach in this PR (but, of course, it's a really nasty type piracy and...
I'll close this PR in favour of #207.
I'm sorry, I think my comment was a bit unclear. The main reason for why I think one should consider changing `isinteger` (and eg. `iseven` and `isodd`) is not to...
> Where should this be fixed? Here? In DiffRules? I didn't follow this issue bit can have a look. Does ForwardDiff use DiffRules for these calculations here? If yes, then...
NaN-safe mode switching via the Preferences.jl system was added in https://github.com/JuliaDiff/ForwardDiff.jl/pull/539.
Apparently @stevengj suggested they should be moved to AbstractFFTs: https://github.com/JuliaMath/AbstractFFTs.jl/issues/47#issuecomment-741914478 and https://github.com/FluxML/Zygote.jl/issues/835#issuecomment-741902634