-
This came up in #366.
And it has proved tricky to debug.
AFAICT there were two issues, one from ChainRules directly (I forgot to put the adjoint for `abs(::Complex)` back in, fixed now).
But second…
-
It is possible to do automatic differentiation through a circuit with conditional gates. I mean variationally optimize the circuit with gates that are applied based on themed-circuit measurement.
rezah updated
2 years ago
-
Hi, Brian! 👋 I am currently using [ForwardDiff.jl](https://github.com/JuliaDiff/ForwardDiff.jl) within [TORA.jl](https://github.com/JuliaRobotics/TORA.jl) for computing Jacobians of user-defined vecto…
-
ChainRules.jl is a language-wide AD definition library. https://github.com/JuliaDiff/ChainRules.jl Plugging into it will give compatibility with a lot of operations for free. You might want to use thi…
-
[This](https://github.com/TuringLang/DistributionsAD.jl/blob/2d566ae83d0ca57966e1f69e3de469da8f9f10ec/src/zygote.jl#L78) definition is very optimistic about the things that it thinks that it can handl…
-
Zygote is missing some `repeat` adjoints;
```julia
# this is OK
julia> gradient(x -> sum(repeat(x, outer=(2,2,2))), reshape(1:8, 2,2,2))
([8 8; 8 8]
[8 8; 8 8],)
# missing rule
julia> gradi…
-
`Zygote.jacobian(eigvals, A)` returns an incorrect answer if `A` happens to be symmetric: it returns an erroneous column of zeros:
```jl
julia> using LinearAlgebra, Zygote
julia> Zygote.jacobian(…
-
I have a custom type for which a corresponding method of `Base.trunc` makes sense to differentiate. But https://github.com/FluxML/Zygote.jl/blob/169563f65f0d9f9221ad528780aab48d157a05fc/src/lib/number…
-
[The docs](https://www.juliadiff.org/ChainRulesCore.jl/dev/api.html#ChainRulesCore.Composite) say that `Composite` users should disregard the `T` parameter as it is an implementation detail.
Howev…
-
Similar to the ChainRulesCore support, we could use `EnzymeCore.EnzymeRules` to define `Forward` and `Reverse` mode rules for Enzyme in an extension.
EnzymeCore requires at least Julia v1.6. Making…