Enzyme.jl
Enzyme.jl copied to clipboard
EnzymeAD
Julia bindings for the Enzyme automatic differentiator
MWE: ```julia using Enzyme, ForwardDiff p = rand(3) y = rand(3) tmp1 = rand(3) tmp2 = rand(3) tmp3 = rand(3*4) tmp3 = SubArray(tmp3, axes(tmp3)) tmp3 = reinterpret(ForwardDiff.Dual{ForwardDiff.Tag{Nothing, Float64}, Float64, 3},tmp3)...
Document behavior of backing arrays when using sparse arrays. Document #373
AD with respect to a struct that has a non-differentiable field fails ```julia struct Foo2{X,Y} x::X y::Y end test_f(f::Foo2) = f.x^2 julia> autodiff(test_f, Active(Foo2(3.0, 2.0))) (Foo2{Float64, Float64}(6.0, 0.0),) julia> autodiff(test_f,...
``` using Enzyme Enzyme.API.printall!(true) function bad(@nospecialize(t)) t = Tuple{t...} return Base.unwrap_unionall(t) end @show bad((Int64,)) flush(stdout) @show Enzyme.fwddiff(bad, Duplicated, Duplicated((Int64,), (Int64,))) #,(Int64,))) ```
```julia using Enzyme n = 10 x = [i/(1.0+i) for i in 1:n] dx = ones(n) rx = zeros(n); drx = zeros(n) y = zeros(1); dy = zeros(1) ry =...
TODO: - [ ] We need to emit write_barriers on write of Julia objects to the Julia managed tape - [ ] Codegen for 1.6/1,7 - [ ] Real runtime...
# Code ```julia import Pkg; Pkg.status() import Random import Enzyme const AV = AbstractVector{T} where T # ===== Set up objective function ===== normal_pdf(x::Real, mean::Real, var::Real) = exp(-(x - mean)^2...
```julia using Enzyme, SparseArrays, LinearAlgebra # setup data θ = rand(3) x = rand(11) ν = rand(10) λ = rand(24) S = sprand(Float64, 6, 7, 0.1) Cp = sprand(Float64, 1,...