ERROR: LoadError: TypeError: in typeassert, expected Float64, got a value of type ForwardDiff.Dual{Nothing, Float64, 2} #636
Comments
|
We should just get rid of the mutation in that function: function jacobian_fn(f, x::AbstractMatrix)
y, back = Zygote.pullback(f, x)
z = Zygote.@ignore similar(y)
Zygote.@ignore fill!(z, zero(eltype(x)))
out = map(1:size(y, 1)) do i
Zygote.@ignore z[i, :] .+= one(eltype(x))
xx = back(z)[1]
end
@show size(Array(VectorOfArray(out)))
return Array(VectorOfArray(out))
endBut |
|
I think the problem comes from sensealg = InterpolatingAdjoint(autojacvec=false)because sensealg = InterpolatingAdjoint()works fine. |
|
Oh interesting. I think it might need to |
|
Yup, I'll have a look later today. |
|
From a quick look in the lines around https://github.com/SciML/DiffEqSensitivity.jl/blob/22b52f82cd8177046ef32adb3baf402983249d61/src/derivative_wrappers.jl#L267 Indeed Wrapping a Is that the solution you meant, @ChrisRackauckas? And is: Line 169 in 73817a5 the definition of the ODE function that needs to be differentiated? (Haven't looked at the ffjord implementation so far) |
|
Yeah that's what I meant. Odd. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
When I run this code:
I get this error:
In this environment:
The text was updated successfully, but these errors were encountered: