A bug in ForwardDiff?

General Usage
, ,
1

Consider this:

using ForwardDiff

function f!(F, x)
    F[1] = (x[1]+3)*(x[2]^3-7)+18
    F[2] = sin(x[2]*exp(x[1])-1)
    return F
end

F = Array{Float64}(undef, 2)
x = [1, 1]

Now let’s compute the Jacobian. The documentation, says:

ForwardDiff.jacobian(f!, y::AbstractArray, x::AbstractArray, cfg::JacobianConfig = JacobianConfig(f!, y, x), check=Val{true}())

Return J(f!) evaluated at x , assuming f! is called as f!(y, x) where the result is stored in y .

since f!(F, x) works, then this should work:

∇f!(F, x) = ForwardDiff.jacobian!(f!, F, x)

however

julia> ∇f!(F, x)
ERROR: MethodError: no method matching jacobian!(::typeof(f!), ::Vector{Float64}, ::Vector{Int64})
Closest candidates are:
  jacobian!(::Union{AbstractArray, DiffResults.DiffResult}, ::Any, ::AbstractArray, ::AbstractArray) at /Users/amrods/.julia/packages/ForwardDiff/QOqCN/src/jacobian.jl:75
  jacobian!(::Union{AbstractArray, DiffResults.DiffResult}, ::Any, ::AbstractArray, ::AbstractArray, ::ForwardDiff.JacobianConfig{T, V, N, D} where {V, N, D}) where T at /Users/amrods/.julia/packages/ForwardDiff/QOqCN/src/jacobian.jl:75
  jacobian!(::Union{AbstractArray, DiffResults.DiffResult}, ::Any, ::AbstractArray, ::AbstractArray, ::ForwardDiff.JacobianConfig{T, V, N, D} where {V, N, D}, ::Val{CHK}) where {T, CHK} at /Users/amrods/.julia/packages/ForwardDiff/QOqCN/src/jacobian.jl:75
  ...
Stacktrace:
 [1] ∇f!(F::Vector{Float64}, x::Vector{Int64})
   @ Main ~/discourseQ4.jl:137
 [2] top-level scope
   @ REPL[1]:1

Is that a bug in ForwardDiff or its documentation?

3 
julia> ForwardDiff.jacobian(f!, F, x)
2×2 Matrix{Float64}:
 -6.0       12.0
 -0.399455  -0.399455

The documentaiton said ForwardDiff.jacobian(..), not ForwardDiff.jacobian!(...).

1 Like
4

Didn’t see that !!!, I guess it’s time for me to take a break…