AD

[ocots]

It seems that the CI does not complete with DifferentialInterface: see here.

For the moment, in CTFlows.jl we have const ctgradient = CTBase.__ctgradient:

https://github.com/control-toolbox/CTFlows.jl/blob/557511f3b433a75bb055910a66d74435d6c6eb93/ext/CTFlowsODE.jl#L15C3-L15C43

with

function ctgradient(f::Function, x::ctNumber; backend=__get_AD_backend())
    extras = prepare_derivative(f, backend, x)
    return derivative(f, backend, x, extras)
end

function __ctgradient(f::Function, x::ctNumber)
    return ForwardDiff.derivative(x -> f(x), x)
end

see here.