Numerical differentiation with bump size

Is there a numerical differentiation package that allows you to specify the size of the bump ?

I’d like to do f(x + d) - f(x) and specify the d (e.g. 0.001)

FiniteDifferences seems to focus on approximating the actual derivative.

You can set the stepsize in FiniteDiff.jl

https://github.com/JuliaDiff/FiniteDiff.jl/blob/master/src/jacobians.jl#L160-L161

1 Like

Many thanks