Function numerical_gradient 

pub fn numerical_gradient<T, F>(
    f: &F,
    x: &[T],
    h: T,
    mode: DifferenceMode,
) -> CoreResult<Array1<T>>
where
    T: Float + FromPrimitive + Debug,
    F: Fn(&[T]) -> T,

Compute the numerical gradient of a scalar function at point x.

• f – function from R^n -> R, taking a slice and returning a scalar.
• x – the point at which to evaluate the gradient.
• h – step size (e.g. 1e-5).
• mode – finite difference mode.

Returns an Array1<T> of the same length as x.