A base step size value, h, will be scaled according to the function input parameter. A common example is hx = h*(1+abs(x)), however
this may vary depending on implementation. This definition only guarantees that the only scaling will be relative to the
function input parameter and not the order of the finite difference derivative.
A base step size value, eps (typically machine precision), is scaled according to the finite difference coefficient order
and function input parameter. The initial scaling according to finite different coefficient order can be thought of as producing a
base step size, h, that is equivalent to RelativeX scaling. This step size is then scaled according to the function
input parameter. Although implementation may vary, an example of second order accurate scaling may be (eps)^(1/3)*(1+abs(x)).