Given a function f, the rule for the derivative of the inverse function allows us to calculate the numerical value of the derivative of f^{-1} at a given input even when the explicit expression of f^{-1} itself is not known.

Let’s see a typical example of this situation.

Example

Consider the function

f(x)=3x+2e^x.


f is defined and differentiable at every \mathbb{R}, and is strictly increasing (because it is the sum of two strictly increasing functions).
So it has an inverse function, which we indicate with the usual symbol f^{-1}, also defined on the whole \mathbb{R}.
Suppose we want to know the value of the derivative of f^{-1} at x=2. The formula for the derivative of the inverse is this:

[f^{-1}]'(2)=\frac{1}{f'(a)},


where a is the only value that satisfies 

a=f^{-1}(2)\quad\text{or equivalently}\quad f(a)=2.


It is easy to observe that f(0)=2, so we deduce that a=0. So we will have

(*)\qquad[f^{-1}]'(2)=\frac{1}{f'(0)}.


We now need the derivative of f, calculated at 0: using the linearity property of the derivative we have

f'(x)=3\frac{d}{dx}x+2\frac{d}{dx}e^x=3+2e^x,


so for x=0 we find

f'(0)=3+2e^0=3+2=5.


Now just substitute in the formula (*) to get the result:

\qquad[f^{-1}]'(2)=\frac{1}{5}.