You are here

Derivatives of polynomials

11 November, 2015 - 10:15

Some ideas in this proof are due to Tom Goodwillie. Theorem: For n= 0, 1, 2, . . . ,the derivative of the function xdefined by x(t) =  t^{n}  is x˙ = nt^{n-1}.

The results for n= 0 and 1 hold by direct application of the definition of the derivative.

For n>1, it suffices to prove \dot{x} (0) = 0 and \dot{x} (1) = n, since the result for other nonzero values of tthen follows by the kind of scaling argument used on page 13 for the n= 2 case.

We use the following properties of the derivative, all of which follow immediately from its definition as the slope of the tangent line:

Shift. Shifting a function x(t) horizontally to form a new function x(t+c) gives a derivative at any newly shifted point that is the same as the derivative at the corresponding point on the unshifted graph.

Flip. Flipping the function x(t) to form a new function x(t) negates its derivative at t= 0.

Add. The derivative of the sum or difference of two functions is the sum or difference of their derivatives.

For even n, \dot{x} (0) = 0 follows from the flip property, since x(t) is the same function as x(t). For n= 3, 5, . . . , we apply the definition of the derivative in the same manner as was done in the preceding section for

n= 3.

We now need to show that \dot{x} (1) = n. Define the function uas

u(t) = x(t+ 1) x(t)

= 1 + nt+ ...   ,

where the second line follows from the binomial theorem, and. . . represents terms involving t^{2} and higher powers. Since we’ve already established the results for n= 0 and 1, differentiation gives

u˙ (t) = n+ ...   .

Now let’s evaluate this at t= 0, where, as shown earlier, the terms represented by . . . all vanish. Applying the add and shift properties, we have

\dot{x} (1) − \dot{x} (0) = n   .

But since \dot{x} (0) = 0, this completes the proof.

Although this proof was for integer exponents n1, the result is also true for any real value of n; see example 24 on p. 41.