Proofs and Derivations

Newton's Method
(For finding roots of equations)

Consider the graph below of the function y = f(x).

Our goal is the find the root, r, of y = f(x).  That is, the value where the graph crosses the x-axis, or where f(r) = 0. (See note below, about when this method fails.)
 

We start with a first guess, x1 (which we know is not the root, but which is a close guess) and construct the tangent to the graph of y = f(x) at the point (x1, f(x1)).  We notice that if we find where this tangent line crosses the x-axis (at x2) , that it is a better approximation to the root than our initial guess, x1. If we continue with this process, x3 is a better guess than x2 and etc.

To generalize this process we proceed as follows.

To find x2 from x1 we first note that the equation of the tangent line f '(x1) is given by
 
 

y - f(x1) = f '(x1)(x - x1),



 

since the slope of the line tangent to the graph of y = f(x) at (x1, f(x1)) is given by f '(x1).

Now, since this tangent line crosses the x-axis at the point (x2, 0), the equation above becomes
 
 

0 - f(x1) = f '(x1)(x2 - x1).



 

If we solve this for x2 it becomes,
 
 
 

x2 = x1 - f(x1)/f '(x1).



 

If we generalize this process we come up with the Newton's method formula:
 
 

xn+1 = xn - f(xn)/f '(xn).



 

NOTE:  Newton's method will not work unless the graph of y = f(x) has the same concavity on either side of the root, r.