What happens if we choose our Initial Guess Foolishly?

Here's the graph of a simple function, a fourth-order polynomial:

f(x)=(x-3)(x-1)(x+1)(x+3)

There are four roots, which are at x values of -3, -1, 1, and 3. Let's color the four roots red, purple, blue and green, respectively. We can make a map, for this function, of where each initial guess leads. For instance, the initial guess x=0.4 leads to the nearby blue root (x=1), so we'll color the point 0.4 in as blue:

However, the initial guess x=0.13 hops over to the rightmost green root (x=3), so we'll color the point 0.13 in as green:

If we continue this process of finding the root for each possible initial x-value, we will get a map that looks like this:

Map of Initial Guesses for f(x)=(x-3)(x-1)(x+1)(x+3)

Wow! It turns out that you can have pockets of initial guesses that lead to faraway roots. These "bad" guesses occur where the derivative gets small, since a shallow slope will shoot the iterated value to a distant point on the axis. This structure -- the fact that the different regions are mixed together near their boundary -- becomes even more surprising if we extend Newton's Method to the Complex Plane.

prev: Can we Break Newton's Method? next: Newton's Method Fractals