Lecture 23

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Lec 23: Newton’s Method

MATH 147 Section 2, Fall Term 2022

I (We will first finish the material from Lecture 22)


I Newton’s Method

Key references: Thomson et al. sec 2


Idea of Newton’s Method

This is used to find approximate roots, i.e., try to solve f (x) = 0.


The idea is to start with a guess x0 and construct a linear approximation to f based at x = x0 .
Unless the slope is zero, it will intersect the x-axis at a point - call this x1 .
Repeat this to generate a sequence {xn } and hope that it converges to a number c such that f (c) = 0.

Some important questions:


1. What kind of functions can you use Newton’s Method on?
2. How do you choose x0 ?
3. Will it always converge if there is a solution?
4. If it converges, is the limit a root?
5. How does it compare with the Bisection Method in these respects?
Theorem (Newton’s Method)
Let f : R → R be twice continuously differentiable. Suppose there exists x ∗ ∈ R satisfying f (x ∗ ) = 0 and
f 0 (x ∗ ) 6= 0. Then there exists r > 0 such that for any x0 ∈ [x ∗ − r , x ∗ + r ], the sequence {xn }∞
n=0 defined
recursively by
f (xn )
xn+1 = xn − 0
f (xn )
for n ≥ 0 converges to x ∗ . Moreover, there exists M ∈ R such that |xn+1 − x ∗ | < M|xn − x ∗ |2 for all n ≥ 1.

We are not yet in a position to prove this theorem so we will simply apply the algorithm and hope it works
provided we satisfy the hypotheses above.
Remarks:
1. Newton’s Method can also be used if f is defined on an interval rather than all of R.
2. Newton’s method establishes “quadratic convergence” to the solution, which is much faster than
“geometric convergence” offered by the Bisection Method.
3. The trade off is that the error bound in Newton’s method requires f to be twice continuously differentiable.
Bisection Method, in contrast, does not even require differentiability.
4. The theorem does not indicate how to determine r , x ∗ , or M. The proof of the theorem suggests a way to
find M but we need to apply some theorems from the future.
Newton’s Method example

The Babylonian Method for calculating square roots coincides with Newton’s Method.

Example

Consider the function f (x) = x 2 − a and apply Newton’s Method to estimate the value of 17.

Question

Why don’t we apply Newton’s Method to the function f (x) = x − a? Will it work?
Newton’s Method fails

Question

Will Newton’s Method work on: f (x) = x 3 ? f (x) = 3
x?

Depending on our starting point x0 , we may not converge to a root even when there is one that satisfies the
hypotheses of the Newton’s Method theorem.

Exercise
Consider the function f (x) = arctan(x). Show that Newton’s Method will fail for any starting point x0 with
|x0 | > π/4.

You might also like