Week 12
Topic 3. Newton’s Method (Stewart, pp. 280 – 285)
1 Introduction
Definition 1. Newton’s Method is an attempt to approximate the value of a solution of
equations in the following from:
f (x) = 0, where f is a differentiable function.
Newton’s Metod generates a sequence of approximations
x1 , x2 , . . . , xn , . . .
where x1 is an initial guess for a root, and the terms that follows are given by the recursive
formula
f (xn
xn+1 = xn − ′
f (xn )
Remark.
• If the sequence of numbers xn converge to a number r, then r is a root of the equation
f (x) = 0.
• xn is called the n-th approximation.
• Newton’s Method is not always guaranteed to work, but for most cases it will work provided
we made a suitable choice for x1 .
• Newton’s Method only provide one approximated value of one solution.
Motivating Newton’s Method
We guess an initial approximation
x1 of an x-intercept of the function f . We then determine the
tangent line at x1 , f (x1 ) :
Where this tangent line
y − f (x1 ) = f ′ (x1 )(x − x1 )
1
cut the x-axis, is our second approximation x2 . That is, with the assumption that f ′ (x1 ) ̸= 0,
f (x1 )
0 − f (x1 ) = f ′ (x1 )(x2 − x1 ) =⇒ x2 = x1 −
f ′ (x1 )
This x2 is closer than
x1 to the x-intercept We repeat this process with x2 : determine the tangent
line at x2 , f (x2 ) and let x3 be the x-intercept of that line:
The formula for x3 is like the previous formula for x2 , but just with x1 replaced by x2 . More
generally, the (n + 1)-th approximation of an x-intercept is
f (xn )
xn+1 = xn − , with f ′ (xn ) ̸= 0
f ′ (xn )
2 Examples
Example 2. Consider the equation
x3 + x − 1 = 0
Use x1 = 1 as a first approximation for a root of the equation and calculate the third approximation
x3 with Newton’s Method.
Solution: Set f (x) = x3 + x − 1. Then f ′ (x) = 3x2 + 1.
From Newton’s Method:
f (xn ) x3n + xn − 1 2x3n + 1
xn+1 = xn − = x n − =
f ′ (xn ) 3x2n + 1 3x2n + 1
By starting with x1 = 1, we get
33 27
2 · 13 + 1 3 2· 43
+1 32 +1 27 + 32 59
x2 = = =⇒ x3 = 32
= 27 = =
3 · 12 + 1 4 3· +1 16 +1 54 + 32 86
42
For interest sake, x33 + x3 − 1 = 0.00894 . . ., which seems very close to a solution.
Remark. Notice that f (0) = −1 < 0 and f (1) = 1 > 0. Thus, according to the Intermediate
Value Theorem, there exists an r ∈ (0, 1) such that f (r) = 0. That explains our initial choice of
x1 = 1. What would happen if we choose x1 = 0?
2
Example 3. Consider the equation
x3 − 5x = 0
(a) What would happen if we apply Newton’s Method with x1 = 1?
(b) What would happen if we apply Newton’s Method with x1 = 2?
Solution: Since
f (x) = x3 − 5x and f ′ (x) = 3x2 − 5
the n-th approximation, according to Newton’s Method, is
x3n − 5xn 2x3
xn+1 = xn − 2
= 2 n
3xn − 5 3xn − 5
(a) with x1 = 1, we get
2·1 2 · (−1)
x1 = 1 x2 = = −1 x3 = =1
3·1−5 3·1−5
x4 = . . . = −1 x5 = . . . = 1
We are stuck in a cycle, but neither x = 1 nor x = −1 is a solution of x3 − 5x = 0! Thus,
Newton’s Method does not work here with x1 = 1 or x1 = −1.
(b) With x1 = 2, we get (with a computer)
x1 = 2.000000 x2 = 2.285714 x3 = 2.237640
x4 = 2.236070 x5 = 2.236068 x6 = 2.236068
which approaches to a root. The x5 is at least 6 decimals correct to a root.
√
6
Example 4. Use Newton’s Method to determine 2 correct to eight decimal places.
√
6
Solution: By definition, 2 is the positive root of
x6 − 2 = 0
We thus set f (x) = x6 − 2. Further, f ′ (x) = 6x5 . The formula for xn+1 is thus
x6n − 2
xn+1 = xn −
6x5n
With a computer and x1 , we get
x1 = 1.00000000 x2 ≈ 1.16666667 x3 ≈ 1.12644368
x4 ≈ 1.12249707 x5 ≈ 1.12246205 x6 ≈ 1.12246205
Since x5 and x6 agrees on eight decimal places, we deduce that
√
6
2 ≈ 1.12246205
3
Remark. For interest sake, there are many methods to determine roots of equations. For example,
to solve
f (x) = x6 − 2 = 0
we could have choose a1 = 1 and b1 = 2. Since f (a2 ) < 0 and f (b1 ) > 0, according to the
Intermediate Value Theorem, there exists a root in [a1 , b1 ]. Let m = (a1 + b1 )/2 be the mean of a1
and b1 . If f (m) < 0, let a2 = m and b2 = b1 , else let a2 = a1 and b2 = m. We thus get an interval
[ab , b2 ] with half the length of [a1 , b1 ] in which the root lies. By repeating
√ this process until an and
bn differs with less that 10−8 , we will also get an approximation of 6 2 to eight decimal accurate.
But here we have to repeat the process almost 30 times, instead of just 6 times with Newton’s
Method! Though, an advantage of this inefficient method, is that it will always give a solution,
provided the function is continuous.
Example 5. Determine the root of the following equation to six decimal places.
cos x = x
Solution: We first rewrite the equation in standard form:
cos x − x = 0
We thus let
f (x) = cos x − x =⇒ f ′ (x) = − sin x − 1
According to Newton’s Method
cos xn − xn cos xn − xn
xn+1 = xn − = xn +
− sin xn − 1 sin xn + 1
With a computer, we get
x1 = 1.000000 x2 ≈ 0.750364 x3 ≈ 0.739113
x4 ≈ 0.739085 x5 ≈ 0.739085
Since x4 and x5 agrees to six decimal places, we deduce that the root of cos x = x, correct to six
decimal places, is
x ≈ 0.739085
