Numerical Analysis

Solution of Exercises : Chapter 21

Roots of Non-linear Equations in One Variable
1. Use the bisection method to find solutions accurate to within 10−3 for the following problems.
(a) x − 2−x = 0 for 0 ≤ x ≤ 1.
(b) ex − x2 + 3x − 2 = 0 for 0 ≤ x ≤ 1.
(c) x + 1 − 2 sin(πx) = 0. for 0 ≤ x ≤ 0.5 and 0.5 ≤ x ≤ 1.
Sol.
(a) Let f (x) = x − 2−x = 0, which is continuous. We have f (0)f (1) < 0. Thus we take a = 0 and b = 1.
Approximations to the roots (middle points) are given in Table:

n a c = (a + b)/2 b sign(f (a).f (c)) |cnew − cold |

1 0 0.5 1 +
2 0.5 0.75 1 − 0.25
3 0.5 0.625 0.75 − 0.125
4 0.625 0.6875 0.75 + 0.0625
5 0.625 0.65625 0.6875 − 0.03125
6 0.625 0.640625 0.65625 + 0.015625
7 0.640625 0.648438 0.65625 − 0.007813
8 0.640625 0.644531 0.648438 − 0.003907
9 0.640625 0.642578 0.644531 − 0.001953
10 0.640625 0.641602 0.642578 − 0.000976 < 0.001

We accept 0.641602 as root with given accuracy.

(b) Let f (x) = ex − x2 + 3x − 2 = 0, which is continuous. Root lies in [0, 1] as f (0)f (1) < 0. Thus we take
a = 0 and b = 1. Approximations to the roots (middle points) are given in the following Table.
Let initial choices are a = 0, b = 1, such that f (0)f (1) < 0.
a+b
Then c = .
2
Approximations to roots (middle points) are given as follows:

n a c = (a + b)/2 b sign(f (a).f (c)) |cnew − cold |

1 0 0.5 1 −
2 0 0.25 0.5 + 0.25
3 0.25 0.375 0.5 − 0.125
4 0.25 0.3125 0.375 − 0.0625
5 0.25 0.28125 0.3125 − 0.03125
6 0.25 0.265625 0.28125 − 0.015625
7 0.25 0.257812 0.265625 − 0.007813
8 0.25 0.253906 0.257812 + 0.003906
9 0.253906 0.255859 0.257812 + 0.001953
10 0.255859 0.256836 0.257812 + 0.000976 < 0.001

We accept 0.256836 as root with preassigned accuracy.

2. Using the bisection method, determine the point of intersection of the curves given by y = 3x and y = ex in
the interval [0, 1] with an accuracy 0.1.
Sol. We consider f (x) = 3x − ex = 0, which is continuous. Root lies in [0, 1] as f (0)f (1) < 0. Thus we take
a = 0, b = 1. We have:
c1 = 0.75
c2 = 0.625
c3 = 0.5625
1 Lecture Notes of Dr. Paramjeet Singh

CONTINUED
 –2–
As|c3 − c2 | = 0.0625 < 0.1, so x coordinate of the point with given accuracy is x = 0.625. Thus y = 3x =
1.875. So point of intersection of the curves is (0.625, 1.875).
√
3. Find an approximation to 3
25 correct to within 10−2 using the bisection algorithm.
Sol. Let f (x) = x3 − 25 = 0 which is continuous. Also f (2)f (3) < 0 which implies that root lies in interval
(2, 3).
a+b
Thus we take a = 2, b = 3, c = and bracket the root at each stage. Below are the iterations to get an
2
−3
accuracy of 10 .

n a c = (a + b)/2 b sign(f (a).f (c)) |cnew − cold |

1 2 2.5 3 +
2 2.5 2.75 3 + 0.25
3 2.75 2.875 3 + 0.125
4 2.875 2.9375 3 − 0.0625
5 2.875 2.90625 2.9375 + 0.03125
6 2.90625 2.921875 2.9375 + 0.015625
7 2.921875 2.929688 2.9375 − 0.007813 < 0.01

We accept 2.929688 as a root with given accuracy.

4. Find a bound for the number of iterations needed to achieve an approximation by bisection method with
accuracy 10−2 to the solution of x3 − x − 1 = 0 lying in the interval [1, 2]. Find an approximation to the
root with this degree of accuracy.
Sol. Let f (x) = x3 − x − 1 = 0 and f (1)f (2) < 0. Also f is a continuous function.
a+b
Thus we take a = 1, b = 2, and then we calculate the middle points c = by bracketing the root.
2
−2
log(2 − 1) − log(10 )
Number of iterations required are N ≥ ≈ 6.6438.
log 2
Thus seven iterations are required to achieve an accuracy of 10−2 , which are given below in table. So root

n a c = (a + b)/2 b sign(f (a).f (c)) |cnew − cold |

1 1 1.5 2 −
2 1 1.25 1.5 + 0.25
3 1.25 1.375 1.5 − 0.125
4 1.25 1.3125 1.375 + 0.0625
5 1.3125 1.34375 1.375 − 0.03125
6 1.3125 1.328125 1.34375 − 0.015625
7 1.3125 1.320313 1.328125 + 0.007812 < 0.01

with given accuracy is 1.320313.

5. Sketch the graphs of y = x and y = 2 sin x. Use the bisection method to find an approximation to within
10−2 to the first positive value of x with x = 2 sin x.
Sol. The graphs of y = x and y = 2 sin x has shown in the next Figure. They intersect near 2.
More precisely f (1)f (2) < 0, thus first positive root lies in (1, 2).
Iterations with desired accuracy are given below. So root with given accuracy is 1.8984375.

6. The function defined by f (x) = sin(πx) has zeros at every integer. Show that when −1 < a < 0 and
2 < b < 3, the bisection method converges to

(a) 0, if a + b < 2
(b) 2, if a + b > 2
(c) 1, if a + b = 2.

CONTINUED
 –3–

n a c = (a + b)/2 b sign(f (a).f (c)) |cnew − cold |

1 1 1.5 2 +
2 1.5 1.75 2 + 0.25
3 1.75 1.875 2 + 0.125
4 1.875 1.9375 2 − 0.0625
5 1.875 1.90625 1.9375 − 0.03125
6 1.875 1.890625 1.90625 + 0.015625
7 1.890625 1.8984375 1.90625 − 0.0078125 < 0.01

Sol.

(a) In this case a + b < 2.

The first step is to add the two inequalities given in the problem to get 1 < a + b < 3 and note that
b−a a+b
c=a+ = .
2 2
Since sin(πa) < 0 and

1 < a + b < 2 =⇒ 1/2 < c < 1, =⇒ c ∈ (1/2, 1) =⇒ f (c) = sin(πc) > 0.

Thus
f (a) · f (c) < 0 =⇒ b = c ∈ (1/2, 1).
The algorithm will now search within the interval [a, c] and 0 is the only root within this new interval.
(b) In this case a + b > 2, f (a) = sin(πa) < 0 and

3 > a + b > 2 =⇒ 3/2 > c > 1, =⇒ c ∈ (1, 3/2) =⇒ f (c) = sin(πc) < 0.

Thus
f (a) · f (c) > 0 =⇒ a = c ∈ (1, 3/2).
The algorithm will now search within the interval [c, b] and 2 is the only root within this new interval.
(c) In this case a + b = 2 and f (a) = sin(πa) < 0.

a + b = 2 =⇒ c = 1, f (c) = sin(πc) = 0.

The algorithm will terminate with c = 1 as the root.

CONTINUED
 – 4 –
−x 1
7. Show that g(x) = 2 has a unique fixed point on , 1 . Use fixed-point iteration to find an approximation
3
to the fixed point accurate to within 10−2 .
Sol. The function g(x) = 2−x is continuous. Further

g(1/3) = 0.793700526
g(1) = 0.5
g 0 (x) = −2−x ln 2
|g 0 (1/3)| = 0.550151282
0
|g (1)| = 0.34657
00
g (x) = 2−x ln2 2 6= 0.
   
1 0 1
Thus g(x) ∈ , 1 and |g (x)| ≤ 0.550151282 < 1. Hence g(x) has a unique fixed point in , 1 . Now with
3 3
x0 = 0.5, we have

x1 = 0.70711
x2 = 0.61255
x3 = 0.65404
x4 = 0.63550
x5 = 0.64372
|x5 − x4 | = 0.00822 < 0.01.

So root is 0.64372.

8. For each of the following equations, use the given interval or determine an interval [a, b] on which fixed-point
iteration will converge. Estimate the number of iterations necessary to obtain approximations accurate to
within 10−2 , and perform the calculations.
5
(a) x = + 2.
x2
(b) 2 + sin x − x = 0 in interval [2, 3].
(c) 3x2 − ex = 0.
Sol.
5 5
(a) Here f (x) = + 2 − x and we take g(x) = 2 + 2.
x2 x
As f (2)f (3) < 0, so root lies in (2, 3). We have
10
g 0 (x) = −
x3
|g 0 (2)| = 1.25 > 1
|g 0 (2.5)| = 0.64 < 1
0
|g (3)| = 0.37037
10
g 00 (x) = − 3 6= 0.
x
As root lies in (2, 3), we have to restrict the interval and need to consider (2.5, 3) in order to |g 0 (x) < 1|.
Also k = max |g 0 (x)| = 0.64. Further
x∈[2.5,3]

g(2.5) = 2.8
g(3) = 2.5556.

CONTINUED
 –5–
Thus g(x) ∈ [2.5, 3] for x ∈ [2.5, 3]. Thus the choice of g will work. We start with x0 = 2.5 and get
x1 = g(x0 ) = 2.8. We can estimate the number of iteration from the formula
kn
|α − xn | ≤ |x1 − x0 | < 10−2
1−k
which gives
0.64n
(0.3) < 10−2
0.36
n > 9.9103 ≈ 10.

We have

x1 = 2.8
x2 = 2.63775510204082
x3 = 2.71862291377498
x4 = 2.67650663176068
x5 = 2.69796453650544
x6 = 2.68690634940763
x7 = 2.69257202494715
|x7 − x6 | < 0.01.

We get the root in 7 iterations with given accuracy although theoretical estimate shows that we need
to perform minimum 10 iterations. By starting near root, we can get a more closer estimate.
(b) We consider f (x) = 2 + sin x − x. Root lies in [2, 3] as f (2)f (3) < 0. Let us take initial guess x0 = 2.5.
We take g(x) = 2 + sin x and get

g 0 (x) = cos x
|g 0 (2)| = 0.4161
|g 0 (3)| = 0.9900
00
g (x) = − sin x 6= 0.

So |g 0 (x) < 1| for x ∈ [2, 3]. Also k = max |g 0 (x)| = 0.99. Further
x∈[2,3]

g(2) = 2.9093
g(3) = 2.1411.

Thus g(x) ∈ [2, 3] for x ∈ [2, 3]. Thus the choice of g will work. We start with x0 = 2.5 and get
x1 = g(x0 ) = 2.5985. We can estimate the number of iteration from the formula
kn
|α − xn | ≤ |x1 − x0 | < 10−2
1−k
which gives
0.99n
|2.5985 − 2.5| < 10−2
1 − 0.99
n > 685.81 ≈ 686.

We have

x1 = 2.5985, x2 = 2.5168, x3 = 2.5849, x4 = 2.5284

x5 = 2.5755, x6 = 2.5363, x7 = 2.5690, x8 = 2.5418
x9 = 2.5644, x10 = 2.5456, x11 = 2.5613, x12 = 2.5483
x13 = 2.5591, x14 = 2.5501.

In 14 iterations we get the solution 2.5501 correct to within 10−2 .

CONTINUED
 –6–
(c) The equation 3x2 − ex = 0 have two positive
q roots, one in [0, 1] and other in [3, 4]. There are numerous
1 x
possibilities to write g(x) and for g(x) = 3 e , on [0, 1] with x0 = 1, we have

1
g 0 (x) = √ ex/2
2 3
|g 0 (0)| = 0.28868
|g 0 (1)| = 0.47594
1
g 00 (x) = √ ex/2 6= 0.
4 3
Thus |g 0 (x) < 1|. Also k = max |g 0 (x)| = 0.47594. Further
x∈[0,1]

g(0) = 0.57735
g(1) = 0.95189.

Thus g(x) ∈ [0, 1] for x ∈ [0, 1]. Thus the choice of g will work. We start with x0 = 0.5 and get
x1 = g(x0 ) = 0.74133. We can estimate the number of iteration from the formula

kn
|α − xn | ≤ |x1 − x0 | < 10−2
1−k
which gives
0.47594n
|0.74133 − 0.5| < 10−2
1 − 0.47594
n > 5.1581 ≈ 6.

We have

x1 = 0.74133
x2 = 0.83641
x3 = 0.87713
x4 = 0.89517
x5 = 0.90328
|x5 − x4 | = 0.00811 < 0.01.

Root is 0.90328.
For another root in [3, 4], we take g(x) = ln 3x2 with x0 = 3.5,
2
g 0 (x) =
x
|g 0 (3)| = 0.66667
|g 0 (4)| = 0.5
2
g 00 (x) = − 2 6= 0.
x
Thus |g 0 (x) < 1|. Also k = max |g 0 (x)| = 0.66667. Further
x∈[0,1]

g(3) = 3.2958
g(4) = 3.8712.

Thus g(x) ∈ [3, 4] for x ∈ [3, 4]. We start with x0 = 3.5 and get x1 = g(x0 ) = 3.6041. We can estimate
the number of iteration from the formula
kn
|α − xn | ≤ |x1 − x0 | < 10−2
1−k

CONTINUED
 –7–
which gives
0.66667n
|3.6041 − 3.5| < 10−2
1 − 0.66667
n > 8.4876 ≈ 9.

We have

x1 = 3.6041
x2 = 3.6628
x3 = 3.6951
x4 = 3.7126
x5 = 3.7221
|x5 − x4 | < 0.01.

Even we get the root in five iterations and root is 3.7221.

9. Show that g(x) = π + 0.5 sin(x/2) has a unique fixed point on [0, 2π]. Use fixed-point iteration to find an
approximation to the fixed point that is accurate to within 10−2 . Also estimate the number of iterations
required to achieve 10−2 accuracy, and compare this theoretical estimate to the number actually needed.
Sol. Since g 0 (x) = 14 cos(x/2), g is continuous and g 0 exists on [0, 2π].
Further, g 0 (x) = 0 only when x = π, so that g(0) = g(2π) = π and g(π) = π + 1/2 and thus g(x) ∈ [0, 2π].
|g 0 (x)| ≤ 1/4, for 0 ≤ x ≤ 2π.
Thus a unique fixed point exists in [0, 2π]. With k = 1/4 and x0 = π, we have x1 = π + 1/2. Now

kn
|α − xn | ≤ |x1 − x0 | < 10−2 .
1−k
For the bound to be less than 0.01, we need n ≥ 4.
Iterations are given by

x1 = 3.6416
x2 = 3.6260
x3 = 3.6270
|x3 − x2 | < 0.01.

Thus x3 = 3.6270 is accurate to within 0.01.

10. Use the fixed-point iteration method to find smallest and second smallest positive roots of the equation
tan x = 4x, correct to 4 decimal places.
Sol. Let f (x) = tan x − 4x = 0. In this case we write

x = tan−1 (4x) + nπ, n = 0, ±1, ±2, · · ·

For first positive root, we take n = 0 and thus g(x) = tan−1 (4x). Clearly g(x) is continuous and root lies
before π/2. So we take x0 = 1.5 and we have

x1 = 1.4056476
x2 = 1.3947829
x3 = 1.3934402
x4 = 1.3932729
x5 = 1.3932520.

So smallest positive root correct to four decimals is 1.3932.

For another root which lies before 3π/2, we start with x0 = 4.5 and we take g(x) = tan−1 (4x) + π.

CONTINUED
 –8–
Iterations are given by

x1 = 4.6568904
x2 = 4.6587566
x3 = 4.6587780

So second smallest positive root correct to four decimals is 4.6588.

11. Find all the zeros of f (x) = x2 + 10 cos x by using the fixed-point iteration method for an appropriate
iteration function g. Find the zeros accurate to within 10−2 .
Sol. A closer look to f (x) = x2 + 10 cos x = 0 or x2 = −10 cos x, shows that this equation has four roots.
Two roots are negative which lies in [−4, −3] and [−2, −1].
By symmetry of both x2 and −10 cos x, two positive roots lies in [1, 2] and [3, 4].
There are many possibilities to write g(x). In particular, we take

f (x) x2 + 10 cos x
g(x) = x − 0
=x− .
f (x) 2x − 10 sin x
Iterations scheme is given by
x2n + 10 cos xn
xn+1 = g(xn ) = xn − .
2xn − 10 sin xn
For root lies in [−4, −3], we take x0 = −3. We have

x1 = −3.1961
x2 = −3.1630
x3 = −3.1620
|x3 − x2 | < 0.01.

Thus root accurate to within 10−2 for the interval [−4, −3] is −3.1620.
Similarly for root lies in [−2, −1], we take initial guess x0 = −1.5. Then iterations are given by

x1 = −1.9240
x2 = −1.9679
x3 = −1.9689
|x3 − x2 | = 0.001 < 0.01.

Thus root accurate to within 10−2 for the interval [−2, −1] is −1.9689.
Just reversing the sign, we get other roots. So positive root in [1, 2] is 1.9689 and second positive root lies
in [3, 4] is 3.1620.

12. Let A be a given positive constant and g(x) = 2x − Ax2 .

(a) Show that if fixed-point iteration converges to a nonzero limit, then the limit is α = 1/A, so the inverse
of a number can be found using only multiplications and subtractions.
(b) Find an interval about 1/A for which fixed-point iteration converges, provided x0 is in that interval.
Sol.
(a) Fixed-point iterations
xn+1 = g(xn ), n = 1, 2, · · · .
If fixed-point converges to the limit α, then

α = lim xn+1 = lim (2xn − Ax2n ) = 2α − Aα2 .

n→∞ n→∞

Solving for α gives α = 0, 1/A. Thus if fixed-point iteration converges to a nonzero limit, then the limit
is α = 1/A, so the inverse of a number can be found using only multiplications and subtractions from
g(x) = 2x − Ax2 ..

CONTINUED
 –9–  
1 3 5
(b) Any subinterval [a, b] of having root α = will work. For example, we take
A , containing 1/A
4A 4A
0
suffices. Now g(x) is continuous and g (x) exists and we have
1
g 0 (x) = 2 − 2Ax = 0 =⇒ x=
  A
1 1
g =
A A
 
3 15
g =
4A 16A
 
5 15
g =
4A 16A
 
3 1
g0 =
4A 2
 
5 1
g0 =− .
4A 2
   
3 5 3 5 1
Thus if we take x ∈ , , then g(x) ∈ , and |g 0 (x)| ≤ < 1. Thus g satisfy all the
4A 4A 4A 4A 2
condition for fixed point convergence.

13. Consider the root-finding problem f (x) = 0 with root α, with f 0 (x) 6= 0. Convert it to the fixed-point
problem
x = x + cf (x) = g(x)
with c a nonzero constant. How should c be chosen to ensure rapid convergence of

xn+1 = xn + cf (xn )

to α (provided that x0 is chosen sufficiently close to α)? Apply your way of choosing c to the root-finding
problem x3 − 5 = 0.
Sol. In order to ensure convergence of the iterations

|g 0 (x)| < 1,

for x in some interval [a, b] containing root. Now

|g 0 (x)| = |1 + cf 0 (x)| < 1.

Let us start with some initial guess x0 , then

−2
< c < 0.
f 0 (x0 )

Any c satisfying the above condition will work and c = −1/f 0 (x0 ) will provide rapid convergence (which is
actually Newton’s method). Now we apply the way to given problem.
We have f (x) = x3 − 5 = 0 and f (1)f (2) < 0. So we can start with any point, say x0 = 1.5. Then

x3 − 5
g(x) = x − .
3x2
Iterations are given by

x1 = 1.740741
x2 = 1.710516
x3 = 1.709976.

CONTINUED
 – 10 –
14. Show that if A is any positive number, then the sequence defined by
1 A
xn = xn−1 + , for n ≥ 1,
2 2xn−1
√
converges to A whenever x0 > 0. What happens if x0 < 0?
Sol. Do it yourself.

15. Use secant method to find root accurate to within 10−3 for −x3 − cos x = with initial guesses −1 and 0.
Sol. The secant iterations are defined as
xn − xn−1
xn+1 = xn − f (xn ) n = 1, 2 · · ·
f (xn ) − f (xn−1 )

Start with x0 = −1 and x1 = 0, we obtain

0+1
x2 = 0− f (0) = −0.68507
f (0) − f (−1)
x3 = −1.2521
x4 = −0.80720
x5 = −0.84778
x6 = −0.86653
x7 = −0.86546
x8 = −0.86547
|x8 − x7 | = 0.00001 < 0.001.

16. Use secant method to find root of x3 − 0.165x2 + 3.993 × 10−4 = 0 with initial guesses 0.02 and 0.05. Use
the stopping criterion that the relative error is less than 0.5%.
Sol. The secant iterations are given by
xn − xn−1
xn+1 = xn − f (xn ), n = 1, 2 · · ·
f (xn ) − f (xn−1 )

Start with x0 = 0.02 and x1 = 0.05, denote the relative error by εn , we get

x1 − x0
x2 = 0.06461, ε1 = × 100% = 22.62%
x1
x2 − x1
x3 = 0.06241 ε2 = × 100% = 3.525%
x2
x3 − x2
x4 = 0.06238 ε3 = × 100% = 0.0595% < 0.5%.
x3

17. Use Newton’s method to approximate the positive root of 2 cos x = x4 correct to six decimal places.
Sol. Let f (x) = 2 cos x − x4 = 0. Clearly f is continuous differentiable function and f (1)f (2) < 0.
We have f 0 (x) = −2 sin x − 4x3 .
Starting with x0 = 1, we get

f (x0 ) 2 cos(1) − 14
x1 = x0 − = 1 − = 1.014183
f 0 (x0 ) −2 sin(1) − 4(1)3
f (x1 )
x2 = x1 − 0 = 1.013957
f (x1 )
x3 = 1.013957.

Thus root correct to six decimal places is 1.013957.

CONTINUED
 – 11 –
18. Use Newton’s method to approximate, to within 10−4 , the value of x that produces the point on the graph
of y = x2 that is closest to (1, 0).
Sol. Let (x, y = x2 ) be a point on the curve y = x2 .
Let s(x) be the square of the distance of this point from point (1, 0). Then
2
s(x) = (x − 1)2 + (x2 ) = (x − 1)2 + x4 .
Minimizing the square of the distance is equivalent to minimizing the distance.
For finding the minimum, firstly equate first derivate of s(x) to zero, which gives
s0 (x) = 2(x − 1) + 4x3 = 2[(x − 1) + 2x3 ] = 0.
Write the above expression as root finding problem.
f (x) = (x − 1) + 2x3 = 0.
f 0 (x) = 1 + 6x2 .
Newton iteration’s are given by
f (xn ) (xn − 1) + 2x3n 1 + 4x3n
xn+1 = xn − = x n − = .
f 0 (xn ) 1 + 6x2n 1 + 6x2n
Starting with x0 = 1.0, we have
x1 = 0.6
x2 = 0.589873417721519
x3 = 0.589754528506440
x4 = 0.589754512301459.
Therefore root with desired accuracy 10−4 is 0.58975.
Thus the point of minimum distance from (1, 0) is (0.58975, 0.34781).

19. (a) Apply Newton’s method to the function

 √
√ x, x≥0
f (x) =
− −x, x<0
with the root α = 0. What is the behavior of the iterates? Do they converge, and if so, at what rate?

(b) Do the same but with √

3

2 x≥0
f (x) = √x ,
3 2
− x , x<0

Sol. Clearly f (x) = 0 gives root x = 0 in both cases.

(a) For x > 0, the Newton’s iterations are given by
f (xn )
xn+1 = xn −
f 0 (xn )
√
xn
= xn −
√1
2 xn
= xn − 2xn
= −xn .
Similarly for x < 0, we have
√
− −xn
xn+1 = xn −
√−1
2 −xn
= −xn .

CONTINUED
 – 12 –
Thus in both cases
xn+1 = −xn .
Hence sequence is oscillatory and provide no convergence.
(b) In this case for x > 0, the Newton’s iterations are
2/3
xn
xn+1 = xn − 2
1/3
3xn
3
= xn − xn
2
1
= − xn .
2
Similarly for x < 0, we have
2/3
−xn
xn+1 = xn −
− 21/3
3xn
1
= − xn .
2
Thus in both cases
1
xn+1 = − xn .
2
In this case convergence occurs.
Since α = 0 is the root, so from iterations scheme, we can write
1
|xn+1 − 0| = |xn − 0|
2
1
|xn+1 − α| = |xn − α|
2
1
en+1 = en ,
2
where error en = |xn − α|. Thus by definition of order of convergence, order of convergence is one.
Note that the Newton’s method provides second order convergence (minimum) if f 0 (α) 6= 0. In this example,
the function is not differentiable at x = 0. So method may fail (as in part (a)) or method provide slow
convergence (as in part (b)).

20. Apply the Newton’s method with x0 = 0.8 to the equation f (x) = x3 − x2 − x + 1 = 0, and verify that
the convergence is only of first-order. Further show that root α = 1 has multiplicity 2 and then apply the
modified Newton’s method with m = 2 and verify that the convergence is of second-order.
Sol. Successive iterations in Newton’s method are given by

x3n − x2n − xn + 1
xn+1 = xn − , n = 0, 1, 2, · · ·
3x2n − 2xn − 1
Starting with x0 = 0.8, we obtain

x1 = 0.905882, x2 = 0.954132, x3 = 0.977338, x4 = 0.988734.

Since the exact root is α = 1, we have the error in approximations

e0 = |α − x0 | = 0.2 = 0.2 × 100

e1 = |α − x1 | = 0.094118 = 0.94 × 10−1
e2 = |α − x2 | = 0.045868 = 0.46 × 10−1
e3 = |α − x3 | = 0.022662 = 0.22 × 10−1
e4 = |α − x4 | = 0.011266 = 0.11 × 10−1 .

CONTINUED
 – 13 –
Now we compute the ratios to find the order of convergence.

ln(e2 /e1 ) ln(0.46/0.94)

p = = = 0.946532462,
ln(e1 /e0 ) ln(0.94/0.2)
ln(e3 /e2 ) ln(0.22/0.46)
p = = = 1.032107253,
ln(e2 /e1 ) ln(0.46/0.94)
ln(e4 /e3 ) ln(0.11/0.22)
p = = = 0.939734509,
ln(e3 /e2 ) ln(0.22/0.46)

which shows the linear (almost) convergence .

Iterations in modified Newton’s method are given by

x3n − x2n − xn + 1
xn+1 = xn − 2 , n = 0, 1, 2, · · ·
3x2n − 2xn − 1

Starting with x0 = 0.8, we obtain

x1 = 1.011765, x2 = 1.0000034, x3 = 1.000000.

Now we have the error in approximations

e0 = |α − x0 | = 0.2 = 0.2 × 100

e1 = |α − x1 | = 0.011765 = 0.12 × 10−1
e2 = |α − x2 | = 0.000034 = 0.34 × 10−4
ln(e2 /e1 ) ln(0.34 × 10−4 /0.12 × 10−1
p= = = 2.085120871,
ln(e1 /e0 ) ln(0.12 × 10−1 /0.2)

which verifies the second-order convergence.

21. Use Newton’s method and the modified Newton’s method to find a solution of
√ √
cos(x + 2) + x(x/2 + 2) = 0, for − 2 ≤ x ≤ −1

accurate to within 10−3 .

Sol. Newton’s iterations are given by
√ √
cos(xn + 2) + xn (xn /2 + 2)
xn+1 = xn − √ √ , n = 0, 1, 2, · · ·
− sin(xn + 2) + xn + 2

Starting with x0 = −1.5, we get the root in 12 iterations with accuracy 10−3 .

x1 = −1.47855075977922 x2 = −1.46246535074333
x3 = −1.45040193555133 x4 = −1.44135464479384
x5 = −1.43456929092347 x6 = −1.42948032375788
x7 = −1.42566361878740 x8 = −1.42280109901900
x9 = −1.42065421398883 x10 = −1.41904405169404
x11 = −1.41783644052568 x12 = −1.41693074432811.

If we apply modified Newton’s method, then iterations are given by

f (xn )f 0 (xn )
xn+1 = xn − , n = 0, 1, 2, · · ·
f 0 (x 2 00
n ) − f (xn )f (xn )

In this case, we get the answer only in two iterations (with same accuracy)

x1 = −1.41423460337851, x2 = −1.41424162427621.

CONTINUED
 – 14 –
f (xn )
22. Given the iterative scheme xn+1 = xn − 0 , n ≥ 0 with f (α) = f 0 (α) = 0 and f 00 (α) 6= 0. Find the
f (xn )
order of convergence for this scheme.
Sol. We write the Newton’s iterations as fixed point iterations with
f (x)
g(x) = x −
f 0 (x)
[f 0 (x)]2 − f (x)f 00 (x) f (x)f 00 (x)
g 0 (x) = 1 − = .
[f 0 (x)]2 [f 0 (x)]2
(0)
Since f (α) = f 0 (α) = 0, so g 0 (0) has form. Thus by using the l’Hop̂ital rule, we have
(0)
f 0 (x)f 00 (x) + f (x)f (3) (x)
lim g 0 (0) = lim
x→0 x→0 2f 0 (x)f 00 (x)
[f 00 (x)]2 + 2f 0 (x)f (3) (x) + f (x)f (4) (x)
lim g 0 (0) = lim
x→0 x→0 2[f 0 (x)]2 + 2f 0 (x)f (3) (x)
1
g 0 (α) =
6= 0.
2
Thus the order of convergence for the given scheme is one.

23. A particle starts at rest on a smooth inclined plane whose angle θ is changing at a constant rate
dθ
= ω < 0.
dt
At the end of t seconds, the position of the object is given by
e − e−ωt
 ωt 
g
x(t) = − 2 − sin ωt .
2ω 2
Suppose the particle has moved 1.7 ft in 1 s. Find, to within 10−5 , the rate ω at which θ changes. Assume
that g = 32.17 ft/s2 .

Sol. Substituting the appropriate values, we find ω by finding the root of

32.17 eω − e−ω
 
f (ω) = − − sin ω − 1.7 = 0.
2ω 2 2
Here we substitute g = 32.17 ft/s2 and t = 1.
Clearly f is a continuous function and f (−1)f (0) < 0. Solving for ω, by bisection method (we can chose any
method), we get root ω = −0.317055511474609.

24. An object falling vertically through the air is subjected to viscous resistance as well as to the force of gravity.
Assume that an object with mass m is dropped from a height s0 and that the height of the object after t
seconds is
mg m2 g
s(t) = s0 − t + 2 (1 − e−kt/m ),
k k

CONTINUED
 – 15 –
where g = 32.17 ft/s2 and k represents the coefficient of air resistance in lb-s/ft. Suppose s0 = 300 ft,
m = 0.25 lb, and k = 0.1 lb-s/ft. Find, to within 0.01 s, the time it takes this quarter-pounder to hit the
ground.
Sol. When the object hit the ground, the height of the object s(t) = 0. This gives
mg m2 g
s0 − t + 2 (1 − e−kt/m ) = 0
k k
300 − 80.425t + 201.0625(1 − e−0.4t ) = 0
t = (501.0625 − 201.0625 e−0.4t )/80.425 = g(t).

Start with t0 = 3, first three iterations of fixed-point are (within accuracy 0.01 s)

t1 = 5.89184519259234, t2 = 5.99336160678534, t3 = 6.00278549575005.

So 6.0028 is within 0.01 s of the actual time.

25. The circle below has radius 1, and the longer circular arc joining A and B is twice as long as the chord AB.
Find the length of the chord AB, correct to four decimal places. Use Newton’s method.

Sol. Draw a perpendicular from O to AB, meeting AB at M. Let θ = ∠AOM . Standard trigonometry shows
that the length of AB is 2 sin θ. The shorter arc joining A and B has length 2θ, so the longer arc has length
2π − 2θ. The longer arc is twice the chord, and therefore

2π − 2θ = 4 sin θ.

We can use the Newton’s Method to solve this equation as it stands by writing f (θ) = 2 sin θ + θ − π.
Also if we substitute x = π − θ then equation becomes x = 2 sin x. This equation can be solved easily with
initial guess 1.5 and root is 1.895494267.
The length of the chord is 2 sin θ that is, 2 sin x, and that is equal to x.

26. It costs a firm C(q) dollars to produce q grams per day of a certain chemical, where

C(q) = 1000 + 2q + 3q 2/3 .

The firm can sell any amount of the chemical at $4 a gram. Find the break-even point of the firm, that is,
how much it should produce per day in order to have neither a profit nor a loss. Use the Newton’s method
and give the answer to the nearest gram.
Sol. If we sell q grams then the revenue is 4q. The break-even point is when revenue is equal to cost, that
is, when
4q = 1000 + 2q + 3q 2/3 .
Let f (q) = 2q − 3q 2/3 − 1000. We need to solve the equation f (q) = 0.
It is worth asking first whether there is a solution, and whether possibly there might be more than one.
We have f 0 (q) = 2 − q −1/3 . The Newton’s Method yields
2/3 1/3
2qn − 3qn − 1000 qn + 1000qn
qn+1 = qn − −1/3
= −1/3
.
2 − qn 2 − qn

CONTINUED
 – 16 –
Let us start with q0 = 600. Note that f (q) < 0 for small values of q, indeed up to 500 and beyond. Also,
f (1000) > 0. Since f is continuous, it is equal to 0 somewhere between 500 and 1000.
Further, iterations are given as

q1 = 607.6089386
q2 = 607.6067886.

Thus, to the nearest integer, the answer is 608.