Statistics and Numerical Methods Notes

MA8452 STATISTICS AND NUMERICAL METHODS LTPC
4 0 0 4
OBJECTIVES:
• This course aims at providing the necessary basic concepts of a few statistical and numerical
methods and give procedures for solving numerically different kinds of problems occurring
in engineering and technology.
om
• To acquaint the knowledge of testing of hypothesis for small and large samples which plays
an important role in real life problems.
• To introduce the basic concepts of solving algebraic and transcendental equations.
• To introduce the numerical techniques of interpolation in various intervals and numerical

techniques of differentiation and integration which plays an important role in engineering
.c
and technology disciplines.
• To acquaint the knowledge of various techniques and methods of solving ordinary differential
ul
equations.
UNIT I TESTING OF HYPOTHESIS pa 12
Sampling distributions - Estimation of parameters - Statistical hypothesis - Large sample

tests based on Normal distribution for single mean and difference of means -Tests based on
t, Chi-square and F distributions for mean, variance and proportion - Contingency table (test
for independent) - Goodness of fit.
jin
UNIT II DESIGN OF EXPERIMENTS 12
One way and two way classifications - Completely randomized design - Randomized block
design - Latin square design - 22 factorial design.
.re
UNIT III SOLUTION OF EQUATIONS AND EIGENVALUE PROBLEMS 12
Solution of algebraic and transcendental equations - Fixed point iteration method - Newton
Raphson method - Solution of linear system of equations - Gauss elimination method -
w
Pivoting - Gauss Jordan method - Iterative methods of Gauss Jacobi and Gauss Seidel -
Eigenvalues of a matrix by Power method and Jacobi’s method for symmetric matrices.
w
UNIT IV INTERPOLATION, NUMERICAL DIFFERENTIATION AND NUMERICAL

INTEGRATION 12
Lagrange’s and Newton’s divided difference interpolations - Newton’s forward and

w
backward difference interpolation - Approximation of derivates using interpolation

polynomials - Numerical single and double integrations using Trapezoidal and Simpson’s
1/3 rules.
UNIT V NUMERICAL SOLUTION OF ORDINARY DIFFERENTIAL EQUATIONS 12
Single step methods : Taylor’s series method - Euler’s method - Modified Euler’s method -
Fourth order Runge-Kutta method for solving first order equations - Multi step methods :
Milne’s and Adams - Bash forth predictor corrector methods for solving first order equations.
TOTAL : 60 PERIODS
OUTCOMES:
om
Upon successful completion of the course, students will be able to:
• Apply the concept of testing of hypothesis for small and large samples in real life problems.
• Apply the basic concepts of classifications of design of experiments in the field of agriculture.
• Appreciate the numerical techniques of interpolation in various intervals and apply the
.c
numerical techniques of differentiation and integration for engineering problems.
• Understand the knowledge of various techniques and methods for solving first and second order
ordinary differential equations.
ul
• Solve the partial and ordinary differential equations with initial and boundary conditions by
using certain techniques with engineering applications
pa
TEXT BOOKS:
1. Grewal. B.S. and Grewal. J.S., "Numerical Methods in Engineering and Science", 10 th Edition,
Khanna Publishers, New Delhi, 2015.
jin
2. Johnson, R.A., Miller, I and Freund J., "Miller and Freund’s Probability and Statistics for
Engineers", Pearson Education, Asia, 8 th Edition, 2015.
REFERENCES:
.re
1. Burden, R.L and Faires, J.D, "Numerical Analysis", 9 th Edition, Cengage Learning, 2016.
2. Devore. J.L., "Probability and Statistics for Engineering and the Sciences", Cengage Learning,
New Delhi, 8 th Edition, 2014.
w
3. Gerald. C.F. and Wheatley. P.O. "Applied Numerical Analysis" Pearson Education, Asia, New
Delhi, 2006.
4. Spiegel. M.R., Schiller. J. and Srinivasan. R.A., "Schaum’s Outlines on Probability and Statistics
w
", Tata McGraw Hill Edition, 2004.
5. Walpole. R.E., Myers. R.H., Myers. S.L. and Ye. K., "Probability and Statistics for Engineers and
w
Scientists", 8 th Edition, Pearson Education, Asia, 2007.
ii
Contents
om
1 Testing of Hypothesis 1
2 Design of Experiments 3
3 Solution of Equations and Eigenvalue Problems 5
.c
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.2 Solution of algebraic and transcendental equations . . . . . . . . . . . . . . . . . . . . . . 5
3.3 Fixed point iteration method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
ul
3.3.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.3.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.3.3 Anna University Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
pa
3.4 Newton Raphson method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.4.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.4.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.5 Gauss elimination method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
jin
3.5.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.5.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
.re
3.6 Pivoting - Gauss Jordan method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

3.6.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.6.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.7 Iterative methods of Gauss-Jacobi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
w
3.7.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.7.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.8 Iterative methods of Gauss-Seidel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
w
3.8.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.8.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
w

3.9 Matrix Inversion by Gauss Jordan method . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.9.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.9.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
iii
Contents
3.10 Eigen values of a matrix by Power method . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

3.10.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.10.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.10.4 Assignment Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4 Interpolation, Numerical Differentiation and Numerical Integration 51
om
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.2 Lagrange’s interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.2.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.2.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4.3 Newton’s divided difference interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
.c
4.3.1 Properties of divided differences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.3.2 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.3.3 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
ul
4.4 Cubic Splines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.4.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
pa
4.4.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.5 Newton’s forward and backward difference formulae . . . . . . . . . . . . . . . . . . . . . . 69
4.6 Newton’s backward interpolation difference formula: . . . . . . . . . . . . . . . . . . . . . 69
jin
4.6.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.6.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.6.4 Assignment problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
.re
4.7 Numerical Differentiation and Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

4.8 Approximation of derivatives using interpolation polynomials . . . . . . . . . . . . . . . . 77
4.8.1 Newton’s forward difference formula to compute derivative . . . . . . . . . . . . . . 77
4.8.2 Newton’s backward difference formula to compute derivatives . . . . . . . . . . . . 78
w
4.8.3 Maxima and Minima of a tabulated function . . . . . . . . . . . . . . . . . . . . . . 79

4.8.4 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.8.5 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
w

4.9 Numerical integration using Trapezoidal . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
w
4.9.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
4.9.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.10 Numerical integration using Simpson’s 1/3 rule . . . . . . . . . . . . . . . . . . . . . . . . . 85
4.10.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
iv
Contents
4.10.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
4.11 Numerical integration using Simpson’s 3/8 rule . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.11.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.11.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
4.11.3 Single integrals by Trapezoidal, Simpson 1/3 & 3/8 . . . . . . . . . . . . . . . . . . 88
om
4.12 Romberg’s method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
4.12.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4.12.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4.13 Two point Gaussian quadrature formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
.c
4.13.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
4.13.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
4.14 Three point Gaussian quadrature formula . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
ul
4.14.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.14.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
pa
4.15 Double integrals by Trapezoidal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
4.15.1 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
4.16 Double integrals by Simpson’s 1/3 rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
jin
4.16.1 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.16.2 Double integrals by Trapezoidal, Simpson rule . . . . . . . . . . . . . . . . . . . . . 102
.re
5 Numerical Solution of Ordinary Differential Equations 105

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.2 Taylor’s series method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
5.2.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
w
5.2.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

5.3 Taylor’s series method for simultaneous first order differential equations . . . . . . . . . . 114
w
5.3.1 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

5.4 Taylor’s series for II order differential equations . . . . . . . . . . . . . . . . . . . . . . . . 115
w
5.4.1 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

5.5 Euler’s method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5.5.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5.5.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
v
Contents
5.6 Modified Euler’s method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

5.6.1 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.7 Fourth order Runge-Kutta method for solving first order equations . . . . . . . . . . . . . 122
5.7.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
5.7.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
om
5.8 Fourth order Runge - Kutta method for solving II order differential equation . . . . . . . 124
5.9 Fourth order R. K. Method for simultaneous first order differential equations . . . . . . . 126
5.9.1 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
5.10 Milne’s forth predictor corrector methods for solving first order equations . . . . . . . . . 128
.c
5.10.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
5.10.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
ul
5.11 Adams-Bash forth predictor corrector methods for solving first order equations . . . . . . 130
5.11.1 Part A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
5.11.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
pa
5.11.4 Anna University Questions (Taylor’s, RK, Adam, Milne) . . . . . . . . . . . . . . . 131
jin
.re
w
w
w
vi
STATISTICS
UNIT- I TESTING OF HYPOTHESIS
Population:
A population consists of collection of individual units, which may be person’s or experimental
om
outcomes, whose characteristics are to be studied.
Sample:
A sample is proportion of the population that is studied to learn about the characteristics of the
population.
.c
Random sample:
A random sample is one in which each item of a population has an equal chance of being
ul
selected.
Sampling:
pa
The process of drawing a sample from a population is called sampling.
Sample size:
The number of items selected in a sample is called the sample size and it is denoted by ‘n’. If
n 30, the sample is called large sample and if n 30, it is called small sample
jin
Sampling distribution:
Consider all possible samples of size’ n’ drawn from a given population at random. We calculate
mean values of these samples.
.re
If we group these different means according to their frequencies, the frequency distribution so
formed is called sampling distribution.
The statistic is itself a random variate. Its probability distribution is often called sampling
distribution.
w
All possible samples of given size are taken from the population and for each sample, the statistic
is calculated. The values of the statistic form its sampling distribution.
w
Standard error:
The standard deviation of the sampling distribution is called the standard error.
w
Notation:
Pop. mean = ; Pop. S.D = ; P - Pop. proportion
Download Useful Materials from Rejinpaul.com

sample mean = x; sample S.D = s; P' = sample Proportion
Note
Statistic S.E (Standard Error)
om
x
n
Difference of sample 1 1
p1' p2 ' pq
proportions n1 n2
2 2
Difference of sample 1 2
x1 x2
.c
means n1 n2
pq
p' (Sample proportion)
n
ul
Null Hypothesis ( H 0 )
pa
The hypothesis tested for possible rejection under the assumption that it is true is usually called
null hypothesis. The null hypothesis is a hypothesis which reflects no change or no difference. It is
usually denoted by H 0
Alternative Hypothesis ( H1 )
jin
The Alternative hypothesis is the statement which reflects the situation anticipated to be correct if
the null hypothesis is wrong. It is usually denoted by H1 .
.re
For example:
If H0 : 1 = 2 (There is no diff' bet' the means) then the formulated alternative hypothesis is
w
H1 : 1 2
ie., either H1 : 1 2 (or) 1 2

w
Level of significance
It is the probability level below which the null hypothesis is rejected. Generally, 5% and 1% level
w
of significance are used.
Critical Region (or) Region of Rejection

The critical region of a test of statistical hypothesis is that region of the normal curve which
corresponds to the rejection of null hypothesis.
The shaded portion in the following figure is the critical region which corresponds to 5% LOS
om
.c
ul
Critical values (or) significant values pa
The sample values of the statistic beyond which the null hypothesis will be rejected are called
critical values or significant values
Level of significance
Types of test 1% 5% 10%
jin
Two tailed test 2.58 1.96 1.645
One tailed test 2.33 1.645 1.28

.re
Two tailed test and one-tailed tests:
When two tails of the sampling distribution of the normal curve are used, the relevant test is
called two tailed test.
w
The alternative hypothesis H1 : 1 2 is taken in two tailed test for H 0 : 1 = 2
When only one tail of the sampling distribution of the normal curve is used, the test is described
w
as one tail test H1 : 1 2 (or) 1 2
H0 1 2
two tailed test
w
H1 1 2
Type I and type II Error
Type I Error : Rejection of null hypothesis when it is correct

Type II Error : Acceptance of null hypothesis when it is wrong
Procedure for testing Hypothesis:
1. Formulate H 0 and H1
2. Choose the level of significance
om
3. Compute the test statistic Z, using the data available in the problem
4. Pick out the critical value at % level say Z
5. Draw conclusion: If |Z| < Z , accept H0 at % level. Otherwise reject H0 at % level
Test of Hypothesis (Large Sample Tests)
.c
Large sample tests (Test based in Normal distribution.)
Type - I: (Test of significance of single mean)
ul
Let {x1, x 2 ,. . . .xn } be a sample of size (n 30) taken from a population with mean and
S.D . Let x be the sample mean. Assume that the population is Normal.
pa
To test whether the difference between Population mean
not and this sample comes from the normal population whose mean is
and sample mean x is significant or
or not.
H0 : = a specified value
jin
H1 : a specified value
we choose = 0.05(5%) (or) 0.01(1%) as the Level of significance

.re
the test statistic is
x x
Z= N(0.1) for large n.
S .E ( x)
n
w
Note:
s
w
1. If is not known, for large n, S.E x = where 's' is the sample S.D
n
Problems:
w
1. A sample of 900 members is found to have a mean 3.5cm. Can it reasonably regarded as a simple
sample from a large population whose mean is 3.38 and a standard deviation 2.4cm?
Solution:
We formulate the null hypothesis that the sample is drawn from population whose mean is
3.38cm.
i.e., H 0 : = 3.38
om
H1 : 3.38
Hence it is a two-tailed test
Level of significance = 0.05
.c
x
Test statistic Z =
ul
Given x 3.5, =3.38, n = 900, =2.4
3.5 3.32
Z=
2.4
900
1.5
pa
Critical value:
jin
At 5% level, the tabulated value of Z is 1.96
Conclusion:
Since |Z| = 1.5 < 1.96, H 0 is accepted at 5% level of significance

.re
i.e., the sample comes from a population with mean 3.38cm
2. A manufacturer claims that his synthetic fishing line has a mean breaking strength of 8kg and S.D
0.5kg. Can we accept his claim if a random sample of 50 lines yield a mean breaking of 7.8kg. Use 1%
w
level of significance.
Solution:
w
We formulate H 0 : =8
w
H1 : 8
L.O.S = 0.01
x
Test statistic Z =
Given x 7.8, = 8, n = 50, = 0.5
om
7.8 8
Z= 2.828
0.5
50
.c
|Z| = 2.828
Critical value:
ul
At 1% level of significance the table of Z = 2.58
Conclusion:
pa
Since |Z| > 2.58, H 0 is rejected at 1% level
i.e., the manufacturer’s claim is not accepted
3. A random sample of 200 Employee’s at a large corporation showed their average age to be 42.8 years,
jin
with a S.D of 6.8 years. Test the hypothesis H 0 : = 40 versus H1 : > 40 at = 0.01 level of
significance.
Solution:
.re
We set up H 0 : = 40
H1 : 40
w
It is one tailed test.
L.O.S = 0.01
w
x
Test statistic Z =
w
Given x 42.8, = 40, n = 200, = 6.89

42.8 40
Z= 5.747
6.89
200
Critical value:
om
For one-tail test, the table value of Z at 1% level = 2.33
Conclusion:
Since |Z| =5.747> 2.33, H 0 is rejected at 1% level.
.c
i.e., The hypothesis = 40 is accepted at this level.
Type - II:
ul
Test of significance of difference of two means
Consider two samples of sizes n1 and n2 taken from two different populations with population
means 1 and 1 and S.D's 1 and
pa
2
Let x1 and x2 be the sample means and S1 and S2 be the S.D's of the samples
The formulated null and alternative hypothesis is,

jin
H0 : 1 = 2
H1 : 1 2
.re
The test statistic 'Z' is defined by
x1 x2
Z=
S .E ( x1 x2 )
w
x1 x2
ie., Z = N(0,1)
2 2
1 2
w
n1 n2
We use the los is 0.05 (or ) 0.01

w
If |Z| < Z , H0 is accepted at %Los
otherwise, H0 is rejected at %Los

Note:
In many situations, we do not know the S.D’s of the populations (or) population from which the
samples are drawn.
In such cases, we can subs the S.D’s are of samples S1 and S2 in place of 1 and 2
om
x1 x2
The test statistic Z =
s12 s2 2
n1 n2
Problems
.c
The mean of two sample large samples of 1000 and 200 members are 67.5 inches and 68 inches
respectively. Can the samples be regard as drawn from the population of standard deviation of 2.5 inches?
ul
Test at 5% Los
Solution pa
we set up H0 : 1 2
ie., the samples are drawn from the sample population

jin
H1: 1 2
x1 x2
.re
The test statistic Z =

1 1
n1 n2
Given x1 67.5; n1 = 1000

w
x2 68; n2 = 2000; =2.5

w
67.5 68
Z= 5.164
1 1
2.5
w
1000 2000
|Z| = 5.164
We choose the Los =0.05

Critical value:
The table values of Z at 5% Los is Z = 1.96
Conclusion:
om
Since |Z| > 1.96, H 0 is rejected at 5% Los.
The sample cannot be regards as drawn from the same population.
2. Samples of students were drawn from two universities and from the weights is kilogram. The
means and S.D’s are calculated. Test the significance of the difference between the means of two
.c
samples
Mean S.D Sample Size
University A 55 10 400
ul
University B 57 15 100
Solution:
we set up H0 : 1
pa
2
ie., there is no significant difference between the sample means

jin
H1: 1 2 ; = 0.05
x1 x2
The test statistic Z = `
.re
s12 s2 2
n1 n2
Given x1 55; s1 = 10; n1 400

w
x2 57; s2 = 15; n2 =100

w
55 57
Z= 1.265
102 152
400 100
w
|Z| = 1.265
Critical value:
The table values of Z at 5% Los is Z = 1.96
Conclusion:
Since |Z| < 1.96, H 0 is accepted at 5% Los. We conclude that the difference between the means is
not significant.
om
3. The average hourly wage of a sample of 150 workers is plant A was Rs. 2.56 with a S.D of
Rs.1.08. The average wage of a sample of 200 workers in plant B was Rs. 2.87 with a S.D of Rs.
1.28. Can an applicant safety assume that the hourly wages paid by plant B are greater than those
.c
paid by plant A?
Solution:
ul
Let x1 and x2 denote the hourly wages paid to workers in plant A and plant B respectively.
We set up H0 : 1 2 (Plant Bnot greater than Plant A)

pa
H1 : 1 2 (one tailed test)
= 0.05
jin
x1 x2
Z =
s12 s2 2
n1 n2
.re
Given x1 2.56; s1 = 1.08; n1 150
x2 2.87; s2 =1.28; n2 = 200

w
2.56 2.87
Z= 2.453
(1.08)2 (1.28) 2
w
150 200
|Z| = 2.453
w
Critical value:
The table values of Z at 5% in case of one-tailed test is Z = 1.645

Conclusion:
Since |Z| > 1.643, H 0 is rejected at 5% Los.
The hourly wage paid by Plant B are greater than those paid by Plant A
om
4. A sample of size 30 from a normal population yielded 80 and variance 150. A sample of
size 40 from a second normal population yielded the sample mean 71 and variance 200.
Test H 0 : 1 2 2. Versus H1 : 1 2 2
Solution:
.c
H0 : 1 2 2.
ie., the diff 'bet the means of two population is 2
ul
Versus H1 : 1 2 2 (one tailed)
Test Statistic Z =
x1 x2
pa 1 2
s12 s2 2
n1 n2
jin
(80 71) 2
Z= 2.215
150 200
30 40
.re
Critical value:
For one tail test, at 5% Los the table value of z = 1.645
Conclusion:
w
Since |Z| > 1.645, H 0 is rejected.

w
The formulated null hypothesis H0 : 1 2 2 is wrong
5. A buyer of electric bulbs purchases 400 bulbs; 200 bulbs of each brand. Upon testing these
w
bulbs be found that brand A has an average of 1225 hrs with a S.D of 42 hrs. where as brand B
had a mean life of 1265 hrs with a S.D of 60 hrs. Can the buyer be certain that brand B is
Superior than brand A in quality?
Solution:
H0 : 1 2 ;
ie., the two brands of bulbs do not differ in quality
om
ie., they have the same mean life
H1: 1 < 2 (one tailed)
L.o.s : = 0.05
.c
x1 x2 x1 x2
Test Statistic Z =
ul
S .E x1 x2 s12 s2 2
n1 n2
Here, x1 1225;
pa
s1 = 42; n1 200
x2 1265; s2 =60; n2 = 200

jin
1225 1265 40
Z= 7.72
(42) 2 (60) 2 5.18
200 200
.re
| Z | = 7.72
Critical value:
The critical value of Z at 5% Los Z = 1.645.

w
Conclusion:
Since |z| < 1.645 H 0 is rejected.

w
The brand B is superior to brand A in equality.

w
Type - III:
Test of significance of single proportion:
If ‘x’ is the number of times possessing a certain attribute in a sample of n items,

x
The sample proportion p '
n
p' :sample porportion;
p: population proportion.
om
The hypothesis H0 : p = p'
ie., p has a specified value
Alternative hyp: H1 : p p'
.c
p' p
Test statistic Z =
pq
ul
n
Since the sample is large Z N(0,1)

pa
Problems
1. A coin is tossed 400 times and it turns up head 216 times. Discuss whether the coin may be
regarded as unbiased one.
jin
Solution
we set up H0 : coin is unbiased

.re
1 1
ie., p = q = 1- p =
2 2
H1 : coin is biased
w
= 0.05
p' p
Test statistic Z =
w
pq
n
216
w
Here p' ;n 400

400
0.54 0.5
Z= 1.6
1
600
Table value of Z = 1.96
om
Conclusion:
Since z < 1.96, H0 is accepted at 5% Los
Hence the coin may be regarded as unbiased

2. In a city of sample of 500 people, 280 are tea drinkers and the rest are coffee drinkers.
.c
Can we assume that both coffee and tea are equally popular in this city at 5% Los.
Solution:
ul
1
we set up H0 : p =
2
pa
ie., the coffee and tea are equally popular
1
H1 : p
2
jin
p' p
Test statistic Z =
pq
n
.re
280
Here p' 0.56; n 500; p=0.5
500
q = 1-p = 0.5
w
0.56 0.5
Z= 2.68
0.5 0.5
500
w
Conclusion:
Since z > 1.96, H0 is rejected at 5% level

w
Both type of drinkers are not popular at 5% Los.

3. A manfacturing company claims that atleast 95% of its products supplied confirm to the
specifications out of a sample of 200 prodcuts, 18 are defective. Test the claim at 5% Los.
Solution
we set up H0 : The proportion of the products confirming to specification is 95%
ie., p = 0.95
H1 : p < 0.95 (one tailed test)
p' p
om
Z=
pq
n
200 18
Here p' 0.91; n 200
200
p = 0.95 q = 1- p = 0.05
.c
0.91 0.95
Z= 2.595 | Z | 2.595
0.95 0.05
200
ul
Critical value : at 5%Los Z 1.645
Conclusion:
z =2.595 > 1.645, H0 is rejected at 5% Los(Level of significance)
pa
4. A manfacturer claims that only 4% of his products supplied by him are defective.
Sample of 600 products contained 36 defectives. Test the claim of the manufactrer.
Solution:
we set up H0 : p = 0.04
H1 : p > 0.04 (one tailed test)
jin
p' p
Test Statistic Z =
pq
n
Here p =0.04 q = 1-p = 0.96
.re
36
p' 0.06; n 500
600
0.06 0.04
Z= 2.5
0.04 0.96
w
600
Critical value :
The table value of Z =1.645 at 5%L.o.s
w
Conclusion:
Z = 2.5 > 1.645, H0 is rejected
Manufacturer's claim is not acceptable
w
Type - IV: Test of significance for Difference of proportion of success in two samples:
To test the significance of the difference between the sample proportions p1' and p2' .
We formulate the null hypothesis H0 : p1 = p2

ie., the population proportions are equal
The alternative hypothesis is H1: p1 p2
1 1
The standard error of p1' p2' pq
om
n1 n2
x x n1 p1' n2 p2 '
Where p= 1 2
n1 n2 n1 n2
p1' p2'
.c
The test statistic is Z = N (0,1)
1 1
pq
n1 n2
ul
Problems:
1. If a sample of 300 units of a manufactured product 65 units were found to be defective and in
pa
another sample of 200 units, there were 35 defectives. Is there significant difference in the
proportion of defectives in the samples at 5% Los.
Solution:
jin
H0: p1 = p2 (ie., There is no significant difference in the proportion defectives in the samples)
The alternative hypothesis H1: p1 p2

.re
Los: = 0.05
p1' p2'
The test statistic is Z =
1 1
pq
w
n1 n2
65
p1' 0.22; p 2 ' =0.175
w
300
100 1 4
p= q
w
500 5 5
1 1 4 1 1
pq 0.0365
n1 n2 25 300 200
0.22 0.175
Z= 1.233
0.0365
Critical value :
The table value of Z at 5% Level =1.96
om
Conclusion:
Z < 1.96, H0 is accepted at 5% Los.
The difference in the porportion of defectives in the samples is not significant
.c
2. A machine puts out 16 imperfect articles in a sample of 500. After the machine is over-hauled
in puts out 3 imperfect articles in a batch of 100. Has the machine improved?
ul
Solution:
H0: Machine has not been improved
ie., H0 : p1 = p2
pa
The alternative hypothesis H1: p1 p2 (one-tailed)
jin
Los: = 0.05
p1' p2'
.re
1 1
pq
n1 n2
16
Here p1' 0.032; p 2' =0.03
500
w
n1 = 500; n 2 =100
w
19 581
p= and q
600 600
0.032 0.03
w
Z= 0.104
19 581 1 1
600 600 500 100
| Z | = 0.104
Critical value :
The table value of Z for one tailed test Z = 1.645 at 5 Los
Conclusion:
om
Z < 1.645, H 0 is accepted at 5% Los.
The Machine has not improved due to overhaulding.
3. Before an increase in excise duty on tea, 800 perons out of a sample of 1000 persons were
found to be tea drinkers. After an increse is excise duty. 800 people were tea drinkers in a
.c
sample of 1200 people. Test whether there is a significant decrease in the consumption of
tea after the increase in excise duty at 5% Los
ul
Solution:
pa
H0: the proportion of tea drinkers before and after the increase in excise duty are equal
ie., p1 = p 2
H1: p1 p2
jin
Los: = 0.05
p1' p2'
.re
1 1
pq
n1 n2
Here x1 =800; x2 =800; n1 = 1000;

w
800 800
n 2 =1200; p1' 0.8; p2' = 0.67
1000 1200
w
x1 x2 1600 8 3
p= = = q
n1 n2 2200 11 11
w
1 1 24
pq 0.001 0.0008 0.0189
n1 n2 121
0.13
Z= 6.88 |Z| 6.88
0.0189
Critical value: At 5% Los 1.645
Conclusion:
om
Z > 1.645, H0 is rejected.
There is a significance decrease in the consumption of tea due to increase in excise duty.
Type - V: (Test of significance for the difference of S.D’s of two large samples)
.c
Let S1 and S2 be the S.D's of two indepedent samples of sizes n1 and n2 respectively
ul
The null hypothesis H0 : 1 2 ;
The Alternative Hypothesis H1:

pa
ie., the sample S.D's do not differ significantly.
1 2
S1 S2
the test statistic is Z = N (0,1) for large 'n'
jin
S.E (S1 S2 )
ie., If 1 and 2 are known,
S1 S2
.re
Z= N (0,1)
2 2
1 2
2n1 2n2
(or)If 1 and 2 are not known,

w
S1 S2
Z=
S12 S2 2
w
2n1 2n2
If |Z| > Z , H0 is rejected at % level, otherwise H0 is accepted

w
Problems:
1. The sample of sizes 1000 and 800 gave the following results
Mean S.D
Sample I 17.5 2.5
Sample II 18 2.7
Assuming that the samples are indepedent, test whether the two samples may be
om
regarded as drawn from the universe with same S.D's at 1% Level.
Solution:
We set up H0 : 1 2 ;
.c
ie., two samples maybe regarded as drawn from the universe with same S.D’s
H1 : 1 2
ul
S1 S2
Test statistic Z =
S12 S2 2
pa
2n1 2n2
Here n1 = 1000; n2 =800; S1 2.5; S2 = 2.7
2.5 2.7 0.2

jin
Z = =
(2.5)2 (2.7) 2 0.3125 0.455625
2000 1600
| Z | =2.282
.re
Critical value :
At 1% Los, the tabulated value is 2.58
Conclusion:
Since Z < 2.58, H 0 is accepted at 1% Los.
w
The two samples may be regarded as drawn from the universe with the same S.D's
w
2. In a survey of incomes of two classes of workers, two random samples gave the following
results. Examine whether the differences between (i) the means and (ii) the S.D's are
w
significant.
Sample Size Mean annualincome (Rs) S.D in Rs
I 100 582 24
II 100 546 28
Examine also whether the samples have been drawn from a population with same S.D
Solution:
om
(i) We set up H0 : 1 2 ;
ie., the difference is not significant
H1 : 1 2
.c
Here it is two tailed test
x1 x 2 582 546
Test statistic Z =
ul
2 2
S1 S2 (24)2 (28)2
n1 n2 100 100
Z =
pa
360
(24) 2 (28) 2
=9.76
| Z | = 9.76
jin
Critical value :
At 5% Los, the table value of Z is 1.96
Conclusion:
.re
Since Z > 1.96, H0 is rejected at 5% Los.
There is a significant difference in the means in the two samples.

w
(ii) H 0 : 1 2
H1 : 1 2
Here it is two tailed test

w
Los: = 0.05
S1 S2 24 28
w
Test statistic Z =
S12 S2 2 (24)2 (28)2
2n1 2n2 200 200
40
Z = = 1.53
288 392
| Z | = 1.53
Critical value :
om
Conclusion:
Since Z < 1.96, H0 is accepted at 5% Los.
.c
The difference between the sample S.D's is not significant.
Hence we conclude that the two samples have been drawn from population with the same S.D's
ul
3. Two machines A and B produced 200 and 250 items on the average per day with a S.D of
pa
20 and 25 items reply on the basis of records of 50 day's production. Can you regard both
machine's equally efficient at 1% Los.
Solution:
jin
(i) H 0 : 1 2 ; ie., the two machines aer equally efficient
H1 : 1 2
Los: = 0.05
.re
S1 S2
Test statistic Z =
S12 S2 2
n1 n2
w
n1 =200 50; S2 = 25
n2 =250 50; S1 20
w
(20 25) 50 5 50
Z = = 23.57
w
400 625 1 1.25

400 500
| Z | = 23.57
Critical value :
Conclusion:
Since Z > 2.58, H 0 is rejected at 1% Los.
om
We conclude that the both machines are not equally efficient at 1% Los
Small sample Tests (t - Test):

Definition:
Consider a random sample {x1 ,x 2 ,.....x n }of size 'n' drawn from a Normal population with
.c
2
mean and variance .
n
xi
ul
i 1
Sample mean x
n
2
The unbiased estimate of the pop.variance pa is denoted as s 2 .
n
( xi x) 2
s2 i 1
n 1
|x |
The student's t-statistic is defined as t = n , Where n = sample size
s
jin
The degree of freedon of this statistic
V=n 1
Type I:
.re
To test the significance of a single mean (For small samples)
x x
Test Statistic t =
S .D S
w
n 1 n
s = sample S.D and

w
1 ns 2
S2 ( x x)2 (or) S=
n 1 n 1
w
If the computed value of t is greater than the critical value t ,Ho is rejected
(or) if |t| < t , the null hypothesis Ho is accepted at level.

1. A machinist is making engine parts with axle diameter of 0.700 inch. A random sample of
10 parts shows a mean diamter of 0.742 inch with a S.D of 0.40. Test whether the work is
meeting the specification at 5% Los
om
Solution:
Given that n =10; x 0.742 inches
ns 2 10 (0.40)2
= 0.700 inches S= = 0.4216
n 1 9
.c
s = 0.40 inches S = 0.42
ul
Null hypothesis H0 : the product is confirming to specification ie., there is no significant
difference between x and
H0 : 0.700inches
pa
H1 : 0.700inches
jin
|x |
Test Statistic t = n =0.316
s
degrees of freedom = n-1 = 9

.re
Table value of t at 5% level = 2.26
the product is meeting the specification.
2. Ten individuals are chosen at random from a population and their heights are found to be
w
in inches 63, 63, 66, 67, 68, 69, 70, 70, 71, 71. In the light of this data, discuss the suggestion
that the mean height in the universe is 66 inches.

w
Solution:
w
x : 63 63 66 67 68 69 70 70 71 71
(x- x)2 : 23.04 23.04 3.24 0.64 0.04 1.44 4.84 4.84 10.24 10.24
x 678 and (x - x)2 81.6

x 678
x 67.8
n 10
(x - x) 2 81.6
S= 3.011
9 9
om
Let H 0 : = 66 the mean and height if the universe is 66 inches
and H1 : 66
Los = 0.05
|x | 67.8 66
Test Statistic t = n 10 1.89
.c
s 3.011
Table value of t for 9 d.f at 5% Los is t 0 2.2
ul
Since |t| < t 0 , H0 is accepted at 5% level.
The mean height of universe of 66 is accepted. pa

Type II: (Test of significance of difference of mean)
x1 x2
Test Statistic t =
1 1
jin
S
n1 n2
n1s12 n2 s2 2
Where S2 (or)
n1 n2 2
.re
2
( x1 x1 ) 2 ( x2 x 2 )2
S
n1 n2 2
The number of degrees of freedom = V = n1 n2 2

w
The calculated value of t is less than the table value of t for d.f = n1 n2 2, H0 is accepted
w
Otherwise H0 is rejected at the selected Los

w
1. Two independent samples from normal pop's with equal variances gave the following
results
Sample Size Mean S.D

1 16 23.4 2.5
2 12 24.9 2.8
Test for the equations of means.
om
Solution:
(i) We set up H0 : 1 2 ; ie., there is no significant difference between their means
H1 : 1 2
.c
Los: = 0.05
x1 x2 n1s12 n2 s2 2
Test Statistic t = Where S2
n1 n2 2
ul
1 1
S
n1 n2
Given x1 23.4; n1 16; s1 2.5
x2 24.9; n 2 12; s 2
pa 2.8
n1s12 n2 s2 2 16(2.5) 2 12(2.8) 2

S2 =
jin
n1 n2 2 16 12 2
100 94.08
= =7.465
26
.re
S = 2.732
23.4 24.9
t= = -1.438
1 1
2.732
w
16 12
| t | = 1.438
w
Number of degrees of freedom = n1 n2 2 26
Critical value :
w
The table value of t for 26 d.f at 5% Los is
t 0.05 2.056
Conclusion:
Since the calculated value of t is less than table value of t,
H0 is accepted at 5% Los.
om
There is no significant difference between their means
2. Two independent samples of 8 and 7 items respectively had the following values
Sample I : 9 13 11 11 15 9 12 14
.c
Sample II : 10 12 10 14 9 8 10
Is the difference between the means of the samples significant?
ul
Solution:
We set up H0 : 1 2 pa
H1 : 1 2
Hence it is a two tailed test
Los: = 0.05
jin
x1 x2
Test Statistic t =
1 1
S
n1 n2
.re
2
( x1 x1 ) 2 ( x2 x 2 )2
Where S
n1 n2 2
w
w
w
Computation of t:
d1 ( x1 x1 ) d2 ( x2 x2 )
x1 d1 x1 11.75 d1 2
( x1 x1 ) 2 x2 d2 x2 10.43 d 22 ( x2 x2 ) 2
9 -2.75 7.5625
om
10 -0.43
0.1849
13 1.25 1.5625
12 1.57
2.4649
11 -0.75 0.5625
10 -0.43
0.1849
11 -0.75 0.5625
14 3.57
12.7449
.c
15 3.25 10.5625
9 -1.43
2.0449
9 -2.75 7.5625
8 -2.43
5.9049
ul
12 0.25 0.0625
10 -0.43
0.1849
14 2.25 5.0625
d1 3.5
pa
d12 33.5 d2 0.01 d22 23.7143
6
jin
x1 11 11.75
8
3
x2 10 10.43
7
.re
2
2 2
d1 36
( x1 x1 ) d 1 38 33.5
n1 8
2
2 2
d2 9
( x2 x2 ) d2 25 33.5
w
n2 7
33.5 23.71
S2 S = 2.097
w
8 7 2
x1 x2
t=
w
1 1
S
n1 n2
11.75 10.43
=
1 1
2.097
8 7
t = 1.218
om
d.f = 8 + 7 - 2 = 13
Critical value:
The table value of t for 13 d.f at 5% level is 2.16
.c
Conclusion:
Since | t | < 2.16, H0 is accepted
ul
There is no significant difference between the means of the two samples.
Type III:
pa
Testing of significance of the difference in means paired data.
When the two samples are of the same sizes and the data are paired
d
the test statistic is t =
S
jin
n
Where d mean of differences
(d d ) 2
.re
and S =
n 1
Degrees of freedom = n-1
1. Elevan school boys were given a test in painting. They were given a month's further tution
w
and a second test of equal difficulty was held at the end of the month. Do the marks give
evidance that the students have beneifit by extra coaching?
w
w
Boys: 1 2 3 4 5 6 7 8 9 10 11
om
First Test (marks) 25 23 19 22 21 19 22 21 25 18 20
Second test (marks) 26 22 22 19 23 21 24 24 25 22 18
Solution:
H0 : the student have not been benefited by extra coaching.
.c
ie., The mean of the difference between the marks of the two tests is zero
ul
ie., H0 : d 0
H1: d 0 pa
Los: = 0.05 (or) 5%
d
S
jin
n
S. No 1 2 3 4 5 6 7 8 9 10 11
d=x-y -1 1 -3 3 -2 -2 -2 -3 0 -4 2
.re
d-d 0 2 -2 4 -1 -1 -1 -2 1 -3 3
2 0 4 4 16 1 1 1 4 1 9 9
d-d
d
w
11
d 11; d 1
n 11
2
w
d-d 50
(d d ) 2 50
w
S= 5 2.236
n 1 10
d 1
t=
S 2.236
n 11
1
|t|= 1.48
om
0.625
No. of d.f = 11-1 = 10
Critical value:
At 5% Los, the table value of t at 10 degree freedom is 1.812
.c
Conclusion:
| t | < 1.812, H0 is accepted at 5% Los.
ul
The students have not been benefited by extra-coaching.
48 36 37
pa
2. The scores of 10 candidates prior and after training are given below,
Prior : 84 54 69 83 96 90 65
After : 90 58 56 49 62 81 84 86 84 75
Is the training effective?
jin
Solution:
We set up H0 : the training is not effective

.re
ie., H0 : d 0
H1: d 0
d
w

S
n
w
S. No 1 2 3 4 5 6 7 8 9 10
d=x-y -6 -10 -20 -12 -8 -12 -1 10 6 -10
d-d 0.3 -3.7 -13.7 -5.7 -1.7 -5.7 5.3 16.3 12.3 -3.7
w
2 0.09 13.69 187.69 32.49 2.89 32.49 28.09 265.69 151.29 13.69
d-d
d 63
d 63; d 6.3
n 10
2
om
d-d 728.1
(d d ) 2 728.1
S= 80.9 8.994
n 1 9
S 8.994
.c
6.3 6.3
t 2.21
8.994 2.844
ul
10
| t | = 2.21
Degrees of freedom V= n-1 = 10-1 = 9
Critical value:
pa
At 5% Los, the table value of t at 9 degree freedom is 2.262
jin
Conclusion:
| t | < 2.262, H0 is accepted at 5% Los.

.re
There is no effective in the training.
Variance Ratio Test (or) F-test for equality of variances
This test is used to test the significance of two or more sample estimates of population variance
w
The F-statistic is defined as a ratio of unbiased estimate of population variance

2
S2 x1 x1
w
F = 12 ; Where S12
S2 n1 1
2
w
x2 x2
S2 2
n2 1
S12
The distribution of F = 2 S12 S2 2 is given by the following p.d.f
S2
If S12 and S22 are the variances of two sample of sizes n1 and n 2 respectively, the estimate
om
of the population variances based on these samples are respectively
n1s12 n 2s 2 2
S12 ; S2 2
n1 1 n2 1
d.f V1 = n1 1 &V2 = n2 1
.c
While defining the statistic F, the large oftwo variances is always placed in the numerator and
smaller in the denominator
ul
Test of significance for equality of population variances
Consider two independent R, samples x1 ,x 2 ,.......x n1 & y1 ,y 2 ,.......y n 2 from normal populations
The hypothesis to be tested is
pa
"The population variances are same".
jin
2 2
we set up: H0 : 1 = 2
2 2
& H1: 1 2
.re
S12
The test statistic F= 2 S12 S2 2
S2
n n
1 2 1 2
S12 xi x and S12 yj y
n1 1 i 1 n1 1 j 1
w
F distribution with d.f V1 = n1 1 &V2 = n2 1

w
Problems:
1. It is known that the mean diameters o rivets produced by two firms A and B are practically the
same but the standard deviations may differ.
w
For 22 rivets produced by A, the S.D is 2.9 m, while for 16 rivets manufactured by B, the S.D is
3.8 m. Test whether the products of A have the same variability as those of B
Solution:
2 2
H0 : 1 = 2
ie., variability for the two types of products are same.
Los: = 0.05 (or) 5%
S12
The test statistic F= 2 S12 S2 2
S2
om
Given, n1 = 22; n 2 = 16
S1 = 2.9; S2 = 3.8
n1s12 22(2.9) 2
S12 8.81
n1 1 22 1
n 2s 2 2 16(3.8)2
S2 2 15.40
.c
n2 1 16 1
S2 2
F = S2 2 S12
S12
ul
15.40
=
8.81
F = 1.748 pa
Number of degrees of freedom are V1 16 1 15
V2 22 1 21
Critical value:
At 5% Los, the table value of F at d.f (15,21) is F = 2.18
jin
Conclusion:
F< 2.18, H0 is accepted at 5% Los.
Variability for two types of products may be same.
2. Two random samples of sizes 8 and 11, drawn from two normal populations are characterized
.re
as follows
Size Sum of observations Sum of square of observations
Sample I 8 9.6 61.52

w
Sample II 11 16.5 73.26
You are to decide if the two populations can be taken to have the same variance.
w
Solution:
w
Let x and y be the observations of two samples
2 2
we set up: H0 : 1 = 2
2 2
& H1: 1 2
For sample I
2
2
x2 x
s
om
1
n n
2
61.52 9.6
=
8 8
= 7.69 (1.2)2
.c
7.69 1.44
s12 6.25
ul
For sample II
2
2
y2 pa y
s2
n n
2
73.26 16.5
=
11 11
jin
= 6.66 (1.5)2 6.66 2.25

.re
s22 4.41
n1s12 8(6.25)
S12 7.143
n1 1 7
w
n 2s 2 2 11(4.41)
S2 2 4.851
w
n2 1 10
w
S2 2
F = S2 2 S12
S12
7.143
= 1.472
4.851
F = 1.472
Number of degrees of freedom are V1 n1 1 8 1 7
om
V2 n 2 1 11 1 10
Critical value:
The table value of F for (7,10) d.f at 5% Los is 3.14
Conclusion:
Since | F |<3.14, H 0 is accepted at 5% level
.c
Variances of two populations may be same.
Variability for two types of products may be same.
ul
Chi-Square Test
Definition pa
If Oi (i = 1, 2, . . . . . n) are set of observed (experimental) frequencies and Ei (i 1,2,....n)
are the corresponding set of expected frequncies, then the statistic
2
is defined as
jin
2
n
2 Oi Ei
i 1 Ei
.re
The degree of freedom is v = n 1
For fitting Binomial distribution v = n 1
For fitting Poisson distribution v = n 2

w
For fitting Normal distribution v=n 3

w
Chi-square Test of Goodness of fit

2
If the calculated value of is less than the table value at a specified Los.
w
The fit is considered to be good

otherwise the fit is considered to be poor.
2
Conditions for applying Test
For the validity of chi-square test of "goodness of fit" between theory and experiment following
Conditions must be satisfied.
(i) The sample of observations should be independent

(ii) Constraints on the cell frequncies. If any, should be linear.
om
(iii) N, the total frequency should be reasonably large, say greater than 50.
(iv) N0 theoretical cell frequency should be less than 5, If any theoretical cell
frequency less than 5, the for application 2 test
It is pooled with the preceeding or succeeding frequency so that the pooled
frequency is greater than 5 and finally adjust for the d.f lost in pooling.
.c
Problems
1. The following table gives the number of aircraft accident that occured during the various days
ul
of the week. Test whether the acidents are uniformly distributed over the week.
Days : Mon
No.of accidents : 14
Tue
18
Wed
pa
12
Thu
11
Fri
15
Sat
14
Total
84
Solution:
jin
We set up H0 : The accidents are uniformly distributed over the week
Los α = 0.05
2
n
Oi Ei
.re
2
Test Statistic
i 1 Ei
Under the nul hypothesis,
84
w
The expected frequency of the on each day = 14

6
Oi : 14 18 12 11 15 14
w
Ei : 14 14 14 14 14 14
w
2 2 2 2 2 2
2 14 14 18 14 12 14 11 14 15 14 14 14
14 14 14 14 14 14
= 1.143 + 0.286 + 0.643 + 0.071

= 2.143
Number of degrees of freedom V = n 1=7 1=6
Critical value:
om
2
The tablulated value of at 5% for 6 d.f is 12.59
Conclusion:
2
Since < 12.59, we accept the null hypothesis
.c
We conclude that the accidents are uniformly distributed over the week.
2. The theory predicts the population of beans in the four groups A, B , C and D should be
ul
9:3:3:1. In an experiment among 1600 beans, the number in the four groups were 882, 313,
287 and 118. Does the experimental result support the theory?
Solution:
We set up the null hypothesis

pa
H0 : The theory fits well into the experiment
jin
ie., the experimental results supports the theory
Total Number of beans = 1600

.re
Divide these beans in the ratio 9:3:3:1
To calculate the expected frequencies
9
E(882) = 1600 900
w
16
3
E(313) = 1600 300
16
w
3
E(287) = 1600 300
16
w
1
E(118) = 1600 100
16
Oi : 882 313 287 118

Ei : 900 300 300 100
2
n
2 Oi Ei
Test Statistic
i 1 Ei
om
2 2 2 2
2 882 900 313 300 287 300 118 100
900 300 300 100
= 0.36 + 0.563 + 0.563 + 3.24

2
= 4.726
.c
Critical Value:
ul
The table value of at 5% for 3 d.f is 7.815
Conclusion:
Since 2
pa
< 7.815, H0 is accepted at 5% Los.
We conclude that there is a very good correspondent between theory and experiment
3. 4 coins were tossed 160 times and the following results were obtained.
jin
No. of heads : 0 1 2 3 4
Frequency : 19 50 52 30 9 160
.re
0 50 104 90 36 280
2
Test the goodness of fit with the help of on the assumption that the coins are unbiased
Solution:
w
We set up, the null hypothesis, the coins are unbiased:
1
w
The probability if getting the success of heads is p =

2
1
w
q = 1- p =
2
When 4 coins are tossed, the probability of getting 'r' heads is given by,
P(x = r) = n Cr pr qn-r ; r = 0, 1, 2, 3, 4
r 4-r
1 1
= 4Cr
2 2
om
4
1
= 4Cr
2
1
P(x = r) 4C r r = 0, 1, 2, 3, 4
16
.c
1
The expected frequencies of getting 0, 1, 2, 3, 4 heads are given by 1604Cr
16
ul
= 104Cr , r = 0, 1, 2, 3, 4
= 10, 40, 60, 40, 10
Oi : 19 50 52
pa
30 9
Ei : 10 40 60 40 10
26 48 43 26 12
jin
2
n
2 Oi Ei
Test Statistic
i 1 Ei
.re
2 2 2 2 2
2 19 10 50 40 52 60 30 40 9 10
10 40 60 40 40
= 8.1 + 2.5 + 1.067 + 2.5+0.1

w
2
= 14.267
D.f V= n-1=5-1=4
w
Critical value:
2
The table value of for 4 d.f at 5% Los is 9.488
w
Conclusion:
2
Since > 9.488, H 0 is rejected at 5% Los
The coins are biased
4. The follwoing table shows the distribution of goals in a football match
No. of goals : 0 1 2 3 4 5 6 7
om
No. of mistakes : 95 158 108 63 40 9 5 2
Fit a poisson distribution and test the goodness of fit.
Solution:
Fitting of poisson distribution
.c
x: 0 1 2 3 4 5 6 7
ul
f : 95 158 108 63 40 9 5 2
fx 812 and f 480
x
fx
f
812
480
1.7
pa
The expected frequencies are computed by
jin
1.7
e (1.7) r
= 480 r = 0, 1, 2, 3, 4, 5, 6, 7
r!
= 88, 150, 126, 72, 30, 10, 3, 1

.re
We set up H0 : The fit is good
2
n
2 Oi Ei
Test Statistic
w
i 1 Ei
Oi : 95 158 108 63 40 9 5 2
w
16
Ei : 88 150 126 72 30 10 3 1
14
w
2 2 2 2 2 2 2
2O E 95 88 158 150 108 126 40 30 16 14 63 72
=
E 88 150 126 30 14 72
= 0.56 + 0.43 + 2.57 + 3.33 + 1.12 +0.29
2
= 8.30
Number of degrees of freedom V= n-2 = 6 - 2 =4
Critical value:
om
2
The table value of at 5% Los for 4 d.f is 9.483
Conclusion:
2
Since < 9.483, H 0 is accepted at 5% Los.
The fit is good
.c
2
5. Apply the test of goodness of fit to the follwoing data
ul
Oi : 1 5 20 28 42 22 15 5 2
Ei : 1 6 18 25 40 25 18 6 1
Solution:
H0 : The fit is good

pa
= 0.05 (or) 5%
jin
Oi : 1 5 20 28 42 22 15 5 2
6 7
Ei : 1 6 18 25 40 25 18 6 1
.re
7 7
n=7
2
n
2 Oi Ei
Test Statistic
w
i 1 Ei
2 2 2 2 2 2 2
2 6 7 20 18 28 25 42 40 22 25 15 18 7 7
w
7 18 25 40 25 18 7
= 0.143 + 0.222 + 0.36 + 0.1 + 0.36 +0.5 +0

w
2
= 1.685
d.f V= n-1 = 7 - 1 = 6
Critical value:
2
At 5% Los, the table value of for 6 d.f is 12.592
Conclusion:
om
2
Since <12.592, H0 is accepted at 5% Los.
The fit is good

6. The follwoing table shows the number of electricity failures in a town for a period of 180 days
Failures : 0 1 2 3 4 5 6 7
.c
No. of days : 12 39 47 40 20 17 3 2
2
Use , examine whether the data are poisson distributed.
ul
Solution:
Fitting of poisson distribution pa

x: 0 1 2 3 4 5 6 7
f : 12 39 47 40 20 17 3 2
jin
fx : 0 39 94 120 80 85 18 14
fx 450 and f 180
fx 450
.re
x 2.5
f 180
The expected frequencies are computed by
2.5
(2.5) r
w
e
= 180 r = 0, 1, 2, 3, 4, 5, 6, 7
r!
Ei = 15, 37, 46, 38, 24, 12, 5, 2

w
r = 0 1 2 3 4 5 6 7
w
We set up H0 : The fit is good
Oi : 12 39 47 40 20 17 3 2
5
Ei : 15 37 46 38 24 12 5 2
7
2
Test Statistic
2 2 2 2 2 2 2
om
2 12 15 39 37 47 46 40 38 20 24 17 12 5 7
15 37 46 38 24 12 7
= 0.6 + 0.108 + 0.022 + 0.105 + 0.667 +2.083+0.5+1

2
= 4.156
.c
d.f V = n-1 = 7 - 1 = 6
ul
Critical value:
2
Conclusion:
Since 2
< 12.592, H0 is accepted at 5% Los.
pa
The fit is good
jin
Test for Independence of Attributes
Attribute A
.re
A1 A2 ................A j .................. At Total
Attribute B1 O11 O12 ................O1j.................. O1t (B1 )
B B2
w
.
w
Bi Oi1 Oi 2 ................Oij .................. Oit (Bi )

w
. .
. .
Bs Os1 Os 2 ................Osj.................. Ost (Bs )
Total (A1 ) (A2 )................(Ai ).................(At ) N
Attribute A
om
A1 A 2 ................A j .................. A t Total
Attribute B1 O11 O12 ................O1j.................. O1t (B1 )
B B2
.
.
.c
.
Bi O i1 Oi 2 ................O ij.................. Oit (Bi )
. .
ul
. .
. .
Total
Bs O s1
(A1 )
pa
O s 2 ................O sj.................. O st
(A 2 )................(A i ).................(A t )
(B s )
N
Such a table is called (s t) consistency table

jin
Here, N Total Frequency
Oij Observed frequency of (i, j)th cell

.re
The expected frequency eij obtained by the rule
row total Bi Column total A j

eij Where i = 1, 2, 3 ....... s
N
w
j = 1, 2 ........... t
Degrees of freedom associated with s t consistency table = (s - 1) (t - 1)

w
Chi-square table for 2 2 consistency table
a b
w
2
In a 2 2 consistency table where in the frequencies are , the value of is
c d
2 (a b c d )(ad bc)2
(a b)(a c)(c d )(b d )
Problems :
1. An opinion poll was conducted to find the reaction to a proposed civic reform in 100 members
of each of the two political parties theinformation is tabulated below
om
Favorable Unfavorable Indifferent
Party A 40 30 30
Party B 42 28 30
Test for Indepedence of reduction with the party affiliations.
.c
Solution:
We set up H0 : Reactions and party affiliations are independent
ul
The expected frequencies are calculated by
pa Total
40 30 30 100
42 28 30 100
jin
Total 82 58 60 200
Favorable Unfavorable Indifferent
82 100 58 100 60 100

Party A 41 29 30
.re
200 200 200
82 100 58 100 60 100

Party B 41 29 30
w
200 200 200

2
Test Statistic
w
2
n
2 Oi Ei
i 1 Ei
w
Oi : 40 30 30 42 28 30
Ei : 41 29 30 41 29 30
2 2 2 2 2 2
2 40 41 30 29 30 30 42 41 28 29 30 30
41 29 30 41 29 30
= 0.024 + 0.024 + 0.034 + 0.034
om
2
= 0.116
Number of degrees of freedom =(2-1)(3-1) = 2
Critical value:
2
.c
Conclusion:
ul
2
Since < 5.99, H0 is accepted at 5% Los.
The independence of reactions with the party affiliations may be correct.

pa
2. In a locality 100 persons were randomly selected and asked about their educational
achievements. The results are given below.
Education
jin
Middle High School College
Male 10 15 25
Female 25 10 15
Can you say that education depends on sex?
.re
3. The following table gives the classification of 100 workers according to sex and the nature
of work. Test whether nature of work is independent of the sec of the worker.
Skilled Unskilled Total

w
Male 40 20 60
Sex Female 10 30 40
w
Total 50 50
w
Solution:
H0 : Nature of work is independent of the sex of the worker
Under H0 , the expected frequencies are

60 50 60 50
E(40) = 30; E(20) = 30
100 100
40 50 40 50
E(10) = 20; E(30) = 20
100 100
om
2
Test Statistic
2
n
2 Oi Ei
i 1 Ei
.c
Oi : 40 20 10 30
Ei : 30 30 20 20
ul
2 2 2 2
2 40 30 20 30 10 20 30 20
30 30 pa 20 20
= 3.333 + 3.333 + 5 + 5
2
= 16.67
Number of degrees of freedom =(2-1)(2-1) = 1

jin
Critical value:
2
The table value of at 5% Los, for 1 d.f is 3.841
.re
Conclusion:
2
Since > 3.841, H0 is rejected at 5% Los.
We conclude that the nature of work is dependent on sex of the worker.

w
4. From the following data, test whether there is any association between intelligency and
w
economics conditions
Intelligences
w
Excellent Good Medium Dull Total

Economic Good 48 200 150 80 478
Conditions Not Good 52 180 190 100 522
Total 100 380 340 180 1000
Solution:
H0 : There is no association between intelligency and economic conditions.
Los : = 0.05 (or) 5%
om
Under H0 , the expected frequencies are obtained as follows
100 478 100 522

E(48) = 47.8; E(52) = 52.2
1000 1000
380 478 380 522
.c
E(200) = 181.64; E(180) = 198.36
1000 1000
478 340 340 522

E(150) = 162.52; E(190) = 177.48
ul
1000 1000
180 478 180 522

E(80) = 86.04; E(100) = 93.96
1000 pa 1000
Oi : 48 200 150 80 52 180 190 100

jin
Ei : 47.8 181.64 162.52 86.04 52.2 198.36 177.48 93.96
2
Test Statistic
.re
2
n
2 Oi Ei
i 1 Ei
2 2 2 2 2
48 47.8 150 162.52 52 52.2 190 177.48 200 181.64
47.8 162.52 52.2 177.48 181.64
w
w
2 2 2
80 86.04 180 198.36 100 93.96
86.04 198.36 93.96
w
= 0.0008 + 0.9645 + 0.0008 + 0.8832+1.8558+0.4240+1.6994+0.3883

2
= 6.2168
Number of degrees of freedom = (s-1)(t-1) = (2-1)(4-1) = 3

Critical value:
2
Conclusion:
om
2
We conclude that there is no association between intelligency and economic conditions
5. From the following data, test the hypothesis that the flower color is independent of flatness
.c
of leaf
Flat leaves Curved leaves Total

White Flowers 99 36 135
ul
Red Flowers 20 5 25
Total 119 41 160
Solution:
pa
We set up: H0 : flower color is independent of flatness of leaf. Los =0.05 (or) 5%
The given probelm is a 2 x 2 consistency table

jin
2
we use the formula to find is
2 (a b c d )(ad bc)2
.re
(a b)(a c)(c d )(b d )
Here, a = 99; b = 36; e = 20; d=5
2 160(495 720)2 160(50625)

w
(135)(119)(25)(41) 16, 466,625
2
= 0.4919
w
Number of degrees of freedom = (s-1)(t-1) = (2-1)(2-1) = 1

w
Critical value:
2
Conclusion:
2
Flower colour is independent of flatness of leaf.
om
Test for single variance
Chi-square test for population variance

2 2
In this method, we set up the null hypothesis H0 : 0 (with a specified variance)
.c
2 ns 2
The test statistic 2
Where n = sample size
ul
s = sample variance
Note:
= population variance
pa
If the sample size n is large (>30)
jin
2
The test statistic z= 2 2n 1 ~ N (0,1)
We use the usual normal test.

.re
1. A random sample of size 9 from a normal population have the following values 72, 68, 74, 77,
61, 63, 63, 73, 71. Test the hypothesis that the population variance is 36.
Solution:
w
2
Null hypothesis H0 : 36
2
w
Alternative hypothesis H1: 36
Los : 0.05 (or) 5%

w
2 ns 2
The test statistic 2
x : 72 68 74 77 61 63 63 73 71
x 622
x 622; x 69.11
n 9
x x : 2.9 -1.1 4.9 7.9 -8.1 -6.1 -6.1 3.9 1.9
om
x x : 8.41 1.21 24.01 62.41 65.61 37.21 37.21 15.21 3.61
2
x x 254.89
2 ns 2 254.89
7.08
.c
2
36
d. f n-1 = 9-1 = 8
ul
Critical value:
2
The table value of for 8 d.f at 5% Los is 15.51
pa
Conclusion:
2
We conclude that the hypothesis of population variance is 36 is accepted

jin
2. Test the hypothesis that 10, given that s = 15 for a random sample of size 50 from a
normal population
.re
Solution:
Null hypothesis H0: 10
Alternative hypothesis H1: 36

w
We are given n = 50; s = 15

w
2 ns 2 50 225
2
= 112.5
100
w
2
Since 'n' is large (n > 30, the test statistic z = 2 2n 1
= 225 99 15 9.95
z = 5.05
This statistic z follows N(0,1)
Critical value:
om
At 5% Los, the table value of z is 3
Conclusion:
Since | z | > 3, H0 is rejected.
.c
We conclude that 10
3. The standard deviation of the distribution of times taken by 12 workers for performing a Job
ul
is 11 sec. Can it be taken 1 as a sample from a popualation whose S.D is 10 sec.
Solution:
Let H0: 10
pa
ie., the population standard deviation 10
H1: 10
jin
Los : 0.05 (or) 5% Los
Given n = 12; s = 11
.re
The test statistic is
2 ns 2
2
w
12 121
= 14.52
100
w
2
14.52
Degrees of freedom = n - 1 = 12 - 1 = 11
w
Critical value:
2
The table value of for 11 d.f at 5%Los is 19.675.
www.rejinpaul.com
www.rejinpaul.com
Conclusion:
2
Since < 19.675, H0 is accepted at 5% level
The S.D of the time element is 10 sec is supported.
om
ie., the population standard deviation 10
.c
ul
pa
jin
.re
w
w
w

UNIT-II
DESIGN OF EXPERIMENTS
Analysis of variance:
The technique of analysis of variance is referred to as ANOVA. A table showing the source of
om
variance, the sum of squares, degrees of freedom, mean squares(variance)and the formula for the “F ratio
is known as ANOVA table”
The technique of analysis if variance can be classified as
(i) One way classification(CRD)
.c
(ii) Two way classification(RBD)
(iii) Three way classification(LSD)
One way classification:
ul
In one way classification the data are classified on the basic of one criterion
The following steps are involved in one criterion of classification

pa
(i) The null hypothesis is
Ho : 1  2  ......  k
H1 : 1  2  .......  k
jin
(ii) Calculation of total variation
G2
Total sum of squares V   j xij  N
2
.re
Where G    xij (Grand total)

i j
G2
=correction formula
N
w
(iii) Sum of squares between the variates

w
T 2  G2
V1    i   With k-1 degree of freedom
i  ni  N
(iv) Sum of squares within samples
w
V2  V  V1

V1
then the ratio K  1 follows F-distribution with degrees of freedom. Choosing the ratio which is
V2
N K
greater than one, we employ the F-test
om
If we calculated F < table value F0.05 , the null hypothesis is accepted.
ANOVA Table for one way classification
Source of Sum of Degrees of Mean square Variance ratio

variation square freedom
.c
Between V1 K-1 V1 V1
classes K 1 K  1 (or)
V2
N K
ul
V2 V2
N K V2
Within N K N K
classes pa V1
K 1
V N 1
1. To test the significance of the variation of the retail prices of a certain commodity in the four
jin
principal plates A,B,C &D, seven shops were chosen at random in each city and the prices
observed were as follows (prices in paise)
A 82 79 73 69 69 63 61
B 84 82 80 79 76 68 62
.re
C 88 84 80 68 68 66 66
D 79 77 76 74 72 68 64
Do the data indicate that the prices in the four cities are significantly different?
Solution:
w
H0 : 1  2  3  4
H1 : 1  2  3  4
w
i.e., the prices of commodity in the four cities are same.

w
we take the origin at x  80 and the calculation are done as follows.

Calculation of ANOVA (use new values)
Cities
K=4 1 2
Shop(n = 7)
3 4 5 6 7
Ti Ti 2 x 2
n
A 2 -1 -7 -11 -11 -17 -19 -64 585.14 946
om
B 4 2 0 -1 -4 -12 -18 -29 120.14 505
C 8 4 0 -12 -12 -14 -14 -40 228.57 760
D -1 -3 -4 -6 -8 -12 -16 -50 357.14 526
G2
 1196.03
G  183 T i
2
 1290.9
 x
i j
ij
2
 2737
N n
2
G
Total sum of squares V=  x 2

.c
ij
i j N
=2737-1196.03
ul
V=1540.97
Sum of squares between cities pa

Ti 2 G 2
V1   
n N
= 1290.9 – 1196.03
jin
V1  94.87
Sum of squares within cities
V2  V-V1  1540.97  94.87

.re
V2  1446.1
ANOVA Table:
w
Source of Sum of square Degrees of f Mean square F

variation of deviation
V1  94.87 K-1=4-1=3 V1

94.87
w
Between cities K 1 3
60.25
 31.62 
V2  1446.1 N-K=28-4=24 V2 1446.1 31.62
  1.90
w
Within cities N K 24
 60.25
Total V=1540.97 N-1=27

Number of degrees of freedom = (N - K, K - 1) = (24,3)
Critical value:
The table value of F for (24, 3) degree of freedom at 5% Los is 8.64
om
Conclusion:
Since F< 8.64, H o is accepted at 5% Los
 The prices of commodity in the four cities are same
2. Fill up the following Analysis of variance table
.c
Source of Degrees of Sum of squares Mean squares F ratio
variation freedom
Treatments - - 117
ul
-
Error - 704 -
Total 16 938
Solution:
pa
From the given table we have,
V2  704;V=938
jin
degree of freedom (total) N - 1 = 16 => N = 17
V1
mean squares  117
K 1
.re
We Know that V2  V-V1
=> V1  V-V2
w
= 938 – 704
V1  234
w
V1
 117
K-1
w
234 234
  117 =>  K-1
K-1 K-1
K–1=2

degree of freedom (K-1) = 2
=>K=3
Next, N-K = 17-3 = 4
om
V2 938
  50.29
N-K 14
Source of Degrees of Sum of Mean squares F ratio

variation freedom squares
Treatments K-1=3-1=2 V1  234 V1 117
 117
.c
K-1 50.29
Error
N-K=17-3=14 V2  704 V2
 50.29  2.327
N-K
ul
Total 16 V = 938
3.
pa
The following are the number of mistakes made in 5 successive days of 4 technicians working in
a photographic laboratory
Technicians I Technicians II Technicians III Technicians IV

6 14 10 9
jin
14 9 12 12
10 12 7 8
8 10 15 10
11 14 11 11
.re
Test at the 1% Los whether the difference among the 4 samples means can be attributed to chance
Solution:
H0 : 1  2  3  4
w
ie., There is no differences among the 4 samples mean

w
H1 : 1  2  3  4
We take the origin at 12 and the calculation are done as follows

w

Calculation of ANOVA (NEW Values)
Days(5)
Technicians
K=4
Ti Ti 2 x 2
n
1 2 3 4 5
om
I -6 2 -2 -4 -1 -11 24.2 61
II 2 -3 0 -2 2 -1 0.2 21
III -2 0 -5 3 -1 -5 5 39
IV -3 0 -4 -2 -1 -10 20 30
Total G 2 (27) 2 G=-27 49.4 151
=  36.45
.c
N 20
Total sum of squares:
ul
G2
V= xij2 
i j N
 151  36.45
pa
V=114.55
Sum of squares b/w cities:

jin
Ti 2 G 2
V1   
n N
 49.4  36.45
.re
V1  12.95
Sum of squares within cities:

w
V2  V-V1  114.55 12.95
V2  101.6
w
w

Source of Sum of squares Degrees of Mean squares F ratio
variation of deviation freedom
B/W V1  12.95 K-1=4-1=3 V1 12.95
Technicians   4.31
K-1 3
om
6.35

V2  101.6 N-K=20-4=16 4.31
Within
V2 101.6
Technicians   6.35
N-K 16
Total V=114.55 N-1=19  1.473
.c
Degrees of freedom ( (N - K, K - 1) = (16,3)
ul
Critical value:
The table value of ‘F’ for (16,3) degree of freedom at 1% Los is 5.29
Conclusion:
pa
Since F < 5.29, H 0 accepted at 1% level
 There is no difference among the four sample means.

jin
4. The following table shows the lives in hours of four batches of electric lamps.
Batches Lives in hours

1 1610 1610 1650 1680 1700 1720 1800
.re
2 1580 1640 1640 1700 1750

3 1460 1550 1600 1620 1640 1660 1740 1820
4 1510 1520 1530 1570 1600 1680
Perform an analysis of the variance on these data and show that a significant test does not reject
w
their homogeneity
Solution:
w
H0 : 1  2  3  4
I.e., the means of the lives of the four brands are homogeneous.
w
H1 : 1  2  3  4

oldxij  1700
We take the origin xij 
10
Calculation of ANOVA
Brand Lives Ti Ti 2 x 2
om
ij
K=4 1 2 3 4 5 6 7 8 ij
n
1 -9 -9 -5 -2 0 2 10 - -13 24.143 295
2 -12 -6 -6 0 5 - - - -19 72.2 241
3 -24 -15 -10 -8 -6 -4 4 12 -51 325.125 1177
4 -19 -18 -17 -13 -10 -2 - - -79 1040.167 1247
.c
Total G 2 (162) 2
  1009.38 G=-162 =1461.635 2960
N 26
ul
N=n1  n2  n3  n4  7  5  8  6  26
Total sum of squares: pa

G2
V= ( xij )2 
i j N
 2960 1009.38
jin
V=1950.62
Total sum of squares b/w brands:

.re
Ti 2 G 2
V1   
n N
 1461.635 1009.38
V1  452.255
w
Sum of squares within brands:

w
V2  V-V1
 1950.62  452.255
w
V2  1498.365

ANOVA Table:
om
variation freedom
B/W V1  452.255 K-1=4-1=3 V1

452.255
 150.75
Brands K-1 3
150.75

68.11
Within N-K=26-4=22
V2  1498.365 V2 1498.365  2.21
Brands   68.11
.c
N-K 22
Total 1950.62  V N-1=25
ul
Degrees of freedom (3, 22) =3.05
Critical value:
pa
The table value of ‘F’ for (3,22) d.f at 5% Los is 3.05
Conclusion:
jin
Since F<3.05, H 0 is accepted at 5% level
 The means of the lives of the four brands are homogeneous.
ie., the lives of the four brands of lamps do not differ significantly.
.re
Two way classification:
In two way classification the data are classified on the basis of two criterions
The following steps are involved in two criterion of classification

w
(i) The null hypothesis
H 01 and H 02 framed
w
We compute the estimates of variance as follows

w
(ii) G=  xij  Grand total of K  n Observations

i j
G2
S : Total sum of squares  xij 
2
(iii)
N

1 n G2
(iv) S1 :Sum of squares b/w rows (class-B)   R j 
2
K j=1 N
1 K 2 G2
(v) S 2 :Sum of squares b/w (classes A)   Ci  N
n i=1
om
S 3 : Sum of squares due to error (or) Residual sum of squares
(vi) Errors (or) Residual S 3 = S - S1 - S 2
(vii) The degrees of freedoms of
S1  n-1 ;S2  k-1 ;S3  (n-1)(k-1)
S=nk-1
.c
ANOVA Table for two way classification
ul
Source of Sum of Degrees of Mean squares F ratio
variation squares freedom
B/W ‘B’ S1 n-1 S1 QB
classes(rows)  QB F1 
pa n-1 Q AB
d . f  [(n-1)(k-1)(n-1)]
B/W ‘A’ S2 k-1 S2 Q
classes(column)  QA F2  A
k-1 QAB
jin
Residual (or) S3 (n-1)(k-1) S3 d . f  [(k-1),(k-1)(n-1)]
error  Q AB
(n-1)(k-1)
Total S nk-1 - -
.re
Advantages of R.B.D:
The chief advantages of R.B.D are as follows
(i) This design is more efficient or more accurate than CRD. This is because of
w
reduction of experimental error.

(ii) The analysis of the design is simple and even with missing observations, it is not
much complicated
w
(iii) It is Quite flexible, any number of treatments and any number of replication may be
used
(iv) It is easily adaptable as in agricultural experiment it can be accommodated well in a
rectangular, squares(or)in a field of any shape
w
(v) It provides a method of eliminating or reducing the long term effects.

(vi) This is the most popular design with experiments in view of its simplicity, flexibility
and validity. No other has been used so frequently as the R.B.D

Disadvantages:
(i) The number of treatments is very large, than the side of the blocks will increase and
this may introduce heterogeneity within blocks.
(ii) If the interactions are large, the experiments may yield misleading results.
om
1. The following data represent the number of units of production per day turned out by four
randomly chosen operators using three milling machines
Machines.
M1 M2 M3
1 150 151 156
.c
2 147 159 155
Operators 3 141 146 153
4 154 152 159
Perform analysis of variance and test the hypothesis
ul
(i) That the machines are not significantly different
(ii) That the operators are not significantly different at 5% level
Solution:
pa
H 01 : There is no significantly difference bet machine and
H 02 : There is no significantly a difference b/w operator

jin
We take the origin 155 and the calculations are done as follows.
Calculation of ANOVA (using new values)

.re
Operators Machines Row total

x ij
2
Rj j
M1 M2 M3
1 -5 -4 1 -8 42
2 -8 4 0 -4 80
w
3 -14 -9 -2 -25 281

4 -1 -3 4 0 26
Column -28 -12 3 429
w
total -37
Ci
x ij
2 286 122 21 429
w
Here N=12 ; G=-37
G 2 (37) 2
Correction factor   114.08
N 12

G2
S=  xij2 
i j N
 429 114.08
om
 314.92
Sum of squares between operators:
R j2 G2
S1   
.c
j nj N
1
 [(8)2  (4)2  (25) 2 ]  114.08
ul
3
 235 114.08
 120.92
Sum of squares between machines:

pa
 C 2  G2
S2    i  
jin
i  ni  N
1
 (28) 2  (12) 2  (3) 2   114.08
4
.re
 234.25 114.08
S2  120.17
Residual sum of squares:

w
S3  S-S1  S2
 314.92 120.92 120.17

w
 73.83
w

AVOVA Table for two way classification
Source of Sum of Degrees of Mean sum squares F ratio

om
B/W 120.92 n-1=4-1=3 S1
operators QB   40.31
n 1
B/W 120.17 k-1=3-1=2 S 40.31
machines QA  2  1.49
k-1 12.305
 60.09 (3, 6)
Residual 73.83 (n-1)(k-1)=6 S3 60.09
.c
Q AB   4.88
(k-1)(n-1) 12.305
 12.305 (2, 6)
Total 314.92 nk-1=11
ul
Degrees of freedom V1  2; V2  6 (machines)
Degrees of freedom V1  3;V2  6 (operators)

pa
Critical value:
(i) Machines
The table value of ‘F’ for (2,6) d.f at 5% Los is 5.14
jin
(ii) Operators
The table value of ‘F’ for (3,6) d,f at 5% Los is 4.76
Conclusion:
.re
(i) Operators
Since F< 4.76 , H 02 is accepted at 5% level
 The operators are not significantly different
(ii) For Machines
Since F< 5.14, H 01 is accepted at 5% level
w
 The machines are not significantly different
2. An experiment was designed to study then performance of four different detergents, the following
w
“whiteness” readings were obtained with specially designed equipment for 12 loads of washing
distributed over three different models of washing machines.
w

Machines 1 2 3 Total
Detergents
A 45 43 51 139
B 47 46 52 145
om
C 48 50 55 153
D 42 37 49 128
Total 182 176 207 565
Looking on the detergents as treatment and the machines as blocks, obtain the appropriate analysis of
variance table and test at 0.01 level of Significance whether there are differences in the detergents (or) in
.c
the washing machines
Solution:
ul
H 01 : There is no significant different b/w detergent
H 02 : There is no significant different b/w washing machine
We take the origin is 50 and the calculation are done as follows.
Calculation of ANOVA (using new values)

pa
Detergents
Washing machines
Row total
x ij
2
Rj j
jin
M1 M2 M3
A -5 -7 1 -11 75
B -3 -4 2 -5 29
C -2 0 5 3 29
D -8 -13 -1 -22 234
.re
Column total -18 -24 7 367

Ci -35
x
w
2 102 234 31 367

ij
i
w
Here N=12; G=-35
G 2 (35) 2
  102.08
w
Correction factor
N 12
G2
Total sum of squares: S=  xij2 
i j N

 367  102.08
S=264.92
R j2 G2
Sum of squares b/w detergents: S1   j hj

N
om
1
 (11) 2  (5) 2  (3) 2  (22) 2   102.08
3
 213 102.08
S1  110.92
.c
Sum of squares between machines
ul
 C 2  G2
S2    i  
i  ni  N

1
4

pa
(18)2  (24)2  (7)2   102.08
 237.25 102.08
S2  135.17
jin
Residual sum of squares S3  S-S1  S2
 264.92 110.92 135.17

.re
S3  18.83
w
w
w

ANOVA table for two way classification:
Source of Sum of Degrees of Mean squares F ratio

S1  110.92
om
B/W n-1=4-1=3 S1 110.92
detergents QB   QB 36.97
n-1 3 
 36.97 QAB 3.14
B/W S2  135.17 k-1=3-1=2 S
QA  2 
135.17  11.77
machines
k-1 2
 67.59
.c
Residual S3  18.83 (n-1)(k-1)=6 QA 67.59

(or) S3 18.83 QAB 3.14
Error QAB  
(n-1)(k-1) 6  21.52
ul
 3.14
Total S=264.92 nk-1=11
Degrees of freedom V1  2;V2  6 (machines)

pa
Degrees of freedom V1  3;V2  6 (detergents)
jin
Critical value:
(i) Detergents:
The table value of F for (3,6) degree of freedom at 1% Los is 9.78

.re
(ii)Machines
The table value of F for (2,6) degree of freedom at 1% Los is 10.92
Conclusion:
w
(i) For detergents
Since F>9.78, H 01 is rejected at 5% level

w
 The detergents are significantly different
(ii) For machines

w
Since F>10.92, H 02 is rejected at 5% level
 The machines are significantly different

3. To study the performance of three detergents and three different water temperatures the following
whiteness readings were obtained with specially designed equipment.
Water temp Detergents A Detergents B Detergents C

Cold Water 57 55 67
om
Worm Water 49 52 68
Hot Water 54 46 58
Solution:
.c
We set the null hypothesis
H 01 : There is no significant different in the three varieties of detergents
ul
H 02 : There is no significant different in the water temperatures
We choose the origin at x=50 pa

Water temp Row total
Detergents Rj x
j
ij
2
A B C
Cold Water 7 5 17 29 363
jin
Worm Water -1 2 18 19 329
Hot Water 4 -4 8 8 96
Column total
Ci 10 3 43 56 788
.re
66 45 677 788
x
i
ij
2
w
G2
S=  xij  2
w
j i N
(56) 2
 788   788  348.44
w
S=439.56

Sum of squares between detergents:
Ci 2 G 2
S1   
i ni N
1
 (10)2  (3)2  (43)2   348.44
om
3
 652.67  348.44
S1  304.23
.c
Sum of squares b/w temperatures:
R j2 G2
S2   
nj N
ul
j
1
 1266  348.44
3 pa
 422  348.44
S2  73.56
jin
Error sum of squares:
S3  S-S1  S2
 439.56  304.23  73.56

.re
S3  61.77
ANOVA Table:
w

variation freedom
B/W detergents 2 304.23
304.23 152.11
2
w
15.445
 152.11
 9.848
73.56 (2, 4)
B/W
w
temperatures 2 2 36.78
73.55  36.78 15.445
 2.381
Error 61.79 4 15.445
Total 439.56 8

Degrees of freedom (2,4) and (2,4)
Critical value:
The table value of F for (2,4) d.f at 5% Los is 6.94
om
Conclusion:
(i) For detergents:
Since F > 9.85, H 01 is rejected at 5% Los
 There is a significant different between the three varieties detergents,
.c
(iii) For water temperature
Since F<6.94, H 02 is accepted at 5% Level
 There is no significant different in the water temperatures.
ul
4. Four experiments determine the moisture content of samples of a powder, each man taking a
sample from each of six consignments. These assignments are
Observer Consignmentpa
1 2 3 4 5 6
1 9 10 9 10 11 11
2 12 11 9 11 10 10
3 11 10 10 12 11 10
4 12 13 11 14 12 10
jin
Perform an analysis if variance on these data and discuss whether there is any significant different b/w
consignments (or) b/w observers.
Solution:
We formulate the hypothesis
H 02 : There is no significant different b/w observer
.re
H 02 : There is no significant different b/w consignment
We take origin at x=11 and the calculations are done are as follows
w
Calculation ANOVA:
Observer
1
consignments
2 3 4 5 6
Rowtotal
x ij
2
Rj j
w
1 -2 -1 -2 -1 0 0 -6 10
2 1 0 -2 0 -1 -1 -3 7
3 0 -1 -1 1 0 -1 -2 4
w
4 1 2 0 3 1 -1 6 16
Column total C i 0 0 -5 3 0 -3 37
-5
xj
ij
2 6 6 9 11 2 3 37

G2
Total sum of squares   xij  2
j i N
( 5) 2
S=37-  35.96
24
om
(R j ) 2 G2
Sum of squares b/w observers   
nj N
1 25
S1  (6) 2  (3) 2  (2) 2  (6) 2  
6 24
.c
S1  13.13
ul
 Ci 2  G 2
Sum of squares b/w consignments    
 ni  N
S2 
1
4
pa
(0  0  25  9  9) 
25
24
S2  9.71
jin
Error sum of squares S3 = S  S1  S2
= 35.96 - 13.13 - 9.71
S3 = 13.12
.re
Source of Sum of Degrees of Mean ‘F’ ratio

variation squares freedom squares
B/W S1  9.71 n-1=5 9.71 1.94
Consignments 5 0.87
w
 1.94  2.23
(5,15)
B/W observers S2  13.13 k-1=3 13.13 4.38
w
3 0.87
 4.38  5.03
Error S3  13.12 (n-1)(k-1)=15 13.12
w
(3,15)
15
 0.87
Total S  35.96 nk-1=23

Critical value:
(i) For consignments ,

The table value of ‘F’ for (5, 15) d.f at 5% Los is 2.90
(ii) For observers:
The table value of F for (3, 15) d,f at Los 3.29
om
Conclusion:
(i) For observers

Since F>3.29, H 01 is rejected
Hence these is a difference between observers is significant
.c
(ii) For consignment:
Since F<2.33, H 02 is accepted
 There is no significant different b/w the consignments
ul
LATIN SQUARES DESIGN:
A Latin squares is a squares arrangement of m-rows and m-columns such that each symbol
pa
appearly once and only once in each row and column.
In randomized block design the randomization is done within blocks the units in each block
being relatively similar in L.S.D there are two restrictions
(i) The number of rows and columns are equal

jin
(ii) Each treatment occurs once and only once in each row and column.
This design is a three way classification model analysis of variance
The following steps are involved in Latin square design

.re
G2
Correction factor  ; G -> Grand total
N
m
Si 2
S.S b/w rows=Sa    C.F
w
(S.S means Sum of Squares)

i=1 m
m S j2 G2
S.S b/w Columns=Sb   
w
|C.F
j=1 m N
m
Vi 2
S.S b/w Varieties=Sc    C.F
w
i=1 m
Total sum of 
 S   xij  C.F
2
squares  j i

and Sd  S-Sa  Sb  Sc
Here Si =sum of i th row
Sj =sum of jth column
om
Vi =sum of i th variety
ANOVA Table:
Sum of
Source of Degrees of Mean squares ‘F’ ratio
.c
squares
variation freedom
ul
B/W Rows Sa
Sa m-1 R E
m-1
(m-1),(m-1)(m-2)
B/W
pa C
Sb
Columns Sb m-1 C E
m-1
(m-1),(m-1)(m-2)
V
B/W varieties Sc
Sc m-1 V E
m-1
(m-1),(m-1)(m-2)
jin
Error Sd
Sd (m-1)(m-2) E
(m-1)(m-2)
Total
S m2 1
.re
Comparison of LSD and RBD
(i) In LSD, the number of rows and number of columns are equal and hence the number of
replication is equal to the number of treatments there is no such restriction in RBD
(ii) L.S.D is suitable for the case when the number of treatments is b/w 5 and 12 where as R.B.D
w
can be used for any number of treatments and replications

(iii) The main advantage of L.S.D is that it removes the variations b/w rows and columns from
that within the rows resulting in the reduction of experiment error to a large extent
w
(iv) The RBD can be performed equally on rectangular of square plots but for LSD, a mose (or)
less a squares field is required due to (iii) LSD is preferred over RBD
Note: A 2  2 Latin Square Design is not possible. The degree of freedom for error in a m  m
w
Latin squares design is (m-1)(m-2)
For m=2 the degree of freedom is ‘o’ and hence comparisons are not possible.
Hence a 2  2 LSD is not possible.

1. The following is the LSD layout of a design when 4 varieties of seeds are being tested set up the
analysis of variance table and state four conclusion
A B C D
105 95 125 115
C D A B
om
115 125 105 105
D C B A
115 95 105 115
B A D C
95 135 95 115
Solution:
.c
H: There is no significant difference
xij  100
we take the origin as u ij  and the calculations are done as follows
ul
5
Varieties Values Vi
A
B
C
1
-1
5
1
1
3
3
-1
-1
pa7
-1
3
12
0
10
D 3 5 3 -1 10
jin
Columns /
Rows
C1 C2 C3 C4 Row total
x ij
2
Rj i
R1 1 -1 5 3 8 36
R2 3 5 1 1 10 36
.re
R3 3 -1 1 3 6 20
R4 -1 7 -1 3 8 60
Columns 6 10 6 10 G=32 152
total C i
w
x
j
ij
2 20 76 28 28 152
w
G=32 N=16;  x
j i
ij
2
 152
w
G 2 (32) 2
C.F=   64
N 16

G2
Total sum of squares   xij  2
j i N
(32) 2
 152 
16
om
 152  64
S=88
1 2
Sum of squares b/w rows  8  102  62  82   64
4
.c
 66  64
ul
Sa =2
1 2
Sum of squares b/w columns  6  102  62  102   64
4
Sb  68  64
pa
Sb  4
jin
1
Sum of squares b/w Varieties  122  02  102  102   64
4
 86  64
.re
Sc  22
Error sum of squares Sd = S  Sa  Sb  Sc

w
= 88  2  4  22
Sd  60
w
w

ANOVA Table:
Sum of
Source of Degrees of ‘F’ ratio
squares Mean sum of squares
variation freedom
om
Sa 2 0.67
B/W rows Sa  2 m-1=4-1=3   0.67  0.067
m-1 3 10
Sb 4 1.33
B/W columns Sb  4 m-1=4-1=3   1.33  0.133
m-1 3 10
Sc 22 7.33
B/W varieties Sc  22 m-1=3   7.33  0.733
m-1 3 10
.c
(m-1)(m-2) Sd
Error Sd  60  10 -
=3  2=6 (m-1)(m-2)
Total S  88 m 2  1  15 - -
ul
Number of degrees of freedom V1  3 ; V2  6
Critical value:
pa
The table value of F for (3, 6) d.f at 5% Los is 4.76
Conclusion:
jin
Since F<4.76, for all the case.
 There is no significant difference for the varieties
2. Analyse the variance in the following Latin squares of fields (in keys) of paddy where A,B,C,D
.re
denote the difference methods of calculation
D122 A121 C123 B122

B124 C123 A122 D125
A120 B119 D120 C121
w
C122 D123 B121 A122

Examine whether the different methods of cultivation have given significantly different fields.
Solution:
w
Re arrange the table in order
A121 A122 A120 A122

w
B122 B124 B119 B121

C123 C123 C121 C122
D122 D125 D120 D123

We take the origin 122 and the table is
Letter Values Vi
total
A -1 0 -2 0 -3
B 0 2 -3 -1 -2
om
C 1 1 -1 0 1
D 0 3 -2 1 2
Calculation of LSD:
Columns 1 2 3 4 Row total

x 2
.c
ij
/ Rows j
1 0 -1 1 0 0 2
2 2 1 0 3 6 14
ul
3 -2 -3 -2 -1 -8 18
4 0 1 -1 0 0 2
Columns 0 -2 -2 2 36
total pa -2
x
i
ij
2 8 12 6 10 36
Here N=16; G=-2

jin
G2 4
Correction factor =   0.25
N 16
.re
G2
Total sum of squares S=  xij2 
i j N
 36  0.25
w
S=35.75
m
Si 2 G 2
Sum of squares b/w rows Sa   
w
i=1 m N
1
 (6)2  (8) 2   0.25
w
 25  0.25
Sa  24.75

m S j2 G2
Sum of squares b/w columns Sb  m
j=1

N
1
 (0)2  (2)2  (2)2  (2)2   0.25
4
om
Sb  2.75
m
Vi 2 G 2
Sum of squares b/w varieties Sc  
i=1 m

N
.c
1
 (3)2  (2) 2  (1)2  (2) 2   0.25
4
 4.5  0.25
ul
Sc  4.25
Error (or) Residual Sd  S-Sa  Sb  Sc

pa
 35.75  24.75  2.75  4.25
Sd  4
jin
LSD Table:
Source of Sum of Degrees of Mean sum of ‘F’ ratio

.re
Sa 24.75
 8.25
B/W rows Sa  24.75 m-1=3 m-1 3  12.31
0.67
 8.25
w
Sb 2.75 0.92
B/W columns Sb  2.75 3   1.37
m-1 3 0.67
 0.92
w
Sc 4.25
 1.42
B/W varieties Sc  4.25 3 m-1 3  2.12
0.67
 1.42
w
Sd
Error (or)
Sd  4.0 6=(m-1)(m-2) (m-1)(m-2)
Residual
 0.67
Total S  35.75 m 1  8
2

Critical value:
The value of ‘F’ for (3,6) d.f at 5% Los is 4.76
Conclusion:
Since F<4.76, we accept the null hypothesis
om
 The difference between the methods of cultivation is not significant.
3. The following data resulted from an experiment to compare three burners A,B, and C,A Latin
squares design was used as the tests were made on 3 engines and were spread over 3 days.
Engine 1 Engine 2 Engine 3
.c
Day 1 A 16 B 17 C 20
Day 2 B16 C 21 A 15
Day 3 C15 A 12 B 13
ul
Test the hypothesis that there is no diff between the burners
Solution:
We take the origin x=15 and the calculation are done as follows
Re arrangement of given table is

pa
A B C
16 17 20
A B C
jin
15 16 21
A B C
12 13 15
.re
Varieties Values Vi
A 1 0 -3 -2
B 2 1 -2 1
C 5 6 0 11
w
Calculation of LSD
Columns/
Rows
C1 C2 C3 Row
total
x
j
ij
2
w
R1 1 2 5 8 30
R2 1 6 0 7 37
w
R3 0 -3 -2 -5 13
Column 2 5 3 10 80
total
x
i
ij
2 2 49 29 80

Here N=9; G=10
G 2 (10) 2
Correction Factor    11.11
N 9
Total sum of squares S=  xij2  C.F
om
j i
 80 11.11
S=68.89
.c
m
Si 2
Sum of squares b/w Rows Sa  
i=1 m
 C.F
ul
 [82  7 2  (5)2 ]  11.11
3
 46 11.11 pa
Sa =34.89
m S j2
Sum of squares b/w columns Sb   m  C.F
j=1
jin
1
 [(2) 2  (5) 2  (3) 2 ]  11.11
3
 1.56
.re
m
Vi 2
Sum of squares b/w varieties Sc  
i=1 m
 C.F
1
w
 [(2)2  12  112 ]  11.11

3
Sc  30.89
w
Error (or) Residual Sd  S-Sa  Sb  Sc

w
 68.89  34.89 1.56  30.89
Sd  1.55

Source of Sum of Degrees of Mean sum of ‘F’ ratio
Sa 34.89
 17.445
om
B/W rows Sa  34.89 m-1=2 m-1 2  22.5
0.775
 17.445
Sb 1.56
 0.78
B/W columns Sb  1.56 m-1=2 m-1 2  1.01
0.775
 0.78
Sc 30.89

.c
15.445
B/W varieties Sc  30.89 m-1=2 m-1 2  19.93
0.775
 15.445
Sd (m-1)(m-2)
ul
Error (or) 1.55
Sd  1.55 (m-1)(m-2) =
Residual 2
pa  0.775
Total S  68.89 m2  1  8
Critical value:
The value of ‘F’ for (2,8) d.f at 5% Los is 4.46

jin
Conclusion:
Since F> the table value for the burners

.re
 There is a significant difference between the burners
and also F> tabulated F for columns the difference b/w the engines is not significant.
Homework:
w
1. Analyse the variance in the following LS:
B C D A
20 17 25 34
w
A D C B
23 21 15 24
D A B C
w
24 26 21 19
C B A D
26 23 27 22

2. Analyse the variance in the following LS:
A C B
8 18 9
C B A
9 18 16
B A C
om
11 10 20
Factorial Experiments
Definition 1:
A factorial experiment in which each of m factors at ‘S’ is called a symmetrical factorial
.c
experiment and is often known as Sm factorial design
Definition 2:
ul
2m - Factorial experiments means a symmetrical factorial experiments where each of the m-factors
is at two levels pa
2 2 -a factorial experiment means a symmetrical experiment where each of the factors is at two levels
Note:
If the numbers of level of the different factors are equal the experiments is called as a symmetrical
jin
factorial experiment.
Uses advantages of factorial experiments:
(i) Factorial designs are widely used in experiments involving several factors where it is
.re
necessary
(ii) F.D allow effects of a factor to be estimated at several levels of the others, giving
conclusions that are valid over a range of experimental conditions
(iii) The F.D are more efficient than one factor at a time experiments.
(iv) In F.D individual factorial effect is estimated with precision, as whole of the experiment is
w
devoted to it.
(v) Factorial designs from the basis of other designs of considerable practical value.
(vi) F.D are widely used in research work. These design are used to apply the results over a wide
range of conditions
w
2 2 -Factorial experiment:
w
A factorial design with two factors, each at two levels is called a 2 2 factorial design

Yates’s notation:
The two factors are denoted by the letters A and B the letters ‘a’ and ‘b’ denote one of the
two levels of each of the corresponding factors and this will be called the second level.
The first level of A and B is generally expressed by the absence of the corresponding letter in
om
the treatment combinations. The four treatment combinations can be enumerated as follows.
Symbols used:
a 0b0 (or)1:Factors A and B both at first level
a1a 0 (or)a:A at second level and B at first level
.c
a0a1 (or) b :A at first level and B at second level
ul
a1a1 (or) ab : A and B both second levels.
Yates’s method of computing factorial effect totals

pa
For the calculation of various factorial effect total for 2 2 -factorial experiments the following table
is need
Treatment Total yield (3) (4) Effect Totals

combination from all
jin
replicates
'1' [1] [1]+[a] [1]+[a]+[b]+[ab] Grand total
a [a] [b]+[ab] [ab]-[b]+[a]-[1] [A]
b [b] [a]-[1] [ab]+[b]-[a]-[1] [B]
.re
ab [ab] [ab]-[b] [ab]-[b]-[a]+[1] [AB]
2 2 -factorial experiment conducted in a CRD

w
Let xij  j observation of i th treatment combinations i=1, 2, 3, 4; j=1,2….(say)

th
i.e., x1  [1]; x2  [a]; x3  [b]; x4  [ab]

w
Where
xi =total of i th treatment combination .

w
G=  xij grand total

i f

n=4r=Total number of observations
G2
TSS=  xij2 -
j i 4r
1. The follwoing table gives the plan and yields of a 22  factorial experiment conducted in CRD
om
Analyse the design and give your comments
(1) a a b
20 28 24 10
ab b ab (1)
.c
23 11 22 17
a b ab (1)
24 15 21 19
Solution:
ul
Arrange the observation as in one-way classification, we proceed as follows
Treatment
Combination
(1) 20
pa 17 19
Total
56
a 28 24 24 76
b 10 11 15 36
ab 23 22 21 66
jin
Total G= 234
G2 2342
Correction Formula =   4563
22  r 4  3
 x
.re
ij
2
 202  17 2  192  282  242  242  102  112  152  232  222  212
j i
 x
j i
ij
2
 4886
w
G2
TSS   xij2 
j i 4r
 4886  4563  323
w
w

The values of SSA, SSB and SSAB are obtained by yate’s method
Treatment Total (3) (4) Divisor Sum of squares

combination (2) (5) (6)
1 [1] [1]+[a] [1]+[a]+[b]+[ab]=[M] - -
a [a] [b]+[ab] [ab]-[b]+[a]-[1]=[A] 4r [A]2 /4r=SSA
om
b [b] [a]-[1] [ab]+[b]-[a]-[1]=[B] 4r [B]2 /4r=SSB
ab [ab] [ab]-[b] [ab]-[b]-[a]+[1]=[AB] 4r [AB]2 /4r=SSAB
SSE = TSS-(SSA+SSB+SSAB)
.c
The analysis of variance table for 2 2 factorial design conducted in CRD
Source of d.f S.S M.S.S F

variation
ul
A 1 SSA MSSA MSSA
MSSE
B 1 pa SSB MSSB MSSB
MSSE
AB 1 SSAB MSSAB MSSAB
MSSE
Error 3(r-1) SSE MSSE -
Total 4r-1 TSS - -
jin
To obtain the sum of squares SSA, SSB, SSAB use yate’s method:
Treatment/ Total (3) (4) Divisor S.S

.re
combination response (5) (6)

(1) 56 56+76=132 132+102=234 4r=12 Grand total
a 76 36+66=102 20+30=50 12 502
 208.33
12
w
b 36 76-56=20 102-132=-30 12 (30) 2

 75
12
ab 66 66-36=30 30-20=10 12 (10) 2
w
 8.33
12
Total 291.66
w

www.rejinpaul.com
www.rejinpaul.com
SSE=TSS-(SSA+SSB+SSAB)
=323-291.66
SSE=31.34
Analysis of variance table:
om
Source of d.f S.S M.S.S F F0.01 (1, 6)
variation
A 1 208.33 208.33 53.15 13.75
B 1 75 75 19.13
AB 1 8.33 8.33 2.09
Error 3(r-1)=6 31.34 3.92
.c
Total 4r-1=11 323
Critical value:
The table value of for (1,6) d.f at 1% Los is 13.75
ul
Conclusion:
Since F> tabulated value of ‘F’ for the main effect A and B, we conclude that the main effects A
pa
and B both are significantly different at 1% Los
jin
.re
w
w
w

3 Solution of Equations and Eigenvalue Problems
om
Solution of algebraic and transcendental equations - Fixed point iteration method - Newton Raphson
method- Solution of linear system of equations - Gauss elimination method - Pivoting - Gauss Jordan
method - Iterative methods of Gauss Jacobi and Gauss Seidel - Matrix Inversion by Gauss Jordan
method - Eigen values of a matrix by Power method.
.c
3.1 Introduction
ul
Solution of equation:
A value satisfies given equation is called solution of that equation.
pa Y=f(x)
x1
jin
-X x2 x3 O x4 x5 X
-Y
.re
Here the curve f (x) with
f (x = x2 ) = f (x = x3 ) = f (x = x4 ) = 0
w
⇒ x2 , x3 and x4 are solutions of f (x).

and f (x) > 0 for x2 < x < x3 and x4 < x < x5
f (x) < 0 for x1 < x < x2 and x3 < x < x4 .
w
3.2 Solution of algebraic and transcendental equations

w
This chapter deals with finding solutions of algebraic and transcendental equations of either of the
forms
f (x) = 0 or f (x) = g(x) (1)

6 MA8452 Statistics and Numerical Methods by K.A. Niranjan Kumar
where we want to solve for the unknown x. An algebraic equation is an equation constructed using
the operations of +, −, ×, ÷, and possibly root taking (radicals). Rational functions and polynomials are
examples of algebraic functions. Transcendental equations in comparison are not algebraic. That is,
they contain non-algebraic functions and possibly their inverses functions. Equations which contain
either trigonometric functions, inverse trigonometric functions, exponential functions, and logarithmic
functions are examples of non-algebraic functions which are called transcendental functions.
Example 1 : (Root of algebraic equation)
132 2 28 147
om
Estimate the solutions of the algebraic equation f (x) = x3 −
x + x+ = 0.
32 32 32
Solution: We use a computer or calculator and plot a graph of the function y = f (x) and obtain the
figure as follows:
.c
ul
132 2 28 147
f (x) = x3 − pa x + x+ =0
32 32 32
One can now estimate the solutions of the given equation by determining where the curve crosses the
x−axis because these are the points where y = 0. Examining the graph in above figure we can place
bounds on our estimates x0 , x1 , x2 of the solutions. One such estimate is given by
jin
−1.0 < x0 < −0.8
1.4 < x1 < 1.6
.re
3.4 < x2 < 3.6
To achieve a better estimate for the roots one can plot three versions of the above graph which have
some appropriate scaling in the neighborhood of the roots.
Finding values for x where f (x) = g(x) can also be approached using graphics. One can plot graphs
w
of the curves y = f (x) and y = g(x) on the same set of axes and then try to estimate where these curves
intersect.
w
Example 2 : (Root of transcendental equation)

132 2 28 147
Estimate the solutions of the transcendental equation x3 − x + x+ = 5 sin x.
32 32 32
Solution: We again employ a computer or calculator and plot graphs of the functions
w
132 2 28 147
y = f (x) = x3 − x + x+ and y = g(x) = 5 sin x
32 32 32
to obtain the following figure

Unit III - Solution of Equations and Eigenvalue Problems 7
om
Graph of y = f (x) and y = g(x)
One can estimate the points where the curve y = f (x) intersects the curve y = g(x). If the curves are
plotted to scale on the same set of axes, then one can place bounds on the estimates of the solution.
One such set of bounds is given by
.c
−1.5 < x3 < −1.0
0.5 < x4 < 1.0
3.0 < x5 < 3.5
ul
By plotting these graphs over a ner scale one can obtain better estimates for the solutions.
Soln. of 1 equation
Fixed point (IM)
pa
Soln. of n equations
Gauss elimination DM(RO)
Matrix Inversion
Gauss Jordan(RO)
Eigen Val. & Vec.
Power method IM(MM)
Newton Raphson (IM) Gauss Jordan DM(RO)
Gauss Jacobi IM(SVS)
jin
Gauss Seidel IM(LVS)
Calculator Usage : Casio fx-991ms
Polynomial: Simultaneous Equations Matrix Entry A: Matrix Entry A&B:
EQN→Deg?(2 or 3) EQN→Unknowns?(2 or 3) Press[MODE] Press[MODE]
.re
Trigonometry: [6:Matrix] [6:Matrix]

Deg.,Rad.,Grads 1:MatA 1:MatA & 2:MatB
−1
Exponential: A : A ∗ B:
Shift ln Shift 4 Shift 4
w
Logarithmic: MATRIX 3 MATRIX 3

log (def. base 10) MATA x−1 MATA * MATB
where IM:Iterative method, DM: Direct method, RO: Row operations, MM: Matrix multiplication.
w
SVS:Step value substitution, LVS:Latest value substitution

π
Deg.:Degree Mode, Rad.:Radian Mode, def.:default 90◦ degrees = radians = 100Grads
2
w
For example:
sin 90◦ ( in degree mode ) = sin (π ÷ 2) ( in radian mode ) = sin (100) ( in Grads mode ) = 1
*** sin (π ÷ 2) , sin π ÷ 2 [∵ sin (π ÷ 2) = 1 & sin π ÷ 2 = 0 ÷ 2 = 0]
Definition of natural logarithm : When ey = x. Then base e logarithm of x is : ln(x) = loge (x) = y
Default base of log is 10 and ln is e. [The e constant or Euler’s number is: e ≈ 2.71828183]
3.3 Fixed point iteration method
Write the given equation as f (x) = 0.

Find x1 , x2 with opposite signs of f (x). Find an relation as x = g(x) such that
g ′ (x1 ) < 1
g ′ (x2 ) < 1
om

 x1 , | f (x1 )| < | f (x2 )| ⇒ f (x1 ) is nearer to zero.



X0 = 
 x2 , | f (x1 )| > | f (x2 )| ⇒ f (x2 ) is nearer to zero.


X1 = g (X0 )
X2 = g (X1 )
.c
..
.
Xn = g (Xn−1 )
Xn+1 = g (Xn )
ul
Note: Stop method if the consecutive values of Xn &Xn+1 are equal upto required place of decimal. i.e.,
Xn = Xn + 1. pa
3.3.1 Part A
1. If g(x) is continuous in [a, b] then under what condition the iterative method x = g(x) has unique
solution in [a, b]. (MJ2010)
jin
′
Solution: |g (x)| < 1 in [a, b] with order of convergence is one.
2. What are the advantages of iterative methods over direct methods for solving a system of linear
equations. (ND2012)
.re
Solution:
1. Iterative methods are suitable for solving linear equations when the number of equations in
a system is very large(more than 100).
2. Iterative methods are very effective concerning computer storage and time requirements.
w
3. One of the advantages of using iterative methods is that they require fewer multiplications
for large systems.
w
4. Iterative methods automatically adjust to errors during study. They can be implemented in
smaller programmes than direct methods.
5. They are fast and simple to use when coefficient matrix is sparse(more zeros).
w
6. Advantageously they have fewer rounds off errors as compared to other direct methods.
7. The direct methods, aim to calculate an exact solution in a finite number of operations.
Whereas iterative methods begins with an initial approximation and reproduce usually
improved approximations in an infinite sequence whose limit is the exact solution.
8. Error is controlled by the number of itertaions and there is no round of error.
9. Iterative methods are self-correcting.
*. Direct methods : Gaussian Elimination, Gauss Jordan Method
Iteration Methods : Fixed-point Iteration, Gauss-Seidel Iteration Method
om
3. What do you mean by the order of convergence of an iterative method for finding the root of the
equation f (x) = 0? (ND2013)
Solution: For Fixed point iteration method, g ′ (x) < 1 in [a, b] , with order of convergence is one.
where f (x) = 0 ⇒ x = g(x)
.c
For Newton-Raphson iteration method, | f (x) f ′′ (x)| < | f ′ (x)|2 , with order of convergence is two.
✎ ☞
Worked Examples
ul
✍ ✌
3.3.2 Part B
pa
Example 3.1. Solve the equation x3 + x2 − 100 = 0 by iteration method.
Solution: Let f (x) = x3 + x2 − 100 = 0
jin
′
f (x = 0) = (0)3 + (0)2 − 100 = −100 < 0 −′ ve

f (x = 1) = (1)3 + (1)2 − 100 = −98 < 0 ′ −′ ve

f (x = 2) = (2)3 + (2)2 − 100 = −88 < 0 ′ −′ ve

f (x = 3) = (3)3 + (3)2 − 100 = −64 < 0 ′ −′ ve

.re
f (x = 4) = (4)3 + (4)2 − 100 = −20 < 0 (′ −′ ve)
f (x = 5) = (5)3 + (5)2 − 100 = 50 > 0 (′ +′ ve)

w
∴ f (x) has a root between 4 and 5.

i.e, f (x) has a positive root.
This equation f (x) = x3 + x2 − 100 = 0 can be written as
w
x3 + x2 − 100 = 0
i.e., x2 (x + 1) = 100
w
10
i.e., x = √

= g (x) , say

x+1
−1
10
∴ g (x) = √ = 10 + 1) 2
(x
x+1
d
Therefore, g ′ (x) =

g (x)
dx
 −1  ! −1 −3
d  −1 −5
  −1
= 10 (x + 1) 2  = 10 (x + 1) 2 = −5 (x + 1) 2 =
dx   2 3
(x + 1) 2
−5 5
∴ g ′ (x) = =
3 3
(x + 1) 2 (x + 1) 2
om
5 5 5 1 1
At x = 4, g ′ (x = 4) = = = √ = √ = = 0.57735 < 1
(4 + 1)3/2 (5)3/2 5 5 5 2.23606
5 5
At x = 5, g ′ (x = 5) = = = 0.34020 < 1
(5 + 1)3/2 (6)3/2
.c
|g ′ (x)| < 1 in (4, 5)
So the fixed point iteration method can be applied. Let x0 = 4

| f (x = 4)| < | f (x = 5)|
Use xn+1 = g (xn )
ul
10 10 10
x1 = g (x0 ) = √ = √ = √ = 4.47214 (round off at 5th decimal)
x0 + 1 4+1 5
10
x2 = g (x1 ) = √ = 4.27486pa
4.47214 + 1
10
x3 = g (x2 ) = √ = 4.35406
4.27486 + 1
10
x4 = g (x3 ) = √ = 4.32174
4.35406 + 1
jin
10
x5 = g (x4 ) = √ = 4.33484
4.32174 + 1
10
x6 = g (x5 ) = √ = 4.32952
4.33484 + 1
10
x7 = g (x6 ) = √ = 4.33168
.re
4.32952 + 1
10
x8 = g (x7 ) = √ = 4.33080
4.33168 + 1
10
x9 = g (x8 ) = √ = 4.33116
4.33080 + 1
10
w
x10 = g (x9 ) = √ = 4.33101

4.33116 + 1
10
x11 = g (x10 ) = √ = 4.33107
4.33101 + 1
w
10
x12 = g (x11 ) = √ = 4.33105
4.33107 + 1
10
w
x13 = g (x12 ) = √ = 4.33105

4.33105 + 1
These values are same at fifth decimal, i.e., x12 = x13 . So stop the method.
∴ The root of the given equation x3 + x2 − 100 = 0 is 4.33105

Note: In above problem, the root is 4.33105(at 5th decimal) or 4.3311(at 4th decimal) or 4.331(at 3rd
decimal).
Example 3.2. Solve the equation x3 + 2x2 + 10x = 20 by fixed point iteration method.
Solution: [x13 = x14 = 1.36880]
Example 3.3. Solve the equation x3 + x + 1 = 0 by fixed point iteration method.
om
h i
Solution: g (x) = (−1 − x)1/3 ; x23 = x24 = −0.682
Example 3.4. Solve the equation 3x − cos x − 2 = 0 by fixed point iteration method correct to
3 decimal places.
Solution: Let f (x) = 3x − cos x − 2 = 0
.c
f (x = 0) = 3 (0) − cos (0) − 2 = 0 − 1 − 2 = −3 < 0 (′ −′ ve)
f (x = 1) = 3 (1) − cos (1) − 2 = 0.459697 > 0 (′ +′ ve)
ul
i.e, f (x) has a positive root. pa
This equation f (x) = 3x − cos x − 2 = 0 can be written as
3x − cos x − 2 = 0
i.e., 3x = 2 + cos x
jin
2 + cos x
i.e., x = = g (x) , say

3
2 + cos x
∴ g (x) =
3
.re
d
Therefore, g ′ (x) =

g (x)
dx " # ! !
d 2 + cos x 1 d 1 − sin x
= = [2 + cos x] = [0 − sin x] =
dx 3 3 dx 3 3
w
− sin x sin x
∴ g ′ (x) = =
3 3
w
sin (0)
At x = 0, g ′ (x = 0) = =0 < 1
3
sin (1)
At x = 1, g ′ (x = 1) = = 0.2804 < 1
w
|g ′ (x)| < 1 in (0, 1)
So the fixed point iteration method can be applied. Let x0 = 0

| f (x = 1)| < | f (x = 0)|
Use xn+1 = g (xn ) Now, put n = 0 in xn+1 = g (xn ), we get

2 + cos (x0 ) 2 + cos (0) 2 + 1
x1 = g (x0 ) = = = =1
3 3 3
2 + cos (x1 ) 2 + cos (1)
x2 = g (x1 ) = = = 0.84677
3 3
2 + cos (x2 ) 2 + cos (0.84677)
x3 = g (x2 ) = = = 0.88747
3 3
2 + cos (x3 ) 2 + cos (0.84747)
x4 = g (x3 ) = = = 0.87713
3 3
om
2 + cos (x4 ) 2 + cos (0.87713)
x5 = g (x4 ) = = = 0.87979
3 3
2 + cos (x5 ) 2 + cos (0.87979)
x6 = g (x5 ) = = = 0.87910
3 3
These values are same at third decimal, i.e., x5 = x6 . So stop the method.
.c
∴ The root of the given equation 3x − cos x − 2 = 0 is 0.879
Example 3.5. Solve the equation cos x = 3x − 1 by iteration method upto 5 decimal.
(Or) Find a positive root of the equation cosx − 3x + 1 = 0 by using iteration method. (AM13)
ul
Solution: [x = 0.60710]
Example 3.6. Solve the equation 2 sin x = x by fixed point iteration method.
Solution:

pa [x = 1.89549]
Hint: |g (x = 1)| > 1, so use |g (x = 1.5)| < 1& |g ′ (x = 2)| < 1
′ ′
Example 3.7. Solve the equation e x − 3x = 0 by iteration method correct to 3 decimal.(MJ12)

Solution: Let f (x) = e x − 3x = 0
jin
f (x = 0) = e0 − 3 (0) = 1 > 0 (′ +′ ve)
f (x = 1) = e1 − 3 (1) = −0.2817 < 0 (′ −′ ve)

.re

This equation f (x) = e x − 3x = 0 can be written as
3x = e x
ex
i.e., x = = g (x) , say
3
w
ex
∴ g (x) =
3
w
d ex ex
" #
′ d
Therefore, g (x) =

g (x) = =
dx dx 3 3
x x
e e
∴ g ′ (x) = =
w
3 3
e0 1
At x = 0, g ′ (x = 0) = = = 0.333 < 1
3 3
e1
At x = 1, g ′ (x = 1) = = 0.906 < 1
3
|g ′ (x)| < 1 in (0, 1)
So the fixed point iteration method can be applied. Let x0 = 0 use xn+1 = g (xn )
Now, put n = 0 in xn+1 = g (xn ), we get
e x0 e0
x1 = g (x0 ) = = = 0.33333 e x7 e0.60727
3 3 x8 = = = 0.61180
3 3
e x1 e0.33333
x2 = = = 0.46520 e x8 e0.61180
3 3 x9 = = = 0.61458
3 3
e x2 e0.46520
e x9 e0.61458
om
x3 = = = 0.53078
3 3 x10 = = = 0.61629
x3 0.53078 3 3
e e
x4 = = = 0.56675 e x10 e0.61629
3 3 x11 = = = 0.61735
3 3
e x4 e0.56675
x5 = = = 0.58751 e x11 e0.61735
3 3 x12 = g (x11 ) = = = 0.61800
e x5 e0.58751 3 3
x6 = = = 0.59983
3 3 e x12 e0.61800
.c
e x6 e0.59983 x13 = g (x12 ) = = = 0.61840
x7 = = = 0.60727 3 3
3 3
These values are same at third decimal, i.e., x12 = x13 . So stop the method.
ul
∴ The root of the given equation e x − 3x = 0 is 0.618
Example 3.8. Solve the equation 2x − log x = 7 by iteration method correct to 5 decimal.
Solution: Let f (x) = 2x − log x = 7 = 0
pa
f (x = 0) = 2 (0) − log10 (0) − 7 = Undefined

′
−′ ve

f (x = 1) = 2 (1) − log10 (1) − 7 = −5 < 0
f (x = 2) = 2 (2) − log10 (2) − 7 = −3 < 0 ′ −′ ve

jin
f (x = 3) = 2 (3) − log10 (3) − 7 = −1 < 0 ′ −′ ve

′
+′ ve

f (x = 4) = 2 (4) − log10 (4) − 7 = 0.39 > 0

.re

This equation f (x) = 2x − log x − 7 = 0 can be written as
2x = 7 + log x
7 + log x
i.e.,x = = g (x) , say
2
w
7 + log x
∴ g (x) =
2
w
" # 1
d d 7 + log x 1
Therefore, g ′ (x) = = x =

g (x) =
dx dx 2 2 2x
1 1
∴ g ′ (x) = =
w
2x 2x
1 1
At x = 3, g ′ (x = 3) = = = 0.166 < 1
2 (3) 6
1 1
At x = 4, g ′ (x = 4) = = = 0.125 < 1
2 (4) 8
|g ′ (x)| < 1 in (3, 4)
So the fixed point iteration method can be applied. Let x0 = 4, use xn+1 = g (xn )
Now, put n = 0 in xn+1 = g (xn ), we get
7 + log (x0 ) 7 + log (4)

x1 = g (x0 ) = = = 3.80103
2 2
7 + log (x1 ) 7 + log (3.80103)
x2 = g (x1 ) = = = 3.78995
2 2
om
7 + log (x2 ) 7 + log (3.78995)
x3 = g (x2 ) = = = 3.78932
2 2
7 + log (x3 ) 7 + log (3.78932)
x4 = g (x3 ) = = = 3.78928
2 2
7 + log (x4 ) 7 + log (3.78928)
x5 = g (x4 ) = = = 3.78928
2 2
These values are same at fifth decimal, i.e., x4 = x5 . So stop the method.
.c
∴ The root of the given equation 2x − log x = 7 is 3.78928
ul
Example 3.9. Solve the equation 3x − log10 x = 6 by fixed point iteration method.
Solution: [x = 2.1080]
3.3.3 Anna University Questions

pa
1. Solve e x − 3x = 0 by the method of fixed point iteration. (AM12)
Solution : [x = 0.618]
jin
2. Find a positive root of the equation cosx − 3x + 1 = 0 by using iteration method. (AM13)
Solution : [x = 0.60710]
.re
3.4 Newton Raphson method
(Newton’s method or method of tangents)

f (xn )
Newton-Raphson method formula is xn+1 = xn − = φ (xn ) , where n = 0, 1, 2, · · ·
f ′ (xn )
w
3.4.1 Part A
1. State the order of convergence and convergence condition for Newton-Raphson method.
w
(Or) What is the criterion for the convergence in Newton’s method? (MJ2011)
(Or) State the order of convergence and the criterion for the convergence in Newton’s method.
(AM2012)
w
(Or) Write down the condition for convergence of Newton-Raphson method for f (x) = 0. (ND14)
Solution: Order of convergence is 2. Convergence condition is
| f (x) f ′′
(x)| < | f ′ (x)|2
√
2. Find an iterative formula to find N, where N is positive number.
√
Solution: Let x = N
⇒ x = N ⇒ x2 − N = 0
2
Let f (x) = x2 − N ⇒ f ′ (x) = 2x
By Newton’s formula
xn2 − N 2xn2 − xn2 + N xn2 + N
om
f (xn )
xn+1 = xn − = xn − = =
f ′ (xn ) 2xn 2xn 2xn
3. Derive Newton’s algorithm for finding the pth root of a number N.

√p
Solution: Let x = N 1/p = N ⇒ x p − N = 0
Let f (x) = x p − N ⇒ f ′ (x) = px p−1
By Newton’s formula
.c
xnp − N px p − xnp + N (p − 1) xnp + N
xn+1 = xn − p−1
= n p−1 =
px px px p−1
ul
4. Establish an iteration formula to find the reciprocal of a positive number N by Newton-Raphson
method. (Or)
Find an iterative formula to find the reciprocal of a given number N(N , 0). (AM2013)
Solution : Let x =
i.e., N =
1
N
1
pa
x
1
Let f (x) = − N
x
jin
′ 1
f (x) = − 2
x
 1
 
− N

f (xn )  xn
 
We know that by Newton’s iterative formula xn+1 = xn − ′ = xn −  
.re
f (xn ) 1 
− 2 

xn
" #
1
= xn + xn2 − N = xn + xn − N xn2 = 2xn − N xn2
xn
xn+1 = xn [2 − N xn ]
w
which is the required iterative formula for the reciprocal of a given number N.
5. Locate the negative root of x3 − 2x + 5 = 0, approximately.

w
Solution: Let f (x) = x3 − 2x + 5 = 0

f (−1) = −1 + 2 + 5 = 6 (+ve)
w
f (−2) = −8 + 4 + 5 = 1 (+ve)
f (−3) = −27 + 6 + 5 = −16 (−ve)
∴ Root lies between −2 and −3.

and root is closer to −2 since | f (−2)| < | f (−3)|
√
6. Evaluate 12 applying Newton formula.
√
Solution: Let x = 12
⇒ x2 = 12 ⇒ x2 − 12 = 0
Let f (x) = x2 − 12 ⇒ f ′ (x) = 2x

f (3) = 9 − 12 = −3 (−ve)
om
f (4) = 16 − 12 = 4 (+ve)
∴ Root lies between 3 and 4.

and root is closer to 3 since | f (3)| < | f (4)|
Take x0 = 3
Newton’s formula is
.c
f (xn )
xn+1 = xn −
f ′ (xn )
f (x0 ) f (3) (3)2 − 12
ul
when n = 0, x1 = x0 − =3− = 3 − = 3.5
f ′ (x0 ) f ′ (3) 2 (3)
f (x1 ) f (3.5) (3.5)2 − 12
when n = 1, x2 = x1 − = 3.5 − ′ = 3.5 − = 3.46429
f ′ (x1 ) f (3.5) pa 2 (3.5)
f (x2 ) f (3.46429) (3.46429)2 − 12
when n = 2, x3 = x2 − = 3.46429 − ′ = 3.46429 − = 3.46410
f ′ (x2 ) f (3.46429) 2 (3.46429)
f (x3 ) f (3.46410) (3.46410)2 − 12
when n = 3, x4 = x3 − = 3.46410 − ′ = 3.46410 − = 3.46410
f ′ (x3 ) f (3.46410) 2 (3.46410)
Here x3 = x4 = 3.46410
jin
∴ The root is 3.46410
7. What is Newton’s algorithm to solve the equation x2 = 12? (ND2010)

2 ′
Solution : Let f (x) = x − 12 ⇒ f (x) = 2x
.re
We know that by
f (xn )
Newton Raphson iteration formula xn+1 = xn −
f ′ (xn )
x2 − 12
= xn − n
2xn
w
2xn2 − xn2 + 12
=
2xn
xn2 + 12
xn+1 =
2xn
w
8. Using Newton’s method, find the root between 0 and 1 of x3 = 6x − 4. (ND2011)

f (xn )
Solution: Newton-Raphson method formula is xn+1 = xn − ′ , where n = 0, 1, 2, · · · (1)
f (xn )
w
Let f (x) = x3 − 6x + 4 ⇒ f ′ (x) = 3x2 − 6

f (x = 0) = 03 − 6(0) + 4 = 4 > 0 ′ +′ ve

′
f (x = 1) = 13 − 6(1) + 4 = −1 < 0 −′ ve

∴ A root lies between 0 and 1.

But | f (0)| > | f (1)|
∴ The root is nearer to 1.
Take x0 = 1
Now, put n = 0 in (1), we get
h i
f (x0 ) (x0 )3 − 6 (x0 ) + 4 13 − 6(1) + 4
x1 = x0 − ′ = x0 − h i =1− h i
f (x0 ) 3 (x0 )2 − 6 3 (1)2 − 6
om
(−1) 1
=1− = 1 − = 0.666 = 0.67 [Corrected to 2 dec. places]
−3 3 h i
f (x1 ) (x1 )3 − 6 (x1 ) + 4 (0.67)3 − 6(0.67) + 4
x2 = x1 − = x1 − = 0.67 −
f ′ (x1 )
h i h i
3 (x1 )2 − 6 3 (0.67)2 − 6
0.28 0.28
= 0.67 − = 0.67 + = 0.73
−4.65 4.68
.c
h i
f (x2 ) (x2 )3 − 6 (x2 ) + 4 (0.73)3 − 6(0.73) + 4
x3 = x2 − ′ = x2 − h i = 0.73 − h i
f (x2 ) 3 (x2 )2 − 6 3 (0.73)2 − 6
0.009 0.009
ul
= 0.73 − = 0.73 + = 0.7320 = 0.73 [Corrected to 2 dec. places]
−4.4013 4.4013
Here x2 = x3 = 0.73
∴ A root of given equation is 0.73 corrected to two decimal places.
9. Evaluate
√
pa
15 using Newton-Raphson’s formula.
√
(AM2014)
Solution : W.K.T. The Newton Raphson iterative formula for N is
" #
1 N
jin
xn+1 = xn +
2 xn
√
To find 15
√
W.K.T. 15 is nearer to 4
.re
∴ Let x0 = 4
" # " #
1 N 1 15
x1 = x0 + = 4+ = 3.87500
2 x0 2 4
" # " #
1 N 1 15
x2 = x1 + = 3.875 + = 3.87298
2 x1 2 3.875
w
" # " #
1 N 1 15
x3 = x2 + = 3.87298 + = 3.87298
2 x2 2 3.87298
Here x2 = x3 = 3.87298
w
∴ The root is 3.87298

w
3.4.2 Part B
Example 3.10. Using Newton-Raphson method, find a ‘ + ’ve root correct to 3 decimal
places for the equation x3 − x − 2 = 0 .
f (xn )
Solution: Newton-Raphson method formula is xn+1 = xn − , where n = 0, 1, 2, · · ·
f ′ (xn )
Let f (x) = x3 − x − 2
⇒ f ′ (x) = 3x2 − 1
f (x = 0) = 03 − 0 − 2 = −2 < 0 ’ − ’ve

f (x = 1) = 13 − 1 − 2 = −2 < 0 ’ − ’ve

f (x = 2) = 23 − 2 − 2 = 4 > 0 ′ +′ ve

om
Take x0 = 1
f (xn )
Now, put n = 0 in xn+1 = xn − f ′ (xn ) , we get
h i
f (x0 ) (x0 )3 − x0 − 2 13 − 1 − 2
x1 = x0 − ′ = x0 − h i =1− h i =2
f (x0 ) 3 (x0 )2 − 1 3 (1)2 − 1
.c
h i
f (x1 ) (x1 )3 − x1 − 2 23 − 2 − 2
x2 = x1 − ′ = x1 − h i =2− h i = 1.63636
f (x1 ) 3 (x1 )2 − 1 3 (2)2 − 1
f (x2 ) (x2 )3 − x2 − 2 1.636363 − 1.63636 − 2
ul
x3 = x2 − ′
= x2 − h i = 1.63636 − h i = 1.53039
f (x2 ) 3 (x2 )2 − 1 3 (1.63636)2 − 1
f (x3 ) (x3 )3 − x3 − 2 1.530393 − 1.53039 − 2
x4 = x3 − = x3 − = 1.53039 − = 1.52144
f ′ (x3 )
h i h i
3 (x3 )2 − 1 3 (1.53039)2 − 1
x5 = x4 −
f (x4 )
′
f (x4 )
= x4 −
(x4 )3 − x4 − 2
h
2
3 (x4 ) − 1
i = 1.52144 − h
pa
1.521443 − 1.52144 − 2
3 (1.52144)2 − 1
i = 1.52138
f (x5 ) (x5 )3 − x5 − 2 1.521383 − 1.52138 − 2

x6 = x5 − ′
= x5 − h i = 1.52138 − h i = 1.52138
f (x5 ) 3 (x5 )2 − 1 3 (1.52138)2 − 1
jin
x5 = x6 = 1.52138 upto 5 decimal places.
∴ A root of given equation is 1.521
Example 3.11. Find the real root of x3 − 3x + 1 = 0 lying between 1 and 2 by Newton Raphson
.re
method.
Solution: Let f (x) = x3 − 3x + 1

f (1) = 1 − 3 + 1 = −1 < 0
f (2) = 8 − 6 + 1 = 3 > 0
w

Since | f (1)| < | f (2)|, the root is nearer to 1.
Let x0 = 1.
w
The Newton Raphson formula is

f (xn )
xn+1 = xn −
f ′ (xn )
w
f (x) = x3 − 3x + 1 ⇒ f ′ (x) = 3x2 − 3
Since f ′ (1) = 3 − 30, Newton’s approximation formula cannot be applied for the initial approximation
at x = 1.
Let us take x0 = 1.5 Now,
f (x0 ) f (x1 )
x1 = x0 − x2 = x1 −
f ′ (x0 ) f ′ (x1 )
f (1.5) f (1.5333)
= 1.5 − ′ = 1.5333 − ′
f (1.5) f (1.5333)
(1.5333)3 − 3(1.5333) + 1
" #
(1.5)3 − 3(1.5) + 1
" # " #
−0.125 = 1.5333 −
= 1.5 − = 1.5 − 3(1.5333)2 − 3
3(1.5)2 − 3 3.75 " #
0.0049
= 1.5 + 0.0333 = 1.5333 − = 1.5333 − 0.0012
4.0530
x1 = 1.5333 x2 = 1.5321
om
f (x2 )
x3 = x2 − ′
f (x2 ) Here x2 = 1.5321 and
f (1.5321)
= 1.5321 − ′
f (1.5321) x3 = 1.5321.
(1.5321)3 − 3(1.5321) + 1
" #
= 1.5321 − ∴The required root,
3(1.5321)2 − 3
correct to four places of decimal is
" #
0.00004
.c
= 1.5321 − = 1.5321 − 0.00001
4.042 1.5321.
x3 = 1.5321
ul
Example 3.12. Solve the equation x3 − 5x + 3 = 0 by Newton-Raphson method.
Solution: [x3 = x4 = 0.65662]
Example 3.13. Solve the equation x4 − x − 9 = 0 by Newton-Raphson method.

pa
Solution: [x6 = x7 = 1.813] upto 3rd decimal.
Example 3.14. Find a negative root of x3 − 2x + 5 = 0 by Newton Raphson method.

Solution: (Home work)
jin

Hint: f (−2) > 0, f (−3) < 0
∴ a root lies between − 2 and − 3
Ans : − 2.095

.re
[Or]
h
Given: x3 − 2x + 5 = 0
Replace x by − x3 + 2x + 5 = 0
i.e., x3 − 2x − 5 = 0
w
Let f (x) = x3 − 2x − 5 = 0
Now, we find the positive root of f (x) = x3 − 2x − 5
w
f (2) = −1 and f (3) = 16

∴ a root lies between 2 and 3
w
The root of f (x) = x3 − 2x − 5 is 2.095

i
∴ The root of x3 − 2x + 5 is − 2.095
Example 3.15. Using Newton-Raphson method, find a ‘ + ’ve root correct to 5 decimal places
for the equation x tan x = 1.28
Solution: Newton-Raphson method formula is
f (xn )
xn+1 = xn − , where n = 0, 1, 2, · · ·
f ′ (xn )
Let f (x) = x tan x − 1.28 ⇒ f ′ (x) = x sec2 x + tan x
′
− ′ ve

f (x = 0) = (0) tan (0) − 1.28 = −1.28 < 0
′
− ′ ve

f (x = 0.5) = (0.5) tan (0.5) − 1.28 = −1.006 < 0
om
f (x = 1) = (1) tan (1) − 1.28 = 0.2774 > 0 ′ +′ ve

∴ A root lies between 0.5 and 1.

Take x0 = 0.5
f (xn )
Now, put n = 0 in xn+1 = xn − , we get
f ′ (xn )
.c
f (x0 ) x0 tan x0 − 1.28 (0.5) tan (0.5) − 1.28
x1 = x0 − = x0 − = 0.5 − = 1.34218
f ′ (x0 ) 2
x0 sec x0 + tan x0 (0.5) sec2 (0.5) + tan (0.5)
f (x1 ) (1.34218) tan (1.34218) − 1.28
x2 = x1 − = 1.34218 − = 1.19469
ul
f ′ (x1 ) (1.34218) sec2 (1.34218) + tan (1.34218)
f (x2 ) (1.19469) tan (1.19469) − 1.28
x3 = x2 − = 1.19469 − = 1.04143
f ′ (x2 ) (1.19469) sec2 (1.19469) + tan (1.19469)
f (x3 ) (1.04143) tan (1.04143) − 1.28
pa
x4 = x3 − = 1.04143 − = 0.95512
f ′ (x3 ) (1.04143) sec2 (1.04143) + tan (1.04143)
f (x4 ) (0.95512) tan (0.95512) − 1.28
x5 = x4 − = 0.95512 − = 0.93871
f ′ (x4 ) ( 0.95512) sec2 (0.95512) + tan (0.95512)
f (x5 ) (0.93871) tan (0.93871) − 1.28
x6 = x5 − = 0.93871 − = 0.93826
f ′ (x5 ) (0.93871) sec2 (0.93871) + tan (0.93871)
jin
f (x6 ) (0.93826) tan (0.93826) − 1.28
x7 = x6 − = 0.93826 − = 0.93826
f ′ (x6 ) (0.93826) sec2 (0.93826) + tan (0.93826)
x6 = x7 = 0.93826 upto 5 decimal places.

.re
Example 3.16. Solve the equation x sin x + cos x = 0 by Newton-Raphson method.

Solution: [x4 = x5 = 2.79839]
Example 3.17. Using Newton-Raphson method, find a root correct to 5 decimal places for
w
the equation e x − 4x = 0
f (xn )
Solution: Newton-Raphson method formula is xn+1 = xn −
f ′ (xn )
Let f (x) = e x − 4x ⇒ f ′ (x) = e x − 4
w
′
f (x = 0) = e0 − 4 (0) = 1 > 0 +′ ve

f (x = 1) = e1 − 4 (1) = −1.28272 < 0 ’ − ’ve

w

Here | f (1)| = 1.28172 > | f (0)| = 1
∴ Root is nearer to 0.
Take x0 = 0
f (xn )
Now, put n = 0 in xn+1 = xn − , we get
f ′ (xn )
f (x0 ) e x0 − 4x0 e0 − 4 (0)
x1 = x0 − = x0 − =0− = 0.33333
f ′ (x0 ) e x0 − 4 e0 − 4
f (x1 ) e x1 − 4x1 e0.33333 − 4 (0.33333)
x2 = x1 − ′ = x1 − x = 0.33333 − = 0.35725
f (x1 ) e 1 −4 e0.33333 − 4
f (x2 ) e x2 − 4x2 e0.35725 − 4 (0.35725)
x3 = x2 − ′ = x2 − x = 0.35725 − = 0.35740
f (x2 ) e 2 −4 e0.35725 − 4
om
f (x3 ) e x3 − 4x3 e0.35740 − 4 (0.35740)
x4 = x3 − ′ = x3 − x = 0.35740 − = 0.35740
f (x3 ) e 3 −4 e0.35740 − 4
Example 3.18. Find the real root of xe x − 2 = 0 correct to three places of decimals using
Newton Raphson method.
.c
Solution: Let f (x) = xe x − 2
f (0) = −2 < 0
f (1) = e − 2 = 0.7183 > 0
ul
Since f (0) < 0 and f (1) > 0.
∴ A root lies between 0 and 1. pa
Since | f (1)| < | f (0)|, the root is nearer to 1.
Let x0 = 1.
The Newton Raphson formula is
f (xn )
xn+1 = xn −
f ′ (xn )
jin
f (x) = xe x − 2 ⇒ f ′ (x) = xe x − e x
Now,
f (x0 )
x1 = x0 −
.re
f ′ (x0 ) f (x1 ) f (x2 )

x2 = x1 − x3 = x2 −
f (1) f ′ (x1 ) f ′ (x2 )
=1− ′
f (1) (0.8679)e0.8679 − 2 (0.8528)e0.8528 − 2
" # = 0.8679 − = 0.8528 −
e−2 (0.8679)e0.8679 + e0.8679 (0.8528)e0.8528 + e0.8528
=1− " # " #
e+e 0.0673 0.0008
" # = 0.8679 − = 0.8528 −
0.7183 4.4492 4.3471
w
=1−
5.4366 = 0.8679 − 0.0151 = 0.8528 − 0.0002
= 1 − 01321
x2 = 0.8528 x3 = 0.8526
x1 = 0.8679
w
Here x2 = 0.853 and x3 = 0.853.

∴ The required root, correct to three places of decimal is 0.853.
w
Example 3.19. Solve the equation cos x = xe x , by x0 = 0.5 by Newton-Raphson method.

Solution: [x2 = x3 = 0.51776]
Example 3.20. Solve the equation xe−2x = 0.5 sin x by Newton-Raphson method.

x2 = x3 = 3.12962 upto 5th decimal

Solution:
Example 3.21. Solve the equation 2x − log10 x = 7 by Newton-Raphson method.

Solution: x5 = x6 = 3.78928upto 5th decimal
1
Example 3.22. Find the Newton-Raphson formula to find the value of where N is a real
N
1
number, hence evaluate correct to 4 decimal places.
26
1
Solution: Let x =
N
1 1
⇒N = ⇒ −N =0
om
x x
1 1
Let f (x) = − N ⇒ f ′ (x) = − 2
x x
By Newton’s formula ,
1
−N !
xn 1
xn+1 = xn − = xn + − N xn2 = xn (2 − N xn )
1 xn
− 2
xn
.c
1
This is the Newton’s-Raphson formula to find the value of .
N
1
To find
26
ul
Put N = 26 and n = 0 in xn+1 = xn (2 − N xn ), we get
x1 = x0 (2 − 26x0 )
pa (1)
!
1 1 1
Take = 0.04 as x0 ∵ is nearer to
25 25 26
∴ (1) ⇒ x1 = x0 (2 − 26x0 ) = 0.04 [2 − 26 (0.04)] = 0.0384

jin
x2 = x1 (2 − 26x1 ) = 0.0384 [2 − 26 (0.0384)] = 0.03846
x3 = x2 (2 − 26x2 ) = 0.03846 [2 − 26 (0.03846)] = 0.03846
∴ A root of given equation is 0.03846 or 0.0385

.re
√
Example 3.23. Find the Newton-Raphson formula to find the value of N where N is a real
√
number, hence evaluate 142 correct to 5 decimal places.
Solution: [x2 = x3 = 11.91638]
1
w
Example 3.24. Find the Newton-Raphson formula to find the value of N p where N is a real
√3
number, hence evaluate 17 correct to 5 decimal places.
Solution: [x4 = x5 = 2.57128]
w

w
1. Solve for a positive root of the equation x4 − x − 10 = 0 using Newton - Raphson method.(MJ2010)
Solution : [x2 = x3 = 1.856]
2. Find the approximate root of xe x = 3 by Newton’s method correcot to 3 decimal places. (MJ2011)
Solution : [x2 = x3 = 1.050]
3. Find the Newton’s iterative formula to calculate the reciprocal of N and hence find the value of
1
. (ND2012)
23
Solution : [x0 = x1 = 0.0435]
4. Using Newton’s method, find the real root of x log10 x = 1.2 correct to five decimal places.(ND2013)
Solution : [x3 = x4 = 2.740646096 ≡ 2.74065]
5. Find by Newton-Raphson method a positive root of the equation 3x − cos x − 1 = 0. (ND14)
om
Solution : [x = 0.607102]
1
6. Find Newton’s iterative formula for the reciprocal of a number N and hence find the value of ,
23
correct to five decimal places.
Solution : [x = 0.04347]
7. Prove the quadratic convergence of Newton’s-Raphson method. Find a positive root of f (x) =
.c
x3 − 5x + 3 = 0, using this method.
Solution : [x = 0.6566]
ul
8. Use Newton’s method to find the real root of 3x − cos x − 1 = 0.
Solution : [x = 0.607102]
9. Obtain the positive root of 2x3 − 3x − 6 = 0 that lies between 1 and 2 by using Newton’s-Raphson
method.
Solution :
pa [1.7838]
Solution of linear system of equations

jin
Let us consider a system of n linear algebraic equations in n unknowns
a11 x1 + a12 x2 + · · · + a1n xn = b1

a21 x1 + a22 x2 + · · · + a2n xn = b2
.re
..
.
an1 x1 + an2 x2 + · · · + ann xn = bn
where the coefficients ai j and the constants bi are real and known. This system of equations in matrix
form may be written as
w
AX = b
w

where A = ai j
n×n
X = (x1 , x2 , · · · , xn )T
w
and b = (b1 , b2 , · · · , bn )T .
A is called the coefficient matrix.

We are interested in finding the values xi , i = 1, 2, · · · , n if they exist, satisfying Equation.
Methods to solve linear system of equations:
1. Gaussian elimination method
2. Pivoting-Gauss Jordan method
3.5 Gauss elimination method
om
Step 1: Write the augmented matrix for the given system of simultaneous equations
 
 a11 a12 a13 b1 
 
 a21 a22 a23 b2 

 
a31 a32 b33 b3
.c
Step 2: Using elementary row operations reduce the given matrix into an upper-triangular matrix
say
 
 c11 c12 c13 d1 
 
ul
 0 c22 c23 d2 

 
0 0 c33 d3
Step 3: By back substitution we get the values for unknowns.

pa
3.5.1 Part A
1. By Gauss elimination method solve x + y = 2, 2x + 3y = 5. (MJ2011)

jin
Solution : Given x + y = 2
2x + 3y = 5
The given system can be written as

.re
    
1 1  x 2

    =  
2 3 y 5
AX = B
w
 
1 1 2
∴ The augmented matrix is [A|B] =  
2 3 5
 
w
1 1 2
∼   (1)
0 1 1 R2 → R2 − 2R1
Now From (1), by back substitution we get

w
0x + y = 1 ⇒ y = 1
x+y=2⇒ x+1=2⇒ x=1
∴ x = 1, y = 1.
2. Solve the equations x + 2y = 1 and 3x − 2y = 7 by Gauss-Elimination method. " (ND2013) #

1
Solution : x = 2, y = − .
2
3. Using Gauss elimination method solve : 5x + 4y = 15, 3x + 7y = 12. (AM2014)
Solution : Given 5x + 4y = 15
3x + 7y = 12
om
The given system can be written as
    
5 4  x 15

    =  
3 7 y 12
AX = B
.c
 
5 4 15
∴ The augmented matrix is [A|B] =   

3 7 12
 
ul

 5 4 15
∼  3 3 3 3

3 − (5) 7 − (4) 12 − (15) R2 → R2 − (R1 )

5 5 5 5
 
pa 5 4 15 
(1)

∼  23 15 

0
5 5
Now From (1), by back substitution we get
23 15 15
0x + y= ⇒ y=
jin
5 5 23
285 57
5x + 4y = 15 ⇒ 5x = ⇒ x=
23 23
57 15
∴x= ,y = .
.re
23 23
3.5.2 Part B
Example 3.25. Solve the system of equations by Gauss elimination method

w
−3x1 + 2x2 − 3x3 = −6

x1 − x2 + x3 = 1
w
2x1 − 5x2 + 4x3 = 5.
Solution: The given system of equations can be written as

w
x1 − x2 + x3 = 1
−3x1 + 2x2 − 3x3 = −6
2x1 − 5x2 + 4x3 = 5
Given system of the form AX = B

Gauss elimination method:
The augmented matrix form is
 
 1 −1 1 1 
 
[A|B] = −3 2 −3 −6
 
2 −5 4 5
om
 
 1
 −1 1 1 

∼ −3 + 3 2 − 3 −3 + 3 −6 + 3 R2 → R2 + 3R1

 
2 − 2 −5 + 2 4 − 2 5 − 2 R3 → R3 − 2R1
 
1 −1 1 1 
 
∼ 0 −1 0 −3
 
.c
0 −3 2 3
 
1 −1 1 1 
 
∼ 0 −1 0 −3

ul
 
0 −3 + 3 2 − 0 3 + 9 R3 → R3 − 3R2
 
1 −1 1 1 
 
∼ 0 −1 0 −3 pa
 
0 0 2 12
∴ By back substitution, (−2, 3, 6) is the solution for the system of equations.
Example 3.26. Solve the system of equations by Gauss elimination method 10x − 2y + 3z = 23,
jin
2x + 10y − 5z = −33, 3x − 4y + 10z = 41.
    
10 −2 3   x  23 
     
Solution: The matrix form is  2 10 −5 y = −33
.re
     

3 −4 10 z 41
i.e., AX = B
The augmented matrix

w
 
10 −2 3 23 
 
[A, B] =  2 10 −5 −33
 
w
3 −4 10 41
 
10 −2 323 
 
∼  0 52 −28 −188 R2 → 5R2 − R1
w
 
0 −34 91 341 R3 → 10R3 − 3R1
 
10 −2 3 23 
 
∼  0 52 −28 −188 
 
0 0 3780 11340 R3 → 52R3 + 34R2
The reduced system of equations is given by
10x − 2y + 3z = 23 (1)
52y − 28z = −188 (2)
3780z = 11340 (3)
11340
From (3) ⇒ z = =3 (∵ by back substitution)
3780
om
From (2) ⇒ 52y − 28(3) = −188
52y = −104
y = −2
.c
From (1) ⇒ 10x + 4 + 9 = 23
10x = 10
x=1
ul
Hence the solution is x = 1, y = −2, z = 3.
Example 3.27. Using Gauss elimination method, solve the system 3.15x − 1.96y + 3.85z = 12.95,
pa
2.13x + 5.12y − 2.89z = −8.61, 5.92x + 3.05y + 2.15z = 6.88.
   
3.15 −1.96 3.85   x  12.95 

     
Solution: The matrix form is 2.13 5.12 −2.89 y = −8.61
     
5.92 3.05 2.15 z 6.88
jin
i.e., AX = B
The augmented matrix

.re
 
3.15 −1.96 3.85 12.95 
 
[A, B] = 2.13 5.12 −2.89 −8.61
 
5.92 3.05 2.15 6.88
 
3.15 −1.96 3.85 12.95 
 
∼  0 20.3028 −17.304 −54.705 R2 → 3.15R2 − 2.13R1
w
 
0 21.2107 −16.0195 −54.992 R3 → 3.5R3 − 5.92R1
 
3.15 −1.96 3.85 12.95 
w
 
∼  0 20.3028 −17.304 −54.705
 
0 0 41.7892 43.8398 R3 → 20.3028R3 − 21.2107R2
The reduced system of equations is given by

w
3.15x − 1.96y + 3.85z = 12.95 (1)

20.3028y − 17.304z = −54.705 (2)
41.7892z = 43.8398 (3)
43.8398
From (3) ⇒ z = = 1.049(correct to 3 decimals)
41.7892
From (2) ⇒ 20.3028y − 17.304(1.049) = −54.705

36.5531
⇒y=− = −1.800
20.3028
From (1) ⇒ 3.15x − 1.96(−1.8) + 3.85(1.049) = 12.95
om
3.15x + 7.5667 = 12.95
5.3834
x= = 1.709
3.15
Hence the solution is x = 1.709, y = −1.8, z = 1.049.
.c
1. Solve the given system of equations by Gauss elimination method: −x1 + x2 + 10x3 = 35.61, 10x1 +
ul
x2 − x3 = 11.19, x1 + 10x2 + x3 = 20.08. " #
1453 837 779
Solution: x= ,y = ,z =
1100 550 220
3.6 Pivoting - Gauss Jordan method

pa
Step 1: Write the augmented matrix for the given system of simultaneous equations
 
 a11 a12 a13 b1 
jin
 
 a21 a22 a23 b2 

 
a31 a32 b33 b3
Step 2: Using elementary row operations reduce the given matrix into a diagonal matrix say
.re
 
 c11 0 0 d1 
 
 0 c22 0 d2 

 
0 0 c33 d3
w
    
  c11 0 0   diagonal matrix 
    
Here
 0 c22 0  is a or
  
   
 
0 0 c33 unit matrix
 
w
Step 2: By direct substitution we get the values for unknowns.

w
3.6.1 Part A
1. For solving a linear system AX = B, compare Gauss elimination method and Gauss
Jordan method.
Solution:
Gauss elimination method Gauss Jordan method

Coefficient matrix is transmitted into Coefficient matrix is transmitted into
upper triangular matrix diagonal matrix or unit matrix
Direct method Direct method
Obtain the solution by No need of
back substitution method back substitution method
2. State the two difference between direct and iterative methods for solving system of
om
equations.
Solution:
Direct Method Iterative Method

It gives exact value It gives only approximate value
Simple, takes less time Time consuming and labourious
.c
This method determine all the roots This method determine only one root
at the same time at a time
ul
3.6.2 Part B
Example 3.28. Solve the system of equations by Gauss Jordan method

pa −3x1 + 2x2 − 3x3 = −6
x1 − x2 + x3 = 1
2x1 − 5x2 + 4x3 = 5.
jin
Solution: The given system of equations can be written as
x1 − x2 + x3 = 1
−3x1 + 2x2 − 3x3 = −6
.re
2x1 − 5x2 + 4x3 = 5
Given system of the form AX = B

Gauss-Jordan method:
The augmented matrix form is
w
 
 1 −1 1 1 
 
[A|B] = −3 2 −3 −6
w
 
2 −5 4 5
 
 1
 −1 1 1 

∼ −3 + 3 2−3 −3 + 3 −6 + 3 R2 → R2 + 3R1

w
 
2−2 −5 + 2 4−2 5 − 2 R3 → R3 − 2R1
 
1 −1 1 1 
 
∼ 0 −1 0 −3
 
0 −3 2 3
 
1 −1 1 1
 
∼ 0 1 0 3 R2 → R2 × (−1)
 
0 −3 2 3
 
1 + 0 −1 + 1 1+0 1 + 3  R1 → R1 + R2
 
∼  0 1 0 3 
 
0 −3 + 3(1) 2 + 3(0) 3 + 3(3) R3 → R3 + 3R2
 
1 0 1 4 
om
 
∼ 0 1 0 3 
 
0 0 2 12
 
1 0 1 4
 
∼ 0 1 0 3
  R3
0 0 1 6 R3 →
.c
2
 
1 0
 0 −2 R1 → R1 − R3


∼ 0 1 0 3 
ul
 
0 0 1 6
∴ x1 = −2, x2 = 3, x3 = 6
⇒ (x1 , x2 , x3 ) = (−2, 3, 6). pa
Example 3.29. Solve the system of equation by (i) Gauss elimination method (ii) Gauss
Jordan method
28x + 4y − z = 32
jin
x + 3y + 10z = 24
2x + 17y + 4z = 35

Solution: (x, y, z) = (0.99359, 1.50697, 1.84855)
.re
Example 3.30. Solve the system of equation by (i) Gauss elimination method (ii) Gauss
Jordan method
p+ q+ r+ s = 2
w
p + q + 3r − 2s = −6
2p + 3q − r + 2s = 7
p + 2q + r − s = −2
w

Solution: (p, q, r, s) = (1, 0, −1, 2)
w
1. Apply Gauss-Jordan method to solve the following system of equations x + y + z = 9, 2x − 3y + 4z =

13, 3x + 4y + 5z = 40. (AM11)

Solution: (x, y, z) = (1, 3, 5)
2. Apply Gauss-Jordan method to find the solution of the following system: 10x+y+z = 12, 2x+10y+z =
13, x + y + 5z = 7 (ND11)

Solution: (x, y, z) = (1, 1, 1)
3. Solve the system of equations by Gauss-Jordan method: 5x1 − x2 = 9; −x1 +5x2 − x3 = 4; −x2 +5x3 = −6
(AM14)
om

Solution: (x, y, z) = (2, 1, −1)
4. Using Gauss-Jordan method to solve 2x − y + 3z = 8; −x + 2y + z = 4, 3x + y − 4z = 0. (ND14)

Solution: x = 2, y = 2, z = 2
.c
5. Solve by Gauss Jordan method, the following system 10x+y−z = 11.19; x+10y+z = 20.08; −x+y+10z =
35.61. " #
1453 837 779
1100 550 220
ul
6. Apply Gauss Jordan method to solve the equations x + y + z = 9; 2x − 3y + 4z"= 13; 3x + 4y + 5z = 40.#
11 19 59
pa 12 6 12
Diagonal system
In the system of linear equations in n unknowns AX = B. If the coefficient matrix A is diagonally
dominant then the system is said to be a diagonal system.
jin
Thus the system of equations
    
a1 b1 c1   x d1 
     
i.e., a2 b2 c2  y = d2 
a1 x + b1 y + c1 z = d1      
a3 b3 c3 z d3
.re
a2 x + b2 y + c2 z = d2
i.e., AX = B
a3 x + b3 y + c3 z = d3
is a diagonal system if
w
|a1 | ≥ |b1 | + |c1 |

|b2 | ≥ |a2 | + |c2 |
w
|c3 | ≥ |a3 | + |b3 |

w
Note : For the Gauss Seidal method to converge quickly, the coefficient matrix A must be diagonally
dominant.
If the coefficient matrix A is not diagonally dominant we must rearrange the equations in such a
way that the resulting matrix becomes dominant, and then only we can apply Gauss Seidel method.
Iterative Methods:
Let us consider a system of n linear algebraic equations in n unknowns
a11 x1 + a12 x2 + · · · + a1n xn = b1

a21 x1 + a22 x2 + · · · + a2n xn = b2
..
.
an1 x1 + an2 x2 + · · · + ann xn = bn
om
where the coefficients ai j and the constants bi are real and known. This system of equations in matrix
form may be written as
AX = b
.c

where A = ai j
ul
n×n
X = (x1 , x2 , · · · , xn )T
and b = (b1 , b2 , · · · , bn )T . pa
A is called the coefficient matrix.
We are interested in finding the values xi , i = 1, 2n if they exist, satisfying Equation.
P
A square matrix A is called diagonally dominant if |Aii | ≥ |Ai j | for all i.
j,i
i.e.,
jin
X
|A11 | ≥ |A1 j |
j,1
X
|A22 | ≥ |A2 j |
.re
j,2
X
|A33 | ≥ |A3 j |
j,3
P
A square matrix A is called strictly diagonally dominant if |Aii | > |Ai j | for all i.
j,i
w
i.e.,
X
|A11 | > |A1 j |
w
j,1
X
|A22 | > |A2 j |
j,2
X
|A33 | > |A3 j |
w
j,3
The following methods to solve the system of equations by iterative methods:

(a) Gauss Jacobi Method
(a) Gauss-Seidel Method
3.7 Iterative methods of Gauss-Jacobi
Step 1: Let the system of equations be
a11 x + a12 y + a13 z = b1

a21 x + a22 y + a23 z = b2
a31 x + a32 y + a33 z = b3
om
with diagonally dominant.
Step 2: The above system should write into the form

1
x= (b1 − a12 y − a13 z) (1)
a11
1
y= (b2 − a21 x − a23 z) (2)
a22
.c
1
z= (b3 − a31 x − a32 y) (3)
a33
Step 3: Start with the initial values x(0) = 0, y(0) = 0, z(0) = 0 for x, y, z and get x(1) , y(1) , z(1)
ul
∴ (1),(2),(3) become
1
x(1) = b1 − a12 y(0) − a13 z(0)
a11
1
pa y(1)
z(1)
=
=
a22
1
a33

b2 − a21 x(0) − a23 z(0)
b3 − a31 x(0) − a32 y(0)

Step 4: Using this x(1) for x, y(1) for y, z(1) for z in (1),(2),(3) respectively, we get
jin
1
x(2) = b1 − a12 y(1) − a13 z(1)
a11
(2) 1
y = b2 − a21 x(1) − a23 z(1)
a22
1
z(2) = b3 − a31 x(1) − a32 y(1)
.re
a33
Continuing in the same procedure until the convergence is confirmed.
The general iterative formula of Gauss-Jacobi is
1 h i
x1( j+1) = b1 − a12 x2( j) + a13 x3( j) + · · · + a2n xn( j)
a11
w
1 h i
x2( j+1) = b2 − a21 x1( j) + a23 x3( j) + · · · + a2n xn( j)
a22
( j+1) 1 h i
x3 = b3 − a31 x1( j) + a32 x2( j) + · · · + a2n xn( j)
a33
w
..
.
1 h i
xn( j+1) = bn − an1 x1( j) + an2 x2( j) + · · · + an,n−1 xn−1( j)
w
ann
Note : Suppose n equations with n unknown variables x1 , x2 , · · · , xn , then
In Gauss Jacobi iteration method, for xn( j+1) , use xn( j) values only.

Start with x1 = x2 = · · · = 0

In Gauss Seidel iteration method, for xn( j+1) , use latest values of xn( j) or xn( j+1) .

Start with x2 = x3 = · · · = 0

3.7.1 Part A
1. Why use Jacobi ?

Solution : Because you can separate the n−equations into n independent tasks; it is very well
suited to computers with parallel processors.
3.7.2 Part B
om
Example 3.31. Solve by Gauss-Jacobi iterative method
x + y + 54z = 110
27x + 6y − z = 85
6x + 15y + 2z = 72
.c
Solution: This can be written in diagonally dominant matrix as
27x + 6y − z = 85 (1)
ul
6x + 15y + 2z = 72 (2)
x + y + 54z = 110 (3)
(1) ⇒ x =
(2) ⇒ y =
1
27
1
pa
(85 − 6y + z)
(72 − 6x − 2z)
(4)
(5)
15
1
(3) ⇒ z = (110 − x − y) (6)
54
jin
Iteration 1: Put the initial values x(0) = y(0) = z(0) = 0 in RHS of (4),(5),(6)
1 h i 1
(4) ⇒ x(1) = 85 − 6y(0) + z(0) = [85 − 6(0) + 0] = 3.14815
27 27
1 h i 1
(5) ⇒ y(1) = 72 − 6x(0) − 2z(0) = [72 − 6(0) − 2(0)] = 4.80000
.re
15 15
1 i 1
(6) ⇒ z(1) = 110 − x(0) − y(0) = (110 − 0 − 0) = 2.03704
54 54
Iteration 2: Put the intial values x = 3.14815, y = 4.80000, z = 2.03704 in RHS of (4),(5),(6)
1 h i 1
w
(4) ⇒ x(2) = 85 − 6y(1) + z(1) = [85 − 6(4.80000) + 2.03704] = 2.15693

27 27
1 h i 1
(5) ⇒ y(2) = 72 − 6x(1) − 2z(1) = [72 − 6(3.14815) − 2(2.03704)] = 3.26914
15 15
1 h 1
w
i
(6) ⇒ z(2) = 110 − x(1) − y(1) = (110 − 3.14815 − 4.80000) = 1.88985
54 54
Iteration 3:
w
1 h i 1
(4) ⇒ x(3) = 85 − 6y(2) + z(2) = [85 − 6(3.26914) + 1.88985] = 2.49167
27 27
1 h i 1
(5) ⇒ y(3) = 72 − 6x(2) − 2z(2) = [72 − 6(2.15693) − 2(1.88985)] = 3.68525
15 15
1 h i 1
(6) ⇒ z(3) = 110 − x(2) − y(2) = (110 − 2.15693 − 3.26914) = 1.93655
54 54
Iteration 4:
1 h i 1
(4) ⇒ x(4) = 85 − 6y(3) + z(3) = [85 − 6(3.68525) + 1.93655] = 2.40093
27 27
1 h i 1
(5) ⇒ y(4) = 72 − 6x(3) − 2z(3) = [72 − 6(2.49167) − 2(1.93655)] = 3.54513
15 15
1 h i 1
(6) ⇒ z(4) = 110 − x(3) − y(3) = (110 − 2.49167 − 3.68525) = 1.92265
54 54
Iteration 5:
om
1 h i 1
(4) ⇒ x(5) = 85 − 6y(4) + z(4) = [85 − 6(3.54513) + 1.92265] = 2.43155
27 27
1 h i 1
(5) ⇒ y(5) = 72 − 6x(4) − 2z(4) = [72 − 6(2.40093) − 2(1.92265)] = 3.58328
15 15
1 h i 1
(6) ⇒ z(5) = 110 − x(4) − y(4) = (110 − 2.40093 − 3.54513) = 1.92692
54 54
Iteration 6:
.c
1 h i 1
(4) ⇒ x(6) = 85 − 6y(5) + z(5) = [85 − 6(3.58328) + 1.92692] = 2.42323
27 27
1 h i 1
(5) ⇒ y(6) = 72 − 6x(5) − 2z(5) = [72 − 6(2.43155) − 2(1.92692)] = 3.57046
ul
15 15
1 h i 1
(6) ⇒ z(6) = 110 − x(5) − y(5) = (110 − 2.43155 − 3.58328) = 1.92565
54 54
Iteration 7: pa
1 h i 1
(4) ⇒ x(7) = 85 − 6y(6) + z(6) = [85 − 6(3.57046) + 1.92565] = 2.42603
27 27
1 h i 1
(5) ⇒ y(7) = 72 − 6x(6) − 2z(6) = [72 − 6(2.42323) − 2(1.92565)] = 3.57395
15 15
1 h i 1
(6) ⇒ z(7) = 110 − x(6) − y(6) = (110 − 2.42323 − 3.57046) = 1.92604
jin
54 54
Iteration 8:
1 h i 1
(4) ⇒ x(8) = 85 − 6y(7) + z(7) = [85 − 6(3.57395) + 1.92604] = 2.42527
27 27
.re
1 h i 1
(5) ⇒ y(8) = 72 − 6x(7) − 2z(7) = [72 − 6(2.42603) − 2(1.92604)] = 3.57278
15 15
1 h i 1
(6) ⇒ z(8) = 110 − x(7) − y(7) = (110 − 2.42603 − 3.57395) = 1.92593
54 54
Iteration 9:
w
1 h i 1
(4) ⇒ x(9) = 85 − 6y(8) + z(8) = [85 − 6(3.57278) + 1.92593] = 2.42553
27 27
1 h i 1
(5) ⇒ y(9) = 72 − 6x(8) − 2z(8) = [72 − 6(2.42527) − 2(1.92593)] = 3.57310
15 15
w
1 h i 1
(6) ⇒ z(9) = 110 − x(8) − y(8) = (110 − 2.42527 − 3.57278) = 1.92596
54 54
Iteration 10:
w
1 h i 1
(4) ⇒ x(10) = 85 − 6y(9) + z(9) = [85 − 6(3.57310) + 1.92596] = 2.42546
27 27
1 h i 1
(5) ⇒ y(10) = 72 − 6x(9) − 2z(9) = [72 − 6(2.42553) − 2(1.92596)] = 3.57299
15 15
1 h i 1
(6) ⇒ z(10) = 110 − x(9) − y(9) = (110 − 2.42553 − 3.57310) = 1.92595
54 54
Iteration 11:
1 h i 1
(4) ⇒ x(11) = 85 − 6y(10) + z(10) = [85 − 6(3.57299) + 1.92595] = 2.42548
27 27
1 h i 1
(5) ⇒ y(11) = 72 − 6x(10) − 2z(10) = [72 − 6(2.42546) − 2(1.92595)] = 3.57302
15 15
1 h i 1
(6) ⇒ z(11) = 110 − x(10) − y(10) = (110 − 2.42546 − 3.57299) = 1.92595
54 54
Iteration 12:
om
1 h i 1
(4) ⇒ x(12) = 85 − 6y(11) + z(11) = [85 − 6(3.57302) + 1.92595] = 2.42547
27 27
1 h i 1
(5) ⇒ y(12) = 72 − 6x(11) − 2z(11) = [72 − 6(2.42548) − 2(1.92595)] = 3.57301
15 15
1 h i 1
(6) ⇒ z(12) = 110 − x(11) − y(11) = (110 − 2.42548 − 3.57302) = 1.92595
54 54
.c
Iteration 13:
1 h i 1
(4) ⇒ x(13) = 85 − 6y(12) + z(12) = [85 − 6(3.57301) + 1.92595] = 2.42548
27 27
1 h 1
ul
i
(5) ⇒ y(13) = 72 − 6x(12) − 2z(12) = [72 − 6(2.42547) − 2(1.92595)] = 3.57302
15 15
1 h i 1
(6) ⇒ z(13) = 110 − x(12) − y(12) = (110 − 2.42547 − 3.57301) = 1.92595
54 54
Iteration 14:
(4) ⇒ x(14) =
1 h
pai
85 − 6y(13) + z(13) =
1
[85 − 6(3.57302) + 1.92595] = 2.42548
27 27
1 h i 1
(5) ⇒ y(14) = 72 − 6x(13) − 2z(13) = [72 − 6(2.42548) − 2(1.92595)] = 3.57302
15 15
jin
1 h i 1
(6) ⇒ z(14) = 110 − x(13) − y(13) = (110 − 2.42548 − 3.57302) = 1.92595
54 54
∴ Solution is (x, y, z) = (2.42548, 3.57302, 1.92595)
.re
3.8 Iterative methods of Gauss-Seidel
Step 1: Let the system of equations be
a11 x + a12 y + a13 z = b1

w
a21 x + a22 y + a23 z = b2

a31 x + a32 y + a33 z = b3
w
with diagonally dominant.
Step 2: The above system should write into the form

w
1
x= (b1 − a12 y − a13 z) (1)
a11
1
y= (b2 − a21 x − a23 z) (2)
a22
1
z= (b3 − a31 x − a32 y) (3)
a33
Step 3: Start with the initial values y(0) = 0, z(0) = 0 for y, z and get x(1) from the first equation.
1
∴ (1) becomes x(1) = b1 − a12 y(0) − a13 z(0)
a11
Step 4: Using this x(1) in (2), we use z(0) for z and x(1) for x instead of x(0) , we get
1
∴ (2) becomes y(1) = b2 − a21 x(1) − a23 z(0)
a22
Step 5: Substitute x(1) , y(1) for x, y in the third equation.

1
∴ (3) becomes z(1) = b3 − a31 x(1) − a32 y(1)
om
a33
Step 6: To find the values of unknowns, use the latest available values on the right side. If x(r) , y(r) , z(r)
are the rth iterate values, then the next iteration will be
1
x(r+1) = b1 − a12 y(r) − a13 z(r)
a11
1
y(r+1) b2 − a21 x(r+1) − a23 z(r)
.c
=
a22
1
z(r+1) = b3 − a31 x(r+1) − a32 y(r+1)
a33
ul
Step 7: This process of continued until the convergence is confirmed.
The general iterative formula of Gauss-Seidel is

pa
1 h i
x1( j+1) = b1 − a12 x2( j) + a13 x3( j) + · · · + a2n xn( j)
a11
1 h i
x2( j+1) = b2 − a21 x1( j+1) + a23 x3( j) + · · · + a2n xn( j)
a22
1 h i
x3( j+1) = b3 − a31 x1( j+1) + a32 x2( j+1) + · · · + a2n xn( j)
jin
a33
..
.
1 h i
xn( j+1) = bn − an1 x1( j+1) + an2 x2( j+1) + · · · + an,n−1 xn−1
( j+1)
ann
.re
Note : If either method converges, Gauss-Seidel converges faster than Jacobi.
3.8.1 Part A
w
1. Write the first iteration values of x, y, z when the equations 27x + 6y − z = 85, 6x + 15y + 2z =
72, x + y + 54z = 110 are solved by Gauss-Seidel method.
w
Solution: Since the given equations are in the diagonally dominant and find the x, y, z as follows
1 1 85
x= 85 − 6y + z = [85 − 6(0) + (0)] = = 3.14815 (by putting y = z = 0)
27 27 27
w
1 1
y = [72 − 6x − 2z] = [72 − 6 (3.14815) − 2 (0)] = 3.54074 (by putting x = 3.14815, z = 0)
15 15
1 1
z= 110 − x − y = [110 − 3.14815 − 3.54074] = 1.91317 (by putting x = 3.14815, y = 3.54074)
54 54
∴ First iteration values of x, y, z is (x = 3.14815, y = 3.54074, z = 1.91317)

3.8.2 Part B
Example 3.32. Solve by Gauss-Seidel iterative method [AM 2011]
x + y + 54z = 110
27x + 6y − z = 85
6x + 15y + 2z = 72
om
Solution: This can be written in diagonally dominant matrix as
27x + 6y − z = 85 (1)
6x + 15y + 2z = 72 (2)
x + y + 54z = 110 (3)
1
.c

(1) ⇒ x = 85 − 6y + z (4)
27
1
(2) ⇒ y = [72 − 6x − 2z] (5)
15
1
ul

(3) ⇒ z = 110 − x − y (5)
54
Iteration 1: Put Initial values y(0) = z(0) = 0 in (4)
1 h pai 1
(4) ⇒ x(1) = 85 − 6y(0) + z(0) = [85 − 6(0) + 0] = 3.14815
27 27
1 h i 1
(5) ⇒ y(1) = 72 − 6x(1) − 2z(0) = [72 − 6(3.14815) − 2(0)] = 3.54074
15 15
1 h i 1
(6) ⇒ z(1) = 110 − x(1) − y(1) = [110 − 3.14815 − 3.54074] = 1.91317
54 54
jin
Iteration 2: [Substitute the latest values of x, y, z in (4), (5), (6)]
1 h i 1
x(2) = 85 − 6y(1) + z(1) = [85 − 6 (3.54074) + 1.91317] = 2.43218
27 27
1 h i 1
y(2) = 72 − 6x(2) − 2z(1) = [72 − 6 (2.43218) − 2 (1.91317)] = 3.57204
.re
15 15
1 h i 1
z(2) = 110 − x(2) − y(2) = [110 − 2.43218 − 3.57204] = 1.92585
54 54
Iteration 3:
1 1
x(3) = [85 − 6y(2) + z(2) ] =
w
[85 − 6 (3.57204) + 1.92585] = 2.42569

27 27
1 h i 1
y(3) = 72 − 6x(3) − 2z(2) = [72 − 6 (2.42569) − 2 (1.92585)] = 3.57294
15 15
1 h 1
w
i
z(3) = 110 − x(3) − y(3) = [110 − 2.42569 − 3.57294] = 1.92595
54 54
Iteration 4:
w
1 h i 1
x(4) = 85 − 6y(3) + z(3) = [85 − 6 (3.57294) + 1.92595] = 2.42549
27 27
1 h i 1
y(4) = 72 − 6x(4) − 2z(3) = [72 − 6 (2.42549) − 2 (1.92595)] = 3.57301
15 15
1 h i 1
z(4) = 110 − x(3) − y(3) = [110 − 2.42549 − 3.57301] = 1.92595
54 54
Iteration 5:
1 h i 1
x(5) = 85 − 6y(4) + z(4) = [85 − 6 (3.57301) + 1.92595] = 2.42548
27 27
1 h i 1
y(5) = 72 − 6x(5) − 2z(4) = [72 − 6 (2.42548) − 2 (1.92595)] = 3.57301
15 15
1 h i 1
z(5) = 110 − x(4) − y(4) = [110 − 2.42548 − 3.57301] = 1.92595
54 54
∴ Solution is (x, y, z) = (2.42548, 3.57301, 1.92595)
om
Note : Iterations of the above given problem are given as follows:
Iteration x y z
1 3.14815 3.54074 1.91317
2 1.91317 3.57204 1.92585
3 2.42569 3.57294 1.92595
.c
4 2.42549 3.57301 1.92595
5 2.42548 3.57301 1.92595
Example 3.33. Solve by Gauss-Seidel iterative method
ul
8x − 3y + 2z = 20
6x + 3y + 12z = 35
pa 4x + 11y − z = 33
Solution: [(x, y, z) = (3.016, 1.985, 0.911)]
Example 3.34. Apply Gauss-Seidel method to solve the system of equations

20x + y − 2z = 17, 3x + 20y − z = −18, 2x − 3y + 20z = 25
jin
Solution: [(x, y, z) = (0.999969 ≈ 1, −1064 ≈ −1, 102 ≈ 1)]
Example 3.35. Solve the following system by Gauss-Seidal method :

28x + 4y − z = 32; x + 3y + 10z = 24; 2x + 17y + 4z = 35
Solution: [x = 0.9936, y = 1.507, z = 1.8486]
.re
Example 3.36. By using Gauss-Seidel method, solve the system of equations

6x + 3y + 12z = 35; 8x − 3y + 2z = 20; 4x + 11y − z = 33.
Solution: [(x, y, z) = (3.016, 1.985, 0.911)]
Example 3.37. By using Gauss-Seidel iteration method, solve the following system of
w
equations upto four decimals.

10x − 2y − z − w = 3; −2x + 10y − z − w = 15; −x − y + 10z − 2w = 27; −x − y − 2z + 10w = −9.
Solution: [x = 3.017, y = 1.986, z = 0.91]
w
Example 3.38. Using Gauss Seidel iteration method solve the following system start with
x = 1, y = −2, z = 3: x + 3y + 52z = 173.61, x − 27y + 2z = 71.31, 41x − 2y + 3z = 65.46.
w
Solution: Given
x + 3y + 52z = 173.61
x − 27y + 2z = 71.31
41x − 2y + 3z = 65.46
Since the coefficient matrix A of the given system of equations is not diagonally dominant, we rewrite
the equations
41x − 2y + 3z = 65.46
x − 27y + 2z = 71.31
x + 3y + 52z = 173.61
From the above system of equations
om
1
x= [65.46 + 2y − 3z]
41
1
y = − [71.31 − x − 2z]
27
1
z= [173.61 − x − 3y]
52
.c
Let the initial values be x = 1, y = −2, z = 3
Iteration 1: Iteration 2:
1 1
x(1) = [65.46 + 2(−2) − 3(3)] = 1.2795 x(2) = [65.46 + 2(−2.3715) − 3(3.4509)] = 1.2284
ul
41 41
1 1
y(1) = − [71.31 − 1.2795 − 2(3)] = −2.3715 y(2) = − [71.31 − 1.2284 − 2(3.4509)] = −2.34
27 27
(1) 1 (2) 1
z = [173.61 − 1.2795 − 3(−2.3715)] = 3.4509
pa z = [173.61 − 1.2284 − 3(−2.34)] = 3.4477
52 52
Iteration 3: Iteration 4:
1 1
x(3) = [65.46 + 2(−2.34) − 3(3.4477)] = 1.2302 x(4) = [65.46 + 2(−2.3402) − 3(3.45)] = 1.23
41 41
1 1
y(3) = − [71.31 − 1.2302 − 2(3.4477)] = −2.3402 y(4) = − [71.31 − 1.23 − 2(3.45)] = −2.34
27 27
jin
(3) 1 (4) 1
z = [173.61 − 1.2302 − 3(−2.3402)] = 3.45 z = [173.61 − 1.23 − 3(−2.34)] = 3.45
52 52
From third and fourth iterations
x = 1.23, y = −2.34, z = 3.45

.re
1. Use Gauss-Seidel iterative method to obtain the solution of the equations 9x−y+2z = 9, x+10y−2z =
15, 2x − 2y − 13z = −17. (MJ10)
w
(5) (6)
Solution : [(x, y, z) = (x, y, z) = (0.917, 1.647, 1.195)]
2. Apply Gauss-Seidel method to solve the system of equations 20x + y − 2z = 17, 3x + 20y − z = −18,
(ND10)
w
2x − 3y + 20z = 25.
(3) (4)
Solution : [(x, y, z) = (x, y, z) = (1, −1, 1)]
3. Solve by Gauss-Seidel iterative method x + y + 54z = 110, 27x + 6y − z = 85, 6x + 15y + 2z = 72.(AM11)
w
Solution : [(x, y, z)(3) = (x, y, z)(4) = (2.426, 3.573, 1.926)]
4. Solve the following system by Gauss-Seidal method :

28x + 4y − z = 32; x + 3y + 10z = 24; 2x + 17y + 4z = 35 (ND11)
(5) (6)
Solution : [(x, y, z) = (x, y, z) = (0.9936, 1.507, 1.8486)]
5. Apply Gauss-Seidel method to solve the system of equations

20x + y − 2z = 17, 3x + 20y − z = −18, 2x − 3y + 20z = 25 (MJ12,ND10)
(3)
Solution : [(x, y, z) = (x, y, z)(4) = (1, −1, 1)]
6. Solve the following system of equations using Gauss-Seidel method:10x+2y+z = 9, x+10y−z = −22,
−2x + 3y + 10z = 22 (ND12)
(3) (4)
Solution : [(x, y, z) = (x, y, z) = (1, −2, 3)]
om
7. Solve by Gauss-Seidel method, the equations 27x + 6y − z = 85, 6x + 15y + 2z = 72, x + y + 54z = 110.
(AM13)
(3) (4)
Solution : [(x, y, z) = (x, y, z) = (2.426, 3.573, 1.926)]
.c
8. Solve the equations by Gauss-Seidel method of iteration. 10x + 2y + z = 9, x + 10y − z = −22, −2x +
3y + 10z = 22. (ND13)
(3) (4)
ul
Solution : [(x, y, z) = (x, y, z) = (1, −2, 3)]
9. Using Gauss-Seidel method, solve the following system of linear equation s 4x+2y+z = 14; x+5y−z =
pa
10; x + y + 8z = 20. (AM14)
Solution : [(x, y, z) = (2, 2, 2)]
10. Apply Gauss-Seidal method to solve the system of equations 20x + y − 2z = 17, 3x + 20y − z = −18,
jin
2x − 3y + 20z = 25. (ND14)
Solution : [(x, y, z) = (1, −1, 1)]
.re
3.9 Matrix Inversion by Gauss Jordan method
Given Matrix An×n ,

Form
w
Row
w
h i
[An×n | In×n ] ∼ In×n | A−1
n×n
Operations
w
3.9.1 Part A
 
1 3
1. Find the inverse of A =   by Gauss-Jordan method. (ND14)
2 7
 
1 3 1 0
Solution : Consider [A|I] =  
2 7 0 1
 
1 3 1 0
∼  
0 1 −2 1 R2 → R2 − 2R1
 
1 0 7 −3
∼   R1 → R1 − 3R2
om
0 1 −2 1 
 
 7 −3
∴ A−1 =  
−2 1
3.9.2 Part B
.c
Example 3.39. Using Gauss-Jordan method, find the inverse of the matrix
 
1 0 −1
ul
 
3 4 5 
 

0 −6 −7

  pa
1
 0 −1

Solution: Let A = 3 4 5 
 
0 −6 −7
 
1 0 −1 1 0 0
jin
 
Consider [A|I] = 3 4 5 0 1 0

 
0 −6 −7 0 0 1
 
 1
 0 −1 1 0 0 

∼ 3−3 (1) 4−3 (0) 5−3 (−1) 0−3 (1) 1−3 (0) 0−3 (0) R2 → R2 − 3R1
.re

 
0 −6 −7 0 0 1
 
1 0 −1 1 0 0
 
∼ 0 4 8 −3 1 0

  R2 → R2
0 −6 −7 0 0 1 4
w
 
1 0 −1 1 0 0
 
∼ 0 1 
2 −3/4 1/4 0
 
w
0 −6 −7 0 0 1 R3 → R3 + 6 (R2 )
 
1
 0 −1 1 0 0 

∼ 0 1 2 −3/4 1/4 0 
w
 
0 −6 + 6 (1) −7 + 6 (2) 0 + 6 (−3/4) 0 + 6 (1/4) 1 + 6 (0)
 
1 0 −1 1 0 0
 
∼ 0 1 2 −3/4 1/4 0
  R3
0 0 5 −9/2 3/2 1 R3 →
5
 
1 0 −1 1 0 0  R1 → R1 + R3
 
∼ 0 1 2 −3/4 1/4 0  R2 → R2 − 2 (R3 )
 
0 0 1 −9/10 3/10 1/5
 
 1 + 0 0+0 −1 + 1 1 + (−9/10) 0 + (3/10) 0 + (1/5) 
 
∼ 0 − 2 (0) 1 − 2 (0) 2 − 2 (1) (−3/4) − 2 (−9/10) (1/4) − 2 (3/10) 0 − 2 (1/5)
 
0 0 1 −9/10 3/10 1/5
 
1 0 0 1/10 3/10 1/5 
om
 
∼ 0 1 0 21/20 −7/20 −2/5
 
0 0 1 −9/10 3/10 1/5
h i
= I|A−1
 
 1 3 2 
1  
∴ A−1
.c
= 21/2 −7/2 −4
10  
−9 3 2
 
 1 1 3 
ul
 
Example 3.40. Find the inverse of a matrix  1 3 −3 by Gauss-Jordan method.

 
−2 −4 −4
 
  
pa   12 4 6 
A−1 = 1 
  
Solution: −5 −1 −3
 4  
−1 −1 −1

 
2 2 6 
 
jin
Example 3.41. Using Gauss-Jordan method, find the inverse of the matrix 2 6 −6

 
4 −8 −8

  
 12 4 6 
A−1 = 1 
  
Solution: 1 5 −3
 56  
.re
 
5 −3 −1

 
1 2 −1
 
Example 3.42. Find the inverse of the matrix 4 1 0  by using Gauss-Jordan method.

 
2 −1 3
 
w
  
 −1/5 1/3 −1/15
  
Solution: A−1 = 
4/5 −1/3 4/15

  
2/5 −1/3 7/15
  
w
 
4 1 2 
 
Example 3.43. Find the inverse of the matrix by Gauss-Jordan method: A = 2 3 −1.

w
 
1 −2 2
 
  
 −4/3 2 7/3 
  
Solution: A−1 = 
  5/3 −2 −8/3 

  
7/3 −3 −10/3

 
 0 1 2 
 
1. Find the inverse of the matrix by Gauss - Jordan method:  1 2 3 . (AM10)
 
3 1 1
  


 1/2 −1/2 1/2 

   

 −1 
 
Solution: A =

−4 3 −1
 




 
 


  
 
5/2 −3/2 1/2 
  
om

 
4 1 2 
 
2. Find the inverse of A = 2 3 −1 by using Gauss-Jordan method. (ND10)
 
1 −2 2
  


 −4/3 2 7/3  

   

.c

 −1  
Solution: A =

5/3 −3 −8/3
 




 
 
 


 

7/3 −3 −10/3 

  
 
 1 1 3 
ul
 
3. Using Gauss Jordan method, find the inverse of the matrix  1 
 3 −3. (MJ12)
 
−2 −4 4
pa   


  12 4 6  

 −1 1 


  


Solution: A = −5 −1 −3 
 
4


 

 
 
 

−1 −1 −1 

 
 
1 2 6 
jin
 
4. Using Gauss-Jordan method, find the inverse of the matrix 2 5 15. (ND12)
 
6 15 46
  


  5 −2 0  


   
 −1 
 
Solution: A =

−2 10 −3 
.re




  



0 −3 1 

 
 
4 1 2 
 
5. Find, by Gauss-Jordan method, the inverse of the matrix A = 2 3 −1. (MJ13)
 
1 −2 2
w
  


 −4/3 2 7/3  


  
 −1 
 
Solution: A =  5/3 −3 −8/3 



 
  
 
w
  
7/3 −3 −10/3 

 
 
 3
 −1 1 

6. Find the inverse of the matrix −15 6 −5 using Gauss-JOrdan method. (ND13)
w


 
5 −2 2
  




2 0 −1 


 −1 
   

Solution: A = 5


 1 0 


  



0 1 3 

 
 
2 2 6 
 
7. Using Gauss-Jordan method, find the inverse of 2 6 −6. (AM14)
 
4 −8 8
  


  0 1/5 3/20  


   
 −1 
 
Solution: A = 1/8 1/40


 −3/40 


  



1/8 −3/40 −1/40 

 
om
3.10 Eigen values of a matrix by Power method
Suppose a given square matrix is A.

 
1
 
Let X1 = 0 be an arbitrary initial eigen vector of the matrix A.
 
0
.c
    
 value 1   numerically largest value   1 
     
AX1 =  value 2  =   = λ [X ]
of matrix A say   value 2/value 1
 
 1 2
  
ul
value 3 value 1 value 3/value 1
 
    
 value 1   numerically largest value   value 1/value 2 
     
 = λ [X ]
AX2 =  value 2  =  of matrix A say   1  2 3
 pa     
..
.

AXn = λn [Xn+1 ] If λn = λn+1 and Xn+1 = Xn+2 are same





AXn+1  upto required decimals,then stop iteration.
= λn+1 [Xn+2 ] 

jin
∴ Dominant eigen value = λn (or) λn+1 and the corresponding
eigen vector = Xn+1 (or) Xn+2
.re
To find the smallest eigen value of A

 
1
 
Form B = A − λI & Y1 = 0 be an arbitrary eigen vector of the matrix B.
 
0
w
    
 value 1   numerically largest value   1 
     
BY1 =  value 2  =   = λ [Y ]
of matrix B say   value 2/value 1
   1 2
  
 
w
    
 value 1   numerically largest value   value 1/value 2 
     
BY2 =  value 2  =   = λ [Y ]
of matrix B say 1
 
   2 3
  
w
  
..
.

BYn = λn [Yn+1 ] If λn = λn+1 and Yn+1 = Yn+2 are same





BYn+1  upto required decimals,then stop iteration.
= λn+1 [Yn+2 ] 





 Dominant eigen value of B


∴ The smallest eigen value of A = 


 +


 Dominant eigen value of A

To find third eigen value of A:
λ1 + λ2 + λ3 = Sum of the main diagonal elements.
3.10.1 Part A
om
1. What type of eigen value can by obtained by using power method.
Solution: We can obtain dominant eigen value of the given matrix.
2. Write down all possible initial vectors of a matrix 2 × 2 order.

Solution: The initial vectors are (0, 1)T , (1, 0)T , (1, 1)T .
.c
   
4 1 1
3. Find the dominant eigen value of A =   by power method upto two decimals and choose  
1 3 1
as the initial eigen vector.
ul
 
1
Solution: Let X1 =  
1
      
4 1 1 5
pa  1 
AX1 =     =   = 5   = 5X2
1 3 1 4 0.8
      
4 1  1  4.8  1 
AX2 =     =   = 4.8 
0.71 = 4.8X3

1 3 0.8 3.4
      
jin
4 1  1  4.71  1 
AX3 =     =   = 4.71 
0.67 = 4.71X4
 
1 3 0.71 3.13
      
4 1  1  4.67  1 
AX4 =     =   = 4.67    = 4.67X
 5
1 3 0.67     3.01   0.65
.re
      
4 1  1  4.65  1 
AX5 =     =   = 4.65 
0.63 = 4.65X6

1 3 0.65 2.95
      
4 1  1  4.63  1 
AX6 =     =   = 4.63 
0.62 = 4.63X7
 
1 3 0.63 2.89
w
      
4 1  1  4.62  1 
AX8 =     =   = 4.62    = 4.62X
 8
1 3 0.62    2.86   0.62
w
 
 1 
∴ The eigen value = λ = 4.62 and the corresponding eigen vector = X =   .

0.62
w
3.10.2 Part B
 
−4 −5
Example 3.44. Find the dominant eigen value and eigen vector of the matrix   by
1 2
power method.
 
1
Solution: Let X1 =   be an arbitrary initial eigen vector.
0
        
−4 −5 1 (−4 × 1) + (−5 × 0) −4  1 
−0.25 = −4X2 , say
AX1 =     =   =   = −4  
1 2  0  (1 × 1) + (2 × 0)   1 
        
−4 −5  1  (−4 × 1) + (−5 × (−0.25)) −2.75  1 
 = −2.75 
AX2 =  −0.18182 = −2.75X3
   =   =   
1 2  −0.25  (1 × 1) + (2 × (−0.25))   0.5 
om
      
−4 −5  1  −3.0909  1 
AX3 =     = 
   = −3.0909 
   = −3.0909X
 4
1 2 −0.18182
    0.63636   −0.20588
      
−4 −5  1  −2.9706  1 
 = −2.9706 
AX4 =  −0.06666 = −2.9706X5
   =  
1 2  −0.20588  0.58824 
      
−4 −5  1  −3.6667  1 
.c
AX5 =     = 
   = −3.6667 
   = −3.6667X
 6
1 2 −0.06666     0.86668   −0.23637
      
−4 −5  1  −2.81815  1 
 = −2.81815 
AX6 =  −0.18709 = −2.81815X7
   =  
1 2  −0.23637  0.52726 
ul
      
−4 −5  1  −3.06455  1 
 = −3.06455 
AX7 =  −0.19579 = −3.06455X8
   =  
1 2  −0.18709  0.62582 
AX8 = 

−4
1
−5 
2 −0.19579

 
 
1
pa  
 −3.02105
 = 



 0.60842

 = −3.022105 






1
−0.20133


 = −3.022105X
 9
Dominant Eigen value = λ = −3 and the corresponding

jin
   
 1   1 
Eigen vector = X =    ≈ 
  

−0.19529 −0.2
Example 3.45. Find by power method, the largest eigen value and the corresponding eigen
 
.re
 1 3 −1
 
vector of the matrix  3 2 4 .
 
−1 4 10
 
 
0.02404
 
Solution: Dom. Eigen value = λ = 11.72 and Eigen vector = X = 0.42582
w
 
1
 
1 6 1

w

Example 3.46. Find all the eigen values of the matrix 1 0 by power method.

2
 
0 0 3
 
w
Solution: Dominant Eigen value of A = λ1 = 4, Dominant Eigen value of B = −5

Smallest Eigen value of A = λ2 = Dom. Eig. val. of A + Dom. Eig. val. of B = 4 + (−5) = −1
Other Eigen value of A = λ3 = Sum of the main diagonal elements of A − (λ1 + λ2 ) = 6 − 3 = 3
The required eigen values of the given matrix are 4, −1, 3

 
 1 1  6
 
1. Find the dominant eigen value and the corresponding eigen vector of the matrix A =  1 2 0 .
 
0 0 3
 (AM10)
 


  1  



   

th
Solution: In 8 iteration, Dominant E. value = 4, corresponding E. vector = 0.5
   

 
om


  
  

0 

 
2. Find, by power method, the largest eigenvalue and the corresponding eigenvector of a matrix
 
3 −1
 1
 
2 4  with initial vector (1 1 1)T .
A =  3 (ND10)
 
4 10 −1
.c
n o
Solution: In 8th iteration, Dom. E. value = 11.663, corres. E. vector = (0.025 0.422 1)T
 
5 0
1

ul

3. Using Jacobi’s method, find all the eigenvalues and eigenvectors for the given matrix 0 −2 0.
 
1 0 5
 1 (AM11)
 √   √1 
pa  

   

  0    
2 2

   


       
      
 
Solution: E. values = {−2, 4, 6}, E. vectors = −1 ,  0  ,  0 
      


 

    1   1 
      

0 − √   √ 


 


 

2 2
 
jin
1 6 1
 
4. Find the largest eigenvalue of 1 2 0 by using Power method. (ND11)
 
0 0 3
      


 −0.9487 -0.8944  −0.2357 
  
.re

 
 
   
   

Solution: Eigen values = {−1, 4, 3}, Eigen vectors = , -0.4472 ,
 
0.3162 −0.2357

 
 
 
 
 




      

      


0 0 0.9428 

 
 
 2 −1 0 
 
5. Determine the largest eigen value and the corresponding eigen vector of the matrix −1 2 −1.
w
 
0 −1 2
(MJ12)     1   
−
 0.5   √2   -0.5 


   


  
w


 

 1     1 


   
 
   

Solution: Eigen value = {0.5858, 2, 3.4142}, Eigen vectors =  √  ,  0  ,  √ 
 
  



  2   1   2 
      

   √    
0.5 -0.5 


  

 
2
w
 √ 
 1 2 2 
 √ √ 
6. Find all the eigenvalues and eigenvectors of the matrix  2 3 2 using Jacobi method.
 √ 
2 2 1
(ND12)
  1     
  √   
  0.5   0.5 


   

2

  


    1   1 
    


Solution: Eigen values = {−1, 1, 5}, Eigen vectors = , − √ , √
 
−0

 
 

 

 

 


       
2 2

 
 
     


  1  

 

  

 





  − √  0.5 0.5 



2
7. Using  Jacobi method find the all eigen values and their corresponding eigen vectors of the matrix
2 3
A =  . (MJ13)

3 2
om
 !T !T 

 1 1 1 1  
Solution: Eigen values = {−1, 5}, Eigen vectors = − √ , √
 
 √ √ 
2 2 2 2 

 
 
 1 3 −1
 
8. Determine the largest eigenvalue and the corresponding eigenvector of a matrix  3 2 4  with
 
−1 4 10
.c
(1 1 1)T as initial vector by power method. (ND13)
n o
Solution: Dominant E. value = 11.662, corresponding E. vector = (0.0229 0.3885 0.9212)T
 
25 1 2 
ul
 
9. Find the numerically largest eigen value of A =  1 3 0  and its corresponding eigen vector by
 
2 0 −4
T
power method, taking the initial eigen vector as 1 0 0 (upto three decimal places). (AM14)
Solution:
n
pa
Dom. E. value = 25.1822, corres. E. vector = (−0.9967 − 0.0449 − 0.0683)T
 
o
25 2  1
 
10. Find the numerically largest eigenvalue if A =  1 3 0  and the corresponding eigen vector.
jin
 
2 0 −4
(ND14)
n o
Solution: Dom. E. value = 25.1822, corres. E. vector = (−0.9967 − 0.0449 − 0.0683)T
 
 15 −4 −3 
.re
 
11. Obtain by power method the numerically largest eigen value of the matrix −10 12 −6 .
 
−20 4 −21
      


 0.1000  0.7068  −0.4399 


      


Solution: E. values = −22.2315, 20, 8.2315, E. vectors = , ,
 

 0.2001 −0.5823 −0.8797
   
 
 


w


 
 
    
   


0.9747 −0.4016 0.1806 

 
 
1 1 3
 
w
12. Solve by power method, to find the dominant Eigen value for the following matrix 1 5 1.
 
3 1 1
  1  

 − √      
  0.5774  0.4082

 
w
2

     



       

Solution: Eigen values = {−2, 3, 6}, Eigen vectors = , ,
 




 −0 

 

 −0.5774 

 

 0.8165 

 
      
 
1

      

0.5774 0.4082
 
√

 

 

 


 

2
[Assignment Problems, P.T.O.]

3.10.4 Assignment Problems
1. Solve e x − 3x = 0 by the method of fixed point iteration. (AM12)
2. Find a positive root of the equation cosx − 3x + 1 = 0 by using iteration method. (AM13)
3. Solve for a positive root of the equation x4 − x − 10 = 0 using Newton - Raphson method.(MJ2010)
4. Using Newton’s method, find the real root of x log10 x = 1.2 correct to five decimal places.(ND2013)
om
5. Solve the given system of equations by Gauss elimination method: −x1 + x2 + 10x3 = 35.61, 10x1 +
x2 − x3 = 11.19, x1 + 10x2 + x3 = 20.08.
6. Solve the system of equations by Gauss-Jordan method: 5x1 − x2 = 9; −x1 +5x2 − x3 = 4; −x2 +5x3 = −6
(AM14)
.c
7. Using Gauss-Jordan method to solve 2x − y + 3z = 8; −x + 2y + z = 4, 3x + y − 4z = 0. (ND14)
8. Using Gauss-Seidel method, solve the following system of linear equation s 4x+2y+z = 14; x+5y−z =
10; x + y + 8z = 20. (AM14)
ul
9. Apply Gauss-Seidal method to solve the system of equations 20x + y − 2z = 17, 3x + 20y − z = −18,
2x − 3y + 20z = 25. (ND14)

 3

10. Find the inverse of the matrix −15 6


−1
pa 
1 

−5 using Gauss-Jordan method.

(ND13)
5 −2 2
 
2 2 6 
jin
 
11. Using Gauss-Jordan method, find the inverse of 2 6 −6. (AM14)
 
4 −8 8
 
1 6 1
.re
 
12. Find the largest eigenvalue of 1 2 0 by using Power method. (ND11)
 
0 0 3
 
25 1 2 
 
13. Find the numerically largest eigen value of A =  1 3 0  and its corresponding eigen vector by
w
 
2 0 −4
T
power method, taking the initial eigen vector as 1 0 0 (upto three decimal places). (AM14)
w
w
4 Interpolation, Numerical Differentiation and
Numerical Integration
om
Lagrange’s and Newton’s divided difference interpolations - Newton’s forward and backward
difference interpolation - Approximation of derivates using interpolation polynomials - Numerical
.c
single and double integrations using Trapezoidal and Simpson’s 1/3 rules.
ul
4.1 Introduction
pa
The process of finding the value of a function inside the given range of discrete points are called
interpolation. We have
1. Interpolation with unequal intervals

jin
2. Interpolation with equal intervals
Methods of Equal or Unequal intervals Methods of only Equal intervals

.re
Lagrange’s interpolation (Lagrange & Inverse Lagrange) Newton’s forward difference method
Newton’s divided difference interpolation Newton’s backward difference method
Cubic Splines
w
4.2 Lagrange’s interpolation
Lagrangian Polynomials(Equal and unequal intervals):

w
Let y = f (x) be a function which takes the values y = y0 , y1 , · · · , yn corresponding to x = x0 , x1 , · · · , xn .

Lagrange’s interpolation formula(x given, finding y in terms of x)
w
(x − x1 )(x − x2 ) . . . (x − xn )
y = y(x) = f (x) = y0
(x0 − x1 )(x0 − x2 ) . . . (x0 − xn )
(x − x0 )(x − x2 ) . . . (x − xn )
+ y1 + . . .
(x1 − x0 )(x1 − x2 ) . . . (x1 − xn )
(x − x0 )(x − x1 ) . . . (x − xn−1 )
+ yn
(x2 − x0 )(x2 − x1 ) . . . (x − xn−1 )
Inverse Lagrange’s interpolation formula(y given, finding x in terms of y)

(y − y1 )(y − y2 ) . . . (y − yn )
x = x(y) = f (y) = x0
(y0 − y1 )(y0 − y2 ) . . . (y0 − yn )
(y − y0 )(y − y2 ) . . . (y − yn )
+ x1 + . . .
(y1 − y0 )(y1 − y2 ) . . . (y1 − yn )
(y − y0 )(y − y1 ) . . . (y − yn−1 )
+ xn
(y2 − y0 )(y2 − y1 ) . . . (y − yn−1 )
Note: Lagrange’s interpolation formula can be used for equal and unequal intervals.
om
4.2.1 Part A
1. State Lagrange’s interpolation formula for unequal intervals. (ND11)

Solution: Lagrange’s interpolation formula for unequal intervals is
(x − x1 )(x − x2 ) . . . (x − xn )
.c
y = y(x) = f (x) = y0
(x0 − x1 )(x0 − x2 ) . . . (x0 − xn )
(x − x0 )(x − x2 ) . . . (x − xn )
+ y1 + . . .
(x1 − x0 )(x1 − x2 ) . . . (x1 − xn )
(x − x0 )(x − x1 ) . . . (x − xn−1 )
ul
+ yn
(x2 − x0 )(x2 − x1 ) . . . (x − xn−1 )
2. Using Lagrange’s formula, find the polynomial to the given data. (MJ13)
pa
X:
Y:
0
5
1
6 50
3
Solution: Lagrange’s formula to find ‘y’ for three sets of given values
(x0 = 0, y0 = 5) , (x1 = 1, y1 = 6) and (x2 = 3, y2 = 50)
jin
(x − x1 )(x − x2 ) (x − x0 )(x − x2 ) (x − x0 )(x − x1 )
y = f (x) = y0 + y1 + y2
(x0 − x1 )(x0 − x2 ) (x1 − x0 )(x1 − x2 ) (x2 − x0 )(x2 − x1 )
(x − 1)(x − 3) (x − 0)(x − 3) (x − 0)(x − 1)
= (5) + (6) + (50)
(0 − 1)(0 − 3) (1 − 0)(1 − 3) (3 − 0)(3 − 1)
5 25
.re
= (x − 1)(x − 3) − 3x(x − 3) + x(x − 1)

3 3
5
= (x − 1)[x − 3 + 5x] − 3 x2 − 3x
3
5
= (x − 1)[6x − 3] − 3 x2 − 3x
3

w
= 5(x − 1)[2x − 1] − 3x2 − 9x

h i
= 5 2x2 − 3x + 1 − 3x2 + 9x
= 7x2 − 6x + 5
w
3. Find the second degree polynomial through the points (0, 2), (2, 1),(1, 0) using Lagrange’s formula.
(ND14)
w
4. What is the assumptions we make when Lagrange’s formula is used?

Solution: Lagrange’s interpolation formula can be used whether the values of x, the
independent variable are equally spaced or not whether the difference of y become smaller or
not.
Unit IV - Interpolation, Numerical Differentiation and Numerical Integration 53
5. What is the disadvantage in practice in applying Lagrange’s interpolation formula?

Solution: Though Lagrange’s formula is simple and easy to remember, its application is not
speedy. It requires close attention to sign and there is always a chance of committing some error
due to a number of positive and negative signs in the numerator and the denominator.
6. What is ‘inverse interpolation’?

Solution: Suppose we are given a table of values of x and y. Direct interpolation is the process if
finding the values of y corresponding to a value of x, not present in the table. Inverse interpolation
om
is the process of finding the values of x corresponding to a value of y, not present in the table.
7. Construct a linear interpolating polynomial given the points (x0 , y0 ) and (x1 , y1 ).
(x − x1 ) (x − x0 )
Solution: y = y(x) = f (x) = y0 + y1
(x0 − x1 ) (x1 − x0 )
8. What is the Lagrange’s formula to find ‘y’ if three sets of values (x0 , y0 ) , (x1 , y1 ) and (x2 , y2 ) are
.c
given.
(x − x1 )(x − x2 ) (x − x0 )(x − x2 ) (x − x0 )(x − x1 )
Solution: y = y0 + y1 + y2
(x0 − x1 )(x0 − x2 ) (x1 − x0 )(x1 − x2 ) (x2 − x0 )(x2 − x1 )
ul
9. Find the second degree polynomial fitting the following data:
pa x 1 2 4
y 4 5 13
Solution: Here x0 = 1, x1 = 2, x2 = 4
jin
y0 = 4, y2 = 5, y2 = 13
By Lagrange’s formula for three points is
(x − x1 )(x − x2 ) (x − x0 )(x − x2 ) (x − x0 )(x − x1 )

y = y0 + y1 + y2
.re
(x0 − x1 )(x0 − x2 ) (x1 − x0 )(x1 − x2 ) (x2 − x0 )(x2 − x1 )

(x2 − 6x + 8) (x2 − 5x + 4) (x2 − 3x + 2)
y = (4) + (5) + (13)
(−1)(−3) ( 1)(−2) (3)(2)
(x2 − 6x + 8) (x2 − 5x + 4) (x2 − 3x + 2)
= (4) + (5) + (13)
3 −2 6
1 h 2 i
8x − 48x + 64 − 15x2 + 75x − 60 + 13x2 − 39x + 26
w
=
6
1 h 2 i
= 6x − 12x + 30
6
w
y = f (x) = x2 − 2x + 5
4.2.2 Part B
w
Example 4.1. Using Lagrange interpolation formula, find f (4) given that f (0) = 2, f (1) =
3, f (2) = 12, f (15) = 3587.
x x0 = 0 x1 = 1 x2 = 2 x3 = 15
Solution: Given
y = f (x) y0 = 2 y1 = 3 y2 = 12 y3 = 3587
Lagrange interpolation formula is
(x − x1 )(x − x2 )(x − x3 ) (x − x0 )(x − x2 )(x − x3 )

y = y0 + y1
(x0 − x1 )(x0 − x2 )(x0 − x3 ) (x1 − x0 )(x1 − x2 )(x1 − x3 )
(x − x0 )(x − x1 )(x − x3 ) (x − x0 )(x − x1 )(x − x2 )
+ y2 + y3
(x2 − x0 )(x2 − x1 )(x2 − x3 ) (x3 − x0 )(x3 − x1 )(x3 − x2 )
(4 − 1)(4 − 2)(4 − 15) (4 − 0)(4 − 2)(4 − 15)
f (4) = (2) + (3)
(0 − 1)(0 − 2)(0 − 15) (1 − 0)(1 − 2)(1 − 15)
(4 − 0)(4 − 1)(4 − 15) (4 − 0)(4 − 1)(4 − 2)
+ (12) + (3587)
om
(2 − 0)(2 − 1)(2 − 15) (15 − 0)(15 − 1)(15 − 2)
(3)(2)( − 11) (4)(2)( − 11)
= (2) + (3)
( − 1)( − 2)( − 15) (1 )(−1 )( − 14)
(4)(3)( − 11) (4)(3)( 2)
+ (12) + (3587)
(2 )( 1)( − 13) (15)(14)(13)
= 77.99 = 78
.c
Example 4.2. Find polynomial f (x) by using Lagrange formula from the given data and find
f (8).
ul
x 3 7 9 10
f (x) 168
pa 120 72 63
Solution: Lagrange polynomial f (x) = x3 − 21x2 + 119x − 27.

y(x=8) or y(x = 8) or f (x = 8) = 93
Example 4.3. Use Lagrange’s formula to fit a polynomial to the data

jin
x −1 0 2 3
f (x) −8 3 1 12
.re
and hence find y(1).

Solution: Lagrange polynomial f (x) = 2x3 − 6x2 + 3x + 3.

y(x=1) = y(x = 1) = f (x = 1) = 2.
w
Example 4.4. Using Lagrange’s formula, prove that
y1 = y3 − 0.3(y5 − y−3 ) + 0.2(y−3 − y−5 ).

w
Solution: From the equation, the values of x are

w
x0 = −5 x1 = −3 x2 = 3 x3 = 5
y0 = y−5 y1 = y−3 y2 = y3 y3 = y5
The x values are not equally space, so use Lagrange’s formula to find y = f (x).
Lagrange’s formula for a set of 4 pair of values is
(x − x1 )(x − x2 )(x − x3 ) (x − x0 )(x − x2 )(x − x3 )

y = y x = f (x) = y0 + y1
(x0 − x1 )(x0 − x2 )(x0 − x3 ) (x1 − x0 )(x1 − x2 )(x1 − x3 )
(x − x0 )(x − x1 )(x − x3 ) (x − x0 )(x − x1 )(x − x2 )
+ y2 + y3
(x2 − x0 )(x2 − x1 )(x2 − x3 ) (x3 − x0 )(x3 − x1 )(x3 − x2 )
(x + 3)(x − 3)(x − 5) (x + 5)(x − 3)(x − 5)
= y−5 + y−3
(−5 + 3)(−5 − 3)(−5 − 5) (−3 + 5)(−3 − 3)(−3 − 5)
(x + 5)(x + 3)(x − 5) (x + 5)(x + 3)(x − 3)
+ y3 + y5
om
(3 + 5)(3 + 3)(3 − 5) (5 + 5)(5 + 3)(5 − 3)
Put x = 1, we get
(1 + 3)(1 − 3)(1 − 5) (1 + 5)(1 − 3)(1 − 5)
y1 = y−5 + y−3
(−5 + 3)(−5 − 3)(−5 − 5) (−3 + 5)(−3 − 3)(−3 − 5)
(1 + 5)(1 + 3)(1 − 5) (1 + 5)(1 + 3)(1 − 3)
+ y3 + y5
(3 + 5)(3 + 3)(3 − 5) (5 + 5)(5 + 3)(5 − 3)
.c
= −0.2y−5 + 0.5y−3 + y3 − 0.3y5
= −0.2y−5 + 0.2y−3 + 0.3y−3 + y3 − 0.3y5
y1 = y3 − 0.3(y5 − y−3 ) + 0.2(y−3 − y−5 )
ul
Example 4.5. Find the age corresponding to the annuity value 13.6 from the given table
Age (x)
pa 30 35 40 45 50
Annuity value (y) 15.9 14.9 14.1 13.3 12.5

x13.6 or (y = 13.6) = 43

Solution:
Example 4.6. Find x for which y = 7, given

jin
x 1 3 4
y 4 12 19

Solution: x(y = 7) = 1.8565
.re
1. Use Lagrange’s formula to find a polynomial which takes the values f (0) = −12, f (1) = 0, f (3) = 6
and f (4) = 12. Hence find f (2). (AM10)
w
h i
3 2
Solution: f (x) = x − 7x + 18x − 12 ⇒ f (2) = 4
2. Using Lagrange’s interpolation formula to fit a polynomial to the given data f (−1) = −8, f (0) =
3, f (2) = 1 and f (3) = 12. Hence find the valur of f (1). (ND10)
w
h i
3 2
Solution: f (x) = 2x − 6x + 3x + 3 ⇒ f (1) = 2
3. Find the expression of f (x) using Lagrange’s formula for the followin data. (AM11)
w
x: 0 1 4 5
f (x) : 4 3 24 39
h i
Solution: f (x) = 2x2 − 3x + 4
4. Find the value of x when y = 20 using Lagrange’s formula from the following table. (AM11)
x: 1 2 3 4
y = f (x) : 1 8 27 64

Solution: f (y = 20) = x(y = 20) = 2.84675
5. Using Lagrange’s interpolation, calculate the profit in the year 2000 from the following data:
om
(MJ12)
Year: 1997 1999 2001 2002

.
Profit in Lakhs Rs. : 43 65 159 248

Solution: f (x = 2000) = 100
.c
6. Use Lagrange’s method to find log10 656, given that log10 654 = 2.8156, log10 658 = 2.8182, log10 659 =
2.8189, and log10 661 = 2.8202. (ND12)

Solution: f (656) = 2.8168
ul
7. Apply Lagrange’s formula, to find y(27) to the data given below. (MJ13)
x: 14
pa 17 31 35
y: 68.8 64 44 39.1

Solution: f (27) = 49.3
8. Use Lagrange’s formula to find the value of y at x = 6 from the following data: (ND13)
jin
x: 3 7 9 10
y: 168 120 72 63

Solution: f (6) = 147
.re
9. Using Lagrange’s interpolation formula, find y(2) from the following data: y(0) = 0; y(1) = 1; y(3) =
81; y(4) = 256; y(5) = 625 (AM14)

Solution: f (2) =
w
10. Find the polynomial f (x) by using Lagrange’s formula and hence find f (3) for the following values
of x and y: (ND14)
w
x 0 1 2 5
y 2 3 12 147
w

Solution: f (x) = , f (2) =
11. Find the Lagrange’s polynomial of degree 3 to fit the data : y(0) = −12, y(1) = 0, y(3) = 6 and
h i
y(4) = 12. Hence find y(2). Ans: f (x) = x3 − 7x2 + 18x − 12; y(2) = 4
12. Find the missing term in the following table using Lagrange’s interpolation.
x 0 1 2 3 4
Ans : 31
y 1 3 9 − 81
13. Find the value of x corresponding to y = 100 from the table.
x 3 5 7 9 11
Ans : 8.656

y 6 24 58 108 174
om
4.3 Newton’s divided difference interpolation
First divided difference for arguments x0 , x1 :

f (x1 ) − f (x0 )
= f (x0 , x1 ) = ∆| f (x0 ) = [x0 , x1 ](or)[x1 , x0 ] = ∆| f (x1 )
x1 − x0 x1 x0
.c
First divided difference for arguments x1 , x2 :
f (x2 ) − f (x1 )
= f (x1 , x2 ) = ∆| f (x1 ) = [x1 , x2 ](or)[x2 , x1 ] = ∆| f (x2 )
x2 − x1 x2 x1
ul
Second divided difference for arguments x1 , x2 , x3 :
∆| f (x2 ) − ∆| f (x1 )
x3 x2
= f (x1 , x2 , x3 ) = ∆| 2 f (x1 ) = [x1 , x2 , x3 ]
x3 − x1
pa x3 x2
Third divided difference for arguments x0 , x1 , x2 , x3 :

∆| 2 f (x1 ) − ∆| 2 f (x0 )
x3 x2 x2 x1
= f (x0 , x1 , x2 , x3 ) = ∆| 3 f (x0 ) = [x0 , x1 , x2 , x3 ]
x3 − x0 x3 x2 x1
Newton’s divided difference formula is

jin
f (x) = f (x0 ) + (x − x0 ) f (x0 , x1 ) + (x − x0 ) (x − x1 ) f (x0 , x1 , x2 )
+ (x − x0 ) (x − x1 ) (x − x2 ) f (x0 , x1 , x2 , x3 ) + · · ·
.re
4.3.1 Properties of divided differences
1. The divided difference are symmetric functions of their arguments.
For example, (1) ∆| f (x) = ∆| f (y)

w
y x
(2) ∆| 2 f (x) = ∆| 2 f (y) = ∆| 2 f (z)

yz xz xy
w
2. The nth divided differences of a polynomial of degree n are constants.
3. The divided difference operator(∆| ) is a linear operator.

w
∆| [ f (x) + g(x)] = ∆| [ f (x)] + ∆| [g(x)] and
∆| [c f (x)] = c∆| [ f (x)], where c is constant
Note : For n set of (x, y) values, we get upto (n − 1)th divided difference values.
4.3.2 Part A
1. Form the divided difference table for the data (0,1) , (1,4) , (3,40) and (4,85). (AM10)
Solution: Newton’s divided difference table is
x f (x) ∆| f (x) ∆| 2 f (x) ∆| 3 f (x)
0 1
4−1
=3
1−0
om
18 − 3
1 4 =5
3−0
40 − 4 6.75 − 5
= 18 = 0.44
3−1 4−0
45 − 18
3 40 = 6.75
4−0
85 − 40
= 45
4−3
.c
4 85
1
2. Find the first and second divided differences with arguments a, b, c of the function f (x) = .(ND10)
x
1 1
ul
Solution : If f (x) = ⇒ f (a) =
x a
1 1
1 f (b) − f (a) − 1
f (a, b) = ∆| = = b a =−
b a
1
f (a, b, c) = ∆| 2 =
f
b−a
(b, c) −
pa
f (a,
b−a
b)
1
− +
ab
1
= bc ab =
−a + c
=
1
bc a c−a c−a abc(c − a) abc
1 1
∴ ∆| 2 =
bc a abc
jin
3. Find the divided differences of f (x) = x3 − x + 2 for the arguments 1, 3, 6, 11. (AM11)
Solution: Newton’s divided difference table is
.re
x f (x) = x3 − x + 2 ∆| f (x) ∆| 2 f (x) ∆| 3 f (x)
1 2
26 − 2
= 12
3−1
62 − 12
3 26 = 10
6−1
212 − 26 20 − 10
w
= 62 =1
6−3 11 − 1
222 − 62
6 212 = 20
11 − 3
1322 − 212
= 222
w
11 − 6
11 1322
4. Construct the divided difference table for the following data: (MJ12)
w
x: 0 1 2 5
f (x) : 2 3 12 147
Solution: [Ref : Part A : Example 1 (AM10)]

5. Find the divided differences of f (x) = x3 − x2 + 3x + 8 for arguments 0, 1, 4, 5. (ND13)

Solution: [Ref : Part A : Example 3 (AM11)]
1
6. Find the second divided difference with arguments a, b, c, if f (x) = . (AM14, ND10)
x
Solution: [Ref : Part A : Example 2 (ND10)]
7. Prove that ∆| 2 x3 = x + y + z.
yz
Solution : Given the function f (x) = x3 and the arguments are x, y, z.
om
f (y) − f (x) y3 − x3 (y − x)(x2 + xy + y2 )
∆| f (x) = = = = x2 + xy + y2
y y−x y−x y−x
Similarly, ∆| f (y) = y2 + yz + z2
z
∆| f (y) − ∆| f (x)
.c
2 z y y2 + yz + z2 − (x2 + xy + y2 )
Now, ∆| f (x) = =
yz z−x z−x
z − x2 + yz − xy (z + x)(z − x) + y(z − x)
2
= = = x+y+z
z−x z−x
ul
!
1 1
8. Show that ∆| 3
=−
abcd abcd
1 1
Solution : If f (x) = , f (a) =
x a
pa 1 1
1 f (b) − f (a) − 1
f (a, b) = ∆| = = b a =−
b a b−a b−a ab
jin
1 1 1 1
Similarly, f (b, c) = ∆| = − , f (c, d) = ∆| = −
c b bc d c cd
1 1 1 1
∆| − ∆| − +
1 c b bb bc ab −a + c 1
.re
f (a, b, c) = ∆| 2 = = = =
bc a c−a c−a abc(c − a) abc
1 1
Similarly, f (b, c, d) = ∆| 2 =
cd b bcd
1 1 1 1
∆| 2 − ∆| 2
w
1 b a −
= bcd abc
cd bc
∴ f (a, b, c, d) = ∆| 3 =
bcd a d−a d−a
a−d 1
= =−
w
abc(d − a) abcd
4.3.3 Part B
w
Example 4.7. Construct the divided difference table for the following data and find the
value f (2).
x 4 5 7 10 11 12
y = f (x) 50 102 296 800 1010 1224
Solution: Newton’s divided difference formula is
f (x) = f (x0 ) + (x − x0 ) f (x0 , x1 ) + (x − x0 ) (x − x1 ) f (x0 , x1 , x2 )

+ (x − x0 ) (x − x1 ) (x − x2 ) f (x0 , x1 , x2 , x3 ) + · · ·
Newton’s divided difference table is

2 3 4 5
x f (x) ∆| f (x) ∆| f (x) ∆| f (x) ∆| f (x) ∆| f (x)
4 50
102−50
om
= 52
5−4
97−52
5 102 = 15
7−4
296−102 14.2−15
= 97 = −0.133
7−5 10−4
168−97 −.617+.133
7 296 = 14.2 = −.069
10−5 11−4
800−296 10.5−14.2 −.158+.069
= 168 = −.617 = −.011
10−7 11−5 12−4
.c
210−168 −1.7+.617
10 800 = 10.5 = −.158
11−7 12−5
1010−800 2−10.5
= 210 = −1.7
11−10 12−7
214−210
11 1010 =2
ul
12−10
1224−1010
= 214
12−11
12 1224
pa
f (x) = 50 + (x − 4) (52) + (x − 4) (x − 5) (15) + (x − 4) (x − 5) (x − 7) (−0.133)
+ (x − 4) (x − 5) (x − 7) (x − 10) (−0.069)
+ (x − 4) (x − 5) (x − 7) (x − 10) (x − 11) (−0.011)
f (x = 2) = 50 + (2 − 4) (52) + (2 − 4) (2 − 5) (15) + (2 − 4) (2 − 5) (2 − 7) (−0.133)
jin
+ (2 − 4) (2 − 5) (2 − 7) (2 − 10) (−0.069)
+ (2 − 4) (2 − 5) (2 − 7) (2 − 10) (2 − 11) (−0.011)
= 47.19
.re
Example 4.8. If f (0) = 0, f (1) = 0, f (2) = −12, f (4) = 0, f (5) = 600, f (7) = 7308, find a polynomial
that satisfies this data using Newton’s devided difference formula. Hence find f (6), f (−1).
Solution: Hint : f (x) = x(x − 1)[x3 − 2x2 − 13x + 32]

f (6) = 2940
w
f (−1) = 84
Example 4.9. Find the third divided difference with arguments 2, 4, 9, 10 of the function f (x) =
w
x3 − 2x.
Solution: [Ref : Part A : Example 3, 5]
w
1. Find the function f (x) from the following table using Newton’s divided difference formula:
x: 0 1 2 4 5 7
(AM10)
f (x) : 0 0 −12 0 600 7308
h i
Solution: f (x) = x5 − 3x4 − 11x3 + 33x2 − 20x ⇒ f (6) = 2580
2. Given tha table
x: 5 7 11 13 17
f (x) : 150 392 1452 2366 5202
Evaluate f (9) using Newton’s divided difference formula. (AM11)

h i
Solution: f (x) = x3 − x2 + 24x − 70 ⇒ f (9) = 794
om
3. Determine f (x) as a polynomial in x for the following data, using Newton’s divided difference
formulae. Also find f (2). (ND11)
x: −4 −1 0 2 5
f (x): 1245 33 5 9 1335
h i
Solution: f (x) = 3x4 − 5x3 + 6x2 − 14x + 5 ⇒ f (2) = 9
.c
4. Use Newton’s divided difference formula to find f (x) from the following data. (MJ13)
x: 1 2 7 8
ul
y: 1 5 5 4
" #
1 3
Solution: f (x) = 3x − 58x2 + 321x − 224
pa 42
5. Find f (3) by Newton’s divided difference formula for the following data: (AM14, ND2004)
x: −4 −1 1 2 5
y: 1245 33 5 9 1335
h i
jin
Solution: f (x) = 3x4 − 5x3 + 6x2 − 14x + 5 ⇒ f (3) = 125
6. By using Newton’s divided difference formula find f (8), gives (ND14)
x 4 5 7 10 11 13
f (x) 48 100 294 900 1210 2028
.re
7. If f (0) = f (1) = 0, f (2) = −12, f (4) = 0, f (5) = 600 and f (7) = 7308, find a polynomial that satisfies
this data using Newton’s divided difference interpolation formula. Hence, find f (6). (MJ2007)
[Ans : f (6) = 2580]
w
8. Given the values
x 5 7 11 13 17
f (x) 150 392 1452 2366 5202
w
Evaluate f (9) using Newton’s divided difference formula. (N/D2007)

h i
Ans : f (x) = x3 − x2 + 24x − 70; f (9) = 794
w
9. Using Newton’s divided difference interpolation, find the polynomial of the given data
x −1 0 1 3
(AU Nov/Dec 2007)
f (x) 2 1 0 −1
" #
1 3
Ans : f (x) = x − 25x + 24
24
4.4 Cubic Splines
Interpolating with a cubic spline

The cubic spline interpolation formula is
1
S (x) = y(x) = y = [(xi − x)3 Mi−1 + (x − xi−1 )3 Mi ]
6h
1 h2 1 h2
+ (xi − x)[yi−1 − Mi−1 ] + (x − xi−1 )[yi − Mi ]
h 6 h 6
om
where
6
Mi−1 + 4Mi + Mi+1 = [yi−1 − 2yi + yi+1 ]
h2
n = number of data
i = number of intervals [ i.e., i = 1, 2, 3, , (n − 1)]
.c
h = length of interval = interval length.
′′
h ′′ ′′
i
Note : If Mi and yi values are not given, then assume M0 = Mn = 0 or y0 = yn = 0 , and find
ul
M1 , M2 , · · · , Mn−1 in 1st interval, 2nd interval, · · · , (n − 1)th interval value.
Note : Order of convergence of the cubic spline is 4.
4.4.1 Part A
pa
1. Define a cubic spline S (x) which is commonly used for interpolation. (AM10)
Solution: Definition(Cubic Spline Interpolation): Given a function f defined on [a, b] and
a = x0 < x1 < · · · < xn = b, a cubic spline interpolant S for f is a function that satisfies the following
jin
conditions:
1. For each j = 1, · · · , n, S (x) is a cubic polynomial, denoted by S j (x), on the subinterval [x j−1 , x j ].
2. S (x j ) = f (x j ) for each j = 0, 1, · · · , n.
.re
3. S j+1 (x j+1 ) = S j (x j+1 ) for each j = 0, 1, · · · , n − 2.
4. S ′j+1 (x j+1 ) = S ′j (x j+1 ) for each j = 0, 1, · · · , n − 2.
5. S ′′j+1 (x j+1 ) = S ′′j (x j+1 ) for each j = 0, 1, · · · , n − 2.
6. One of the following sets of boundary conditions is satisfied:

w
i. S ′′ (x0 ) = S ′′ (xn ) = 0 (natural or free boundary);
ii. S ′ (x0 ) = f ′ (x0 ) and S ′ (xn ) = f ′ (xn ) (clamped boundary).

w
2. Define cubic spline function. (ND11)

2
Solution : A function s ∈ C [a, b] is called a cubic spline on [a, b], if s is a cubic polynomial si in
w
each interval [xi , xi+1 ]. It is called a cubic interpolating spline if s(xi ) = yi for given values yi .
3. For cubic splines, what are the 4n conditions required to evaluate the unknowns. (MJ12)
Solution: We need 4n conditions to fix the coefficients:
1. S (x j ) = f (x j ) for each j = 1, · · · , n ⇒ ( n conditions)

2. S n?1 = yn , 1 condition ⇒ ( 1 condition)
3. S j (x j+1 ) = S j+1 (x j+1 ) for each j = 1, · · · , n − 1.⇒ ( n − 1 conditions)
4. S ′j (x j+1 ) = S ′j+1 (x j+1 ) for each j = 1, · · · , n − 1.⇒ ( n − 1 conditions)
5. S ′′j (x j+1 ) = S ′′j+1 (x j+1 ) for each j = 1, · · · , n − 1.⇒ ( n − 1 conditions)
⇒ Totally (4n − 2 conditions)
These are 4n − 2 conditions. We need two extra.
om
We can define two extra boundary conditions.
Natural Spline: M0 = S 0′′ (x0 ) = 0 and Mn = S n′′ (xn ) = 0. (2 conditions)
∴ 4n − 2 + 2 = 4n conditions required to evaluate the unknowns for cubic spline.
4. Define cubic spline. (ND12, ND11)
.c
5. What is a cubic spline?
Solution : A cubic spline which has continuous slope and curvature is called a cubic spline.
6. What is a natural cubic spline?
ul
Solution : A cubic spline fitted to the given data such that the end cubics approach linearity at
their extremities is called a natural cubic spline.
pa
7. State the conditions required for a natural cubic spline.
Solution : A cubic spline g(x) fits to each of the points is continuous and is continuous in slope
′′ ′′
and curvature such that M0 = S 0 = g0 (x0 ) = 0 and Mn = S n = gn−1 (xn ) = 0 is called a natural cubic
spline. Let us assume that (xi , yi ), i = 0, 1, 2, · · · , n are data points.
jin
8. What are the advantages of cubic spline fitting?
Solution : Cubic spline provide better approximation to the behavior of functions that have
abrupt local changes. Further, spline perform better than higher order polynomial
approximation.
.re
9. Write the end conditions on Mi (x) in natural cubic spline.

Solution : M0 (x) = 0, Mn (x) = 0.
10. Write the relation between the second derivatives Mi (x) in cubic splines with equal mesh spacing.
Solution :
w
6
Mi−1 + 4Mi + Mi+1 = [yi−1 − 2yi + yi+1 ], i = 1, 2, · · · , n − 1.
h2
w
Or
6
Mi−1 + 4Mi + Mi+1 = [ fi−1 − 2 fi + fi+1 ], i = 1, 2, · · · , n − 1
h2
w
4.4.2 Part B
Example 4.10. Find the cubic spline approximation for the function f(x) given by the data:
x 0 1 2 3
y = f (x) 1 2 33 244
with M0 = 0 = M3 . Hence estimate the value f (0.5), f (1.5), f (2.5). {AU2010}

Solution: We know that cubic spline interpolation formula for xi−1 ≤ x < xi , i = 1, 2, 3 is
1 h i
S i (x) = y (x) = y = (xi − x)3 Mi−1 + (x − xi−1 )3 Mi
6h
h2
" #
1
+ (xi − x) yi−1 − Mi−1
h 6
h2
" #
1
+ (x − xi−1 ) yi − Mi (1)
h 6
om
6
where Mi−1 + 4Mi + Mi+1

= 2 yi−1 − 2yi + yi+1 (2)
h
n = number of data = 4
i = number of intervals = 3 i.e., i = 1, 2, 3.
h = length of inteval = 1
.c
Given M0 = M2 = 0, so find M1 , M2 .
′′
Suppose Mi or yi values are not given, then assume M0 = M3 = 0 and find M1 , M2 .
To find M1 , M2
ul
6
When i = 1, (2) ⇒ M0 + 4M1 + M2 =

2
y0 − 2y1 + y2
1
⇒ 4M1 + M2 = 180
pa
⇒ 0 + 4M1 + M2 = 6 [1 − 2 (2) + 33]
(3)
6
jin

When i = 2, (2) ⇒ M1 + 4M2 + M3 = y1 − 2y2 + y3
12
⇒ M1 + 4M2 + 0 = 6 [2 − 2 (2) + 244]
⇒ 4M1 + M2 = 180 (4)
.re
Solving (3)&(4), (3) ⇒ 4M1 + M2 = 180

4 × (4) ⇒ 4M1 + 16M2 = 4320
i.e., (3) + 4 × (4) ⇒ −15M2 = 4140
w
⇒ M2 = 276
(3) ⇒ 4M1 = 180 − 276
⇒ M1 = −24
w
To find Cubic spline

When i = 1,
w
Cubic spline in xi−1 ≤ x ≤ xi

i.e., x0 ≤ x ≤ x1
i.e., 0 ≤ x ≤ 1
i.e., Cubic spline in 0 ≤ x ≤ 1 is

1 h i
y1 (x) = S 1 (x) = (x1 − x)3 M0 + (x − x0 )3 M1
6 (1)
" # " #
1 1 1 1
+ (x1 − x) y0 − M0 + (x − x0 ) y1 − M1
1 6 1 6
1h i
= (1 − x)3 (0) + (x − 0)3 (−24)
6
+ (1 − x) [1 − 0] + (x − 0) [2 − (−24)]
om
= −4x3 + (1 − x) + 6x
= −4x3 + 5x + 1
When i = 2,
.c
i.e., x1 ≤ x ≤ x2
i.e., 1 ≤ x ≤ 2
ul
1 h i
y2 (x) = S 2 (x) = (x2 − x)3 M1 + (x − x1 )3 M2
6 (1)
pa " # " #
1 1 1 1
+ (x2 − x) y1 − M1 + (x − x1 ) y2 − M2
1 6 1 6
1h i
= (2 − x)3 (−24) + (x − 1)3 (276)
6 " # " #
1 1
+ (2 − x) 2 − (−24) + (x − 1) 33 − (276)
6 6
jin
= −4 (2 − x)3 + 46 (x − 1)3 + 6 (2 − x) − 13 (x − 1)
= 50x3 − 162x2 + 162x − 53
When i = 3,
.re

i.e., x2 ≤ x ≤ x3
i.e., 2 ≤ x ≤ 3
w

1 h i
y3 (x) = S 3 (x) = (x3 − x)3 M2 + (x − x2 )3 M3
6 (1)
w
" # " #
1 1 1 1
+ (x3 − x) y2 − M2 + (x − x2 ) y3 − M3
1 6 1 6
1h i
= (3 − x)3 (276) + 0
w
6 " #
1
+ (3 − x) 33 − (276) + (x − 2) [244 − 0]
6

= 46 27 − x3 + 9x2 − 27x − 13 (3 − x) + 244x − 488
= −46x3 + 414x2 − 985x + 715
∴ Cubic spline is




 S 1 (x) = y1 (x) = −4x3 + 5x + 1, 0≤x≤1


S (x) = 
 3 2

 S 2 (x) = y2 (x) = 50x − 162x + 167x − 53, 1≤x≤2

 S 3 (x) = y3 (x) = −46x3 + 414x2 − 985x + 715, 2 ≤ x ≤ 3


When x = 0.5, y1 (x = 0.5) = S 1 (x = 0.5) = −4 (0.5)3 + 5 (0.5)2 + 1 = 3
om
When x = 1.5, y2 (x = 1.5) = S 2 (x = 1.5) = 50 (1.5)3 − 162 (1.5)2 + 167 (1.5) − 53 = 1.75
When x = 2.5, y3 (x = 2.5) = S 3 (x = 2.5) = −46 (2.5)3 + 414 (2.5)2 − 985 (2.5) + 715 = 121.25
x 1 2 3
Example 4.11. From the following table . Find cubic spline and
y = f (x) −8 −1 18
.c
compute y(1.5), y′ (1), y(2.5) and y′ (3).
Solution:

3
 S 1 (x) = y1 (x) = 3(x − 1) + 4x − 12, 1≤x≤2
ul



S (x) = 
 S 2 (x) = y2 (x) = 3(3 − x)3 + 22x − 48,

 2≤x≤3
&
y(x = 1.5) = S 1 (x = 1.5) = −

45 ′
8
′
pa
′
, y (x = 1) = S 1 (x = 1) = 4
′
y(x = 2.5) = S 2 (x = 2.5) = 7.375, y (x = 3) = S 2 (x = 3) = 22.
x 0 1 2 3
Example 4.12. Fit a natural cubic spline for the following data:
jin
y = f (x) 1 4 0 −2
{AU 2008}
Solution: Assume M0 = 0 = M3 .


 S 1 (x) = y1 (x) = −2x3 + 5x + 1, [0, 1]
.re




S (x) =  S 2 (x) = y2 (x) = 3x3 − 15x2 + 20x − 4,


 [1, 2]

 S 3 (x) = y3 (x) = −x3 + 9x2 − 28x + 28,


[2, 3]

w
1. If f (0) = 1, f (1) = 2, f (2) = 33 and f (3) = 244, find a cubic spline approximation, assuming M(0) =
M(3) = 0. Also, find f (2.5). (AM10)
Solution: Hint :
w




 S 1 (x) = y1 (x) = −4x3 + 5x + 1, x ∈ [0, 1]


S (x) = y(x) =  S 2 (x) = y2 (x) = 50x3 − 162x2 + 1670x − 53,


 x ∈ [1, 2]

 S 3 (x) = y3 (x) = −46x3 + 414x2 − 985x + 715,


x ∈ [2, 3]
w
3 2
f (x) = −46x + 414x − 985x + 715 x ∈ [2, 3]
f ′ (x) = −138x2 + 828x − 985
f ′ (x = 2.5) = −138(2.5)2 + 828(2.5) − 985
= 222.5
2. Find the natural cubic spline approximation for the function f (x) defined by the following data:
x: 0 1 2 3
f (x) : 1 2 33 244
(ND10)
Solution: Hint : Ref : Previous AU 1 (AM10) 



 S 1 (x) = y1 (x) = −4x3 + 5x + 1, x ∈ [0, 1]
om


S (x) = y(x) =  S 2 (x) = y2 (x) = 50x3 − 162x2 + 1670x − 53,


 x ∈ [1, 2]

 S 3 (x) = y3 (x) = −46x3 + 414x2 − 985x + 715, x ∈ [2, 3]


h i
f (x) = −46x3 + 414x2 − 985x + 715 ⇒ f (x = 2.5) = y(x = 2.5) = 121.25, ∵ x = 2.5 ∈ [2, 3]
3. Find the cubic spline approximation for the function y = f (x) from the following data, given that
y′0 = y′3 = 0. (AM11)
.c
x: −1 0 1 2
y: −1 1 3 35
ul
Solution: Hint : 



 S 1 (x) = y1 (x) = −2x3 − 6x2 − 2x + 1, x ∈ [−1, 0]


S (x) = y(x) =  S 2 (x) = y2 (x) = 10x3 − 6x2 − 2x + 1,

pa 
 x ∈ [0, 1]

 S 3 (x) = y3 (x) = −8x3 + 48x2 − 56x + 19,


x ∈ [1, 2]
4. The following values of x and y are given:
x: 1 2 3 4
jin
y: 1 2 5 11
Find the cubic splines and evaluate y(1.5) and y′ (3) (MJ12)
Solution: Hint :  1 3
x − 3x2 + 5x ,
.re


 S 1 (x) = y1 (x) = x ∈ [1, 2]


 3

 1 3
S (x) = y(x) =  x − 3x2 + 5x ,

S 2 (x) = y2 (x) = x ∈ [2, 3]


 3

 1
−2x3 + 24x2 − 76x + 81 , x ∈ [3, 4]

 S 3 (x) = y3 (x) =

3
1 3
y(x) = x − 3x2 + 5x ⇒ y(1.5) = 1.375, x ∈ [1, 2]
3
w
1 2
y′ (x) = 3x − 6x + 5 ⇒ y′ (3) = 4.666666667, x ∈ [2, 3]
3
(or)
w
1
y′ (x) = −6x2 + 48x − 76 ⇒ y′ (3) = 4.666666667, x ∈ [3, 4]
3
x: −1 0 1 2
5. Obtain the cubic spline for the following data to find y(0.5). (ND12)
w
y: −1 1 3 35
Solution: Hint : 



 S 1 (x) = y1 (x) = −2x3 − 6x2 − 2x + 1, x ∈ [−1, 0]


S (x) = y(x) =  S 2 (x) = y2 (x) = 10x3 − 6x2 − 2x + 1,


 x ∈ [0, 1]

 S 3 (x) = y3 (x) = −8x3 + 48x2 − 56x + 19,


x ∈ [1, 2]
6. Using cubic spline, compute y(1.5) from the given data. (MJ13)
x: 1 2 3
y: −8 −1 18
Solution: Hint :
S (x) = y(x) = 3x3 − 9x2 + 13x − 15
45
y(1.5) = 3(1.5)3 − 9(1.5)2 + 13(1.5) − 15 = − = −5.625, x ∈ [1, 2]
8
om
7. Find the natural cubic spline to fit the data:
x: 0 1 2
f (x) : −1 3 29
Hence find f (0.5) and f (1.5). (ND13)
.c
Solution: Hint :
11 3 3
S (x) = y(x) = x − x − 1, x ∈ [0, 1]
2 2
y(0.5) = −1.0625, x ∈ [0, 1]
ul
11 5
S (x) = y(x) = (2 − x)3 − (2 − x) + 29x − 29, x ∈ [1, 2]
2 2
pa y(1.5) = 13.9375, x ∈ [1, 2]
8. Fit the cubic splines for the following data. (16)

x: 1 2 3 4 5
(AM14)
y: 1 0 1 0 1
Solution : Assume M0 = 0 = M4 . 
S 1 (x) = y1 (x) = 2 − x, 1≤x≤2
jin




1



S 2 (x) = y2 (x) = [−5x3 + 45x2 − 123x + 106], 2≤x≤3



7


f (x) = S (x) = 

 1

 S 3 (x) = y3 (x) = [6x3 − 72x2 + 275x − 332], 3≤x≤4
7



1


S 4 (x) = y4 (x) = [−5x3 + 75x2 − 363x + 772],

4≤x≤5



7
.re
9. Obtain the cubic spline approximation for the function y = f (x) from the following data, given
that y′′0 = y′′3 = 0. (ND14)
x −1 0 1 2
w
y −1 1 3 35
10. Find the cubic Spline interpolation. (AU N/D, 2007,AM2014)

w
x 1 2 3 4 5
f (x) 1 0 1 0 1
11. Given the following table, find f (2.5) using cubic spline functions : (AU May/June 2007)
w
x 1 2 3 4
f (x) 0.5 0.3333 0.25 0.2

Ans: S 2 (2.5) = 0.2829

Solution :
12. Fit the st.line for the data.
x 0 1 2 3
f (x) 1 2 9 28
Solution :  4 4


 y1 (x) = x3 − x + 1, 0≤x≤1


 5 5

 1 3
f (x) =  − 18x2 + 19x − 1,

y (x) = [10x 1≤x≤2
om
2


 5
 y3 (x) = −2x3 + 102 x2 − 333 x + 159 ,



 2≤x≤3
5 5 5
4.5 Newton’s forward and backward difference formulae
Newton’s forward and backward difference formulae for Uniform (or) equal intervals only.
.c
Newton’s forward interpolation difference formula:

If y required x near to x0 =? and use ∆

ul
y(x) = f (x) = f (x0 + uh)
u u(u − 1) 2 u(u − 1)(u − 2) 3
= y0 + ∆y0 + ∆ y0 + ∆ y0 + · · ·
pa 1! 2! 3!
x − x0
where u = , h = length of interval.
h
4.6 Newton’s backward interpolation difference formula:

jin

If y required x near to xn =? and use ∇
y(x) = f (x) = f (xn + vh)

v v(v + 1) 2 v(v + 1)(v + 2) 3
.re
= yn + ∇yn + ∇ yn + ∇ yn + · · ·
1! 2! 3!
x − xn
where v = , h = length of interval.
h
w
4.6.1 Part A
1. When to use Newton’s forward interpolation and when to use Newton’s backward interpolation?
(ND10)
w
Solution :
Use of Newton’s forward interpolation : The formula is used to interpolate the values of
w
y near the begining of the table value and also for extrapolation the values of y short distance
ahead (to the left) of y0 .
Use of Newton’s backward interpolation : The formula is used to interpolate the values of y
near the end of the table value and also for extrapolation the values of y short distance ahead (to
the right) of y0 .
2. State Newton’s backward difference formula. (ND12)

Solution : Newton’s backward interpolation difference formula:

If y required x near to xn =? and use ∇

y(x) = f (x) = f (xn + vh)

v v(v + 1) 2 v(v + 1)(v + 2) 3
= yn + ∇yn + ∇ yn + ∇ yn + · · ·
1! 2! 3!
om
x − xn
where v = , h = length of interval.
h
3. State Newton’s forward interpolation formula. (MJ13)

Solution : Newton’s forward interpolation difference formula:

If y required x near to x0 =? and use ∆
.c
y(x) = f (x) = f (x0 + uh)
u u(u − 1) 2 u(u − 1)(u − 2) 3
= y0 + ∆y0 + ∆ y0 + ∆ y0 + · · ·
ul
1! 2! 3!
x − x0
where u = , h = length of interval
h
Solution :
pa
4. State Newton’s forward difference formula for equal intervals. (ND13)
[Ref : Part A : 3 (MJ13)]
5. State Newton’s backward formula for interpolation. (ND14,ND12)

jin
6. What advantage has Lagrange’s formula over Newton?
Solution : The forward and backward interpolation formulae of Newton can be used only when
the values of the independent variable x are equally spaced can also be used when the differences
.re
of the dependent variable y become smaller ultimately. But Lagrange’s interpolation formula can
be used whether the values of x, the independent variable are equally spaced or not and whether
the difference of y become smaller or not.
w
7. Derive Newton’s forward difference formula by using operator method. (or) Derive Gregory -
Newton forward difference interpolation formula.
w
Solution : Pn (x) = Pn (x0 + uh) = E u Pn (x0 ) = E u y0

= (1 + ∆)u y0
u u(u − 1) 2 u(u − 1)(u − 2) 3
w
= y0 + ∆y0 + ∆ y0 + ∆ y0 + · · ·
1! 2! 3!
x − x0
where u = .
h
8. Derive Newton’s backward difference formula by using operator method.

Solution :
Pn (x) = Pn (x0 + vh) = E v Pn (xn ) = E v yn

= (1 − ∇)−v yn where E = (1 − ∇)−1
!
v(v + 1) 2 v(v + 1)(v + 2) 3
= 1 + v∇ + ∇ + ∇ + · · · yn
2! 3!
v v(v + 1) 2 v(v + 1)(v + 2) 3
= yn + ∇yn + ∇ yn + ∇ yn + · · ·
1! 2! 3!
om
x − xn
where v =
h
4.6.2 Part B
Example 4.13. Using Newton’s forward interpolation formula, find the cubic polynomial
which takes the following values
.c
x 0 1 2 3
.
y = f (x) 1 2 1 10
ul
Evaluate f (4). (AU 2000, 2009)
Solution: WKT, Newton’s forward formula to find the polynomial in x.
There are only 4 data given. Hence the polynomial will be degree 3. Newton’s forward formula is
pa
y(x) = f (x) = f (x0 + uh)
u u(u − 1) 2 u(u − 1)(u − 2) 3
= y0 + ∆y0 + ∆ y0 + ∆ y0 + · · ·
1! 2! 3!
x − x0
where u =
h
jin
h = length of interval.
.re
w
w
w
Newton’s divided difference table is
x y = f (x) ∆y ∆2 y ∆3 y
x0 = 0 y0 = 1
2−1=1
x1 = 1 y1 = 2 −1 − 1 = −2
1 − 2 = −1 10 − (−2) = 12
x2 = 2 y2 = 1 9 − (−1) = 10
om
10 − 1 = 9
x3 = 3 y3 = 10
x x (x − 1) x (x − 1) (x − 2)
f (x) = 1 + (1) + (−2) + (12)
1! 2! 3!
.c
= 2x3 − 7x2 + 6x + 1
When x = 4,
ul
f (4) = 2 × 43 − 7 × 42 + 6 × 4 + 1 = 128 − 92 + 24 + 1 = 41
Example 4.14. The population of a city in a census takes once in 10 years is given below.
Estimate the population in the year 1955.
Year
Population in lakhs 35
1951
42
1961
58
1971
pa 1981
84
Solution: y(x = 1955) = f (x = 1955) = 36.784
jin
Example 4.15. From the table given below find sin 52◦ by using Newton’s forward
x 45◦ 50◦ 55◦ 60◦
interpolation formula.
y = sin x 0.7071 0.7660 0.8192 0.8660
Solution: y(x = 52) = sin 52◦ = 0.788 approximately.
.re
Example 4.16. From the data given below find the number of students whose weight is
Weight in lbs 0-40 40-60 60-80 80-100 100-120
between 60 and 70.
Number of students 250 120 100 70 50
Solution: Let weight be denoted by x and
Number of students be denoted by y, i.e., y = f (x).
w
Use Newton’s forward formula to find y where x lies between 60 − 70.

Newton’s forward formula is
w
y(x) = f (x) = f (x0 + uh)

u u(u − 1) 2 u(u − 1)(u − 2) 3
= y0 + ∆y0 + ∆ y0 + ∆ y0 + · · ·
1! 2! 3!
w
We rewrite the table as cumulative table showing the number of students less than x lbs.
x Below 40 Below 60 Below 80 Below 100 Below 120
y 250 370 470 540 590
Newton’s forward difference table is
x y = f (x) ∆y ∆2 y ∆3 y ∆4 y
Below 40 250
120
Below 60 370 −20
120 −10
Below 80 470 −30 20
om
70 10
Below 100 540 −20
50
Below 120 590
.c
x − x0 x − 40
Here u = = .
h 20
70 − 40
When x = 70, u = = 1.5
20
ul
1.5 1.5(1.5 − 1) 1.5(1.5 − 1)(1.5 − 2)
y(x = 70) = 250 + (120) + (−20) + (−10)
1! 2! 3!
1.5(1.5 − 1)(1.5 − 2)(1.5 − 3)
+ (20)
pa 4!
= 423.59 = 424
∴ Number of students whose weight is 70 = 424

∴ Number of students whose weight is between 60 − 70 = 424 − 370 = 54.
jin
Example 4.17. Use Newton’s backward difference formula to construct as interpolating
polynomial of degree 3 for the data. !
1
f (0.75) = 0.07181250, f (0.5) = 0.024750, f (0.25) = 0.33493750, f (0) = 1.10100. Find f − .
3
.re
!
1
Solution: y(x) = x3 + 4.001x2 + 4.002x + 1.101, f − = 0.174518518.
3
Example 4.18. From the following data, find θ at x = 43 and x = 84.

x 40 50 60 70 80 90
. Also express θ in terms of x.
θ 184 204 226 250 276 304
w
Solution: θ(x = 43) = 189.79 (by Newton’s forward formula)

w
θ(x = 84) = 286.96 (by Newton’s backward formula)

θ(x) = 0.01x2 + 1.1x + 124 (by Newton’s forward formula)
w
Example 4.19. Find a polynomial of degree two for the data by Newton’s forward difference
x 0 1 2 3 4 5 6 7
method : (AU May/June, 2007)
f (x) 1 2 4 7 11 16 22 29
" #
1 2
Solution: Ans : y(x) = x +x+2
2
1. Find the value of tan 45◦ 15′ by using Newton’s forward difference interpolation formula for
x◦ 45 46 47 48 49 50
◦
tan x 1.00000 1.03553 1.07237 1.11061 1.15037 1.19175
(ND10)
Ans: tan 45◦ 15′ = 1.00876, by Newton’s forward difference formula

Solution :
om
2. Derive Newton’s backward difference formula by using operator method. (MJ12)
3. Find the value of y when x = 5 using Newton’s interpolation formular from the following table:
x: 4 6 8 10
.c
y: 1 3 8 16
(ND12)

Solution : y(x = 5) = 1.625, by Newton’s forward difference formula
ul
4. Fit a polynomial, by using Newton’s forward interpolation formula, to the data given below. (8)
x:pa 0 1 2 3
y: 1 2 1 10
(MJ13)
Solution :
y(x) = 2x3 − 7x2 + 6x + 1, by Newton’s forward & backward formula
jin
y(x = 4) = 41
5. Using Newton’s forward interpolation formula, find the cubic polynomial which takes the
following values: (ND14)
.re
x 0 1 2 3
f (x) 1 2 1 10
w
w
w
4.6.4 Assignment problems
1. Use Lagrange’s formula to find a polynomial which takes the values f (0) = −12, f (1) = 0, f (3) = 6
and f (4) = 12. Hence find f (2). (AM10)
2. Find the value of x when y = 20 using Lagrange’s formula from the following table. (AM11)
x: 1 2 3 4
1 8 27 64
om
y = f (x) :
3. Using Lagrange’s interpolation, calculate the profit in the year 2000 from the following data:
Year: 1997 1999 2001 2002

(MJ12)
Profit in Lakhs Rs. : 43 65 159 248
.c
4. Use Lagrange’s method to find log10 656, given that log10 654 = 2.8156, log10 658 = 2.8182, log10 659 =
2.8189, and log10 661 = 2.8202. (ND12)
ul
5. Find the function f (x) from the following table using Newton’s divided difference formula:
x: 0 1 2 4 5 7
(AM10)
f (x) : 0 0 −12 0
pa 600 7308
6. Given tha table
x: 5 7 11 13 17
f (x) : 150 392 1452 2366 5202
jin
Evaluate f (9) using Newton’s divided difference formula. (AM11)
7. Find f (3) by Newton’s divided difference formula for the following data: (AM14, ND2004)
.re
x: −4 −1 1 2 5
y: 1245 33 5 9 1335
8. By using Newton’s divided difference formula find f (8), gives (ND14)

w
x 4 5 7 10 11 13
f (x) 48 100 294 900 1210 2028
w
9. If f (0) = 1, f (1) = 2, f (2) = 33 and f (3) = 244, find a cubic spline approximation, assuming M(0) =
M(3) = 0. Also, find f (2.5). (AM10)
10. Find the natural cubic spline to fit the data:

w
x: 0 1 2
f (x) : −1 3 29
Hence find f (0.5) and f (1.5). (ND13)

11. Fit the cubic splines for the following data. (AM14)
x: 1 2 3 4 5
y: 1 0 1 0 1
12. Obtain the cubic spline approximation for the function y = f (x) from the following data, given
that y′′0 = y′′3 = 0. (ND14)
x −1 0 1 2
om
y −1 1 3 35
13. Find the value of tan 45◦ 15′ by using Newton’s forward difference interpolation formula for
x◦ 45 46 47 48 49 50
(ND10)
◦
tan x 1.00000 1.03553 1.07237 1.11061 1.15037 1.19175
.c
14. Find the value of y when x = 5 using Newton’s interpolation formular from the following table:
x: 4 6 8 10
(ND12)
ul
y: 1 3 8 16
15. Fit a polynomial, by using Newton’s forward interpolation formula, to the data given below.
x:
y:
0
1
1
2
2
1
3
10
pa (MJ13)
16. Using Newton’s forward interpolation formula, find the cubic polynomial which takes the
following values: (ND14)
jin
x 0 1 2 3
f (x) 1 2 1 10
.re
w
w
w
4.7 Numerical Differentiation and Integration
Numerical Differentiation Numerical Integration

Newton’s forward difference formula Trapezoidal rule
to compute derivatives (Equal interval) [n = 1 in Quadrature formula]
Newton’s backward difference formula Simpson’s one third rule
to compute derivatives(Equal interval) [n = 2 in Quadrature formula]
om
Lagrange’s Interpolation formula Simpson’s three eighth rule
(Equal or unequal intervals) [n = 3 in Quadrature formula]
Newton’s divided difference Interpolation Romberg’s method
formula (Equal or unequal intervals)
Maxima and minima of a tabulated function Two point Gaussian’s quadrature formula
Three point Gaussian’s quadrature formula
.c
Double integrals by Trapezoidal rule
Double integrals by Simpson’s 1/3 rule
ul
4.8 Approximation of derivatives using interpolation polynomials
Numerical Differentiation
Given y = y(x) = f (x) [in a table]
dy
pa
= y ′ (x) = f ′ (x) is the first numerical derivative
dx
d2 y
= y ′′ (x) = f ′′ (x) is the second numerical derivative
dxn 2
jin
d y
= y1···n (x) = f 1···n (x) is the nth numerical derivative
dxn
4.8.1 Newton’s forward difference formula to compute derivative

.re
WKT, Newton’s forward difference interpolation formula is
u u (u−1) 2 u (u−1) (u−2) 3 u (u−1) (u−2) (u−3) 4

y (x) = f (x0 +uh) = y0 + ∆y0 + ∆ y0 + ∆ y0 + ∆ y0 +· · ·
1! 2! 3! 4!
u u2 −u 2 u3 −3u2 +2u 3 u4 −6u3 +11u2 −6u 4
= y0 + ∆y0 + ∆ y0 + ∆ y0 + ∆ y0 +· · ·
1! 2! 3! 4!
w
x − x0
where u =
h
First derivative
w
!
dy dy du dy 1 1 dy
= = =
dx du dx du h h du
w
3u2 −6u+2 3 4u3 −18u2 +22u−6 4

" ! ! ! #
dy 1 2u−1 2
i.e., = ∆y0 + ∆ y0 + ∆ y0 + ∆ y0 + · · ·
dx h 2 6 24
" # " #
dy 1 1 1 1 1
  = ∆y0 − ∆2 y0 + ∆3 y0 − ∆4 y0 + ∆5 y0 − · · ·
dx  at x = x0  h 2 3 4 5
   
⇒u=0
 
Second derivative
d2 y 1 d2 y
" # " # " # " # " # " #
d dy d dy du d 1 dy d du 1 dy d 1 1 dy 1 d 1 dy
= = = = = = =
dx2 dx dx dx du dx dx h du du dx h du du h h du h du h du h2 du2
2
" ! #
1 12u − 36u + 22 4
= 2 ∆2 y0 + (u − 1) ∆3 y0 + ∆ y0 + · · ·
h 24
" 2 # " #
d y 1 2 22 5
2
  =
2
∆ y0 − ∆3 y0 + ∆4 y0 − ∆5 y0 + · · ·
dx  at x = x0  h
 
24 6
 
om
 
⇒u=0
 
Third derivative
d3 y
" ! #
1 3 24u − 36 4
= ∆ y0 + ∆ y0 + · · ·
dx3 h3 24
" 3 # " #
d y 1 3 36
.c
  = ∆ y0 − ∆4 y0 + · · ·
dx3  at x = x0  h3 24
  
⇒u=0
 
ul
4.8.2 Newton’s backward difference formula to compute derivatives
WKT, Newton’s backward difference interpolation formula is
y (x) = f (xn + vh) = yn +

v
1!
∇yn +
v (v + 1) 2
2!
∇ yn +
v (v + 1) (v + 2) 3
3!
pa
∇ yn +
v (v + 1) (v + 2) (v + 3) 4
4!
∇ yn + · · ·
v2 + v 2 v3 + 3v2 + 2v 3 v4 + 6v3 + 11v2 + 6v 4
! ! !
v
= yn + ∇yn + ∇ yn + ∇ yn + ∇ yn + · · ·
1! 2! 3! 4!
jin
x − xn
where v =
h
First derivative
!
.re
dy dy dv dy 1 1 dy
= = =
dx dv dx dv h h dv
3v2 + 6v + 2 3 4v3 + 18v2 + 22v + 6 4
" ! ! ! #
1 2v + 1 2
= ∇yn + ∇ yn + ∇ yn + ∇ yn + · · ·
h 2 3! 4!
" # " #
dy 1 1 2 6
  = ∇yn + ∇2 yn + ∇3 yn + ∇4 yn + · · ·
dx  at x = xn  h 2 3! 4!
w
   
⇒v=0
 
Second derivative
w
d2 y 1 d2 y
" # " # " # " # " # " #
d dy d dy dv d 1 dy d dv 1 dy d 1 1 dy
= = = = = =
dx2 dx dx dx dv dx dx h dv dv dx h dv dv h h dv h2 dv2
w
d2 y 12v2 + 36v + 22 4
" ! ! #
1 6v + 6 3
2
= 2 ∇2 yn + ∇ yn + ∇ yn + · · ·
dx h 3! 4!
" 2 # " #
d y 1 2 6 3 22 4
  = ∇ yn + ∇ yn + ∇ yn + · · ·
dx2  at x = xn  h2 3! 4!
  
⇒v=0
 
Third derivative
d3 y
" ! #
1 6 3 24v + 36 4
= ∇ yn + ∇ yn + · · ·
dx3 h3 3! 4!
" 3 # " #
d y 1 6 3 36 4
  = ∇ yn + ∇ yn + · · ·
dx3  at x = xn  h3 3! 4!
  
⇒v=0
 
4.8.3 Maxima and Minima of a tabulated function
om
(If the intervals are same)
WKT, Newton’s forward difference interpolation formula is

u u (u − 1) 2 u (u − 1) (u − 2) 3
y = y0 + ∆y0 + ∆ y0 + ∆ y0 + · · ·
1! 2! 3!
.c
First derivative
3u2 − 6u + 2 3
" #
dy 1 2u − 1 2
= ∆y0 + ∆ y0 + ∆ y0 + · · · (1)
dx h 2! 3!
ul
substitute h, ∆y0 , ∆2 y0 , ∆3 y0 , ∆4 y0 , · · · gives the equation
dy
pa = an equation in u (2)
dx
For maxima or minima is obtained by the equation by
dy
= 0 (or) y ′ (x) = 0
dx
jin
Put RHS of (1) = 0, find u = u1 , u2 , · · ·
d2 y
Find 2 from (2):
dx
d2 y
!
Suppose = −ve ⇒u1 is maximum point
dx2 at u1
.re
⇒y (u1 ) = maximum value of y
d2 y
!
Suppose = +ve ⇒u2 is minimum point
dx2 at u2
w
⇒y (u2 ) = minimum value of y
4.8.4 Part A
w
1. Write down the formulae for finding the first derivative using Newton’s forward difference at
x = x0 Newton’s backward difference at x = xn . (ND2010)
w
Solution : First derivative by Newton’s forward difference interpolation formula is

" # " #
dy 1 1 2 1 3 1 4
  = ∆y0 − ∆ y0 + ∆ y0 − ∆ y0 + · · ·
dx  at x = x0  h 2 3 4
 
 
 ⇒u=0 
First derivative by Newton’s backward difference interpolation formula is

" # " #
dy 1 1 2 2 3 6 4
  = ∇yn + ∇ yn + ∇ yn + ∇ yn + · · ·
dx  at x = xn  h 2 3! 4!
 
 
 ⇒v=0 
dy d2 y
2. Write down the expression for and 2 at x = xn by Newton’s backward difference formula.
dx dx
(AM2014)
om
First derivative by Newton’s backward difference interpolation formula is
" # " #
dy 1 1 2 2 3 6 4
  = ∇yn + ∇ yn + ∇ yn + ∇ yn + · · ·
dx  at x = xn  h 2 3! 4!
   
⇒v=0
 
.c
Second derivative by Newton’s backward difference interpolation formula is
d2 y
" # " #
1 2 6 3 22 4
  = ∇ yn + ∇ yn + ∇ yn + · · ·
dx2 
 at x = xn  h2 3! 4!
ul
 

⇒ v = 0 
 
4.8.5 Part B pa
Example 4.20. The following data gives the velocity of a particle for 20 seconds at an
interval of 5 seconds.
Time(sec) 0 5 10 15 20
Velocity (m/sec) 0 3 14 69 228
jin
Find (a) Initial acceleration using the entire data (b) Final acceleration.
Solution: The difference table is
Time t = x Velocity v = y(x) ∆2 y (x) ∆3 y (x) ∆4 y (x)

.re
∆y (x)
0 (= x0 ) 0 (= y0 )
3 (= ∆y0 )

5 3 8 = ∆2 y0

11 36 = ∆3 y0
w

10 14 44 24 = ∆4 y0 or ∇4 yn

55 60 = ∇3 yn

15 69 104 = ∇2 yn
w
159 (= ∇yn )
20 228 (= yn )
w
Here x0 = 0, h = interval length = 5

dv
(a) WKT, acceleration = = rate of change of velocity
dt ! !
dv dv
To find initial acceleration, put = .
dt t=t0 dt t=0
i.e., initial acceleration exists at t = 0 = x0 [which is nearer to beginning of the table], so we use
Newton’s forward difference formula for first derivative.

∴ Newton’s forward difference interpolation formula is
u u (u − 1) 2 u (u − 1) (u − 2) 3
y (x) = f (x0 + uh) = y0 + ∆y0 + ∆ y0 + ∆ y0 + · · ·
1! 2! 3!
Now, Newton’s forward difference formula for first derivative at x = x0 = 0 [⇒ u = 0]
! " #
dv 1 1 1 1
i.e., y ′ (x0 ) = = ∆y0 − ∆2 y0 + ∆3 y0 − ∆4 y0 + · · · [Here h = 5]
dt t=0 h 2 3 4
om
" #
1 1 1 1
= 3 − (8) + (36) − (24) = 1
5 2 3 4
⇒ y ′ (0) = 1
∴ Initial acceleration (acceleration when t = 0) is 1 m/sec2 .

(b) Final acceleration exists at t = 20 = x4 [Nearer to ending of table], so use Newton’s backward
.c
! !
dv dv
difference interpolation formula for first derivative, put = .
dt t=tn dt t=20
WKT, Newton’s backward difference interpolation formula is
v v (v + 1) 2 v (v + 1) (v + 2) 3
ul
y (x) = f (xn + vh) = yn + ∇yn + ∇ yn + ∇ yn
1! 2! 3!
v (v + 1) (v + 2) (v + 3) 4
+ ∇ yn + · · ·
" 4! #
pa 1 1 2 6
y ′ (xn ) = ∇yn + ∇2 yn + ∇3 yn + ∇4 yn + · · ·
h 2 3! 4!
" #
1 1 2 6
= 159 + (104) + (60) + (24)
5 2 6 24
1
= (237) = 47.2 m / sec2
5
jin
Example 4.21. Consider the following table of data :
x 0.2 0.4 0.6 0.8 1.0

f (x) 0.9798652 0.9177710 0.8080348 0.636093 0.3843735
.re
Find (1) f ′ (0.25) using Newton’s forward difference approximation

(2) f ′ (0.95) using Newton’s backward difference approximation
Solution: Here h = 0.2
x − x0 x − 0.2 0.25 − 0.2
(1) u = = = = 0.25
w
h 0.2 0.2
∴ f ′ (0.25) = −0.25828
w
x − xn 0.95 − 1
(2) u = = = −0.25
h 0.2
f ′ (0.95) = −1.367948
w
Example 4.22. Find the value of f ′ (8), f ′′

(9), maximum and minimum value from the
following data, using an approximate interpolation formula.
x 4 5 7 10 11
f (x) 48 100 294 900 1210
Solution: The values of x are unequally spaced.

To find f (x), we use Newton’s divided difference formula (or) Lagrange formula.
WKT, Newton’s Divided difference formula is
f (x) = f (x0 ) + (x − x0 ) f (x0 , x1 ) + (x − x0 ) (x − x1 ) f (x0 , x1 , x2 ) + (x − x0 ) (x − x1 ) (x − x2 ) f (x0 , x1 , x2 , x3 )

+ (x − x0 ) (x − x1 ) (x − x2 ) (x − x3 ) f (x0 , x1 , x2 , x3 , x4 ) + · · · (1)
x f (x) ∆| f (x) ∆| 2 f (x) ∆| 3 f (x) ∆| 4 f (x)
om
4 48
52
5 100 15
97 1
7 294 21 0
.c
202 1
10 900 27
310
ul
11 1210
h i
pa
∴ (1) ⇒ f (x) = 48 + (x − 4)(52) + (x − 4)(x − 5)(15) + (x − 4)(x − 5)(x − 7)(1)
= 48 + 52x − 208 + 15 x2 − 9x + 20 + x3 − x2 (16) + x(83) − 140
= x3 + x2 (−16 + 15) + x (83 − 135 + 52) − 208 + 300 − 140 + 48
= x3 − x2 + x(0) + 0
jin
∴ f (x) = x3 − x2
f ′ (x) = 3x2 − 2x, f ′ (8) = 176
′′ ′′
f (x) = 6x − 2, f (9) = 52
.re
To find maximum and minimum

2
Put f ′ (x) = 0 ⇒ 3x2 − 2x = 0 ⇒ x = 0, x =
3
′′
At x = 0, f (x = 0) = 6 (0) − 2 = −2 < 0
w
⇒ x = 0 is a maximum
! point
! & maximum value is f (x = 0) = 0.
2 ′′ 2 2
At x = , f x= =6 − 2 = 2 > 0.
3 3 3 !
2 2 4
w
⇒ x = is a minimum point & minimum value is f x = =−

3 3 27
Example 4.23. Evaluate y ′ and y ′′ at x = 2 given
w
x 0 1 3 6
y 18 10 −18 40
Solution: " #
70 2 15 187 ′′ 22
Ans : y (x) = x3 − x − x + 18, y ′ (x = 2) = − , y (x = 2) = −
9 9 9 9
Example 4.24. Find the value of cos(1.747) using the values given in the table below :
x 1.70 1.74 1.78 1.82 1.86

sin x 0.9916 0.9857 0.9781 0.9691 0.9584

Ans : − 0.175

Solution:
Example 4.25. Find sec 31◦ from the following data :
θ 31 32 33 34
om
tan θ 0.6008 0.6249 0.6494 0.6745
π
Solution: Ans : sec2 31 = 1.3835 ⇒ sec 31 = 1.174, Hint : 1◦ = = 0.017453292
180
.c
1. Given the following data, find y ′ (6) and the maximum value of y (if it exists). (AM10)
x: 0 2 3 4 7 9
y: 4 26 58 112 466 922
ul
2. Find f ′ (x) at x = 1.5 and x = 4.0 from the following data using Newton’s formulae for
differentiation. pa
x: 1.5 2.0 2.5 3.0 3.5 4.0
(MJ12)
y = f (x) : 3.375 7.0 13.625 24.0 38.875 59.0
3. Find the first three derivatives of f (x) at x = 1.5 by using Newton’s forward interpolation formula
to the data given below.
jin
x: 1.5 2 2.5 3 3.5 4
(MJ13)
y: 3.375 7 13.625 24 38.875 59
Numerical integration
.re
Newton-Cote’s quadrature formula is

a+nh
n (n − 2)2 3
Z " #
n n (2n − 3) 2
f (x)dx = nh y0 + ∆y0 + ∆ y0 + ∆ y0 + · · ·
2 12 24
a
w
where n = number of intervals

h = length of interval
w
4.9 Numerical integration using Trapezoidal

w
Trapezoidal rule(n = 1 in quadrature formula)

h


(y0 + yn ) + 2 (y1 + y2 + y3 + · · · + yn−1 )




x0R+nh



 2
f (x)dx = 

 (or)
x0


 h
first term + last term + 2 remaining terms




2
4.9.1 Part A
1. Write down the Newton-Cote’s formula for the equidistant ordinates. (MJ2011)
Solution : The general Newton-Cote’s quadrature formula is
a+nh
n (n − 2)2 3
" #
R n∆ n (2n − 3) 2
f (x) dx = nh y0 + y0 + ∆ y0 + ∆ y0 + · · ·
a 2 12 24
om
This is known as the general Gauss-Lagendre integration formula. Putting n = 1 and omitting
second and hyper differences in the above, we get
a+h
R h h
f (x) dx = f (a) + f (a + h) = (y1 + y2 )
a 2 2
.c
which is the trapezoidal rule.
R1
dx
2. Evaluate using Trapezoidal rule. (ND2012)
ul
0 1 + x2
1 1
Solution : Let y = y(x) = f (x) = 2
, h = 0.2 =
1+x 5
y=
x
1
1 + x2
0
1
0.2
0.96154
pa 0.4
0.86207
0.6
0.73529
0.8
0.60976
1
0.5
y0 y1 y2 y3 y4 y5
jin
h




xZ
0 +nh





 2
f (x)dx = 


 (or)
x0


 h
(y0 + yn ) + 2 (y1 + y2 + y3 + · · · + yn−1 )
.re



2
h
= (y0 + y5 ) + 2 (y1 + y2 + y3 + y4 )
2
0.2
= [(1 + 0.5) + 2 (0.96154 + 0.86207 + 0.73529 + 0.60976)]
2
≡ 0.78373
w
Rπ
3. Evaluate sin xdx by Trapezoidal rule by dividing ten equal parts. (AM2013)
w
0
Solution : [1.9843]
R6 dx
4. Evaluate using Trapezoidal rule. (ND2013)
w
0 1 + x2
Solution : [1.41079950]
R2 dx
5. Taking h = 0.5, evaluate using Trapezoidal rule. (AM2014)
1 1 + x2
Solution : [0.3289]
4.9.2 Part B
R1 1
Example 4.26. Using Trapezoidal rule, evaluate dx by taking eight equal intervals.
−1 1 + x2
(MJ13)
1
Solution: Let y = y(x) = f (x) =
1 + x2
Lenth of the given interval [a, b] = b − a = 1 − (−1) = 2
2
om
Lenth of the 8 equal intervals = h = = 0.25
8
1
Form the table for the ordinates of the function y = y(x) = f (x) =
1 + x2
x −1 −0.75 −0.5 −0.25 0 0.25 0.5 0.75 1

1
y= 0.5 0.64 0.8 0.94118 1 0.94118 0.8 0.64 0.5
.c
1 + x2
y0 y1 y2 y3 y4 y5 y6 y7 y8
h

ul


xZ0 +nh





2
f (x)dx = 


(or)
x0


h


 (y0 + yn ) + 2 (y1 + y2 + y3 + · · · + yn−1 )
2 pa
h
= (y0 + y8 ) + 2 (y1 + y2 + y3 + y4 + y5 + y6 + y7 )
2
0.2
= [(0.5 + 0.5) + 2 (0.64 + 0.8 + 0.94118 + 1 + 0.94118 + 0.8 + 0.64)]
2
≡ 1.56559
jin
R1 1
1. Using Trapezoidal rule, evaluate dx by taking eight equal intervals. (MJ13)
−1 1 + x2
.re
Solution: [1.5656]
4.10 Numerical integration using Simpson’s 1/3 rule
Simpson’s one third rule (Simpson’s 1/3 rule) (n = 2 in quadrature form)

w
 h


 (y0 + yn ) + 4 (y1 + y3 + y5 + · · ·) + 2 (y2 + y4 + y6 + · · ·)
3





(or)


w
x0R+nh




f (x) dx = 
  
  first term + last term 
x0 h 


  

odd suffices

+4

  
3 


 


+2 even suffices
 
w
4.10.1 Part A
1 1
1. When do you apply Simpson’s rule, and what is the order of the error in Simpson’s rule.
3 3
(MJ2011)
Solution : Let n = number of intervals

1
Simpson’s rule : The number of ordinates is odd (or) the intervals number is even.
3
The error in the Simpson’s 1/3 rule is of the order h4 .
2. State Simpson’s one-third rule. (ND2011,AM2013)

Solution : Let DC be the curve y = f (x) and DA, CB be the terminal ordinates. Let OA = a and
OB = b. Divide AB into even number(say 2n) of equal parts, equal to h. Let x1 , x2 , · · · , x2n+1 be the
abscissae of the points, A, A1 , B and y1 , y2 , · · · , y2n+1 be the corresponding ordinates. Then
om
Zb
h
ydx is approximately = (A + 4B + 2C) (1)
3
a
where A = y1 + y2n+1 = sum of the first and last ordinates.

B = y2 + y1 + · · · + y2n = sum of the even ordinates and
.c
C = y3 + y5 + · · · + y2n−1 = sum of the remaining ordinates (1)
known as Simpson’s rule or Simpson’s one - third rule.
ul
1
3. State the local error term in Simpson’s rule. (ND14)
3
4.10.2 Part B
pa
Example 4.27. A curve passes through the points (1, 2), (1.5, 2.4), (2, 2.7), (2.5, 2.8), (3, 3), (3.5, 2.6)
& (4, 2.1). Obtain Area bounded by the curve, x axis between x = 1 and x = 4. Also find the
volume of solids of revolution by revolving this area about x− axis.
jin
Solution:
Zb Z4
WKT, Area = ydx = ydx
a 1
Z4
.re
h
Simpson’s 1/3 rule, Area = ydx = (y0 + y6 ) + 4 (y1 + y3 + y5 ) + 2 (y2 + y4 )
3
1
= 7.783
Zb Z4
2
Volume = π y dx = π y2 dx
w
a 1
2
To find y :
w
x 1 1.5 2 2.5 3 3.5 4

y 2 2.4 2.7 2.8 3 2.6 2.1
y2 4 5.76 7.29 7.84 9 6.76 4.41
w
Z4 ( h i)
2 h 2 2

2 2 2

2 2
∴ Volume = π y dx = π y + y6 + 4 y1 + y3 + y5 + 2 y2 + y4 [by Simpson’s 1/3 rule]
3 0
1
= 64.07 cubic units

1. The velocity υ of a particle at a distance S from a point on its path is given by the table below:
S (meter) 0 10 20 30 40 50 60
υ (m / sec) 47 58 64 65 61 52 38
Estimate the time taken to travel 60 meters by Simpson’s 1/3rd rule and Simpson’s 3/8th rule.
om
(AM10)

Solution: By S(1/3) : I = 1.06338, By S(3/8) : I = 1.06425
4.11 Numerical integration using Simpson’s 3/8 rule
.c
Simpson’s three eighth rule (Simpson’s 3/8 rule) (n = 3 in quadratic form)
  



  (y0 + yn ) 




 3h 
 +2 (y3 + y6 + · · ·)




ul



 8 




+3 (y1 + y2 + y4 + y5 + y7 + · · ·)



x0R+nh




f (x) dx = 

 (or)
x0


  



 pa  first term + last term 
3h 


 

 +2 suffices with a multiple of 3


 
8


 
 
 
+3 remaining terms


 
4.11.1 Part A
jin
1. Under what condition, Simpson’s 3/8 rule can be applied and state the formula. (AM2012)
Solution: Condition : The number of intervals should be multiple of three.
  


  (y0 + yn ) 
.re



 3h  


  +2 (y3 + y6 + · · ·) 



 8  

+3 (y1 + y2 + y4 + y5 + y7 + · · ·)



x0R+nh




f (x) dx = 

 (or)
x0


  



  first term + last term 
3h 


 
w
 +2 suffices with a multiple of 3


 

8 


 
 
+3 remaining terms



w
4.11.2 Part B
R2 dx
Example 4.28. Compute the value of using Simpson’s 3/8 rule
1 x
w
1
Solution: Let y = f (x) = ,
x
h = 1/3
The tabulated values of y = f (x) are

x0 x1 x2 x3
x 1 4/3 5/3 6/3 = 2
1 1 1 1
f (x) =1 = 3/4 = .75 = 0.6 = 0.5
1 4/3 5/3 2
y0 y1 y2 y3
WKT, Simpson’s 3/8 rule is
Zb
om
3h
f (x) dx = (y0 + y4 ) + 3 (y1 + y2 ) + 2 (y3 )
8
a
Z2 !
1 3 (1/3)
i.e., dx = [(1 + 0.5) + 3 (0.75 + 0.6) + 2 (0)]
x 8
1
= 0.69375
.c
By actual integration,
Z2
ul
1
dx = loge x 21 = (ln 2 − ln 1) = 0.69315

x
1
4.11.3 Single integrals by Trapezoidal, Simpson 1/3 & 3/8
Example 4.29. Compute the value of

R2 dx
using
pa
1 x
(a) Trapezoidal rule (b) Simpson’s 1/3 rule (c) Simpson’s 3/8 rule
jin
Solution: Here h = 0.25,
1
y = f (x) =
x
The tabulated values of y = f (x) are

.re
x0 x1 x2 x3 x4
x 1 1.25 1.50 1.75 2
1 1 1 1 1
f (x) =1 = 0.8 = 0.66667 = 0.57143 = 0.5
1 1.25 1.5 1.75 1.2
y0 y1 y2 y3 y4
w
(a) WKT, Trapezoidal rule is
Zb xZ
0 +nh
w
h
f (x) dx = f (x) dx = (y0 + y4 ) + 2 (y1 + y2 + y3 )
2
a x0
Z2
w
!
1 0.25
⇒ dx = [(1 + 0.5) + 2 (0.8 + 0.66667 + 0.57143)]
x 2
1
0.25 0.25
= [1.5 + 4.0762] = [5.5762]
2 2
= 0.697025
(b) WKT, Simpson’s 1/3 rule is
Zb
h
f (x) dx = (y0 + yn ) + 4 (y1 + y3 ) + 2 (y2 )
3
a
Z2 !
1 0.25
⇒ dx = [(1 + 0.5) + 4 (0.8 + 0.57143) + 2 (0.66667)]
x 3
1
= 0.693255
om
(c) WKT, Simpson’s 3/8 rule is
Zb
3h
f (x) dx = (y0 + y4 ) + 3 (y1 + y2 ) + 2 (y3 )
8
a
.c
Z2 !
1 3 (0.25)
i.e., dx = [(1 + 0.5) + 3 (0.8 + 0.66667) + 2 (0.57143)]
x 8
1
= 0.66024
ul
pa
Z2
1
1
x
dx = loge x 21 = (ln 2 − ln 1) = 0.69315

R10 dx
Example 4.30. Evaluate by dividing the range into 8 equal parts by (a) Trapezoidal
jin
0 1+ x
rule (b) Simpson’s 1/3 rule (c) Simpson’s 3/8 rule
1
Solution: Here h = 1.25, y = f (x) =
1+x
R10 dx
(a) I = 2.51368 , (b) I = 2.42200 , (c) I = 2.41838 , Actual integration, = 2.39790.
0 1+x
.re
R1
Example 4.31. Evaluate x e x dx taking 4 equal intervals by (a) Trapezoidal rule (b)
0
Simpson’s 1/3 rule (c) Simpson’s 3/8 rule
Solution: (a) 1.02307 , (b) 1.00017 , (c) 0.87468 , AI = 1
w
Rπ
Example 4.32. Calculate sin3 xdx taking 7 ordinates (6 intervals) using a) Trapezoidal rule
0
(b) Simpson’s 1/3 rule (c) Simpson’s 3/8 rule
w
Solution: (a)1.33467 , (b)1.32612 , (c)1.30516 , AI = 1.3333.

w
S (meter) 0 10 20 30 40 50 60
υ (m / sec) 47 58 64 65 61 52 38
(AM10)

S(1/3) : I = 1.06338, S(3/8) : I = 1.06425

Solution:
R6
1
2. Evaluate I = dx by using (i) direct integration (ii) Trapezoidal rule (iii) Simpson’s one-third
0 1 + x
rule (iv) Simpson’s three-eighth rule. (ND11)

Solution: By Dir. Int. : I = 1.9459, Trap. : I = 2.022, S(1/3) : I = 1.9587, S(3/8) : I = 1.966
om
π/2
R
3. Compute sin xdx using Simpson’s 3/8 rule. (ND12)
0
S(3/8) : I = .9999988 ≈ 1

Solution:
4. The velocities of a car running on a straight road at intervals of 2 minutes are given below:
Time(min): 0 2 4 6 8 10 12
.c
Velocity(km/hr): 0 22 30 27 18 7 0
1
Using Simpson’s −rd rule find the distance covered by the car. (ND13)
3
ul

By S(1/3) rule, the distance covered by the car : I = 3.55 km,

Solution:
R1.3 √
5. Taking h = 0.05 evaluate xdx using Trapezoidal rule and Simpson’s three-eighth rule.(AM14)
1
Solution:
pa
By Trap. : I = 0.32147, S(3/8) : I = 0.321485354
6. The velocity υ of a particle at a distance α from a point on its path is given by the table: (ND14)

s( f t) 0 10 20 30 40 50 60
jin
υ 47 58 64 65 61 52 38
3
Estimate the time taken to travel 60 feet by using Simpson’s rule.
8
.re
4.12 Romberg’s method

Rb
 
 
Romberg’s method for a given interval I = f (x)dx

a
b−a
when h = , by trapezoidal rule, we get I1
w
2
b−a
4
b−a
w

8
(I2 − I1 )
Romberg’s formula for I1 &I2 = IRM1,2 = I2 +
w
3
(I3 − I2 )
Romberg’s formula for I2 &I3 = IRM2,3 = I3 +
3
If IRM1,2 = IRM2,3 , then we can equal I = IRM1,2 = IRM2,3
Note : Check, use actual integration, we get IAI = I = IRM1,2 = IRM2,3

4.12.1 Part A
1. State the Romberg’s integration formula with h1 and h2 . Further, obtain the formula when h1 = h
h
and h2 = . (MJ2010)
2
I1 h22 − I2 h21
Solution :I = , where I1 the value of the integral with h1
h22 − h21
I2 the value of the integral with h2
om
h
If h1 = h & h2 = we get
2
4I2 − I1 1
I= = I2 + (I2 − I1 )
3 3
Rb
2. State Romberg’s integration formula to find the value of I = f (x)dx for first two intervals.(ND14)
.c
a
4.12.2 Part B
R2 dx
ul
Example 4.33. Evaluate using Romberg’s method. Hence obtain an approximate
0 x2 +4
value for π.
Solution: To find I1 pa
2−0 1
When h = = 1, y = f (x) = 2
2 x +4
Z2
dx
Let I = 2
x +4
0
jin
The tabulated values of y are
x 0 1 2
1 1
f (x) = 2 = 0.25 0.2 0.125
x +4 02 + 4
.re
y0 y1 y2
Using Trapezoidal rule,
Z2
dx h
I1 = = (y0 + y2 ) + 2 (y1 )
x2 + 4 2
w
0
1
= [(0.25 + 0.125) + 2 (0.2)]
2
= 0.3875
w
To find I2
2−0 1
h= = = 0.5
4 2
w
x 0 0.5 1 1.5 2
1
f (x) = 2 0.25 0.23529 0.2 0.160 0.125
x +4
y0 y1 y2 y3 y4

Z2
dx h
I2 = = (y0 + y4 ) + 2 (y1 + y2 + y3 )
x2 +4 2
0
= 0.25 [(0.25 + 0.125) + 2 (0.23529 + 0.2 + 0.160)]

= 0.39136
To find I3
om
2−0 1
h= = = 0.25
8 4
x 0 0.25 0.5 0.75 1 1.25 1.5 1.75 2

1
f (x) = 2 0.25 0.24615 0.23529 0.21918 0.2 0.17918 0.160 0.14159 0.125
x +4
y0 y1 y2 y3 y4 y5 y6 y7 y8
.c
Z2
dx h
I3 = = (y0 + y8 ) + 2 (y1 + y2 + y3 + y4 + y5 + y6 + y7 )
ul
x2 +4 2
0
h i
= 0.125 (0.25 + 0.125) + 2 (0.24615 + 0.23529 + 0.21918 + 0.2 + 0.17918 + 0.16 + 0.14159)
= 0.39237
Romberg’s formula for I1 &I2 is

pa
(I2 − I1 ) (0.39136 − 0.39136)
IRM1,2 = I2 + = 0.39136 +
3 3
jin
= 0.39265
Romberg’s formula for I2 &I3 is

(I3 − I2 ) (0.39237 − 0.39237)
IRM2,3 = I3 + = 0.39237 +
3 3
.re
= 0.39271
∴ IRM1,2 0.3927
IRM2,3 0.3927
w
Here IRM1,2 &IRM2,3 are almost equal and I = 0.3927 (1)

Z2
dx 1 −1 x x=2 1 h −1
w
i
−1
= tan = tan (1) − tan (0)
x2 + 4 2 2 x=0 2
0
1 π π
= −0 = (2)
w
2 4 8
From (1) & (2),
π
= 0.3927
8
⇒ π = 8 (0.3927) = 3.1416
R1 1
Example 4.34. Using Romberg’s method, evaluate dx correct to 3 places of decimals.
0 1+x
Solution: I1 = 0.7083
I2 = 0.6970
I3 = 0.6941
IRM1,2 = 0.6932
om
IRM2,3 = 0.6931
∴ I = 0.693

R1 dx
1. Using Romberg’s integration to evaluate . (AM10)
.c
0 1 + x2
Solution: [0.7854]
R1 1
2. Using Romberg’s rule evaluate dx correct to three decimal places by taking h = 0.5, 0.25,
ul
0 1+x
and 0.125. (ND10)
Solution: [0.6931]
R1
3. Evaluate
Solution:
0
1
1 + x2
pa
dx using Romberg’s method. (AM11,AM10)
[0.7854]
1 R1
4. Use Romberg’s method to compute 2
dx correct to 4 decimal places. Also evaluate the same
0 1+x
jin
integral using tree-point Gaussian quadrature formula. Comment on the obtained values by
π
comparing with the exact value of the integral which is equal to . (MJ12)
4
By Rom. : I = 0.7854, Dir. Int. : I = 0.7853982

Solution:
1
.re
R2 x
5. Evaluate dx correct to three decimal places using Romberg’s method. (AM14)
0 sin x
Solution: [0.5070676]
R1 dx
6. Evaluate and correct to 3 decimal places using Romberg’s method and hence find the value
0 1+x
w
of loge 2. (ND14)
4.13 Two point Gaussian quadrature formula

w
Two point Gaussian quadrature formula : (Guass two point formula)

Rb
Given I = f (x)dx
w
a
Case (i) If a = −1, b = +1, then
R1
" # " #
1 1
I= f (x)dx = f − √ + f √
−1 3 3
Case (ii) If a = 0, b = 1, then
R1 1 R1
I= f (x) dx = f (x) dx, if f (x) is an even function
0 2 −1
b−a b+a
Case (iii) If (a , −1&b , 1), then x = z+ = mz + c ⇒ dx = mdz
2 2
Zb Z1 Z1
I= f (x)dx = f (z) mdz = m f (z) dz
a −1 −1
" ! !#
1 1
=m f −√ + f √
om
3 3
4.13.1 Part A
dx R1
1. Use two-point Gaussian quadrature formula to solve 2
& also find error. (MJ2010,AM2012)
−1 1 + x
Solution : Given interval is −1 to 1 so we apply formula
R1
! !
.c
−1 1
f (x) dx = f √ + f √
−1 3 3
1
Here f (x) =
1 + x2
ul
!
−1 1 1 3
f √ = = !=
3 1 4 4
1+
3 3
pa
1
f √ =
3
!
1
1+
1
3
=
1
4
3
!=
3
4
Z1
1 3 3 3
∴ dx = + = = 1.5
1 + x2 4 4 2
jin
−1
But actual integration

Z1
1 h i1
2
dx = tan−1 x = tan−1 (1) − tan−1 (−1)
1+x −1
−1
.re
= tan−1 (1) + tan−1 (1)

= 2 tan−1 (1)
π π
=2 =
4 2
= 1.5708
w
Here the error due to two-point formula is 0.0708.

R2 2
2. Evaluate e−x dx by two point Gaussian quadrature formula. (ND2010)
w
0
2
Solution : Here f (x) = e−x [Given range is not in exact form]
b−a b+a
Let x =

Here a = 0, b = 2

z+
w
2 2
2−0 2+0
= z+
2 2
x = z + 1 when x = 0 ⇒ z = −1
dx = dz when x = 2 ⇒ z = 1
Z2 Z1
2
−x2
e dx = e−(z+1) dz
0 −1
# "" #
−1 1
= f √ +f √ (1)
3 3
2
Here f (z) = e−(z+1)
−1 2
 

"
−1
# √ +1
−
om
f √ =e 3 = e−0.1786 = 0.8364
3
2
 1


" # − √ +1
1 3
f √ =e = e−2.488 = 0.0831
3
Z2
.c
2
∴ (1) ⇒ e−x dx = 0.8364 + 0.0831
0
= 0.9195
ul
3. Write down two point Gaussian quadrature formula. 1 (ND2011)
" # " #
 R 1 1 
Solution:  f (x)dx = f − √ + f √ 
pa −1 3 3
−x
R2
4. Evaluate e 2 by Gauss two point formula. (ND2013)
−2
Solution: [4.6854]
jin
4.13.2 Part B
R2 dx
Example 4.35. Evaluate by using Gaussian two point formula.
1 x
.re
Solution: Here a , −1, b = 2 , 1,

(b − a) z + (b + a)
so use x =
2
x = (z + 3) /2 ⇒ dx = dz/2
∴ The above integral becomes

w
R2 dx R1 dz/2 R1 1 R1 1
I= = = dz = f (z) dz, where f (z) =
1 x −1 (z + 3)/2 −1 z + 3 −1 z+3
w
∴ By Gaussian two point formula
Z1 ! !
1 1
f (z) dz = f z = √ + f z = − √
w
3 3
−1
! !
1 1
= +
z + 3 z= √1 z + 3 z=− √1
3 3
= 0.693
π/2
R
Example 4.36. Using Gaussian two point formula evaluate log(1 + x)dx
0
Z1 π π
Solution: I = log 1 + (1 + z) dz
4 4
−1
" ! !#
π 1 1
= f −√ +f √
4 3 3
om
= 0.858
R2 dx
Example 4.37. Evaluate by using Gaussian two point formula.
1 1 + x3
Solution: I = 0.2544.
.c
4.14 Three point Gaussian quadrature formula
ul
Z1   r   r 
5  − 3  +
   3  8
Case (i) f (x)dx =  f f   + f (0)
9 5 5  9
−1 pa
Z1 Z1
1
Case (ii)

for even function f (x)

f (x) dx = f (x)dx
2
0 −1
   r   r  
 5 
1

− 3  +
   3  8 

 + f (0)
=−  f f 
2
9  5 5 9



jin
b−a b+a
Case (iii) (a , −1&b , 1) , then x = z+ = mz + c ⇒ dx = m dz
2 2
Zb Z1    r   r  
 5  

 3   3  8 

.re
I= f (x)dx = f (z) m dz = m   f − + f   + f (0)

   
9  
 5 5 9 

a −1
4.14.1 Part A
R1
1. Write down the three point Gaussian quadrature formula to evaluate f (x)dx. (ND2012)
w
 −1 r   r 
R1
 
 5   3   3  8 
Solution: Ans : f (x)dx =  f −  + f   + f (0)
−1 9 5 5 9
w
4.14.2 Part B
R2 1
Example 4.38. Evaluate dx by using Gaussian three point formula.
w
1 1 + x3
(b − a) z + (b + a)
Solution: Using the substitution x = , a = 1, b = 2
2
z+3
x=
2
Z1 !
1 dz
I= !3
z+3 2
−1 1+
2
Z1
1 1
=
2 !3 dz
z+3
−1 1+
2
Z1
8 1
om
= dz
2 8 + + 3)3
(z
−1
Z1
4 4
= dz, where f (z) =
8 + (z + 3)3 8 + (z + 3)3
−1
( h )
5 p p i 8
= f − 3/5 + f 3/5 + f (0)
9 9
.c
5 8
= [0.27505] + (0.11429)
9 9 " #
5 8
= 0.25439 ∵ = 0.5555, = 0.8888
9 9
ul
I = 0.02544
R1 dx
pa
Example 4.39. Evaluate by two and three point Gaussian quadrature formula &
0 1 + x2
hence find the value of π.
Solution:
Z1 Z1
dx 1 1
jin
Now , I = =
1 + x2 2 1 + x2
0 −1
Z1
1 1 1
= f (x)dx, where f (x) =
2 2 1 + x2
−1
.re
By Gaussian two point formula
Z1 " ! !#
1 1 1 1
I= f (x)dx = f √ +f −√ = 0.75 (1)
2 2 3 3
−1
w
By Guassain three point formula
Z1 " #
1 1 5h p p i 8
w
I= f (x)dx = f − 3/5 + f 3/5 + f (0) = 0.79166 (2)

2 2 9 9
−1
By actual integration
w
Z1
dx h i1
I= 2
= tan−1 (x)
1+x 0
0
π
= tan−1 (1) − tan−1 (0) = (3)
4
From (2) & (3),
π
0.79166 =
4
⇒ π = 3.16664
π
R2
om
Example 4.40. Find sin xdx by two & three point Gaussian quadrature formula.
0
π(z + 1)
Solution: x =
4
by two point formula

I = 0.9985
.c
by three point formula

I = 1.0000
R1 1
Example 4.41. Find dt by using Gaussian three point formula.
ul
0 t
Solution: [I = 1.6027].

pa
R2 1
1. Evaluate dx using Using Gauss three point formula. (AM11)
1 1 + x3
jin
Solution: [0.2544]
R2 x2 + 2x + 1
2. Evaluate 2
dx by Gaussian three point formula. (MJ13)
0 1 + (x + 1)
.re
Solution: [1.5363]
R1 sin x
3. Apply three point Gaussian quadrature formula to evaluate dx. (ND13)
0 x
Solution: [0.94616]
w
R2 dx
4. Evaluate using 3 point Gaussian formula. (ND14)
1 + x2
w
1
w
4.15 Double integrals by Trapezoidal
Double integration by trapezoidal rule is

Rd Rb
Given f (x) dx dy[= I(S ay)]
c a
D(a, d) C(b, d)
 
sum of values of f at the four corners

 

 
k l


   


  sum of values of f at the nodes 
hk 

  


i.e., I = +2
 
  
4 


  on the boundary except the corners  





 


 
 +4 sum of the values at the interior nodes

 


A(a, c) ↔ B(b, c)
h
om
b−a d−c
where h = , k=
n m
where n = is number of equal intervals in (a, b) .
where m = is number of equal intervals in (c, d) .
.c
4.15.1 Part B
R1.4 R2.4 dxdy
Example 4.42. Evaluate the integral using Trapezoidal rule. Verify your results
1 2 xy
by actual integration.
ul
1
Solution: f (x, y) = , x varies from (2, 2.4)
xy
y varies from (1, 1.4)
pa
Divide the range of x and y into 4 equal parts.
2.4 − 2 1.4 − 1
h= = 0.1, k = = 0.1
4 4
The values of f (x, y) at the nodal points are given in the table :
jin
❅ x
❅ 2 2.1 2.2 2.3 2.4
y ❅
❅
1 0.5 0.4762 0.4545 0.4348 0.4167
.re
1.1 0.4545 0.4329 0.4132 0.3953 0.3788

1.2 0.4167 0.3698 0.3788 0.3623 0.3472
1.3 0.3846 0.3663 0.3497 0.3344 0.3205
1.4 0.3571 0.3401 0.3247 0.3106 0.2976
By Trapezoidal rule for double integration
w
 
 Sum of values of f at the four corners 
   
hk 
  Sum of values of f at the nodes  
w
I=  + 2   

4  on the boundary except the corners  
 


+4 Sum of the values at the interior nodes

 
w
 (0.5 + 0.4167 + 0.2976 + 0.3571) 

   
 +2  0.4762 + 0.4545 + 0.4348 + 0.3788 + 0.3472 + 0.3205
   

(0.1) (0.1) 
 
=
 
 +0.3106 + 0.3247 + 0.3401 + 0.3846 + 0.4167 + 0.4545 
4    
  0.4329 + 0.4132 + 0.3953 + 0.3623 + 0.3344  
 +4 
  

+0.3497 + 0.3663 + 0.3698 + 0.3788 
= 0.0614
Z1.4 Z2.4 Z1.4 Z2.4 Z1.4

 
1  1  1 1
dxdy =  dx dy = log x 2.4
2 dy
xy  x  y y
1 2 1 2 1
= log 2.4 − log 2 log y 1.4

om
1
= 0.0613
.c
R1.2 R1.4 dxdy
1. Evaluate by trapezoidal formula by taking h = k = 0.1. (AM10)
1 1 x+y
Solution: [0.0349]
ul
R5 R4
 
1 
2. Evaluate  dx dy by Trapezoidal rule in x−direction with h = 1 and Simpson’s one-third
1 1 x + y
rule in y−direction with k = 1. (ND10)
Solution:
pa
By Trap. : I = 2.4053, Simp. : I = 2.122

R1 R1 1
3. Evaluate dx dy by using Trapezoidal rule taking h = 0.5 and k = 0.25. (AM11)
0 0 x+y+1
jin
Solution: [0.5319 ≈ 0.532]
R2 R2 dxdy
4. Using Trapezoidal rule, evaluate 2 2
numerically with h = 0.2 along x−direction and k =
1 1 x +y
.re
0.25 along y−direction. (MJ12)

Solution: [0.2643]
R2.4 R4.4
5. Evaluate xy dx dy by Trapezoidal rule taking h = k = 0.1. (ND13)
2 4
w
Solution: [1.4784]
R1 R1 1
6. Evaluate dxdy. (ND14)
w
0 0 1+x+y
w
4.16 Double integrals by Simpson’s 1/3 rules
Rd Rb
Double integration by Simpson’s 1/3 rule is Given f (x) dx dy[= I(S ay)]
c a
 




 sum of values of f at the four corners 





   



  sum of values of f at the odd positions  


+2 


  


 
on the boundary except the corners

  



 



   




  sum of the values of f at the even positions  



+4 


   


 
on the boudary except the corners

  



 



   




  sum of the values of f at odd positions  



hk 

 +4   


i.e., I = on the odd rows of the matrix except boundary rows 
  
9 
 
om

   




  sum of the values of f at even positions  


 
+8 


  


 
on the odd rows of the matrix except boundary rows 

  



 



   




  sum of the values of f at odd positions  



+8 


   


 
on the even rows of the matrix except boundary rows 

  



 



   




  sum of the values of f at even positions  





 +16   


.c
on the even rows of the matrix except boundary rows 


  


ul
4.16.1 Part B
pa R1.4 R2.4 dxdy

Example 4.43. Evaluate the integral using Simpson’s rule. Verify your results by
1 2 xy
actual integration.
1
Solution: f (x, y) = ,x varies from (1, 1.4)
xy
jin
y varies from (2, 2.4)
Divide the range of x and y into 4 equal parts.

.re
2.4 − 2 1.4 − 1
h= = 0.1, k = = 0.1
4 4
w
The values of f (x, y) at the nodal points are given in the table :
w
❅ x
❅ 2 2.1 2.2 2.3 2.4
y ❅
❅
1 0.5 0.4762 0.4545 0.4348 0.4167
w
1.1 0.4545 0.4329 0.4132 0.3953 0.3788

1.2 0.4167 0.3698 0.3788 0.3623 0.3472
1.3 0.3846 0.3663 0.3497 0.3344 0.3205
1.4 0.3571 0.3401 0.3247 0.3106 0.2976
By Extended Simpson’s rule

 
 Sum of the values of f at the four corners 
 
 + 2 Sum of the values of f at the odd positions on the boundary except the corners
 


 + 4 Sum of the values of f at the even positions on the boundary except the corners
 
   
 + 4  Sum of the values of f at the odd positions
   
 

on the odd rows of the matrix except boundary rows 
 
 
hk    
I=   Sum of the values of f at the even positions  
om
9  + 8   
  on the odd rows of the matrix except boundary rows  

   
  Sum of the values of f at the odd positions  

 +8    
  on the even rows of the matrix except boundary rows  

   

  Sum of the values of f at the even positions  
 +16   
 on the even rows of the matrix except boundary rows  
.c
 
 (0.5 + 0.4167 + 0.2976 + 0.3571) 
 
+2 (0.4545 + 0.3472 + 0.3247 + 0.4167)
 
   
0.4762 + 0.4348 + 0.3788 + 0.3205 + 0.3106
 

  


ul
 
 +4   
(0.1) (0.1)  +0.3401 + 0.3846 + 0.4545 
=  
9  +4 (0.3788) 

 
 +8 (0.3968 + 0.3623)
 
pa

 +8 (0.3497 + 0.4132)

+16 (0.3663 + 0.3344 + 0.4329 + 0.3953)




0.01
= (55.2116) = 0.0613
9
jin
Z1.4 Z2.4 Z1.4 Z2.4 Z1.4
 
1  1  1 1
dxdy =  dx dy = log x 2.4
2 dy
xy  x y y
1 2 1 2 1
= log 2.4 − log 2 log y 1.4

.re
= 0.0613
There is no error(or deviation) netween actual integration and Simpson’s rule.

R0.5 R0.5 sin xy
Example 4.44. Evaluate I = dxdy by using Simpson’s rule with step size 0.25.
0 0 1 + xy
w

Ans : I = 0.000216(Simpson’s rule) .

Solution:
4.16.2 Double integrals by Trapezoidal, Simpson rule

w
R1.4 R2.4 dxdy

Example 4.45. Evaluate the integral using Trapezoidal rule and Simpson’s rule.
1 2 xy
Verify your results by actual integration.
w

I = 0.0614(Trapezoidal rule), I = 0.0613(Simpson’s rule)

Solution:
π π
R2 R2
Example 4.46. Evaluate sin(x + y)dxdy by using trapezoidal rule, Simpson’s rule and also
0 0
by actual integration.
Solution: Divide into 2 intervals (3 ordinates) on x&y
By Trapezoidal rule, I = 1.7975

Simpson’s rule, I = 2.0080
Actual integration, I = 2
om
R5 R4
 
1 
1. Evaluate  dx dy by Trapezoidal rule in x−direction with h = 1 and Simpson’s one-third
1 1 x + y
rule in y−direction with k = 1. (ND10)

Solution: By Trap. : I = 2.4053, Simp. : I = 2.122
R2 R1 1 1
2. Evaluate 4xydxdy using Simpson’s rule by taking h = and k = . (ND12)
0 0 4 2
.c
Solution: [3.1111]
R1.4 R2.4 1
3. Evaluate dxdy using Simpson’s one-third rule. (MJ13)
1 2 xy
ul
Solution: [0.0613]
1 1
1 R2 R2 sin(xy)
4. Taking h = k = , evaluate dxdy using Simpson’s rule . (AM14)
4
Solution:
pa
0 0 1 + xy
[0.0141]
1. Given the following data, find y ′ (6) and the maximum value of y (if it exists). (AM10)
jin
x: 0 2 3 4 7 9
y: 4 26 58 112 466 922
2. Find f ′ (x) at x = 1.5 and x = 4.0 from the following data using Newton’s formulae for
.re
differentiation.
x: 1.5 2.0 2.5 3.0 3.5 4.0

(MJ12)
y = f (x) : 3.375 7.0 13.625 24.0 38.875 59.0
w
3. Find the first three derivatives of f (x) at x = 1.5 by using Newton’s forward interpolation formula
to the data given below.
x: 1.5 2 2.5 3 3.5 4

w
(MJ13)
y: 3.375 7 13.625 24 38.875 59
R1 1
w
4. Using Trapezoidal rule, evaluate dx by taking eight equal intervals. (MJ13)

−1 1 + x2
S (meter) 0 10 20 30 40 50 60
(AM10)
υ (m / sec) 47 58 64 65 61 52 38
1R6
6. Evaluate I = dx by using (i) direct integration (ii) Trapezoidal rule (iii) Simpson’s one-third
0 1 + x
rule (iv) Simpson’s three-eighth rule. (ND11)
R1.3 √
7. Taking h = 0.05 evaluate xdx using Trapezoidal rule and Simpson’s three-eighth rule.(AM14)
1
om
1 R1
8. Use Romberg’s method to compute dx correct to 4 decimal places. Also evaluate the same
0 1 + x2
integral using tree-point Gaussian quadrature formula. Comment on the obtained values by
π
comparing with the exact value of the integral which is equal to . (MJ12)
4
1
R2 x
.c
9. Evaluate dx correct to three decimal places using Romberg’s method. (AM14)
0 sin x
R1 dx
10. Evaluate and correct to 3 decimal places using Romberg’s method and hence find the value
ul
0 1+x
of loge 2. (ND14)
R2 x2 + 2x + 1
11. Evaluate 2
dx by Gaussian three point formula. (MJ13)
0 1 + (x + 1)
12. Apply three point Gaussian quadrature formula to evaluate

pa R1 sin x
0 x
dx. (ND13)
R1 R1 1
13. Evaluate dx dy by using Trapezoidal rule taking h = 0.5 and k = 0.25. (AM11)
x+y+1
jin
0 0
R2 R2 dxdy
14. Using Trapezoidal rule, evaluate 2 2
numerically with h = 0.2 along x−direction and k =
1 1 x +y
0.25 along y−direction. (MJ12)
.re
R2.4 R4.4
15. Evaluate xy dx dy by Trapezoidal rule taking h = k = 0.1. (ND13)
2 4
1 1
1 R2 R2 sin(xy)
16. Taking h = k = , evaluate dxdy using Simpson’s rule . (AM14)
4 0 0 1 + xy
w
w
w
5 Numerical Solution of Ordinary Differential
Equations
om
Single Step methods − Taylor’s series method − Euler’s method − Modified Euler’s method
.c
− Fourth order Runge − Kutta method for solving first order equations
Multi step methods − Milne’s and Adams − Bash forth predictor corrector methods
ul
for solving first order equations
5.1 Introduction
pa
Single Step methods Multi step methods
jin
Taylor’s series method Milne’s forth predictor corrector method
Taylor’s series method for simultaneous first ODE Adams-Bash forth predictor corrector method
Taylor’s series method for II order ODE
Euler’s method
.re
Modified Euler’s method

Runge-Kutta method for first order ODE
Runge - Kutta method for solving II order ODE
R. K. Method for simultaneous first order ODE
w
5.2 Taylor’s series method

w
dy
Given = y ′ = f (x, y) with y (x0 ) = y0
w
dx
d2 y ∂ f ∂ f dy
⇒ 2 = +
dx ∂x ∂y dx
i.e., y ′′ = f x + fy y ′

lllly find y ′′′ , y ′′′′ , · · ·

Taylor’s series expansion of y(x) above x = x0 is given by

(x − x0 ) ′ (x − x0 )2 ′′ (x − x0 )3 ′′′
y (x) = y (x0 ) + y (x0 ) + y (x0 ) + y (x0 ) + · · ·
1! 2! 3!
(x − x0 ) ′ (x − x0 )2 ′′ (x − x0 )3 ′′′
= y0 + y0+ y0 + y 0 + ···
1! 2! 3!
(or)
h ′ h2 ′′ h3 ′′′
y (x1 ) = y1 = y0 + y + y + y + ··· , where x1 = x0 + h, h = x1 − x0
1! 0 2! 0 3! 0
om
h h2 ′′ h3 ′′′
y (x2 ) = y2 = y1 + y ′1 + y + y + ··· , where x2 = x1 + h, h = x2 − x1
1! 2! 1 3! 1
5.2.1 Part A
1. Using Taylor series method find y(1.1) given that y ′ = x + y, y(1) = 0. (MJ2011)
′
Solution: Given y = x + y, y(x = 1) = 0 ⇒ x0 = 1, y0 = 0, x1 = 1.1, h = x1 − x0 = 1.1 − 1 = 0.1
.c
∴ Taylor’s series formula is
h ′ h2 ′′ h3 ′′′ h4 ′v
y1 = y0 + y + y + y + y 0 + ··· (1)
1! 0 2! 0 3! 0 4!
ul
y′ = x+y y ′0 = x0 + y0 (x0 =1,y0 =0) = 1 + 0 = 1 (2)
′′ ′ ′′ ′
y =1+y y 0 = 1 + y0 = 1+1=2 [by (2)] (3)
y ′′′
y ′v = y ′′′
=0+y ′′
=y ′′
y
y
′′′
pa ′′
0 = y0 = 2
′v ′′′
0 = y0 = 2
[by (3)]
[by (4)]
(4)
0.1 (0.1)2 (0.1)3 (0.1)4

∴ (1) ⇒ y1 = 0 + (1) + (2) + (2) + (2)
1! 2! 3! 4!
jin
(0.1)3 (0.1)4
i.e., y (1.1) = (0.1) + (0.1)2 + + + ···
3 12
= 0.1103083
≡ 0.11031
.re
2. What is the major drawback of Taylor series method? (AM2012)

Solution: The major drawbacks of Taylor series method are :
dy
1. In the differential equation, = f (x, y) may have a complicated algebraic structure.
dx
2. The evaluation of higher order derivatives may become tedious.
w
dy
3. Find y(0.1) if = 1 + y, y(0) = 1 using Taylor series method. (ND2012)
dx
′
Solution: Given y = 1 + y, y(x = 0) = 1 ⇒ x0 = 0, y0 = 1, x1 = 0.1, h = x1 − x0 = 0.1 − 0 = 0.1
w

h ′ h2 ′′ h3 ′′′ h4 ′v
y1 = y0 + y + y + y + y 0 + ··· (1)
1! 0 2! 0 3! 0 4!
w
y′ =1+y y ′0 = 1 + y0 (x0 =0,y0 =1) = 1 + 1 = 2 (2)

′′ ′ ′ ′′ ′
y =0+y =y y 0 = y0 = 2 [by (2)] (3)
y ′′′ = y ′′ y ′′′ ′′
0 = y0 = 2 [by (3)] (4)
y ′v = y ′′′ y ′v ′′′
0 = y0 = 2 [by (4)]
Unit V - Numerical Solution of Ordinary Differential Equations 107
0.1 (0.1)2 (0.1)3 (0.1)4

∴ (1) ⇒ y1 = 1 + (2) + (2) + (2) + (2)
1! 2! 3! 4!
i.e., y (1.1) = 1.210341667
≡ 1.21034
4. Find y(1.1) if y ′ = x + y, y(1) = 0 by Taylor series method. (AM2013)
om

Solution: Re f : (MJ2011)
5. State the advantages and disadvantages of the Taylor’s series method. (AM2014)
Solution:
The advantages of the Taylor’s series method are :
1. It is a powerful single step method if we are able to find the successiove derivatives easily.
.c
2. This method gives a straight forward adaptation of classic calculus to develop the solution
as an infinite aseries.
ul
3. This method will be very useful for finding the starting values for powerful methods like
Runge-Kutta method, Milne’s method etc.
The disadvantages of the Taylor’s series method are :

pa
dy
1. In the differential equation, = f (x, y) may have a complicated algebraic structure.
dx
2. The evaluation of higher order derivatives may become tedious.
5.2.2 Part B
jin
dy
Example 5.1. Using Taylor’s series method, find y at x = 0.1, if = x2 y − 1, y (0) = 1
dx
Solution: Given y ′ = x2 y − 1, x0 = 0, y0 = 1, h = 0.1 (G)
.re
h ′ h2 ′′ h3 ′′′ h4 ′v
y1 = y0 + y + y + y 0 + y 0 + ··· (1)
1! 2! 3! 4!
y ′ = x2 y − 1 y ′0 = x02 y0 − 1(x0 =0,y0 =1) = 0 − 1 = −1 [by (G)] (2)

′
′′ 2 ′ ′′
y = 2xy + x y y 0 = 2x0 y0 + x02 y0 = 0 [by (G) & (2)] (3)

′′′
= 2 x0 y ′0 + y0 + x02 y ′′0 + y ′0 2x0
w
y 0
y ′′′ = 2 (xy ′ + y) + x2 y ′′ + y ′ 2x [by (G) & (3)] (4)
=2
yiv ′ ′′ ′ ′′
0 = 2y 0 + 2x0 y 0 + 2y 0 + 2x0 y 0
yiv = 2y ′ + 2xy ′′ + 2y ′ + 2xy ′′
w
+x02 y ′′0 + y ′′0 (2x0 ) + 2y ′0 [by (G) & (4)]

+x2 y ′′ + y ′′ (2x) + 2y ′
= −6
w
0.1 (0.1)2 (0.1)3 (0.1)4

∴ (1) ⇒ y1 = 1 + (−1) + (0) + (0) + (−6)
1! 2! 3! 4!
(0.1)3 (0.1)4
i.e., y (0.1) = 1 − (0.1) + − + ···
3 4
= 0.900308
dy
Example 5.2. Use Taylor series method to solve = x − y2 , y(0) = 1 and hence compute y(0.1)
dx
and y(0.2) correct to 4 places of decimals.
Solution: Given y ′ = x − y2 and x0 = 0, y0 = 1.
Take h = 0.1
x1 = x0 + h = 0 + 0.1 = 0.1
y1 = y (x1 ) = y(0.1)
om
To find y1
The Taylor’s series formula for y1 is
h ′ h2 ′′ h3 ′′′ h4 ′v
y1 = y0 + y + y + y 0 + y 0 + ··· (1)
1! 2! 3! 4!
.c
y ′0 = x0 − y20
y ′ = x − y2
= 0 − 1 = −1
y ′′0 = 1 − 2y0 y ′0
y ′′ = 1 − 2yy ′
ul
= 1 − 2(1)(−1) = 3
′′′ ′′ ′ ′ ′′′
y = −2 (yy + y y ) y 0 = −2y0 y ′′0 − 2y ′2
0
= −2yy ′′ − 2y ′2 = −2(1)(3) − 2 = −8
yiv
= −6y ′ y′′ − 2yy ′′′
pa
= −2 (yy ′′′ + y ′′ y ′ ) − 4y ′ y′′ yiv
0 = −6y ′0 y ′′0 − 2y0 y ′′′
0
= −6(−1)(3) − 2(1)(−8) = 34
From (1)
0.1 (0.1)2 (0.1)3 (0.1)4

∴ (1) ⇒ y1 = 1 + (−1) + (3) + (−8) + (34)
jin
1 2 6 24
= 1 − 0.1 + 0.015 − 0.001333 + 0.0001417
= 0.9138
∴ y(0.1) = 0.9138
.re
To find y2
x2 = x1 + h = 0.1 + 0.1 = 0.2

y2 = y (x2 ) = y(0.2)
w
h ′ h2 ′′ h3 ′′′ h4 ′v
w
y2 = y1 + y + y + y + y 1 + ··· (2)
1! 1 2! 1 3! 1 4!
y ′1 = x1 − y21
w
y ′ = x − y2
= 0.1 − (0.9138)2 = −0.735
y ′′1 = 1 − 2y1 y ′1
y ′′ = 1 − 2yy ′
= 1 − 2(0.9138)(−0.735) = 2.3433
y ′′′
1 = −2y1 y ′′1 − 2y ′2
1
y ′′′ = −2 (yy ′′ + y ′ y′ )
= −2(0.9138)(2.3433) − 2(−0.735)2
= −2yy ′′ − 2y ′2
= −5.363
yiv
0 = −6y ′1 y ′′1 − 2y1 y ′′′
1
yiv = −2 (yy ′′′ + y ′′ y ′ ) − 4y ′ y′′
= −6(−0.735)(2.3433) − 2(0.9138)(−5.363)
= −6y ′ y′′ − 2yy ′′′
= 20.1354
From (2)
om
0.1 (0.1)2 (0.1)3 (0.1)4
y2 = 0.9138 + (−0.735) + (2.3433) + (−5.363) + (20.1354)
1 2 6 24
y2 = 0.8512
∴ y(0.2) = 0.8512
Example 5.3. Using Taylor series method, find y to five places of decimals when x = 1.3 given
.c

that dy = x2 y − 1 dx and y = 2 when x = 1.
Solution: Given y ′ = x2 y − 1 and x0 = 1, y0 = 2.
Given h = 0.3
ul
x1 = x0 + h = 1 + 0.3 = 1.3
y1 = y (x1 ) = y(1.3) pa
To find y1
h ′ h2 ′′ h3 ′′′ h4 ′v
y1 = y0 + y + y + y + y 0 + ··· (1)
jin
1! 0 2! 0 3! 0 4!
y ′0 = x02 y0 − 1
y ′ = x2 y − 1
= 12 (2) − 1 = 1
y ′′0 = x02 y ′0 + 2x0 y0
.re
y ′′ = x2 y ′ + 2xy
= 12 (1) + 2(1)(2) = 5
y ′′′ = x2 y ′′ + 2xy ′ + 2 xy ′ + y y ′′′ = x02 y ′′0 + 4x0 y ′0 + 2y0

0
= x2 y ′′ + 4xy ′ + 2y = 12 (5) + 4(1)(1) + 2(2) = 13
y ′v = x2 y ′′′ + 2xy ′′ + 4 xy ′′ + y ′ + 2y ′ y ′v = x02 y ′′′ ′′ ′

0 0 0 + 6x0 y 0 + 6y 0
w
= x2 y ′′′ + 6xy ′′ + 6y ′ = 12 (13) + 6(1)(5) + 6(1) = 49
From (1),
w
0.3 (0.3)2 (0.3)3 (0.3)4

y1 = 2 + (1) + (5) + (13) + (49)
1! 2! 3! 4!
= 2 + 0.3 + 0.225 + 0.0585 + 0.01654
w
= 2.60004
Example 5.4. Using Taylor series method find y at x = 0.1 correct to four decimal places from
dy
= x2 − y, y(0) = 1 with h = 0.1. Compute terms upto x4 .
dx
Solution: Given y ′ = x2 − y and x0 = 0, y0 = 1.
Given h = 0.1
x1 = x0 + h = 0 + 0.1 = 0.1
y1 = y (x1 ) = y(0.1)
To find y1
h ′ h2 ′′ h3 ′′′ h4 ′v
om
y1 = y0 + y + y + y + y 0 + ··· (1)
1! 0 2! 0 3! 0 4!
y ′0 = x02 − y0
y ′ = x2 − y
= 02 − 1 = −1
y ′′0 = 2x0 − y ′0
y ′′ = 2x − y ′
= 2(0) − (−1) = 1
.c
′′′
y 0 = 2 − y ′′0
y ′′′ = 2 − y ′′
=2−1=1
y ′v = −y ′′′
ul
0 0
y ′v = −y ′′′
= −1
From (1),
y1 = 1 +
0.1
1!
(−1) +
2!
pa
(0.1)2
(1) +
(0.1)3
3!
(1) +
(0.1)4
4!
(−1)
= 1 − 0.1 + 0.005 + 0.000167 − 0.000004

= 0.905163
jin
Example 5.5. Using Taylor series method, find y(1.1) and y(1.2) correct to four decimal places
1
dy
given = xy 3 and y(1) = 1.
dx
1
.re
′
Solution: Given y = xy 3 and y(1) = 1.
Given h = 0.1
x1 = x0 + h = 1 + 0.1 = 1.1
y1 = y (x1 ) = y(1.1)
w
To find y1
w
h ′ h2 ′′ h3 ′′′
y1 = y0 + y + y + y + ··· (1)
1! 0 2! 0 3! 0
w
1
1
′ y ′0 = x0 y03
y = xy 3
1
= 1(1) 3 = 1
2 1 1 1
′′ 1 − ′ 1 2 −3 3
y = x y 3 y + y3 y ′′
0 = x0 y0 + y0
3 3
1 1 1 1
1 2 − 1 − 4
= x y 3 + y3 = (1)2 (1) 3 + (1) 3 =
3 3 3
2 1
1 − 2 −
y ′′′
0 = − x02 y0 3 y ′0 + x0 y0 3
! 4 1 9 3
 
1  2 1 − ′ − 2
 
y ′′′ =  x − y 3 y + y 3 (2x)

−
1
+ y0 3 y ′0
om
3 3 
2 3
2
1 − 1 2 −
+ y 3y ′ = − (1) (1) 3 (1)
3 9
4 1 2 1 2
1 2 − ′ 2 − 1 − ′ 2 − 1 −
3
= − x y y + xy + y y3 3 + (1)(1) 3 + (1) 3 (1)
9 3 3 3 3
8
=
9
.c
From (1),
ul
(0.1)2 4 (0.1)3 8
! !
0.1
y1 = 1 + (1) + +
1! 2! 3 3! 9
pa = 1 + 0.1 + 0.00666 + 0.000148
= 1.10681
∴ y(1.1) = 1.10681
To find y2
jin
x2 = x1 + h = 1.1 + 0.1 = 1.2
y1 = y (x2 ) = y(1.2)
.re
h ′ h2 ′′ h3 ′′′
y2 = y1 + y + y + y + ··· (2)
1! 1 2! 1 3! 1
1
w
1
y ′1 = x1 y13
y ′ = xy 3
1
= (1.1)(1.10681) 3 = 1.13785
w
1 1
1 2 −3
1 1 y ′′1 = x1 y1 + y13
1 2 − 3
y ′′ = x y 3 + y3 1 1
3 1 2
−
= (1.1) (1.10681) 3 + (1.10681) 3
w
3
= 1.42433
4 1 2
′′′ 1 2 −3 ′ 2 −3 1 −3 ′
y 1 = − x1 y1 y 1 + x1 y1 + y1 y 1
9 3 3
4
4 1 1 −
1 − 2 − = − (1.1)2 (1.10681) 3 (1.13785)
y ′′′ = − x2 y 3 y ′ + xy 3 9
9 3 1
2 2 −
1 − + (1.1)(1.10681) 3
+ y 3y ′ 3
3 2
1 −
3
+ (1.10681) (1.13785)
om
3
= 0.92979
From (2),
.c
0.1 (0.1)2 (0.1)3
y2 = 1.10681 + (1.13785) + (1.42433) + (1.92979)
1! 2! 3!
= 1.2277
ul
∴ y(1.2) = 1.2277
Example 5.6. Find the Taylor series solution with three terms for the initial value problem
dy
= x3 + y, y(1) = 1.
dx
Solution: Given y ′ = x3 + y and x0 = 1, y0 = 1.
pa
(x − x0 ) ′ (x − x0 )2 ′′ (x − x0 )3 ′′′
y (x) = y (x0 ) + y (x0 ) + y (x0 ) + y (x0 ) + · · · (1)
1! 2! 3!
jin
y ′0 = x03 + y0
y ′ = x3 + y
= 13 + 1 = 2
y ′′0 = 3x02 + y ′0
y ′′ = 3x2 + y ′
.re

= 3 12 + 2 = 5
y ′′′
0 = 6x0 + y ′′0
y ′′′ = 6x + y ′′
= 6(1) + 5 = 11
Substituting in (1)
w
x−1 (x − 1)2 (x − 1)3

y1 = 1 + (2) + (5) + (11)
1! 2! 3!
5 11
= 1 + 2(x − 1) + (x − 1)2 + (x − 1)3
w
4 6
Example 5.7. Using Taylor’s method, compute y(0.2) and y(0.4) correct to 4 decimal places
dy
given (= y ′ ) = 1 − 2xy & y (0) = 0.
dx
w
Solution: y(0.2) = 0.194752003, y(0.4) = 0.359883723

dy
Example 5.8. Find the Taylor series solution of y(0.1) given that + y2 = e x , y(0) = 1. Compute
dx
using first five terms.
Solution: Given y ′ = e x − y2 and x0 = 0, y0 = 1.
Take h = 0.1
x1 = x0 + h = 0 + 0.1 = 0.1
y1 = y (x1 ) = y(0.1)
To find y1
h ′ h2 ′′ h3 ′′′ h4 ′v h5 v
om
y1 = y0 + y + y + y + y 0 + y0 + · · · (1)
1! 0 2! 0 3! 0 4! 5!
y ′0 = e x0 − y20
y ′ = e x − y2
= e0 − 12 = 0
y ′′0 = e x0 − 2y0 y ′0
y ′′ = e x − 2yy ′
.c
= e0 − 2(1)(0) = 1
h i
y ′′′ = e x − 2 yy ′′ + y ′ y′ y ′′′ = e x0 − 2 y0 y ′′0 − y ′2

0 0
h i h i
= e x − 2 yy ′′ + y ′2 = e0 − 2 1(1) + 02 = −1
ul
h i
yiv = e x − 2 yy ′′′ + y ′′ y ′ + 2y ′ y′′ yiv = e x0 − 2 y ′0 y ′′′ ′ ′′

0 0 + 3y 0 y 0
= e x − 2 y ′ y′′′ + 3y ′ y′′ = e0 − 2[1(−1) + 3(0)(1)] = 3

h h i
yv = e x − 2 yy ′v + y ′′′ y ′
pa yv0 = e x0 − 2 y ′0 y ′v ′ ′′′ ′′2
0 + 4y 0 y 0 + 3y 0
i
+3 (y ′ y′′′ + y ′′ y ′′ ) = 1 − 2[1(3) + 4(0)(−1) + 3(1)2 ]
h i
= e x − 2 y ′ y′v + 4y ′ y′′′ + 3y ′′2 = −11
jin
From (1)
0.1 (0.1)2 (0.1)3 (0.1)4 (0.1)5

y1 = 1 + (0) + (1) + (−1) + (3) + (−11)
1 2 6 24 120
= 1 + 0 + 0.005 − 0.000167 + 0.000125 − 0.000000917
.re
= 1.00496
∴ y(0.1) = 1.00496
Example 5.9. Using Taylor’s series method, with the first five terms in the expansion; find
dy
w
y(0.1) correct to 3 decimal places, given that = e x − y2 , y (0) = 1

dx
Solution: y(0.1) = 1.0049891 1.005 (correct to 3 decimal places)
w

dy
1. Evaluate the value of y at x = 0.1, 0.2 given = x2 y − 1, y(0) = 1, by Taylor’s series method upto
dx
w
four terms. (ND10)

Solution: y(0.1) = 0.900308, y(0.2) = 0.802269
2. Using Taylor series method to find y(0.1) if y ′ = x2 + y2 , y(0) = 1. (MJ13)

Solution: y(0.1) = 1.1115
3. Obtain y by Taylor series method, given that y ′ = xy + 1, y(0) = 1, for x = 0.1 and 0.2 correct to four
decimal places. (ND13)

Solution: y(0.1) = 1.1053, y(0.2) = 1.2224
dy
4. Using Taylor’s series method, find y at x = 1.1 by solving the equation = x2 + y2 ; y(1) = 2.
dx
Carryout the computations upto fourth order derivative. (AM14)

Solution: y(0.1) = 2.64333
om
dy
5. Using Taylor’s series method, find y at x = 0 if = x2 y − 1, y(0) = 1. (ND14)
dx
5.3 Taylor’s series method for simultaneous first order differential

equations
.c
dy dz
Given = f1 (x, y, z) , = f2 (x, y, z) with initial conditions y (x0 ) = y0 , z (x0 ) = z0
dx dx
ul
5.3.1 Part B
dy dz
Example 5.10. Solve the system of equations pa = z − x2 , = y + x with y (0) = 1, z (0) = 1 by
dx dx
taking h = 0.1 to get y (0.1) and z (0.1).
Solution: Given x0 = 0, y0 = 1, z0 = 1
WKT, Taylor’s series for y1 is
h ′ h2 ′′ h3 ′′′ h4 iv
jin
y1 = y (0.1) = y0 + y + y + y + y0 + · · · (1)
1! 0 2! 0 3! 0 4!
& Taylor’s series for z1 is
h ′ h2 ′′ h3 ′′′ h4 iv
z1 = z (0.1) = z0 + z + z + z + z0 + · · · (2)
1! 0 2! 0 3! 0
.re
4!
dy dz
y′ = = z − x2 y ′0 = z0 − x02 = 1 − 0 = 1 z′ = = x+y z′0 = x0 + y0 = 0 + 1 = 1
dx dx
y ′′ = z′ − 2x y ′′0 = 1 z′′ = 1 + y ′ z′′0 = 1 + y ′0 = 2
y ′′′ = z′′ − 2 y ′′′
0 =0 z′′′ = y ′′ z′′′
0 =1
w
yiv = z′′′ yiv

0 =1 ziv = y ′′′ ziv
0 =0
(0.1) (0.1)2 (0.1)3 (0.1)4

w
∴ (1) ⇒ y (0.1) = 1 + (1) + (1) + (0) + (1)

1! 2! 3! 4!

correct to 4 decimal places

= 1.1050
w
(0.1) (0.1)2 (0.1)3 (0.1)4

Now (2) ⇒ z (0.1) = 1 + (1) + (2) + (1) + (0)
1! 2! 3! 4!
= 1.110167

correct to 4 decimal places

≃ 1.1102
5.4 Taylor’s series for II order differential equations
5.4.1 Part B
Example 5.11. By Taylor’s series, find y(0.1) and y(0.2) given that y ′′ = y + xy ′ , y (0) = 1, y ′ (0) = 0.
Solution: Given x0 = 0, y0 = 1, y ′0 = 0 & y ′′ = y + xy ′
WKT, Taylor’s series formula is
om
x2 ′′ x3 ′′′ x4 ′v
y (x) = y0 + xy ′0 + y + y + y 0 + ··· (1)
2! 0 3! 0 4!
y ′′ = y + xy ′ y ′′0 = y0 + x0 y ′0 = 1 + 0 (0) = 1
y ′′′ = y ′ + xy ′′ + y ′ = 2y ′ + xy ′′ y ′′′
0 = 2(0) + (0)(1) = 0
yiv = 2y ′′ + xy ′′′ + y ′′ = 3y ′′ + xy ′′′ yiv
0 = 3(1) + (0)(0) = 3
.c
x2 x4
∴ (1) ⇒ y (x) = 1 + 0 + (1) + 0 + (3) + · · ·
ul
2! 4!
x2 x4
=1+ + + ···
2 8
(0.1)2 (0.1)4
y (0.1) = 1 + + + ···
pa 2
= 1.0050125 ≃ 1.0050
8
(correct to 4 decimal places)
(0.2)2 (0.2)4
y (0.2) = 1 + + + ···
2 8
= 1.0202
jin
5.5 Euler’s method

.re
Given y ′ = f (x, y) , x0 , y0 , h
Euler algorithm is
y1 = y0 + h f (x0 , y0 )
w
y2 = y1 + h f (x1 , y1 )
y3 = y2 + h f (x2 , y2 )
..
w
.
In general, yn = yn−1 + h f (xn−1 , yn−1 )
w
5.5.1 Part A
dy
1. Use Euler’s method to find y(0.2) and y(0.4) given = x + y, y(0) = 1. (MJ2010)
dx
Solution : Given f (x, y) = x + y, x0 = 0, y0 = 1, x1 = 0.2, x2 = 0.4. (Here h = x1 − x0 = x2 − x1 = 0.2)
By Euler algorithm,

y1 = y0 + h f (x0 , y0 ) = 1 + (0.2) x0 + y0 = 1 + (0.2) [0 + 1]
i.e., y1 = y (0.2) = 1.2

y2 = y1 + h f (x1 , y1 ) = 1.2 + (0.2) x1 + y1 = 1.2 + (0.2) [0.2 + 1.2]
= 1.2 + 0.28
i.e., y2 = y (0.4) = 1.48
om
dy
2. Find y(0.1) by using Euler’s method given that = x + y, y(0) = 1. (ND10)
dx
dy
Solution : Given, = x + y = f (x, y), x0 = 0, y0 = 1, x1 = 0.1, h = 0.1
dx
By Euler algorithm,
.c
y1 = y0 + h f (x0 , y0 )

= 1 + (0.1) x0 +y0
= 1 + (0.1) [0 + 1]
ul
= 1 + 0.1
= 1.01
pa
i.e., y (0.1) = 1.01
3. Find y(0.2) for the equation y ′ = y + e x , given that y(0) = 0 by using Euler’s method. (AM11)
x
Solution : Given, f (x, y) = y + e , x0 = 0, y0 = 0, h = 0.2
By Euler algorithm,
jin
y1 = y0 + h f (x0 , y0 )
= 0 + 0.2 f (0, 0)
h i
= 0.2 0 + e0 = 0.2
.re
i.e., y (0.2) = 0.2
dy
4. State Euler’s method to solve = f (x, y) with y (x0 ) = y0 . (ND11)
dx
Solution: yn = yn−1 + h f (xn−1 , yn−1 )
w
dy
5. Using Euler’s method, find the solution of the initial value problem = log(x + y), y(0) = 2 at
dx
x = 0.2 by assuming h = 0.2. (MJ12)
w
dy
Solution : Given, = log(x + y) = f (x, y), x0 = 0, y0 = 2, h = 0.2
dx
By Euler algorithm,
w
y1 = y0 + h f (x0 , y0 )
= 2 + 0.2 f (0, 2)

= 2 + 0.2 log(0 + 2) = 2 + 0.2 log 2 = 2 + 0.2(0.30103)
i.e., y (0.2) = 2.060206
6. State Euler’s formula. (MJ13)

Solution: yn = yn−1 + h f (xn−1 , yn−1 )
7. Using Euler’s method find the solution of the initial value problem y ′ = y − x2 + 1, y(0) = 0.5 at
x = 0.2 taking h = 0.2. (ND13)
′ 2
Solution : Given, y = y − x + 1 = f (x, y), x0 = 0, y0 = 0.5, h = 0.2
By Euler algorithm,
om
y1 = y0 + h f (x0 , y0 )
= 0.5 + 0.2 f (0, 0.5)
h i h i
= 0.5 + 0.2 y0 − x02 + 1 = 0.5 + 0.2 0.5 − 02 + 1 = 2 + 0.2(1.5)
i.e., y (0.2) = 0.8
.c
8. Given y ′ = x + y, y(0) = 1, find y(0.1) by Euler’s method. (ND14)
5.5.2 Part B
ul
dy
Example 5.12. Using Euler’s method, find y(0.2), y(0.4) and y(0.6) from = x + y, y (0) = 1 with
dx
h = 0.2.

pa
Solution: Given y (0) (or) y0 = 1 ⇒ x0 = 0
&x1 = 0.2, x2 = 0.4, x3 = 0.4, h = 0.2
y′ = x+y
jin
We have to find y (0.2) = y1
Now, by Euler algorithm,
y1 = y0 + h f (x0 , y0 )
.re

i.e., y (0.2) = 1 + (0.2) (x + y) x=x0 = 1 + 0.2 (x0 + y0 ) = 1 + (0.2) (0 + 1)
y=y0
y (0.2) = 1.2 = y1

y2 = y1 + h f (x1 , y1 ) = 1.2 + (0.2) x1 + y1
= 1.2 + 0.2 [0.2 + 1.2] = 1.48 ⇒ y (0.4) = 1.48
w

y3 = y2 + h f (x2 , y2 ) = 1.48 + (0.2) x2 + y2
= 1.856 ⇒ y (0.6) = 1.856
w
∴ The result of the problem:
x y
w
x0 = 0 y0 = 1
x1 = 0.2 y1 = y0 + h f (x0 , y0 ) = 1.2
x2 = 0.4 y2 = y1 + h f (x1 , y1 ) = 1.48
x3 = 0.4 y3 = y2 + h f (x2 , y2 ) = 1.856
′
Example 5.13. Using Euler’s method solve y = x + y + xy, y(0) = 1 compute y at x = 0.1 by
taking h = 0.05.
Solution: Given y ′ = x + y + xy, x0 = 0, y0 = 1
h = 0.05.
x1 = x0 + h = 0 + 0.05 = 0.05
y1 = y (x1 ) = y(0.05)
om
To find y1
By Euler algorithm
y1 = y0 + h f (x0 , y0 )
= 1 + 0.05 f (0, 1)
.c
= 1 + 0.05[0 + 1 + (0)(1)]
y1 = 1.05
∴ y(0.05) = 1.05
ul
x2 = x1 + h = 0.05 + 0.05 = 0.1
y2 = y (x2 ) = y(0.1)
To find y2
pa
By Euler algorithm
jin
y2 = y1 + h f (x1 , y1 )
= 1.05 + 0.05 f (0.05, 1.05)
= 105 + 0.05[0.05 + 1.05 + (0.05)(1.05)]
y2 = 1.10762
.re
∴ y(0.1) = 1.10762
Example 5.14. Using Euler’s method find y(0.3) of y(x) satisfies the initial value problem
dy 1 2
= x + 1 y2 , y(0.2) = 1.1114.
dx 2
w
1 2
Solution: Given f (x, y) = x + 1 y2 and x0 = 0.2, y0 = 1.1114.
2
Take h = 0.1
w
x1 = x0 + h = 0.2 + 0.1 = 0.3

y1 = y (x1 ) = y(0.3)
w
To find y1
By Euler’s method
y1 = y0 + h f (x0 , y0 )
= 1.1114 + 0.1 f (0.2, 1.1114)
" #
1
= 1.1114 + 0.1 (0.2)2 + 1 (1.1114)2
2
y1 = 1.1756
∴ y(0.3) = 1.1756
dy y
Example 5.15. Using Euler’s method with h = 0.1 to solve the equation = , y(0) = 2 in
dx 1+x
the range 0 ≤ x ≤ 1.
y
om
Solution: Given f (x, y) = and x0 = 0, y0 = 2, h = 0.1
1+x
x1 = x0 + h = 0 + 0.1 = 0.1
y1 = y (x1 ) = y(0.1)
To find y1
.c
By Euler algorithm
y1 = y0 + h f (x0 , y0 )
ul
= 2 + 0.1 f (0, 2)
" #
2
= 2 + 0.1
1+0
pa y1 = 2.2
∴ y(0.1) = 2.2
Similarly
jin
x2 = x1 + h = 0.1 + 0.1 = 0.2 x3 = x2 + h = 0.2 + 0.1 = 0.3 x4 = x3 + h = 0.3 + 0.1 = 0.4
y2 = y (x2 ) = y(0.2) y3 = y (x3 ) = y(0.3) y4 = y (x4 ) = y(0.4)
∴ y2 = y1 + h f (x1 , y1 ) ∴ y3 = y2 + h f (x2 , y2 ) ∴ y4 = y3 + h f (x3 , y3 )
.re
= 2.2 + 0.1 f (0.1, 2.2) = 2.4 + 0.1 f (0.2, 2.4) = 2.6 + 0.1 f (0.3, 2.6)
" # " # " #
2.2 2.4 2.4
= 2.2 + 0.1 = 2.4 + 0.1 = 2.6 + 0.1
1 + 0.1 1 + 0.2 1 + 0.3
y2 = 2.4 y3 = 2.6 y4 = 2.8
∴ y(0.2) = 2.4 ∴ y(0.3) = 2.6 ∴ y(0.3) = 2.8
w
y5 = y4 + h f (x4 , y4 ) y6 = y5 + h f (x5 , y5 ) y7 = y6 + h f (x6 , y6 )

= 2.8 + 0.1 f (0.4, 2.8) = 3 + 0.1 f (0.5, 3) = 3.2 + 0.1 f (0.6, 3.2)
w
" # " # " #

2.8 3 3.2
= 2.8 + 0.1 = 3 + 0.1 = 3.2 + 0.1
1 + 0.4 1 + 0.5 1 + 0.6
y5 = 3 y6 = 3.2 y7 = 3.4
w
∴ y(0.5) = 3 ∴ y(0.6) = 3.2 ∴ y(0.7) = 3.4

y8 = y7 + h f (x7 , y7 ) y9 = y8 + h f (x8 , y8 ) y10 = y9 + h f (x9 , y9 )

= 3.4 + 0.1 f (0.7, 3.4) = 3.6 + 0.1 f (0.8, 3.6) = 3.8 + 0.1 f (0.9, 3.8)
" # " # " #
3.4 3.6 3.8
= 3.4 + 0.1 = 3.6 + 0.1 = 3.8 + 0.1
1 + 0.7 1 + 0.8 1 + 0.9
y8 = 3.6 y9 = 3.8 y10 = 4
∴ y(0.8) = 3.6 ∴ y(0.9) = 3.8 ∴ y(1) = 4
om
dy
Example 5.16. Solve = y + e x , y (0) = 0 for x = 0.2, 0.4 by using Euler’s method.
dx
Solution: y1 = 0.2, y2 = 0.484281
.c
dy
Example 5.17. Using Euler’s method, find the solution of the initial value problem =
dx
log (x + y) , y (0) = 2 at x = 0.2 by assuming h = 0.2.
ul
Solution: [y(0.2) = 2.0602]
1. Solve y ′ =
y−x
y+x
pa
, y(0) = 1 at x = 0.1 by taking h = 0.02 by using Euler’s method. (MJ13)

Solution: y(0.02) = 1.02, y(0.04) = 1.0392, y(0.06) = 1.0577, y(0.08) = 1.0756, ∴ y(0.1) = 1.0928
jin
5.6 Modified Euler’s method
Modified Euler’s method formula is

.re
" #
h h
In general, yn+1 = yn + h f xn + , yn + f (xn , yn )
2 2
" #
h h
When n = 0, y1 = y0 + h f x0 + , y0 + f (x0 , y0 )
2 2
w
" #
h h
When n = 1, y2 = y1 + h f x1 + , y1 + f (x, y)
2 2
" #
h h
When n = 2, y3 = y2 + h f x2 + , y2 + f (x2 , y2 )
2 2
w
5.6.1 Part B
w
Example 5.18. Solve y ′ = 1 − y, y (0) = 0 by modified Euler method.

Solution: Given y ′ = f (x, y) = 1 − y, x0 = 0, y0 = 0.
Let h = 0.1, x1 = 0.1, x2 = 0.2, x3 = 0.3
We have to find y1 , y2 , y3 .
By Modified Euler’s method:

" #
h 1
yn+1 = yn + h f xn + , yn + h f (xn , yn )
2 2
" #
h h
Here f (x0 , y0 ) = f (x = x0 , y = y0 ) = 1 − y0 = 1 − 0 = 1

y1 = y0 + h f x0 + , y0 + f (x0 , y0 )
2 2
" #
0.1 0.1
∴ y1 = 0 + (0.1) f 0 + ,0 + (1)
2 2
= 0.1 f [0.05, 0.05]
om

= (0.1) 1 − y x=0.05 = 1 − 0.05
y=0.05
y1 = 0.095
" #
h h
Now y2 = y1 + h f x1 + , y1 + f (x1 , y1 )

Here f (x1 , y1 ) = 1 − y1 = 0.905

.c
2 2
" #
0.1 0.1
i.e., y2 = 0.905 + (0.1) f 0.1 + , 0.095 + f (0.1, 0.095)
2 2
y2 = 0.18098
ul
" #
h h
Now y3 = y2 + h f x2 + , y2 + f (x2 , y2 ) Here f (x2 , y2 ) = 1 − y2 = 1 − 0.18098 = 0.81902
2 2pa
" #
0.1 0.1
i.e., y3 = 0.18098 + (0.1) f 0.2 + , 0.18098 + f (0.2, 0.18098)
2 2
y3 = 0.258787
∴ y0 = 0, y1 = 0.095, y2 = 0.18098, y3 = 0.258787

jin
dy x−y
Example 5.19. By using modified Euler method, find y (1.8) if = , y (2) = 1
dx x+y
Hint h = −0.2

Solution: [y(1.8) = 0.9349].

.re
dy
Example 5.20. Given + y − x2 = 0, y (0) = 1, y (0.1) = 0.9052, y (0.2) = 0.8213, find correct to four
dx
decimal places y (0.3), using modified Euler’s " method. " # #
h
′ 2
i h h
Solution: Hint y = x − y Ans : y (0.3) = y3 = y2 + h f x2 + , y2 + f (x2 , y2 ) = 0.7493
2 2
w

dy
1. Using Modified Euler’s method, find y(4.1) and y(4.2) if 5x + y2 − 2 = 0; y(4) = 1. (ND12)
dx
w

Solution: y(4.1) = 1.005, y(4.2) = 1.0098
2. Apply modified Euler’s method to find y(0.2) and y(0.4) given y ′ = x2 + y2 , y(0) = 1 by taking h = 0.2.
w
(ND14)
5.7 Fourth order Runge-Kutta method for solving first order equations
Fourth order Runge-Kutta method for solving I order Differential Equations

[Single step method]
dy
Given = y ′ = f (x, y)
dx
& y (x0 ) = y0
We have to find y (x1 ) =? = y1 , y (x2 ) =? = y2 , y (x3 ) =? = y3 , · · ·
om
To find f (xn+1 ) = y (xn+1 ) = yn+1 :
yn+1 = y (xn + h) = y (xn ) + ∆y = yn + ∆y

1
where ∆y = [k1 + 2k2 + 2k3 + k4 ]
6
.c
where k1 = h f (xn , yn )
!
h k1
k2 = h f xn + , yn +
2 2
!
h k2
ul
k3 = h f xn + , yn +
2 2
k4 = h f (xn + h, yn + k3 )
To find f (x1 ) = y (x1 ) = y1

pa
k1 = h f (x0 , y0 )
!
h k1
k2 = h f x0 + , y0 +
2 2
!
h k2
jin
k3 = h f x0 + , y0 +
2 2
k4 = h f (x0 + h, y0 + k3 )
1
∆y = [k1 + 2k2 + 2k3 + k4 ]
6
.re
∴ y1 = y (x0 + h) = y (x0 ) + ∆y = y0 + ∆y
5.7.1 Part A
1. State the fourth order Runge - Kutta algorithm. (ND2012)

w
Solution: The fourth order Runge - Kutta algorithm is
k1 = h f (x0 , y0 )
w
!
h k1
k2 = h f x0 + , y0 +
2 2
!
h k2
k3 = h f x0 + , y0 +
w
2 2
k4 = h f (x0 + h, y0 + k3 )
1
∆y = [k1 + 2k2 + 2k3 + k4 ]
6
∴ y1 = y (x0 + h) = y (x0 ) + ∆y = y0 + ∆y
5.7.2 Part B
dy
Example 5.21. Given = x + y2 , y (0) = 1, find y(0.1) & y(0.2) by Runge-Kutta method for IV
dx
order.
Solution: Given y ′ = f (x, y) = x + y2 , h = 0.1

&x0 = 0, y (x0 = 0) = 1
om
To find y(0.1) = y1
h i
k1 = h f x0 , y0 = (0.1) f (0.1) = (0.1) 0 + 12 = 0.1

" #
h k1 h i
k2 = h f x0 + , y0 + = (0.1) f [0 + 0.05, (1 + 0.05)] = (0.1) 0.05 + (1.05)2 = 0.11525
2 2
.c
" # " #
h k2 0.11525 h i
k3 = h f x0 + , y0 + = (0.1) f 0.05, 1 + = (0.1) 0.05 + (1.057625)2 = 0.116857
2 2 2
h i
k4 = h f x0 + h, y0 + k3 = (0.1) f [0.05, 1 + 0.116857] = (0.1) 0.05 + (1.116857)2 = 0.134737

ul
1 1
∆y = [k1 + 2k2 + 2k3 + k4 ] = [(0.1) + 2 (0.1) + 2 (0.11525) + (0.116857)]
6 6
= 0.11649
pa
∴ y (0.1) = y1 = y0 + ∆y = 1 + 0.11649 = 1.11649 ⇒ y1 = 1.11649
To find y (0.2) = y2

k1 = h f x1 , y1 = (0.1) f [0.1, 1.11649] = (0.1) [0.1 + (1.116492)] = 0.1347
jin
" # " #
h k1 0.1347
k2 = h f x1 + , y1 + = (0.1) f 0.1 + 0.05, 1.11649 + = (0.1) f [0.15, 1.18385] = 0.1552
2 2 2
" # " #
h k2 0.1552
k3 = h f x1 + , y1 + = (0.1) f 0.1 + 0.05, 1.11649 + = (0.1) f [0.15, 1.1941] = 0.1576
2 2 2
.re

k4 = h f x1 + h, y1 + k3 = (0.1) f [0.1 + 0.1, 1.11649 + 0.1576] = 0.18233
1
∆y = [k1 + 2k2 + 2k3 + k4 ] = 0.1571
6
y2 = y1 + ∆y = 1.11649 + 0.1571 = 1.27359
w
i.e.,
x 0 0.1 0.2
w
y 1 1.11649 1.27359
dy 2xy + e x
Example 5.22. Using R.K. method of fourth order solve for x = 1.4, from = with
dx x2 + xe x
w
y = 0 at x = 1.

Solution: y1 = y(1.2) = 0.140, ∴ y2 = y(1.4) = 0.27, with h = 0.2
dy p
Example 5.23. Solve y for x = 0.2, 0.4 from the given O.D.E. = x2 + y and y = 0.8 at x = 0.
dx
Solution: y1 = y (0.2) = 0.99029, y2 = y (0.4) = 0.120828
1. Using Runge-Kutta method of order four, find y when x = 1.2 in steps of 0.1 given that y ′ = x2 + y2
and y(1) = 0.5. (ND13)

y1 = y(1.1) = y0 + ∇y = 0.6428, ∴ y2 = y(1.2) = y1 + ∇y = 0.8278 with h = 0.1

Solution:
5.8 Fourth order Runge - Kutta method for solving II order differential
om
equation
Given f (y ′′ , y ′ , y) = g(x) with y (x0 ) , y ′ (x0 ). Find y (x1 ) = y1 , y ′ (x1 ) = y1 ′ [ or z (x1 ) = z1 ].
Let us set y ′ = z = f1 (x, y, z)

y ′′ = f2 (x, y, z)
.c
To find y1 To find z1
ul
k1 = h f1 (x0 , y0 , z0 ) ℓ1 = h f2 (x0 , y0 , z0 )
" # " #
h k1 ℓ1 h k1 ℓ1
k2 = h f1 x0 + , y0 + , z0 + ℓ2 = h f2 x0 + , y0 + , z0 +
2 2 2 2 2 2
h k2 ℓ2
i h k2 ℓ2
i
h h
k3 = h f1 x0 + 2 , y0 + 2 , z0 + 2 pa ℓ3 = h f2 x0 + 2 , y0 + 2 , z0 + 2

k4 = h f1 x0 + h, y0 + k3 , z0 + ℓ3 ℓ4 = h f2 x0 + h, y0 + k3 , z0 + ℓ3
1 1
∆y = [k1 + 2k1 + 2k2 + 2k3 ] ∆z = [ℓ1 + 2ℓ2 + 2ℓ3 + ℓ4 ]
6 6
y1 = y0 + ∆y z1 = z0 + ∆z
jin
Example 5.24. Consider the II order initial value problem y ′′ −2y ′ +2y = e2t sin twith y (0) = −0.4
and y ′ (0) = −0.6 using fourth order Runge-Kutta method, find y(0.2), z(0.2).
.re
Solution: Let t = x. ∴ Given equations can be written as
y ′′ − 2y ′ + 2y = e2x sin x (1)

y (x = 0) = −0.4
w
y ′ (x = 0) = −0.6
Here h = 0.2
w
Setting y ′ = z, y ′′ = z′
∴ (1) ⇒ z′ = 2z − 2y + e2x sin x

w
dy
Let f1 (x, y, z) = = z,
dx
2
d y dz
& f2 (x, y, z) = 2 = = 2z − 2x + e2x sin x
dx dx
ℓ1 = h f2 (x0 , y0 , z0 ) = (0.2) f2 (0, −0.4, −0.6)

k1 = h f1 (x0 , y0 , z0 ) = (0.2) f1 (0, −0.4, −0.6) h i
= (0.2) 2z0 − 2y0 + e2x0 sin x0
= (0.2) (z0 ) h i
= (0.2) 2 (−0.6) − 2 (−0.4) + e2(0) sin (0)
= (0.2) (−0.6) = −0.12
= −0.08
" # " #
h k1 ℓ1 h k1 ℓ1
k2 = h f1 x0 + , y0 + , z0 + ℓ2 = h f2 x0 + , y0 + , z0 +
" 2 2 2 # " 2 2 2 #
0.2 0.12 0.8 0.2 0.12 0.08
= (0.2) f1 0 + , (−0.4) − , (−0.6) − = (0.2) f2 0 + , −0.4 − , −0.6 −
om
2 2 2 2 2 2
= (0.2) f1 [0.1, −0.46, −0.64] = (0.2) f2 [0.1, −0.46, −0.64]
h i
= (0.2)(−0.64) = (0.2) 2 (−0.64) − 2 (−0.46) + e2(0.1) sin (0.1)
= −0.128 = −0.0476
" # " #
h k2 ℓ2 h k2 ℓ2
k3 = h f1 x0 + , y0 + , z0 + ℓ3 = h f2 x0 + , y0 + , z0 +
" 2 2 2 # " 2 2 2 #
0.2 0.128 0.0476 0.2 0.128 0.0476
.c
= (0.2) f1 0+ , −0.4− , −0.6− = (0.2) f2 0+ ,−0.4− , −0.6−
2 2 2 2 2 2
= (0.2) f1 [0.1, −0.464, −0.62381] = (0.2) f2 [0.1, −0.464, −0.62381]
h i
= (0.2) [−0.62381] = (0.2) 2 (−0.62381)−2 (−0.464)+e2(0.1) sin (0.1)
ul
= −0.12476 = −0.03954

k4 = h f1 x0 + h, y0 + k3 , z0 + ℓ3 ℓ4 = h f2 x0 + h, y0 + k3 , z0 + ℓ3
= (0.2) f1 [0+0.2,−0.4−0.12476,−0.6−0.03954]
pa = (0.2) f2 [0+0.2,−0.4−0.12476,−0.6−0.03954]
= (0.2) f1 [0.2, −0.52476, −0.63954] = (0.2) f2 [0.2, −0.52476, −0.63954]
h i
= (0.2) [−0.63954] = (0.2) 2 (−0.63954)−2 (−0.52476)+e2(0.2) sin (0.2)
= −0.12791 = 0.013366
1 1
∆y = [k1 + 2k1 + 2k2 + 2k3 ] ∆z = [ℓ1 + 2ℓ2 + 2ℓ3 + ℓ4 ]
6 6
jin
= −0.12557 = −0.04015
y1 = y (0.2) = y0 + ∆y z1 = z (0.2) = z0 + ∆z
= −0.52557 = −0.64015
Example 5.25. Given y ′′ + xy ′ + y = 0, y (0) = 1, y ′ (0) = 0, find the value of y(0.1) by using R. K.
.re
method of fourth order.

Solution: [y(0.1) = 0.9950].
Example 5.26. Given y ′′ − x (y ′ )2 + y2 = 0, y (0) = 1, y ′ (0) = 0, find the value of y(0.2) by using R.
K. method of fourth order.
w
Solution: [y(0.2) = 0.9801].

w
1. Find the value of y(0.1) by Runge-Kutta method of fourth order given y ′′ + xy ′ + y = 0, y(0) = 1 and
y ′ (0) = 0. (ND10)
w

Solution: y1 = y(0.1) = y0 + ∇y = 0.9950
2. Given y ′′ + xy ′ + y = 0, y(0) = 1, y ′ (0) = 0. Find the value of y(0.1) by using Runge-Kutta method of
fourth order. (ND11)

Solution: y1 = y(0.1) = y0 + ∇y = 0.9950
′′
3. Consider the second order initial value problem y − 2y ′ + 2y = e2t sin t with y(0) = −0.4 and
y ′ (0) = −0.6 using Fourth order Runge Kutta algorithm, find y(0.2). (MJ12)

Solution: y1 = y(0.2) = y0 + ∇y = −0.5159
4. Using Runge-Kutta method find y(0.2) if y ′′ = xy ′2 − y2 , y(0) = 1, y ′ (0) = 0, h = 0.2. (MJ13)

Solution: y1 = y(0.2) = y0 + ∇y = 0.9801
om
5. Given y ′′ + xy ′ + y = 0, y(0) = 1, y ′ (0) = 0 find the value of y(0.1) by Runge-Kutta’s method of fourth
order. (ND14)

Solution: y1 = y(0.1) = y0 + ∇y = 0.9950
.c
ul
5.9 Fourth order R. K. Method for simultaneous first order differential
equations
dy dz
pa
Solving the equation = f1 (x, y, z) & = f2 (x, y, z) with the initial conditions y (x0 ) = y0 , z (x0 ) = z0 .
dx dx
Now starting from with increments, (x0 , y0 , z0 ) with increments, ∆y & ∆z in y & z respectively.
jin
Use formula
k1 = h f1 (x0 , y0 , z0 ) ℓ1 = h f2 (x0 , y0 , z0 )
.re
! !
h k1 ℓ1 h k1 ℓ1
k2 = h f1 x0 + , y0 + , z0 + ℓ2 = h f2 x0 + , y0 + , z0 +
2 2 2! 2 2 2!
h k2 ℓ2 h k2 ℓ2
k3 = h f1 x0 + , y0 + , z0 + ℓ3 = h f2 x0 + , y0 + , z0 +
2 2 2 2 2 2
k4 = h f1 (x0 + h, y0 + k3 , z0 + ℓ3 ) ℓ4 = h f2 (x0 + h, y0 + k3 , z0 + ℓ3 )
1 1
∆y = (k1 + 2k2 + 2k3 + k4 ) ∆z = (ℓ1 + 2ℓ2 + 2ℓ3 + ℓ4 )
w
6 6
y1 = y0 + ∆y z1 = z0 + ∆z
w
5.9.1 Part B
w
dz dy
Example 5.27. Solving the system of differential equation = xz + 1, = −xy for x = 0.3
dx dx
using fourth order R. K. method, the initial values are x = 0, y = 0, z = 1.
Solution: Given x0 = 0, y0 = 0, z0 = 1, h = 0.3.
f1 (x, y, z) = xz + 1 f2 (x, y, z) = −xy

k1 = h f1 (x0 , y0 , z0 ) = (0.3) (x0 z0 + 1) ℓ1 = h f2 (x0 , y0 , z0 ) = (0.3) [−0 (0)]
= (0.3) (0 + 1) = 0.3 =0
! !
h k1 ℓ1 h k1 ℓ1
k2 = h f1 x0 + , y0 + , z0 + ℓ2 = h f2 x0 + , y0 + , z0 +
2 2 2 2 2 2
= 0.345 = −0.007
! !
h k2 ℓ2 h k2 ℓ2
k3 = h f1 x0 + , y0 + , z0 + ℓ3 = h f2 x0 + , y0 + , z0 +
2 2 2 2 2 2
om
= 0.3448 = −0.0078
k4 = h f1 (x0 + h, y0 + k3 , z0 + ℓ3 ) ℓ4 = h f2 (x0 + h, y0 + k3 , z0 + ℓ3 )
= 0.3893 = −0.031032
1 1
∆y = (k1 + 2k2 + 2k3 + k4 ) = 0.34482 ∆z = (ℓ1 + 2ℓ2 + 2ℓ3 + ℓ4 ) = −0.01011
6 6
y1 = y0 + ∆y = 0 + 0.34482 z1 = z0 + ∆z = 1 + (−0.01011)
.c
⇒ y (0.3) = 0.34482 ⇒ z (0.3) = 0.98989
dy dz
Example 5.28. Solve the simultaneous differential equation = 2y + z, = y − 3z, y (0) =
dx dx
0, z (0) = 0.5 for y(0.1) and z(0.1) using R. K. method of fourth method.
ul
Solution: Given x0 = 0, y0 = 0, z0 = 0.5, h = 0.1
[y (0.1) = 0.04814, z (0.1) = 0.37263].
pa dy dz
Example 5.29. Using the R. K. method, tabulate the solution of the system = x + z, =
dx dx
x − y, y = 0, z = 0, when x = 0 at intervals of h = 0.1 from x = 0.0 to x = 0.2.
Solution: Given x0 = 0, y0 = 0, z0 = 1, h = 0.1
[Hint: y (0.1) = 0.1050, z (0.1) = 0.9998

jin
y (0.2) = 0.2199, z (0.2) = 0.9986]

dy dz
.re
1. Solve y(0.1) and z(0.1) from the simultaneous differential equations = 2y + z; = y − 3z; y(0) =
dx dx
0, z(0) = 0.5 using Runge-Kutta method of the fourth order. (ND12)
Solution: Given x0 = 0, y0 = 0, z0 = 0.5, h = 0.1
[y (0.1) = 0.04814, z (0.1) = 0.37263].
Multi step methods

w
w
h(= Interval length)

Given (x0 , y0 ) , x1 , x2 , x3 , x4 : y4,p = known by Multistep predictor method
y1 , y2 , y3 are given (or) known by Singular method y4,c = known by Multistep corrector method
w
i.e.,
Given x : x0 x1 x2 x3 x4 x4
Given or find y : y0 y1 y2 y3 y4,p y4,c
find by Multistep predictor method find by Multistep corrector method
Multi step methods:
1. Milne’s forth predictor corrector methods for solving first order equations
2. Adam’s Bash-forth predictor corrector methods for solving first order equations
5.10 Milne’s forth predictor corrector methods for solving first order
om
equations
Milne’s predictor and corrector methods (multistep method)

Milne’s predictor formula
4h h ′ i
yn+1,p = yn−3 + 2y n−2 − y ′n−1 + 2y ′n
3
.c
Milne’s corrector formula
hh ′ i
yn+1,c = yn−1 + y n−1 + 4y ′n + y ′n+1
ul
3
5.10.1 Part A
pa
1. What are multi-step methods? How are they better than single step methods?
Solution : 1. Milne’s predictor - correction method.

(ND2010)
2. Adams-Bashforth predictor - correction method.

jin
In the single step methods, it is not possible to get any information about truncation error.
In the multi step methods, it is possible to get easily a good estimate of the truncation error.
2. State the Milne’s predictor and corrector formulae. (AM2014)

.re
3. State the Milne’s predictor-corrector formulae. (ND14)
5.10.2 Part B
dy 1
Example 5.30. Given = [x + y] , y (0) = 2, y (0.5) = 2.636, y (1) = 3.595, y (1.5) = 4.968 by Milne’s
w
dx 2
method to find y(2).
Solution: Given
w
x0 = 0 x1 = 0.5 x2 = 1 x3 = 1.5 x4 = 2
y0 = 2 y1 = 2.636 y2 = 3.595 y3 = 4.968 y4 =?
1
w
Here h = 0.5 Given y ′ = x + y = f (x, y) By Milne’s predictor formula,

2
4h h ′ i
yn+1,p = yn−3 + 2y n−2 − y ′n−1 + 2y ′n
3
4h h ′ i
when n = 3, y3+1,p = y0 + 2y 1 − y ′2 + 2y ′3 (1)
3
1 1
Now, y ′1 = x1 + y1 = [0.5 + 2.636] = 1.568
2 2
′ 1 1
y2 = x2 + y2 = [1 + 3.595] = 2.2975
2 2
′ 1 1
y3 = x3 + y3 = [1.5 + 4.968] = 3.234
2 2
4 (0.5)
(1) ⇒ y4,p = 2 + [2 (1.568) − 2.2975 + 2 (3.234)]
3
= 6.871 (= y4 , say)
om
By Milne’s corrector formula,
hh ′ i
yn+1,c = yn−1 + y n−1 + 4y ′n + y ′n+1
3
hh ′ i
when n = 3, y4,c = y2 + y 2 + 4y ′3 + y ′4 (2)
3
.c
1 1
Now, y ′4 = x4 + y4 = [2 + 6.871] = 4.4355
2 2
0.5
(2) ⇒ y4,c = 3.595 + [2.2975 + 4 (3.234) + 4.4355]
3
ul
0.5
= 3.595 + (19.669) = 6.8732
3
∴ Corrected value of y at 2 is y (2) = 6.8732.
dy
Example 5.31. Given
dx
pa
= x3 + y, y (0) = 2, the values of y (0.2) = 2.073, y (0.4) = 2.452, y (0.6) = 3.023
are got by R.K. Method. Find y (0.8) by Milne’s method.
h i
Solution: y4,p = 4.1659, y4,c = 3.7953
Example 5.32. Solve y ′ = x − y2 , 0 ≤ x ≤ 1, y (0) = 0, y (0.2) = 0.02, y (0.4) = 0.0795, y (0.6) = 0.1762
jin
by Milne’s method to find y(0.8) and y(1).

Solution: y(0.8) = 0.3046, y(1) = 0.4515 .

.re
1. Use Milne’s
predictor-corrector formula to find y(0.4), given
2 2
dy 1 + x y
= , y(0) = 1, y(0.1) = 1.06, y(0.2) = 1.12 and y(0.3) = 1.21. (AM10)
dx 2 h i
Solution: y4,p = 1.2771, y4,c = 1.2797
w
dy 1
2. Given that = 1 + x2 y2 ; y(0) = 1; y(0.1) = 1.06; y(0.2) = 1.12 and y(0.3) = 1.21, evaluate y(0.4)
dx 2
and y(0.5) by Milne’s predictor corrector method. (ND11)
h i
Solution: y4,p = 1.2771, y4,c = 1.2797
w
dy
3. Given that = 1 + y2 ; y(0.6) = 0.6841, y(0.4) = 0.4228, y(0.2) = 0.2.27, y(0) = 0, find y(−0.2) using
dx
Milne’s method. (ND12)
w
h i
Solution: y4,p = −0.2003, y4,c = −0.2027
1
4. Use Milne’s method to find y(0.8), given y ′ = , y(0) = 2,y(0.2) = 2.0933, y(0.4) = 2.1755, y(0.6) =
x+y
2.2493. (ND13)
h i
Solution: y4,p = 2.3162, y4,c = 2.3164
5. Given 5xy ′ + y2 = 2, y(4) = 1, y(4.1) = 1.0049, y(4.2) = 1.0097, y(4.3) = 1.0143. Compute y(4.4) using
h i
Solution: y4,p = 1.01897, y4,c = 1.01874
5.11 Adams-Bash forth predictor corrector methods for solving first

order equations
om
Adams predictor and corrector methods(efficient multistep method) [Adam’s-Bashforth
method]
Adam’s predictor formula :
h h i
yn+1,p = yn + 55y ′n − 59y ′n−1 + 37y ′n−2 − 9y ′n−3
24
.c
Adam’s corrector formula :
h h ′ i
yn+1,c = yn + 9y n+1 + 19y ′n − 5y ′n−1 + y ′n−2
24
ul
5.11.1 Part A
1. Write the Adam-Bashforth predictor and corrector formulae. (MJ2010)

(Or) State Adam’s predictor-corrector formulae.
pa
Solution : Adam’s predictor and corrector formulas are
h h i
(ND2011)
yk+1,p = yk + 55y ′k − 59y ′k−1 + 37y ′k−2 − 9y ′k−3

24
h h ′ i
yk+1,c = yk + 9y k+1 + 19y ′k − 5y ′k−1 + y ′k−2
jin
24
5.11.2 Part B
dy
Example 5.33. Given = x2 (1 + y) , y (1) = 1, y (1.1) = 1.233, y (1.2) = 1.548, y (1.3) = 1.979, evaluate
dx
.re
y(1.4) by Adam’s-Basforth method.

Solution: Given
x0 = 1 x1 = 1.1 x2 = 1.2 x3 = 1.3 x4 = 1.4
y0 = 1 y1 = 1.233 y2 = 1.548 y3 = 1.979 y4 =?
Here h = 0.1
w
By Adam’s predictor formula,

h h i
yn+1,p = yn + 55y ′n − 59y ′n−1 + 37y ′n−2 − 9y ′n−3
w
24
h h i
when n = 3, y4,p = y3 + 55y ′3 − 59y ′2 + 37y ′1 − 9y ′0 (1)
24
w
Here y ′0 = x02 (1 + y0 ) = 12 [1 + 1] = 2
y ′1 = x12 (1 + y1 ) = (1.1)2 [1 + 1.233] = 2.70193
y ′2 = x22 (1 + y2 ) = (1.2)2 [1 + 1.548] = 3.66912
y ′3 = x32 (1 + y3 ) = (1.3)2 [1 + 1.979] = 5.0345
0.1
(1) ⇒ y4,p = 1.979 + [55 (5.0345) − 59 (3.66912) + 37 (2.70193) − 9 (2)]
24
= 2.5723
By Adam’s corrector method,
h h ′ i
yn+1,c = yn + 9y n+1 + 19y ′n − 5y ′n−1 + y ′n−2
24
h h ′ i
when n = 3 , y4,c 9y 4 + 19y ′3 − 5y ′2 + y ′1

= y3 + (2)
om
24
y ′4 = (x4 )2 (1 + y4 ) = (1.4)2 [1 + 2.5871] = 7.0017
0.1
∴ (2) ⇒ y4,c = 1.979 + [9 (7.030716) + 19 (5.0345) − 5 (3.60912) + (2.70193)]
24
= 2.5749
.c
Example 5.34. Using Adam’s Bashforth method, find y(4.4) given 5xy ′
+ y2 = 2, y(4) = 1,
y(4.1) = 1.0049, y(4.2) = 1.0097 & y(4.3) = 1.0143.
2 − y2 h i
Solution: Here y ′ = , ∴ y4,p = 1.0186&y4,c = 1.0187 .
ul
2x
Example 5.35. Evaluate y(0.9), using Adam Bashforth’s predictor-corrector method, given
1
dy
that = xy 3 , y(1) = 1, y(1.1) = 1.106814, y(1.2) = 1.22787 and y(1.3) = 1.36412.
dx
Solution: Here h = −0.1,
pa h i
∴ y4,p = 0.906518&y4,c = 0.906520 .

dy
jin
1. Given = xy + y2 , y(0) = 1, y(0.1) = 1.1169, y(0.2) = 1.2774, y(0.3) = 1.5041. Use Adam’s method to
dx
estimate y(0.4). (AM10)
h i
Solution: y4,p = 1.8341, y4,c = 1.8389
2. Using Adam’s method to find y(2) if y ′ = (x + y)/2, y(0) = 2, y(0.5) = 2.636, y(1) = 3.595, y(1.5) = 4.968.
.re
(MJ13)
h i
Solution: y4,p = 6.8708, y4,c = 6.8731
3. Using Adam’s Bashforth method, find y(4.4) given that 5xy ′ + y2 = 2, y(4) = 1, y(4.1) = 1.0049, y(4.2) =
1.0097 and y(4.3) = 1.0143. (AM14)
w
h i
Solution: y4,p = 1.0186, y4,c = 1.0187
5.11.4 Anna University Questions (Taylor’s, RK, Adam, Milne)

w
dy
1. Given = xy+y2 , y(0) = 1, y(0.1) = 1.1169 and y(0.2) = 1.2774, find (i) y(0.3) by Runge-Kutta method
dx
of fourth order and (ii) y(0.4) by Milne’s method. (ND10)
w
h i
Solution: (i)y(0.3) = 1.5041, (ii)y4,p = 1.8344, y4,c = 1.8387
2. Using Runge-Kutta method of fourth order, find y for x = 0.1, 0.2, 0.3 given that y ′ = xy + y2 , y(0) = 1.
Continue the solution at x = 0.4 usign Milne’s method. (AM11)
h i
Solution: y(0.1) = 1.11689, y(0.2) = 1.2774, y(0.3) = 1.5041; y4,p = 1.8344, y4,c = 1.8387
3. Solve y ′ = x − y2 , y(0) = 1 to find y(0.4) by Adam’s method. Starting solutions required are to be
obtained using Taylor’s method using the value h = 0.1. (AM11)
h i
Solution: TS:Py(0.1) = 0.9138, y(0.2) = 0.8512, y(0.3) = 0.8076; MMy4,p = 0.7799, y4,c = 0.7797
4. Given that y ′′ + xy ′ + y = 0, y(0) = 1, y ′ (0) = 0 obtain y for x = 0.1, 0.2 and 0.3 by Taylor’s series
method and find the solution for y(0.4) by Milne’s method. (MJ12)
h i
Solution: TS: y(0.1) = 0.995, y(0.2) = 0.9802, y(0.3) = 0.956; MM: y4,p = 0.9232, y4,c = 0.9232
om
dy
5. Using Runge Kutta method of fourth order, find the value of y at x = 0.2, 0.4, 0.6 given =
dx
3
x + y, y(0) = 2. Also find the value of y at x = 0.8 using Milne’s predictor and corrector method.
(AM14)
h i
Solution: RK: y(0.2) = 2.073, y(0.4) = 2.452, y(0.6) = 3.023; MM: y4,p = 4.1664, y4,c = 3.79536
.c
1. Using Taylor series method to find y(0.1) if y ′ = x2 + y2 , y(0) = 1. (MJ13)

dy
2. Using Taylor’s series method, find y at x = 1.1 by solving the equation = x2 + y2 ; y(1) = 2.
ul
dx
Carryout the computations upto fourth order derivative. (AM14)
dy
3. Using Taylor’s series method, find y at x = 0 if = x2 y − 1, y(0) = 1. (ND14)
dx
4. Solve y ′ =
y−x
y+x
pa
, y(0) = 1 at x = 0.1 by taking h = 0.02 by using Euler’s method. (MJ13)
dy
5. Using Modified Euler’s method, find y(4.1) and y(4.2) if 5x + y2 − 2 = 0; y(4) = 1. (ND12)
dx
jin
6. Apply modified Euler’s method to find y(0.2) and y(0.4) given y ′ = x2 + y2 , y(0) = 1 by taking h = 0.2.
(ND14)
7. Using Runge-Kutta method of order four, find y when x = 1.2 in steps of 0.1 given that y ′ = x2 + y2
and y(1) = 0.5. (ND13)
.re
′′
8. Consider the second order initial value problem y − 2y ′ + 2y = e2t sin t with y(0) = −0.4 and
y ′ (0) = −0.6 using Fourth order Runge Kutta algorithm, find y(0.2). (MJ12)
9. Using Runge-Kutta method find y(0.2) if y ′′ = xy ′2 − y2 , y(0) = 1, y ′ (0) = 0, h = 0.2. (MJ13)

w
10. Given y ′′ + xy ′ + y = 0, y(0) = 1, y ′ (0) = 0 find the value of y(0.1) by Runge-Kutta’s method of fourth
order. (ND14)
dy dz
w
11. Solve y(0.1) and z(0.1) from the simultaneous differential equations = 2y + z; = y − 3z; y(0) =
dx dx
0, z(0) = 0.5 using Runge-Kutta method of the fourth order. (ND12)
w
12. Use Milne’s

predictor-corrector formula to find y(0.4), given
2 2
dy 1 + x y
= , y(0) = 1, y(0.1) = 1.06, y(0.2) = 1.12 and y(0.3) = 1.21. (AM10)
dx 2
1
13. Use Milne’s method to find y(0.8), given y ′ = , y(0) = 1, for x = 0.1 and 0.2 correct to four
x+y
decimal places. (ND13)
14. Given 5xy ′ + y2 = 2, y(4) = 1, y(4.1) = 1.0049, y(4.2) = 1.0097, y(4.3) = 1.0142. Compute y(4.4) using
15. Using Adam’s method to find y(2) if y ′ = (x + y)/2, y(0) = 2, y(0.5) = 2.636, y(1) = 3.595, y(1.5) = 4.968.
(MJ13)
16. Using Adam’s Bashforth method, find y(4.4) given that 5xy ′ + y2 = 2, y(4) = 1, y(4.1) = 1.0049, y(4.2) =
1.0097 and y(4.3) = 1.0143. (AM14)
om
dy
17. Given = xy+y2 , y(0) = 1, y(0.1) = 1.1169 and y(0.2) = 1.2774, find (i) y(0.3) by Runge-Kutta method
dx
of fourth order and (ii) y(0.4) by Milne’s method. (ND10)
dy
18. Using Runge Kutta method of fourth order, find the value of y at x = 0.2, 0.4, 0.6 given =
dx
3
x + y, y(0) = 2. Also find the value of y at x = 0.8 using Milne’s predictor and corrector method.
(AM14)
.c
ul
pa
jin
.re
w
w
w
w
w
w
.re
jin
pa
ul
.c
om

Statistics and Numerical Methods Notes

Uploaded by

Copyright:

Available Formats

Statistics and Numerical Methods Notes

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Statistics and Numerical Methods Notes

Uploaded by

Copyright:

Available Formats

MA8452 STATISTICS AND NUMERICAL METHODS LTPC

• To introduce the basic concepts of solving algebraic and transcendental equations.

• To introduce the numerical techniques of interpolation in various intervals and numerical

UNIT I TESTING OF HYPOTHESIS pa 12

Sampling distributions - Estimation of parameters - Statistical hypothesis - Large sample

UNIT III SOLUTION OF EQUATIONS AND EIGENVALUE PROBLEMS 12

UNIT IV INTERPOLATION, NUMERICAL DIFFERENTIATION AND NUMERICAL

Lagrange’s and Newton’s divided difference interpolations - Newton’s forward and

backward difference interpolation - Approximation of derivates using interpolation

", Tata McGraw Hill Edition, 2004.

Scientists", 8 th Edition, Pearson Education, Asia, 2007.

3 Solution of Equations and Eigenvalue Problems 5

3.6 Pivoting - Gauss Jordan method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

3.8.3 Anna University Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

3.10 Eigen values of a matrix by Power method . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

4 Interpolation, Numerical Differentiation and Numerical Integration 51

4.7 Numerical Differentiation and Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

4.8.3 Maxima and Minima of a tabulated function . . . . . . . . . . . . . . . . . . . . . . 79

4.8.6 Anna University Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

5 Numerical Solution of Ordinary Differential Equations 105

5.2.2 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

5.3.1 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

5.4.1 Part B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

5.6 Modified Euler’s method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

UNIT- I TESTING OF HYPOTHESIS

A population consists of collection of individual units, which may be person’s or experimental

Pop. mean = ; Pop. S.D = ; P - Pop. proportion

Download Useful Materials from Rejinpaul.com

ie., either H1 : 1 2 (or) 1 2

of significance are used.

Critical Region (or) Region of Rejection

One tailed test 2.33 1.645 1.28

Two tailed test and one-tailed tests:

The alternative hypothesis H1 : 1 2 is taken in two tailed test for H 0 : 1 = 2

as one tail test H1 : 1 2 (or) 1 2

Type I and type II Error

Type I Error : Rejection of null hypothesis when it is correct

Procedure for testing Hypothesis:

Test of Hypothesis (Large Sample Tests)

Type - I: (Test of significance of single mean)

we choose = 0.05(5%) (or) 0.01(1%) as the Level of significance

the test statistic is

Hence it is a two-tailed test

Level of significance = 0.05

Since |Z| = 1.5 < 1.96, H 0 is accepted at 5% level of significance

i.e., the sample comes from a population with mean 3.38cm

Given x 7.8, = 8, n = 50, = 0.5

i.e., the manufacturer’s claim is not accepted

It is one tailed test.

Given x 42.8, = 40, n = 200, = 6.89

Since |Z| =5.747> 2.33, H 0 is rejected at 1% level.

The formulated null and alternative hypothesis is,

The test statistic 'Z' is defined by

We use the los is 0.05 (or ) 0.01

If |Z| < Z , H0 is accepted at %Los

otherwise, H0 is rejected at %Los

ie., the samples are drawn from the sample population

The test statistic Z =

Given x1 67.5; n1 = 1000

x2 68; n2 = 2000; =2.5