R.K. Jain 0 S.R.K.

Iyengar

Advanced
Engineering Mathematics

@
Alpha Science International Ltd.
Pangbourne England
R.K. Jain
Professor
S.R.K. Iyengar
Professor and Former Head
Department of Mathematics
Indian Institute of Technology
New Delhi 110 016, India

Copyright © 2002

Alpha Science International Ltd.

P.O, Box 4067, Pangbourne RG8 SUT, UK

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system
or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or
otherwise, without the prior written permission of the publishers.

ISBN 1-84265-086-6

Printed in India.
 To Our Parents

Bhagat Ram Jain and Sampati Devi Jain

&
S.T.V. Raghavacharya and Rajya Lakshmi
whose memories had always been an inspiration
 Preface
i
mathematics
This book is based on the experience and the lecture notes of the authors while teaching
of Technolo gy, Delhi for more than three
courses to engineering students at the Indian Institute
inspirati on for introduct ion of
decades. A number of available textbooks have been a source of
of these books for their
concepts and formulation of problems. We are thankful to the authors
indirect help.
ng
This comprehensive textbook covers syllabus for two courses in Mathematics for engineeri
on the presen-
students in various Institutes, Universities and Engineering Colleges. The emphasis is
in an intelligibl e and easy to understan d manner.
tation of the fundamentals and theoretical concepts
to make it an effective tool to arouse
Each chapter in the book has been carefully planned
ng and scientific problems.
interest in the study and application of mathematics to solve engineeri
sets of
Simple and illustrative examples are used to explain each theoretical concept. Graded
to understan d
examples and exercises are given in each chapter, which will help the students
problems in the
every important concept. The book contains 682 solved examples and 2984
and hints for difficult problems are given at the end of each
exercises. Answers to every problem
students for self-learn ing. While some problems emphasize the
chapter which will motivate the
these concepts
theoretical concepts, others provide enough practice and generate confidence to use
and lucid presentat ion of both the theory and
in problem solving. This textbook offers a logical
problem solving techniques so that the student is not lost in unnecessa ry details.
all the engineering
We hope that this textbook will meet the requirements and the expectations of
students.
/exclusion
We will gratefully receive and acknowledge every comment, suggestions for inclusion
of topics and errors in the book, both from the faculty and the students.
ement
We are grateful to our former teachers, colleagues and well wishers for their encourag
thankful to our students for their feed back. We are
and valuable suggestions. We are also
grateful to the authorities of {IT Delhi for providing us their support.
Publishing
We extend our thanks to the editorial and the production staff of M/s Narosa
enthusia sm in the preparati on
House, in particular Mr. Mohinder Singh Sejwal, for their care and
of this book.
Vinod Jain
Last. but not the least, we owe a lot to our family members, in particular, our wives
and rejuvenat ing.
and Seetha Lakshmi whose encouragement and support had always been inspiring
We appreciate their patience during our long hours of work day and night.

R.K. Jain
New Delhi
S.R.K. [yYENGAR
October 2001

Seaeaeienitioemtlen el - - = —_ Fer) eens tay sechapehaRemyay a:hya nen ei ee

re re ae) RB ER = OSESFAAP “8S eg oa re oes
RL
i
Se

Contents

Preface Vu

1. Functions of a Real Variable

1.1 Introduction /
1.2 Limits, Continuity and Differentiability /
1.2.1 Limit ofa Function /
1.2.2 Continuity of a Function 3
1.2.3. Derivative of a Function 5
1.3 Application of Derivatives and Taylor's Series /0
1.3.1 Differentials and Approximations /0
1.3.2 Mean Value Theorems //
1.3.3 Indeterminate Forms /4
1.3.4 Increasing and Decreasing Functions /5
1.3.5 Maximum and Minimum Values of a Function /7
1.3.6 Taylor's Theorem and Taylor's Series /9
1.4 Integration and its Applications 26
1.4.1 Indefinite Integrals 26
1.4.2 Definite Integrals 26
1.4.3 Areaofa Bounded Region 28
144 Are Length of a Plane Curve 32
1.4.5 Volume ofa Solid 33
1.4.6 Surface Area of a Solid of Revolution 40
1.5 Improper Integrals 45
1.5.1 Improper Integrals of the First Kind 45
1.5.2 Improper Integrals of the Second Kind 49
1.5.3. Absolute Convergence of Improper Integrals 54
1.5.4 Beta and Gamma Functions 56
1.5.5 Improper Integrals Involving a Parameter 63
1.6 Answers and Hints 69
Functions of Several Real Variables 76
2.1 Introduction 76
L2 Functions of Two Variables 76
2.2.4 Limits 79
2.2.2 Continuity 8&2
23 Partial Derivatives 87
2.3.1 Total Differential and Derivatives 92
2.3.2 Approximation by Total Differentials 97

0 eeee eee ee
x Contents
te

2.3.3. Derivatives of Composite and Implicit Functions 99

2.4 Higher Order Partial Derivatives 107
2.4.1. Homogeneous Functions //0
2.4.2 Taylor’s Theorem //3
2.5 Maximum and Minimum Values of Functions /2/
2.5.1 _ Lagrange Method of Multipliers /26
2.6 Multiple Integrals 130
2.6.1 Double Integrals /30
2.6.2 Triple Integrals /40
2.6.3 Change of Variables in Integrals 43
2.7 Answers and Hints 1/50
3. Matrices and Eigenvalue Problems 157
3.1 Introduction /57
3.2 Matrices 1/57
3.2.1 Matrix Algebra /59
3.2.2 Some Special Matrices /6/
3.2.3 Determinants /63
3.2.4 Inverse of a Square Matrix /68
3.2.5 Solution of n x n Linear System of Equations /70
3.3 Vector Spaces 177
3.3.1 Subspaces /80
3.3.2 Linear Independence of Vectors /83
3.3.3 Dimension and Basis /85
3.3.4 Linear Transformations /88
3.4 Solution of General Linear System of Equations 199
3.4.1 Existence and Uniqueness of the Solution 200
3.4.2 Elementary Row and Column Operations 20/
3.4.3 Echelon Form of a Matrix 202
3.4.4 Gauss Elimination Method for Non-homogeneous Systems 205
3.4.5 Homogeneous System of Linear Equations 209
3.4.6 Gauss Jordan Method to Find the Inverse of a Matrix 2/0
3.5 Eigenvalue Problems 2/3
3.5.1 Eigenvalues and Eigenvectors 2/4
3.5.2 Similar and Diagonalisable Matrices 22?
3.5.3. Special Matrices 228
3.5.4 Quadratic Forms 232
3.6 Answers and Hints 238
4. Ordinary Differential Equations of First Order 247
4.1 Introduction 247
4.2 Formation of Differential Equations 248
4.3 Solution of a Differential Equation 249
4.4 Initial and Boundary Value Problems 25/
4.5 Solution of Equations in Separable Form 253
4.5.1 Equations Reducible to Separable Form 256
4.6 Exact First Order Differential Equations 26/
4.6.1 Integrating Factors 264
 Contents Xi
ree LEE

4.7 Linear First Order Equations 27]

4.8 Some Special First Order Equations 275
48.] Bernoulli Equation 275
4.8.2 Riccati Equation 2/7
4.8.3 Clairaut’s Equation 278
4.9 Orthogonal Trajectories of a Given Family of Curves 280
4.10 Existence and Uniqueness of Solutions 255
4.10.1 Picard’s Iteration Method of Solution 289
4.11 Answers and Hints 292
Linear Differential Equations 300
5.1 Introduction 300
5.2 Solution of Linear Differential Equations 30/
5.2.1] Linear Independence and Dependence 303
5.3 Methods for Solution of Linear Equations 308
5.3.1 Differential Operator D 308
5.3.2 Solution of Second Order Linear Homogeneous Equations
with Constant Coefficients 3/0
§.3.3. Method of Reduction of Order for Variable Coefficient
Linear Homogeneous Second Order Equations 3/5
5.3.4 Solution of Higher Order Homogeneous Linear Equations
with Constant Coefficients 3/8
5.4 Solution of Non-homogeneous Linear Equations 324
5.4.1 Method of Variation of Parameters 326
5.4.2 Method of Undetermined Coefficients 33/
5.4.3 Solution of Euler-Cauchy Equation 336
5.5 Operator Methods for Finding Particular Integrals 343
5.5.1 Case r(x)=e% F344
5.5.2 Case r(x )=cos (ox) orsin(ax ) 348
§.5.3 Caser(x)=x",@>OQOand Integer 35]
5.6 Simultaneous Linear Equations 354
5.6.1 Solution of First Order Systems by Matrix Method 357
5.6.2. Method of Undetermined Coefficients to Find Particular Integral 36/
5.6.3. Method of Diagonalisation to Find Particular Integral 364
5.7 Answers and Hints 367

Series Solution of Differential Equations 377

6.1 Introduction 377
6.2 Ordinary and Singular Points of an Equation 377
6.3 Power Series Solution 350
6.4 Series Solution about a Regular Singular Point: Frobenius
Method 39/
6.5 Answers and Hints 404

Legendre Polynomials, Bessel Functions and Sturm-Liouville Problem 410

7.1 Introduction 4/0
7.2 Legendre Differential Equation and Legendre Polynomials 4/0

SE mS RNS SEE PO 2 =8 2a ce eo a lh i ern ae be oe dee ee a ec ne ae ae EE

x11 Contents
Sr eee
7.2.1 Rodrigue’s Formula 4/2
7.2.2 Generating Function for Legendre Polynomials 4/4
7.2.3, Recurrence Relations for Legendre Polynomials 4/5
7.2.4 Orthogonal and Orthonormal Functions 4/7
7.2.5 Orthogonal Property of Legendre Polynomials 420
7.2.6 Fourier-Legendre Series 42]
7.3 Bessel’s Differential Equation and Bessel’s Functions 424
7.3.1 Bessel’s Function of the First Kind 429
7.3.2 Bessel’s Function of the Second Kind 435
7.4 Sturm-Liouville Problem 44]
7.4.1 Orthogonality of Bessel Functions 447
7.4.2 Fourier-Bessel Series 450
7.5 Answers and Hints 454

8. Laplace Transformation 459

8.1 Introduction 459
8.2 Basic Theory of Laplace Transforms 459
8.3 Laplace Transform Solution of Initial Value Problems 465
8.3.1 Laplace Transforms of Derivatives 465
8.3.2 Laplace Transforms of Integrals 469
8.4 Translation Theorems (Shifting Theorems) 472
8.4.1 Heaviside Function or Unit Step Function 475
8.5 Laplace Transform of Dirac-delta Function and More Properties of
Laplace Transforms 483
8.5.1 Laplace Transform of Dirac-delta Function 483
8.5.2 Differentiation of Laplace Transform +86
8.5.3 Integration of Laplace Transform 490
8.5.4 Convolution Theorem 492
8.6 Laplace Transform of Periodic Functions 497
8.7 Laplace Transform Method for the Solution of Some Partial
Differential Equations 503
8.8 Answers and Hints 5/9
9. Fourier Series, Fourier Integrals and Fourier Transforms 529
9.1 Introduction 529
9.2 Fourier series 529
9.2.1 Fourier Series of Even and Odd Functions 535
9.2.2 Convergence of Fourier Series 537
9.3 Fourier Half-range Series 544
9.3.1 Complex Form of Fourier Series 546
9.4 Fourier Integrals 549
9.5 Application of Fourier Series: Separation of Variables Solution
of Linear Partial Differential Equations 557
9.5.1 Classification of Linear Second Order Partial Differential Equations 557
9.5.2 Separation of Variables Method (Fourier Method) 559
9.5.3. Fourier Series Solution of the Heat Equation 560
9.5.4 Fourier Series Solution of the Wave Equation 566
9.5.5 Fourier Series Solution of the Laplace Equation 576
 Contents X11]
se

9.6 Fourier Transforms 582

9.6.1 Fourier Transform Solution of Some Partial Differential Equations 595
9.7 Answers and Hints 60/

10. Functions of a Complex Variable: Analytic Functions 611

10.1 Introduction 6/1]
10.2 Complex Number System 6//
10.2.1 Algebra of Complex Numbers 6/2
10.2.2 Polar Form of aComplex Number 6/3
10.2.3 Powers and Roots of a Complex Number 6/5
10.3 Sets of Points in the Complex Plane 622
10.4 Functions of a Complex Variable 625
10.5 Elementary Functions 632
10.5.1 Exponential function 632
10.5.2 Trigonometric and Hyperbolic Functions 634
10.5.3 Logarithm Function 639
10.5.4 General Powers of Functions 64/
10.5.5 Inverse Trigonometric and Hyperbolic Functions 642
10.6 Limits and Continuity 646
10.6.1 Limit of a Function 646
10.6.2 Continuity of a Function 652
10.6.3 Uniform Continuity 655
10.7 Differentiability and Analyticity 658
10.7.1 Cauchy-Riemann Equations 662
10.8 Harmonic Functions 675
10.9 Answers and Hints 6380

11. Integration of Complex functions 692

11.1 Introduction 692
11.2 Definite Integrals 692
11.2.1 Curves in the Complex Plane 694
11.2.2 Contour Integrals (Line Integrals in the Complex Plane) 697
11.3 Cauchy Integral Theorem /7//
11.3.1 Extension of Cauchy Integral Theorem for Multiply
Connected Domains 7/8
11.3.2 Use of Indefinite Integrals in the Evaluation of Line Integrals 723
11.4 Cauchy Integral Formula 727
11.5 Cauchy Integral Formula for Derivatives 734
11.6 Answers and Hints 74]
12. Power Series, Taylor and Laurent Series 747
12.1 Introduction 747
12.2 Infinite Sequences 747
12.2.1 Real Sequences 747
12.2.2 Complex Sequences 752
12.2.3 Sequences of Functions 755
12.2.4 Uniform Convergence 75/7

ka eR
TU PAREN OTP rea Se rea ure mms meena
pe me ep a ee ei ee ae amy el ee eee pe
XIV. Contents

12.3 Infinite Series 759

12.3.1 Tests for Convergence 76]
12.3.2 Uniform Convergence for Series of Functions 766
12.4 Power Series 768
12.5 Taylor Series 774
12.6 Laurent Series 783
12.7 Answers and Hints 793
13. Zeros, Singularities and Residues 802
13.1 Introduction 802
13.2 Zeros and Singularities of Complex Functions 802
13.3 Residues 8]/
13.4 Evaluation of Contour Integrals Using Residues 8/9
13.5 Evaluation of Real Integrals Using Residues 824
13.5.1 Real Definite Integrals Involving Trigonometric Functions 824
th

13.5.2 Improper Integrals of the Form fixjdx 828

13.5.3. Improper Integrals of the Form
=

{ cos(ax) f(x)dx or | sn(ax) fi(x)dx 834

[3.5.4 Improper Integrals with Singular Points on the Real Axis 838
13.6 Answers and Hints 849
14. Bilinear Transformations and Conformal Mapping 858
14.1 Introduction 858
14.2 Linear and Inverse Transformations 8&8
14.2.1 Linear Transformation 858
14.2.2 Inverse Transformation 864
14.3 Bilinear Transformations §87/
14.4 Conformal Mapping 885
14.5 Answers and Hints 89]
15. Vector Differential and Integral Calculus 896
15.1 Introduction 896
15.2 Parametric Representations, Continuity and Differentiability of
Vector Functions 897
13.2.1 Motion of a Body or a Particle on a Curve 903
15.3 Gradient of a Scalar Field and Directional Derivative 905
15.4 Divergence and Curl of a Vector Field 9/3
15.5 Line Integrals and Green’s Theorem 920
15.5.1 Line Integrals Independent of the Path 925
15.5.2 Green's Theorem 929
15.6 Surface Area and Surface Integrals 937
15.6.1 Surface Area 937
15.6.2 Surface Integrals 942
 Contents xv

15.7 Divergence Theorem of Gauss and Stokes’s Theorem 951]

15.7.1 Divergence Theorem of Gauss 95/
15.7.2 Stokes’s Theorem 958
15.8 Answers and Hints 966
SOME REFERENCE TEXTBOOKS 975
INDEX 977
 Chapter 1
Functions of a Real Variable
eee

1.1 Introduction
A teal valued function y = f(x) of a real variable x is a mappin
g whose domain S is a set of real
numbers (in most of the applications S is an open interval (a, b) or
a closed interval [a, b]) and whose
codomain is R, the set of real numbers. The range of the function
is the set {y= f(x): x€ R} , which
is a subset of R.
A real valued function f is said to be bounded above if |f (x) | < M and bounded below
if
Ff @) | 2m for all xe Sy,.S CoS.
The real positive finite numbers M and m are respectively called the upper bound
and the lower
bound of the function.

1.2 Limits, Continuity and Differentiability

Let f be a real valued function defined over S C IR. We define the distance function as

A(x, X2) =| x.-x, |,x1,%,€ R. ges

Let a be any real number. Then, the open interval Ns(a) = (a — 6, a + 5), 5 > 0 is called a
6-neighborhood of the point a. The interval 0 < |x-a|< dis called a deleted neighborhood of a.

1.2.1 Limit of a Function

The function fis said to tend to the limit / as x — a, if for a given positive real number € > 0, we can
find a real number 6 > 0, such that

|f(x)-l|<6€, whenever 0<|x-al<6. (1.2)

Symbolically, we write lim f(x) =f, (1.3)

Let f and g be two functions defined over S and let a be any point, not necessarily in S. Let

lim f(x) =; and lim g(x) = 1,

A t X= il
1.2 Engineering Mathematics

following properties:
exist. Then, we have the
= ¢|,,¢ a real constant.
(i) lim [cf(x] =e jim f(x)
g(*) =1, th.
(ii) lim [f(x)+ g(x)] = fim f(r) + lim

_ ae vax h | Bes (x)= bl

veow, i | =| lim f(x) | lim g(x) | =4h.
(iii) lim (f(x) g(x) L474 rod |
k—d

| = |,/l,, provided 1, # 0.
(iv) lim [f(x)/g(x)J = im, f(49|/|fig()

(v) lim(£1 = (1).

Right hand limit Let x>aandx—a from the right hand side. If
or lim f(x)=/; (| A)
lf (x-l|<€& a<x <a+6,
x—a*

then, /, is called the right hand limit.

Left hand limit Let x < a and x — a from the left hand side. If
<a, oF lim f(x)=h (1.5)
\f(x) -h| <6 a-d<x
ra

then, [, is called the left hand limit.

If /; = i, then lim f(x) exists. If the limit exists, then it is unique.

Example 1.1 Show that lim sin (1/x) does not exist.
=

Solution For different values of x in the interval 0 < |x | < 6, the function sin (1/x) takes values
between —| and |. Since lim sin (1/x) is not unique, limit does not exist.
x=

Example 1.2 Show that lim, Lx? + 1] does not exist, where |-| is the greatest integer function.

Solution Let h>0. We have

lim f(4 +h) =L(4+h)? + 1J=L17+ h(h+ 8)J=17 if Ath+8)<1

or (h+4)°<17, or he V17-4
and lim f(4 -h) =L(4-h)? + 1J=L17 + Ath-8)|=16 if A(h—8)>-1
Or
(h-4)?>15 or h>4- +15.
Theref ore, dim, f(x) = 17 and lim f(x) = 16.
x— 47
The limit does not exist.
 Functions of a Real Variable 1.3

1.2.2 Continuity of a Function

aaee a real valued function of the real variable x. Let xo be a point in the domain of f and let f be
in some neighborhood of the point x9. The function f is said to be continuous at x = x9, If
(i) lim f(x) =I exists and (ii) im f(x) =f (%). (1.6)
Alternately, ’ fis said to be contitinuous at a pointXp € J, if
if given
gi any real positive } ;
exists a real 6 > 0, such that ° . , , a nt

ss | f (x) —f (xo) |< €& whenever |x —-2x9|< 6. (1:7)

Note that 6 depends on both € and the point xo. If a function f is continuous at every point in an
interval J, then f is said to be continuous on /. |
A point at which f is not continuous is called a point of discontinuity.
If ae f(x) = I exists, but im, f(x) #f (xo), then the point xp is called a point of removable

discontinuity. In this case, we can redefine f(x), such that f(xo) = J, so that the new function is
continuous at x = x9. For example, the function
(sinx)/x, x #0
fo= {4 " saedt

lim, f(x) = 1 and a new function can be defined as

has a removable discontinuity at x = 0, since x=

pops| (sinx)/x,
ree x #0

so that it is continuous at x = 0.
Continuous functions have the following properties:
x = 9. Then,
1. Let the functions f and g be continuous at a point
any constant.
(i) cf, ft g and f'g are continuous at x = Xo, where c is
# 0.
(ii) f/g is continuous at X = Xo, if g (Xo)
1s
us at f (xo), then the composite function o(f (x))
2. If fis continuous atx = Xo and g is continuo
continuous at x = Xp.
point in (a, 5),
fun cti on f is con tin uou s in a closed interval [a, 5], if it is continuous at every
3. A
=f(6).
lim f(x)=f(a) and lim f(x)xd”
x—@a .

at an interior point c of a closed interval [a, b| and f(c) # 0, then there exists
4, If fis continuous
(x) has the same sign as f (c).
a neighborhood of c, throughout which f
attains its bou nds at least
cont inuo us in a clos ed inte rval [a, b], then it 1s bounded there and
5. Iffis
once in [a, 5]. n there exists
inte rval [a, 5] and f(a) , f(b) are of opposite signs, the
6. If fis continuous inaclose d
b] such that f (c) = 0.
at least one point c € [a, ry Value between
in a clos ed inte rval (a, b] and f (a) #f (6), then it assum es eve
7. If fis continuo us
wn as intermediate value theorem).
f (a) and f (b). (This result is kno /, if the
on f (x) is sai d to be piecewise continuous in an interval
Piecewise c ontinuity A func ti
 s
1.4 Engineering Mathematic

interval can be subdivided into a finite number of subintervals such that f (x) is continuous in each
end points of each subinterval are finite.
of the subintervals and the limits of f(x) as x approaches the
Thus. a piecewise continuous function has finite jumps at one or more points in J.
For example, the function
Ll, O<x<I]
2, Lsx<2
Se 3, 2Lx 65
4. 3S 2 5.4

is continuous in each of the subintervals and has finite jumps at the points x = 1, 2, 3. The magnitude
of these jumps is 1.
Uniform continuity In the definition of continuity given in Eq. (1.7), the value of 6 depends both
on € and the point x9. However, if a value of 5 can be obtained which depends only on € and not on
the choice of the point xp in J, then the function fis said to be uniformly continuous. Thus, a function
fis uniformly continuous on an interval J, if for a given real positive number € > 0, there exists a real
6 > 0 such that
| f(%2) —f (x) |< & whenever |x.-2x,|<6 (1.8)
for arbitrary points x,, x in I.
Obviously, a function which is uniformly continuous on an interval is also continuous on that
interval.

_ Example 1.3. What value should be assigned to

f(x) = 1 54x21
1 ly
at x = 1, so that it is continuous at x = 1.

Solution Let x = 1 + 4. Then, h > 0 as x > 1. We have

CC ~
a 1B a, ) re Pa
te 1-[1+5h+ 00")
Therefore, lim f(x) = 3.
x—]
Hence, if we assign f (1) = 3, the function will be conti
nuous at x = l.
Example 1.4 Prove that the function f defined by
fixi= r when
x x is
iS irrational]
irrati
—1, when x is rational
is discontinuous at every point.
Solution Let x = a be any rational number so that
an infinite number of rational and irra
f (a) = — |. Now, in any given interval there lie
tional numbers. Therefore, for each positive integer n, we can
choose an irrational number a, such
that |a, —a |< Ln. Thus, the sequence {a,,} converges to a. Now,
f (a,) = 1 for all n and f(a) =- 1. Hence,
at all rational points. jim f (a, ) # f(a). Therefore, the function is discontinuous
 Functions of a Real Variable |.5

Now, let
slate’ x = b be anv |; a |
aaeeahs amepanes number and f(b) = 1. For each positive integer n, we can choose
1 for all n and f (b) a on Hence,b lim< I/n. Thus, the sequence {b,} converges to b. Now,
f(b,) =—
:irrational points. = 1. nme |
at all & aif (On) # f(b). Therefore, the function is discontinuous

Hence, the given functionts discontinuous at all points.

Exa
xample 1.5 Show that the function f (x) = x°: is uniformly continuous on
[-1, 1].
Solution We have

If 2) — f (x) |= | x5 ~*7 |= |35 =x, || xp +x, |<2| x --x, | ae,

whenever |x. — x, |< €/2 = 6,

Thus, for as €/2, |f(x2) — f (x1) |< €, whenever |x2 — x; |< 6 for arbitrary x,, x. € [-1, 1]. Hence
the function x* is uniformly continuous on [-1, I].
| |
Example 1.6 Show that the function f (x) = sin (1/x) is continuous and bounded on (0, 2/7), but it
is not uniformly continuous there. | |
Solution since both the functions 1/x and sin (1/x) are continuous for all x # 0, the given function
1s continuous for all x # 0. Also, since |sin(1/x) |< 1, it is bounded there. Thus, on (0, 2/7),
f (x) is
continuous and bounded. Let € > O be given. Choose € = 1/2. Consider the points x, - In)
and x» = 2/[2n + 1)z]. Both the points lie in the given interval, that is x, x. € (0, 2/7). We have

. (2n4+1 |
sin ASDF _ sina |=1>€= 3
lf (x2) - f(a) | =

. . oe er or
no
can be made arbitrarily small by choosing n sufficiently large. Therefore, | x.— x, | < 0. However,
|< €. Thus, the function
matter how small &5> 0 may be, |x2 — x, |< 0 cannot ensure | f(x2) — f (x)
sin (1/x) is not uniformly continuous on (0, 2/7).

1.2.3 Derivative of a Function

/ and let x9 be a point in J. Then, if
Let a real valued function f (x) be defined on an interval
an Tes f (xq + Ax) —f(xX0) Ag)
_ f(x) ~ fo)
xX — XQ Ax—-0 Ax
x 3X0

rentiable at xj and / is called the derivative of f(x)

exists and is equal to /, then f(x) is said to be diffe
in the interval (a, b), then f(x) is said to be
at x = Xo. If f(%) is differentiable at every point
we consider one-
renti able in (a, 5). If the inter val [a, b] is closed, then at the end points a and b,
diffe t P gives the slope of the tangent I
ine
of f (x) at a given poin
sided: limits. Geometrically, the derivative
nt P.
to the curve y =/ (x) at the poi
satisfied by the differentiable functions.
The following properties are
be differentiable at a point x9. Then,
1. Let the functions f and 8
¢ any constant.
(i) (cf’) (%o) = cf’ (xo).
£ 8°).
(ii) (f+ g) (Xo) = f’(Xp)
1.6 Engineering Mathematics

(iii) (fg) (xo) =F Oo) 8 (Xo) + FX) 8'(%0).

(iv) (zf\] (xp) = ne. a 8'(%0)

Xo) g(x,) #0.
(Xo)
2. If f is differentiable at x9 and g is differentiable at f (x9), then the composite function
h = g(f(x)), is differentiable at x9 and h’(xo) = g°(f(%)) f’ (x9).
3. If the function y = f(x) is represented in the parametric form x = Q(t), y= W(t), J)<t<Tand
o’(t), w(t) exist, then

dyldt _ y(t)
f(x)= dxidt ~Oty’ , Q(t) #0. (1.10)

Higher order derivatives The derivative of f (x) at any point x, if it exists, is again a function of
x, say f"(x) = g(x). If g(x) is differentiable at x, then we define the second order derivative of f (x) as
d2
f(x) = g(x) = oe

Similarly, we define the nth order derivative of f as

f(x) = (SF
dx"
I-as
dx"
The existence of the th order derivative f(x), implies the existence and continuity of f, f’ a
f-» in a neighborhood of the point x.
Leibniz formula Let f and g be two differentiable functions. Then, the nth order derivative of the
product fe is given by the Leibniz formula as

Fg) = "Co f(x) g(x) + "Cy FOP) a(x) + CFR) BR) + + "C, Fe) 2D).
(1.41)
This formula can be proved by induction.
Example 1.7 Show that the function

peay= | eee x #0
, x=
is differentiable at x = 0 but f(x) is not continuous
at x= 0,
Solution We have lim f(x) = 0 =f(0). Therefore, f (x) is continuous
at x = 0. Now,

f’'(0) = lim mi)

= lim [x cos (1/x)] = 0
Hence, f (x) is differentiable at x = 0 and f’(0)
= 0. Forx# O, we have

f , (x) = 2x cos Gl |
+2] sin (2)|-+ = 2xcos (+) + sin (+).
 Functions ofa Real Variable 1.7

lim, sin (1/x) does not exist. Therefore, f(x) Is not continuous at
Now, fim f’(x) does not exist as 1—
x= 0.
Example 1.8 Find the derivative of f(y) =x|x|,- 1S 4S 1.
Solution We have

f(xy) = ae

The function is continuous for all x in [—l, 1].

For x € [—l, 0], we get

l 4 +
) = fi
f ie lim se i a ~ 2x.
: F(x — ax) — f(x) :

Ax 30

Hence, f’(O-) = 0.
For x € [0, 1], we get

al . f(x + Ax) — f(x) ©

Xt | a
(x AX )
git. AX
+0 Ax
AX

Hence, f’(0*) = 0.
f (x) is differentiable for all x in [- 1, 1] and the derivative function is given by
We find that

. mie ge
f(x) = ax, ED of fC)
.
=2 |x|

Example 1.9 Find the equations of the tangent and the normal to the ellipse x 2 = | at the
2 + or
point (xg, yo) on the ellipse.

Solution The slope of the tangent to the ellipse at the point (xy, yo) is given by m= (dy/dx); <4 yo)
and the slope of the normal to the ellipse at the point (x9, yo) is given by m, = — I/m.
Differentiating the equation of the ellipse, we obtain
2
a b*x
m=(2) ~ "i and hy Ri
FX 7}x9,¥0) a” Yo° m Bx

Hence, equation of the tangent at (xo, yo) is given by

7 a } + +

= — b°X xy + O° XxG
“
(X-—X9) OF yyga* —a°yy
b* xo
y-yYo=- Dip

4 ,
Y¥o . Xp _ *Q Yo
ee ee eh ae l.
or b a a* b-

Equation of the normal at (xo, Yo) 1s given by

2|
a” yo 2 2
y- Yo = bP (x—Xg) OF YAQd* ~ b*xg vg = aXVG — a™XO¥O
0)
 1.8 Engineering Mathematics

wi b*xoy — a°yox = (b° — a”) xyyo,

Exampl
xX
e 1.10
*
Find the fourth order derivative of e“* sin bx at the point x = Q.
Solution Let f(x) = e"*, g(x) = sin bx and F(x) =f (x)g(x). Using the Leibniz formula, we obtain

FIO (x) = a (e“sin bx) = *Cy(e™)

sin bx + *C, (e™)® (sin bx)’
x
+ *C,(e%")” (sin bx)” + 4C,(e%) (sin bx) + 4C, e&® (sin bx)
= e“*[a* sin bx + 4a°b cos bx — 6a7b ? sin bx — 4ab3 cos bx + b'sin bx}
Hence, F'(0) = 4a*b ~ 4ab? = 4ab(a? — b’).

Exercise 1.]

From the first principles, show the following.

! ) iA - 2.
L. lim - sin {4 =o lim — = |.
h ar l+eo3

3. lim (x - 3)" sen (x — 3) = 0,where sen is the sien function

1—+3

LE 2£>0
sen(x)=4 0, x=0
-I,x<9.

4, lim xLl + xJ=0, where |xJis the greatest intege

i - r function.

5S» lim e*) does not exist.

De

Obtain lim f(x), if it exists, in problems 6 to 16.

r-tg

lia
6. f(x)= a= 0. in? a
7 f(x)= ee = |.
8. f(x)
=sin (e""),
a= 9. 9. f(x) =[v+]x| J/x,a=0.
Vi¢x-1]
10. f(x) = — —. a=0. HN. f(x) = (x-2)2 eM? 2
Gin
an —!
12. f(x) = BO Ll <0, |
13. fixy= SOPH =
In +ge)
}x-9¢j
14. f(x)= Teco a=. 15. F(x) = SOARTS n—|
tan ixta,
-_
CRE
—
ale +234. 4 ps ht Boa |
16. J(aj=
a a= co,
[1° +2? + cchy*P

17. Show that the function

f defined as f(x)= | X, x rational
—X, X uTational
IS NOU Continuous at
any point x # 0. Is
jt continuous at x
= 0?
 Functions of a Real Variable 1.9

18. Find the points of discontinuity of the function f (x) = Lx], where |x| is the greatest integer function
19, Discuss the continuity of the function
f (x) =LxJ+ Vx—|x| atx = 1.
|.xf x;..4
#0
20. Show that the function f (x) = | is bounded in IR , though it is not continuous over any
. | ,#=
interval containing x = 0.
21. Determine the values of a, b, c so that the function

'sin(a+1)x +sinx
. » ee O
(x)= c , x=0
x + bx? — Sx
er i . eg

is continuous for all x.

22. Show that the function f (x) = x° is uniformly continuous on [0, 1] but not on [0, -).
23. Show that the function f (x) = ~/x is uniformly continuous on [0, 1].

24. Show that the function

f (x) = -— : is not uniformly continuous on (2, 3).
25. Show that
f (x) = cos x is uniformly continuous on [0, °).

26. Show that

f (x) = V1—x? is uniformly continuous on [0, 1].

In problems 27 to 31, find the derivative of the given function at the given point.
27. x tan”! x + sec”! (1/x), at x = 0. 28. x cosh x — sinh x, at x = 2.
29. (x + 1)? x? atx = 4. 30. e* In (cosec x), at x = 77/6.

31. yx+ yx++¥x+.., , at any point x.

In problems 32 to 37, find dy/dx, where y is defined implicitly.

32. xy + xe + ye* — x’ = 0 at any point (x, y).
x+y -—3 a xy =0, ais a constant, at x = a and y #a.
y —cos (x—y) =O at x= 7/2 andy #0. 35. x” y* = 1 at any point (x, y).
wey%a=(xt+y)* atx=ly=l. 37. (tan7'x)” + (y)'* = 1 at any point (x, y).

Find dy/dx when x = a(t — sin ft), y= a(1 — Cos 1) att = 7/2.
Find dy/dx when x = a(cos 0 + 6 sin 8), y = a(sin O— @ cos 8).
Find the derivative of sin x° with respect to x.

Find the derivative of sin™' € =) with respect to Vx.

Find the derivative of (x)*"* with respect to (sin x)*.

x € IR. Also, show
Show that the function f (x) = x’ sin (1/x), x # 0, f (0) = 0 is differentiable for all
that f’(x) is not continuous at x = 0.
Find all values of a and J, so that the function
tanx, x<
fo =|
ax+b, x20
 atics
1.10 Engineering Mathem
continuous al x = 7.
and its derivative / (x) are
t the function
45. Show tha
(x— 1) tan (x/2), x #1
pare} , eee

is not differentiable at x = 1.
lfx
Xe
x#0
46. Show that the function f(x) = 4 1 +e! |
0 q x=0
is not differentiable at x = 0.
d*y cos Xx
47. Ify= tan x + sec x, x # 7/2, then show that aot asina)?

48. Ifx = (1+ In a/r? and y = (3 + 2 In d/t, then find d?y/dx?.

49. Find the fourth order derivative of f (x) = e~ cos x.
50. Find the nth order derivative of f (x) = (ax + b)", m > Nn.
51. If y= xe”, then find d"y/dx" at x = 0.

52, Find the nth order derivative of f (x) = Vax+b.

53. Find the nth order derivative of f (x) = e™ sin (bx + c).
54. If y = cosx, — (a/2) <x < (7/2), then find d”y/dx" at x = 0.
55. If y = e45”'*, then find d"y/dx" at x = 0.

1.3 Application of Derivatives and Taylor Series

We now discuss some applications of derivatives like finding approximate values of a function, mean
value theorems, increasing and decreasing functions, maximum and minimum values of a function
and series representation of a function.

1.3.1 Differentials and Approximations

Let y mp (x) be u real valued differentiable function and xg be a point in its domain. Let x) + Ax be
a point in the neighborhood of xp. Then, Ax may be considered as an increment in x. The corresponding
increment in f (x) is given by

Afo = Af (xo) = f (x + Ax) — f (x).

From the definition of derivative, we have (see Eg. (1.9))

f = |. f(Xo :+ Ax) —f(x 0 ) A f

f'(x0) Ar-30 Ax = jim a (LIZ)
Since f’(xp) exists, we can write from Eq. (1.12) that

Afo _,
Ax ~f()+@ or Afy=f'(xp) Ax + @ Ax (1.13)
where @ is an infinitesimal
| quanti
. uantity dependent on Ax and ter ! —%
 Functions of a Real Variable 1.11

(1) Principle part f’(x9) Ax, which is called the differential of f.

(ii) Residual part @ Ax which tends to zero as Ax — 0.
In the limit, the differential is also written as

df (Xo) = dyg = f (Xo) ax. (1.14)

Hence, an approximation to f(x + Ax) can be written as

f(x + Ax) = f(xo) + f’(%)
dx. (1.15)
Differentials have application in calculating errors in functions due to small errors in the independent
variable. We define | dy | as the absolute error; dy/y as the relative error and (dy/y) X 100 as the
percentage error in computations.
Example 1.11 Find an approximate value of

y = 3(4.02)? — 2(4.02)>” + 8/+/4.02 .

Solution Let a function be defined as

y =f) = 3x" — 2x + 8) f/x.

Let x) = 4 and Ax = 0.02. Then, we need an approximation to f (xp) + Ax) = f (4.02). The approximate
value is given by (see Eq. (1.15)).
f (4.02) = f (4) + (0.02) f°(4)
We have f (4) = 48 — 2(8) + 8/2 = 36,
f'(x) = 6x — 3x!” — 44°” and f’(4) =24- 6-4/8 = 35/2.
Therefore, the required approximation is
f (4.02) = 36 + 0.02 (35/2) = 36.35.

Example 1.12 If there is a possible error of 0.02 cm in the measurement of the diameter of a sphere,
then find the possible percentage error in its volume, when the radius is 10 cm.
Solution Let the radius of the sphere be r cm. Volume of the sphere = V = 4zr7/3 and
dr =+0.01 when r= 10 cm.
Differentiating V, we obtain dV = 47 r'dr.
When r = 10, we get from Eq. (1.14), dV = 47(10)* (+ 0.01) = + 477.
Hence, the percentage error in volume is
dV x 100 '= 100 Para
(4) +127 |_= + 0.3 cubiccm.
;

1.3.2 Mean Value Theorems

We now state some important results.

Theorem 1.1 (Rolle’s theorem) Let a real valued function f be continuous on a closed interval
4 point c € (a, b)
[a, b] and differentiable in the open interval (a, 6). If f(a) = f(b), then there exists
such that f’(c) = 0.
(See Appendix 1 for proof).
1.12 Engineering Mathematics

Remark 1
(a) Differentiability of f (x) in an open interval (a, b) is a necessary condition for the applicability
of the Rolle’s theorem. | .
For example, consider the function f (x) =|x|,-1 <x 1. Now, f (x) - continuous on [-1, |]
and is differentiable at all points in the interval (—1, 1) except at the point x = 0. Now,

rot sea
L. #29

does not vanish at any point in the interval (—1, 1). This shows that the Rolle’s theorem cannot
be applied as the function f (x) is not differentiable in (—1, 1).
= 0.
(b) Rolle’s theorem gives sufficient conditions for the existence of a point c such that f(c)
For example, the function
0 [Sx 2
j= |
Da 2S a]

is not continuous on [1, 3], but f’(c) = 0 for all c in [1. 3].
(c) Geometrically, the theorem states that if a function satisfies the conditions of Rolle’s theorem
and has the same value at the end points of an interval [a, b], then there exists at least one point
c,a<c <b where the tangent to the curve y = f(x), a <x < b is parallel to the x-axis.
Theorem 1.2 (Lagrange mean value theorem) Let f be a real valued function which is continuous
on a closed interval [a, b] and differentiable in the open interval (a, b). Then there exists a point c,
a<c<_6 such that

fi(c2 (1.16)
(See Appendix | for proof).
Remark 2
(a) If f(a) = f (6), then Lagrange mean value theorem reduces to the Rolle’s theorem.
(b) Geometrically, Lagrange mean value theorem states that there exists a point (c, f (c)) on the
curve C: y=f (x), a <x <b, such that the tangent to the curve C at this point is parallel to the
chord joining the points (a, f (a)) and (b. f (b)) on the curve.
(c) Using Eq. (1.16), we can write

f(b) - f(a)
min f‘(x)< “hag ,
acrsb “et J Ok
gSxeh (1.17)
Theorem 1.3 (Cauchy mean value theorem) Let f(x) and
8(x) be two real valued functions defined
on a closed interval la, b] such that (i) they are conti
nuous on la, b], (ii) they are differentiable in
(a, b) and (iii) g’(x) # 0 for every x in (a, b). Then,
there Exists a point c € (q 5) such that
f(b)-f(a)_ f’(c)
a<c<b.
8(b)— g(a) g’(c)’ (1.18)
(See Appendix | for proof).
Remark 3
(a) For g(x) = x, Cauchy mean value
theorem reduces to Lagrange mean valu
e theorem
 Functions of a Real Variable | 13
(b) Let a curve C be
represente d parametrically asx=f(t),
value theorem states that th y=p) nerehT oo
. fO.¥ 8(#),a <t< b. Then, Cauchy mean
—_ 8 , (c)/f’(c
; ere exists a point (f(c), 2(c)), ce (a, b) on the curve such that the
) of the tangent to the curve at this Point is equal to
joining the end points of the curve. Hence, the slope of the chord
Cauchy mean value theorem has the same geometri
interpretation as the Lagrange cal
mean value theorem.
(c) Cauchy mean value theorem cann
ot be proved by applying the Lagran
separately to the numerator and ge mean value theorem
denominator on the left side of Eq.
Lagrange mean value theorem to the (1.18). If we apply the
numerator and the denominator sepa
rately, we obtain
f(b) - f(a) _ FFtex)
g(b)- g(a) g’(e) 9<1 <9 <2 <b # ep.
Example 1.13 A twice differentiable function f is such
that f(a) = f(b) = 0 and f(c) > 0 for
a<c <b. Prove that there is at least one value
¢,a<€&<hb for which f”(& <0.
Solution Consider the function f(x) defined on
[a, b]. Since f(x) exists, both fand f’ exist and are
continuous on [a, b]. Let c be any point in (a, b).
Applying the Lagrange mean value theorem to F(x)
on [a, c] and [c, 5] separately, we get

Lie) ~f(a) = f'(€)), as E < Cc, and Oi) = f’(€2), c< G2 < b.

Using f(a) = f(b) = 0, we obtain from the above equations

f(Ey= f(c)
2S and f(g) 2-2.
c-d@

Now, f"(x) is continuous and differentiable on [&,, &]. Using the Lagran ge mean value theorem again,
we obtain

Lon) =ftov) = f'"(6), 61 <6 < &2.

G2 — Si
Substituting the values of f’(&,) and f’(€3), we get
‘ FOP) fod yg hy OE) ee,
a ee ee gore (b-c)(c-a)(€,-&) ~

Example 1.14 Using the Lagrange mean value theorem, show that
| cosb -—cosa|<|b-a|.
Solution Let f(x) = cos x, a <x <b. Using the Lagrange mean value theorem to f (x), we obtain

cos b= c0S4 ~ ¢’(¢) =~ sinc, or

b-
sae
—_
d =|-sinc|
<1.
| |
Hence, the result.

Example 1.15 Let fiy= WG= x*) and f (0) = 1. Find an interval in which f (1) lies.

Solution Using Eq. (1.17), we obtain for a = 0 and b = 1

(1) — f(0) é
f’(x)
min f'(x)S LM) =f < max
2-FO)
 L.14 Engineering Mathematics

. ) ¥ ae |
or vain, |5 z]s fi
— i<jmax, [st

oT +| Sf(-1S5,l oF z4 eSJ I<

S 9:
: . ne v-acng.0<@<
Example 1.16 Let C be acurve defined parametrically as v= a cos 0.y=a sin 6,0 ‘s0 ; 7/2. Deter-
mine a point P on C, where the tangent to C is parallel to the chord joining the points (a, 0)and (0, a).

Solution We have x = f(8)=acos’@ and y=g(@)=4 sin°@. Using the Cauchy mean value
theorem, we have at some point 0 1 (0
slope of tangent to C = slope of the chord joining the points (a, 0) and (0, a)

g(@) 3asin*?@cos@ _ a — 0
4 ;
. f'(@)
a
-—3acos*@sin@ Q0-a

or —~tan@=-1, or 0=/4.

Therefore, the required point is (a/2¥2. a/2V2).

1.3.3 Indeterminate Forms

Consider the ratio f (x)/g(x) of two functions f (x) and g(x). If at any point x = a, f (a) = g(a) = 0,
then the ratio f (x)/g(x) takes the form 0/0 and it is called an indeterminate form. The problem is to
determine ‘im Lf(xVe(x)]. if it exists. Since f(a) = g(a) = 0, we can write

fl) A= Fla) _ 4 @)-f@Wx-a) £)

ry g(a) be g(x) g(a) ~ 5h [e(x) ~g(a)(a—a) ~ xt5e g(x)
provided the limit on the right hand side exists. This result is known as L’ Hospital’s rule.
L’ Hospital’s rule Suppose that the real valued functions f and g are differentiable in some open
interval containing the point x = a (except may be at the point x = a) and f(a) = 0 = g(a). Then,
_, £) zi a f(a)
roa g(x) xd g(x) g(a) (1.19)
Suppose now that f’(a) = 0 = g’(a). Then, we repeat the application of L’ Hospital’ ) ee
and obtain ospital’s rule on f’(x)/g (x)

F(x) _ FX) f(x) _ f(a)

li |
roa g(x)
g(x) * g’(x) + ne
e’(x) g(a)
provided
ad the limits exist. This applicati
ee tg pplication of the rule can be continued as long as the indeterminate

Whencanboth
nile f(a)=+ We-write
bé applied. and g(a) = + o we get another indeterm; ate form. In
this case also L Hospital’s

f(x) [I/g(x)]
roo g(x)
lj

FG]
=_ ee

which is of 0/0 from.

Remark 4

(a) L’Hospit al’’s rulelec: can be used

Pital only when the ratio is of indeterminat
e form, that is, either it is
 1.15
Functions of a Real Variable

of the form 0/0 or c0/ce.

(b) The other indeterminate forms are 0 - 00, 0°,00”, 1° and ce — oo. In each of these cases, we can
reine the ratio function to the form 0/0 or 7 ad use this rule. For the indeterminate forms
0°, co” and 1”, we take logarithm of the given function and then take the limits.
(c) When the function isO of the form 0”, - 0, 2 + 00, ©” or oo, it is not of indeterminate form
and we cannot apply L’Hospital’s rule. We note that 0™ = 0, oc - 00 = ce, oc + ce = co, oo” = cc and

(d) L’ Hospital’s rule can also be applied to find the limits as x > + =.
Example 1.17 Evaluate the following limits

In (1 + x) is a ae s eX
(1) lim eal (11) lim [x (In ¥));, (111) jim - |

Solution Using L’Hospital’s rule, we get

(i im |PEF
x30 sin x |=x30
im MEX
COSX -
” 3 aaa [ie]. ge Se”
: 0:
(ii) lim [x”-(In-x)] = limbret] em ott]

usc , e~ |_ 4: e _—
a) pe. = =jim| |- )
Example 1.18 Evaluate lim x”.
x0

Solution The given limit is of the form 0° which is an indeterminate form. Let y = x".
Then, In y = x In x. Now,
Inx |
wiiticctitie |! Gee we! wx= ty (xn x] = fig |
ling(In y} = In|timy | BE)

= x0in [-]/x?]
= tin
x0
x=
lim y = e° = 1.
Therefore, x0

Example 1.19 Evaluate lim x tan (1/x).

limSornars which 1s
Solution As x — ©, the function takes the form °° - 0. We first write it as
of the form ce/ee. Applying the L’Hospital’s rule, we obtain
—— {
lim x tan (1/x) = |ye cot (1/x) ~ oe (1/x7) cosec? (1/x)
3 9 Fede A
sin? (1/x) _ tiny HOY — tien (SZ) 21
— —— Ss 1m a Wm a? a

yoo (1/x)? yO y yoo )

1.3.4 Increasing and Decreasing Functions

domain of the function f (2)- Le
Let y = f(x) be a function defined on an interval / contained in the
 1.16 Engineering Mathematics

X), X» be any two points in /, where x,, x2 are not the end points of the interval.
]> +2 On the interval | the
; >
function f (x) is said to Ue
(i) an increasing function, if f (x,) < f (x) whenever x, < x5.
(ii) a strictly increasing function, if f (x,) <f (x.) whenever x, < x5.
(iii) a decreasing function, if f (x,) > f (x.) whenever x, < x.
(iv) a strictly decreasing function, if f (x,) > f (x2) whenever x, < x.
A function which is either increasing or decreasing in the entire interval / is called a monotonic
function.
Let a real valued function f defined on an interval /, have a derivative at every point x in /. Then,
using the Lagrange mean value theorem, we have

I (x2) — f(x) = f’(c), x, ZOR hy;

AQ AY

Therefore, we conclude that

(1) fincreases in J if f’(x) > 0 for all x in J,

(ii) f decreases in J if f’(x) < 0 for all x in J.
Thus, a differentiable function increases when its graph has positive slopes and decreases when its
graph has negative slopes. Now, if f’(x) is continuous, then f(x) can go from positive to negative
values or from negative to positive values only by going through the value 0. The values of x for
which f’(x) = 0 are called the turning points or the critical points. At a turning point, the tangent
to
the curve is parallel to the x-axis. On the left and right of a turnin g point, tangents to the curve
have
different directions.
Example 1.20 Find the intervals in which the function f (x) = sin 3x, 0 < x < 7/2 is increasing or
decreasing.
Solution We have f’(x) = 3 cos 3x. Now, f(x) = 0 when 3x = 77/2, 37/2... . for
positive x.
Hence, x = 7/6 is the only turning point in (0, 2/2). We conside
r the intervals (O, 77/6) and
(70/6, 77/2). We have in

O<x < 7/6: f’(x) = 3 cos 3x > 0, f (x) is an increasing function,

mM/6<x< 1/2: f’(x)=3 cos 3x < Q, f (x) is a decreasing function.
Example 1.21 Show that for all x > 0
2
L—* <6" «logy 2.
2
Solution Let f (x) = e* + x - 1. Now,
£@)= 1-—e" 5 0 forall x> 0.
Hence, f (x) is an increasing function for all x >
0. Therefore,
f(x) >f(0)=0, or e*4+x_] >QOor
e*>1- x.
2
Now, consider g(x) = e*-1+x-

We have g(x)=l-x-e* <0 for all x > O.

Hence, g(x) is a decreasing function for all x > 0. Therefore,
 Functions of aReal Variable |.|7

Hr) < ¥(O) =), mw et cf —x~4+ 5

Oombining the above Iwo
(' 2
results, we obtain
é ‘jf #

£
b-¥e etr A
[= e4 ZF 2>0

1.3.5 Maximum and Minimum

Values of a Function
Let a real
_ cea, a valued func
function f(x) bye CONUAMUGUS ON 4 closed interval la. db}. Sincé a continuous function a
oan the penh 7seaie and attains these bounds
at least once in the interval, we wish to deter-
; a AMS
ae where
a f (x) attains these
} Oe bounds. Let int inj (a, b)
Xy be a point . and I, = (x, —h +h)
be an infinitesimal} interval around ty. Then, the function f (x) is said to have a , ’ |
lacal maximum (or a relative maximum) at the point
xy, if f (c,) = f (x), for all x in J.
local minimum (or a relative minimum) at the point
45, if f (xy) < f (x) for all x in I.
The points of local maximum and local minimurn are
called the critical points or the stationary
points, The values of the function at these points are
called the extreme values. °
The following theorem gives the neces sary Condition for the existence
of a local maximum/minimum.
ia hs 1.4 | (First derivative test) Let f (x) be differentiable at x,
€ (a, b). Then, a necessary
condition for the function § (x) to have a local maximum or a local minimum
at Xy is that f’(x9) = 0.
Ata critical point, f'(x) changes direction. Thus, to find the local maximum/mi
nimum values of
the function in an interval /, we find the critical points in J by solving f’(x) = 0. B y studying
the sign
of f (x) as it passes through the critical point, we decide whether it is a point
of local maximum
(f(x) changes sign from positive to negative) or a point of loca! minimum (f’(x) changes
sign from
negative to positive).
Example 1.22 Examine the functions
(1)
f (x) = x -3x+3,x%€ R , Gi fj=sinx,0O<cxcen
for maximum and minimum values.
Solution We have
(i) f(x) = 3x? - 3. Now, f’(x) = 0 gives x = 1,-1.

Por x < 1, f(x) < O and forx > 1, f(x) > 0. Since f’(x) changes sign from negative to positive as
it passes through the critical point x = 1, the function has a local minimum value f(1) = 1 at x= 1.
For x <- 1, f(x) > 0 and for x> — 1, f’(x) < 0. Since f’(x) changes sign from positive to negative
as it passes through the critical point x = — 1, the function has a loca] maximum value f (-1) = 5
atx =— |.
(it) f'(x) = 2 sin x cos x = sin 2x = O at x = 7/2.
Por x < 1/2, f(x) > 0 and forx> 1/2, f(x) < 0. Since f(x) changes sign from positive to negative
f(7/2) = |
as it passes through the critical point x = 7/2, the function has a local maximum value
atx = n/2.
Theorem 1.5 (Second derivative test) Let f (x) be differentiable at x). a < x) < b and let
f' (xg) = 0. If f"Cc) exists and is continuous in a neighborhood of xo, then
 1.18 Engineering Mathematics

af
xq) ’
= lim
: f'(xo +h): ~f'(%0) :_ jn LOW = h =LOW 450
f" (x0) h->0 Fgrer,

Therefore,

(i) f (x) has a maximum value at x = x9, when f”(%9) < 9:

(ii) f (x) has a minimum value at x = xo, when f”(%o) > 9.
decide whether x = Xp is a point of local maximum
When f(x) = 0, further investigation is needed to
or minimum. In this direction, we have the following result.
Let
Theorem 1.6 Let f(x) exist for x in (a, b) and be continuous there.

=fi) = 0 LM) =O and f°"(%o) # 0

f(a)
Then,

when n is even, f (x) has a maxim um if f(x) < 0 and a minim um if f@) + 0.
(i)
(ii) when n is odd, f (x) has neither a maximum, nor a minimum.
Absolute maximum/minimum values of a function f(x) ina n interval
[a, b] are defined as follows:

Absolute maximum value = max { f (a), f (b), all local maximum values}.

Absolute minimum value = min {f (a), f (), all local minimum values}.
Example 1.23 Find the absolute maximum/minimum values of the function
f(x) = sin x (1 + cos x), 0<x < 27.

Solution We have

f (x) = sin x (1 + cos x) = sinx + : sin 2x, f’(x) = cos x + cos 2x.

Setting f’(x) = 0, we get

cosx+cos2x=0, or cosx+2cosx-l1=0, or cosx=-1, 1/2.
Therefore, the critical points are x = 2/3, 7 and 57/3.
Now, f(x) =-— sin x — 2sin 2x.

At x = 2/3, f"(m/3) = — 3V3/2 <0. Hence, f (x) has a local maximum at x = 77/3 and the local
maximum value is f (77/3) = 3-¥3/4.
Atx= 7, f”(z) = 0. We find that
f(a) =-cosx-—4cos2x and f”(2) =~—3+#0.
Since, f (2) # 0 and n = 3 is odd, the function has neither maximum nor minimum at x = 7
At x = 57/3, f"(Sm/3) = 33/2 > 0. Hence, f (x) has a local minimum at x = 52/3. The local
minimum value is f (52/3) = -3-/3/4.
We also have f (0) = (22) = 0. Therefore,

absolute maximum value of f (x) = max { f (0), f(27), local maximum value at x
= 77/3}
= max {0, 0, 33/4} = 3./3/4.
absolute minimum value of f (x) = min {f (0), f (27), local minimum
value at x = 57t/ 3}
 Functions of a Real Variable |\.\9

= min {0, 0, — 3./3/4} =- 3./3/4.

Example 1.24 Find a right angled triangle of maximum area with hypotenuse h.
Solution Let x be the base of the right angled triangle. The area of the right angled triangle is

A(x) = axvh* —x*2, O<x<h.

tro|—

Now, A*(x) =
— x , = 2x""
ht ==2x
t|— jh? =x? [p2 — x? 2-/h? — x?

Setting A’(x) = 0 we obtain the critical point as x = h/-/2.

Now, A(x) > 0 for x < h/-/2 and A’(x) < 0 for x > h/-j2.
Therefore, A(x) is maximum when x = A/-/2 and the maximum area is A(h/ J2) = h/4.

1.3.6 Taylor’s Theorem and Taylor’s Series

A very useful technique in the analysis of real valued functions is the approximation of continuous
provides
functions by polynomials. Taylor’s theorem (Taylor’s formula) is an important tool which
the mean
such an approximation by polynomials. Taylor’s theorem can be regarded as an extension of
and
value theorems to higher order derivatives. Mean value theorems relate the value of the function
and its higher
its first order derivative, whereas the Taylor’s theorem relates the value of the function
order derivatives.
defined and have continuous derivatives
Theorem 1.7 (Taylor’s theorem with remainder) Let f(x) be
s expansion of the function
upto (m + 1)th order in some interval /, containing a point a. Then, Taylor’
f(x) about the point x = a is given by
|

p(x) = f(a) + FO pray + SSP pat. AAA f(a) + Ry (x) (1.20)

ees i _ a

— (eal const) (1.

71 )
R, (x) = (n + 1)! f (c), accex
where

nsion.
‘s the remainder or the error term of the expa
ee n, which satisfies the conditions
Proof We first find a polynomial P,,(x), of degr
P, (a) = f(a), Pha) = f(a), k= 1,2, 050
polynomial as
approximation to f (x). Write the required
In a certain sense, P,,(x) is a polynomial
P, (x) = Co + C(% — a) + C4(X - a’ +... +¢,(x - a)".

obtain
Substituting P,(x) in the given conditions, we
P.(a) = f(a) = Co, Pia) = f(a) = C1, Palla) = f(a) = 220-0
P<(a) = f(a) = (a!) Cn.
=:
= as gl) (a) K=O) Ly
C,ee Qisasg ti|
Hence, we have
 1.20 Engineering Mathematics

on -a)* x-a)"
Uherefore, 429 pr,a) +Se
f(x) = P(x) =f(a)+ SAE + e "(ays ~ me at te,

The error of approximation is given by R(x) = f (x) — P(x). Therefore,

I(x)= P,(x) + R,(x)= 2 (ea) h(a) + R,(x).

Now, we derive a form of R,(x). Write R,(x) as

R,(x)
= 4 (x-a yt
h(x)

where A(x) is to be determined.

Consider the auxiliary function

(x-1)" Wx)
F(t) = f(x) - fo + (x- Of) +... == ("+ (n+ DE ac i<=

We have f as a variable and x is fixed. The function F(t) has the following properties
(1) F(t) is continuous in a < ¢ < x and differentiable in a < t < x,
(ii) F(x) =0

(x =)"
+ Oe ms)
(x —a)" nye
(iti) F(a) = f(x) - fa +(x-—a)f(a)+...

=f (x) -f (x) = 0
Hence, F(t) satisfies the hypothesis of the Rolle’s theorem on [a, x]. Therefore, there exists a point
c,a<e<x such that F’(c) = 0. Now,

F(t) =0- mn~ f(t) + (x- NF" - 2 F(t) +.

aes ne (n+ 1)(x-t

+ Sa peorny- (n+ 1)!
(+Di-0" 0 [=
es
(x [A(x) —fM"FD (ty)

and F(c)=0= F=f) —fey

We obtain A(x) =f tle), Therefore,
a yttl
R, (x) = 158) _ pte} acc<x.
(n+ 1)!
The error term can also be written as

(
R,,(x)= caf (a + O(x-a)),0<0<1 (1.22)
(
which is called the Lagrange form of the remai
nder.
Ita=0, we get
 Functions of a Real Variable | 2]

F(x) = Ff Q) Tel (QO) + x

pik f* 7 FO)
wv # +=
x" FO) )
+ eile
zal
). <c<x (1.23)
which is called the Maclaurin’s theorem
with remainder.
If we neglect the error term in Eq.
(1.20), we obtain

(Wa Pie F=0 Yee

n . Mm

mi! f'"(a) (1.24)

which is called the nth degree Taylor's polynomial
approximation to f (x).
eaeSincece c¢ or 9 in
in the
the remai
remain de term 7 (see Eqs. (1.21), (1.22)) iis not known, we
nder cannot evaluate
n(x) exactly for a given x in the interval /. However, a bound on the error
can be obtained as
| | 2 (x = a)" . Se n+] _

Rn) |= | FOP Co) |S max SE |max |f(x) | eo)

For a given error bound ¢, we can use Eq. (1.25) to determine
(i) n for a given x and a,
or (11) x = x* for a given n and a such that |R,(x*) |< €.

Remark 5
(a) Writing x = a + h in Eq. (1.20), we obtain

f(a+hy~flay+ Apays Spray... + HKM(a), (1.26)

The error of approximation simplifies as
_ prt ki |
R, (x) = fat pil (c) a<c<ath. (1.27)

(b) Another form of the remainder R,(x) is the Cauchy form of remainder which is given by
n+l : | |
Ry (x) = = (1 - 6)" f"" (a+ Oh), O< O< 1. (1.28)
(see Appendix | for proof).
(c) Another form of the remainder R,(x) is the integral form of remainder, which is given by
rath

R,(x) = “ | (a+h—s)" f""'(s) ds (1.29)

(see Appendix 1 for proof).

Example 1.25 The function f (x) = sin x is approximated by Taylor’s polynomial of degree three
about the point x = 0. Find c such that the error satisfies | Ry(x) | < 0.001 for all x in the
interval [0, c].

Solution We have
4
i 3

f 0) =f (0) + x f (0) + %rf") + art 0).

For f (x) = sin x, we obtain
f’(x) = cos x, f(x) =-sinx, f(x) =—- cos x and f (x) = sin X.

Hence, £0) = 1, £"(0) =sinS.

=0,f’"(0) = - 1 and f(E)
1.22 Engineering Mathematics

vescwuil@incan. Hy ee x33
approximation is f(x) = sin x = x - -
The required
The maximum error in the interval [0, c] is given by

|R3(x)|=
pag, inal
sing] <x. [5 02
Now, c is to be determined such that
4
C_<09001 or c”* 4 < 0.024.
74 = 0.00

We obtain c ~ 0.3936. Hence, for all x in the interval [0, 0.3936], this error criterion is satisfied.

Taylor’s Series
In the Taylor’s formula with remainder (Eqs. (1.20), (1.21)), if the remainder R,(x) — 0 as n - 00,
then we obtain

fodv=Ftal + (2—
me
FR2
=a)"
9)pr(qy 4 (x7 a)" " pay... +FRA
_ == a)"
fM(a) +...
(n)
(1:30)
which is called the Taylor’s series. When a = . we obtain the Maclaurin's series
2 n |
FR) =f0)+ F/O + L'O +... + HMO +. (1.31)
Since it is assumed that f(x) has continuous derivatives upto (n + 1)th order, f ("+ 1)(~) is bounded in
the interval (a, x). Hence, to establish that lim |R,(x)|=0, it is sufficient to show that
hn—} oo
|x -a lade .
im “Grol = 0 for any fixed numbers x and a. Now, for any fixed numbers x and a,| we can
ri? co é

always find a finite positive integer N such that | x —a |< N. Denote g = |x -—a|/N. Then

(x-a)™" | _ X—-al||x-a i X—a@ | x-a

(n 4 | f l 2 N-] N n+]

(x-a)N7
ee fe
el ‘0... = | (x
2c -—a)"- !
Sd
(N-TP
a
[7°99 |"Naar |4
N | (x-a)*- |. , 2.
ow, “iN. DE 1s a finite quantity and is independent of
n. Also q < |. Hence

(x —a)"*!
(n+ I)! |
= 0 for any fixed x and a, and lim |R, (x)|=0

Example 1.26 Obtain the Taylor’s pol

x = 1/4, Show that the error term wigS tO Zero expansion of the functionf(x) = sin x about the point
aS n > co for any real x. Hence, write the Taylor’s
series expansion of f (x).
Solution For f (x) = sin x, we have

f(x) = (- 1)" sin x and f@ +

(x) = (-1)" cos x
for any integer py. Therefore,
 Functions of a Real Variable 1,23

f (m14) = (-1)VV2 and f+ (/4) = (-1)7 V2.

Hence, the Taylor’s expansion of f (x) = sin x about x = 7/4 is given by

tix) =s(4)+ [+- 2)7{2)+...+ 4 (x-4) f( £)+R, (x).

Now,

1
= 2n+l I - 2n+]

|Ron (x) | =
axa 2-8)
4 £6)
soe (*-8)
since fO""') = |(-1)” cos c |< 1. Hence, R2,(x) > 0 as n > ~.

Similarly, we find that R,,,,(x%) >~ 0 asn > ~.

Therefore, the required Taylor’s series expansion is given by
7) 3
we ty tf. @)1 [,.#) 1 (x #) 7
It
etal ea G f) at ;) Bni2\* 4) ~
9 3
1 1 tn
=f fte(s-8)-a(2- 3] -4(x-4)

1.3.7 Exponential, Logarithmic and Binomial Series

Exponential series
n about the point x = 0 for the function
Consider the Taylor’s polynomial approximation of degree <
is given by
f (x) = e*. The Taylor's polynomial approximation
:
HO.
n

f(x) =fO+2S/'O+ Fl’ Or... +

2

For f(x) =e%, we obtain

f*"*” (x) =e".
e* f'(0) = Lor= 0; gen _,nand
fl@=

Hence,

inder, we get
Using the Lagrange form of the rema
n+l n+l ;

R= Gael” = meme
_ x tl Ax |
= Ty ty © ; 0<@< 1.
R(X)
OF aS

x"! +1
@x xX : Jen= 0

Therefore, jim |Rn (x) |= lim (n+ 1 ~ im i + 1)!

 1.24 Engineering Mathematics

for all x, since e9x is bounded for a given x.

Hence. we obtain the exponential series

+t 32)
Example 1.27 For the Taylor’s polynomial approximation of degree <1 about the point x = 0 fo,
the function f (x) = e*, determine the value of n such that the error satisfies |Ry(x) | < 9.005, when
=| Sx dL.

Solution We have the Taylor’s polynomial approximation of e* as

? n

f(x) =e
— ow
=k PESx _&
Py
=.

The maximum error in the interval [-1, 1] is given by

|Rn (x) |= for (c)

x n+l Ft ¥ a

(71+ 1)! $ E ay | eas ens Ge

Now, 7 is to be determined such that

— |
fa + 1! < 0.005 or |)! > 200 e€«€
(n+ 1)

We find that n > 5. Hence, we will require at least 6 terms in the Taylor’s polynomial approximation
to achieve the given accuracy.
Example 1.28 Obtain the fourth degree Taylor’s polynomial approximation to f (x) = e** about
x = 0. Find the maximum error when0O <x <0.5.
Solution We have

f(x) = £(0) + xf'O) + 5,£0) + 5 FO) + FO)

2 3

For f (x) = e*, we obtain f(x) = 2’e*, f (0) = 2", r=0, 1,2,... and fc) = 32 &*.
‘ 4x* 8x? 16x4
Therefore, f (x) = e~ Ht hbo et el + ee De + 3x3 + Ext.
The error term is given by
5.
R4(x) = 7" @ = 3 Se, O0<c<x.

32
and | Ra(2) | 120 |max. P [om . = 120°

Logarithmic series

Consider the Taylor’s polynomial approximation of degree <n about the point x = 0 for the function
f (x) = In (1 + x). The Taylor’s polynomial approximation is given
by
 Functions of a Real Variable 1,25

f(x) =O) + xf") + 2 f"(O) +... + 2 60),

For f (x) = In (1 + x), we obtain

f(x) =In(1 +x), f(0)=0; f(x)= ! f’0)= 1;

1+x’

F(x)r -1)""(r -
=§
II 1)!
OO) =A Dh 28a
(n+1) _ (-1)"n!
pax) Aiea

_1)2 (9! _y\r-1 ~ :

=
Hence, f(x) =In(Q ities! ey? aS eS oe ..ope 1) a Ly!

2 3 ‘
=X ; + 3 ...+(-l) a (1.35)

We note that f (x) and all its derivatives exist and are continuous for -—1 <x < 1.
Using the Lagrange form of the remainder, we get

Ru (X=
a ar) atte (—1)" n°
yl (6x) = (n+1)! Ferd

n n+l

i “ ;,; Pee= 1.
~ (n+1)| 1+ 0x

We consider the following two cases:

Case1l 0 < @< 1, we have
Let 0<-x < 1. Since

O<O@x<x<1 and —*—e<1.

1+ 60x

Therefore, we obtain

n+1
. =" | x :
Pe = Bee oltees| =o
Case 2 Let —-1 <x <0. Since0 < 6<1,|x/(1 + @x) |may or may not be less than 1. Hence, we
cannot use the Lagrange form of the remainder to find lim |R,, (x) |.

Now, consider the Cauchy’s form of the remainder. We have

n+]
R, (x) = 2] (1 —— 9)"6)" fr"F“" (8x)
(x)= (ls Ox)
(-1)” (1 = 0)" yt
n!
 1.26 Engineering Mathematics

(-1)” (1-@ | 0< 6< 1.

~ (1 + Ox)\ 14Ox
Sj .
inc 1-8
1+Ox |>>"
]
3
d _1l ie,
| ia Gx 1 -|x|

|
we have
a | tlt |1-@|"|29
inRy (2)|< im
|AEG 1 + Ox |

Hence, we obtain

2 3 ni xe _ <1. :
In(l+x)=x-4-+4--...+(C0) 1S 4 yang SL EwS I (1.34)

Note that for |x|> 1, lim | R, (x) | = ©.

Writing 1+x=y or x=y-—1 in Eg. (1.34), we get

) ae | n—| ” |

Iny =(y- 1) - 509-1)? t... 4S —Ky-)) Hiney OSES (1.35)

The series given in Eqs. (1.34) and (1.35) are called the logarithmic series.

Binomial series
Consider the expansion of the function (x + y)". We can write
(x + yy" = x™ [1 + Q/x)J™ = x1 + z), where z = y/x.
Therefore, it is sufficient to obtain the expansion of the function f(z) =(1 + z)™ or f(x)
= (1 + x)”.
Consider the Taylor’s polynomial approximation for the function f (x) = (1 + x)”
about the point
x = 0. We consider the following two cases.
Case 1 When m is a positive integer, f(x) = (1 + x) posse
sses continuous derivatives of all orders
and f'"(x)=0, r>m+ 1 for all x. We have

f(x)=(1+x)", fO)=1; f’(x)=m(l4+x)™. f’(0) =m;

f(x) = m(m - 1) (1 + x)": f°(0) = m(m — 1): ...

f°" (x) = m(m-1)...2.1, f™(0=

)mi and f(x) <0 ps.
Therefore, we obtain

POV= (1+ x)" = F0) + xf0) + FF"

7 * , 2 |
) 4... 4 rh (0) m
 Variable 1.27
Functions of 4 Real

m(m— 1) Hx”
—-1+mxt ca yr eee

("| "x3 ("ee oo "|" (1.36)

We also have
fr(c) = 0, n2m.
xn!

Rn(*) = Ta 1)! of all

= (1 + x)™ possesses continuous derivatives
positive integer, f(x)
Case 2 When m is nota
orders provided x # -I.
Let -1 <x < 1. We have
FO) (x) = m(m — 1). -- (m - n)(1 +x
ye

ainder, we get
Using the Cauchy’s form of the rem
n+l ? (Ox)
R,, (x) — =—-( = @)" fo"

= 2" (1 - 6)" m(m - 1)...(m—n)(1 + Ox

not

DD—ee.(m—
PAE — 1)..
_“ m(m 1-6)"
SEn) |(a :
_ntl,
(1+ oxy™m-1tx™! | O< O<1. (1.37)

<4 and
Now, for! x|< 1, 0< 11-0
+ Ox

limixi™ =0, n—limoo m(m—1)... (m—") | _,

fm (2-8) <0, neo n!
nol 1 + OX

ent of n and bounded, we obtain from Eq. (1.37)

where ais a finite quantity. Since (1 + x)M-I is independ
lim 1 R, (x) | = 0.
and | x | < 1, we obtain
Therefore, when m is not a positive integer
I _ 4
4 MOD yn (1.38)
flea 14 met MDa,

Alternative proof
x".
Consider the series » a, where @, = mn)”
Using the ratio test, we get |

= |
_ | i
(m-n-—l)
A S s ix.
nw | Ane dD
1.28 Engineering Mathematics
im,

Hence, the senes (1.38) converges for =] < ¢ < |

Further, it can be shown that the binomial series (1.38) converges atv
= | when m > — 1,
For example. consider the series
- , 1
(lL + x) _ l—-xt2x -xX +.

Fora = 1, the senes on the right hand side has two limit points 0 and | and hence the series
ig poy
eOonvergent.
We have the following binomial series

(1 +x)" = 1 — 2x 4 3? -—42° + =] 6.x]

Th | P39 1°3°5 23 — fe
ee 1.
(Ltsy" =1 S447 "a4. 6" +. F

(1 +27)
Bi =l+gx-y7es
a foe Ne ae es FE +39
a Be dw ailing
oS I.

Exercise 1.2

Find the approximate values of the following quantities using differentials.

1. (1005)'*. 2. (999)!9.
3. (1.001) + 2(1.001)*5+ 5. 4. sin 60° 10’.
5. tan 45° 5’ 30”.
6. State why Rolle’s theorem cannot be applied to the following functions.
(i) f (x) = tan x in the interval [0, z], (ii) f(x) = |x] in the interval [-1/2, 3/2],
_ x ;« OSes ]
wd sve eS
7. It is given that the Rolle’s theorem holds for the function fw
=x + bx? + cx, 1 <x <2 at the
point x = 4/3. Find the values of b and c.
8. The functions f (x) and g(x) are continuous on [a, b]
and differentiable in (a, 6) such that f (a) =4,
f (6) = 10, g(a) = 1 and g(b) = 3. Then, show that P'(c) = 3g'(c),a<ce<b.
9. Prove that between any two real roots of e* sin x =
l, there exists atleast one root of e*cosx+1=
0).
10. Let f’(x), g’(x) be continuous and differentiable functions
on [a, 6]. Then, show that for a<c<b
f(b) -f(a)-(b-a) f(a) fc) i
g(b)- (a) =(b—a)g(a) g%eyr8 () #0.
Il. Let f(x) be continuous on [a - 1, a +
1] and differentiable in (@—1,
0 < @< 1 such tha f (a
t — I)-2f(a)+f@+l)=f(ad 0) a+ 1). Show that there exists a 0,
-f"(a- 6).
12. Using the Lagrange mean value theorem,
show that
(i) e* >1+x, x>0;
(ii) In (1 tx)<x, x>0-
(iii) x<sin'x<x/VI-x? ,Ocxe].
13. Suppose that f(x) is differentiable for al]
values ofx such that f(a) =a, f(—a
all x, Show thaft(0) = 0. ) =—¢ and |f’(x) |< 1 for
14. Let F(x) and G(x) be two function
s defined on [a, b]
with G(x) ¥ 0 for any x in [a, Satisfying the hypothesis of the mean valu
b]. Show that ther € €Xxists e theorem
a point c in (a, b) such
that
 Fib)-Flay Fcjy! Gi lc)
(i(b) . (sla) G te) C10 GOB,

Evaluate the limits in problems 15 to 2%

15. lim -— 16. jim ce’ ~ 20 e+e”

t+] x" = | afi KX afi Zz

In (sina) :
17, lim: —
Boni (fp Dy)?
18. lim 24-42
so aggun’ x

19. lim (1 = 4) tan (mx/2) 2. lim 7

Sal Se Roe
to! s-02 1 x= 2 Infx—})

fie lim hl ZZ lm (una)?

g— | roel

23.
93 lim S——=—!,
1Xlim — £0,say =0= a
1-9) (x) f | 2A. im F
f
Qa £

25. re
lim [1 +f(x)|'!, iv
lim fix) =0 2%. iam
lim |1++x |
ee ff |

27. lim prriya

[-—7™ y X—
x+4In.
CO" x
28. lum
icv A 2 J

In problems 29 to 36, find the intervals in whichf (x) is increasing or decreasing.

29. In (2 + x) —2x/(2
+ x), xe IR 3%. xsixL
x€ R.

31. tan! x+x, xe RR. 32. snx+|sinx|, O< x5 2%

33. In(sinx), O< «<7. 34. (Inxs, x>O

35. sinx(1 +cos x), 0< 4 < m2. 36. «*, «>0.
37. Leta>b>O and n be a positive integer satisfying n 2 2. Prove that a’ - b’" <(a-b)"™
In problems 38 to 43, find the extreme values of the given function f (x).
38. (x - 1) (x + 1)’. 39. sin x + cos x
40. x'* 41. (sin xj”

2sinx+cos2x, 0S x5 22. 43. sin*x sin 2x + cos*x cos 2x, O< x<
42.
44. Show that the function f(x) = (ax + by/cx + d) has no extreme value regardless of the values of a, b.c, d.
_ | —~x? + [(b’ ~h* +b-1)(b? +3b+2)|, OF «<1

sas Lats had 2x-3 , Se $3.

Find all possible real values of 6 such that f (x) has minimum value al x = 1
[In problems 46 to 50, obtain the Taylor's polynomial approximation of degree n to the function f (x) about the
point x = a. Estimate the error in the given interval.
46. fix)= Vx.n=3,a=1,1524915, 47. fixyzeo! .n=3,a=0,-18191
48. fj =xsinxs,n=4,a=90,-1 x61. 9. f(x)ax'e' .n=4,a=1,05Sx5 15

50. f(x) = 1 ~x),n=3,a=0,05x%50.25.

f (x) about the
In problems 51 to 54, obtain the Taylor's polynomial approximation of degree n to the tuncuion
Hence, write its Taylor's senes.
point x = a. Find the error term and show that it tends to zero as n + ~
=sin 3x, a= 0.
51. f(x) §2. f(x) =sin*x, a=0.
53. flxy =x? Inx, a=. 54. fixy=2, a=.
 1.30 Engineering Mathematics

“ achiem (given
55. Show that the number @ which occurs in the Taylor’s formula with sham ine
in Eg. 1.22)) after n terms approaches the limit I/(m + 1) as h — O provide UOUS and
| .
not zero at x = d.

56. Find the number of terms that must be retained in the Taylor’s polynomial approximation about the point
x = 0 for the function cosh x in the interval [0, 1] such that |Error |< 0.001. a
57. Find the number of terms that must be retained in the Taylor’s polynomial ae about the point
x = 0 for the function sin x cos x in the interval [0, 1] such that |Error |< _— —
Find n such
58. The function In (1 — x’) is approximated about x = 0 by an nth degree Taylor’s polynomial.
that |Error |< 0.1 onO<x<0.5. | mr epolvnonial exeanas
59. The function sin*x is approximated by the first two non-zero terms 1n the Taylor's polyn n
about the point x = 0. Find c such that |Error |< 0.005, when O<x<c. | ine |
60. The function tan7!x is approximated by the first two non-zero terms in the Taylor's polynomial expansion
about the point x = 0. Find c such that |Error |< 0.005, when 0<x<c.
Obtain the Taylor’s series expansions as given in Problems 61 to 65.

61. (xIna)?—
a* =1+xIna+—y— (xIna)?
7 ‘ ., eo <x < 09,

6. » infl=x)=
In = =e 2-4...
5 5 .- <ee
1l+x gt _. x?
63. in = = 2]2+=—+—
5 + r + |—] <x <1

| - x-1 realy .afs-TP

fed inx=al( 24) + 3(SS4) +3(254) +...| x >0

1.4 Integration and Its Applications

Let f(x) be defined and continuous on a closed interval [a, b]. Let there
exist a function F (x) such that
F’(x) =f (x), a<x <b. Then, the function F (x) is call
-derivative of f (x). We observe that
if F (x) is an anti-derivative of f(x), then F (x) + c,
where c is a n arbitrary constant, is also
derivative of f (x). an anti-

1.4.1 Indefinite Integrals

If F(x) is an anti-derivative of f (x) on [a,
6], then for an y arbitrary consta
indefinite integral of f(x) on [a, b] and nt c, F (x) + c is called the
is written as

|fe0de- F(x) +c.

(1.39)
In this case, we say that the fun
ction f (x) is integrable on [a,
Integrable. For example, the fun b). We note that, not eve
ction f (x) defined as ry function is
|

rexy={ 1,0, xisxis irra

rational
tional
 Functions of a Real Variable | 3]

on [O, 1] does not have an anti-derivative and hence is not integrable. We have the following result.

Theorem 1.8 Every function which is continuous on a closed and bounded interval is integrable.
We can evaluate an indefinite integral directly or by using the method of substitution, or integration
by parts etc.

1.4.2 Definite Integrals

Let f(x) be a continuous function on [a, b]. Let

m= {min f(x) and M= {max f(x).

Divide the interval [a, 6] into n subintervals [Xp. X1], [x], x2], . . ., [x,-). X,], Where xp = a, x, = b and
A= X'Sy SZ - SM, = OF ig. 1.1). Let Ax,= x; -2x.), 621, 2, . ..., nm Let

m = fa min (x) an d = ee -o Lg 1.9Beus accel

et M,

Xp 1, 12 Nj-) F Xj
SI

Fig. 1.1. Definite integrals.

and &, be any point in the interval [x,_;, x,]. Corresponding to this partition, define
n

lower sum = |,(f ) = 2 m, Ax, (1.40)

upper sum = u,(f ) = 2 M; Ax, (1.41)

and S,(f) = af (¢;) Ax;. (1-42)

Using these definitions, we obtain
If) S Sn) S Unf ).
The sum S,(f) depends upon the way in which the interval [a, b] is sub-divided and also upon the
choice of the points €; inside the corresponding subintervals. Let n — ee such that max (Ax;) — 0. If
lim 1,(f) = Jim. Sx (f) = jim. u,(f)
1.32) En gineering Mathematics

for any choice of the sequence of subdivisions of the

interval [a, b] and any &€, in the interval
| X:

then this limit is called the definite integral of f (x) over the interval [a, b] and bb Xi],
is written as F(x)
X)dy
Remark 6

(a) For integrability, the condition that f (x) is continuous on

[a, b] can be relaxed. The
f (x) may only be piecewise continuous on [a, b]. functio,
(b) The choice of points *0, Xi, - - ., X, 18 arbitrary. One may choose such
that they for, ‘
arithmetic progression or a geometric progression.
(c) Let ,
m and M be the minimum and maximum values of f (x) on [a,
5]. Then,
b
m(b — a) < | f(x)dx $<M(b - a). (1.43)
b

(d) | f (x)dx = (b - a) f(6),a<o <b (1.44)

(mean value theorem of integrals).
(e) If f (x) is bounded and integrable on [a, b], then |f (x) | is also bounded and integrable on
(a, b], and
b b
| f(x)dx | <| |f(x) |dx. (1.45)
(f) The average value of an integrable function f (x) defined on [a, b] is given by
b
_—
oo | F(x)dx.

(g) Let f(x) and g(x) be integrable functions on [a, b] and let f(x) S$ g(x), a< x < b. Then,
b b
J f(xddes | e(x)dx. (1.46)
if

The method of evaluating a definite integral as a limit of sum can be used only when f(x)
is a simple
function. It may not always be possible to find the limit of sum for every integrab
le function f (x). We
use the following result to evaluate definite integrals.
Theorem 1.9 (First fundamental theorem of integrals) Let
f (x) be a continuous function on a
closed and bounded interval [a, b]. Then the function

F(x) = [ f(tat
Is continuous on [a, b], differentiable in (a, b) and F’(x) = f (x).

Theorem 1.10 (Second fundamental theorem of integrals, Newton-Leibniz

formula) Let
F(x) be an anti-derivative of a continuous functio
n Ff (x) on [a, b]. Then,
b
| f(x) = F(b) - F(a). (1.47)
 1.33
Functions of a Real Variable

particular, they can be used to

find the (i) areas of
Definite integrals have many applications. In ( of surfaces of revolution
etc.
ths of curv es, (11!) volu mes of solid s, iv) areas
bounded regions, (iii) leng

1.4.3 Areas of Bounded Regions

lines x = a,x = 5 is given
of the reg ion bou nde d by the curve y =f (x), the x-axis and the
C, The area
by (Fig. 1.2)
b b (1.48)
.
Area = | ydx= | f(x)dx

Fig. 1.2. Area of region in Cj.

We note the following:

value of the integral given in
a<x<b is above the x-axis, then the
(i) If the curve y = OO), x- axis then
, the value of the integral
If the cu rv e y=f(x),aSxSb is below the
Eq. (1.48) 1s pos iti ve.
ntity, we take the magnitude of this value.
in Eq. (1.4 8) is neg ati ve. Since area is a positive qua
given below the x-axis in the
erval a<x<cand
If the cur ve y = f (x) is above the x-axis in the int
(ii)
n we write
int al c <x <5, the
erv b
[foods
Cc
| +
Area = | f(x)dx
Cc
ad

the y-axis and the lines y

=c,y=dis given
d by the curve x= 0(y),
bounde
C, The area of the region
by (Fig. 1.3) d d (1.49)
Area= | rdy= | o(y)dy.
the lines x= a,x =5
los ed be tw ee n the cur ves y = f (x), y = g(x) and
enc
C, The area of the region
is given by (Fig. 1-4)
ab (1.50)
where f (x) 2g (x) in [a, b].
Area = | [f(x) - g(x)ld,
 s
1.34 Engineering Mathematic

Fig. 1.3. Area of region in C2-

i]
ry
2{?_ -—
* ‘,

J es NOK ON
. ~y : 7 \
. ‘\ et \

. \
% : Ny ‘. : % :
\
| ‘.
‘ ’ a ‘
"

ys g\

| dad:
b
0

Fig. 1.4. Area of region in C3.

d(x)S g(x) in [c, b],

C, If f(x) = g(x) in [a, c] anf a<c <b, then we write the area as (Fig. 1.5)
c b

area= | (s(x) - colar + | [e(x) —f(x)Jdx. (1.51)

Area bounded by a curve represented in parametric form

Let the curve y = f (x) be defined in parametric form as
x= (1),y= W(@),astsb

where (1) and w(t) are continuous functions of f in the interval [a, b]. Let xp = o(a) and x, = 9(9).
Then, from Eq. (1.48), the area is given by
x b
area= [ yar= | W(t) O'(t)dt. (1.52)
x0 a
 Functions of a Real Variable | 4%

| ye f(xy)
KN y= yr)

AN | | .
NX Sy nn > — \

AAS ye f(x)
y = @(x)
|
|

= 4 eX
; b

Fig. 1.5. Area of region in C4.

Area of a sector
Let the curve be defined in polar form as
r=f(0)a<s0<fP (1.53)
by the curve
where f (@) is a continuous function in |a, [ |. Let A be the area of the sector bounded
and the radial lines 6 = @ and 0 = B (Fig. 1.6).
OPN, with
In an element area, we approximate area of the sector OPQ, by the area of the triangle
base PN = rd6 and height ON = OP = r. (PN is perpendicular to OQ). Then
p
dA = srd0, and Area=A = 5 | r*do. (}.54)
ff

Fig. 1.6. Area of sector.

the curves y = Jx andy=x-

Example 1.29 Find the area of the region enclosed between
x = 1.
Vx =x’, or x* — x =0, that is atx = 0 and
Solution The curves intersect at the points where
as
Since Vx > x2 when 0 $x$ |, we obtain the area

prea = [)[Ye — 2? Jae= 3 ~ 4= square units

1 /

P A »
0
 1.36 Engineering Mathematics

Example 1.30 Find the area of the region enclosed by the curve x = a(t — sin t), Y=a (1 ~¢¢ t),
O<t<s27n.
Solution As ¢ varies from 0 to 27, x varies from 0 to 27a. Hence,

2a
Area = | [a(1 — cos t)] [a(1 — cos r)]dt
0

2a 2ma 1
=a’ | (1 — 2cos t+ cos*1)df = a?
| f — 2cos ft + 5 (1 + cos 21) at
0 0
- i. ;
a? 2ma : 1 r 2na
= _
= =
(3 — 4cost+ cos 2r)dt = 7 3 4sint+ 5 sin [
9) I

= 3a” square units.

Example 1.31 Find the area of the region that lies inside the circle r = a cos @ and outside the
cardioid r= a (1 — cos @).

Solution The region is given in Fig. 1.7. The curves intersect at 9 = + 2/3. Let r; = a cos @ and
r2 = a (1 - cos 6). Therefore, the required area is given by
n/3 I mii
Area= 3 | (7° -—r7)dO= =a? | [cos*@— (1 - cos 0)” ]d@
—7/3 2 —m/3

ri3

=a?|0 (2 cos8 1)d8

=a*[2 sin8-B15” (2)
|
; 5]
3

= = [3/3 — 2] square units.

Fig. 1.7, Example 1.31,

 Functions of a Real Vartable | ,37

1.4.4 Arc Length of a Plane Curve

Consider a portion of the curve y = f (x) between x = a and x = b. Then, the length of the arc of the
curve between x = a and x = b is given by

BY ax
sf fi+(2) (1.55)

If the curve is defined by x = @(y), c < y < d, then the length of the arc is given by

d | bee
ee | 14 (=) dy. (1.56)

Arc length of a curve represented in parametric form

Let the parametric form of the curve be given by
x= (1), y= Wt),t9 St Sty

= b, then
where @(t) and y(t) are continuously differentiable functions on [to, t)]. If O(to) =a and o(r,)
from Eq. (1.55), the arc length is given by

sof (Bla f(a) ga fa) (Sle oe [<>2

; | nh, |t- \2
:
_
b
|
:
dy 2 = i ae dx _

Arc length of a curve represented in polar form

f (6), a< @ < PB, where f (9) is continuously
Consider the portion of the curve defined by r =
d In parametric form as
differentiable on [@, 8]. The curve can be represente
6.
x=rcos @0=f(@)cos 6 y=rsin 6=f(0)sin 8, as O0<

Therefore,

dx = f’(@) cos 6 — f(@) sin 8 Be = f’(@) sin 8 + f(@) cos @

dO * d@
dx - dy : ” P 4) el .
e + (| a ( F'(8)|” + fF" (@).
and
aand @= Bas
of the portion of the curve between @=
Using Eq. (1.57), we obtain the length
, B “ed
B (1.58)
r? +(45) de
s= | J £7 (8) + [f'(O)I° dd = |
a
|
a

f(B) 2 |
e= + 1 dr. (1.59)
ie
f(a@)
1.38 Engineering Mathematics

Example 1.32 Find the total length of the curve r = a sin?(0/3).

Solution The curve is defined when 0 < 6 < 37. We have

f(@) =a sin’ (2).7@) =a sin?(2)cos @ and f2(0) + [f(8)? = a? sin* ($)|

5) ad = : Q
| :

Therefore,
30 | @\ 10. nm = 1
3

s=[ JP) + (FOR AO=a | sin?($)d0= 30 | sin“ odd

where @ = 3¢. Integrating, we obtain

— -con2p
r=32[1 8[o-
3a[,_ fine2a],== 34%39.
rdo
1an.rg|

1.4.5 Volume of Solids

In this section we discuss methods for finding the volume of solids.

Method of slicing
Let a solid be bounded by two parallel planes x = a and x = b (Fig. 1.8). Divide the interval [a, 5]
for x into n subintervals [xo, x;], [x, X2], - - «> [%n-1, Xn], Where @ = x9 < x1 < x2... <x, =D. Let
Ax, = X_—Xj-1, k= 1, 2, .. ., n. Draw the planes x = x9, x = x), ..., x = X,. This will cut the solid into
slices of thickness Ax,. We now approximate the volume of the sliced solid part S, by the volume
of a cylinder with base as a cross section of the sliced solid S, and the height as Ax,. Therefore, an
approximation to the volume of the sliced solid between x = x,_; and x = x, is given by

Vy = A(€,) AX p, Xp) < Cy SX,

Lo \ @,
se
2S
=o a H
=
a

Fig. 1.8. Method of Slicing,

 Functions of a Real Variable 1.39
where A(€,) is the cross-sectional area of the sliced solid. Now,
consider the sum of all the approximate
volumes of the sliced solids. We obtain

n A

in = Py ve = Pa A(Se) Ary:
Let n — co such that max Ax; — 0. In the limit V,, > V, volume of the solid and the summa
tion reduces
to an integral. Therefore, volume of the solid is given by
b
v- | exhib (1.60)
If the solid is bounded by the planes y = c and y = d, then volume of the solid can be written as
d
y= | A(y)dy

where A(y) is the cross-sectional area.

Example 1.33 The cross sections of a certain solid made by planes perpendicular to the x-axis are
circles with diameters extending from the curve y = 3x” to the curve y = 16 — x’. Find the volume of
the solid which lies between the points of intersection of these curves.
Solution At the points of intersection of the curves, we have 3x” = 16 — x’, or x* =4, that isx=+2.
Therefore, the points of intersection of the curves are (— 2, 12) and (2, 12) (Fig. 1.9).
Any point on the curve y = 16 — x” is R(x, 16 — x’).
Any point on the curve y = 3x’ is S(x, 3x’).

Fig. 1.9. Problem 1.33.

Diameter of the circle = RS = 16 - 4x’.

Area of the circle (R=
A(x)= = 44 (RS)*
ircle==A(x) 4n( *
=S)
4n(4 - x2)?
¥.

Since the solid is symmetric about the y-axis, the required volume is obtained as
2 2 ;
V=2 | A(x)dx = sx| (4 ~x*)* dx = sn 16_ 8 3 - =|
0 0 | 0

64 32]
_ 2048 a
= $132 -_> + 2 an (a 7 cubic units.
 1.40 Engineering Mathematics

revolution
Volume of a solid of eenx=a and fa — b. Consider the area boundeg
f(x)> 0, betw
a curvey = f (x),
LetAB be the portion of , and the lines x = @ and x = b. A solid is generated py
by the arc AB of the curve y = f (x), the x-axis
revolving this area about the x-axis (Fig 1.10).

Fig. 1.10. Solid of revolution.

Divide the arc AB into n parts by considering the subintervals [xp, x1], [x,, X2], ~~ -s [Xp-1. Xq];
where a =X) <X| <X%)... <x, =5. Let Ax,=x,-—xy_), k= 1,2, ..., n. Consider a typical subinterval
[x,_1, x] of length Ax, = MN. A solid is generated by rotating the area MNQP about the x-axis. The
volume V, of this solid lies in magnitude between the volumes generated by revolving the rectangular
areas MNRP and MNQS about the x-axis. Now,

MP = yy) =f (%e1) and NQ=y,= f(x,).

Hence, the volume V, of the typical solid is bounded as

Tye, Ax, <S Vi Ss Ty? Ax,.

Adding the inequalities corresponding to all the subintervals, we get

> 4 A < a n -

Myo
Mk Ok S 2M S Vie Ate.
Let n — © such that max Ax, — 0. In the limit, we obtain the
volume of the solid of revolution as
b
v= | My*dx.
ad
(1.61)
Similarly, if the area bounded by the arc AB of the curve
y = @is revolved about the y-ax
x = (y), the y-axis, and the lines y = c and
is, then the volume of the solid of
revolution can be written as
 1.41
Functions of a Real Variable

| (1.62)
v= f mx dy.
c

Remark 7
x= b is revolved
the curv e y = f (x), th e li ne y = p and the lines x= a,
(a) If the area bounded by solid of revolution is
= p (a line para llel to the x- axis ), then the volume of the
about the line y
given by
b
(1.63)
ven (y — p)* dx.
a

revolved
the cur ve x = g(y ), the line x = q and the lines y=c, ¥ = d is
(b) If the area bounded by y-ax is), then the volume of solid of rev
olution is given
= (a line para llel to the
about the line x
by
(1.64)
y= x | (x —q)° dy.
. |

c
nded by the
vol ume of the soli d gen era ted by revolving the finite region bou
Example 1.34 Find the
line x = 4
curves y = x’ + 1, y = 5 about the
1.11.
Solution The required region is given in Fig.

y=5
=

oe SAAR i te, ie, x=3

RAO, SS iia ite Eh, a

. 7," Ny, ~* i, i+ |
Nie ah |
ie

(x1, Yt) eee SPT

1.34.
Fig. 1.11. Region of revolution in Example

The volume is given by

5 5
[3+ yy-1)?-G- y—1)? |dy
y=n| (xf - x3 )dy = |
¥ |
|

.
— 1)22]? = 82(8) = 647 cubic units.
5

= on | Jy - ldy= 12n( 3][(y

|

revolving the region bounded by the curves

Example 1.35 Find the volume of the solid generated by
y= 3-2 and y =~ | about the line y = — I.
 1.42 Engineering Mathematics

ion is gi in Fig. 1.1 9. The region PAQ is revolved abou the lin
Solution The required region is given in Fig y- axis, the volume is: obtained as ]
y =— 1. Since the region is Symmetri cal about the

Fig. 1.12. Example 1.35.

V = 2 [volume of the solid obtained by revolving the region PAR about the line y = — 1]
= 2
=2n | (1+y)de=2n | (1+3—x7)* ax
0 0

2
=2n| (16 -— 8x2 + x4)de= 2]16x— 8x?
2 '

80rn ra | = 212 mz cubic units.

0 5 6 15
Example 1.36 Find the volu
me of the solid generated by revolving an arch
* = a(t—sin t), y = a(1 —cos t) and x-axis abou of the cycloid
t the x-axis.
Solution * Setting y = 0, we obtain cos t =
Lorr=0 U, and t = 27. Hence, one
Intersects the x-axis at the poin arch of the cycloid
ts (0, 0) and (27a, 0). Therefore, the
required volume is given by
2a | 27
V= | y*dx = “| a*(l — cos r)? [a(1
— cos t)|dt
0

: 2X F
X
= Ta
—

I 8 sin (5Ja- on
a’ | sin®T dT
1 6
|

ni2
= 32na° | sin°T dT = 32
: ial 13.3. l 2
ak ak 5 = 52*a? cubic units.
Volume of solid of re
volution by the meth
od of cylindrical Sh
Suppose that a region in ells
the x-y plpla
annee bounded by the
_" 5 is revolved abou
— 1
curve |
t the y-axis. Divide . e i

the interval [a, b] i

| y = q I

oaiees
~
=

Sea ae ad
—= i ¥ — ¥ i oo» fb a | 1. OnS] er
|
 Functions of a Real Variable 1.43

M(x;_1, 9) and N(x,, OJ.

a NOP (Fig. 1.13), where MN = Ax, and the coordinates of M, N are
thin walled shell of inner radius
ea ee strip about the y-axis, it generates a hollow
shell is a ring bounde d by the concentric
ote ae oodwe and volume AV,. The base of this | '
id is g iven
+ Ax,. A cross-section of this sol
= sa Pe
§ x.) and outer radius x, = x4-1
“s inner radius x,
in Fig. 1.14.

on.
Fig. 1.14. Cross-section of solid of revoluti
Fig. 1.13. Region of revolution.

The area of this ring is given by

= 2% Ey AX,
AA, = IX; — THe = U(X, 5 Xp_1) (xX, — X,-1)

r and outer boundaries of

&, = (x, + X4- 1)/ 2 is the rad ius of the circle midway between the inne tant height y
where Now , ‘f we take a cylindrical shell of cons
its cir cum fer enc e.
the ring and 27 ¢;, is as AV, = (AA,y. Since f (x) ts continuous,
y can take any
e, we obt ain vol ume
standing on this bas of f (x) on [xy 1, x]. If we take
y = f (Tp),
m and maximum values
value between the minimu shell as
Xb the n we can wri te app roximately the volume of the
Xp SM S
Poa © 2; si nl.
;
AV; = (27 E.)f (Nm) AX

ervals, we obtain
esponding to all the subint
Adding the volumes corr

M) Xe:
V, = Py AV, = &. (270 4) f(

me of the solid as
ma x Ax , — 9. In the limit, we obtain the volu
Let n — ce such that
b
(1.65)
v= | Imax f(x)dx.
ra
revolution is
about the x- axis, the volume of the solid of
1.15 is revolved
If the region given in Fig.
obtained as
a (1.66)
V= | 2my g(y)dy
c
 s
1.44 Engineering Mathematic

Fig. 1.15. Region of revolution.

where x = g(y) is the equation of the bounding curve AP QB.

jx,
ted by revolving the region bounded by y=
Example 1.37 Find the volume of the solid genera
y = 0 and x = 9 about the y-axis.
l strip of the area between
Solution The region is plotted in Fig. 1.16. When we revolve the vertica
circumference
the lines at distancesxand x + Ax from the y-axis, we generate a cylindrical shell of inner
e as
2x, inner radius x, inner height y and wall thickness Ax (Fig. 1.16(a)). We obtain the volum

2), 5279 _ 9722

V= |: 2mxydx = an |° xx dx = 2n(2 [x2], = —— cubic units.
0 0
y y

(a)
(b)
Fig. 1.16. Example 1.37,
Alternative If we revolve the horizontal str;
as nia’ stp about: the y-axis (Fig. 1.16(b)), we obtain the volume
_
3 4
3

#0
where x, = 9, x, = y*. Therefore,

3 3
V= x | 81 dy~ | y*dy = [243 - 243 es 9727
0 0 a cubic units.
 Functions of a Real Variable 1.45

lving the region bounded by the curves

Example 1.38 Find the volume of the solid generated by revo
y=1+ Vx and y=1 +x about the y-axis.
x = I. The points of
Solution The curves intersect when | + x = 1 + 4x, or when x = 0 and
When we revolve the vertical strip
intersection are (0, 1) and (1, 2). The region is plotted in Fig. 1.17.
s, we generate a cylindrical shell
of the area between the lines at distances x and x + Ax from the y-axi
+ vx )—-(1+x)= Jx —x and wall
of inner circumterence 27x, inner radius x, inner height y* = (1
thickness Ax (Fig. 1.17(a)). We obtain the volume as
1 5/2 3
1
x a = 2% i cubic units.
v= | nx yt dx=2n | x(Vx —x)dx= 22 5/2 3
0 0

(0, 1)

(a)
Fig. 1.17. Example 1.38.
1.17(b)), we obtain the volume
Alternative If we revolve the horizontal strip about the y-axis (Fig.
aS
2 2

ven dyn] x5 dy
l 1

where x, = y — 1 and x2 = - 1)’. Therefore,

2
O-De _ 11m
ven[ (y-b?-O- Dilan
2
(gad?
= : ]
= 113 —5|= 75 cubic units.
l

1.4.6 Surface Area of a Solid of Revolution

this curve be revolved about the
Let y =f (x), f (x) 2 0 between x = a and x = b define a curve. Let
tervals [Xo, +1],
x-axis to generate a surfac e S (Fig. 1.18). Divide the interval [a, b] into n subin
<x, = b. Let Axy =X — Xe and Ay; = ye- Veet
[xy Rebs s « vx igpete Sel where a =X <X1 < X72...
curve PQ in the interval [xj-1, %«]- Let 5x
= f (x,) —f (4-1), ¥= 1,2, ...,n. Consider the portion of the
of curve about the y-axis. In this interval
be the area of the surface generated by revolving this portion we
If we revolve the chord PQ about the x-axis,
[xp-1. X;], We approximate arc (PQ) = chord (PQ). is approximated by the area
obtain a frustum of a cone (Fig. 1.19). Now, the area of the surface S,
of the surface of the frustum of the cone. We have

PM = yp.ON = Yo PQ == J (Ax,)*+ (Ay,)?

 1.46 Engineering Mathematics

Fig. 1.19. Surface of revolution,

Fig. 1.18. Surface of revolution.
—_— r
and Se = 1 (yp, typl = Mer + Yd) V(Axn)° + (Ave)

Ay k ‘
= Uy + Ye) 1+ (Be) Axy

Adding the approximations corresponding to each of the subintervals [x,_;, x,],k = 1, 2,...,n, we
obtain

fn fl A

= 2S = - Emons tye) fie (Be) AXx,.

k=l k=l k
Let n — co such that max Ax, > 0. In the limit, we obtain
the surface area of the solid of
revolution as
|

= 2my ,j1+ ae dx = 2my ds

Ji x , (1.67)

where ds = 4/1 + (dy/dx)*

dr.
If the given region is revolved about
the y- axis, then we obta
in the Surface area
as
d
2
=|f 2x
| 1 ax
dy dy= )° 2x ds
(1.68)
where ds = y/1 + (dx/dy)?
dy.
If the curve js given
in parametric form aS
X= P(t), y = VW), t<
] —_t<t
— Ff]; then we have

ds = dx
ay\2
[ey(3 dt.
(1.69)
 Functions of a Beal Variatle | 4)

If the curve is given in polar form r= { (9), ,< O< G;, then we have
ay \?
a= yApt ia
+ Ee | dG
7a
|
(}. 79)
ds

Example 1.39 Find the surface area of the solid generated by revolving the circle x“ + (y by =a’
b 2 a about the x-axis.
Solution The equation of the circle can be written in parametric form as
x=acost,y=bsasint,Ostt27

We obtain

a ae 2
as 1) «| 2) = /(-asint)’ 4 (acnt) =a
dt \\ dt) dt} *

Therefore,
27
square units
S- | 2m(b+asintjadt = 2 nal bt _ a cos t\2" = 4 xab
0)

1/4, is revolved about the x-axis

Example 1.40 The part of the lemniscate P =2a* cos 20,0< O<
Find the surface area of the solid generated.
Solution We have x = rcos 6, and y = sin 6. We find that

‘4 ; 4
4 int DOF
| \" + 4a* sin* 26) = o..
=r? + So sin 28 _ 5.[4a*cos? 26
ds* =r" (5)

emis / ) 4
mia

2m y dg = | Inrsin@| —— \dé
Therefore, s= |
f) _.
0

9: Wy LF
2m a°(2 — ~2) square units.
mia oh . |=
, I oF
= 4na7|-cos 8] = 41a

x = sin* t, y = cos“ t, 0 <t < 77/2 is revolved about the y-axis. Find
Example 1.41 The line segmenera t
the sol id gen ted.
the surface area of
en by
Solution The surface area is giv SS TT

| . | ( dx\"
s= | ie wt) dy.
r

We have
| . d . dx
GX <2 sint cost, = = ~ 2acostsint and ae it,

| to 0. Therefore,
As t varies from 0 to 1/2, y varies from
| air ;
sin? (--2 sint cos t)dt
S§= { aJInx dy = - | 2-42
0 fj
1.48 Engineering Mathematics

zf2
-_ ‘ Helir cnt)
———
=
=
22 equate vans
= 43 2| sin“rcostdr= 42% )/7 S18 Ll —

Exercise 1.3
Cie”
2 ear al

In problems 1 to 9. find the area of the region boundedDy tae grvem

* - et .
= o 7 :;
5

l. y= x — Sx + 6. the x-auis and the mes x = 0. 2 =

/-
-_

Pr -_= ~=—- 741 mee yeas}

2 y= ¥x and the ines x
y=x, =0.2=! 3. y =2 am) =4
2
“
a
?
x2
a.
; S wyx+47=
_— —_
anc tie CoUPOiidit 2a
-

4.
a” b* D — iy —

6 y= ex In x and y = In x/(ex)
7. x= 241, £: < 1/2
7 =4¢-1.-1/2 and the x-as.
-_ ioe ~*~ Fh 0<6= 22

& x=acost, y=bsim1.0<: —

—
-com 29).92°92 22

the curve » =-2—r.

10. Find the area of the region enclosed between the peas and the orden: oS tie
points where the curve has local mummum
11. Find the area of a loopof the curve x00 +) = alr - yr)
ecos 26 anc oulsiae ©.
7 = Za
12. Find the area of the region inside the curv
rerae
-——<—
f= €

13. Find the area that is inside the circle , = a and outside the cardiac r= 2 -
—-— fF Te. eT
oe ae =

In problems 14 to 26, find the Jength of the indicated pornon of tine curve
14. 9x?=y, from x=Otox=9.

15. x77 + y** =a”, from x = 0 to x =a in the first quadrant

16. x=4tyt41,. from y= lwo y=2.
4 8
17. x° + y’ — 2ax = 0 and above the line y = a/2.a > 0.
18. yal: Vcost dt, from x=0 tox = 2/2. 19. y = In [(e*+ 1 yWle*
— 1) . from =

20. x = 3ar’, y = alt - 39), from 1=Otor= ].

21. x =a(t—sin
1), y= a(l—cos!), fromit=Qto1r=2z
22. x=" cost, y=e” sins, fromi=Ow1=I.
23, x = (In (a’ + P)Y2,
y = tan” (t/a), fromt=Owr1=a
24. x = 2c0os t+ cos 26+), y=2 sins + sin 2¢, fromr=Otwr-7
25., r= a6, from r=r, tor=ry.

26. r=ae™, from r=r, tor = rp

2 y2 2
27. Find the volume of the ellipsoid a3 ~ re: ~ = ie 1.
28. The base of a certain solid is the circle x7 + y? = a*. Each cross-section of the solid cx og by 2 Dae
perpendicular to the X-akiS 18 a Square with one side of the Square in the base
of the solid Fmd
volume of the solid.
29. The base of a certain solidis the circlex* + y* = a*. Each cross-section of the
solid cat out by a pame
perpendicular to the x-axis is an isosceles right triangle with one of the eine cides in the bast of OF
solid. Find its volume. SO
WW. The base of a certain solid is the region between the
%-axis and the curve y = cos x between7 x=
xX => 0 ra
 Functions of a Real Variable 1.49

x = 7/2. Each cross-section of the solid cut out by a plane perpendicular to the x-axis is an equilateral}
triangle with one side in the plane of the solid. Find its volume.

In problems 31 to 36, find the volume of the solid of revolution generated by revolving the specified region about
the given axis. |
31. Region bounded by y = cos x, y= 0 from x = 0 to x = #/2 about the x-axis.
32. Region bounded by y = Jx, y = 0 from x = 0 to x = 4 about the x-axis.
33. Region bounded by y= 4x, y = 0 from x = 0 to x = 4 about the line y = 2.
34. Region bounded by y = x’ + 1 and y = 3 — x about the x-axis.
35. Region bounded by x = a sin’ t, y=a cos’? 1,0<1< 2/2,x=0, y = 0 about the x-axis.
36 . Region bounded by x = 24 + 1, y = 427 - 1, -1/2 <1 < 0, y = 0 about the line x = 1.
In problems 37 to 41, use the method of cylindrical shells to find the volume of the solid generated by revolving
the specified region about the given axis.
37. Region bounded by y = x, y = 2 and x = 0 about the y-axis.
38. Region bounded by y = 2x — x’ and y = x about the y-axis.
39. Region bounded by y = x’ and y = x about the y-axis.
40. Region inside the triangle with vertices at (0, 0), (a, 0) and (0, b) about the y-axis.
41. Region inside the circle x* + y’ = a’ about the line y= b, b>a>0.
In problems 42 to 50, find the surface area of the solid generated by revolving the curve C about the given line.
2
“
1,a2b, y2 0 about the x-axis.
2 1

42. (x —b)* + y? =a’, b> a about the y-axis. 43. %-+2-=-

aal
b

2 : 4 .
44, x” +> =1,a>5b, x >0 about the y-axis. 45. yet |
>» 1 <x <2 about the line y=— 1.
|

oox

a
46. 292+ line x
about theline
ie qt eT=? < 2 about x = -1.

47. x=a(t—sin?t), y=a(l-cosft),0<rs 22 about the x-axis.

48. x=acos’t, y=a sin’ t, 0 <1< 2/2 about the x-axis.

49. x=e' cost, y=e' sint, 0<t< 2/2 about the y-axis.
50. r=a(1+cos 6), 0 < @< about the initial line.

1.5 Improper Integrals

b

While defining the definite integral | f(x)dx, we had assumed that

(i) a and 5b are finite constants.
(ii) f(x) is bounded for all x in [a, 5].
but f(4) becomes
If in the above integral, (i) a or 5 or both a and b are infinite, or (11) a, b are finite
finite integral
infinite at x = a or x = b or at one or more points within the interval (a, b), then the de
is respectively called
(i) improper integral of the first kind.
(ii) improper integral of the second kind.
To define the improper integrals, we assume the following:
 1.50 Engineering Mathematics

(i) The integrand f (x) is of the same sign within its range of integration. Without any loss of
generality, we assume that f (x) > 0 (when f (x) < 0, we can write g(x) = — f (x) so th
8 (x) 2 0). We shall discuss later, the case when f(x) changes sign within its range
of integration
(1) f (x) 18 continuous over each finite subinterval [a, 6] contained in the range of INte gration
Hence, there exists a positive constant K independent of a and f such that

B
| f(x)dx<K,
(t

The improper integrals are evaluated by a limiting process.

1.5.1 Improper Integrals of the First Kind (Range of Integration is Infinite)

We shall now discuss methods to evaluate improper integrals of the form
oo b oo

(1) | f(x)dx, (ii) | f(x)dx, and (iii) | f(x)dx

if they exist. We define these improper integrals as follows:

o2 p
(i) f(x)de= lim| f (x)dx. (1.71)
If the limit exists and is finite, say equal to l,, then the improper integral converg
es and has the
value /;. Otherwise, the improper integral diverges.

(ii)
b b

= oo
f(x)dx = lim | f(x)de. (1.72)
p
If the limit exists and is finite, say equal to l,, then the
improper integral converges and has the
value /,. Otherwise, the improper integral diverges.
eo c
(iii)
b
[_ Fladar = lim, I P(x)+de
lim | f(x)dx (1.73)
where c is any finite constant includin
g zero. If both the limits on the righ
and are finite, say equal to l; and /, respecti t hand side exist separately
vely, then the improper integral converge
value /; + /4. If one or both the limits do not exist or are infi s and has the
nite, then the improper integral diverges.
Example 1.42 Evaluate the following improper inte
grals, if they exist.
= dx
(i) | ee (ii) [ (iii) . x
0 a* +x? ] x (E i | es ax,

(iv) | x sin x dx, (v) | e-“cospx dx, a > 0, P constant

0 I
Solution |

(i) [ +45 =Jim ° dx =< 1G L cand |b 1

0 atx? bom Jy gr? oy? oe |g Glas
Therefore, the improper integral converges to 7/(2a).
 Functions of a Real Variable 1.5]

: = | a
dx
c O64 go =]

he ref b

feel ames _,

= lim [sec™! nen —_ b — sec"! c]

_ “i 1 _ .
= SE€Ccl ¢ — sec”! l + > — sec londt
2
Therefore, the improper integral converges to 7/2.
b
(iil) { e*dx = | «ar 2 jim | e~*dx= lim (1 -e7’) = 1.
})—} 00 0 bc

Therefore, the improper integral converges to |.

oo b

(iv) | xsinx dx = lim | x sinxdx = lim (sinb — bcos b).

0 Ta AiO a
Since this limit does not exist, the improper integral diverges.
(v) Using the result

|e~ “cos px dx = ———_; (p sin px — a cos px),

a“ +p-
we obtain
b
b — (it

e~**cos px dx = —___ (p sin px — a cos P|

a~ + p* 0
0

= —— fe" \(p sin bp — a cos bp) + a]

-ab _¢
tim, e~* = (0). Hence,
a 43

Now, sin bp and cos bp have finite value s and

b .
eo
lim I e“cosprp ik =
sdx = jim San
j e~™*‘on copx
Bie: 2
the impro per integr al conve rges to a/(a° + p’),
Therefore,
ce of the improper integral
Example 1.43 Discuss the convergen
|2
) x?

Solution We have

Now, lim [be | : QO.

 1.52 Engineering Mathematics

ral converges if p > 1 and diverges if p< 1.

Therefore, the improper integ
For p = 1, we have
ea b
ax = lim | ax lim [In x]? = lim In b.

Since the limit does not exist, the improper integral diverges. Hence, the given improper integra)
converges to I/(p — 1) for p > 1 and diverges for p < 1.
oo

Example 1.44 Discuss the convergence of the integral | xen* dx.

Solution We write

oo , c 4 said 2

=| xe? d= | 4o°* ax+ | xe ade

— oo — Cc

where c is any finite constant. We have

c b
I= lim | xe7™ dx + im | xe~* dx
d——o b— co ‘
a

= lim E (e7® — ee |+ jim E (e-" =e

a——ee

-l eer 4 e- )=O,
2
Therefore, the given improper integral converges
to 0.
It is not always possible to Study the converge
nce/divergence of an improper integral by
evaluating
itaswas done in the previous examples. pies. AA simp
sj le| example is the inte
I gral © e~*
g-? dx which cannot
be evaluated directly. We now present some
results which can be used to
i

discuss the convergence or

-

divergence of improper integrals. In this ca

0

Se, we cannot find the value of the improper

is the value to which it converges. Howev ef, integral, that
we may be able to find a bound
of the integral.
Theorem 1.11 (Comparison Test 1) If 0 < Ff (x) €
g(x) for all x. then
sa]

(1) | J (x)dx converges if | 8(x)dx converges.

(11) / &(x)dx diverges if | f(x)dx diverges.

Theorem 1.12 (Comparison Test 2) Suppose that f(x) and g(x) are Positive functions and let

lim
f(x }_ 7 ¢
X—> 09
Folle — (1.74)
 Functions of a Real Variable | 53

Then, the improper integrals | f(x

)dx and | 8(x)dx converge or diverge toge
ther.
a
Example 1.45 Discuss the convergence of the fol
lowing improper integrals
| =
(i) I = dx
3 en” dx. (ii | p=
(11) ea>e iii
(111) ax
; nx’

(iv) | . ae (v) © XtanTtx

7) x(in x)? dias
+ x3

Solution

‘ ee
(1) We have e-*" < e™
=
forall x> 1. Consider the impr
'
=

oper integral | e* de:

l

We have
—

| e“*dx =lim | e“*de=Jim [1 - e-] =1.

=
b

—)

Therefore, the integral | e~~ dx is convergent. By Compa

rison Test 1 (i). the given integral
is also convergent. Further, its value is less than 1.

) ] | re]
(11) Let f(x) = ————._ and
F(a) (e * + 1)x? ame ake) x?

Now lim Fal = lim ——+___ =| = lim = 1

x40 g(x) | ios (e-* a 1)x ] Xe eT * 4 |

Also, [ 2(x)dx = | —z
3 converges to | (see Example 1.43). Therefore, by Comparison
Test 2, the given improper integral is also convergent. Its value is less than 1.

Alternative We have ——
< + for all x = 1. The improper integral | J is
(e-* + 1)x x x2
, ; l
convergent. Therefore, by Comparison Test | (i), the given improper integral converges.

(111) We have In x < x for all x > 0. Hence,

ajuda a> | dx
Inx <x 5 Inx 3 2

Let g(x) = 1/(In x) and f (x) = 1/x. We have g(x) > f (x). Now, the integral

| f(x)d& = | a is divergent (see Example 1.43).

2 2
oo oc ax ; | |

Therefore, by Comparison Test | (ii), the integral | e(x)dx = | ae also divergent.

2 2
L.S4 Engineering Mathematic
s

Qv) Substitute In ye 7. We pet

l= [ a [. dt
» wV(Inw)? ing oe?
Which is Convergent tor P> | and divergent for ps | (see Mxample 1.43),
- | antl
iW) Let f(x) = Xan - 1a. and g(x) © be.
v4 bx" Vx VI hax Vx
We find that lim Lie) lim —pilendeer = 2,
ror BON) vdeo Vi¢e4x-3
5

Hence, by Comparison Test 2, the integrals =

| f(x)dx and | R(x )dx CONVETBE OF diverge

|
together. Now, | &(x)dx is divergent.
Therefore, | f(x)dx is also divergent.

1.5.2 Improper Integral of the Second

Kind

b
Now consider an improper integral of the
form | f(x)dx, where a, b are finite constants,
f (x) is
continuous in (a, 5) and has infinite disc
ontinuity (becomes infinite) at (i) x = a,
(iil) x= a and x= b. or (iv) f(x) is or (li) x = b, or
continuous in (a, b) except atx=c,a<c
infinite discontinuity. e< b, where f(x) has an
If f (x) has a finite number of points of
discontinuity, Oly CQ, ss ley ASC)
then we write the integral as Seg... <0, $b,

[ f(x)dx = [ f(x)dx + [ Sed tees f f(x)dx

a a ¢ (1.75)
a
and consider each integral on the right hand
side ‘eandely
Infinite discontinuity at x=a Since the function f (x) is continuous at
all points except at x = 4,
the integral f(x)dx is a proper integral and
ate
exists for every €,0<ée€<hb—a.
We evaluate the improper integral as
b b
f(x)dx= lim | (x)dx
J ane ate d ,
If this limit exists and is finite, say equal to /;, then the
improper integral converges to /,. Otherwise,
it diverges.
Infinite discontinuity at x = b Since the func
tion f(x) is continuous at all points except at x
b-€ = b, the
integral f(x)dx is a proper integral and
exists for every €,0<e<h-— a.
if

We evaluate the improper integral as

b b-e
| f(x)dx = lim | f(x)dx.
 Functions of a Real Variable | 55
Infinite discontinuity at x = a and x = h We write
the improper integral as
b (x by

| f(xrde = | f(x)dx+ | f(x)de

where @ 1s any finite constant between a and b at which
f is defined. We evaluate the improper
integral as -
b Ct b-—€
| f(x)ax= lim f(x)dx + lim F(x)dx.
a ot d+eé 60 Jy
If both the limits exist and are finite, then the improper integral converges. Otherwise,
it diverges.
Infinite discontinuity at x = c, a <c<b We write the improper integral as
b c b c-e b
| f(x)dx = | f(x)dx+— fix)dx = him | fx)dx + tim | f(x)dx.
a a *C = a i o+e

The given improper integral converges, if both the integrals on the right hand side converge. If one
or both the integrals on the right hand side diverge, then the given improper integral diverges.
| The following tests can be used to discuss the convergence or divergence of the above improper
integrals. In this case, we cannot find the value of the improper integral, that is the value to which it
converges. However, we may be able to find a bound of the integral.
Theorem 1.13 (Comparison Test 3) If 0 < f(x) < g(x) for all x in [a, b], then
b b
(i) | f(x)dx converges it | g(x)dx converges.
(l a

b b
(ii) | g(x)dx divergesif | f (x) dx diverges.
a ra

Theorem 1.14 (Comparison Test 4) If f (x) and g(x) are two positive functions and
(i) a is a point of infinite discontinuity such that

fing 2S!
a ge) == Cee
pee ee
etat+hy) =1,,0<1,
eS <©

or (ii) b is a point of infinite discontinuity such that

_ f(x)_,. f(b—-h)
=14,,0<l, < 00
Aes g(x) a g(b —h)
b b

then, the improper integrals | f(x)dx and | g(x)dx converge or diverge together.
a a

Example 1.46 Evaluate the following improper integrals, if they exist:

4 2

(i) |0 =
Vx
(i) |0 ah
V¥4—x?
156 En gineering Mathematics

1 ~
i ]+zx 3
il dx dx
ay) f [-x (iv — 9
i)3x -—x*

{v) : ax
SS
(vi
V1 So dx
-| -" gee eee
6 x? — 3x+2
Solution

0 7h) 4 ma NEI=4
Therefore, the improper integral converges to 4.
2
a 2-€
dx
(41) | , = jim ———
dx ; si = —
0 J4 x2 E—)
= lim sin { 3) =sn*] =
5 l4 _ x? E—rf) 2 NIA
Therefore, the improper integral
converges to 77/2.

’ ties mies edgy

(iy

me i; Leo —— — ' j

I Vi-; "2 [. i
fe ae
21,
!

For
—_ \ !
. a dni
j |

-l

= lim [{sin-'(1 - €) — sin~!(_1)} —

{(Vl-(l-e)? - VT— 1}]
= sin” (1) — sin“(_]) = 9 sin
(l)=7
Therefore, the improper int
egral con Verges to 77.
(1v) Here, the integran f (x)
d has infinite discontinuity,
take any point, Say x = c, at both the end points x
inside the interval of inte = 0 and x = 3. We
gration, at whic fh(x) is defined. We write
[ dx -[ dx ; , ods , ©
Jo 3x — x?
dx 3 dy
Fi) 3x — x? ena = Him
| ee + bi } x(3-—x)

- 1; f —_)\ 1, e
s4|'(s3)] Yr
+5 yn[in( 522)
€

oe or = E . _¢ ge
= 5 im | 3-¢ (3a) | Som [in(258)
=]

We write
7 | of integration
i 0 | =
= = dx dx ~ lim . dx . I dx
~{ * , x? 2 90) xe * pm é x2
 Functions of a Real Variable 1.57

Therefore, the improper integral diverges.

2, both of which lie inside the
(vi) The integrand has infinite discontinuities at x = | and x =
interval of integration. We write

i dx 7 dx : dx +f dx
, y= tre) és (x-1Iyx-2)- (x — 1)(x -2) ‘ (x — 1)(x - 2)

= I + I, ~~ I.

We find that
(a) the integrand in /, has infinite discontinuity at x = 1,
2. We
(b) the integrand f (x) in /, has infinite discontinuity at both the end points x = 1 and x =
f (x) is defined. We also find
take any point, say x = c inside the limits of integration, at which
f(x) < 0 when 1 <x <2. We write g(x) = ~ f (x) so that g(x) >O when l<x< 2. Therefore,
that
we can write

ae ; dx | te dx
ee / G@-—-De-x) J, @-DC-2)
(c) the integrand in J; has infinite discontinuity at x = 2.
Hence, we can write

| in. dx . dx
: dx
2 Se42 = jim
4M do (x—-2-2)
x-—1f1x
(x— 4) &ua 1463 (x -1)(2-2x)
9 x=

- ie dx , 2 dx
= om 3 (z—1)(2—x) * 4-20 _ (x — 1)(x - 2)

| ( E>
7
|
; _ 4; : fe}
€, +1)
= lim, in(<L+": —In 2 jim, in|7 4
_t In | 2 i

“anfe(252)—»(522) aml) -m(

'

Since the limits do not exist, the improper

: b= Ey

integral diverges.
el! : I AE ( Es ]

oper
. We could have concluded that the impr
Note that the improper integral /; diverges
ergence/divergence of /2 and I.
integral diverges without discussing the conv
b dx
improper integra! | fea p> 0.
Example 1.47 Discuss the convergence of the
at x = a. We write
Solution The integrand has infinite discontinuity

de | l
b
y (b- a)? er"
eW Jase (x-a)? 1 -p £0)
« tt- a)?

_ fit -py(b-a)?"], ifp<!

7 | O°, ifp> |.
1.58 Engineering Mathematics

For
p = 1, weget
b | b
ax = lim dx| _ jim in|?
r By a)=e.
gre ee) x-8 E—0

Therefore, the improper integral converges for p < 1 and diverges for p 2 |
n/2
Example 1.48 Show that the improper integral | tan x dx is divergent.
-7/2

Solution The integrand has infinite discontinuity at x = + 7/2. We write

ai2 (2/2)—€
| tan
x dx = lim
m | tanx dx + im | tan
x dx
¢ —1/2
é0
(m/2)+€ c

(2/2)-€
= lim [-In ( cos ¥) | erase + eer [-In (cos x),

—7 Incos- - + e)~ In [cos 3)

7 lim i co (+- ) — In |cos co}

Since the limits do not exist, the improper integral diverges.

Note that if we write

ml2 (1/2)-E
tan x dx = lim tan x dx
-n/2 £0 J_(n/2)+¢
m/2
we get | tan x dx = 0, which is not the correct solution.
—ml2

Example 1.49 Discuss the convergence of the following improper integ

rals
2 Vx
x dx.
(i). | x~ (ii) [-ane
n/2

Solution
(i) We have f(x) = (¥x/In x) > 0, 1<x<2. The pointx= 1 isthe only point of
infinite discontinuity:
Let g(x)= I/(x In x). Then, we have

f(x) fl+h)_,, ich

Bae g(x) = g(1 +h) = jn E (1 + q (1 +h) In (1 + A)]
ios \V2
= lim (1 + h)"* = 1.
 Functions of aReal Variable 1.59

(x)dx converge OF diverge together.

Therefore, both the integrals [ f(x)dx and [ g
:

| : dx | : dx = 5
Now,
‘ | gixidx=
| = | lox = lim _ vie
lim =| lim (InIn (Inx))y,,
l as.

9 ~.
= lim [In (In 2) —-In(In(1 + €))J
i—rJ

4.
Since le( x )dx isdivergent, the given integral [9 x)dx isalsodivergent byComparison Test
| ]

i ape
x f
| . ' |
(ii) Wehave f(x)= ee = —— a2 gt. since sin x/x is bounded and (sin x/x) < l,
AY ST * Vx J x

O<x< n/2. Let g(x) = Ux. We have f(x) < g(x), 0<x< 2/2.
Now, g(x) has a pomt of discontinuity at x = 0. Hence,
arz mf2 dx Fe
| g (x)dx= | =Vx = iim| — = lim [|/2a - 2Ve] = 2.
e e—)

gI2 xi2
Since | g{x)dx is convergent, the improper integral | f(x)dx is also convergent by
6 0
Comparison Test 3 (i).
xi2 .
cos” x
Example 1.50 Show that the improper integral dx converges when n < 1.
0
Solution We have f(x) = —=ate +, 0 <x < 7/2. The point x = 0 is the point of infinite
x x
discontinuity of f(x). Let g(x) = 1/x". Then f(x) < g(x).
exi2 | af2 dx
Since the integral [ g(xj)dx= { 7-7 is convergent for n < | (see Example 1.47), the given
9 r
integral is also convergent for n < 1 by Comparison Test 3(i).

1.5.3 Absolute Convergence of Improper Integrals

In the previous sections, we had assumed that f (x) is of the same sign throughout the interval of
integration. Now, assume thatf (x) changes sign within the interval of integration. In this case, we
consider absloute convergence of the improper integral.
b
Absolute convergence The improper integral | f (x)dx is said to be absolutely convergent-if

eb
| | f(x) |dx is convergent.
 1.60 Engineering Mathematics

b
Theorem 1.15 An absolutely convergent improper integral is convergent, that is if |f(x) |dx
b a

converges, then | f(x)dx converges.

ra

Since, |f |is always positive within the interval of integration, all the comparison tests can be
used
to discuss the absolute convergence of the given improper integral.

Example 1.51 Show that the improper integral |" —;

sin (I/x) 4x converg
: es absolute
EY ly for SOLD < I,
0 xX

Solution The integrand changes sign within the interval of integration. Ne we consider the
absolute convergence of the given integral. The function f (x) = sin (1/x)/x? has a point of infinite
discontinuity at x = 0. We have

lf(x)|=
sin (1/ x)
< 2%
xP xP

I
Since { dx converges for p < 1, the given improper integral converges
Oo x absolutely for p < |.

Example 1.52 Show that the improper integral | ~Sin x dx

converges.
ii RE
Solution The integrand f(x) changes sign
within the interval of integration. Hence,
absolute convergence of the given integral. we consider the
We have

II |= | sinx
in.
gy < f- -sinx
|4,
c a b
= lim | nx |dx + lim Sa i
Now,

/, = lim |“ sin x
at dx=< lim
Sa a 1+x?
:
a——co 3 1 + x2
= = l —| ide
jim [tan Cc — tan a] = tan ete,

an b
Ia = jim| |ox Id <xlim a
pe Je |1 + x? | “
pe Jo 1+x2 = pm [tan 6 — tan ‘c| ~ 5 —tan lec.
Hence, |I|<l+h<x
Therefore, the given improper integral
converges
1.5.4 Beta and Gamma Functions

Beta and Gamma functions are

names a: improper integrals which are
and enginee ring applicat ions, These functions are used commonly encoun ; —
in e : tered in many scienc
valuating many defi
Gamma function nite integrals.

Consider the improper integr

al [(a) = |
, x*le-Xdx, a>.
(1.16,
76)
 Functions of a Real Variable 1.6]

c oo

We write the integral as J[(a@) = | x%1e*dx + | xe Fdx= I) +1p,0<c<@.

0 c

rvs rn J, is an improper integral of the second kind as the integrand has a point of discontinuity
om j whenever 0 < a< 1. For a2 1, it is a proper integral. The integral /; is an improper integral
of the first kind as its upper limit is infinite. We consider the two integrals separately.

Convergence at x = 0, 0 < a < 1, of the first integral /,

ii .. ——<—=1
= lim ee?
In the integral /,, let
f(x) = x%-le* and g(x) = x*!. Now, lim
x0t B(x x30 Xx

| c Cc
dx
Since | g(x)dx= | na converges when 1 — @ < 1, or @ > O, the improper integral /, is
0
convergent for all a > O.

Convergence at ©, of the second integral /,

Without loss of generality, let c > 1. Otherwise, the integral can be written as the sum of two integrals
with the intervals (c, 1), (1, ©). The first integral is a proper integral.
Let n be a positive integer such that n > a— 1, a> 0. Then,
x*'!<x" and le * 2 z"e".. 1 x
xf eee.
a—-l<n,
oo »b
= lim | xe" ax
Therefore, | x° 9" * dex | x*e* dx b— poo
c c

eG qb
= lim le {polynomial of degree n in x, F (x)}]_

Pn(0)
= b—limoo [e~® P,(b) — e~*Py(c)| = — e*

since lim [b*/e] = 0 for fixed k.

erges for a > 0.
The limit exists and the integral / conv ral is
the give n impr oper inte gral (Eq. (1.76)) converges when & > 0. This improper integ
Hence
denoted by I'(@). Therefore,
called the Gamma function and is
(1.77)
T(a)= | x*'e*dx,a>0.
0

a functions
Some identities of Gamm
(1.78)
1 rd= | etdx = 1.
0
(1.79)
2. T(a+ 1) = al (@).
s, we get
Integrating Eq. (1.76) by part
dx = al (a).
- jeter i ¥ a| x! e-*
r(at+ l= | x%e *dx= | 0
0
positive integer m.
3. Pim + 1) = m!, for any
 | ms
ics
1.62 Engineering Mathemat
= ==m
m (m—I1)..) -. = m!im
1) = EC) = mm — DT (m= 1)
ha veFe| m +
e eha
sv

4. ri/2)= V7:
We have
x
" du. (set= 4)
ee 1qx=2|, on
pot
T (1/2) = J x

We write
oy = fe ee
| 2 os
| e7! | —-4 ) I é
(3) = 2| en | E
0 0
lv = =rr drar dO n and
sin 0, we obtain dudv
tes u = r cos 6, v =r
Changing to polar coordina
2 a
oo

re’ dr= ale”

oo
Jo =7%-
r é =4 {" | re’ dr d@ = 2n|
0
- @=0 r=0

Hence, (1/2) = vz.

double integrals and change of variables.)
(In Chapter 2, we shall discuss evaluation of
(1.82)
5. T(-1/2)=—-2V a.
get
We have ['(a) = [I (a + 1)]/a. Substituting a = -1/2, we
r 5)= cay
~ 2} (-1/2) ~

Beta function
Consider the improper integral

i= | x™1(1 —x)"!dx,0<m<1,0<n< 1. (1.83)

0
Note that J is a proper integral for m 2 1 and n 2 1. The improper integral has points of infinite
discontinuity at (i) x = 0, when m < 1 and (ii) x = 1, when n < 1. When m < 1 and n < 1, we take any
number, say c between 0 and | and write the improper integral as
c ]
[= [
, x™!1(1 — x)" dx+ | x” (1 —x)""! dr = I,) +1,:
c

i=
c
m=-1¢]
1
_ »\n-l | gm ‘as
where 1 ) xe" (L-x)""dx and I, | x™" (1 — x)! dy.
c

q is an improper integral, since x = 0 is a point of infini

te discontinuity, while J, is an improper
integral, since x = | is a point of infinite discontinuity. We consider these
two sitegrils separately
Convergence at x = 0, 0 < m< 1, of the integral /,
In the integral J), let f (x) = x”"'(1 — xy"! and g(x) = "1.
 Functions of a Real Variable 1.63

Now, lim f(x) _ lin xm" -x)"t a

x30t 2(X) — x-40+ ge

‘ie g(x)dx
and | = |cd — is convergent only when 1 -m<1,or m>0O.
0 0 x
Therefore, the improper integral /, converges when m > 0.

Convergence at x = 1, 0 <n <1, of the integral /,

In the integral /5, let f(x) = x—'(1 — x)""! and eG) = (1x).

Now, lim f(x) = lim xm =x)" 1

x3i- §(%) xe (1 - a

1 I
and g(x)dx = | _ ax converges when 1|—n<l,orn>0O.
c - (Lean)
Therefore, the improper integral J, converges when n > 0. Combining the two results, we deduce that
the given improper integral (Eg. (1.83)) converges when m > 0 and n > O. This improper integral is
called the Beta function and is denoted by B(m, n). Therefore,
l
B(m, n) = | x™1(] —x)""dx, m>0,n>0. (1.84)
J0

Some identities of Beta functions

1. B(m, n) = B(n, m) (1.85)

(substitute x = 1 — ¢ in Eq. (1.84) and simplify).
mid i2
2. B(m,n)=2 | sin*”—! (@) cos*""! (9) d@ = 2 | sin*""! (8) cos?”"! (8) dé. (1.86)
0 0

(substitute x = sin? @ in Eq. (1.84) and simplify).

3.. B(m,n) [ ed ‘
men. = |. ee
Came ee (1.87)
:

(substitute x = t/(1 + #) in Eq. (1.84) and simplify).

4. B(m,n) = (mm) P(r

(1.88)
We can prove this result using double integrals and change of variables. We have

(mm) = | fet den | y2m-l p-u? du, (setx = u’)

0 0
og

T(n) -| xMle-x dx=2 | p2n-l p-v? dv, (set x = v2)

y() 0
 1.64 Engineering Mathematics

T(m) (nn) = 4 | | y2mnl py2n-l p-(u? +0"du dv.

Changing to polar coordinates, u = r cos 6, v = r sin 8, we get

Tt pe — ee
C(m)T(n) = 4 | | cos*”""! (@) sin-” '(@) p2mt2n-l oT" dr d@
6=0 #r=0

Z n/2
= 4 I p2mt2n-l y-r* tr| cos”! (8) nO)
0 0

= 2 B(m, n) [ p2m+2n-le-r’ dr (using Eq. (1.86))

0
We also have

(m+n) = [ Fae
wy Gee: | pomten-lo-r" dr. (set x = ah I
0 0

Combining the two results, we obtain

['(m)T(n) = B(m, n)U (m+n), Pom) G

or B(m,n)= ee

5S. B(m, n) = B (m+ 1,n) + Bim n+ 1),

We have
ni2
/2 rr
B(m+1,n)=2 | sin*”*! (9) cos"! (9)d@
= 2 | sin”! () sin? cos2""! (@)d0
0 0
mi2
=? [ sin’ (8) cos?! (9)(1 — cos*@)d@
0
mi2
m/2
=2 [ sin"(9)cos?" (9)49—
0 9 | sin*”"! (@) cos?"*! (9)dé
0
ce B(m, n) = B(m, n+
1)

Therefore, (m,n) = B(m + 1,n) + Bom, n+

1).
-]
Example le 1.53 Given th x
Given that I l+x
=2
sinpm’ show that (pT
np
(1 ~P)= ox
_— . — g |
Solution Let wee Solvin for x, we get x =_ =sy and dx = . _ dy:

| a i
Then, 1= [ ax |»? (1 - y) NPBA (pra -
~ gh EE origrit gi
Hence, the result.
 Functions of a Real Variable | 65
Example 1.54 Evaluate the following improper integrals

(1) | Vx en? dx, (11) { e-* dx

0 0
in terms of Gamma functions.

Solution
(1) Substitute x = Jt. We get dx = dt/(2Vt) and

te [Vee Pace dP
I 2 J, e terad[" 1oe-teare 2 J, ! edt a Lr(3) a}

(11) Substitute x = 2'”. We get dx = 41? dt and

| - wail | | al 4 : | cs] 7 7 | a

/ — =| dxti=-=a. 2/3 e ‘dt

, 7a
=_=_— (1/3)-1 e ‘dt
P ==>5 | 5
E )

Example 1.55 Using Beta and Gamma functions, evaluate the integral

[= | (1 — x*)" dx, where n is a positive integer.

-|
]

Solution We have fs | (l+x)"(1 —x)" dx

-|

Let | + x = 2rf. Then, dx = 2d? and 1 -— x = 2(1 — r). We obtain

|

I= 278") | (1 —1)"dr=2-"*! Bin + ln +1)

0

er | 2n+l Cin+ 1) PT(n+ 1) _ 2%**! (nt)?

(2n + 2) ~ (2n+ 1)! °
_— <a!

Example 1.56 Express | x™(1 —x?)"dx in terms of Beta function and hence evaluate the
0

integral | x%2¢1 ~ Vx)? dx.

0

Solution Let x? = y. Then px” ~ldx = dy. We obtain

l |
io | x™ (1 = zr y dx= E | yi Ptr ey ed y)" dy

0 P Jo
e | f |

= 2P [Jo yorwora -yyrdy=4p(

p\ "=# net),
1.66 Engineering Mathematics

/2 dx _ iven
with the giv integral, we fing
ring the integral en in that
Now, comparing 8 |
: x? (1- vx)
m = 3/2, p = 1/2 and n = 1/2. Therefore,

I 3 21 (5) (3/2)
| PP(1— Jx)'"dx= 2AI5.4 ~
0
T1372)
I3)_U 9.7.5.3
3
>|- 10395 (5).
— rs) =at=24,0( >]= 3-3-3 2 (3 32 (2
2(24)(32)T(3/2) _= 1536 12
Hence, f=
= “703951 (3/2) +-10395 3465
Example 1.57 Using Beta and Gamma functions, show that for any positive integer m

ae (2m — 2)(2m-4)...2
) ) sin" (8)d0= (2m —1)(2m—3)...3°
ae,
(11) j
[Ossin? am ane (2m - 1)(2m-3)...17
=a >

Solution From Ec. (1.86), we obtain

] aha l l m/2
p{m >|=2 [ sin*”-!(@)d@ and p{m + 5, | = 2 |
2 P . ; sin?” (0)d@.

ee | ml2 (2-1) _1 L\ _ [(m)T (1/2)

(1) I= | Sin (@) d@ = 5b{m. 5)~ 20 (m + 1/2)’

ode
8).els
We have I'(m) = (m — 1)!, and

ry fy 3

“pelam—Heam—3)..
.3-yr(2)
Therefore,

- (mi— 112" T(1/2) m-I _

(2m — 1)\Qm—3)...3-]
_ (2m ~2)(2m~4)...4.2
~ (2m — 1) (2m =3)...3- 7°
mie
(ii) =| sin" (9)d0 = 7 B[m+ | =i (e+ 2) rae
0 2 73 |°S>.
20°
(m + 1)
 Fuacnons of aReal Varatie =1.67

—_ =

|fee
(2mm — 1)(2m
-—ps 3)...3 4 (2m — 1) (2m - >) ws r)
Se (yx)* =
[mim — 1)...2-1
=
2™
=
2(m!) | 2” :

_ Qm- 1) Qm-3)...3-laz
(2m)(2m—2)...4-2 2

Example 1.58 Evaluate | 2-9" dx using the Gamma function.

0
Solution We write

f= | 5-91 dx= | eer a2 dx

) 0

= W722.
Substitute 9x* In 2 =y. Then, x= Ee wai des 2 2 .
3VIn2 6 In 2

Therefore. j=— | yt edly = 1_ | yr te"? dy

6ViIn2 Jo 6vIn2 Jo

_ rda/2 _1 | 9

6VIn2 6 \ln2
Example 1.59 Show that
a
P(Qn)==— Pines (nm) (1.89)

ri\—{3 ~
and rl) r(z) = V2. (1.90)

Solution From Eg. (1.88), we have

I (m)P(n) meg ‘
a = Bim, n)=2 J sin(8) cos 7! (0) 8. (1.91)

Setting
m =n, we get
[T(n)}? iJ a .

T(r) Bin, n) =2 sin *(8) cos*' (6)6

: #0
1 aj2

“str | sin (20) de.

Substituting, 20 = 5 — , we get d@= -+ dg. Hence, we obtain

 1.68 Engineering Mathematics

[T(n)}? _ -l {- cos2""! (6)do

P(2n) 21 Jon
fs do=— i cos"! (6)d@
i 9 2n-1 J COS dad = qin-l 0 (1.92)

since cos @ is an even function.

Setting m = 1/2 in Eq. (1.91), we obtain
a/2
T(n)tP(d/2)
le) -3 | cos“! (8)dé. (1.93)
IT (n + 1/2) 0

Comparing Eqs. (1.92) and (1.93), we have

(Tinh 4 Pn
T(2n) 22™1 | T'(n + 1/2)

| _ 2 2n-| ” : es _
or [(2n) = ae ['(n) T(n + 1/2), | since r|5 | Va|

which is the required result.

Setting n = 1/4 in Eq. (1. 89), we obtain

1 7 4-2 S (2
(3) = = rary
or r
a)! (g) = 92.
7)r(>)- V2

a/2
Example 1.60 Show that [
. Vtanx = 4.
2
Solution We have
mi2 mlz
I= | Vianxdx = | sin’”* x cos-'/2
0 0
x dx = 1, 31
2 \4'4
_ 1 T4104) V2 7
7 (1)
|
== = Va (using Eq. (1.90).

1.5.5 Improper Integrals Involving a Parameter

Often, we come across integrals of the form

2b(a)
O(a) = f(x, @) dx (1.94)
"a(a)

where @ is a parameter and the integrand f is such

that the inte
methods. We can evaluate some of these integrals by differen gral cannot be evaluated
evalua by standard
tiating the integral with respect to

.
 Functions of a Real Variable |

ana then
parameter, that 1s first obtain @ ‘(c), evaluate the integral (that is integrate with respect © 2)
we
integrate @’(4) with respect to & Note that f is a function of two variables x and &@ When
differentiate f with respect to @, we treat x as 4 constant and denote the derivative 2s Of 9a |pau a)
derivative offwith respect to @ Chapter 2 discusses partial derivatives in detail). We assume thai/.
fla, a(a) and b(@) are continuous functions of @.
We now present the formula which gives the derivative of Of).
Theorem 1.16 (Leibniz formula) If a(c), b(@). f (x. @) and df/og are continuous functions of a.
then

do ("” of db da a
AG
— = |. —
a /( x, A)dx +
4 fb. OT
jee tae fla, a)er i 1.95)

Proof Let Ac be an increment in «@ and Aa. Ab be the corresponding increments im a and 5. We

have
b+ Ab
Aob= (4+ Ad) -—O(a)= | f(x.a@+ Agjdx- | f(x. ajdx
a+ 0

=
dl u o* oo | oo as

=| flxa+ dade | flz.a+ dade + | flx.a+ Sajdx - | f(x, ajax

i+Aa :

or
Ap [7 4
An re |. ee f(x,oes
cA a+ AGjdx +
f l if ver » ,a+
| Li——I[flx — ajlidx
Aa) —flx,
f

+|oe i
Ag sf6% 4+ 4ajdx 96

Using the mean value theorem of integrals

5S
| f(x)dx
= (x, - x6) (6),49 < $<5;
xy

Aa). a< ¢, <a+ Aa

f(x, @+ Aajdx = -Aaf(,,a+ (1.97)
. -

we get |
a+Aa

b+ Ab |
and | f(x, a@+ Aajdx = Abf(¢,,a+ Aa), b< €- < b+ Ab. (1.98)
b
Using the Lagrange mean value theorem, we get
d
f(x, a+ 4a) - f(x, a) = Aa SFtr.&)a<b:<a+ sa (1.99)

We note that

jim, 51 = jim, 62 = band jim, £5 = « 1100)

Taking limits as Aa — 0 on both sides of Eq. (1.96) and using the results in Eas. (1.97) to (1.100).
we obtain
do f° of - db da
— = —(x,@j)dx — — ;—.
da J da” i i da f(a, @) da
 1.70 Engineering Mathematics

Remark 8
(1.95)
constants, then we obtain from Eq.
(a) If the limits a(a@) and b(a@) are
do. f SL (x, a)dx. (L190)

then we obtain from Eq. (1-95)

(b) If the integrand fis independent of a,

4 - 7H) 8 - fa Sy lime
certain types of improper integrals.
(c) Leibniz formula is often used to evaluate

Example 1.61 Evaluate the integral |

~ e@ ™ sinx 4, g > and deduce that
0
co ” sin ax
sin Xx 1 = : 4 IC Q.

e =" Sip:2" . (1.103)

Solution Let @(q@) = | —— ax.
4 :

The limits of integration are independent of the parameter a. We obtain

aa
do
ee a 0 ee e* sin
ee x
lay
= —
i xe 64
—$—$$§$§$<$§—
sin x
dy = -—
-
e-"*
:
sin Fax.
da” J, 0a x # () x 0

Using the result [e« sinx dx = — = am (@ sin x + cos x), we obtain

i & ax 5

do —ax =
a gr (asin x + 00s x)| = |
l+a ‘ 1+a?
Integrating with respect to @, we get

@ (&) =— tan” a@ + c, where c is the constant of integ

ration.
From Eq. (1.103), we get the condition @(cc)
= 0. Hence,
Mo=0=—tano4+e¢ of c=.

Therefore, a | "en ex sin

9 (a) ) ae te = 5 —tan'q.

(i) Setting @ = 0, we obtain [ sin x soil

5 x an 2" (1.104)
(ii) Substituting x = ay on the
left hand side of Eq. (1.104),
we obtain

| ay dx= | BOY a
o * J y Zi’
 Functions ofaReal Variable 1.71

, evaluate the integral

| o-*’ cos (20x) dx.
Example 1.62 Using the result | ras dx = ae 0
0

(1.105)
Solution Let @(@) = | e-** cos (2ax)dx.
0
the parameter @. Hence,
The limits of integration are independent of

ae [.<s [e7* cos (2ax)|dx = | (- 2x)e* sin (2ax)dx

da J, da 0

o.
= [e7* sin(2 ax)]o —2a | ox” cos (2a@x)dx = —2a
0

| d , 7 nik
equatio n < + 2a = 0, we obtain 9(@) =ce
Integrating the differential

From Eg. (1.105), we get the condition ¢(0) = | e7*” dx = ae,

0

Using this condition, we obtain 9 (Q) = ae ='C.

Jn a? |
oo

Therefore, @(a@) = | e-** cos (2ax)dx = ie

2
P

co =
Evaluate the integral | LD) dx. a>Oanda # l.
Example 1.63
9 x(1+x*)

oe -1
Solution Let @(a)= | Ca (1.106)
0 x(l -x*)
We have

dg = “ oO tan™ (ax) ax= [.en

da 0 da} x(1+ x?) 9 (+ x*)(1 + a* x?)

—_ [l= -Els
a? —1Jy La?x?+1 14x? |"

= a?
]
[{atan“'(ax)}G — {tan(x))§]
ax)}f ~ “iyon =2|4GS. 1| |sea
___

Integrating with respect to a, we obtain

d(a)=>In(a+ l) +e.

From Eq. (1.106), we get the condition (0) = 0. Using this condition, we obtain @(0) = 0:
= ¢.
1.72 Engineering Mathematics

oo ta -| | :

0(a)= | =e dx= 4 In(a + 1).

Therefore, Jo (1 - )

1.5.6 Error Functions

certain types of partial differentj.)
Error functions arise in the theory of probability and solution of
equations (see section 8.7).
probability distributio,
Let us first consider the following function that arises in defining the normal
(the case when mean = p = 0 and variance = 0” = 1)
l oe = .
(1.107)
f(x) = Jon é

by
tion. The bell shaped normal curve defined
This function is also called the Gaussian func
r the curve and above the x-axis 1s given by
Eq. (1.107) is given in Fig. 1.20. The area unde

i |© et 2dx . (1.1»1U008)
if= [: f(x)dx
| =
; = |

f(x)

Fig. 1.20 Normal curve.

Setting u = x//2, we get

zl.
l “ 2
f=—— e" du.

It was shown in equation (1.81), that

pice. t= | f(x)dx= = [_e~” du = 7 af ~ i

that is, the total area under the normal curve is 1. Since the area is symmetric about the y-axis, We ge
0 on

| fix)dx= | f(x)dx= 1%
in ‘()

The area under the curve, from — ° to any point z, is given by

 1.73
Real Variable
Functions of 4

(1.109)
: ed.
oa) =e |
the normal
is cal l ed the di st ri bution function of
= 1/2. The function $(Z) = — dy and
Hence, by definition ¢(0) tin g x =— Y in Eq . (1.109), we get dx
variance 1. Set
distribution with mean 0 and
] | eddy

Jon
oY dy =
9(Z) = “Ts { —<

“del
—<

en 2dy — ] [_«av 2 dy
jan
= 1 — o(-z)
(1.110)

or o(—z) = 1 — $(z)- the

Z. Further, the area under
ribu tion func tion are tabu lated for various values of
Values of the dist
.
curve from x = 0 to x = z is given by
0
: dx |= o(z) - > (1.111)
dx = 2 [ en 2x — | eX
I(z) = = [ e-* oo —o
4{ 2% 0 a 270

or o(z) = “ + 1(Z).

Error function erf (x)

integral function. It is defined by
The error function is also called the error
(1.112)
erf (x) = x), en" dt.
| | 1 du, and
Let 2 =u. Then, dt = y du= oi;
9 —_ 1 —

;
i.” 2 (1.113)
erf (x) =cope| ua" du,

we obtain
This is another form of the error function. Using this definition,

erf(00)= |, s\-+
r(3
=TaN
witertdy ——
Fe We) | (1.114)

Let #2 = u2/2 in Eq. (1.112). Then,

d
2tdt=udu, dt= esdu = = and

V2x
—u/2 du _ af) 2x wate
erf (x) = = e =z) edu. (1.115)
 Usrme Ec {]- 11:

35
ing 2d x= =
———— [ e* Fdx= O(V2 x) — 5-
¥— 0

Theretore.
:
erf (x)
i>
= 2i(w2 x)= -Otv=-F)
> — 341472 x) — 1. (1.116)
using this relation.
Hence. the error function can be evaluated
|
Complementary error function <7c (=)
Usme the deFIMO n of the error function given
in Eqs. (1-113) and (1.114). we write

A
ae 2
=

-e *du —_ | yw 66h €E + du

| uo -e *du =
l un
.
~~ Kk 6 vi I

(1.117)

where we define
(1.118)
erfc(
; x)= ==
Va Je u~Y*e-*du.

This function erfc (x) is callea the =mplementary error function

Using Eqs. (1.117), (1-112) and (1.114), we can write

; erfce(xy=1—erf(x)=1->= e dt
wi df

2 fF" 2, 39 £ 12 2 ft
oo e dt -—= Par= 2 | eo#2 dt. |
(1.119)
VK j wT Jo V1 Vs

Equations (1.112) and (1.119) are the commonly used definitions of error function and complementary
error function respectively. The graphs of ef (x) and erfc(x) are given in Fig. 1.21.

Some properties of error functions

1. erf (—x) = — erf (x). (1.120)

Sse ee Fe Se ee ee eee ce
fe 2S eS Se -=—=—_—=

x
0

Fig. 1.21. Error function and complementary error function.

 Functions of aReal Variable | 75

Using the definition given in Iq. (1.112), we get

erf ( *)= | e' dt= + | e“ (~du) (settf=

ing—u)

= — erf (x).

2. erfc (-x) = | + erf (x) = 2 — erfe (x). (1.121)

Using equation (1.117), we get

erfc (-x) = 1 — erf (-x) = | + erf (x)

= 1+ (1 - erfc (x)] = 2 - erfe (x).

3. Derivative of error function We have

7, bert (ax) = 7" (1 .d2Z)

From the definition, we have

5 (£0 '

erf (a n= | er at. (1.123)

Vm Jo
Eq. (1.94)
Consider x as a parameter. Comparing Eq. (1.123) with
b(a)
o(a) = | f(x, @)da, where @ is a parameter (1.124)
a( ce)

we get f(t, x)= qeen! , B(x) = ax, a(x) = 0, G(x) = erf (Ax).
Using Eq. (1.95)
b( a) da
ae. OS x, ada + f(b, a) 2 - f(a, a) (1.129)
da a(a@)

we obtain

(ax)Ijante (| le” Dat

)dt + f(Ax, x) £
A (ax)
(ax) -
rr a

Lert Itt I ae -0

26 po?
NE
Integral of error function We have
A,

[er
i
rf (Ax)dx = terf terf(at)+ =e-[e
rdx _= 7 | ~ce*y¢ - | | (1.126)

s, we obtain
Inteprating the left hand side by part
1.76 Engineering Mathematics
f d
xa [erf (ax)]dx
f
| | - erf(ax)dx =([xerf(ax)], - I
'
0

2a i =x
= rerf(at)-e | 4 dx

*) Hence,
dul(2a”).
dx = ; du or x dx = dul(2a
using Eq. (1.122). Let a?x* = wu. Then, 9a2x
2,2
2a ] ae ee
7|sss
sa ]f e- “du
jerf (ax) dx =terf (at) - (

| (a
ae
oeie leen \,
abe

= terf (at) + ain” L_je 1).

Exercise 1.4
In problems 1 to 25, discuss the convergence or divergence of the given improper integral. Find its value if
exists.

"i Inx dx, 3. | +&

” I 4+x° o [*
x2
x? 4.2%"

~
4. = xdx |
I
I ode os 6. ndx, a>,
“4x gibx
re" dx, a> 0,

7. { xe dx, . [| —#_
9,
x? S
S
--.;

= |0 — nu. | 2° dx, 0 x+4 dx,

| toy OY SP <I, 12. [<r

"dx ' 14
13. [ ™
o we tl [*
woe
16 we
: dx
. ‘ cos x’ 17. [ In x dx, "

a maj
' dx ;

_ xt
” J * 20. |; sg
rar
_ ty

R/2 sin”

‘2 |
Ig dx
x 23, [
3 x(I
nx2_@
_sinx) 24. SBE ae
 kunctions of a Real Variable 1.77

25, | x! ds. jq20, WnNjyq<0

) t t x!

In problems 27 to 40, evaluate the intevrals using the Beta and Gamma functions

ni} ai?
| dx
2h, 27, | sin’Ocos* Odd 25. | sin’@ cos°@dé@.
) AAT

[
i}

|
| dx
24. Con”
O dO, m inleper ). | xVa® - x? dx 31s
f) () ‘I -x? |

tf

33. 4
x
32.
9 J -Inx |

35.
(o x? ey dx, kx 0
36.
0
Ax
l +x"
37. | BX,
fj
<*
a
a>.
|

3H. [ ue a (,8§ > Ok oO. | t*e"

~,2
dt. | gh?
4 ..o2
o>” dx.
= ( f)

(
Establish the following results.
mang+| !
Al. x™(p4 -xtyt de = (
P Bln } m, n, p,q are positive constants
q q

42,
j (x—a)y™'! (b= x)"'dx =(b-ay”"! Bim,n), m,n,a, b are positive constants.

fee4(5
min ]

————_—_—_———— |, a, b, m,n are positive constants.

43, E sin ‘aaa me

JM My ae ee iy
| enh n! é
m_~x?y" = ———————,, nis a positive integer.
44. = B
\|

[=x" dx =m"! B(m,n-+ 1), m,n are positive constants.

0

|x" ¢@
| r(m+ 1!} m,n, & are positive constants.
"dx = nam bin | oA

47. [.e™ (Ll-e" ‘)"dx

= Bh(m,n+ 1), m,n, are positive constants.
>

(=1)" a!
44. em
(Inx)"” n dx mae eee yar m > | and n is a positive integer.

pel
9
ey x O<pe< [eventha f (ll dx =
49, sin pi |
~ or Tay ele (pRTD | 70 1+.
1.78 Engineering Mathematics

50. For large n, n! = .{2an n" e-" (Stirling’s formula).

lg"-l Gosh |
cS a oe
od j (a+ bx)”"*" ~ a" (a+b)” B(m,n)

Using the concept of differentiation of integrals (assuming that the differentiation is valid) evaluate the following
integrals:
re e~ax _ ex a x? ’
52. dx, a>0,b>0. 53. [ Sqgias a> b>-t.
0 x 0 log x

| n/2a

34. |x" (log x)* dx, n any integer > — 1. 55. | asin ox dx.
0 0
3 oo : a3
-
56. | xe"* dx, where | e° dx 7, 57. | cot! (x/ar? )dx.
— oo —oo 0

sg, [9» ——©0Sx___ ay given tat [

»
__
at
ae
bcosx —
at be
gh bo0,
(at+bcosx)?

” ME
Ix) qx a> 0, given that | e-* dx= =
59. | e-*’-(@?
0 0 :
ri ao dx 7 a

60. | log (1 _ a? sin 2 x) dx, |x | < | 61. | Ge +1)" n any positive integer.
0 0

d 2a ey.
z
62. Show that = [erfc(ax)] = a =e

9 x3 5 7
63. Show that erf (x) = fols-3) + 3a - FG Ps |

f
_ { ry:
64. Show OW that
tna ax)dx=terfc
|exte x) rf(at)
( ) —- mm € — 1}.]

65. Show that | e-M2atgp— <2 e” [1 - erf(a)].

0)

66. The relation | e-*” cos (2bx)dx = az e-”” is given (see Example 13.40). Deduce the result for
0

| et *x? cos (px)dx. Integrate this result with respect p, taken as a parameter, from p = 0 to p=5 and
0
a sete sin (sx) _ Ss-
show that |. ( x Jax- 2 ot (= }
 Oy [en UR 8 “ Chapter2

Functions of Several Real Variables

rmA | <

2.1 Introduction Yo ' =<) ae

Py i = ; , nN.

In Chapter 1 we studied the calculus of functions of a single real variable defined by y = f(x). In this
charter we shall extend the concepts of functions of one variable to functions of two or more
variables.
If to each point (x, y) of a certain part of the x-y plane, xe IR, ye IR or(x,y)€ IR xX R= R’,
there corresponds a real value z according to some rule f(x, y), then f (x, y) is called a real valued
function of two variables x and y and is written as

z=f(x,y),xe R,ye Rror(x, ye R*,ze R. (2.1)

We call x, y as the independent variables and z as the dependent variable.

In general, we define a real valued function of n variables as

ih = f (%1, XQ, + 2 +9 Xn) (x1, XQ, s+ 5 Xn) c IR", ac JR (2.2)

where x), X2, - - -» Xn are the n independent variables and z is the dependent variable. The point
ieee. --;X,) 1S called an n-tuple and lies in an n-dimensional space. In this case, the function f maps
IR" into IR. - A+ @ |
The function as defined by Eq. (2.2) 1s called an explicit function, whereas a function defined by
Sikes. -; X,) = 018 called an implicit function.
We shall discuss the calculus of the functions of two variables in detail and then generalize to the
case of several variables.

2.2 Functions of Two Variables

Consider the function of two variables
oS 7%, 3): (2.3)
setfuncti
ofThethe
(x, y) in the x-y plane for which f(x, y) is defined is called the domain of definition
of points
on and is denoted by D. This domain may be the entire x-y plane or a part of the x-y plane.

The collection of the corresponding values ofz is called the range of the function. The following are
some examples
 2.2 Engineering Mathematics
> m “ P _.

> >. > 2 “2 1.

|
12 - = is real. Therefore, we have 1 — x~ — y~ => 0, or x“ + y~ <1, that ts, the domain
is the region x° + y- < 1. The range is the set of all real, positive numbers.
The domain is the set of all points (x, y) such that x* — y’ #0, that is y#+-x. The
range is IR.

z= log (x+y): The domain is the set of all points (x, y) such that x + y > OQ. The range is IR.

The domain of a function and its natural domain can be different. For example, we have
f (x, y) = area of a triangle = x y/2
where x and y are respectively the base and the altitude of the triangle. The domain ts x > 0, y > 0,
whereas the natural domain of the function is the entire x-y plane.
Consider the rectangular coordinate system Oxyz ( Fig. 2.1).

— — _—_-—
=

O(0, 0, 0)

(a) Y)
Fig. 2.1. Function of two variables.
At each point P (Xp, Yo; 0) in the x-y plane, constr
uct a perpendicul ar to the x
Z - it - = PO = Zo =f (Xo; Yo). This gives a point O(x, -y plane. Take a point
:
esYor Zo), OF Q(X, yo, F (Xp, Yo)) in space.
€ 10Cus Of all such points (x, y, Z) Satisfying z=f(x
re ) cn
ae peer (<©=J, y) is called a surface.
For ‘ss
. the ey z=x+y’, (x,y) € IR* is the paraboloid
perpen cular to the x-y plane intersects the surface z=f (x, ofy)revolu tion as Piao tae = on
at exactly i. elas cok
at no point if (x, y) €D. ; “acUy one point if (x, y) € D and

: Theabas3
graph of z = SS
f (x, y) = c,5 where c is| a real
ai Constant a le
paraboloid of revolution z = x7 + y?, the level curves wetheEc eee oat eaemple, for the
We define the following: _™ Sie @.

fwo points Let P(x», yo) and O(%, y;) be any two points in IR*. Then
ees : af, 2 ='FOT= JG,1— Xo)”
2 +€y; — yo)?
_ IS
Called the distance between the poinP
tsand O. - a (2.4)
% a
 Functions of Several Real Variables 2.3

Neighborhood of a point Let P(x, yo) be a point in IR*. Then the d-neighborhood of the point
P(xo, Yo) is the set of all points (x, y) which lie inside a circle of radius 6 with centre at the point
(xo, Yo)» (Fig. 2.2). We usually denote this neighborhood by NV5(P) or by N(P, 0).

Fig. 2.2. Neighborhood of a point P(g, yo).

Therefore,

Me(P) = (2,9): {= 20)? + = 0)? < 5. (2.5)

Since |x-—xo |< y(x-x0)* +(y-—yo)* and |y-—yo|< ¥(x-2x9)? +(y—yo)?.
the neighborhood of the point P(x, yo) can also be defined as

, BletP) = {, y):|x- x0| <0 and |y—yo|<5d}. (2.6)

that is, the set of all points which lie inside a square of side:20 with centre at (xo, yo) and sides parallel
_ to the coordinate axes (Fig. 2.2). mee
bi If the point P(xq,yo) is not included in the set, then it is called the deleted 6-neighborhood of the
point, that is, the set of points which satisfy

0< V(x-x0)? +(y—y0)? <5 (2.7)

is called the deleted neighborhood of P(x, yo).
Open domain A domain D is open, if for every point P in D, there exists a S> 0 such that all points
in the 6-neighborhood of P are in D.

Connected domain A domain D is connected, if any two points P, Q € D can be joined by finitely
many number of line segments all of which lie entirely in D.

ded domain A domain D is bounded, if there exists a real finite positive number M (no matter
how large) such that D can be enclosed within a circle with radius M and centre at
the origin. That
__ is, thedistance of any point P in D from the origin is less than M, |OP |< M.
A closed region is a bounded domain together
with its boundary.
 2.4 Engineering Mathematics

Bounded function A function f (x, y) defined in some domain D in IR? is bounded, if there exists
a real finite positive number M such that |f (x, y) |< M for all (x, y) € D.
Remark 1

(a) The domain of a function of n variables z = f (x), x, . .. , X,) is the set of all n-tuples in IR”
for which f is defined.
(b) For functions of three variables, we need a four-dimensional space and an (n + 1) - dimensional
space for a function of n variables, for its graphical representation. Therefore, it is not possible
to represent a function of three or more variables by means of a graph in space.
(c) Fora function of n variables, we define the distance between two points P(x;9,
X09... ., Xno)
and O(x)1, NO [so «oy Xn1) in Ik” as

“| PQ|= (x1) - X19) ;

+ (Xo; — X29)? Fe
;
HN ong)
2

and the neighborhood N;(P) of the point P(x19, x29, . - -, X,9) is the
set of all points
(X}, X2, . . ., X,) inside an open ball

(x1 — 10)? :+ (x2: — x29)?2 +... 4 (x, —Xn0)2 2

<6.
2.2.1 Limits
Let z= f(x, y) be a function of two variables defined
in a domain D. Let P(x, yg) be a point in D. If
for a given real number € > 0, however small, we
can find a real number 6 > 0 such that for every
point (x, y) in the 6-neighborhood of P(X; Yo)

|f(% y)-L|<e, whenever f(x — x9)? +(y — yo)? <6 (2.8)

then the real, finite number L is called the limit
of the function f(x, y) as (x, y) > (xo, Yo).
We write it as Symbolically,

lim
(x, ¥) > (x9, yo)
f(x,y) =L.
Note that for the limit to exist, the function
f(x, y) may or may not be def
not defined at P(x9, yo), then we write ined at (Xo, Yo). If f(x, y) is

|f@, y)-L|<e, whenever O< V(—x

x9)? + (yy)? <6.
This definition is called the d-¢
approach to study the existence
of limits.
Remark 2

a) lim I (yy), if it exists is unique. (The proof

| (x, ¥) > (x9, yo) is similar to the case of functi
ons of one
7 variable),
nf (b) Letx=r cos 0,y=rsin @ so
thatx2 +y=r and @=tan"!(y/x), Then, we can define the
Sy om given in Eq. (2.8) as
“a 5 lim|f(r cos er sin@)-—L|
<é, whenever
pe 0, independent of
@.
a aS
a
 Functions of Several Real Variables 2.5

(c) Since (x, y) — (9, yo) in the two-dimensional plane, there are infinite number of paths joining
(x, y) to (Xo, Yo). Since the limit is unique, the limit is same along all the paths, that is the limit
is independent of the path. Thus, the limit of a function cannot be obtained by approaching the
point P along a particular path and finding the limit of f (x, y). If the limit ts dependent on a
path, then the limit does not exist.
Let u = f(x, y) and v =7 (x, y) be two real valued functions defined in a domain D. Let
(x, ee, yo) f(t) =1, and lim g(x,y)=L,.
(x,y)
— (40; Yo)
Then, the following results can be easily established.

(i) oa) eS Lk f(x, y)] =kL, for any real constant k.

(ii) lim (f(x,y) g(x,y] =k, + ly.

(x,¥) — (x6, yo)

(iii) (x,y)
lim(x9. yo)
[f(x, y) a(x, y= LL.

(iv) (tim Lf ya (x, y)] = Li! Lz, Le #0.

Remark of
3

Let z=f (x, X,..-, X,) be a function of n variables defined in some domain D in IR”. Then, for any
fixed point Po(X10; X20 ++ +9 Xn0) in D

plim f(*19%2> es

if |f(x, x2, . - -, X,) - L | < & whenever (x; = Kio)? bilks —Xog)? Fe. (Xe — Keg)? < 6
where P(x, x2, - - -» Xn) is a point in the neighborhood or the deleted neighborhood of Pp.
Example 2.1 Using the 6-€ approach, show that
( ) cw yhile. mex +4y) == 10,
10, lim
er se nNx7 +2y)=
oe

s') Here f(x, y) = 3x + 4y is defined at (2, 1). We have

| f@, y) - 10| =| 3x + 4y - 10] =|3@-2)+4@-1)|<3|x-2|+4]y-1].
—
aE: we take | x —2|<6 |f(x, y)- 10|<78<€, which is satisfied when
dand|y—1|<6, we get
s eae
¥ = 10.
Pe < limwa »

fe rs -Mice Reid, ye 7;
: = GN 2G— 142 —1)|<|x-1 |?+2(x-1])*2|y-1|
If wi Ba, < dand|y— 1/<6, we get|f(x, y) — 3 |< 6° +46< which iis satisfied when
| (642)! <€44 or6< Jer 4-2.

E
,
a ;
 2.6 Engineering Mathematics

Hence, lim f(x, y= 3. !

(x, yy (1,1)
We can also write .y)-3|<6°+46<50<€
_which is satisfied when 0= €/5.

Example 2.2 Using 0-€ approach, show that

: Nyt AZ YZ
= | 00) lim = (Q,
Oo eghe.o Veep? | a 0.20 fx? +y? +27
Solution

(i) Here f (x, y) = xy/(./'x7 + y”) is not defined at (0, 0). We have

, 2 2 ; 7
mo) | Se... cb OY 1 fa? 45? <8, (zy) 40.0)
x? + y* Nig ay : [2 +?

since eS
|xy |< (x?
ee + y*)/2. If we choose 6 < 2¢, then we get

xy 2 | , .
reg -Q|<ێ, whenever 0< Vx* 2 +y*<6.
2
x°+y

Hence, li a =).
(x,
y) > (0,0) 2 a y?

Alternative Writing x = r cos 0, y =r sin 0, we obtain

lim
(x, ¥)— (0, 0)

which is independent of 6.
2xy
Abe en © |
= him
r— >()
r* sin Ocos @
r 8
(ii) Heref(x, y, z) = (xy + xz + yz)/ yx? ty?
+2? is not defined at (0, 0, 0).
Since |xy |< (x7 + y*)/2, |xz |S$ (x?
+ 27/2, |yz |< G* + 2/2, we get

xy + xz + yz 2
of4] pe ee tye
ie re | Wate ya? |<e.
2
: : sue ATE aa )
x? + y? + 2?
Tf we choose 6 < €, we obtain

: xy
AM + XZ + V2 a, <€ whenever
Pa ¢ O< [x2 4524 22-9

fence AY + XZ + yz |
aS

” €5;,3 Ae (0,
(x,y,z) ot 0, 0)
ee

|me + =i
y2 ge = 0.
SS ———— ll

Variables 2.1
Functions of Several Real

Example 2.3 Show that the following limits

» 2 a a
fa) lim ade iow ev pret
FO>
YO l (lim
vo O
X* + y? | 5A hie 9 -y Poe
Ve
ih lim iis a : an~'(2) man v
(x,y) (0,0) x® 4 y? Aes 1)

do not exist.
ular path.
Solution The limit does not exist if it is not finite, or if it depends on a partic
x — 0. Therefore
(i) Consider the path y = mx. As (x, y) — (0, 0), we get
2
| mx m
(x,y) 0,0) x2 4 y? x90 (L4m?)x* 1+ m?
limits. Hence, the limit
which depends on m. For different values of m, we obtain different
does not exist.
setting *= COS 8, y=r sin 0, we obtain
Alternative

ne = in
Pe Are
sin Ocos | G = sin @cos 0
(x,y3)(0, 0) x2 + y* r>0 r2
.
which depends on @. Hence, the limit is dependent on different radial paths 8 = constant
Hence, the limit does not exist.
(ii) Choose the path y = mx’. As (x, y) > (0, 0), we get x — 0. Therefore,
Hay 6. Veealin
HM pe eee
(x,y) (0,0) x2 +y x90 (1+ m)x
Since the limit is not finite, the limit does not exist.
(iii) Choose the path y = mx’. As (x, y) — (0, 0), we get x > 0. Therefore
x *y
——>= lim mx ® zs m
lim #=£o

limit
which depends on m. For different values of m, we obtain different limits. Hence, the
does not exist.
ff We have

| lim tanSE Ase = tan (toc) cle

= + |
(x,y)(, 1) x
1.
depending on whether the point (0, 1) is approached from left or from right along the line y =
If we approach from left, we obtain the limit as —77/2 and if we approach from right, we obtain
_ the limit as 2/2. Since the limit is not unique, the limit does not exist as (x, y) — (0, 1).
sig *
2.2.2 Continuity
Afunction z = f (x, y) is said to be continuous at a point (x9, yo), if

ed at the Pppoint (xo,» Yo),

Yo (11) limy
ii yy ew? exists, <
xists, and
lim, fey) =f (20. Yo).
 2.8 Engineering Mathematics

If any one of the above conditions is not satisfied, then the function is said to be discontinuous a;
the point (x, Yo).
Therefore, a function f (x, y) is continuous at (x9, yo) if |

|f (x, ») —f (xo, Yo) |< € whenever (x- x0) + (¥- Yo ree. (2.9)
If f (%p, Yo) is defined and lim f(x,y) =L exists, but f (xo, Yo) # L, then the point (x, yp)
(x,y) (x0, Yo) h nib ¢ -
is called a point of removable discontinuity. We can redefine the function at the point (xo, yo) as
f (Xo, Yo) = L so that the new function becomes continuous at the point (x9, yo).
If the function f (x, y) is continuotis at every point in a domain D, then it is said to be continuous
in D.
In the definition of continuity, lim f(x, y) =f(%0, Yo) holds for all paths going to the
(x,y) (10, Yo) |
point (x9, yo). Hence, if
continuity
the of a function is to beproved, we cannot choose a path and find
the limit. However, to show that a function is discontinuous, it is sufficient tochoose a path and show
that the limit does not exist. hae
A continuous function has the following properties:
Pl A continuous function in a closed and bounded domain D attains atleast once
its maximum
value M and its minimum value m at some point inside or on the bounda
ry of D.
P2 For any number yp that satisfies m < lu < M, there exists
a point (x, yo) in D such that
f (Xo, Yo) = M.
P3 A continuous function, in a closed and bounded domain D, that attains
both positive and
negative values will have the value zero at some point
in D.
P4 If z=f(x, y) is continuous at some point P(x,
yo) and w = 8(z) is a composite function
at7 =f (Xo, Yo),then the composite function 8 (f(z)) is also defined
functions e*, log (x* + y’), sin (x + y) etc, continuous at P. For exam ple, the
are continuous functions.
Example 2.4 Show that the following functions
are continuous at the point (0, 0).
4 4
zy 2x7 + 3y" | 2x(x? = y?
Ft 9) = Ye spy (x,y) # (0, 0) (ii” f(x,y)= ers (x, y) ¥ (0, 0)
0 (5=00,)
0) ~ 0 » (x, y) = (0, 0),
Sine” (e422
| Ci) f(x,y) = 4 tan7'(2x+44y)’ (x,y)# (0, 0)
1/2 » (x,y
= (0,
)0). by
Solution
, ME thed) - Py Jay
(i) Letx=r cos 6, y=rsin @. Then, r= x2 +y2 40.
We have “oedingbe, 3-9
ale
|f(x,y) - £0, 0) |= oe EM | . |.r4(2 cos4+3 sin4g)
Be ty a
r* (cos?@ + sin2 @)
< r*[2|cos4 O|+3 |
sin4 O|)<Sr<e
 2.9
Functions of Several Real Variables

or r=Jx?t+y’< al e/5.

< Je/5, we find that |f(x, y) -—f, 0)! < €, whenever 0< x? +y°<6.
If we choose fi
ore, a My 9) O59) = aig 0) =0. Hence le is continuous at (0, 0).
,f (x, y) is
Theref
herefo

(ii) Let x=rcos 6,y=rsin @ Then, r= x? +y? #0. We have

If (x,y) =f, 0) 1 =
2x(t7"y*)
mery
os?0 —6 sin“) 8)cos@
| | 21?2(c
— os2 + sin?
_
=|27cos2@cos O\S2r<-E 4 —

or raaix? ty? <e/2.

If we choose 6 < €/2, we find that

lf (x,y) -f(,0)1<e, whenever 0 <x? +y°< 6.

Therefore, Rt F(5-y) = Oy 0) -=10.. Hence, f (x, y) is continuous at (0, 0).

(x,y) (0, 0)

(iii) Let x + 2y = t. Therefore, t > 0 as (x, y) > (0, 0).

We can now write
Y/t '

i |il
|
. etl

ee et eer |Ma
preied

fea 2°
(x,y) 90, a IS tan} 2¢ te (2t)/(2t) | l2t)
10

Since i. f(x,y) =f, 0) = 5. the given function is continuous at (x, y) = (0, 0).
(x,y) ©, 0)
_
- mple 2.5 Show that the following functions are discontinuous at the given points

(x,y)# (0, 0) — TS
i)
Gi) ear he x+y (x,y) # (0, 0)
0. (4 y)=@,0) 0 , (xy) =@,0)
at the point (0, 0).

)
+ y)# (2,2
, a (xy= 2, 2)

3 7=
th :
- a7 } a=whee Vor)
fies i aes
y ie x !a
f il yenee age ag eat

|
_

¥
 2.10 Engineering Mathematics
. ganction
‘
ig
*
not continuous
a | i afé t (Q) (VU, ()()),
, the functio
which depends on m. siege the limit does not exist
ore,
, We get x _» 0, Theref
(ii) Choose the path y = m?x*. As (x, y) — (0, 0)
" (Le mye”. T= eh
lim Ss Hall
y
ng
im ———
x90 (l+m )
= +m Ss
(xyy)9(0,0) xy?
action js not continuous at (0, 0).)
which depends on m. Since the limit does not exist, )the function is not

(111)
i es
x, yD, 2) Fi, y) = an yum. 2)
(x+y) (et D
(x fe y)
bag 1A D9.
(x,y aM, 2)

x
Since a yim, i f(x, y) # f(2,nor2), the function
ation 1sia not not con
continuous at (2, 2).
Note that the point (2, 2) is a point of removable discontinuity.

f | ;

Example 2.6 Let f(x,y) = eee (x;y) (0,0)

4. #£ 0, (ay) =@,0).
Find a 6 > 0 such that | f(x, y) ~/(0, 0)I < 0.01, whenever
Solution We have

If(x,y) -£@,0)| = *

Substituting x = r cos @, y = r sin 6, we obtain

*
If(%,y)-f0,0)1=]-

-*3 Gl <<
ac SP as (ee) oa? x F er = a =
3a f Te ee yoen ee‘ eeLo fi Vote
‘ ae i 5%
oa J ' Mi fi

bee BPO NAb ’

é
.
ie Dee See
et
iy|
ST
enn)
:
s
x
=
7)
fs
Ene
a
5
5)
ye We
i
~
“+
I
’
~
:
i
:

| TherefPoe J: : <¢ .
- iy i 7. a FF:

i i= TiOiT
re
e, -
= zy y? 0.0 iF5=0. i a an _
site oy ; va, . o ‘ ye ‘= 0.002. Hi ‘hkia, = é pels . Pte

— Ps ‘ 7 . ‘oewilh a icnc
la g,. ‘»< 0.1 N JZ, a

- U
 Y

Functions of Several Real Variables 2.1)

Determine the following limits if they exist.

| 3
' y3
7:
:
lim v
(x.y) (0,0)
fa
—
BM
8: |im ee ae |
A
|x? oh y? (x,yy>0,-1) w= ¥

|
” lim [1s 4 .
| 10 Le
fie
1m cot
rs] ) :
(x,y)—>(a,0) y (x, v)—(0,0) |x? + y?

(y — 1) tan? x (x—1)siny
11. im =——————_.. i —_——-
(ay )—(0,1) x? (y? ae L) 12. (z.9)-301,0) y In x

l-x-y
eis iy see
(990.0) 2 4 y? oe eyo x? +y*

3 42
a 16k ne ——

i 1 * v6 ‘ xY + a
Me lim _
(x,¥,2)—(0,0,0)
log} — |. 18. ar
xy (x,y,2)9(0,0,0) 4 + y + ne

Gyyey! , x(x +y+Z)

(x,y,z) (0,0,0F yt + y* 4+ z8 (x,9,2)3(0,0,0) ¥- + yp" 47°

Discuss the continuity eethe following functions at the given points.

(e- We
Be Ha) 75,77 -
OYTO” 22. f(x,y)=| rem+y*, (x,y) # (0,0)
0 , (x, y) = (0, 0) 0 » (x,y) = (0, 0)
at (0, 0). | at (0, 0).
eel
ae 2

24.f(x, y= xy
‘ys, ai
0 : (x, y) = (0, 0)
| eh 4 (es y) = (0, 0)

at (0, 0).

x* ~2xy+y?
7 -, ary) # (0, 0) ’ (x, i ) oe 1G —l1)
26. TUS y)= Mire

(x,y)= (0,0) 0 ? (x, y) = (,-l)

at (1, =1).

PrN
| eae (x, y) # (0, 0)
28. f(x,y) =
0 (%, y= (;.0)
a

» (x,y) #0, 0)
Te ae ate 1 |

Pees 4
ius

yt ijeeu =i (0, 0)
7oy =, 0)

>
I~ — tM Engineering Mathematics

(x,y) 4 0,0)
5

x*y? . oe x
[x?=¥
2+y
———_——_ X, = 0, 0
31, f(x,y) — x? of, y? ' (x y) ( ) 327 f( XxX; y) = s 0 (x, y) a (0, ())

0 , (hy) =@9
at (0, 9).
at (0, 0)
x*y
x#-1
33. f(x,y)= 4 | +x’
y , (x,y) =CI, @)
at (-1, a).
XYZ
———_——, (4, ),2) (0, 0, 0)
D ii +y? +7" ( y
34, F(x), 2) =

0 (x,y,z) = ©, 0, 0)
at (0, 0, 0).

2xy
———,, (x, y, z) # (0, 0, 0 )
x ae 37? ce y ( )
35. fies et

0 , (x,y=, ) 0)
,z0,
at (0, 0, 0).

2.3 Partial Derivatives

The derivative of a function of several variables with respect
to one of the independent variables
keeping all the other independent variables as constant is called the partial derivative of the function
with respect to that variable. ——
Consider the function of two variables z = f (x, y) defined i
; 7 n some do _ !
Let y be held constant, say y = yo. Then, the functi main D of the x-y plane.
an interval about x, that is f(x, yo) is a function of 0

A, z=f(x+ Ax, yo)

—f (x. Yo)
is called the partial increment in z with respect to x (2.10)
s ok. an
Similarly, if x Is held constant, say x = xo, then the isda function of
fdueteg ae. of x and Ax,
defin ed in some interval about y, that is f(xo, y) is a funct; noe rie ¥) depends only on y and
and (xo, y + Ay) be in D, where Ay is an increment in is
the independent varicty- Let the points (Xo, y)
3 Variable y. Then
Ayz=f (xo, y + Ay) — f (xo, y) :
is called the partial increment in z with (2.11)

-
This increment is called the total increment in z
and
Z+A. ZF
nn vene ‘al, Az #2 ee is
- es of x, y, Ax and Ay
soa oe SXaMpl e, conside r
We have € function 7 — 3
On Z =F (x, y) = xy and a point (x, yo)
.
A,z = (x + Ax) yo — X0Yo = yoAx
~ ~
7? a4
:
. Functions
.
of Several Real Variables
‘
2.15
o ;

- ) | |
Ayz=Xo(Yo + Ay) — xo yo = xg Ay
Az = (x9 + Ax) Oo + Ay) — Xoo = XpAy + yoAy + AxAy # A,z+ Ayz

Now, consider the limit

lim A,z = fin f(%o + AX, Yo) — f (0; Yo) aa (2.13)

Ax30 Ax Ax-0 Ax

If this limit exists, then this limit is called the first order partial derivative of z or f (x, y) with respect
to x at the point (xo, yo) and is denoted by z, (xp, Yo) Or f-(Xo, Yo) OF (OF /Ax) (xo, Yo) oF (9z/Ax) (Xo, Yo).
Similarly, if the limit

lim ByZ s. f(%0,¥o0 + Ay) — f (Xo, Yo) (2. 14)

Ayo Ay 0 Ay
exists, then this limit is called the first order partial derivative of z or f (x, y) wiih respect to y at the
point (x9, yo) and is denoted by z, (xo, yo) or f,(%, Yo) or (dz /dy) (xo, Yo) or (OF [Ay) (Xo; Yo):
Remark 4

Let z = f (x, x2, - . ., x,) be a function of n variables defined in some domain D in IR”. Let
Po(x}, X2, . - ., X,) be a point in D. If the limit

i Ce CR ee | Ais a 83.5
Ax;j>0 Ax; Axj20 Ax,
exists, then it is called the partial derivative of f at the point Py and is denoted by (¢f/dx;) (Pp).
Remark 5

The definition of continuity, fim f(x, y) =f(%o, ¥q) can be written in alternate forms. Set
x, ¥)>(x9, Yo
X=Xx)+ Ax, y=yo+ Ay. Define Ap= ./(Ax)*+ (Ay)* . Then, Ax > 0, Ay > 0 implies that A p—> 0.
We note that |Ax |< Ap and |Ay |< Ap.
The above definition of continuity is equivalent to the following forms:

(i) Jaa eee [f(xo + Ax, yo + Ay) —f (x0, yo)) = 0.

(ii) ah, [f(xo + Ax,yo + Ay) —f(%0, ¥o)] =

(iii) lim, Az=0.

we have

eg
~ Ox Ax->0
et Oe)Ax IY) 2,
Ax—>0
Ue +Ax)? +y?+(xAx+ Ax) -[x2 +y? +2]
maa + 1)A+ (Ax)?
fee Ax s arttg 2% + 14+ Ax] = 2241.
2.14 Engineering Mathematics
2 -

| 2 + Ay)” 2 +21 —[x* +y + | x]

OF iy LEA F) e
dy Ay 30 Ay Ay0

Ay-0 Ay Ay-0

= - Ax ee =
ee oye? li — =— ye
oe eee
oy Obie ti DE ae a,
avo As
Of lim i PAye = ye ee
dy Ay30 Ay
iti Of _ |, sin (2(x + Ax) + 3y) - sin(2x+ 3y) _ ae cos(2x + 3y + Ax) sin Ax
a dx , Ag=SO | Ax Ax-0 Ax

= 2 cos(2x + 3y).
Of 7 sin (2x + 3(y + Ay)) — sin (2x + 3y) 2 cos (2x + 3y + 3Ay/2) sin (3 Ay/2)
—= ili : = ha eeFeeeFSFSMMMMMMM—SSSSSS———
dy Ay—0 Ay Ay—-0 Ay

sin (3 Ay/2)
- sim, [3 cos (2x + 3y + 3Ay/2)] Gay l) = 3 cos (2x + 3y),

Example 2.8 Show that the function

, : l
(x + y)sin : } x+y#0O0
f(x,y) = wy :
0 , xt+y=0
is continuous at (0, 0) but its partial derivatives f,
and 18 do not exist at (0, OQ).
Solution We have

|f(x,y) —f, 0) |= (x+y) sin{

S|x+y| <|x|+/y) < 24/x? tyrce.
If we choose 6 < €/2, then

|f (x, y)-O|<€ whenever Oc jx? Hye 5.

Therefore, iy jim f(x,y) =0=f(0, 0).
— (0,0)

Hence, the given function is continuous at (0, 0).

FD
Now, at (0, 0), the limit
Ky

lim S22 = lim TAs) =F0,0) | lim AX sin (/ Ax)

Ax>0 Ax Ax30 Ax agen . to = lim sin (4)
does not exist. Therefore, the partial derivative f, does not exisee. 0, 0) Ax->0 Ax
Similarly at (0, 0), the limit
aS

ayoAy 50 Ay ee
Ayz caf f (0, Ay) — f (O, 0) .
 Functions of Several Real Variables 249

does not exist. Therefore, the partial derivative f, does not exist at (0, 0).
Example 2.9 Show that the function

pea. =|Teh 409 0

0 me ee y= (0, 0)

is continuous at (0, 0) but its partial derivatives f, and f, do not exist at (0, 0).
Solution We have

x? +" | 12
|f(x,y) — (0, 0) | = oe
eLiel tI Lei 4 [y| sayz? +y?<.
[x}+y|
Taking 6 < €/2, we find that

| f(x, y)-0|<e, whenever 0< Jx* +y*< 6.

Therefore, dun f(x, y) = 0 =f (0, 9).

(x, ¥)— (0,0)

Hence, the given function is continuous at (0, 0).

Now, at (0, 0) we have

im et — tim £6429 -f0, Se ee | -{ x > 0

1, when Ax
Ax30 Ax Ax230 Ax ~ Ax-—30 |Ax | —1, when Ax < 0.

Hence, the limit does not exist. Therefore, f, does not exist at (0, 0).

Also at (0, 0), the limit

f (0, Ay)——
Ayf = lin ——— —f(0,9) AY
_ inim ——= 1, when Ay
|
> 0
lim |
. Ay-0 Ay ~ Ay—0 Ay ~ Ay30 | Ay| —l, when Ay < 0

does not exist. Therefore, f, does not exist at (0, 0).

Examp| 10 Show that the function

Be ase| eyxyge
Se ee aes
|
x, 9) LO)A ( ‘y

_ nat : 0 sO, y) = (0, QO)

$ not continuous at (0, 0) but its le derivativeshy and f, exist at (0, 0).

2
f( ey) A ee
(x) jo, 0) u a (l+2m?)x? 1+2m?
1 m, the function is not continuous at (0, 0). We now have

pcmey a
2.16 Engineering Mathematics

fy, Q) = Aim,
700, Ay)=£00.0)_ jim 9=0 <0.
Ay dy7o0
————

Ay

fy exist at (0, 0)
Therefore, the partial derivatives f, and
| e Viti Piss
yr continuity) erhtias
A sufficient -ondition for a function
con : f (x, y) | to
heorem 2.1 (Sufficient condition for bale Mey ‘vatives exists and is bounded
+ eae- Hi at a point (xp, Yo) is that one of its first order partial derivatives
ay | al bie ' r = ‘

in the neighborhood of (xo, Yo) and that the other exists at (%, Yo)- -
, of
of tlthe p oint (X%q, Yo) and
0» Yo)
Proof Let the partial derivative f, exist and be bounded in the neighborhood
fy exist at (x9, Yo). Since fy exists at (xo, yo), we have

lim f (xo, yo + AY) - f(%0+ Yo) _ ¢ (x5, yo).

Ay-0 Ay

Therefore, we can write

f (xo, yo + Ay) —f (os Yo) = AY Fy| o> Yo) + &1 AY (2.15)

ZAS)

where €, depends on Ay and tends to zero as Ay — 0. Since f, exists in the neighborhood of (x9, yo),
we can write using the Lagrange mean value theorem

f (% + Ax, yo + Ay) —f (Xo, Yo + Ay) = Ax f, (Xo + 8 Ax, yo + Ay), O< O< 1. (2.16)

Now, using Eqs. (2.15) and (2.16), we obtain

f (x9 + Ax, yo + Ay) —f (Xo, Yo) = [f (xo + Ax, yo + Ay) —f (xo, yo + Ay)] + LfGo, Yo + Ay) —Ff (Xo, Yo)]

= Ax f,(%9 + OAx, yo + Ay) + Ay F(X, Yo) + & Ay. (2.17)

Since f, is bounded in the neighborhood of the point (x9, yo), we obtain from Eg. (2.17)

Ae tthees6 I(%0, 4x,yo + Ay) =f (x9, yo).

Hence, the function f (x, y) is continuous at the point (Xo, yo).

Geometrical interpretation of partial derivatives

Let z= (x, y) represent a surface as shown in Fig. 2.3. Let the
plane x =
Xo = constant intersect the
surface z = f (x, y) along the curve z = f (x, y). Let
P(X, y, 0) be a particular point in the x-y plane
and R(x, y, Z) be the corresponding point on the surface, where Z=f (Xo,
4 point in the x-y plane in the neighborhood of y). Let Q(x, y + Ay Q) be
aks P and S(%, y+ Ay, z+ A
‘ce a: ve
. BR
z) be the corresponding
Aoipoint
y). From Fig. Leds we find that Ay — PQ —_ RS’ ame the function Zz is increased
a et Gye) —Z= Ayz. Now, let 0* be the angle which the
y-axis. Then from ARS ee oe & ¢
€ chord R makes
RS | it the positive
with iti

tan @* = 55_ _
Ayz
et Ay + 0. Then, A,z > 0.
Hence,
5
in —— Ge. i
aya tan @
—

Ay>0 Ay
—
 Functions ofSeveral Real Variables ON 9

where in the limit, 6 is the angle made by the tangent to the curve z =f (xo, y) at the point R(x, y, Z)
on the surface z = f (x, y) with the positive y-axis.

Fig. 2.3. Geometrical representation of partial derivatives.

Now, consider the intersection of the plane y = yo = constant with the surface z = f (x, y). Following
the similar procedure, we obtain dz/dx = tan @, where @ is the angle made by the tangent to the curve
z = f (x, yo) at the point (x, yo, z) on the surface z = f (x, y) with the positive x-axis.
It can be observed that this representation of partial derivatives is a direct extension of the one
dimensional case.

2.3.1 Total Differential and Differentiability

Let a function of two variables z =f (x, y) be defined in some domain D in the x-y plane. Let P(x, y)
be any point in D and (x + Ax, y + Ay) be a point in the neighborhood of (x, y), in D. Then,

Az =f
(x + Ax, y + Ay) -f (x, y)
iscalled the total increment in z corresponding to the increments Ax in x and Ay in y.
The function z = if(x; y) is od to be wa a at the aia (, y), if at this point Az can be
vee Salve of te ‘
O° ie : KS—)
A
a] PA

ae

b are inde Banden of Te, A

here2 3, = ax Ay), & = &(Ax, Ay) are infinitesimals and
and E :
me fine 5of Ax, Ay such that €; > 0, & — 0 as (Ax, Ay) — (0, 0).
Fey
| he first part a Ax + b Ay in Eq. (2.18) which is linear in Ax and Ay is called the total differential
orsing
oly‘the differential of z at the point (x, y) and is denoted by dz or df. That is
dz=aAx+bAy or dz=adx+b dy,
, %
ee
 2.18 Engineering Mathematics

Let Ay = 0 in Eq. (2.18). Then, Az = a Ax + €,Ax. Dividing by Ax and taking limits as

Ax + (
we obtain a = 0z/dx. Similarly, letting Ax = 0 in Eq. (2.18), dividing by Ay
and taking limits q.
Ay — 0, we obtain b = 0z/dy. Therefore,

igen 2 as =f, Ax +J,Ay

Ox dy (2.19)

assuming that the partial derivatives exist at P. Hence,

Az=dz+ &Ax + &,Ay.

Therefore, existenceo artial derivatives f, and f, at a point P(x, y) is a necessary
differentiabili ) Of f(x, y) at P.
The second part €,Ax + €Ay is the infinitesimal nonlinear part and is of higher order relative to
Ax, Ay or Ap= 4{(Ax)*+ (Ay)? .Note that (Ax, A y) — (0, 0) implies Ap — 0. Eq. (2.20) can be
written as

LAc) “(ae|‘ eo Zp)

Now, if f (x, y) is differentiable, then as A p — 0, €; > 0
-
and & —> 0.
Taking the limit as Ao > 0 in Eg. (2.21), we obtai
n

mee on. ai( a} + |} =(

4p90 =AP Ap 50.| '| Ap aN Ap | 799 )
(2.22
since |Ax/Ap| < 1 and |Ay/Ap| < 1.
Therefore, to test differentiability at a poin
t P(x, y), we can use either of the follow
ing two approaches.
(i) Show that lim 42—% _ 9 .
Ap—0 Ap—» 2x) + Oy) (2.23)
(ii) Find the expressions for E\(Ax,
Ay), &(Ax, A y) from Eq. (2.20)
Z., and lim & > 0Oas (Ax, A ) — (0 and then show that lim E, > 0
NY ShoNot
u e that ” bol l par tia l der C ri
,0) or Ap
e
> 0,
that the function f (x, y) may n SBand ore te conhno ys ot uy)
Efiflerennable a a point P(x, y),
deriva tives f., fy exist at P (see Exa even if the partial
mple 2.12)~ H t ever , if the
continuous at the point P, then the firs t order Partial derivatives are
functionis differentiable at P.
We present this result in the follow |
ing theorem.
_- Theorem 2.2 (Sufficient condit
ion for differentiability) If the
~~ first order partial derivatives fu Nnetion z = f (x, y) has continuo
at a point P(x, y) in D, then F(x us
» ») is differentiable at P.
_ Proof Let P(x, y) be a fixed point in D,
By the Lagrange mean val
ue theorem, we have
}
EAS AE, We fH, Wx Mh
£04 O,4x, 9), 0< 6, <
and |
I(x + Ax, y + Ay) —f(x+ Ax, y)
= Ay f(x + Ax, y + 0, Ay),0<
Sincef,and f, are continuous at 0, <1,
(x, y), We can write
Sx + OAx, y) = f.(x, y) +
|
fy+(Ax%
, y + @,Ay) =f,(x, y) +e
P
 al Variables 2.19
Functions of Several Re
7! _» 0, that is,
: ee of Ax, Ay and tend to zero as Ax — 0, Ay
where €, &2 are infinites imals, are functions

as AP = (ax) (Ay)*> 0. Therefore, we have

(2.24)
f(x+ Ax, y)-f(% y) =AxfiG y) + €, Ax
Ay (2.25)
f(x+ Ax, y+ Ay)-—f(w+ Ax, y)= Ay fy, y) + &
and
Now, the total increment is given by
Az=f (x + Ax, y+ Ay) —f(@ y)
=[f(x+ Ax, y) —f (x, y)]+ Cy a mgs—f (x + Ax, y)I.
Sol
Using Eqs. (2.24) and (2.25), we obtain
(2.26)
Az=f, Ax+f, Ay +& Ax+ & Ay
y). Hence, f (x, y) is differentiable at P.
where the partial derivatives are evaluated at the point P(x,
Remark 6
rential as
(a) For a function of n variables z = f (x1, x2, . - -, Xn), we write the total diffe
dz=f,,dx + fi, dea +... + fe, 4%n- (2.27)
n
(b) Note that continuity of the first partial derivatives f, and f, at a point P is a sufficient conditio
for differentiability at P, that is, a function may be differentiable even if f, and f, are not
continuous (Problem 5, Exercise 2.2).
he conditions of Theorem 2.2 can be relaxed. It is sufficient that one of the art order partial
derivatives is continuous at (xo, yo) and the other exists at (x9, yo).

Example 2.11 Find the total differential of the following functions cil
y
al

2 = tan”! (x/y), (x, y) # (O, 0), (il) = xz+ x) ZF O,

Solution . ;

Sil) 7 l l y
(i)
j f(x,y)
: = n
tan ; (=) fe
atk ee fc (>|
| a |
aye

| 2
/ J ce ee ee
oo Sy Ty yy | A ay
Therefore, we obtain the total differential as
path
Sen 1)
a
= ag de — 20).

*.. y y-l
(ii) f(x, y,z) = [xz+ *) Sy = y(x2 + =) (2es =|
é Z

, ha(aet) (2+ 3)£-o(x+2)"(x-4)

° = , y

ul 7 f.= i rife oe ‘a

Therefore, we obtain the total differential as

,,

J« (=+ x)" o( rtd oft - Jr)ae|+|(x

+ +2) in [c+] |o
 2.20 Engineering Mathematics

Example 2.12 Show that the function

x? tay
, (x,y) # (0, 0)
f(x,y) =35 x2 +y?
0 , (x,y)
= (0,9)
(i) 1S continuous at (0, 0),
(ii) possesses partial derivatives f,(0, 0) and f,(0, 9),
(if) 1s not differentiable at (0, 0).
Solution
(1) Letx=rcos @and y=rsin 0. We have

|f(x, ¥) — f(0, 0) | = ee < r[|cos sin°A| J

+2]?@|

S3r= 3x7 +y7 <é.

Taking 6 < €/3, we find that

ae O|<é, whenever 0< ./x? +y°<6.

erefore, 2 one
Theref l 0) FCs, y))= 0 = F(0; 0).

Hence, f (x, y) is continuous at (O, 0).

(i) f,(0,
0) = lim, Te lin Ax —0
Ax>0 Ax =e

F(0, Q) = lim FO, Ay) =— LO, O) lim 2 Ay — 0 Af.

Ay 4550 9A y
: Therefore, the partial deriva
tives f.(0, 0) and f,(0, 0)
exist
(ili) We have dz = Ax + 2A
y. Using Eq. (2.20), we get
|
Az= Ax +2Ay + E;Ax +
&,Ay
Let Ap= (Ax)? + (Ay)?. Now,
x*
neat
<*
ey eo) Az= f(Ax, Ay)—-
—f(0, 0) = {Ax)" + 2(Ay)3
A 2

Bees,
‘< Hence he ee
= Az
: 47 tim, — - lim —L | (4x)*
+ 2(ay)3
ak | Ps 4p>0 Ap (4x)? + (Ay)? ~ (Ax + 24») oa
Bee.
s # = Aplim - |Bed (ay + 2)
0 {(Ax)? 4 (Ay)2}372
a 5a
P>X A
Functions of Several Real Variables pa

Let Ax = 7 COS QO and Ay = rsin @. As (Ax, Ay) > (0, 0), Ap = r > 0 for arbitrary 9. Therefore,
Az —dz a) O)|
|
4p30 ee
Ap es
Bm [cos @sin: 0 (sin 8 + 2 cos

— [cos @sin O(sin 8 + 2 cos @)] |

is not
The limit depends on @ and does not tend to zero for arbitrary 8. Hence, the given function
differentiable. Alternately, we can write

Ae Me 1 |Arla
s2640)? Ay) 6 (44)40,(32]
Ap Ap (Ax)?+ (Ay)? ale 2p
(Ay)? O( Any
€) = — —_—__
where 1 (An? + (Ay)? ape So Smee Ne
CAR)? = (Ay)?
Substituting Ax =r cos 0, Ay = r sin @, we find that €, and €, depend on @ and do not tend to zero
for arbitrary 9, inthe limit as r > 0. .
Example 2.13 Show that the functien

| xo ye
/ Foe Le
ple AL hy)
O° Gy = 1)
is continuous and differentiable at (1, —1).

Solution We have
2 = y 2 :

Sa el xy) y) =0 =fF((1, =1).)

(.y)>30,.-1) X-y ey we

Therefore, the function is continuous at (1, —1).

The partial derivatives are given by

£-0,- -1) =m,

as Ax
ie ee
| Aa +1 mL| i
Ax | 1 +Ax)
2 te
me a ey,

=,Be
$,(.-1)lim srl +Ay)—f LD) 5)
a3 ayo Ay |T-Cl+4y) ee
9|=5A,
Therefore, the first order partial derivatives exist at (1, —1).
; Nov i WE aave

CeO0)
eee ay ty? ey?
Gay) aye. Gay Cx, y) # Cu. —1)

Ai (x, y) = Ly (x, y) — le —]).

y) 1 @= ay)?
a fa spi29) = (OD E=y)F ! piste 1)
 2.22 Engineering Mathematics

the partial derivative /, is continuous at (1, —1). Also f,(1, —1) exists. Hence,f (x, y) is differentiah|,
at (1, —-1),
Alternately, we can show that eo [(Az — dz)/Ap] = 0.

2.3.2 Approximation by Total Differentials

From Theorem 2.2, we have for a function f (x, y) of two variables

f (x + Ax, y + Ay) —f @, y) =f, Ax + f, ay

or f (xt Ax, y + Ay) =f (x, y) +f, Ax +f, Ay (2.28)
where the partial derivatives are evaluated at the given point (x, y). This result has applications jp
estimating errors in calculations.
Consider now a function of n variables x), x, . . ., x,. Let the function z = f(x), %,.. .. X,) be
differentiable at the point P(x, x2, . . ., x,). Let there be errors Ax;, AX, ..., AX, in measuring the
values of x), %2, . . ., x, respectively. Then, the computed value of z using the inexact values of the
arguments will be obtained with an error

Be fi Morty Ans... Am,) f(r, X, « + 5X). (2.29)

When the errors Ax), Ax», . . ., Ax, are small in magnitude, we obtain (using
the Remark 6 (a).
Eq. (2.27))
Fy FAR, ty +AXs, . 2%, + AX) =f (X1, Xo, vey Xn) + fy, Any + fy, Axg
+... t AX, (2:30)
where the partial derivatives are evaluated at the point (x,
%, ..., X,). This is the generalization of
the result for functinns of two variables given in Eq. (2.28)
.
Since the partial derivatives and errors in argument
s can be both positive and negative, we define
the absolute error as (using Eq. (2.29))
lAzl=ldzl=|dfl= ieee + fx, Ax Tie
th nAe,
Then, |df sil x | Ax, 1+|f,, in| Ax>
[ a oe |Toe | | AS; |.
(2.31)
gives the maximum absolute error
in z. If max | Ax; |< Ax, then we
can write
UaVS SA ie Us| + coe Lf
The expression | ¢ ae
e error and [| df\/| f 1] x 100 is called the
ai

lice. oe
ae —-s ‘

ee
relative
J .
:

percentage error.
The maximum relative error can
also be written as

Wala fi |Ax, 1+ "sn 74) ae ieSnes on tl

’ <|ax
a2tinisti|tax 14 aot in | f17 lA
an a 20. Fin Lf} |1 ry I
“J ey ae
r _ Example 2.14 Find the ,
total increment and the total differential ;
¥ at the point (1, 2) for Ax = 0.1 and Ay = of the function z = x + y + x)
— 0.2. Find the maximum absolut
_ relative error. Olute error and the maximu| m
 Functions of Several Real Variables 2.23

Solution We are given that f (x, y)=x+y4 xy; (x; y) = (1, 2).

therefore, f(1, 2)=5, f,(1, 2)=3, £,(1, 2) =2. We have

total increment = f (x + Ax, y + Ay) — f (x, y)

= [(x
+ Ax) + (y + Ay)+ (x + Ax) (y+ Ay)] —[x t+y + xy]
= Ax+ Ay+x Ay+yAx+ Ax Ay.
At the point (1, 2) with Ax = 0.1 and Ay = — 0.2, we obtain

total increment = 0.1 — 0.2 + 1(—0.2) + 2(0.1) + (0.1) (- 0.2) =-— 0.12

total differential = PALL, QV As + f,01, 2) Ay = 3(0.1) + (2) (0.2) = -0.1

maximum absolute error = af \= of |

| Ax |+ of |
|Ay |= 3(0.1) + 2(0.2) = 0.7
: dy

maximum relative error ia = md) Btee \

Example 2.15 Using differentials, And an approximate value of

(i) f (4.1, 4.9), where f (x, y) = x3 + xy.

(ii) f (2.1, DL) where f (x, y) = x, (log 2= 0.3010).

Solution
Me Let (x, y) =(4, 5), Ax = 0.1, Ay = — 0.1. We have

: 3
GGy)= 4x? +x7y, fG,5)=12, f(y) = pe F2(4;5)= *,
oe +x*y

—_*__.,
Dalx? + xy
f, (4,5)
=4 2 |

| =12+4 (3) (0.1) + (3 (-0.1) = 12.3.

J eis: J(4.1,
4.9)= 12.3.

ee > &
Ok , Ay = 0.2. We have
iz?
. ’

ee ' 7 i]
 2.24 Engineering Mathematics

Therefore, f(21, 3.2) =/ (2, 3) 40, 3) Axct f, (2, 3) Ay

= 8 + 12(0.1) + (2.408) (0.2) = 9.6816.

The exact value is f (2.1, 3.2) = 10.7424.
Example 2.16 Find the percentage error in the computed area of an ellipse when an error of 2%
made in measuring the major and minor axes. 2+).
Solution Let the major and minor axes of the ellipse be 2a and 2b respectively. The errors Aa ay
Ab in computing the lengths of the semi major and minor axes are
Aa = a(0 02) = 0.02 a and Ab = B(0.02) = 0.02 b.
The area of the ellipse is given by A = 2 ab. Therefore, we have the following:
Maximum absolute error in computing the area of ellipse is

|dA|=
_| 9A
da |Aa |+ i ||Ab |= 1b(0.02a) + 2a(0.02b) = 0.04 rab.
Maximum relative error is

|
a = (0.04 mab) aS = 0.04.
Tab

Percentage error = |
wt |x 100 = 4%.

‘3.3 Derivatives of Composite and Implicit Functions

(Chain Rule)
Let z= f(x, y) be a function of two independent variables x and
y. Suppose that x
functions of some independent variable t, say x= O(t), y= W(t). Then, z= F{O( and
),
y are themselve:
w(2)]is a composit
function of the independent variable t. Now, assume that
the partial derivatives f f, are continuou:
functions of x, y and @(t), w(t) are differentiable functions of f. aa.
Let Ax, Ay and Az be the increments respecti
vely in x, y and z colresponding to the incr
in t. Then we have ement A!

of of
Aree a Ay +& Ax + Ey Ay.
Dividing both sides by At, we get

Az _ of Ax
—a of Ay
e Ax Ay
At Ox At” dy At Te Ag (Q
12 Ar 2-42)
Now as At 0; Ax 40, Ay 0 and E| = => Ay
0.25 Ga |
e | t -|
— 0. Therefore, taking limits on both
sides in Eq. (2.32) as At — 0, we obtain :
: a _ of dx | Of dy |
“? ;
‘" Sis dt dx dt aah ate (2.33)
=@

ew let x and y be functions of two independent

*
n,Z=f[d(u,v), Wu, v )] isa com variables uv and VU, Say xX = O(u, Vv), y = Wu, V).
me:
@ posite functi on of two independent variables
u and v. Assume
 Functions of Several Real Variables 2.25

to
that the functions f (x; y), @(u, v), w(u, v) have continuous partial derivatives with respect
their arguments. Now, consider v as a constant and give an increment Au to u. Let A,x and A,y
be the corresponding increments in x and y. Then, the increment Az in z is given by (using
Eq. (2.23))

Az= Za, r+ Tsny + €;4,x

+ E2 Avy

where €,, € > O as Au > 0.

Dividing both sides by Au, we get

= _ Ff Aim , Of Ay. ae AUX Auy (2.34)

~ Ox Au dy Au Au
Taking limits on both sides in Eq. (2.34) as Au — 0, we obtain

oz _ Of ax , Of oy (2.35)
Ou ox du dy Ou
Similarly, keeping u as constant and varying v, we obtain

dz _ of Ox ohOf ey
dy 2.36
do ax ov dy OD =
The rules given in Eqs. (2.35) and (2.36) are called the chain rules. These rules can be easily extended
to a function of n variables z = f (x;, Xo, - - ., X,). If the partial derivatives of f with respect to all its
arguments are continuous and x), x2, . . ., x, are differentiable functions of some independent variable
t, then

(2.37)

Example 2.17 Find df/dt at t = 0, where

(i) f (x, y) =x cos y + e*siny,x=t?+1l,y=t? +t.
(ii) f(x, y, ax txert+y +xyZx=e, y= Cost, z= £3.
Solution
(i) When t = 0, we get x = 1, y = 0. Using the chain rule, we obtain

of dy
df _ of dx me = (cos y + e* sin y) (2t) + (-x sin y + e* cos y) (317+ 1).
Geo On dt oy at
) = e.
Substituting t = 0, x = 1 and y = 0, we obtain (df/dt

(ii) When t = 0 , we get x = 1, y= 1, z= 0. Using the chain rule, we obtain

ase_ of dx , of dy | of dz
max at oy at dz at

: = (3x7 + z* + yz)(e’) + (3y? + xz) (— sin t) + (2xz + xy) (37).

he ‘Substituting t = O,x= 1. = 1; a 0, we obtain (df/dt) =3
 -.26 Engineering Mathematics
>
- ®
ie

Example 2.18 If z=f (x, y),x=e“+ 6, yao 4 e*”, then show that
ee 72| Ld z|
du ou dx By
Solution Using the chain rule, we obtain

of _ Of ax of dy _ 95m OF yn 2u Of
du doxdu dydu ~ ox dy

Therefore, of- of= 2(e7" +e?) of—2(e™ +?”)

:ps of variables

Suppose that f (x, y) is a function of two independent variables x, y

and x, y are functions of two new
independent variables u, v given by x = O(u, v), y= w(u, v). By
chain rule, we have
of _ of dx | Of dy and of of ox Of oy
du Ox du dy du dv dx dv dy du
We want to determine df/0x, df/dy in terms of df/du and of
/dv. Solving the above system of equations
by Cramer’s rule, we get

Of ldx = of /dy i 1]
Of dy of dy Of dx Of ax ax ay
eS cee
dx dy
du dv
—————_—

dvudu
oe

odvdu
SO

odudv.
eee

du Ov dv Ou
The determinant

|ax/du ax/dv] a(x, y)

dy/du dy/dv| (u,v)
is called the Jacobian of the variables of tran
sformation. Similarly, We write
Of dy of dy _Of,y) |Of/du daf/av
du dv
du dv du du
Ov du u,v
O(u,v)
) ~ dy/du —dy/dv
| Of dx Of
and =~ 5 > Ox = A(x,f) |dx/du ax/dv SS O(
eef,x)
dv du du dv Au,v) Of/du daf/av O(u,v) -
Hence, we obtain
of _1[Of,y)] Cs WOCh.x)
ox ES “4 ey (2.38)
OY se va
Similarly, if f (x, y, z) is a function of three independent variables x, y, z and x, y, z are
functions of
three new independent variables u, v, w given by x = F(u, v, w), y = G(u, v, w), z = H(u, v, w), then
_by chain rule, we have

a
 Functions of Several Real Variables 2.27
cok’
so? Of of.
of Or oh
of dy aor oe
4 oO du dx du dy du Oz Ou

—
Of _ Of dx , Of oy | Of daz
dv dx du dy du dz ov
of | Of dx , Of oy , OF dz
Ow dx dw dy dw dz dw
Solving the above system of equations by Cramer’s rule, we get

ofldu af/av af/aw

of_ 1 | Of. 9,2) =+)ay/au ady/av day/aw
ax J |O(av,w) | J
dz/du dz/dv dz/ldw

afldu ofldv dof/aw

aegeedx/du odx/dv adx/dw
dy
5 J7 | seh).
d(u,v,w) O(u,v,aew).
J |ppseuiuass)
Oz/du dz/dv dz/dw.

5 - 5 Of/du odfldv odfldw

dx/du dx/dv dx/dw (2.39)
bl = Bp deD) |
v, w)_
J |d(u, A(u,v,w) |Sik
| ie) J )
dy/du dy/dv dy/dw

AOx/du dx/dv dx/dw

where J=|dy/du dyldv dy/dw
dz/du dz/dv = az/dw

is the Jacobian of the variable ansformation.

Example 2.19 If z=/ (x, y),x =r cos 0, y =r sin 6, then show that

of f of\* 1 far?
(z) (¥) (3) (35)
Solution The variables of transformation are r and @. We have

dx/dr odx/d0 _|cos@ -—-rsin@

~lay/ar ay/d0| sin @ rcos @

acf.y) _|ef/or _|af/ar —afiae

af/oe ; af 3
d(r,9) |dy/dr dy/d0 sin@ rcos@ 56
5, ~ Sin Gan

_ | flor of/de : Of/dr odf/de of of

—rsin @ =-—rsin@ ar — cos Oa.
| dx/dr 0x/00 / cos 6
» y
: '
e
= ‘ad a t aa
7 ‘a
ic. ¥ i wil 1 '1 yo a be
‘ 38),
a : we obtain
=. yy
a i ae
"
7
 2.28 Engineering Mathematics

of - 4)343) -cos Ooo. sin@ Of

ox J d(r, 8) or rr 00

Of O(f, x) ~ AF . cosO OF
aera | mS a ar BO
Squaring and adding, we obtain the required result.
Example 2.20 If u=/f (x, y, z) andx=rsin @cos @, Mie sin @sin @, z =rcos G, then show tha

of) (a) , (a (ery

Solution
(XY «(2 +2) +B AB) ate
The variables of transformation are r, @ and @. We have
2 (ar l (xy

| sin 8cos ) rcos @cos Wy) —rsin @sin 0

— 9%, YsZ) =|sin@sing rcos@sing@ rsin@cos@ |=r7sin @
O(r,0,@) |
cos @ —rsin@ Q

of /dr of/d0@ df /d@

OF, vue)
¥»2) sin@sing rcos@sing@ rsin @cos @
a(r, 0,0)
0, o)
cos @ —rsin @ 0
= r* sin?@cos off + rsin @cos @cos oe —rsin @ io

Of /dr of/de of/dd

ATES
of, ), sin 8 co S@ rcos@
: cos@ -rsin Osin wv)
a(r, 0,0)
6, o)
|
.
@os @ —rsin @ 0

= —r* sin?@sin of —rsin @cos

@sin oof —rcos ov

: Of /dr df /d@ of/dd

Af aay)!
noe. | |
sin@cos@ rcos @cos @ -—rsin @sin @
sin@sing rcos@sing rsin Ocos @

— r :a SI ELOS 0 -=<-
of
—_— a OF
r sin 2 O=-.
ee

Using Eq. (2.39), we obtain

Of _ 7 eee |-Cats: of
cos Ocos @ OF sin
Tsim g of
ox A(r, 0, d) r 06 ~=rsinO ao
Of _ O(f, x, z) ae cos Osin@ Cae co 6
dy aback eee eo of 5
Of O(f, x,y) _ _ sing of
az | 3 0, | -
et rE Oey
____ Squaring and adding, we obtain the We
i result.
a
 - Functions of Several Real Variables 2.29

balienof implicit functions

The functionf (x, y) = 0 defines implicitly a function y = @(x) of one independent variable,x.Then,
we can determine dy/dx using the chain rule. From f (x, y) = 0, we get
BE dy _
# 2 Of 4 of
~ Ox dy dx

or | dy x,(x.y)y)
oe _ ~f(fy yy’ Fy (%Y) = 0. (2.40)

The functionf (x, y, z) = 0 defines one of the variables x, y, z implicitly in terms of the other two
variables. Using differentials, we obtain

t= Fdx+ Lay4 L dz= 0. (2.41)

If we take y = constant, then dy = 0 and we get from Eq. (2.41)

of
> dx + of
> dz=0) dz\
‘or (#e ; =
_ (df/dx)
EER (2.42)

If we take x = constant, then dx = 0 and we get from Eq. (2.41)

=
a Pc—dz=0, on
or er (aflay) - 43
(2.43)

If we take z = constant, then dz = 0 and we get from Eq. (2.41)

Of ech dx\ _ (df/dy)

oe dx + By
— dy= 0, or SE) SS aPTOx) (2.44)

Multiplying Eqs. (2.42), (2.43) and (2.44), we obtain

dx dy | ax) ( dy dz) |
BA al eal = (2.45)
cal ral Fall _—

Example 2.21 Find dy/dx, when

2
¥ =
(i) flx,y) = 4 + 2-1-0.

a) f(x y) = In(x2 + y’) + tan7!(y/x) = 0.

of _ 2x and of _ 2y =—-—_—-

ax a? | dy b2"

Colne Oflox _ 1 b2x

Bee tone, ay 27”
Seat Le a0 (| ee J ek 2K= y
TT 4 y2/x2 ( 4 x? + y? x? +y x2 +?
 2.30 Engineering Mathematics

|
ZY +5

dy xy? 1 4ytle? (so eartNatnghe ep) SereYS:

:
Therefore, 1y a
dy Ofif /ddlox - 2x - ie y-2x
Tay x
Der are
dx of/fdy 2y+x

Exercises 2.2

1. Show that the function

od
F(xy)=4 fx? +y2-
(x, y) ca (0, 0)

0 Ge y= «0, 0)
has partial derivatives f, (0, 0), f, (0, 0), but the partial derivatives are not continuous at (0, 0).
2. Show that the function

2 2
| eee a (x,y)
f(x,yJ= 4 *-y Q) # (0,

0 , (%,y) = (0, 0)
possesses partial derivatives at (0, 0), though it is not continuous at
(0, 0).
3. For the function

y(x* —y?)
FOQsINRS pe? gy?” Es
0 » (x,y) = (0,0)
compute /,(0, y), f,(x, 0), f.(0, 0) and f, (0, 0),
if they exist.
4. Show that the function (HESS) een y2 is not differentiable at (O,
0).
5. Show that the function

(x? +y*) cos |—_1__ > \(X,¥) (0, 0)

f(x,y) = x* + y?

0 » (x,y) = (0, 0)
is differentiable at (0, 0) and that Deady are
not continuous at (0, Q). Does this resu
lt contradict Theorem 2.2?
Find the first order partial der
iva tives for the following functi
ons at the specified point:
6. f (x, y) =x* — x2y? + y* at | (—1, 1).
7. f(x, y) = In(x/y) at (2, 3).
8. f(x, y)=x*e"” at (4, 2).
9 f(x,y) = x/afx? + y? “at 6; 7).
10. f (x, y) = cor (x+y) at (1, 2). 11. f(x, y) = In Fon
e at (3, 4).
x? +y2 4%
tte CY, 2) (x? + 5? 427)-12 (QeT ay 13. f (x,y, z) = e*4)
“Ki J ; MP Rie oy
eit f(x, Y, 2) =(xy)
yee ary) Sin : +s
* at (3, 5, 27/2).
.
.
ez! ae, 1,23,
13. f Gy 2) =In (x+ Sy? : +22) at (2, 3,
4).
ws
 — eee tti(“‘i‘<i— SO
ee

Functions of Several Real Variables 2.31

Find dw/dt in the following problems.

16. wart y,x=(?- it, y= (2 + 1) atta 1.
“47, wart y* + 2°, xX = cos y=In(t+1),z=e' atr=0.

18. w=e* sin (y+ 22), x=t, y=1/t,z=

7. 19. waxyt yet myx, yale, Zale’.
AQ. w=ziny+y Inz+xyz,x=sint,y=Pr+ l,z=cos t att=0.

Verify the given results in the following problems:

21. Ifz=/ (ax

+ by), then oe a oe 10}
Ox dy

ain [(x* — y”)/(x? + y?)], then rays 0.

y

Ps Ifu=f(a%-y, y-—z, z—x), then UES UG

| ax dy dz
24. If z=f (x, y), x =rcosh 6, y=rsinh @, then

az)"
ax}
_(a2)"a _(8)
“Wey Ver
_r24 (\22)"
00)
| x OZ
25. Ifiz=y
+f (wu), as then baaitan Hu

26. Ifw=f (u,v), u= fx? +y2 ,v = cot (y/x), then

(2%) + sole) +3}

ee
27. lfz=f (x, y), x =u cos M—-v sin a, y=u sin @+ v cos G, where @ is a constant, then

az az
2B. Ifz=In(u2+0v), u=e**”, v=x+y’, then ay

Beit w= x? +y? 477 ,x=uUcosvU, y=U SiN DV, z= uD, then

wom
ow -— UD
Ow _ u
Ou Ov See. :

a
t+y? e
+z7)/@ +y +2), then
307 Tfw = sin” “u
Ow aw ow
a, +—— = tan w.
y——'Zz
Y Oy dz

Using implicit differentiation, obtain the iets

OL, 2, when x’ + y* = G, @ any constant, x > 0, y > 0.

= pa =. when cot” (x/y)+y° + 1=0,x>0,y>0.

ate, se

oy), , when Cos xy

ind ($) + cos yz + cos zx = 1,
ax),cand
eek[ a2)
33.
L352 Engineering Mathematics

(BB, m 58)
34. ss) and a , when x + 3xy —2y? + 3xz+ 2° =0.
dx) dy a

Ox Ox | 28)
35.

Using differentials, obtain the approximate values of the following quantities:

36. 4/(298)? + (401)? . 37. (4.05)! (7.97).

=. a + ee ae
38. cos 44° sin 32°. 39. Vis + 307 J9.01

40. sin 26° cos 57° tan 48°.

41. A certain function z =f (x, y) has values f (2, 3) = 5, f,(2, 3) =3 and f, (2, 3) = 7. Find.an approximate
value of f (1. 98, 3.01).
42. The radius r and the height h of a conical tank increases at the rate of (dr/dt) = 0.2°/hr and
(dh/dt) = 0.1"/hr. Find the rate of increase dV/dt in volume V when the radius is 5 feet and the height
is 20 feet.
43. The dimensions of a rectangular block of wood are 60”, 80” and 100” with possible absolute error of 3”
in each measurement. Find the maximum absolute error and the percentage error in the surface area.
44 - Two sides of a triangle are measured as 5 cm and 3 cm and the included angle as 30°. If the possible
absolute errors are 0.2 cm in measuring the sides and 1° in the angle, then find the percentage error in
the computed area of the triangle.
45. The sides of a rectangular box are found to be a feet, 6 feet and c feet with a possible error of
1% in
magnitude in each of the measurements. Find the percentage error in the volume of the box caused
by
the errors in individual measurements.
46. The diameter and the altitude of a can in the shape of a right circular cylinder
are measured as 6 cm and
8 | cm respectively. The maximum absolute error in each measurement
is 0.2 cm. Find the maximum
absolute error and the percentage error in the computed value of
the volu me.
47. The power consumed in an electric resistor is given by P =
E’/R (in watts). If E = 80 volts and
K = 5 Ohms, by how much the power consumption will c hange
if E is increased by 3 volts and R is
decreased by 0.1 Ohms.
If two resistors with resistence R, and Ry in
Ohms are connected in parallel, then the resistence of the
resulting circuit is R = [(1/R,) + (1/R3)]"'.
Find an approximate value
resistence that results by changing R, from 2 to 1.9 of the percentage change in
Ohms and R> from 6 to 6.2 Ohms.
49. Suppose that u = xze” and x, y, z can be
ae
respectively. = . measured with maximum absolute errors 0,1, 0.2 and 0.3
Find the percentage error in the computed value
of u from the measured values x = 3;
y=In2 andz=5. ,
50. If the radius r and the altitude A of acone are measured
with an absolute error of 1% in each measurement,
then find the approximate percentage chan
r= 3 feet and h = 4 feet.
ge in the lateral area of the€ cone
cone if
i the measured
| lues are
valu

2.4 Higher Order Partial Derivatives

Let z = f (x, y) be a function of two variable S and
let its first order partial derivatives exist
points in the domain of definition D of th e function f. Then at all the
, the first order partial derivatives are also
functions of x and y. We define the seco nd order
partial derivatives as
\
a
 Functions of Several Real Variables 2.33

. i ;

Set oe | |=fon 9) = sim f paceeeo vec)

oF 19 fof

*f _ a | of SLI pe
ee sty REY
Oy Ox 7 oy | = Syx (xX; y) =

(differentiate partially first with respect to x and then with respect to y)

eo. J i [pele EE Nata

dcau
oy an =I (x,
os a Ax |
(differentiate aie first with respect to y and then with respect to x)

(x, Ay) - ;
2 g e n F e a
of? ~ Oy
dy aEaE=fyy (x,y)= fima
“a
the TMS exist. The derivatives ry and f,, arc RYmixec Ves aie
ve s. T fr y ANG fyy éi are
point F(x, y), then at this poif,, | deri va ti
nt==f,x« That is, the order of differentiation material inris
case. There are (out partial derivatives of second order for f (x, y). If all the second order partial
derivatives exist at all points in D, then these derivatives are also functions of x and y and can be
further differentiated.

as 2.22 Find all the second order partial derivatives of the function
f (x, y) = In(x? + y*) + tan7! (y/x), (x, y) # (0, 0).
Solution We have

Be= s 4 + y ] + (y/x)?
I ye
y e
2x =
x ieee y-

2y l a
x, aa —|=
Jy ( > 2 ye + ae (4 ie + y?

Cx + y*) (-1) — (2x - y) (2y) jee — x2 - Axy

Syx (x, y)= cs(fy -2(755
aaa es + yee
(x* + y?)?

fiy(*,y) = 2 (fy) =————

mls
Ox
ae
re SONI
= (x*
ee + y*) (1) — (2y + x)
(x? + y?)?
aa (2x)
=
y?
—x* —4xy
(x° +97)?
£
2x-y ) _ (*+y7)@)-Qx-y) 2x) 2y? ~2x2 42
xx (x, y= =
= x (f x ae2wiih
=%,) - =
oa as (x? + y*)? (x? + y2)2

- jee. 0. | 2 ex (x* + y*)(2)-(2y+x

)(2y) 2x2 — L.
. ia”y yo t
Oy ex? Fy? OY )
(x* + y2)2
_ 2x0(x?— Dy? =Dy
ty?)
We note that tay=Sox:
Exan | le 2.23 For the function

— 3y?)
xy(2x*
f(x,y) = a ay? a! (x, y) # (0, 0)
0 > (x, y) = (0, 0)
Show that fy (0, 0) # yx (0, 0).
.
2.34 Engineering Mathematics

Solution We obtain the required derivatives as

fx (0, 0) = GAR 0, fy (0, 0) = Jim, ae =(

f,(x.0) Hai 1)= Jim, peer 22%.

= jim,
Now,

hiv. m= FF) SE ee —_
Jim, 2Ar— 0

fy 0, 0) = = Eee = lim, (ae = Jim, = ei

Hence, f,, (0, 0) # f,, (0, 0).
Example 2.24 Compute fry (0, 0) and fyx (0, 0) for the function

5.2
a re
0” » (x,y)= (0, 0).
Also discuss the continuity of fry and f,, at (0, 0). ,
Solution We have

—_ Iz f (Ax, 0) — f(0, 0) : ) Lees .

BOO = lin = 0, Fy (0, 0) = Jim LOO) = F0.0) _ 9
Ay
Bry tot SOE POY = in er
Ax Ax—0 [Ax + y? ] Ax —

fy(x,0) = fim, FAL) iggy 249)? y Ay 0 [x + (Ay)?] Ay

fy. = fim, 26°)Ax F0.0) |

| ~
Fyx (0, O) =
pr Sx, Ay) — f, (0,0
aim, So im Ay

Since f,,(0, 0) # Fyx(0, 0), fy and

fx are not continuous at (0, 0).
Alternative We find that for (x, y) # (0,
0)
y +5
Syn (%,tas. ee= Pey (Xs ¥).
 Functions of Several Real Variables hep,

Along the path x = my’, we obtain

y(t om) 2 1+ Sm
(2,y 0) Fox 2 =
9 Me y°(1 +m)? ~ (Ll+m)>-
Since the limit does not exist, f,, is not continuous at (0, 0).
Example 2.25 For the implicit function f(x, y) = 0 of one independent variable x, obtain
y= d*y/dx*. Assume that Try=Lx:

Solution Taking the differential of f (x, y) = 0, we obtain : ee ok, -2

ae ae Ai
{= dy | Ix

Therefore,

dy _ afd|-- pt) ft) -f EH)

“ -£|9 aban f2 =

= Ty life + (Son =Selicy + Cy )y ]

ie |
Substituting y’ = —f,/f,, we obtain
d*y Te des = Dike Sy Say + feTe
SS SS Lo Le Le fae ee since = :
dx* 1 Fyx yy

2.4.1 Hemegeneous Functiens

A function f (x, y) is said to be homogeneous of degree n in x and y, if it can be written inbeny
pees the following
one of Pee FOMOWE forms ,

DFGxAy) =a" fin 4

TALI ( T™ my (2.46)
(ii) f (x, y) =x" g(y/x) - a SE ue (2.47)
(iii) f (x, y) = y" g(x/y). Nol Awe’ x | (2.48)
Similarly, a function f (x, y, z) of three variables is said to be homogeneous. of degree n, if it can be

written as f (Ax, Ay, Az) =A" (x, y, 2), or f(x, y, Zz) = s"e(2 2)etc.
Some examples of homogeneous functions are the following:

f | | degree of homogeneity
x? + xy 2
tan“! (y/x) 0
I/(x + y) —]
Wet + y'42°) 4
tgs er +2) 3
—1/2
 “90 Engineering Mathematics

The function f (x, y) = (x? + y)/(x + y?) is not homogeneous.

An important result concerning homogeneous functions is the following.
Theorem 2.4 (Euler’s theorem) If f(x, y) is a homogeneous function of degree n in x and yandh
continuous first and second order partial derivatives, then

(i) 0
x Sy
id
ao = rf.
(2.4!

(ii) O°
2 f + 2xy
XY Conia 2 et7 e 2.5
|

x
*@

a sa =a(n— lf
; Ps be
—= —_ —— ii 4
(2.51
2.

Proof Since f(x, y) is a homogeneous function of degree n in x and y, we can wri

f(x, y) = x"g(y/x).
Differentiating partially with respect to x and y, we get

— meri 2 +x"9 (2 (-] = nx

0 | ¥
il=

(2
2) — yx
N= f y

“8 (2).

of Rial ok | _— «hl 2s J.
y =2'8(2) (4) = (3):
Hence, we obtain

]
roy = 0 mete (2) -aerte{2) eomte(2) | = mete 2)=o |

Differentiating Eq. (2.49) partially with respect

to x and y, we get
pot of Oe. of
PGEe 00x um
*” Gna, "oy ee
and af or ee of
“@ydx dy *? ay?” ay’ oe.
Multiplying Eq. a.51) by x and Eq. aie
(2.52) by y and adding, we obtain
PS oy of of | a*f orF 0? LG
ax2 +(ra |
G] @] :
: ty] tyoak eh). "3p n(x Fay X
:
or oo oof O*f Our
| i,
A lege dy Yar =Mn— DS.
Example 2.26 If u(x,y) = cos~ (zexX +
nesta
on: O<x, y< 1, then prove that

)
tty ses 5 cot u.

pervtion For all x, yO x, yrs], (x ‘ig yf Vx + Vy] < 1, so that u(x, y) is defi
ned. The given
ion can be written as
 Functions of Several Real Variables 2.37

Conn = x[1 + y/x] E/E 1 + (y/x)

Vx+Jfy
Vx [1+ f/x] 1+ «/y/x

Therefore, cos u is a homogeneous function of degree 1/2. Using the Euler’s theorem for f= cos u and
y = 1/2, we obtain

x (cosu) + YF (cos u) = 5 COS1

Ou Ou ] Ou du
or
| =e (sin u) =
S : —_— y (sin uv) Jy = 7 COS Ms or x a as
[ i | i r SS — — _—— 5] cot u
ier | *

Example 2.27 If u(x, y) = x* tan“! (y/x) - y’ tan! (x/y), x > 0, y > 0, then evaluate

x?2 ==
07u 07u + 2 0*u
AS ey Pe
Solution We have u(Ax, Ay) = A*u(x, y). Therefore, u(x, y) is a homogeneous function of degree 2.
Using Theorem 2.4 (ii) for f= u and n = 2, we obtain

x
2
£8 42m OPits) selOLwll_aneo)
BE OU ue. Saal ayes
l)u = 2u.

Example 2.28 Let u(x, y) = [x° + y*]/[x + y], (x, y) # (O, QO). Then evaluate

xX
07u d*u du
axe ex Oy. Ox.
*[1 + (y/x)? function of
Solution We have u(x, y) = Sree Therefore, u(x, y) is a homogeneous

degree 2. Using Euler’s theorem, we get _

Ou Ou_
xa ar = 2.

Differentiating partially with respect to x, we obtain

Ou Oru O*u _ 9 du or Ofu O“u 2on
see. exe OXOR, ON ~ on oid
Example 2.29 Let f(x, y) and g(x, y) be two homogeneous functions of degree m and n respectively
where m # 0. Leth=f+ oh |. Oh
8. If eS 0, then show that f= @g for some scalar a
Solution Since f and g are homogencous functions of degrees m and n respectively, we obtain on
using Euler’s theorem

eSCS eaeaia 98 | oy
and Pee Og TE
Adding the two results, we get
_
of | 9g’ of a
 2.38 Engineering Mathematics

or x Oh, y Oh mp ng =0, where h=f+ g.

Ox dy

Therefore/ f= — 7~ b=
lere Ores f=- 9= ag, 2&8, where
whe a= — zs
— isa scalar.

2.4.Y Taylor’s Theorem

1 section 1.3.6 we have derived the Taylor’s theorem in one variable. If f (x) has continuc
derivatives upto (nm + 1)th order in some interval containing x = a, then

f(x) =f (a) + (x-a) f’(a)+...4+ (5-3) 60(a) + R, (x) (2.!

where R,(x) is the remainder term given by

Ry (x) = FOO pone 6) = FAO praevia g O(x-a)],a<&<x,0<0<1. (25

We now extend this theorem to functions of two variables.

Theorem 2.5 (Taylor’s theorem) Let a function f (x, y) defined in some domain D
in R?* ha
continuous partial derivatives upto (n + 1)th order in some neighborhood of a point
P(xp, yo) in !
Then, for some point (xg + A, yo + k) in this neighborhood, we have

F (% +h, yo + k) =f (x9, yo) + Ge + t2) F020.90) “ n(

r2 + Ke) f(%o, yo)

tet ii(pe
(d cGime
ae rt)” d)gene f(X0,¥0) es
+ R, @355
where R,, is the remainder term given by

1 5 5 n+l

“n= re dl Gaur J(%9 + Oh, yo + Ok),0< O<1. (2.56

Proof Let x = x9 + th Y = Yo + tk, where the
| parameter ¢ taketakes
Define a function @(t) as (1) = f (x, y); =f (x9 +: th, yo + tk), f in the
values in the i rval [0, !].
inte
Using the chain rule, we get

4,Ffdx Of dy of of
a FS oe a at kay = (422 +k : 2)
o”( t))= Ga
“(t oO usof * Pca
OF n+ = . (12402)a 7
dy
eck ylor’s theorem for a function of one variable (see Eq.
(2.53)) with = 1 afd a = 0, we

; 0(1) = (0) + p’(0) + e @°(O) +... + A

o”)
(0) + Le es gl) (9) (2.57)
Kae (n+ 1)!
 Functions ofSeveral Real Variables 2.39

where ~(O) = f (x9, Yo)

O(1) =f (x9 + h, yo + k)

= (toed) f(%9,¥0),@=1, 2,.

| n+]
gi"*D(g) = [1 + ‘2 | f (xo + 9h, yo + Ok),O< O< 1.
Ox dy
Substituting the expressions for (1), (0), ¢’(0),.. ., 9(0) and 9*"(@) in Eq. (2.57), we obtain
the Taylor's theorem for functions of two variables as given in Eqs. (2.55) and (2.56).
Substituting x = x) + h, y = yo + kK in Eq. (2.55), we can also write the Taylor’s theorem as

f(x,y) = f(xo,
yo) + [(x-a 2 0-90) 2[f0.90)

+ 37 |(4 — x0) SL= rat

+ apt
y -Yo. oy flXO» YO )

at a= Hlo-me+o-wZ
0d a |" ff(0, Yo) + Ry, (2.58)

where
co i
7 (n + I)! ee
0” Ox Se
caer
AO |
n) (2.59)
oe

and €= (1 — @)xo + Ox, nN=(1 — 8) yo + Oy, 0< O< 1.

For n = 1, we get the linear polynomial approximation to f (x, y) as

f (x, y) =f (Xo Yo) + %-— 0) fe + — yo)fy (2.60)

where the partial derivatives are evaluated at (Xp, yo). This equation is same as the equation (2.28)
which was obtained using differentials.
n = 2, we get the second degree (quadratic) polynomial approximation to f (x, y) as
For
f (x, y) =f or Yo) + (* — x0)fe+ Y — Yo)fy
+ [Ce 0)? fart 22 - x0 Y- Yo )fey + (9 — ¥0)? fy] (2.61)
where the partial derivatives are evaluated at (xo, yo).

aa ?* = f(0,l
0) + cssi
9 eO)+ >ietyxy
f (0, 0)

Masi
eit, lalla
ne ni
ale crtd) f(0, 0) + R, (2.62)
2.40 Engineering Mathematics

n+l

where Pitz
R, = ERG cHO
> +y sys? OxOy)- Oy),
2) f(@x, 0 BX 1,

(b) When lim R, = 0, we obtain the Taylor’s series expansion of the function f (x, y) about the
n—>00
point (x9, yo).
ylor’s theorem can be easily extended to functions of m variables f (x), x2, . . ., x,,).

fror estimate

Since the point (€, 7) or the value of @ in the error term given in Eq. (2.59) is not known, we canno
evaluate the error term exactly. However, it is possible to find a bound of the error term in a giver
rectangular region R: |x — xy |< 6,, |y — yo| < 63. We assume that all the partial derivatives of the
required order are continuous throughout this region.
For n = | (linear approximation), the error term is given by

R, = * [(x — x0) * fox + 2(x — x9) (y — yo) int OC — yo) fyy] (2.63)
where the partial dirivatives are evaluated at the point (¢, 7) = [xp + A(x — x9), yo
+ OL = yo)I,
0 < 6< 1. Hence, we get

PRLS 5 C140? Lf + 21 *-2011¥-yol lfyl +1 y—yo |? Lf 1

If we assume that

B= max [| fx |, |fy |,|Ay |] for all (x, y) in

R, then we obtain
B |
[Ri |< > []x-x0P +2|x-x9||y-y0 |+] y—yo PI

B,
= 7 (lx-x0l+ly-yl
_
Ps 3 16+ 57 (2.64)
This value of | R, | is called the maximu
m absolute error in the linear approxima
about the point (xo, yo). tion of f (x, y)
|
For n = 2 (quadratic approximation), the
error term is given by

Ra = =; (Ce —%5)*fone+ 3(x — x0) (9 — yp) Frxy + 30 — 45) ( — yo)" fay

+ (9 — Yo)? fryy] (2-65)
where the partial derivatives are evaluate
d at the point
(S, 1) = [xo + A(x — Xo), Yo + O(y
—yp)], 0 < Oe].
From Eq. (2.65), we get

rae. 2
|Rols ¢ Lleol" fer |+3 |x x01? | y - yo) |fay +3 |x — a0 11» yo
[? Lf
+] y—yo I?Ufees V1

- Be 3 12
Pa ml |
+3) 2>x| IY =¥o| +3 |x-xy1|y=yo |? +1 yy [7]
 Functions ofSeveral Real Variables 2.41

= ¢ (tml +ly-wI P< 2 6, +5, (2.66)

where B = max [ Weve , |Beccy , Foy .
fyyy |] for all points (, y) in R.
Remark 8

In a similar manner, we can obtain error estimates for approximations of functions of three or more
variables. For example, if f (x, y, z) is to be approximated by a first degree polynomial (linear
approximation) about the point (Xo, Yo, Zo), then we have

f (, y, Z) = P(x, y, z) =f (X%, Yo: 29) + [x - x) f+ O-yo A + (2 - WF]

where the partial derivatives are evaluated at (Xo, Yo. Zo): Lhe error associated with this approximation
is given by

R, = > [= x9)" fa + (y — 99) fry + (Z — 20) fez+ 2(% — X0) (Y — Yo) fay

+ 2(x — X9) (Z — 2)fee + 2(y — Yo) (z — Z0) Jaz, I

If we consider the region R: | x — x9 |< 6), | y- yo |< do, |Z— Z| < 63
and assume that B= max [| fx |, | Ay |, |fz ly | |fel Ac 1)
for all points (x, y, z) in this region, we can write

Ri |S 5 (1x-x0l+ly-yP<
l+|
2 6,z-z
+52+l
65)
Example 2.30 Find the linear and the quadratic Taylor series polynomial approximations to
the
function f(x, y) =2x° + 3y° — 4x? y about the point (1, 2). Obtain the maximum
absolute error in the
region | x —1|< 0.01 and|y—2|<0.1.
Solution We have
Tee y) = 2x7 + 3y? — 4x*y : FC,2) =18

fi,(&, y) = 6x"— 8xy | #2) =—10

fy (x, y) = Oy? — 4x° ote fC 32
fx (, y) = 12x — By ee 2 ——
fry (X, y) = — 8x BiG yess
Ty %,.¥) = 18y | biyy (h2)'= 36
Ton, ¥) = 12, [email protected])=-8, Sny™ y) =0,f,,@, y) = 18.
The linear approximation is given by

f(y) =f (1, 2) + (@- DAC, 2) + © = 2) 6.1, 2))

| = 18 + (x— 1) C10) + (y — 2) (32) = 18 - 10~@~ 1) + 32(y — 2).
a Ma ximum absolute error in the linear approximation is given by
ee [Riis 3 Ule-11+ly-217 s 2 (01) + 0.172 = 0.0
“=i
0605 B
2.42 Engineering Mathematics

on
where B = max [ |fx |, |fy , |fy | ] in the given regi| x - 1 |< 0.01,|y-—2|< 9.1.

Now, -2) -4 |
max | f,, | = max | 12 x - 8y | = max | 12(7- 1) —8(y
< max [ 12|x-1|+8|y-2|+4]
=4.92
max | f,, | = max | — 8x | = max |8(x — 1) + 8 | S max [8 | x -—1 | + 8] = 8.08
max | f,, | = max | 18y | = max[ 18(y - 2) + 36] Ssmax [18 |y—2|+ 36 ] = 37.8.
Hence, |B |= 37.8 and |R, |< 0.00605(37.8) = 0.23.
The quadratic approximation is given by

fay) =fC, 2)+164-DAC, 2+ 0-DAC, 2)

+ 4 [@- 1)? £0, 2) + 2-1 (W—2) fy (1. 2) + © — 2)’ fy CL, 20]

= 18 —10(¢— 1) + 32G'—2)+ = [-4(x— 1)*- 16(@- 1) (v- 2) + 369- 2)"]

= 18- 10@-1)4+32(y -—2)-2[@- 1)*
+ 4(¢ - 1) (y — 2) = Oy — 2y').

Using Eq. (2.66), the maximum absolute error in the quadratic approximation is given by

|RoI< SU|x-1]+]y-21 P< 2 uy = 2 001331)

where B= max [ |Fro |;ies |,kbs , kecse |] = max [ 12, 8, 0, 18 ] = 18.
Hence, we obtain

| 2 |< B (0.001331) = 0.004.

Example 2.31 Expand f (x, y) = 21 + x—-20y + 4x7 + xy + 6y” in Taylor series of maximum order
about the point (—1, 2).
Solution Since all the third order partial derivatives of f (x, y) are zero, the maximum
order of the
Taylor series expansion of f (x, y) about the point (—1, 2) is
two. We obtain

f(x,y) =f(-1, 2) + (.+ 1) L + (y — 2) 2 hc, 2)+ 4 ( +1) 2 + (y — 2) 3] fC, 2).

We have
F(-l, 2)=6, f,@, y)=14 8x4+y, f.(-1, 2) =-5.
fy y)=-204+x+4 12y, fl, 2) =3,
Fur % ¥) = 8, fy (& y) = 1, fy (x, y) = 12.
Therefore,

F(% y) = 6 —S(x + 1) + 3-2)

+4 +1)?
+ x+ 1) (y — 2) + 6(y - 2).
 iables 2.43
Functions of Several Real Var

This is an rearrangement of the terms in the given function.

Example 2.32 The function f (x, y) = x* — xy + y’ is approximated by a first degree Taylor's
polynomial about the point (2, 3). Find a square | x - 2|< 6,| y- 3 | < 6 with centre at (2, 3)
all points within this
such that the error of approximation is less than or equal to 0.1 in magnitude for
square.

Solution y,
We havef, = 2x — fy =2y-2, f,=2, fy =- 1, ie a).
The maximum absolute error in the first degree approximation is given by

[Ri |< 2 (|x-2|+ly-31P

where B = max [ eee | Wiese |, |Fas |] = max [2 Ite 2| =H

We also have | x - 2| < 6,| y—3| <6. Therefore, we want to determine 6 such that

< 0.1, or 5< 0.025

|R, |< 5 [5+ 6]* <0.1, or 462 = 0.1581.

Example 2.33 If f (x, y) = tan”! (xy), find an approximate value of f (1.1, 0.8) using the Taylor’s
series (1) linear approximation and (11) quadratic approximation.
Saintion Let (xp, yo) = (1.0, 1.0), k= 0.1, K=— 0.2, Then
f(1.1, 0.8) =f (1 + 0.1, 1 = 0-2).
(1) Using the Taylor series linear approximation, we have

se he
FfCUA, 0.8)= fC, D+ eS
[ Ap +k ra, 3)

From
f (x, y) = tan '(xy), we get

fd, 1) =tan“"(1) = 2/4 = 0.7854

(% Y) = <a
fx Ix, N= 4, A yi= aoe Ad. 1) = t

Therefore,

f (1-1, 0.8) = 0.7854 + ‘31 +46 02)| Biren,

(ii) Using the Taylor series quadratic approximation, we have

‘We have

Pay) 2xy° 5 fx (l,l)

= ———_—_—— = — =; f, (x, y) = — 2x7y
(1 fouls I) = -5

aes) Cry) 2 (11 roe

— xy ye 12 1) = 0.
(1 ate ea)

-_ -Therefore, using the result of (i), we obtain

 Engineering Mathematics

f (1.1, 0.8) = 0.7354 + 4 {0.01 (-3 + 2(0.1) (— 0.2) (0) + (0.0 4) - )

2 tf

= 0.7354 — 0.0125 = 0.7229.

The exact value of f(1.1, 0.8) to four decimal places is 0.7217. Thus, the accuracy Increases
as the order of approximation increases.

Exercises 2,3

Find all the partial derivatives of the specified order for the following functions at the given point:

f (x, y) = [v- y]/ [x + y], second order at (1, 1).

f(x y) =x In y, third order at (2, 3).
Ff (x, y) = In [U/x) - (1/y)], second order at (1, 2).
f (x, y) = e* In y + (cos y) In x, third order at (1, 77/2).
f(y) = e8™@, second order at (77/2,1).
f (x, y, z) = [x + y)/[x + z], second order at (1, -1, 1).
ye 2
- JG, y,z) = e* *” ** , second order at (-1, 1, —1).
- f @, y, Z) = sin xy + sin xz + sin yz, second order at (1, 2/2, 7/2).

eC
NY
Fw~f(xy,2= i + —Foe
eon
Aum + =, second order at (1, 2, 3).

a7 Ff
= EQ yy 2) == xy?
_— cya a) ; at any point (x, y, z) # (0, 0, 0).
' t ie |

x? y(x-y) ;
11. For the function 7(2y)= x7 4+y? ’ (x, y) # (0, 0)

Oise) =1(0):0)
show that f,, # f,, at (0, 0).
Ez. Show that Sey = fx for all (x, y) # (0, 0), when Fi, y) =
x’.
13. Show that fy = f,. for all (x, y) # (0,
0), whenf(x, y) = log [x + Vy? 4x2],
14 Show that f,,. = fyzx for all (x, y, z), whenf(x, y, z) = e®” sin Z,
Is. Show that fryyz = fyyxe for all (x, y, z), whenf(x, y, z) = z2ex+)?-
16. If z = e* sin y + e” cos x, where x and
y are implicit functions of 7 def; ned by the equations
eo e+ eth +t— -1=Oand yt +yst+1+y=0, then find dz/dt at
t= 0.
17. If x and y are defined as functions of u, v
by the implicit equations x* — y* + 2y?+ 3y?
and 2x? — y? — 4? + 4y2- 2 = 0, then find Ax/d — 1= 0
u, Ay/du, A2x/A u2 and a “vidu?.
18. If u and v are defined as functions of x and y by
the implicit equations 4.x? + 3y* —z* yu? +0 = S
Bx° = 2y* $ 22 + ue? + 2p? = 14, then find
(Au/dx),. and (dv /Oy),, atx = 1. y=-l,z
Assume that u > 0, v > 0. =2.

If x y1—y? +yV1—x? =c, c any constant, |x|<1,| y| <1, then find dyldx and d2y /dx’
Find dy/dx and d*y/dx* at the point (x, y) = (1, 1), for e” —e*
+ xXp= |.
 s 2.45
Functions of Several Real Vartable

21. Ifz=u', u=(x/y), v = xy, then find 0°z/0

x’.

22. Ifu=In(I/r), r= 4(x-a)? +(y—b)* , then show that u,, + Uyy = 0.

23. If F=f (u,v), 4=y+ax,v=- y — ax, a any constant, then show that F,, = a Pyys
24. If f(x, y) =x log (y/x) , (x, y) # (0, 0), then show that xf, + 2xy fry + ¥ Sy = 0.
25. If f (x,y) = yl’ + y’), (x, y) 4 (0, 0), then show that f,. + Ay = 9.
26. Find @ and B such that u(x, y) = e+” satisfies the equation U,, — 7Uyy + 12uyy
= 0.
27. Ifz=f(u, v), u = x/(x* + y’), pe wie 4 y?), (x, y) # (0, 0), then show that

uu + Zp =X" + Y°)? (2 + By):

28. Ifx=rcos 0, y=r sin @, then show that
020 cos 2 0 2 7
dx dy r? au) dx dy De

Using Euler's theorem, establish the following results.

: x? +y? @)
29. fe =sin— | — | then ee yee Stan
X+y ax dy

ae 2
ey du Ou
30. If u = log
| |————— |,
:thenen x—+y—=0O.
oe Yay

: as% Dos Ou Ou
31. If w= y? —¥- sin"'{ *) then ae al

eo
ys aes 0u 0u 1 2
20° 2
O*u 2_ 0O-u
a2
32. If «= —— then x —+y— =u and
x* ——+2xy —— + — -— ()
: yet eer Ox > Oy dx? ” Ox dy f dy?
3 3 9
33. If tan u = eae then pot ee Ee 2u and
x-y Ox dy
2 2
x _ + 2xy — +y? at = (1 —4sin*u) sin 2u.
x
34. Obtain the Taylor’s series expansion of the maximum order for the function
Gy) = x” + 3y* — 9x — 9y + 26 about the point (2, 2).
35. Obtain the Taylor’s linear approximation to the function f(x, y) = 2x? - xy + y? + ee 4y +1
about the point (1, 1). Find the maximum error in the region |x + 1] <0.1,| y-1 |< 0.1.
36. Obtain the first degree Taylor’s series approximation to the function f(x, y) =e” In (x + y) about the point
(1, 0). Estimate the maximum absolute error over the rectangle | x — 1 |< 0.1, |y |< 0.1.
37. Obtain the second order Taylor’s series approximation to the functionf(x, y) = xy? + y cos (x — y) about
the point (1, 1). Find the maximum absolute error in the region |x — | |< 0.05, | y-1]<0.1.
38. Expand f
(x, y) = xfx+y in Taylor’s series upto second order terms about the point (1, 3). Estimate the
maximum absolute error in the region | x - 1 |< 0.2, | y-3]<0.1.
39. Obtain the Taylor’s series expansion, upto third degree terms, of the function f (x, y) = e?** about the
point (0, 0). Obtain the maximum error in the region |x |< 0.1, |y |< 0.2.
40. Expand f (x, y) = sin (x + 2y) in Taylor’s series upto third
order terms about the point (0, 0). Find the
| maximum error over the rectangle |x|<0.1,]y |<0.1. )
2.46 Engineering Mathematics
‘1a in Tavior’s series ‘der terms
upto second order terms aabout the point (71/4, 7/4), ip
| f (x,
41. Expand iy y) —<sinx:
= sin x sin y in Taylor's series up
0.1, |y— — 2/41/4) |< 0,1.
the maximum error in the regi| on |x - 2/4 |<
+y :
+2 2 | , in Taylor series; upto first order
“rst order
) re the point (2. |
abou
about
termsSer
terms 9
nd
f (x,
42. Expa y,z)= yx
|<0.1,}z |
Obtain the maximum error in the region | x - 2 |< 0.1, |y-2
about the point (1, 3, 3/7
J Apanc f (x,
43.ye Expand y, z) -= «xy
‘4 J “ + :yz +xz in, Taylor's series upto first order terms
ee 4/9 , 0) : |

Obtain the maximum error in the region |x-1[<0.1,|y-3|<9.1,]2 |

sich i about the point (0, 0. ¢
44. Expanfd(x, y, z) = e* sin (x + y) in Taylor's series upto second
; rh a yr » TM i! |
Ls ‘

Obtain the maximum error in the region |x |< 0.1, |y |< 0.1, |z |
point (0, |, 1/2
45. Expand f (x, y, z) = e* sin (yz) in Taylor’s series upto second order terms Ske
Obtain the maximum error in the region |x| <0.1,|y-1|<01,] 2-7/2] <9}.

2.5 Maximum and Minimum Values of a Function

Let a function f (x, y) be defined and continuous in some closed and bounded region R. kt (a, oy
an interior point of R and (a + h, b + k) be a point in its neighborhood and lies inside K. We defin
the following.
(i) The point (a, b) is called a point of relative (or local) minimum, if

f(at+h, b+ k) 2 f(a, b) (2.67%

for all h, k. Then, f (a, 6) is called the relative (or local) minimum value.
(ii) The point (a, 5) is called a point of relative (or local) maximum, if

f(a+h,b+k) < f(a, b) (2.676

for all h, k. Then f (a, b) is called the relative (or local) maximum value.
A function f (x, y) may also attain its minimum or maximum values on the bounda
ry of the region
The smallest and the largest values attained by a function over the entire
region including th
boundary are called the absolute (or global) minimum and absol
ute (or global) maximum value
respectively.
The points at which minimum / maximum values of the
function occur are also called points 0,
extrema or the stationary points and the minimum and the maximum value
s taken together are callec
the extreme values of the function.
We now present the necessary conditions
for the existence of an extremum of a func
ton.
Theorem 2.6 (Necessary conditions for a function
to have an extremum) Let the function
F (x, y) be continuous and possess first order partial derivatives
conditions for the existence of an
at a point P(a, b). Then, the necessary
extreme value of f at t
he point P are f,(a, b) = 0oeand f,(a, b) = 9
Proof Let (a+h,b+k)
of maximum: if 1oth
bea poinnt t in
in t
the neighborhood
ent hood of
of of thethe point
int P(a,
P(a. b).
} Then, P will bee aaa popolpoint
ill be

Af=f(at+h,b+k)-f(a,b)<0 forall h, k
and a point of minimum, if (2.68)
Af=f(a+h, b + k)—f (a, b)>
0 for all h, k. (2.69)
Using the Taylor’s series expansion
about the point (a, b), we obtain
 Functions of Several Real Variables 2.47

flat+h, 0+ k)= f(a, b) 4 (Af. + Kfy

oapy + 5 (hf. + 2hkfey + kf yy) cap) dear. eet)

Neglecting the second and higher order lerms, we pet

Afe hf a, b) + kf,(a, b). (2.71)
The sign of Afin Eq. (2,71) depends on the sign of hf,(a, b) +k f,(a, b) which is a function ofh and
k, Letting A — O, we find that Af changes sign with k. Therefore, the function cannot have an
extremum unless/, = 0, Similarly, letting k —» 0, we find that the function fcannot have an extremum
unlessf,= O |
Therefore, the necessary conditions for the existence of an extremum at the point (a, b) is that

I(a,b)=0 and f, (a, b) = 0. (2.72)

A point P(a, b), where f, (a, b) = 0 and f(a, b) = 0 is called a critical point or a stationary point. A
point P 1s also called a critical point when one or both of the first order partial derivatives do not exist
at this point,
Remark 9
To find the minimum/maximum values of a functionf,we first find all the critical points. We then
examine each critical point to decide whether at this point the function has a minimum value or a
maximum value using the sufficient conditions.
Theorem 2.7 (Sufficient conditions for a function to have a minimum/maximum) Let a function
/ (x, y) be continuous and possess first and second order partial derivatives at a point P(a, b). If P(a, b)
Is a critical point, then the point P is a point of
relative minimum if rt—s* >Oandr>0 (2:73a)
relative maximum if rt—s? > 0 and r<0 (2.73b)
where r = uni Q, bh), S = fry(a, b) and ¢ = fyy(a, b).

No conclusion about an extremum can be drawn if rt — 5? = 0 and further investi gation is needed.
If rt —s* < 0, then the function f has no minimum or maximum at this point. In this case, the point
P is called a saddle point.

Proof Let (a +h, b +k) be a point in the neighborhood of the point P(a, b). Since P is a critical point,
we have f,(a, b) = 0, andf,(a, b) = 0. Neglecting the third and higher order terms in the Taylor’s series
expansion of f(a + h, b + k) about the point (a, b), we get
) Lica |
Af=f(a+ h,b+ k)-fla, b)= > [h*fx(a, b) + 2hk fry (a, b) + k*f,y (a, b)]

= 4 (Wr + 2hks + 1 = 5 (hr? + 2hkrs + rt)

= - (hr + ks)? + (rt = 8*)). (2.74)
(h+
rks)’ > 0, the sufficient condition for the expression (hr + ks)? + (rt — s’) to be
positive
is that rt — 5? > 0.
=

2 Af>0 if r>0
y and Af<O if r<Q.
Therefore, a sufficient condition for the critical point P (a, b) to be a
5 ’q
2.48 Engineering Mathematics

point of relative minimum is rf— s*>0O and r>0

point of relative maximum Is rt — s*>0O and r<0O.
If rt — s* < 0, then the sign of Af in Eq. (2.74) depends on A and k. Hence, no maximum/minimy,
of fcan occur at P(a, 6) in this case.
If rt —s* =O orr=t=s = 0, no conclusion can be drawn and the terms involving higher orde
partial derivatives must be considered.
Remark 10
(a) We can also write Eq. (2.74) as

Af = = (ker? + Qhkst + h?rt] = * (kt + hs)? + (rt- s*)h?].

Hence, a sufficient condition for a critical point P(a, b) to be a

point of relative minimum is rt — s* > 0 and r > 0

point of relative maximum is rt — s* > 0 and t < 0.
From these conditions and Eqs. (2.73a, 2.73b), we find that when an extremum exists, ther
rt—s* > 0, and both r and t have the same sign either positive or negative.

(b) Alternate statement of Theorem 2.7

A real symmetric matrix A = (a,) is called a positive definite matrix, if
x’ Ax>0 for all real vectors x # 0
fl fn

or 2 Py ay x; x; > 0 for all x;, x; (see section 3.5.3).

A sufficient condition for the matrix A to be positive definite is that the minors of all its
leading
submatrices are positive. Now we state the result. Let

heiress
AY t

where r= f,,(a, b), 5=f,y(a, b) =f,,(a, b) and t=f, (a, b). Then, the function f (x, y) has a relative
minimum at a critical point P(a, b), if the matrix A is positive definit
e. Since all the leading
minors of A are positive, we obtain the conditions r > 0 and rt— 52>
0.
The function f(x, y) has a relative maximum at P(a, b), if the matrix B =— A = oe | is
=e +t
positive definite. Since all the leading minors of B are
and 2 3
rt — 5° > 0, that is r<O and rt—s?>0.
positive, we obtain the conditions — 7 > 9
|
_ This alternative statement of the Theorem 2.7 is useful when we consider the extreme values of
the functions of three or more variables. For example, for the function f(x, y,
z) of three variables.
we have

tiny Ixy Tixz

A = Fyx Syy Sys |

Tet Fry Sez
 Functions of Several Real Variables 2.49

where fx =f tu = Sra fy = fiz. The matrix A or the matrix B = — A can be tested whether it 1s
positive definite, to find the points of minimum/maximum. Therefore, a critical point (a point at
whichf, = 0 =f = f;)
(i) is a point of relative minimum if A is positive definite and f,,, fyy,fr,are all positive.
(ii) is a point of relative maximum if B =— A is positive definite (that is, the leading minors of
A are alternately negative and positive) and fi... fy» fe, are alll negative.

Example 2.34 Find the relative maximum and minimum values of the function

f (x, y) = 2(x? - y*) — x* + y’.

Solution We have
f= 4x -—4° = 0, or x = 0, +1
fy =— 4y + 4y’ = 0, or y =0, 41.
Hence, (0, 0), ( 0, +1), (+1, 0), (1, +1) are the critical points. We find that

r =f =4-12x*, s=fy=0, t=fy=-4+ 12y?

and rt —s* =—16(1—3x°) (1 — 3y°).
At the points (0, 1) and (0, —1), we have rt — s* = 32 > 0 and r=4 > 0. Therefore, the points (0, 1)
and (0, —1) are points of relative minimum and the minimum value at each point is —1.
At the points (—1, 0) and (1, 0), we have rt — s* = 32>0Oandr=—8<0. The points (—1, 0) , (1, 0)
are points of relative maximum and the maximum value at each point is 1.
At (0, 0), we have rt — s’> = — 16 <0. At (+1, +1), we have rt — 5? =— 64 < 0. Hence, the points
(0, 0) (+1, +1) are neither the points of maximum nor minimum.
Example 2.35 Find the absolute maximum and minimum values of
f (x, y) = 4x* + Sy? — 8x — l2y
+4
over the rectangle in the first quadrant bounded by the lines x = 2, y = 3 and the coordinate axes.
Solution The function f can attain maximum/minimum values at the critical points or on the
boundary of the rectangle OABC (Fig. 2.4).

a) ra
. Fig. 2.4, Region in Example 2.35,
2.50 Engineering Mathematics

We have f, = 8x - 8 = 0, f, = 18y — 12 = 0. The critical point is (x, y) = (1, 2/3). Noy

r=f,=8, S=fy=0, t=fy = 18, rt—s°= 144.
Since rt — s? > 0 and r> 0, the point (1, 2/3) is a point of relative minimum. The minimum value ;,
Fh, 223) ==4.
On the boundary line OA, we have y = 0 and
f (x, y) =f (x, 0) = g(x) = 4x? — 8x + 4, which js ,
function of one variable. Setting dg/dx = 0, we get 8x —- 8 =0 or x= 1. Now d*g/dx*=8>0. Therefore
at x = 1, the function has a minimum. The minimum value is g(1) = 0. Also, at the corne,,
(0, 0), (2, 0), we have
f (0, 0) = 2(0) =4, f (2, 0) =9(2) =4.
Similarly, along the other boundary lines, we have the following results:
x=2: h(y) = 9y*- 12y +4: dhidy = 18y - 12 = 0 gives y = 2/3: d*hldy’ = 18 > 0. Therefore.

y = 2/3 is a point of minimum. The minimum value is f (2, 2/3) = 0. At the corner (2, 3), we
havef (2, 3) = 49.
y=3: g(x) =4x* — 8x + 49; dg/dx = 8x — 8 = 0 gives x = 1; d*gldx* = 8 > 0. Therefore, G=1 isa
point of minimum. The minimum value is f (1, 3) = 45. At the corner point (0, 3), we have
f (0, 3) = 49.
x=0: h(y) = 9y* - 12y + 4, which is the same case as for x = 2.
Therefore, the absolute minimum value is — 4 which occurs at (1, 2/3) and the absolute maximum
value is 49 which occurs at the points (2, 3) and (0, 3).

Example 2.36 Find the absolute maximum and minimum values of the function
f (x, y) = 3x7 + y* — x over the region 2x7 + y*< 1.
Solution We have f, = 6x — 1 = 0 and f, = 2y = 0. Therefore, the critical point is (x, y) = (1/6, 0).
Now, r=f,=0, s=f,=0, t=f, =2, rt—s*=12>0.
Therefore, (1/6, 0) is a point of minimum. The minimum value at this point is f(1/6, 0) = — 1/12.
On the boundary, we have y* = 1 — 2x”, _1/,/2<x< 1/-/2. . Substituting in f (x, y),
we obtain
f (% y) = 3x + (1 - 2x7-x
) =1-x 422
=p(x)
which is a function of one variable. Setting de/dx = 0, we get

£8 -2-1=0,08 Kas. Also #8 230

For x = 1/2, we get y’ = 1 - 2x? = 1/2 or y= + L/J2. Hence. the points (1/2, + 1/V2)
are points of minimum.: The minimum value is FOR, £ iV2)}.= 3/4.
5 = 3/4. At th € vertice
ces,s, wewe have
ha
f (2,0) = G— V2)/2, f(—WV2,0) = 3 + V2)/2, FO, + 1) = 1. Therefore, the given function
has absolute minimum value — 1/12 at (1/6, 0) and absolut
e maximum value (3 + /2)/2 at (- 1/ 2, 0).
Example 2.37 Find the relative maximum/minimum values of
the function
f (,y,2) = x* + y4 + 24 yy,
Solution We have

oe
fx= 4x" — 4yz =0, f,= 4y? — daz = 0, f= 42 — dry = 0.
ae ae Sys |
Therefore, x = VY =x P= xy or xyz = x*y*z? or xy*z" (xyz —- 1) = 0.
 Functions of Several Real Variables 2.51

Therefore, all points which satisfy

xyz = O or xyz = | are critical points. The solutions of these
equations are (0, 0, 0), (1, 1, 1), (41. £1, 1), (1, £1, £1), 1, 1, +1)
with the same sign taken for the
two coordinates. Now,

fax = 12x", Jyy =a ye, fz = 122°, Ty =—42, fz =— 4), Fyz = re

At (0, 0, 0), all the second order partial derivatives are zero. Therefore, no conclu
sion can be drawn.
12° 42 = 45)
We have A= |'—47 "129" “=4y)
—4y <=4y 1277
Depending on whether A or B=— A is positive definite, we can decide the points of minimum or
maximum. The leading minors are

IDs 44 ; ;
Mj = 1237,M,—| N= 1GKeryy = 22)
=47 12'y?
and M; =| A |= 192x? (9y*z* — x?) — 19224 — 64 xyz— 64x yz— 192 y*
= 192 [9x*y2z? — (x4 + y* + z4)] — 128xyz.
At all points (1, 1, 1), (£1, +1, 1), (£1, 1,+1), (1, +1, 1) with the
same sign taken for two coordinates,
we find that M, > 0, M, > 0 and M; > 0. Hence, A is a positive definit
e matrix and the given function
has relative minimum at all these points, since ln Of > 0, and £, > Q. The
relative minimum
value at all these points is same and is given by f(1, 1, 1) =—1.

Conditional maximum/minimum

In many practical problems, we need to find the maximum/minimum value of a function

Ff(%, x, . . . , X,) when the variables are not independent but are connected
by one or more constraints
of the form
(41,
%2.- +X) =0, 1=1,2,...,k
Where generally n > k. We present the Lagrange method of multipliers to find
the solution of such
problems.

2.5.1 Lagrange Method of Multipliers

We want to find the extremum of the function f (x;, x2, . . ., x,) under the conditions
Qj (%12,,+++ Xn) =O, (= 1,2,...,k, (2.75)
We construct an auxiliary function of the form
k
F(x}, X4,. - = #9 An» Ai, A>, oe o8y Ay) =f (1, AD, s hoy Xn) Us Pp»A; 0; (x, +X 3,2 * 8s Bene (2.76)

where A,’s are undetermined parameters and are

J _the stationary points of F, we have the necessary known as Lagrange multipliers. Then, to determine
conditions
 i Lith Engineering Mathematics

k Od.
LN ae eee tr ee (2.7
Ox; f=I Ox;

From Eqs. (2.75) and (2.77), we obtain (n + k) equations in (n : k) unknowns

My BQ, .. ., y
Ai, Abs + « & Ax Solving these equations, we obtain the required stationary points ijn Xo, . .. ee),
which the function f has an extremum. Further investigation is needed to determine
the exact natu;
of these points.
Example 2.38, ’ ur
Find the minimum value of x”2 + y”) + z” 2 subject to the condition
| “43
xyzwee
= a?
Solution Consider the auxiliary function
FG; y,z, 2) = x? + y" ee A (xyz — a>)
We obtain the necessary conditions for extremum as
lay
5
——= OF _» y
2x +XS Ayz == 0, —a = +Axz=0,Py oe 2 tay =
a—=2z+ XY ‘

From these equations, we obtain

Ayz = — 2x or Axyz = — 2x"
A xz =— 2y or Axyz
= — 2y ”
A xy =— 2z or Axyz = — 22”.
Therefore, x* = y? = z?. Using the condition xyz = a°, we
obtain the solutions as (a, a, a).
(a, — a, — a), (— a, a, — a) and (-a, — a, a). At each of these points, the value
of the given function is
x? + y* + 27 = 3q?,
Now, the arithmetic mean of x2, y*, 2 is AM = ons
y? + z7)/3
the geometric mean of x”, y?, z* is GM = (x*y7z*)
15 — gq?
Since, AM > GM, we obtain x? + y? + 22 > 3,2.
Hence, all the aboye points are the points
of constrained minimum and the minimum
x+ y + 2 is 3a. value of
Example 2.39 Find the extreme value
S of f (x, V5 2) = Qe 4 3y + z Such that x° + y? = 3
and x+z= 1.
Solution Consider the auxiliary function
F(x, y, z, A, Ay) = 2x + Sy+Z
+A, (x7 + y?— 5) + Ay (x+7
z- 1).
For the extremum, we have the nec
essary conditions
OF _ OF
Oy 2 + 2Aix + Az =0;
ay omy
OF
=, oe ot Ae = 0.
From these equations, we get

ee 3+2A,y=0 and 1+2A,x=0

or
x=— I/(2A,) and y =~ 3/(
2A,).
Substituting in the constraint x2 + y?
= 5, we get
 Functions ofSeveral Real Variables 2.53

i 9 | ] !

For A, = 1/2,
we get x = - J2/2, y=— 32/2. c=1-x= (2+ V2)2

.
Eset)=~912~~ ms Se Shoe ee j = 54/2
For A, =-— 1/2, we get “= 2/2. y = 32/2, z=l=ex=(C- 2 )/2

abe canes 2+ LONE ay Saf.

and F (ey Z) a faee

Example 2.40 Find the shortest distance between the line y = 10 — 2x and the ellipse
(x/4) + (y7/9) = 1.
Solution Let (x, y) be a point on the ellipse and (u, v) be a point on the line. Then, the shortest
distance between the line and the ellipse is the square root of the minimum value of

f (%, y, u,v) = (x—u)?

+(y-v)
subject to the constraints
x2 y?
(x, y)= +> -1=0 and @,(u,v)=2u+vu—10=0.

We define the auxiliary function as

2 2
F(x, y, u, Vv, Ay, Ap) = (x— uw)? + (y—v)? + a(t + > - 1+ Az(2u+v — 10).

For extremum, we have the necessary conditions

of 2(x - u) + +A, =0, or A, x = 4(u — x)

x = Ayy-v)+ 2d, =0, or A,y= 9(—

uy)

OF = -2(x-u) + 2A, = (js Or A, =X -—UHu

oe —2(y —v) + A, =0, or Az =2(y - v).

&zl eee g A, and A, from the above equations, we get

J
A(u —x)y =9(U -—y)x and x-u= 2(y — v).

theDive equations, we obtain 8y = 9x. Substituting in the equation of the ellipse, we get

i" » t e+ Ova),
ee 21,
64 or x = 64
Oe
5
re,x=+ 8/5 and y = + 9/5. Corresponding to x = 8/5, y = 9/5, we get
 $-u=2{2-v), OT 20 = =. 2 or u=2U—- Nh
5

Substituting in the equation of the line 2u + v — 10 = 0, we get u = 18/5 and v = 14/5.

Hence, an extremum is obtained when (x, y) = (8/5, 9/5) and (u, v) = (18/5, 14/5). The dis
between the two points is //5.
Corresponding to x = — 8/5, y = — 9/5, we get u — 2u = 2. Substituting In the equ:
2u + v — 10 = 0, we obtain uw = 22/5, v = 6/5. Hence, another extremum Is obtained y
(x, y) = (-8/5, — 9/5) and (u, v) = (22/5, 6/5). The distance between these two points is 34/5,
Hence, the shortest distance between the line and the ellipse is V5.

Exercise 2.4

Test the following functions for relative maximum and minimum.

1. xy + (9/x) + (3/y). 2. a2 —x? -y? a>0.
3. x? + Qbxy + y’. A, x + xy + y° + (1A) + (I/y).
Sx + 2/(x*y) + ae 6. cos 2x + cos y+ cos (2x + y),O<x,y<7.
7a Axe dy l2xy — 6y + z. 8. 18xz — 6xy — 9x* — 2y* — 5427. :
9, x* + y+ 27 + Axyz, 10. 2Inw+y+z)—O?+y?+27),x+y+7z>0.
Find the relative and absolute maximum and minimum values for the following functions in
the given clc
region R in problemes 11 to 20.

Al. x? — y? — 2y, R: x2 + y? $1, 12. xy, Rix + y? < 1.

13. x+y,R: 4x? + Dy? < 36, 14. 4x7 + y?-2¢ 41, R: ax ey? 1
15. x° + y’-x-y + 1, R: rectangular region; 0< x <2. 0 Sy 2.
16. 2x° + y° — 2x — 2y — 4, R: triangular region bounded by the lines
x = 0, y = 0 and 2x + y = 1.
17. x’ + y> — xy, R: triangular region bounded by the lines x= 1,
y= 0 and y = 2x.
18. 4x° + 2y? + 4xy — 10x — 2y — 3, R: rectangular regio
n; OS x <3,-4<y<2,
19. cos x + cos y + cos (x + y), R: rectangular region;
0< x < 7,0 < y= 7
20. cos x cos y cos (x + y), R: rectangular region;
O<x<7,0< YS.
21. Show that the necessary condition for the exist
ence of an extreme value of f (x, y) such that
is that x, y satisfy the equation Ix @(x, y) =
—fy % = 0.
22. Find the smallest and the largest value of xy
on the line segment x + 2y=2,x>0,
y>0.
23. Find the smallest and the largest value of x +
2y on the circle x7 + y? = 1,
24. Find the smallest and the largest valu
e of 2x — y on the curve x — sin y=0,0<
y<2z.
25. Find the extreme value of x? + y2 when x4 + yt =1.
26. Find the points on the curve x2 + xy + y* =
16, which are nearest and farthest from the origin
.
27. Find the rectangle of constant perimete
r whose diagonal is maximum.
28. Find the triangle whose perimeter is
constant and has largest area.
29. Find a point on the plane Ax + By + cz = D whic
h is nearest to origin.
30. Find the extreme value of xyz, when x + y+tz=a
a>0.
 Functions of Several Real Variables 2.55

3], Find the extreme value of a?x2 + b*y? + c3z* such that x! + y'! + 2! =1, wherea>0,b>0,c>09.
32. Find the extreme value of x? + y? + z? on the surface x? + y? + z? = 1, where 0 < p < q,
x¥> 0, y > 0, z> 0.
33. Find the extreme value of x° + 8y3 + 6473. when xyz = 1.
34, Find the dimensions of a rectangular parallelopiped of maximum volume with edges parallel to the
coordinate axes that can be inscribed in the ellipsoid (x7/a’) + (y*/b*) + (ic
y=,
35. Divide a number into three parts such that the product of the first, square of the second and cube of the
third is maximum.
36. Find the dimensions of a rectangular parallelopiped of fixed total edge length with maximum surface area.
37. Find the dimensions of a rectangular parallelopiped of greatest volume having constant surface area S.
38. A rectangular box without top is to have a given volume. How should the box be made so as to use the
least material.
39. Find the dimensions of a right circular cone of fixed lateral area with minimum volume.
40. A tent is to be made in the form of a right circular cylinder surmounted by a cone. Find the ratios of the
height H of the cylinder and the height h of the conical part to the radius r of the base, if the volume V
of the tent is maximum for a given surface area S of the tent.
41. Find the maximum value of xyz under the constraints x? + z? = 1 and y —x = 0.
=
42. Find the extreme value of x” + 2xy + z* under the constraints 2x + y = 0 and x + a ee ame Ie
43. Find the extreme value of x* + y? + z* + xy + xz + yz under the constraints x + y + z= ] and
x + 2y + 3z = 3.
44. Find the points on the ellipse obtained by the intersection of the plane x + z = 1 and the ellipsoid
x* + y* + 2z* = 1 which are nearest and farthest from the origin.
45. Find the smallest and the largest distance between the'points P and Q such that P lies on the plane
x+y +z= 2a and Q lies on the sphere x* + y* + z* = a’, where a is any constant.

2.6 Multiple Integrals

b
In the previous chapter, we studied methods for evaluating the definite integral f(x)dx, where
| : i ) : a
the integrand f(x) is piecewise continuous on the interval [a, b]. In this section, we shall discuss
methods for evaluating the double and triple integrals, that is integrals of the forms

{|f(x, y)dx dy and if] F(x,

y, z)dx dy dz.
J 5 “

Weassume that the integrand f is continuous at all points inside and on the boundary of the region
R or T. These integrals are called multiple integrals. The multiple integral over IR” is written as

{J } | F(X, %2, Bair ,X,)dx,dx5 a Mae

he

2.6.1 Double Integrals

saley, t f(x,
y) be a continuous function in a simply connected, closed and bounded region R in a two
jensional space IR? , bounded by a simple closed curve C (Fig. 2.5).
 2.56 Engineering Mathematics

Fig. 2.5. Region R for double integral.

Subdivide the region R by drawing lines x = Xp Y=Ye,kK=1,2,..., m, parallel to
the coordinate
axes. Number the rectangles which are inside R from 1 to n. In each such rectangle,
take an arbitrary
point, say (¢,, 7,) in the kth rectangle and form the sum

Jn = 2 f(x. Ne DAA,
where AA, = Ax, Ay, is the area of the kth rectangle and d, =
./(Ax,)? + (Ay,.)? is the length of
the diagonal of this rectangle. The maximum len gth of the diagon
al, that is max d, of the subdivisions
is also called the norm of the subdivision. For different
values of n, Say MN), M2,...,M,,..., We obtain
a sequence of sums Dap Sagedess Jim >++ ++ Let n — ©, such that the length
of the largest diagonal
d, — 0. If tim. J, exists, independent of the choice
of the subdivision and the point (€,, 7,), then
we
say that f (x, y) is integrable over R. This limit is
called the double integral of f (x, y) over R and is
denoted by

J= {| f(x, y)dx dy.

(2.78)
R
Evaluation of double integrals by two
successive integrations
A double integral can be evaluated by
two successive integrations. We evaluate
variable (treating the other variable it with respect to one
as constant) and reduce it to an inte
there are two possible ways to evaluate gral of one variable. Thus,
a double integral, which are the foll
owing:

i> If f(x, y)dy dx = {J Lf (x, y)dy] dx: first integrate with respect toyand then integrate
R R with respect to x.

oT «= {|f(x, y)dx dy = {| Lf (x, y)dx] dy:

first integrate with respect tox and then integral
R e
R with respect to y.
Let fbe a continuous function over R. We
consider the following cases.
 Functions of Several Real Variables 2.57

Case 1 Let the region R be expressed in the form

R= {(x, y): O(%) Sys wi), asx <b} (2.79)

— and w(x) are integrable functions, such that @(x) < w(x) for all x in [a, b]. We write
(Fig. £-9)

Fig. 2.6. Region of integration.

b y (x)
v= | | | f(x, y)dy |dx. (2.80)
x=a | y= (x)
While evaluating the inner integral, x is treated as constant.
Case 2 Let the region R be expressed in the form
R={@ )): eO) SxS kh). eS y¥ Sa} (2.81)
where g(y) and h(y) are integrable functions, such that g(y) < A(y) for all y in [c, d]. We write ( Big. 2./)
d A(y)
= | f(x, y)dx | dy.
(2.82)
y= L= RY)

a
L_=
a uating the inner integral, y is treated as constant.

R "sion of integration. |
2.58 Engineering Mathematics

n, the
: region
i . : canno
ck that it
R may be such nentsited
not be represented in peaaiied
in either of
either of ttheeligi zi Ve
forms give
E or met In such cases, the region R can be subdivided such i een + meee Car
Secours in either of the forms given in Eqs. (2.79) or (2.81). For example, 7 aK: —
drown in Fig, 2.8 and we write R = R, U R, where R;, R2 have no common in Ir points.
y= W(x)

Fig. 2.8. Region of integration.

Then, we have

{|F(x, y)dy dx = {|I(x, y)dy dx + {|f(x, y)dy dx

R Ry 59)

=] Uf tera] aes|b ||py(x) sesyvay |ae

p W(x) :

a Qi (x) 92 (x) : p
2.8:
In the general case, the region R may be
subdivided into a number of Parts so
that
m

{|F(x, y)dydx = 2 {|I(x, y) dy dx

(2.84
R alge :
Where each region R; is simply connected and
can be expressed in either of the forms given
Eqs. (2.79) or (2.81). If
Remark 11
(a) If the limits of integration are constant
s (for example, when R is a rectangle bou
lines x = a,x = b and y=C,y=d), then nded by the
the order of integration is not important.
can be evaluated using either The integ! al
Eq, (2.80) or Eq. (2.82).
(b) Sometimes, it is convenient
to e valuate the integral by chan
such cases, limits of integration ging the order of integration. !"
are suitably modified.
Properties of double integrals
1 If f(x, y) and g(x, y) are integrable funct
ions, then
 Functions of Several Real Variables 2.59

{| Lf (x,¥) + g(x, y)]dx dy = \|f(x, y)dx dy + \| g(x, y)dx dy.

oF .
R

ant.
2. {jnt YIEE ay = k| I(x, y)dx dy, where k is any real const
R R
3. Whenf (x, y) is integrable, then |f (x, y) |is also integrable, and

(2.85)
x,y) | dx dy.
\|f (x, y)dx dy = {||f(
R R

\ f(x, y)dx dy = f(¢, mA (2.86)

4.
R
where A is the area of the region R and (€, n) is any arbitrary point in R. This result is called the
mean value theorem of the double integrals.
If m < f (x, y) < M for all (x, y) in R, then

mAS < MA.

{|f(x, y)dxdy (2.87)
R
5. If 0 < f(x, y) < g(x, y) for all (x, y) in R, then

{|f(x, y)dx dy S \]g(x, y)dx dy. (2.88)

R R

6. If f (x, y) 2 O for all (x, y) in R, then

{|f(x, y)dx dy 2 0. (2.89)

R

Application of double integrals

e some of them.
Double integrals have large number of applications. We stat
1. If f (x, y) = 1, then {|dx dy gives the area A of the region R.
R
See For example, if R is the rectangle bounded by the lines x = a, x = b, y=c and y = d, then
i, d pb CT i | i
in a= | | dxdy = | i ts|iy= 0-a) f dy = (b—a)(d-c)

area of the rectan gle.

the s
give
ts =f (,)y) is a
2.12 surface, then
es 2) 3: J zdxdy or {Jf(x, y)dx dy
R R
 2.60 Engineering Mathematics

f (x, y) and above the x-y plane.

gives the volume of the region beneath the surface z =

For example, if z=

| | | ah ae ee:
gives the volume of the eae) x+y t+z2=a,722 0.
f (x, y) = p(x, y) be a density function (mass per unit area) of a distribution of mass in the
3. Let
x-y plane. Then

M = if(x, y)dx dy (2.90)

R
give the total mass of R.

4. Let f (x, y) = e(4, y) be a density function. Then

ane, = | 7
=a {]xf(x,y)dxdy, y= MW {|y f(x, y)dx dy (2.91)
R R
give the coordinates of the centre of gravity (x,y) of the mass M in R.
5. Let f(x, y) = p(x, y) be a density function. Then

|
fe {|y* f(x, y)dx dy and 1, = {|x* f(x, y)dx dy (2.92)
R R |
give the moments of inertia of the mass in R about the x-
axis and the y-axis respectively,
whereas Ip = I, + I, is called the moment of inertia of
the mas Ss in R about the origin. Similarly,

all (x-a)* f(x, y)dxdy and I, “| (y — b)* f(x, y)dx dy (2.93)

give the moment of inertia of the mas
s in R about the ks x =a and y
= b respectively.
ft IJ
f(x, y)dx dy gives the average value of f(x,y)
over R, where A is the area

of et region R,

Example 2.41. Evalua t the double integral |

uate Jxy dx dy, where R is the region bounde
d by the

x-axis, the line y = 2x and

the parabola y = x Nitsa),
Solution The poin
Olnttss of intersec
| tion
| of| the Curves y = 2x and y = x*/(4a)
The region are (0, 0) and (8a, 16a).

(ae Sy<2x,0<x<8
R= {(, y): (x*/4a) <,
O< x 8a}
a is given in Fig. 2.9.
 2.61
Functions of Several Real Variables

| (8a, 16a)

(a) (b)
Fig. 2.9. Region in Example 2.41.

We evaluate the double integral as

g, 2x 2x
a $a xy? z
t= [|xyaxay= [ | xy dy ax= | Ea dx
9
0 2
R x2 /(4a) ie?)

ee (ee )e=|5
Sa

at xee | = 4096
eet|>—agp
S|,|a*
4

oe= GE at
4 6 oa : =

=| 2 (45 aa 0
Alternative We can evaluate the integral as

l6a | p.f4ay ey ies fay

i= xy dxdy= xydx \dy= — yx dy
0 yf2 9 2 yl/2
R
l6a

i ae eas ac a 0 WY f - $2| = 2048 at

=3, 4.
aoe | > 4)" 2| 3 ~ te), 2 2 fe |
2 |
Example 2.42 Evaluate the double integral {|e* dx dy, where the region R is given by
R
R:2y<x<2and0<y<l.
Solution The integral cannot be evaluated by integrating first with respect to x. We try to evaluate
it by integrating it first with respect to y. The region of integration is given in Fig. 2.10. We have
2 x/2 p2 =

i=] | ay = f yer] ae
0 0) 0 0

1 2 2 ie 2 2
=3| xe" dx= ae | = (e~ 1).

2 py*/2

Example 2.43 Evaluate the integral i}I Z dx dy.

af x? +y7 +1
 2.62 Engineering Mathematics

Fig. 2.10. Region in Example 2.42.

Solution Because of the form of the integrand, it would Ee easier to integrate it first with respect
to y. The point of intersection of the line y = 2 and the curve y* =2x is (2, 2). The region of integration
is given in Fig. 2.11.

Fig. 2.11. Region in Example 2.43.

The given region of integration 0 < ys2and0<x
<
v2x <y<2. Hen ce, b obtain y’/2. can also be written as 0 < x <2 and
we

[= | ae | = [eT]
+] | dr = Piers St va
v2x

J fe 2
= eT FS Since
J Bes) ~ dx v? |
0
=34+2(In5— In V5) ~ 5(9 I)
- =2In5-1.
The cylinderx? + z? = ] is cut
of the region jcm the first BPtAnE. |
u by the planes y= 0, z=O0 and Km y, Find the volume
 Functions of Several Real Variables 2.63

Solution In the first octant we have z = ¥1 — x” . The projection of the surface in the x-y plane is
bounded by x = 0O,x = 1, y= 0 and y =x. Therefore,

ir ake l -
v={[ zdxdy = | J V1 — x7 ay arf V1 =x? \[ylg €%
‘ 0 0 0

=
|

5 @ = re Se! cubic units.

0 xV1 x? dx=
' | 3

2 2 2
Example 2.45 Find the volume of the ellipsoid a a re =
a b> ac
Solution We have volume = 8 (volume in the first octant). The projection of the surface

we y lee 8 : | a ee :
ae .h — 7 re in the x-y plane is the region in the first quadrant of the ellipse —~ + re == tle

Therefore,

a | pby1-x2/a? 2 wiye “a bk 2
v=s| | Cc 1-4-4 dylax=8e | b2-
2 dy |dx
0 0 a b #0 0 b? é

where k? = 1 — (x7/a’). Setting y = b k sin 0, we obtain

a n/2 wa | a rl
V= sc | i 4k? — k*sin*@ (bk cos 0) d@ |dx = 8bc | 1 k* cos*6@ 40 ts
0 [0 | 0 | 0

a a) a
= Abc (2]| c ~ © dx= 2896 | (a? — x?)de
OMe Ne il a 0

b x? ‘ 47 abc
_2Mbe} ,2,_ 4 | =—— == cubic. units.
.

a 3 \o 3
Example 2.46 Find the centre of gravity of a plate whose density p(x, y) is constant and is bounded

Solution The mass of the plate is given by (see Eq. 2.90)

M= {J p(x, y)dxdy=k {| dx dy (0 (x, y) = k constant).

R R
1 e boundary of the plate is given in Fig. 2.12. The line y = x + 2 intersec
ts the parabola y = x at a

the points (—1, 1) and (2, 4). The limits of integration can be written as — 1 < x <2, x?

vf ee
< YS xv 2.

+2 : 5

| I
2.64 Engineering Mathematics

Fig. 2.12. Region in Example 2.46.

The centre of gravity (x,y) is given by (see Eq. 2.91)

B=oy[fsec naea3[|
7 | . | 0) 2 x+2
Payne
R = Bee

2 , 2
34 xt 2—aA)de= 5) ex? 2] =a,
—I
Gas 4 —|

y p(x, y)dx dA>-2f | a ln e2 12

9 x+2 y) 2 x+2

al 9 [PFve lara? f°] ;

9 J.
as
Lf [(
Ae =r
9)? 4
\O i= 36") AX =
l
il|)
(x + 2)?
Se : J
5 =

=]
=5 [3 4-1)-402+0)]- 5 21 =|=%.

Therefore, the centre of gravity is loca

ted at (1/2, 8/5).
Moment of inertia about the x-axis is
given by (see Eq. 2 92)

t= JJ spt sndedy=e [ ‘a 24 a Ay ‘cs

‘ Jeuilideow © yee=k | cae dx
2
al (+2 aya =k] G49) al

—l

(3 _ 423
2) a = orm
 | sie ee 2.6:
Functions of Several Real Variables

Moment of inertia about the y-axis is given by (see Eq. 2.92)

2 x+2 2 .
ly = \|x“ p(x, y)dx dy = k | J dy [as=i | x*(x+2-—x°)dx
R -1 |?x?
os

“I

me 2x xe 15 33] 63

2.6.2 Triple Integrals

Letf(%, y, z) be a continuous function defined over a closed and bounded region T in IR? . Divide
the region T into a number of parallelopipeds by drawing planes parallel to the coordinate
planes. Number the parallelopipeds inside T from 1 to n and form the sum

Jn = % f (xe Yer ze) AVE

where (x;, Yg, 2,) iS an arbitrary point in the kth parallelopiped and AV; is its volume. For different
values of n, Say n;,N2,...,N,,,..., We Obtain a sequence ofsums Dive dances Spe qe as De length

of the diagonal of the Ath parallelopiped is d, = (Ax,)* + (Ay,)* + (Az,)* . Letn > such that
maxd, > 0.If |im J, exists, independent of the choice of the subdivision and the point (xy, y,, Z,),
— oo

then we say that f (x, y, z) is integrable over T. This limit is called the triple integral of f(x, y, z) over
T and is denoted by

J= \) F(x,
y, 2)dx dy dz.
(2.94)

Triple integrals satisfy properties similar to double integrals.

Application of triple integrals

1. If f (x, y, z) = 1, then the triple integral
Ver ff dx dy dz (2.95)
T
gives the volume of the region 1i
2. If f(x, y, z) = p(x, y, z) is the density of a mass, then the triple integral

M = {fj f(x,y, z)dx

dy dz (2.96)
T

X= 4 {VJ xf(xy,z)dxdydz, y= 4 {fi yf (x, y,z)dxdy dz,

t T

tn, {iJ ef(%y.2)dx

dydz
Me de, (2.97)
.66 Engineering Mathematics 7
e e
nt re ofafe
gr rav
a\ ity )
ity ) of the solid of mass M in
of the ce mass (or thnee ce
centre ofae
‘ve e
siv the
the Cor
co or di ate
na tess
th e density function.
heeft y ZS P(X Ms z) is
2. f(x
72) ) f (x y. zjdx dy. dz,_
y, Z)4.-

= |
y2? +22)iB f (x, y, Zax dy 42,
as\irAavaZ |
f il} (x 5ae

4. Pe {|| Re

T
(2.98

i, = {jj (x2 +yf(2 )z)dx dy dz

x, y,
< axis respectively where
T ot 5 d z-
o f 1S
mas s in T abo u t the x-a xis, y-aXx1S and
;
rti
iner
give the moments ofine tiaa of the
density function.
f(x, y,z) =p, y, z) is the

Evaluation of triple integrals

ribed by
eval uate the tripl e inte gral by thre e succ essi ve integrations. If the region T can be desc
We
x, SxSx, yi) Sy S yor), 1A y)SzS Z2(x, y)
then we evaluate the triple integral as
+ | wes
Xp pyg(x) ez2l.y) z(x,y) ib
| | : f(x, y, z)dz |dy |ax (2.99)
flay 2)dedyax= | |
| | y,(x) z(%y) ul
XY y (x) z(x,y) x]

We note that there are six possible ways in which a triple integral can be evaluated (order of variables
of integration). We choose the one which is simple to use.

Example 2.47 Evaluate the triple integral WW y dx dy dz, where T is the region bounded by the
T
surfaces x = y’, x =y +2,4z=x*+y’ andz=y+3.
Solution The variable z varies from (x? + y*)/4 to y + 3. The projection of T on the x-y plane is the
region bounded Py the curves x = y* and x = y + 2. These curves intersect at the points
(1, -1)
and (4, 2). Also, y’Sy + 2 for -1 < y < 2. Hence, the required region can be written
as |
-l<y<2, y?<x<y+2 and [x* + y*\/4] <2 <y 43.
Therefore, we can evaluate the triple integral as
2 y+2 [ @(y+3) | 3

J
=] ‘| | dz\d = ‘it te ay? |
y? (x2 +y2y/4 : ha Si liws2 yt aes aoaax ay
f ( y3 ae y+2
= Ee ay We
eI ie : | 12 | : dy
: y
a iG
ee
ys: :
ies sa. le -¥ ay
y F\OF2~ 5%) 1 yy 429
\(

= re ay?Nae 16
ae 3 2y $A? a
 Functions of Several Real Variables 2.67

| 42
she und’4°... yt. ay? Sy
ee IS el ee ae ie
valuate the integral illz dx dy dz, where T is the region bounded by the cone
Example 2.48 Ff

=x’ tan? a + y* tan*B and the planes z = 0 to z =A in the first octant.

Solution The required region can be written as

Osz< 4x? tan?’a+ y* tan’B, 0<<Whk x tan’a)cotB, O<xShcota

: f

Therefore,

h cot a@ (af h2 — x? tan2a@ )cot B

7 (x* tan?a+ y” tan” B)dy | dx
0 0

h cot o 1
= -| ne —x* tan7a)'* tan*a + 5(h? — tan? or)? |cot B dx.

Substituting x tan @ = h sin O, we obtain

m2
y= oot
t
| A? sin?@(h cos @) + =(h cos*6)| h cot acos @d@
0
| n/2 , |
= 5 h*cot Bcot a | (sin*@ cos*@ + ‘vos 0). |
0
m2,
s 4 h*cot B cot a 1 (sin?@ — sin*@a
es.
+
0

;
= 5 h*cot Bcot a a -
+ 2 +z|=44
cot
aie EG acot B.
O
Example 2.49 Find the volume of the solid in the first
octant bounded by the paraboloid
Z = 36 — 4x*— 9y’.
Solution We have
= ! dz dy dx

The projection of the parsboloid (in the

first heh in the x-y plane is the region
in the first quadrant
of the ellipse 4x” + Oy” = 36.
Therefore, the region T is given by
0<z<36-4x°-9y*, O<y< es)CN | =a
re)
© lA + lA tu
|
2.68 Engineering Mathematics

c - x*/3)
V= (36 — 4x? — 9y* )dy |dx
“0 0

0 CE
j j ee a ile aa

a0 32 8 2 3/2 16 ng? 7 ar.

J0
Substituting x = 3 sin 0, we obtain
mi2 /2 ’
y= (27 cos*@)(3 cos 0)d@ = 144 | cos"@d@
0 0

| 4 | 2
== 144 (S
caer SS5 eam4 | 270 cubic units.

Example 2.50 Find the volume of the solid enclosed between the surfaces x2 + y’ = a’ an
v4 2=a?.
Solution We have the region as

—Va* -x? <z< Va =x, = Va’

— x2 <y< va? — x? a —~@SXSa-
Therefore, ff

~S/
4 ABER? nhagt fe d ae
v= | | | de dy ds= 8 | | Nate
— ~ya2 — x2 She 0) 0 ‘ - a
dvd
a
i
= | (a* —x?)dx =8 (a a
2,= =| = 1027
3
0 cubic units.
0 3

2.6.3 Change of Variables in Integrals

b
In the case of definite integrals J (x)dx of one variable, we have seen that
the evaluation of the
integral is often simplified b y using some él

substitution and thus chan

Similarly, the doyble and tri ple integrals ging the variable of integration.
can be evaluated by using s
the variables of integration. ome substitutions and changing

Double integrals
Let the variables x, y define
d in a region R of the x-y pla
ne be transformed as
X=x(u,U), y = y(u, v). (2,100)
We assume that the functions x (u, v), y(u, UV) are defined
and have continuous partial derivatives !"
 '"~S OO
DE E

Functions of Several Real Variables 2.69

. oe ’ .
the region R Coen est in the u-v plane. We also assume that the inverse functions .
u = u(x, y),
y = v(x, y) are defined and are continuous in the region of interest in the x-y plane, so that the mapping
is ene ONE: eee the functio
fn(x,y) is continuous in R, the functi
fon
[x(u, v), y(u, v)] is also
continuous in R*. Then, the double integral transforms as

\]I(x, y)dx dy = \|f[x(u, v), y(u, v)] | J|du du = {|F(u, v)du dv (2.101)
f F
otiaee j= O(x, y) _ Ox/du dx/dv
A(u,v) | dy/du dy/du
is the Jacobian of the variables of transformation.
For example, if we change the cartesian coordinates to polar coordinates, we have

x=rcos @, y=rsin@, n<r<nm, 4<O0<O

-dx/dr dx/00 : cos 6° —rsiné@ |

dy/or dy/de| sin | poos@l (2.102)
Therefore,

{|f (x, y)dx dy = {|f(rcos 8,rsin @) rdrd@= {|F(r, 0) rdrd@

R R* R*
Where R* is the region corresponding to R is the r-@ plane.

Triple integrals
Analogous to double integrals, we define x, y, z as functions of three new variables
x=x(u, VU, WwW),
y= yu, V, Ww), Z= z(U, V, W). (2.103)
Then 7

{| f(x,y,z)dx dy dz = {| flx(u, v, w), yu, Vv, Ww), Z(u, Vv, w)) |J |dudu dw (2.104)
7 ) T™

Ox/du dx/dv-~—s ax/ dw

where J= sane =|dy/du dy/dv_ dy/dw
ee! Aziou “deldve \azidw
is the Jacobian of the variables of transformation.
For example, if we change the cartesian coordinates to cylindrical coordinates, we have
x=rcos 6,y=rsin'@
2—2

dx/dr ax/d@ adx/dz cos@ -rsing 0

y= | ay/or dy/0@ ody/dz| =| sin @ reos@ O|=r (2.105)
dz/dr oaz/dd dz/dz 0 0 lui

{fJ F(X Y> z)dxdy dz = {J f(rcos 0,rsin @,z) rdrd@dz

ne 7
2.70 Engineering Mathematics
: s ; : — 7 1 3 '

we have (Fig. 2.13)

If we change the cartesian coordinates to spherical coordinates,
OF OS7
x=rsin Ocos @ y=rsin sin 6, z=r cos 9, O0<@<2m7,
" < <4
a .

ax/d@ sin@cos@ rcos@cos@ —-r sin @ sin 6

idx/dr dx/d@
dy/de@ dy/dO|=|singsin@ rcose? sin 0 rsin @cos 6
J=\ay/ar
dz/dr dz/dd dz/d0 cos @ —rsing 0
sin“ sin*@)
= cos o[r sin @cos ¢cos7@+ 7° sin dcos ¢ sin-9]+rsin o[r sin-@ cos-@ + r
(2.1
= r*[sin @ cos7@ + sin?¢] =r’ sin ?

and \| f(x,y, z)dx dy dz = {| F(r, 8, 9) r*sin @dr dO dQ.

T Ba

Fig. 2.13. Spherical coordinates.

E xamplele 2.51
2.: | te the
Evalua the integra
1 l {| (a* : — x* — y*) dx dy,
| where R is the region x + y": $4
R
Solution We can evaluate the integral directly by writing
it as
4 ince
[= 2 2
if = 5 (Coa — y*)dy dx.

Howev
LOWEVET,er, It
it is is easierr to
toevevaluate,> 9 if
j we change to r
er ks AE We Cli Olar coord : inate
ate’
to polar coordinates, we have (see Eq. 2.102) : te eeosian coor’
SF COMO. y= r'sini'O, J =.
Therefore,
a pln

— yl (a r )rdr d= J I, 40)(ar Pyar

 Functions of Several Real Variables 2.7)

ad

a | pH By ANA 4
zy an | (a*r - r\deo ope) 2
0 | fu A = :

Anos 2.52 Evaluate the integral \| (x — y)* cos*(x + y)dx dy, where R is the rhombus with
R
successive vertices at (7, 0), (27, 7), (x. 27) and (0, 72).
Solution The region R is given in Fig. 2.14. The equations of the sides AB, BC, CD and DA are
respectively

X-YV=N, X+y=3nm, x-y=-—M and x+y=T.

C (7, 27)

A (7, 0)

Fig. 2.14. Region in Example 2.52.

Substitute y—x = uandy+x=vU. Then, —-wS usa and mS<u< 37. We obtain

x=(v—u)/2, y=(u + w)/2

ax/du | % =i if? yry1 ieee

ICE Win |dx/du
and die dy/du dy/dv 1/2) 2 2

Therefore,
3m pT

j= {](x — y)* cos*(x + y)dx dy = * | | u*cos*v du dv

3m 3 fez 4
== | cos? du = = | (1 + cos 2v) dv = =.
5 t 7

Example 2.53 Evaluate the integral \)Ax? + y* dx dy by changing to polar coordinates, where

= 9.
an by the circles x? + y? = 4 and x7 + y*
Ris the region in the x-y piane weavi
dy=rdrd@, and
Solution Using x=rcos 9 y=" sin 0, we get dx
| 2 3 3 2%
| a= 12| do = 232
afiz “4 J," rerdrdoy= |
Seep eo
 2.72 Engineering Mathematics

, ike integral! ay zdxdy

| 7 where 7 is the hemisphere of -
Example
: 2.54 Evaluate
-valu the g | dz, W

rT
r+ +2=a’,z20.
Solution Changing to spherical coordinates
=
x=rsin7 dcos 6, y=rsinns @sine 8, z=rcos¢, » 8< 27% 0<@<
0<0<6<22 x72 — .

we obtain dx dy dz =r sin @ dr d@ d@ (see Eq. 2.106). Therefore,

21 {2 a _4 pIx
—
gzf2
:
i= | | | (r cos @) r*sin @dr do dO = = | | sin 9@cos 0dod
0 40 #0 9 40
4 20 mif2 4 27 COS 20 qFl2 s a* ¥ Tae
ee
= I | - 290 d0d6:jee 3eyI | ——
sin 7 I d§ =>g J. a0 = ——
Z
=

Example 2.55 A solid fills the region between two concentric spheres of radii a and 6.0 <2<
The density at each point is inversely proportional to its square of distance from the origin. Fir
total mass.
Solution The density is p = k/(x* + y* + 2”), where k is the constant of proportionality. Therefor
the mass of the solid is given by

M = [J p dx dy dz = If ee a gy
J a ty 40287
T T
where a* < x * + y’ + z* < b*. Changing to spherical coordinates,
we obtain
x=rsin @cos 8, y=rsin@sin 6, z=rcos ¢, 24 ¥+2=r, a<r<b
dx dy dz=r° sin o dr dO dd, 0< O< 2m, O<o<7
Therefore,

Fel |.
2m 1 b r-sin o
2 px
2 dr do dO = k(b — a) j I
() /(0) a sin @d@d@
| 20
2%
= k(b -a) ) [—cos 9]5 dO
= 2k(b — a) | d0=4zk(b
| 2).
0
2.6.4 Dirichlet Integrals
Let T be a closed region in the
first octant
and the coordinate planes. Then, in IR’, led
by theessurface
.
an integral of thebounded (x/a)? + (y/b)?, + (z/c)
tefcy= }
form

<= {[f xO PI 21 dxdydz

T (2.107
is called a Dirichlet integral,
where all the constants a,
Positive. 8
a, b, c and p, q, r are assumed to &
 Functions of Several Real Variables Gul
* - “)
2
p " i
|

We now show that

l= {If x Fly Bp Loy | dx dy dz = at*bP et P(alp)T(Biq) i (y!r)

(2.108)
J. Ts ) |
’ Pp q r
x \? y q 2 r
Let =] = u, (=| =U, (=) =W,orx= aul? y= bu'4, z= cw’.

The Jacobian of the transformation is given by

Ox/du dx/dv = ax/dw (al pul?) () ()

j= dy/du Ay/ du JOy/dw = 0 (b/g)? 0

Oz/du dz/du dz/dw () () (clrywr-

= ADE i p)~\yVa)=1 r=
pqr

and dxdydz = |J| dudvdw = 226 y/r)-1y/a)-l

WN du dudw.
pqr
Now,x20,y20,z20 gives u2>0,v 20, w 20 respectively.
Hence, we obtain

—| 5 i— :
Te Ml [au?) |" [bua |?few) om yp )-l yy(/q)-1 (Ir) -1 dududw

= Aor cr {y w@/P-yBa-lyY/)-| du dydw

par
R
where R is the region in the uuw-space bounded by the plane u + v + w = 1 and the wu, vw and uw
coordinate planes, (Fig. 2.15), that is, R is defined by
O<swsl-u-v, Osvsl—-u, O<uK<il.

Fig. 2.15. Dirichlet integral.

2.74 Engineering Mathematics

Therefore, we get

[= ee | [.
—— (a/p)-1 y(Bla)=1 WO =l dududw
u=0 “v=

l-u-v
hee cn pcBlay-1} wir)
wo dudv
~u=0 7) jy

atbPor
wit f l°
ybPr-l y(Blay-l () — yw — vv)’ dudv
Pqy w=0

Substituting v = (1 — u) t, dv = (1 — u)dt, we obtain

a b8 y fi M “i re
fa Se | | yi e/p)-1 (= u)'Bla)+(y/r)) t(B/q)-1 Ch = ry") dudt
Pqy =0 #/=0 |

Sin
theclim
eits are constants, we can write

= £0 |CGP) (yy BIa) +r ay | ¢(Bla)-1(yp) (!7) ay

ol

bry
 Functions of Several Real Variables 2.75

= a“bPoY I(alp)I'(B/lq)l(y/r)
rare $$
r{2 + i + le+ 7
P q iF
which is the required result.

Example 2.56 Evaluate the Dirichlet integral

ik {ff x? y3z3dxdydz
mi

the coordinate
aa I is the region in the first octant bounded by the sphere x? + y* + z” = 1 and
planes.

Solution Comparing the given integral with Eq. (2.107), we get

Q=P=y=4,p=q=r=2,a=b=ce=l.
Substituting in Eq. (2.108), we obtain

I
Le ae = ——
(7) 8(6!) 5760
ince I’(n + 1) =n!, when n is an integer.

xample 2.57 Evaluate the Dirichlet integral

= i} 2 yi? 2M? edydz

Where T is the region in the first octant bounded by the plane x + y + z= 1 and the coordinate planes.

Solution Comparing the given integral with Eq. (2.107), we get

C= B= V= 3/2. p—G=rT=la=—=b=¢= |.

Substituting in Eq. (2.108), we obtain

ae Le Gir
Py

Using the results, F(a + 1) = @ I (@) and (1/2) = V7, we obtain

= C
BPT
H~D G0I/2))°D
(0/2)
ADG
F UD
ID
47
~945
Exercises 2.5
we ny, Find the area bounded by the curves y = x y=4—32

>0, y>0.
. 2 ae.
>
Find the area bounded by the curves y= 4 — 2x,x
2.76 Engineering Mathematics

4. Find the area bounded by

the curves Puay,x=y. el wipe De = Ay ~.
es x + y 2y,xX +y y, x
find the area boun ded by the curv
cha ngi ng to pol ar coo rdi nat es,
5. By
and x = 0.
ible
: lowing double
uate the fol integgrals.
Change the order of integration and eval
, l y+4 9) 1
/ e y dx dy | 7. {| sy +
7 dx dy.
DONE yoo c
ly? + y? y= xs=\

;
1 pn
|

| | e* dx dy.
ofA Whe [eo
2
n
x7/2 |

oe fx? +y? +1
y=0 *x=y

: l-x rae | d = uv)

10. | | e”(**¥)
dy dx (use the substitution x + y= 4 and y =
x=0 # y=0

11. Find the volume of the solid which is below the plane z = 2x + 3 and above the x-y plane and bound
by y’ =x, x =0 and x =2. d
12. Find the volume of the solid which is below the plane z = x + 3y and above the ellipse 25x* + 16y* = 4(
x20,y20.,
Vee Find the volume of the solid which is bounded by the cylinder x ? + y’ = 1 and the planes y + z=
and z= 0.
14. Find the volume of the solid which is bounded by the paraboloid
z = 9 — x* — 4y” and the coordina
planes
x >0, y>0,z>0.
15. Find the volume of the solid which is enclosed betwe
en the cylinders x* + y* = 2ay and 7 = 2ay.
16. Find the volume of the solid which is bounded
by the surfaces 2z = x’ + y* and z= x.
17. Find the volume of the solid which is bound
Oe aa ed byy the Surfaces z= z= 0.0, 373z == x* + y?2 andand the
the cylin
cvlind
18. eee
Find the:es
volumume of the solidi which
ich isis in
j the firstSant
octant bounded by the cylind
. ers x7 + y?——
= a’ an
19, Find the volume of the solid which
is bounded by t | id
and the cylinderx? + y? = 2 x. :
¥ the paraboloid 4z = x? + y’, the cone 27 =x - +)
20 Find the volume of the solid which
and x+y = 1. 1sj Common to the
ha right
z

tioht circul
x: ar cylinders x2 + 2=l,y+7=:
21 Find the volume of
the solid whi ch is
¥+ y+(-aPagq 2
above the cone
z* = ,2 + y* and inside
the sphere
22 Fi
o: ndothe vo
alu
eme S
ofhthe
|ae
solidmi
whivhe
chich is
; belo Ww the surf :
ace z = 4x? + 9)?
and above the Squa
23. Find the volume re with verlic| e>
of the solid which
is bo unded by
th e Paraboloids
z = x2 4 y and z=4- 3(x? + y’):
aesFind
;
the volume of the soli
is bounded b
ne aa vy +z = Va and the coordinate planes.
at. ntained between the
26. Fi
esnd the volum € of the regio
cone 2 = 207 + y?) and the hyperbolo!
|
OS O< m/4. sion under the cone z = 3r and over the
27, Find the volume r Ose petal with boundary r = sin 44
of th
Vertex
ate
ertex at
Oo
at the
the aric;
€ Portion of the unit
:
origin and making an angle q Sphe in]
- “ee rae eg ee
inside the right Circular cone having '*
| e Z-axis.
 Functions of Several Real Variables 2.77

28. Find the volume of the region under the plane z = 1 + 3x + 2y, z= 0 and above the region bounded
y = 2,2
by x= 1,x=2, y=x’, and
29. Find the volume of the portion of the sphere x2 + y2 + z” < 2ay between the planes y = 0 and y = a.
30. Find the moment of inertia about the axes, of the circular lamina x? + y? <a”, when the density function
is P=Jx* +y?.
31. Find the total mass and the centre of gravity of the region bounded by pos ye =a”,
x20, y 20, when the density is constant k.

: Show that [=2 {|ayy

32. dx dy h integer, R: x* + y* > 1 converges for p > 1.
R
Hence, evaluate the integral.
Evaluate the following integrals (change the variables if necessary) in the given region.

33. {|(x* + y”)dx dy, boundary of R: triangle with vertices (0, 0), (1, 0), (1, 1).
R

34. {|x* dx dy, boundary of R: y=x*, y=x+2.

R

35. {| (x? + y*)dx dy,R:0<y< Vl—-x? ,0<x<1.

R

|
36.
i
ary pa a) ono
x? y?

37. [[ eer? aed, ei Pe yP24 P9525, yon x20,y20,

R

38. [[ Pracd, Rite yest x20 y20

$

39, [[ raed. e: V+ Wo= Vax 0920

ag JJee 2 1422)» Downey of X; (he. square with vertices (£1, 0), (0, 1)
i R

(ct ange COOr dinates: x-y=u,x+y=v).

7 at. ed pa os sy , . :
ast. |(x + y)? dxdy, boundary of R : parallelogra . .
m with sides x + y=l,xt+y=4,x-2y=-2,
y
2.78 Engineering Mathematics

43, \|xy dx dy, region (in polar coordinates) R : r= sin 20,0 < 6s 7/2.
R

44. \))
x*y*zdxdydz,T: x+y" <1,0<z1.

dx dy dz ;
45. , boundary of 7: x= 0, y= 0, z= 0, x+ yt z= 1.
bie rere he ia fon al ; ; ani:

46.
> {I (x+3y—2z)dxdydz,T: 0<y<x7,0<zsxt+y,0<xS1.

47.
— [J|s4:dvde, boundary of Ti y=x, y=x4 2, 4z=x°+y’,z=x+3.

48. {| (2x-—y-—z)dxdydz,T:0<x<1, O<9=37 0<7s x+y.

T

49. ee
Ge
a e: b ound) ary DD! 1 2°
Sens
of T: x° + y°+ =a", x*+y?4+72=a>b.
b2

50.
= Biosscsi gay + y, x+y" +27 = I.

51. IJ] 1-4Zi -5-5

73 :te 2 =].
dx dy dz, boundary of T : X—

32. fff fx? + y? + z* dx dy dz, T: x? + y? +

Zty)
T

33. If * +y)ded
Ly, (2 }dx dy dz, boundaryenh ofte T : as
Die) is 7
he Piet SM > a a I and the coordinate planes.
f

54, {ff (y* +z? )dx dy dz, boundary of T: y?+ 2<@2Q<;< h

r a"

aluate th > following Dirichlet integrals.

56. ||| xyzdxdydz, T: Region |

ee ed by x + y+ z= 2 and the coordinate planes.
 Functions of Several Real Variables 2.79

57. {IIxy?z? dxdydz,T: Region bounded by x + y + z = | and the coordinate planes.

Ot

58. ll xyz dx dydz,T: Region bounded by xr + y +z? = 8 and the coordinate planes.
oF

59. illxy'*z dxdy dz, T: Region bounded byxt+y+7e=l1.

T
f 2
60. \\]x*y dx dy dz, T: Region bounded by = + = + a = 1.
L

2.7 Answers and Hints

Exercise
2.1
1. | f(x, y)-—11l=l@-1¥%
+ G- 1) + 2@-1) + 20-D!
2\e—1. 2+ pit Foleyilee
2)/2 - 1]
dis used, we get 26° + 46< €or b< [/(e+
(i) if|x—11<6,ly—11<
(ii) if 6* < dis used, we get 6< €/6

(iii) if (x — 1)? + (y— 1)? < & and| x-11<6,| y—11< dis used, we get 6< Ver e=7%,
@-
+ 6@—
2. \f @y)—71=1@-2)- 20-P1)
2)-1)
Bix 2 Pe lye TPs Ole <2 12 y=1 bee
dix —21< 6, ly-1 | < Sis used, we get 26°
+ 86<¢, or d< (e+ 8)/2 -2.
mt| (i) if é°< dis used, we get o< e/ 10.

Gi) if @- 2)? + (y— 1)? < 6° and| x-21<6,ly—dis

11< used, we get d< Je+16—4.

a
=!
| fe—
= 2,
<lxtyl<laltlyl< 2yx?- +y2 <e Take d<e/2.
m ies ery ia 1|
efore
4, p= rieeas @, } die r sin 0. Ther
LetJs
=;

ee
+e Take 6 < e€/2.
| °~—
"ee =

Se aro y)-0ai ie+lyl< 2, < *a5 <€ Take 6< é/2.

= 2 et) 39
ie as
a a).
ST]

12. ]i,
 - Letx=rcos @ y=rsin@; —| Sarees
cos’@
=a
=)
ae
_» cas r > 0. Limit does not exist.
cos?@+ sin’ @
16. Choose the path y = mx*. Limit does not exist.

17. Choose the path z = x*, y = mx. Limit does not exist.
18. Choose the path y = mx, z = mx. Limit does not exist.

19. Choose the path z = Vx , y= mx. Limit does not exist.

20. Choose the path z = 0, y = mx. Limit does not exist.

21. Choose the path y = mx. Discontinucus.
22. Limit is 0 for x > 0 and 1 for x <0, Discontinuous.
23. Discontinuous, 24. Choose the path y = mx. Discontinuous.
25. Choose the path y = mx. Discontinuous. 26. Cancel (x — y). Discontinuous.
27. Letx=rcos 6, y=rsin @. Continuous. 28. Choose the path y* = mx. Discontinuous.
: 7 7
29. Since x“ +y*>2|x| Lyi we have ———_ 1
_- < >
] |sin 4/ Ixy] = J ly! |
therefote; |f(x, yyIl< : : J
it toys 21 xy V2 | xy!
Continuous.
30. Since 2 < 3 + sin x < 4, we have [1/(3
+ sin x)] < 1/2. Therefore, | f (x, y) | < [(2x2
Continuous. + y’)/2] < x? + y-.
|
31. The function is not defined along
the path y = — x. Discontinuous.
xo =p Fes) esd (x? + y2)5/ +(x? + 2 5/2
32. 2 =p | = = 9
x +y* x* + y*> S
x? +, y? : : Continuous 2.

33. Function is unbounded in

any neighborhood of x = —
|. Discontinuous.
34, Since | xl, 1 y |, 1 zl are all
< VP etVet seal fl < je?
207 Continuous.
35. The function is unbounde
d along x = V¥3z- Discontinu
ous.
Exercise 2,2

I. f,(0, 0) =0, £0, 0) = 0.

For (*, y) # (0, 0), find f,,
as (x, y) > (0, 0). f. anc’ choose the path y= mx. The limits do not exis!
2. f(x, y) is unbounded as (zy) =>
(O, 0), for example al
3. f.(0, 0) = 0, f,(0, 0) = — 1, f.(0, ong x = yy:
(0, Q) —
y) = 0,f,, 0) = 1.
Eo, 0) == i

4. JO, 0) = 1, f,(0, ()) =. kk dz =

|
Ax + Ay, jim, (A
z — 4z)/Ap) does
not exist.
5. fr, 0) =0=£,(0, 0), de =0
- Jim) [(Az ~ dz)/Ap) =
0.
No contradiction since con
tinuity of f fy is
only a sufficient co
In problems 6 to 15, f. » fy and f, ndition.
are given in that Or
der at the given
6. —_ ea 2. point
A. 1 2e 173,
8. 6e'?, de!” |

10. = 1/10, ~ 1/10. 9. 49/(85)3?, _ 49/(¢5)32

(859.
 Functions of Several Real Variables 2.81

11. f(y) = 21n [J x? +y* —x]-—2 In y, —2/5, 3/10.

12. — 2/27, — 1/27, — 2/27. 13. e¢, —2e, e.

14. 5, 3, 0. 15, 1/7, 3/35, 4/35.
16. 0. | aie
18. e*[sin (y+ 2z) + (41° - 1)?} cos
(y +22]. 19. Ay+z)t+ (tz) (t+ Det+ (ety) (1-de".
20. (7/2)
— (2/7). 23. Sets=x-y,v=y—Z,w=Z-x.
31. — [yx?™' + y*In y)/[xy*! + x In x], 32. y/[x + 3y’Q? + y’)].

33. | =~ se) _ _ y(sin xy) + z(sin xz) (=) ted soe. | x(sin xy) + z(sin yz)
Ox y dfldz y(sin yz) + x(sin xz)’ | Oy ‘ 0 fldz yt y(sin yz) + x(Sin xz)”

(= __(9flox\ _— 3x*+3y+3z (dz) _ (dfldy\) _ 3x-4y

34. | Ox : dfldz} — 3x+2z " \dy). \dfldz) 3x4 2z
35. Let u=z/y,
v = x/y; then f (u, v) = 0; x. 36. 499.6.

37. 4.02. 38pee aeSS) l+ fie |

ey ev3i+1)).

39. 1.81. 40. =a~ [180 + 2(6 - ¥3)] = 0- 2686.

41. 5.01. 42. V=nr°h/3, dVidt = 852/72 = 3.71 fP/hr.
43. § = 2(xy + xz + yz), max. absolute error = 2880 in*, max. relative error = 0.0766 in.
percentage error = 7.66%.

44. A= eS sin @ , percentage error = 13.7%.

45. V = abc, percentage error = 3%. 46. V=ar°h, percentage error = 9.2%.

47. 121.6 watts. 48. 2.92%. 49, 29.33%.

0, di = J(dr)? + (dh)?= 1/20,
50. Lateral length / = Vr? + h?, lateral area =rl, dr =r/100, dh =h/10
percentage error = 2%.

Exercise 2.3
— . At (1, 1): fo = — V/2s fy = %Sy = 12.
BAL 2) 3): fee = Oh Fey = 9 Soy = — V9» Sy = 27.
3. At (1, 2):fer= 9, fey = 1s Sy = — 34.
4. At (1, 12/2): feo, = € In (0/2), fery = elm) +1, fyy =— 4e/m*, fry = 16e/7’.
5. At (2/2, 1): fee =— & Sy = Ol2sSy = - mel.
» At (1, -1, 1): fee = — 2sfey = — V4Se = — WA,fy = 0, f,,= -1/4, £, = 0.
A
a» Atl, 1,1): fx = 66°, Say = ~ 40»Sea= 40° Sry = 6", yz = — 40°, f,, = 60°.
B. At (1, 2/2, 1/2): fx = —0°12 Sry = — 712s fra = — M2, fry = — [1 + (127S/4)], fy, = — [(0°8/4) - cl,
fee = — [1 + (77S/4)], S = sin (7/4), ¢ = cos (17/4).
9. At (1, 2, 3): fer = 9 Sy = — M4See= — Le Sy = 14, f,. = - 1/9,f,= 4/27.
10. f,,=f In (ex) In (ey).
 1S
rh oO Engineering Mathematics
= x, f(0, 0) = 1s fx, 0)=
LL. f(0, 0) = 0,f(0, 0) = 05(0, 9) = 0.40% 9° le y)=fy y) = — yl? + y?)3/2
:
12. fy =f yex! CU +y In x).
Yar en
15. 4(1 + 2y*)z @ x+y"
14. (1 + xy) (cos ze”.
t= 0, we getx= 0, y= 0, dz/dt= — 72.
16. For ; See sin bun?
17. Ox/d = 3u/x,
u u
= 5uly;
dy/d dx/du’= 3(x° -—3u yx", y Ay. 2-16.

18. Forx=1, y=-1,z=2, we getu=4;lv =2; (Quldx)y,z = 9/9; (OVI CY) x2 =

9. 2. > ot ee g
x? [22 ;
dk View de 1 dea
20. dy/dx=(e — 1)e + 1), d?yldx” = 2(e” + Ile + 1).

ae = u’'[1+v(
" In(ex), = n eu)’ J.
Ox =u" (v/u)
21. =
26. @= 3f or a= 4B andB # 0 arbitrary.
27. Note that u? + u? =v2 +? = I(x? +y")?, uy,+ Uyy = 0 = Vy + Vyy. We have

xx TF Zyy== fi Ux t Uyy) + fy (Ux, F Vyy) + Ti (u2 + us ) efi (v? F Uy).

28. Use x* + y? =r’, @= tan“'(y/x) and differentiate.

29. sinu=(x? + yx + y) isa homogeneous function of degree 1.

30.. e = [4x* —y*/x] is a homogeneous function of degre

e 0.
31. u is a homogeneous function of degree 1.
32. u is a homogeneous function of degree |.
33. w= tan u is a homogeneous function of
degree 2.
34. f(%, y) = 6 — S(x -2) + 3(y-2) + (x 2)
? +3y — 292.
35. f(x, y) =-2-2@-1)-(y- 1); B= 4;|E| < 0.08,
36. f(x, y)=(x- 1) + y; B= 4.6912: |E| < 0.0938.
CLEANED) yeaa Wytee ye 1) +&— 1)); B= 5.1; | E| < 0.002
38. f (x, Deis 4 e-tae- 3)] -
— 2 B= 0.0144),
| E| < 0.64 x 10-4, eq > rly1) * 2 — 1) & - 3) + (y- 3)%J;
$C Na1+ Qx+y)+
2 x+y? + 2 (2x4 yy} B=
23.87; |E} S 0.008.
40. L(
f y)
(x, y)==(+ 2y- )— F ws+y)? ay. B= 16 [sin(0
& (x .3)] = 4. 723.’ \E1<0315 x 103

aie
a seon= be He 8). ) | &
Pe a(-17
=
I. x 10,

| t/| 4 4 #af22— It at peli

Ad £1 -
42. f(x , y,z) =3 + 32 Ie = 2) += 2) 4
@
Dk B = 0.3872. |E}<0.017.
43. f(y, 2) =34 2(x- +2 (y
- 4).42 ( 3
a 55
t}e = 0.3985; |E |< 0.0179
 Functions of Several Real Variables 2.63

44. f%yW=xX+ytxz+yz;B= l.11; | E| < 0.005.

2 Ne ae
45. f@YyoD=1+x + fa= (y= 1? -(2-4] —fi(y- »(<-)} | B | = 7.0817;
—
to]

| E| < 0.0319.

Exercises 2.4

I. minimum value 9 at (3, 1). 2. maximum value a at (0, 0).

3. minimum value 0 at (0, 0) BE Bl 4. minimum value (3)*? at (3°-'%, 3-"”)
>: minimum value 5(2)~”° at (+ 23/19, 2-5). 6. minimum value —3/2 at (7/3, 27/3).
7. The matrix A or the matrix B = — A is not positive definite. The function has no relative minimum or
maximum.
. The matrix B = — A is positive definite and fix, fy» fez < 0 at (0, 0, 0). Maximum value is 0.
oo

9. A is positive de‘vinite andfe+.Syy»

Sez > 0 at (-1,-1,-1), (-1, 1, 1), (1, -1, 1), (1, 1, -1). Minimum value
is —1 at all these points.
10. B = — A Is positive definite and frxs
yy» fez < 0 at (W/3, 1/3, 1/3). Maximum value is (log 3) —1.
ll. No relative maximum and minimum. Absolute minimum value — 3 at (0, 1). Absolute maximum value
3/2 at (++/3/2, -1/2).
12. No relative maximum and minimum. Absolute maximum value 1/2 at (1/2, 1/./2 ) and (-1/./2 2D):
Absolute minimum — 1/2 at — 1/2, 1/2) and (1/2, — 1/2).
13. No relative maximum and minimum. Absolute maximum value 13 at (9/V13, 4//13). Absolute
minimum value — /13 at (— 9/4/13, — 4/13).
14, Relative minimum value 3/4 at (1/4, 0). Minimum value 3/2 on the boundary at (1/2, + 1/-/2 ). Absolute
minimum value 3/4 at (1/4, 0).
. Absolute minimum value 1/2 at (1/2, 1/2). Absolute maximum value 5 at (2, 2).
. Absolute minimum value — 93/18 at (1/6, 2/3). Absolute maximum value — 4 at (0, 0).

_ Absolute minimum value — 1/27 at (1/3, 1/3). Absolute maximum yalue 7 at (152);
. Absolute minimum value — 23/2 at (2, —3/2). Absolute maximum value 37 at (OA):
Absolute minimum value — 3/2 at (27/3, 27/3). Absolute maximum value 3 at (0, 0).
Absolute maximum value 1 at (0, 0), (0, 7), (7, 0) and (z, 7). Absolute minimum value —1/8 at
(1/3, 1/3), (22/3, 27/3).
F = f(x, y) + A@G, Y) => f,+Aod,=9 andfy + Ad, = 0. Eliminate A.
=- 1/2, (, y) = (1, 1/2); maximum value is 1/2; minimum value is 0.

A= 5/2. (x.y) = CVS, - 2/./5), minimum value is —./5:

A= -V5/2, (x, y) = (V5, 2/V5), Maximum value is ¥5.

Maximum valueis 3-¥3 - 7)/3 at (V3/2, 7/3). Minimum value ~ (33 + 52)/3 at (- 3/2, 51/3).
me value
. Extre V2.
26. : The points (4, ma A); (- 4, 4) are farthest, d? = 32. The points
(4/./3, 4/./3), (— 4/./3, _ 4/./3) are

nearest, d* = 32/3.
 Engineering Mathematics

. Rectangle must be a square.

. Triangle must be an equilateral triangle.
C2.
. The point is (AD/t, BD/t, CD/t), t=A*+ B+
_ Extreme value is a°/27 at (a/3, a/3, a/3).
t=at bt+e.
31. Extreme value is (a + b+ c)° at (t/a, t/b, t/c),
32. Extreme value is 3°7?"” at (¢, t, 1), t= 304.
33. Extreme value is 24 at (2, 1, 1/2).
34. Maximise V = 8xyz such that (x7/a”) + (y*/b?) + (z7/c*) = 1. We get (x, y, z) = (2alJ3, 2b/J3, 21 V3),
35. Maximise xy’z° such that x + y + z=, a constant. We get x= a/6, y= 4/3, z= al2.
36. Maximise 2(xy + xz + yz) such that 4(x + y + z) = a, a@ constant. We get x = y =z = a@/12, that is the
parallelopiped is a cube.
37. Maximise V = xyz such that xy + xz + yz = S/2, we getx=y=z= JS/6-
38. Minimise S = xy + 2xz + 2yz such that xyz =a. We getx=y= (2a)'* and z = x/2.
39. Maximise V = 2r7h/3 such that mrl = a where 1= yr? +h? . We geth = V2r.
40. Maximise V = 2r7H + (mr*h)/3 such that 2arH + mrl =S,1= Vr2 +h? . We get A/r = 2/+/5 and
Hir = WAN5.
41. Maximum value is 2/(3+/3) at (+ 2//3, + 2/./3, 1/3).
42. Extreme value is 3/2 at (1/2, -1, 3/2). 43. Extreme value is 11/12 at (1/6, 1/3, 5/6).

44. Farthest point (1, 0, 0), d = 1; nearest point (1/3, 0, 2/3) d= 5/3.
45. The coordinates of the points P and Q are obtained as (2a/3, 2a/3, 2a/3) and (ta/V3, +a/V3,
+a [3).
Shortest distance : d* = a? (7 — 4.3)/3; Largest distance : d* = a2(7 + 4/3)/3.
Exercise 2.5

1. Curves intersect at x = +4/2 , y=2* 16/9/73.

2. Curves intersect at (1, 1) and (4, —2);
9/2. 3. 8/3.
4. Curves intersect at (0, 0) and (1,
1); 1/10.
9. Ri m4
< O< m/2, Asin OS r <4 sin 6; 3(1
+24
ones
6. J=
f--f
ap f(x, y)dy dx + ifI I(x, y)dy dx, where f (x, y=ee
JI 2/2.

: og x7 +y
Te f=
¥ j.f(x, Yhdy aes| | I(x, y)dy dx, where f(x y)= aut
x=4 y=x-—4 : x + i ~

20 In (5) — 18 In (6) + (7/2),

> -

be eV dyde=(e-22.
oaeee 9, [= iate
V2)

10.
me[ue
du’
dv = 5 (e- LY, Wl. 14/275.
 Functions of Several Real Variables tJ -
2 La

12. 380/3. 13. x. 14. 817/16.

15. 128 a/15. 16. zi. 17. 2727/2.

18. 2a°/3. 19. (256 — 27m)/72. 20. 8(2- V2).

21. Ta. 22. 208/3. 23. 27.
24. a/90. 25. 4m} (4/2-1)/3. 26. 1/3.
27. 2m(1 — cos &)/3. 28. 1931/60. 29. 2na7/3.
30. Lae nls =I. 31. M=3n2ka/32, T=¥ = 8ka’/(105 M).
32. Evaluate the integral over 1 < x? + y? < q* and take the limit as a 3 ~. J = m/(p — 1).
33. 1/3. 34. 63/20. 35. 7/8.
36. 2mab/3. 37. (@° — &) m/16. 38. 1/96.
39. a*/280. 40. 4/3. 41. 21.
42. 8/3. 43. 1/15. 44, 1/48.
45. (8 In 2 — 5)/16. 46. 11/42. 47. 837/160.
48. 8/35. 49. 47 In (a/b). 50. 7/8.
51. t-abcl4. 52. 7/10. 53. abc(a* + b*)/60.
54. tha*/2. 55. 0.
In problems 56 to 60 compare the given integral with Eq. (2.107).
36. a= BP=y=2,a=b=c=2,p=q=r=1,1=4/45.
37. a=?) B=3, y=4,a=b=c=1,p=q-r- eer

38. 6442 7/81.

a= B=7=3/2,a=b=c=1,p=q=r=3,1=
59. a= 2, B= 3/2, y=2,a=b=c=1p=1,9=3,r=4,1=
1/288.
60. a=3,B=2,y=1,a=1,b=2,c=3,p=qar=2,1=78.
 Chapter ‘

Matrices and Eigenvalue Problem:

3.1 Introduction
In modern mathematics, matrix theory occupies an important place and has applications in almost
al
branches of engineering and physical sciences. Matrices of order m Xn form a vector space and
they
define linear transformations which map vector spaces consisting of vectors in IR” or C” into anothe:
vector space consisting of vectors in IR” or C” under a given set of rules of vector addition
anc
scalar multiplication. A matrix does not denote a number and no value can be assi gned to it. The usua
rules of arithmetic operations do not hold for matrices. The rules defining the operations
on matrice:
are usually called its algebra. In this chapter, we shall discuss the matrix al gebra and its use
in solving
linear system of algebraic equations Ax = b and solving the eigenvalue problem Ax
= Ax.

3.2 Matrices
An m X n matrix is an arrangement of mn objects (not necessarily distinct) in m rows and n columns
in the form

oh D 412 “ee! (GER

a2) 422 wee EO
A=]| . (3.1)

Am] Am? cae Amn

We say that the matrix is of order m Xn (m by n). The objects aj), a)2,..., a», are called the elements
of the matrix. Each element of the matrix can be a real ora complex number or a function
of one or
More variables or any other object. The element a, which is common to the ith row and the jth column
is called its general element. The matrices are usually denoted by boldface uppercase letters A, B, C.
-.. @tc. When the order of the matrix is understood, we can simply write A = (a;;). If all the elements
Of a matrix are real, it is called a real matrix, whereas if one or more elements of a matrix are
complex, it is called a complex matrix. We define the following particular types of matrices.
Row vector A matrix of order | x 7, that is, it has one row and n columns is called a row vector or
4 Tow matrix of order n and is written as
[aj;) 412 --> @m], or [a, a ... ay]
3.2 Engineering Mathematics

in which a@,; (or a) is the jth element.

| +8 Paes) at ix. 3 m rows and one column is Called a cy),
umn
Column vector A matrix of order m X 1, that 1s, it has
vector or a column matrix of order m and is written as

by

b>,
. fy OF

Dm

in which 6, (or 5;) is the jth element.

The number of elements in a row/column vector is called its order. The vectors are usually denoteg
by boldface lower case letters a. b.c,.. etc.Ifa vector has n elements and all its elements are reaq|
numbers, then it is called an ordered n-tuple in IR", whereas if one or more elements are complex
numbers, then it is called an ordered n-tuplein C”.
Rectangular matrix <A matrix A of order m X n, m # n is called a rectangular matrix.

Square matrices A matrix A of order m Xn in which m= n, that is number of rows is equal to the
number of columns is called a square matrix of order n. The elements a;;, that is the elements a,,,
@>>, - - . , @,, are called the diagonal elements and the line on which these elements lie is called
the principal diagonal or the main diagonal of the matrix. The elements a;;, when t #7 are called the
off-diagonal elements. The sum of the diagonal elements of a square matrix is called the trace of the
matrix.

Null matrix A matrix A of order m Xn in which all the elements are zero 1s called a null matrix or
a zero matrix and is denoted by 0.
Equal matrices Two matrices A = (2;;) mxn and B = (0; ))pxq are said to be equal,
when
(1) they are of the same order, that is m = p, n = q and
(1i) their corresponding elements are equal, that is aj;
= b; for all i, j.
Diagonal matrix A square matrix A in which all the off- diagonal eleme
3 | : nts a;.. jo | i+AJjaare zero iiss calle
called
a diagonal matrix. For example os

ay) 0
Ain ao
= 's a diagonal matrix of order n.

0 Ann

A diagonal matrix is denoted

by D. It is also written
If allI the
the el
elements of a diag
as diag lay)
ane onal matrix: of order n are equal, that is a;;= a for alleu.i, then
a eT ]

the matt!:

1S Called a scalar matrix

of order n.
 Matrices and Eigenvalue Problems 3.3

if all the elements of a diagonal matrix of order n are 1, then the matrix

's Called an unit matrix or an identity matrix of order n.

An identity matrix is denoted by I.

Submatrix A matrix obtained by Omitting some rows and/or columns from a given matrix A is
called a submatrix of A. As a convention, the given matrix A is also taken as a submatrix of A.

3.2.1 Matrix Algebra

The basic operations allowed on matrices are
(i) multiplication of a matrix by a scalar,
(ii) addition/subtraction of two matrices,
(iii) multiplication of two matrices.
Note that there is no concept of dividing a matrix by a matrix. Therefore, the operation A/B where A
and B are matrices is not defined.

Multiplication of a matrix by a scalar

Let @ be a scalar (real or complex) and A = (a;;) be a given matrix of order m x n. Then
B = aA = Q(a;) = (@a;) for all i and j. (3.2)
The order of the new matrix B is same as that of the matrix A.

Addition/subtraction of two matrices

Let A = (a,;) and B =:(b,) be two matrices of the same order. Then
C= (cj) =A+B= (jj) 45 (bj)) = (aj r bij), for all i and j (3.3a)
and D = (d,;) =A- B = (a,j) — (6) = (ay — 5,), for all i and j. (3.3b)
The order
of the new matrix C or D is the same as that of the matrices A and B. Matrices of the same
Order are said to be conformable for addition/subtraction.
s A be... A, ate p matrices which are conformable for addition and Qi, &,..., O, are any

C= aA, + %A,+...+@,A, (3.4)

7 d linear combination of the matrices A;, A»,..., A,. The order of the
matrix C is same as

e matrix addition and scalar multiplication

A, B, C be the matrices which are conformable for addition and @, 6 be scalars. Then
on. oA. (commutative law)
 3 ee
34 Engineering
- Tr
Mathemati
Tol du ws

t (A +B) + C=A + (B+ ©) (associative law).

3. A +0=A (0 is the null matrix of the same order as A).

4. A+(—A)=0. 5. a(A + B) = @A + GB.

6. (a+ BP)A= aA + BA. 7. Q( BA) = aPA.
S IXA=A and OxA=-0O.

Multiplication of two matrices

The product AB of two matrices A and B is defined
~ only when
) the numbersr of
of oe
CO ns “—
in A ‘e
JS ’equa:
to the number of rows in B. Such matrices are said to be conformable for mul tiplication. Let A = (a,)
be an m X n matrix and B = (6; be ann X p matrix. Then the product matrix

12 ? 1p
b 55 b 2p
= (C;; ) = —

Bn? np

mXn
it X p
is a matrix of order m x p. The general elem
ent of the product matrix C is given by

C= aj); We aj2b3; tt aes =P Ain Dn; = 2, ip dy;.

(3.5)
In the product AB, B is said to be
pre-multiplied by A or A is said
to be post-multiplied by B.
If A is a row matrix of order 1 x
n and B is a column matrix of ord
of order | x 1, that is a single er n X 1, then AB is a matrix
element, and BA is a matrix of
order n x n.
Remark 1
(a) It is possible that for two
given matrices A and B, the
product matrix BA may not be Product matrix AB is defined but the
defined. For example, if A is
matrix, then the product mat a 2 x 3 matrix and B is a
rix AB is defined and is 3 x4
product matrix BA is not define a4 matrix of order 2 x 4, whereas the
d.
| a
th the matrices AB and BA are
square matrices. In general AB * BA,
BA, then the matrices A Thus, the matrix product is not commutative.
and B are Said If AB =
to commute with each oth
(c) If AB=0 it does not always imp] er.
that either A = 0 or B = 0. For
example, let
x OF

y 0 a b
oO" Pe= 9% :
AB =
6 xp and BA =
fee by o|*AB
0
 Matrices and Eigenvalue Problems 3.5

(d) If AB = AC, itdoes not always imply that B = C.

(e) Define A*=AXA.
ia <aid to be bilshone (K times). Then, a matrix A such that A‘ = 0 for some positive integer
k is sald to be Yilpotent. The smallest value of k for which A* = 0 is called the index of
nilpotency of the matrix A.
i If A”= A, then A is called an ide
mpotent matrix.
properties of matrix multiplication

1. If A, B, C are matrices of orders m x n,n X p and p X q respe

ctively, then
(AB)C = A(BC) (associative law)
is a matrix of order m x qd.
2. If A is a matrix of order m x n and B, C are matrices of order n x p, then

A(B+C)=AB+AC (left distributive law).

3. If A, B are matrices of order m x n and C is a matrix of order n x p, then

(A+ B)C=AC+BC (right distributive law).

4. If A is a matrix of order m x n and B is a matrix of order n x p, then
a(AB) = A(@B) = (@A)B
for any scalar @.

3.2.2 Some Special Matrices

We now define some special matrices.
Transposeof a matrix The matrix obtained by interchanging the corresponding rows and columns
ofagiven matrix A is called the transpose matrix of A and is denoted by A’, that is, if

@i1 @12 a2} ee ete

a> 292 a2 m2
A= ;

Am2 42n Amn

By

-_ ATA =(n X m)(m X n) is an n Xn square matrix

cae A.
AA? = (m X n)(n X m) is an m X m square matrix.
ae es |
nn vector b can also be written as iby Os Bp) T

OWN g
i)

ee
results can be easily verified
rr.

ei,
 ineMathematics
1.6 Engineering Ma ey
| " ‘cormable
mable for addition.
BY r— = aTA‘ + 4B’,B’, whenwhe the matrices A and Bare| cant$n
. or f Ene tah [tiplication
tion.
etn
| i raT when the matr ices A and B are conf orma ble fo
4. (AB) =
oe BA‘, when the mé :

‘< defined
, then
If the product Aj Ap . - - A, is defined
PA As acyeg pet
rT T ay se

ane ) of order 1 Xn and a column vector b; = (b); by,

The product of a row vector a; = (aj, 42Peau
- in
i r pro duct of the v ectors ;
or thethe inne
u t or
oduc
a;a; and
and b,, that isS
b,j).| of order n X 1 is called the dot prod

cj = a,°b;= P» Ai OK;
Eq ((3.5) can be written as
in Eq.
which is a scalar. In terms of the inner products, the product matrix C in
a,b, a,: D2 f atie.« a,: Dp

a:b, ay: b> =e az-b, , (3.6)

Ceans ' = & 2 #8

anb, @n,:bz «.-. Dy»by

Symmetric and skew-symmetric matrices A real square matrix 2 = (aj;) is said to be

symmetric, if aj; =a, for all i andj, that is A= A
skew-symmetric, if ay =— aj for all i and j, that is A =— AT
Remark 3
(a) In a skew-symmetric matrix A = (a
ij), all its diagonal elements are zero.
(b) The matrix which is both symmetri c and skew-s
ymmetric must be a null matrix.
(c) For any real Square matrix A,
the matrix A +A’ is always
is always } skew-symmetric, The Symmetric and the matrix
refore, a real A — A’
Symmetric matrix and a Skew-s square matrix A can be written as
ymmetric matrix. That is the sum of a
A=z(A+A 41 (4 =iAT),
Triangular matrices A square matrix A = (q
whenever i < j, that is i) is Called a lower triangular
all elements above t
he principal diagonal
matrix if a; = 0,
matrix if a = 0, whenev are zero and an upper tri
er js J, that is all the angular
elements below the Pri
Conjugate matrix ncipal diagonal are zero.
Let A = (a;;) bea co
Then, the matrix A mplex matrix. Let
ai; denote the
= (4;;) is called th
e conjugate matrix complex conjugate of ay
Hermitian and skew m a
-Hermitian matrice
A=A! or A=(A)?
and a ske mix A is called an Hermitian matrix
He
rmitian matrix is denote it
d by A”
or A*, =—ATor A = —(A)7. Sometimes, 4
Remark4
(a) IfA is_a rea
maltrix, then an Hermitia
Hermitian matrix is sa m n Matrix
e as a Skew-symmetri eanis sq
ates me asas y Mmetrj
{ric matrix and a s kew-
 Matrices and Eigenvalue Problems 3.7
(b) In an Hermitian matrix, all the diagonal elements are real (letaj;=x,+iy,;;
then aj = ay gives
Apt typ = Xj—lyj OF y; = 0 for all j).
(c) In a skew-Hermitian matrix, all th € diagonal elements are either 0 or pure imaginary
3 Met ay = x + by; then ajj = —ajy gives xj + Ly; = — (x; —ty,;) or x; = 0 for all/).
(Jd) For any complex square matrix A, the matrix A + A is always an Hermitian matrix and the
. —T,
matrix A — A is always a skew-Hermitian matrix. Therefore, a complex square matrix| A can
be written as the sum of an Hermitian matrix and a skew-Hermitian matrix, that is

A=3(A+ A’) +2 (A - A’).

Example 3.1 Let A and B be two Symmetric matrices of the same order. Show that the matrix AB
is symmetric if and only if AB = BA, that is the matrices A and B commute.
Solution Since the matrices A and B are symmetric, we have

A’=A and B’=B.

Let AB be symmetric. Then
(AB)’=AB, or B’A’=AB, or BA=AB.
Now, let AB = BA. Taking transpose on both sides, we get
(AB)' = (BA)!= A’ B’ = AB.
Hence, the result.

3.2.3. Determinants

With every square matrix A of order n, we associate a determinant of order n which is denoted by
det (A) or |A |.The determinant has a value and this value is real if the matrix A is real and may be
real or complex, if the matrix is complex. A determinant of order n is defined as

ay| a}2 or Qin

G21 22s a, ,
det (A) = |A |melt ae : (3.7)

| 2 | a n2 “n8 3 Ann

Pikes discuss methods to find the value of a determinant. A determinant of order 2 has two rows
and twocolumns. Its valueisgiven by
a} 12
[A |= a2\ 422
= 4114
— 22
@)2a2).

nf x j : 3 .
We « Waluate higher order determinants using minors and cofactors.
: Pees tisctors Let aj; be the general element of a determinant. If we delete the ith row and
= from thedeterminant, we obtain a new determinant of order (n — 1) which is called the
ats
ANA
Aft a;,.
el)
We denote this minor by M;;. The cofactor of the element a,; is defined as
Ka ie
Ay = (- 17 My (3.8)
—e fsdeterminant of order n through the elements of any row or any column
oeah Lee
bo ce . The value
3.8 Engineering Mathematics

of the determinant is the sum of the products of, the elements ofF the ith (or Jjth column) and the;:
i rowv (or
corresponding cofactors. Thus, we have
nN
|Al= x
j=
+)
(-1)'4 ayy Mij
=
=
2
24 aijAy
ajyAj (3.9;a
when we expand through the elements of the ith row, or

5 It] oe Mis= a; ;Aijj (3.9b)

when we expand through the elements of the jth column. Generally,: we ex ae

: a Serminant
Auaiahe throu:
sees
that row or column which has a number of zeros. We can use one or more of the g properti
of the determinants to simplify the evaluation of determinants.

Properties of determinants
1. tf all the elements of a row (or column) are zero, then the value of the determinant is zero.
2. The value of a determinant remains unchanged if its corresponding rows and columns are
interchanged, that is

|A|=|A"].
3. If any two rows (or columns) are interchanged, then the value of the
determinant is multiplied
by (— 1).
4. If the corresponding elements of two rows (or colum
ns) are same, that is two rows (or
columns) are identical, then the value of the determinant
is zero.
5. If the corresponding elements of two rows (or
columns) are proportional to each other, then
the value of the determinant is zero.
6. The value of the determinant of a diagonal
or a lower triangular or an upper triangular
is the product of its diagonal elements. matrix
7. If each element of a row (or
column) is multiplied by a scalar
determinant is multiplied by the scal a, then the value of the
ar @. Therefore, if Bis a factor of
(or column), then this factor each element of a row
B can be taken out of the det
erminant.
Note that when we multiply
ree a matrix by a
multiplied by a. Therefore, | 7A |= aq" | scalar @, then every element of the matrix is
A |Where A is a
matrix of order n.
8. If each element of any ro
w (or
| column) can be written as the
the determinant can be writt sum of two (or more) terms, then
en as sum of two (or
more) determinants.
2. ifa non-zero constant multiple of the elemen
corresponding elements of some ts of some row (or colu
other row (or column), then the mn) is added to the
remains unchanged. value of the determinant
Remark 5
When ae peene of ee
foW are multiplied by a non-ze
corresponding elements of ro constant k and added to
the ith row, we denote this the
the ith row of |A |.The elements of the ope rat ic 7 SEER, where
Jth row remain unc
ihan egedeewhereas the* Bh: R; is
ith row
ith row getget cha
chan ged. This opera tionion is ele whe
men re
ts of R,
the
nged. Thi Ope ; called
rat an elemen| tary row operation. Simi |
larly, the
 Matrices and Eigenvalue Problems 3.9

operation C; <— C; + kCj, where C, is the ith column of |A |, is called the elementary column
operation. Therefore, under elementary row (or column) operations, the value of a determinant
is unchanged.

40. The sum of the products of elements of any row (or column) with their corresponding cofactors
gives the value of the determinant. However, the sum of the products of the elements of any
row (or column) with the corresponding cofactors of any other row (or column) 1s Zero. Thus,
we have
A | A I i =J :

~ Dip Ajg= (expansion through ith row) (3.10a)

0 i#j
or y ay:THEE
ga Ay; = |ae 3 ion through
(expansion through jthjth ccolumn) ; (3.10b)
J.

11. |A+B |+|A|+]|B|,

in general.

Product of two determinants

If A and B are two square matrices of the same order, then
|AB|=|A
|| B].
Since | A |=| A’ |,we can multiply two determinants in any one of the following ways
(i) row by row, (ii) column by column,

(iii) row by column, (iv) column by row.

The value of the determinant is same in each case.

Rank of a matrix

The rank of a matrix A, denoted by r or r(A) is the order of the largest non-
zero minor of |
Therefore, the rank of a matrix is the largest value of r, for which there exists at least one r X r
submatrix of A whose determinant is not zero. Thus, for an m X n matrix r < min (m, n). For a square
Matrix A of order n, the rank r =n if |A |#0, otherwise r<n. The rank of a null matrix is zero and
a null matrix.
if the rank of a matrix is 0, then it must be
Example 3.2 Find the value of the determinant
He et ah

23. 9
tary column operations.
" (i) us ng elementary row operations, (i1) using elemen

@A yplying
theoperations Ry < Ro — (3/2)R; and R; — Rs + (1/2)R), we obtain
| i ae
|A|=|0 w2 11/72
Be 9726 3/2)
3.10 Engineering Mathematics

Applying the operation R3 <— R3 — 7R, we obtain

2 1 -l,
}AJ=/O 1/2 11/2] = 2(1/2)-37) = -37
On O paG7
since the value of the determinant of an upper triangular matrix is the product of diagonal elements.
(ii) Applying the operations C, < C, — (1/2)C, and C3 < C; + (1/2)C), we obtain
ow 0 0
[AlS| 3 1 t12
en ele.
Applying the operation C; < C, -— 11C), we obtain

2 6 0
‘Al=) 3 ie 0 | = 2(1/2)(—37)
= —37
= 712) 37
since the value of the determinant of a lower triangular matrix is the produc
t of diagonal elements.
Example 3.3. Show that

] 1 ]
JAl=|a@ By |=(a-B\B-yy\y-
a).
a2 B? y?

Solution Applying the operations C, C, — C3 and Co <— C= C3, we get

| 0 0 ] 0 0 ]
|A|=|a-y Boy Y |=(@-y)B-y)}|
Dewiyle |ee
A220 1 l Y
a ey 2 a a ee
]
=(a@-y)(B-y)
a+ y Bt+y =(@— y(B-y)\(B+ y— a - 7)
=(a-y)(B-y)(B- a) = (a- B)(B-y)(y- a),
Example 3.4 Given the matrices

3 9
A=/|4 5 6), B=/] =—s 9
(jm See? l i
verify that | AB [=|A|| BI.
Solution We have

hel 2 Sie Pee 5 8 13

Bee 5. 6 Pg
8
| hye 25 20
te =—B Oly © Ie] 22 117 Q
 Matrices and Eigenvalue Problems Fa

therefore, AB |AB/=/2
|= 3 25)« 29] = gy _ 2340) — 23(0 — 936) + 22(260 — 200)
84 a ais
= — 18720 + 21528 + 1320 =
4198
We can find the value of the product |A|| BI direct
ly as

ie ney. 1) [ae =a — as i
Sahil =5 3 = 28 9 45 (multiplying determinants
=—$§ 9/)/1 ji 14 65 =A0 row
by row)

= 10(360 — 2925) — 28(120 — 1170) + 14(— 135 + 162)

= — 25650 + 29400 + 378 = 4128.
We can also find |A |and |B |and then multiply.
Example 3.5 Without evaluating the determinant, show that

cos(A-—P) cos(A—@Q) cos(A-R)|

D=|'cos(B—P) cos(B=Q) cos(B=—R)|=0.
cos(C-—P) cos(C-—Q) cos(C-R)}]

Solution Expanding all the terms, we have

cosAcosP+sinAsinP cosAcosQ+sinAsinQ cosAcosR+sinAsinR |

D=\|cosBcosP+sinBsinP cosBcosQ+sinBsinQ cosBcosR+sinBsinR
‘cosCcosP+sinCsinP cosCcosQ+sinCsinQ cosCcosR+sinCsinR

sinA 0O|]|cosP sinP O

cosA
sinB O||cosQ sinQ 0/=0x0=0.
=|lcosB
sinC Oj} | cos R sink 0
cosC

is at least
on Since the matrix A is of order 4, r (A) < 4. Now, r (A) = 3, if|A |= 0 and there
a na Si ee: of order 3 whose determinant is not zero. Expanding the determinant through the
 3.12 Engineering Mathematics

:
rc 0] 0) 0)
[Ale al 0 Li lL} 4 () Li Lj me ula — 6) + Ll] -6
| () G |

= Ml - Ou" t+ llu—-6 = (d= 1) = 2)(M — 3).

Setting |A |= 0, we obtain v= 1, 2, 3, For fis 1, 2, 3, the determinant of the leading third orde,
submatrix

Mo | 0
lArJ=/0 aw -ll=u°¢0.
PM eS Te
Hence r(A) = 3, when f= | or 2 or 3. For other values of li, r (A) = 4.

3.2.4 Inverse of a Square Matrix

Let A = (a, be a square matrix of order n. T hen, A is called
a
(1) singular matrix if|A |= 0,
(ii) non-singular matrix if |A |# 0.
In other words, a square matrix of order n is singular
if its rank (A) <n and non-singular if its rank
r (A) =n. A square non-singular matrix A of order
n is said to be invertible, if there exists a
singular square matrix B of order n such that non-

AB = BA =I (3.11)
where I is an identity matrix of order n. The
matrix B is ca lled the inverse matrix of
B= A™ or A= B™. Hence, we say that Aq!
A and we write
is the inversé of the matrix A,
if
AA = AA"! = 1p
The inverse A~' of the matrix A is given by (3.12)

(3.13)
where adj(A) = adjoint matrix of A
= transpose of the matrix of cof
actors of A.
Remark 6
(a) (AB)" = B1A-!,
This result can be easily proved.
We have
(AB)(AB)"! = |.
Premultiplying both sides first by
A~! and then by B™ we obtain
B~'A™' (AB)(AB)"! = B™(A™'A)
B(AB)"! = BI A- or (AB)! = B-4-!
In
In gen
genera
e l, we
Wi hav e
lve
(A, A>..
: . A, “+=
)'= A>AngBe
!a- er
- Bi2a:
by IfAand B are non-singular mat
rices, then AB is also a non-singu
lar matrix.

4
 Matrices and Eigenvalue Problems 3. 13

(c) if
1 AB=O0andA isa= non-singular matrix, then B must be a null matrix, since AB = 0 can be
ie a non-singular matrix, then A must be a null matrix, since
AB= 0 can be AE
‘premultiplied post multiplied by B-.
(d) It
If AB = AC and A is a non-singular matrix, then B = C (see Remark 1(d)).
(e) (A + B)'#A'+B" in general.
——<—<L— LLU

properties of inverse matrices

1. If A” exists, then it is unique,
ee =A
meee = (A~)’,.( From (AA>')7 = 1 = 1, we get (A“')’A’ =I. Hence, the result).
.Let D = diag (d), dy,,... , d,,), d;, #0. Then D™!= diag (1/d,,, I/dy, .. . , Wdyn).
5< The inverse of a Gino. upper_or lower triangular matrix is a an upper or a
lower triangular matrix.
|

6 The iinverse of a non-singularssymmetricmatrix is a symmetric matrix.

——— — a

a a (a7 )"=A” for any positive integer n.

2
Example 3.7 Show that the matrix A=|5 1 0 | satisfies the matrix equation A° — 6A*+
Pn
HNA-I=0 Where I is an identity matrix of order 3. Hence, find the matrix (i) A”! and (ii) A-.
Solution We have
: Oo -l Jai)
Aa |5 15 =a
] 5 4 9

. el -1 3 -6 -19
are )15 1] O;=|35 -4 -30
A “ae ile? 3\ so ia oD
7 g in B= A’ - 6A?+ LIA -I, we get
= _
24 -6 -30| 0 -11 lo ©
30 |— |90 6 -30/+|55 11 O;-|0 1 =O
Parag 24 54 itt 33 Or 0" "1
 3.14 Engineering Mathematics

(i) Premultiplying A? - 6A* + LIA-

| I=0
= by A rh, we ggel
A143 6A7A2 + LIATA-A™ =0

or 111
A=A*-—6A+

5 ie 6 =o) (ih oO 3 !
ap
oy =slel30 6 Ol+| 0 11 Ole) —1o 6
sigs
; 5 a 9 foe 19) Oo © 11 5 2
3,

Bh ad | =i | 0 nn 10

Gi) A-*=(A™)? =|-15 6 ~—5 5° =§'|= |=160 Gin 55)

5 -2 2 — 4, 2 35 = 2] 19

3.2.5 Solution of n x n Linear System of Equations

Consider the system of n equations in n unknowns
41,X%, + Qjz%_2 +... + Ayy,X, = b,

49,X1 + 0x2 XZ +... + Gz, X, = bo

See Ree eee ee er hl

Qty + GpoXo +... a,x, = D,.

In matrix form, we can write the system of equations (3.14)

as
Ax=b

Sy Gig ake} @}, b, 41 1

where A= e ee a b= 72 xa|*
| Any an? mete Ann b,, ee. iN

and A, b, x are respectively called

the coefficient matrix, the right
solution vector. If b # 0, that is, hand Side column vector and th
at least one of the elements b,, b>
of equations is called non-homogene -..,, is not zero, then the syster
ous. If b = 0, then the syst
System of equations is called con ,
em of equati ons is called homogeneous. Th
sistent if it has at |Cast
solution. One solution and inconsistent if it has n'

Non-homogeneous system of equ

ations
The non-homogeneous system
of equations Ax = b can
be solved by the following met
hods.
Matrix method
Let A be non-s ingul
on-sin gularar. remult
Premulti niy; g Ax=b
iplyin a by AW = we Slee.

Xx=A7'b.
(3.16)
 Matrices and Eigenvalue Problems 3.15

The system of equations is consistent and has a unique solution. If b = 0, then x = 0 (trivial solution)
is the only solution.

Cramer's rule

tet A be non-singular. The Cramer's rule for the solution of Ax = b is given by

_ [Ai]
— TAT ea)eon a"eee
ee (3.17))
where |A;| is the determinant of the matrix A; obtained by replacing the ith column of A by the right
hand side column vector b.
We discuss the following cases.

Case 1 When | A |+ 0, the System of equations is consistent and the unique solution is obtained by
using Eq. (3.17).
Case 2 When |A |= 0 and one or more of |A; |, i= 1,2,...,m are not zero, then the system of
equations has no solution, that is the system is inconsistent.

Case 3. When |A |= 0 and all |A;| = 0, i= 1, 2,...,n,.then the system of equations is consistent
and has infinite number of solutions. The system of equations has at least a one-parameter family of
solutions.

Homogeneous system of equations

Consider the homogeneous system of equations

Trivial solution x = 0 is always a solution of this system.

If A issoe stnenlar, then again x = A_'0 = 0 is the solution.
=
A WwW i
= -
| Tt: «all a. eet
a SF.

—rations
s hasinfinite number of solutions.
s
- -..
3 ‘

Exa
Exampl le 3.8 Show that the system of equations
=

ye 1 l

1
~3}=1(1+3)-2-1-1+13-1=10+0.

singular and the given system of equations has a unique

 3.16 Engineering Mathematics

(1) We obtain

4 2 2 4
At=tI)_5 0 5) and b=/0
lis = 24" 3 2

4 2 214 2|
a ='§ Hj » Sko = | —]
Therefore, x=A'b=
1 =f 32 It

Hence,
x = 2, y=— 1 and z= 1.
(ii) We-have

i es| I
|A,|=]0 1 -—3)=4(1 + 3) -0+2@G
—- 1) =20°
2 it l

4 I
[A,/=|/2 O -3]=1(0+ 6) —2(4 -2)+ 1-12 —0)=-10-
2 l

il esjh
| A3|=|2 I 0O}= 12-0) -2(-2 —4)+ 100 -4)= 10.
l LE 2

. ais | Ai| =e | A> | ok | A;eu |Bee

Therefore,
ae a r=
a = oF
ae bapa
Example 3.9 Show that the system of equations

] —] 3 Xy 3

2 3 ] X94 2

3 eZ 4 x3 5

has infinite number of solutions. Hence,

find the solutions
Solutions We find that

Pyeng 3 19)
A 8 Pde Weel ei, a 4 = 0,
ime. 2 4 5 4
ann3 bi 1
ij SEH OF WAG On oo leng
3 4 3 5
 Matrices and Eigenvalue Problems 3.17
nerefore, the system of
Th 4c s al Bitar
y equation has infinite number of solutions. Using the first
two equations
%1-xX,=3- 3X3

2x; + 3X5 = 2— x;
and solving, we obtain xX; =(1] - 10 x4)/5 and x
satisfies the third equation. Pg= (5x; — 4)/5 where x; is arbitrary. This solution
Example 3.10 Show that the System of eq
uations
4 9 31x 6|
2 3 Iiyl=|2
2 6 2iHez is
is inconsistent. |
Solution We find that

4 3 6 3
(Ae 2) 3F TSO TA |2 "3 lao,
2 2 7 2

de Whee 3)
AE? 2 leg.
7 ey ee

Since |A |= 0 and |A; |# 0, the system of equations is inconsistent.

Example 3.11 Solve the homogeneous system of equations

leiligt 2 ale lmao

2 SF SBiy l= On.
rae Be No
Solution We find that | A |= 0. Hence, the given system has infinite number of solutions. Solving
the first two equations
| PET Sie
"we obtain
MM UULGITL xA == 13z y = — 8z where z is arbitrary. This solution satisfies the third equation.
LId, )
)

oa a eee |

_: Sls ail at ee
B|=| A||B |, (ii) |A+B|# |A|+|B|.
ipl5, 7), B= (3: 1s 2) verify that (AB)" = BYA’,
-“
ny
 3.18 s neering g Mathematics
Engi

or Fue
3. Show that the matrix A=|2 1 2] satisfies the matrix equation A* — 4A — 5I = 0. Hence.
7 7 |
find Av’. ia
l ]
4. Show that the matrix A =| 1 2 —3] satisfies the matrix equation A? — 6A* + 5A + III = 0,
2 -!l
Hence, find A“.
|0 1 -1
5. Forthe matrix A=|4 -—-3 4 |, verify that
I —3 4
(i) [adj (A)]" = adj (A"), (ii) [adj (A)J"' = adj (A).
me =—2e J
6. For the matrix A =| —2 3 114, verify that
] I, 46

G@) (A= (Atyt, (ii) (AT) =A.

2 -!l I ro 2 i
7. For the matrices A = | —] 2 -1|,B=|1 3 = 4/, verify that
1 -l 2 2 iO 9
(i) adj (AB) = adj (A) adj (B), (ii) (A+B)'# A! +B
. For any non-singular matrix A = (a) of order n, show that

(i) |adj (A) |=| A |", (ii) adj (adj (A)) =| A |""7A.
- For any non-singular matrix A, show that |A |= 1/| A |.
10. For any symmetric matrix A, show that BAB’ is symmetric, where B
matrix BAB’ is defined
. . re i me e B |is any y matrixi for which
, icht |the product
oduc

11 iL] If A is a symmetric matrix, prove that (BA~!)7(A~B1)-! = T w a | a |

matrices are defined. ( y( ) =I where B is any matrix for which the product
12. If A and B are symmetric matrices, then prove that
(i) A + B is symmetric, (ii) AA’ and ATA are both symmetric
(ili) AB — BA is skew-symmetric.
13. If A and B are non-singular, commut
ati ve and symmetric matrices then
prove that
(i) AB, (ii) AB, (iii) A~ Bo!
are symmetric.
14, Let A be a non-singular matrix. Show that
Gi) if 1+A+A*+...+A"=0, then A-! = A”
(ii) if1-A+A?-...+(-1)"A"= 0, then A-! = (— jy"-lqn
15. Let P, Q and A be non-singular square matrices of order
n and PAQ = I, then show that
16. If 1— A is a non-singular matrix, then show that A~! = QP.

ae. Vey eo ee) a

assuming that the series on
the right hand side converges
.

an
 Matrices and Eigenvalue Problems 3.1 Y

17. For any thre non-singular matrices A, B. C, each of order n, show that (ABC)! = CB! A“!,
pstablish the following identities: |
| ] w Ww |

Es 2 |
18. |’ rf 1 | =0, where w is a cube root of unity.
we 1 W

I be Bib
- _ —a
19. is = 0. 20. |b-c c-a a-b!l=0.
a Cc-a a-b b-e

be a* a? be ab ac =a ab ac
ee el =\ab ac bel 2 | gb 2 be |=4a2h2c2.
ren ie? ab ac be ab ac be Co

| (b + c)? a? a’

oa (a+c)? b> | =2abe(at b+c)3.

c? c? (a+b)?

a b Cc
24. | a? b? c* |=(a-b)(b- c)\(c-—a)\(a+b+c).
b+e cta a+b

12
| Z
%.\y z x| =(x3+y24+23-3xyz)?.
fae OY
1 1 1

Y 8, |=(@- B@- y)(a- 5\(B- y (B= d)(y- 6).

y3 53 |

(a+q@) sin(b+ a) sin(c + @) pan |

| sin (a + i) sin(b+P) sin(e+ B)}=0 foralla,b,c, a, Bandy.
4 :
sin(a+y
|s sin (b+ Y) sin(c + ¥) |
A.
a
ingng systemof equations:
y+z=2, -
2x a) -z= ak ye z= 0.

3x -ytz=3,—xt By t4z=6.
25
ue ; ™ which
isa = the system of equations
.— ky +20, ke +3y-kz=0, 3x+y-z=0
: so
(ii)neon-trivial| solution.
See LIRIMCeTING (VECO MICIEC

33. Find the value of @ for which the system of equations

2(sin 0)x+y—2z2=0, 3x + 2(cos 20)y+3z2=0, Sx +3y—-z=

has a non-trivial solution.

| = ,cx+cy
b6z=0 +z = 0 0, where a, b, ¢ are non-z, ero
34. If the system of equations x + ay + az = 0, bx+y+
and non-unity, has a non-trivial solution, then show that
ad b eerie! mapa. Ts
laa Teh lc
35. Find the values of A and y for which the system of equations
x+2y+z=6, x+4y + 3z = 10, x+4y+Az=yu
has a (i) unique solution, (ii) infinite number of solutions, (iii) no solution.
Find the rank of the matrix A, where A is given by

l 3 -4
36. ; ; MI ai 23 4 i.
s 5 4 a eae

eee “al ee
38. 2 eeatl
f= 3 39. Pp
ys
r

sai ss a ae Tie

ee) ees | | tam

40 3) bee Qe Se 4), (b) a oa Na. (b,
ce.+ 0,7 = 1,2.
Ce Oy ee a
-a, -b, -c, 0
41. Prove that if A is an Hermitian matrix,
then 7A is a skew-Hermitian matrix and if
matrix, then iA is an Hermitian matrix. A is a skew-Hermitian

42. Prove that if A is a real matrix and

A" 0 as n > ~, then I+ A is invertible.
43. Let A, B ben xn real matrices. Then,
show that
(1) Trace (@A + BB) = @ Trace (A)
+ B Trace (B) for any scalars @ and
(ii)
B
Trace (AB) = Trace (BA),
(iii) AB-BA =Tigs never true,
44. If B, C are nxn matrices, A=B+
; C ; BC
= CB and
3 C?=9
=U, th
for any positive integer p. ) +h
Sent ar pee ate + NC)
45. Let A = (a;;) be a square
matrix of order n, such that
|A |=(c-d) q y=4,i# janda,=c, j=; Then show that
""[o + (n - 1)d],
Identify the following matrices
as symmetric, skew
~Symmetric, Hermitian, skew-Hermiti
an or none of these:
] eo 3
Ta hb e
es). 5. 41 PTA Biel ax
4. ¢ ell
a ee
48.
Go +
‘ .
| 2+4i ]-j
Se -c TP -e eos) S| a 5; |
l+i 3+ 5] 6
 Matrices and Eigenvalue Problems 3.21

l 2+4; hei
0 24+ 4i 3
50. —2+4i =5 3-5i |,
Sl. |-2+4; 0 ere
palhea, 3 = Si 6
—-l|-j —-3-S5i 0
: pee
: ae 0 —| 1+;
3 “ak
oe |! Soa ot oe oy 8G
I l 0
.
—l4+i 0 i

: - bk 2 =i
ewe tai SS?) i at)
—-i I1+i 9) : ;

3.3. Vector Spaces

Let V be a non-empty set of certain objects, which may be vectors, matrices, functions or some other
objects. Each object is an element of V and is called a vector. The elements
of V are denoted by a, b,
¢, u, V, etc. Assume that the two algebraic operations
(1) vector addition and (ii) scalar multiplication
are defined on elements of V.
If the vector addition is defined as the usual addition of vectors, then

a b= (Gj, ao, .. « , &,) + (Dj, Do, ... , By) = (ay + Oy, ap + bo,.: «. , a, + B,).
If the scalar multiplication is defined as the usual scalar multiplication of a vector by the scaler a, then
OR= Ody, Ops s1000,14,) = (Bay, Ban... . , ‘Aa,).

The set V defines a vector space if for any elements a, b, ¢ in V and any scalars a, 6 the following
properties (axioms) are satisfied.

Properties (axioms) with respect to vector addition

l. a+b
is in V.
2.a+b=bea. (commutative law)
3. (a+b) +c=a+ (b+). (associative law)

4.a+0=0+a=a. (existence of a unique zero element in V)

(existence of additive inverse or negative vector in V)

(right distributive law)

(existence of multiplicative inverse)
3.22 Engineering Mathematics

The properties defined in 1 and 6 are called the closure

properties. When these two properties are
iplication,
r the vector addition and scalar mult
satisfied, we say that the vector space 1s closed unde
e not always be the usual addition ang
The vector addition and scalar multiplication defined abov need ors, but also on
; on the set V of vect
ds not only
multiplication operators. Thus, the vector space depen :
V.
the definition of vector addition and scalar multiplication on
when the ama :oe puters,
If the elements of Varereal, thenit is called a real vector space
lex fastee a * 2
whereas V is called a complex vector space, if the elements of V oe comp
s @,, | omplex numbers
be real or complex numbers or if the elements of V are real and the scalar
Remark 7
(a) If even one of the above properties is not satisfied, then V is not a vector space. We usually check
the closure properties first before checking the other properties.
(b) The concepts of length, dot product, vector product etc. are not part of the properties to be satisfied.
(c) The set of real numbers and complex numbers are called fields of scalars. We shall consider vector
spaces only on the fields of scalars. In an advanced course on linear algebra, vector spaces over
arbitrary fields are considered.
(d) The vector space V = {0} is called a trivial vector space.
The following are some examples of vector spaces under the usual operations of vector addition and
scalar multiplication.
1. The set V of real or complex numbers.
2. The set of real valued continuous functions f on any closed interval [a, b]. The 0 vector
defined in property 4 is the zero function.
3. The set of polynomials P, of degree less than or equal to n.
4. The set V of n-tuples in IR” or C".
- The set V of all m Xn matrices. The element 0 defined in property
4 is the null matrix of order
mx HR.

The <following are some examples which are not ve

ae. ctor spaces. Assume that usual operations of vector
addition and scalar multiplication are being used.
1. The set V of all polynomials of degree n. Let
P, and Q, be two polynomials of degree n in
Then, aP,+ BQ, need not bea polynomi V.
al of degree n and th us May not be
if Ph=x"+a,x""'+...4a, and Oye ax" + bie 42 bth in VMe°
For exam ple :
amp!
of degree (n — 1), “++ On, then P,, + Q, is a polynomia
2. The set V of all real-valued functions of one
variable x, defined and continuous on the clos
interval la, b) such that the value of ed
the function at b is some non-zero
example, letf (x) and g(x) be two elem constant p. For
ents in V. Now
, f(b) = 2(b)=p. Si b) + 2(b) = 2p:
J (x) + g(x) is not in V. Note that if p =0,
then V forms a ne 5 ao.
es vl ae ae the of all polynomials,
detined by a
with real coefficients, of degree n, where addition
a) and under usual scal ar multiplication. Show that V is
not a vector space.
Solution Let P,, and Q,, be two elements in V. Now, }
2n, whichisnot in V. Therefore, V does not et. ita
7

#
r:Tee is a polynomial of degree
 Matrices and Eigenvalue Problems 3.23
pxample 3.13 Let V be the set of all ordered
pairs (x y), where x, y are real numbers.
Let a= (x, ¥1) and b = (x5, y2) be two elements in V.
Define the addition as
D= (4,99) + Gro, Ya) = (2x A
1 — 3x2, y; — y2)
and the scalar multiplication as

A(x, 1) = (@x,/3, ay, /3).

Show that V Is not a vector space. Which of the properties are not satisfied?
Solution We illustrate the Properties that are not satisfied.

(i) a Ya) + 1, V1) = (2x - 3x, y. — yy) (%1, Yi) + (%2, ya).
Therefore, property 2 (commutative law) does not hold.

(ii) (G1, Yi) + Ca, y2)) + (x3, 3) = (2x, - 3X2, Yi — Y2) + (%3, y3)

= (4x1 — 6x2 — 3x3, y1 — yo — ys)

(x1, Yi) + (2, yo) + (3, y3)) = (41, yy) + (2x2 — 3x3, yo — 3)
= (2%; — 6x2 + 9X3, yi — Yo + Y3)-
Therefore, property 3 (associative law) is not satisfied.
(ili) 1(xy, yy) = 4/3, y,/3) # (1, 1).
Therefore, property 10 (existence of multiplicative inverse) is not satisfied.
Hence, V is not a vector space.
Example 3.14 Let V be the set of all ordered pairs (x, y), where x, y are real numbers. Let
4 =(x,, y;) and b = (x2, y2) be two elements in V. Define the addition as
a+ b= (x1, yi) + 2, Y2) = 1 Xs V1 V2)
and
the scalar multiplication as
ot halk L(x), V1) = (AX), Hy).
cay are not satisfied?
ey
-
may i
‘ =i
t i
a
‘is not a vector space. Which of the properties
fs!

lution Note that (1, 1) is an element of V. From the given definition of vector addition, we find
a 4

i
i

(x1, ¥1) + (ls 1) = Gr yi).

‘ ‘ nae 2?
the role of 0 element
this ( 1). Therefore, the element (1, |) plays
is true only for the element ent (1,
‘ed in proverty 4. Now, there is no element in V for which (a) + (— a) = 0 = (1, 1), since

a
Cent ees oC ap.—y1) # G, D.
sei

*
aia oe

Py a 8 7 | ty 5 isnot satisfied.
two scalars. We have
ie

i 2%
(3x), 394)
(a+ BOM y¥1) = 3@4» Yi) =
ar; .'
—
3.24 Engineering Mathematics

ee). ae (Pye 2
and a(x), y;) + BO, 1) = Ly, Y1) + 201, YW) = Gas Yi) + 2x1, 21) = (227, 27)

Therefore, (@ + B)(x;, y1) # @ (x), y,) + BCX, yi) and property 7 is not satisfied.
Similarly, it can be shown that property 9 is not satisfied. Hence, V is not a vector space.

3.3.1 Subspaces
Let V be an arbitrary vector space defined under a given vector addition and scalar multiplicatioy
A non-empty subset W of V, such that W is also a vector space under the same two operations of vecto
addition and scalar multiplication, is called a subspace of V. Thus, W is also Opies am De tw
given algebraic operations on V. As a convention, the vector space V 1s also taken as a Subspace of |

Remark 8
To show that W is a subspace of a vector space V, it is not necessary to verif y all the 10 properties 4
given in section 3.3. If it is shown that W is closed under the given definition of vector addition an
scalar multiplication, then the properties 2, 3, 7, 8, 9 and 10 are automatically satisfied because thes
properties are valid for all elements in V and hence are also valid for all elements in W. Thus, we nee
to verify the remaining properties, that is, the existence of the zero element and the additive invers
in W.
Consider the following examples:

1. Let V be the set of n-tuples (x, x.... x,) in IR” with usual addition and scalar multiplication
Then
(i) W consisting of n-tuples (x; x ... x,) with x; = 0 is a subspace
of V.
(ii) W consisting of n-tuples (x; x, ... X,) with x; 2 0 is not a
subspace of V, since V
is not closed under scalar multiplication (@x, when q@is a negative
real number, is no
in W),.
(ili) W consisting of n-tuples (X} Xx X,) With x. =x, + lis nota subspace of V, sinc
W is not closed under addition.
(Letx=(x,; x2 ... x,) with x2 =x, + 1 andy= O1 yo Yn) with y2 = y1 + |
be two elements in W. Then

RP=O
V ty, wot yo... Xn + Yn)
iS not in Was x2+ y2=x +y,+24#x, 4 y, + 1),
2. Let V be the set of al] real
polynomials P of degre < | + 7
multiplication. Then eree Sm with usual addition and scalat
(i) W consisting of all real pol
ynomials of degree < m with
P(O) = 0 is a subspace of V.
(ii) W consisting of all rea] polynomials
of degree < m with P(0) = 1 is not a subspace of V.
since W is not closed y nder
addition (If P and Qe W, then P + O¢ W).
(iii) W consis
: ting of all pol ynomials of degree < m al real
with
Vsj : | : 1 p positiIVE: C 3 jents J NOt {a«
1S
hen =P ¢ W). der scalar multiplication (If P is an element of W:

3. Let V be the set of all p xX

oie n real square matrices with usu:
multiplication. Then sual matrixTare eon and scala
additi scalar
 Matrices and Eigenvalue Problems 3.25
i) Wconsisting
(i) of all
ih g OF sy, Anan:
all ‘ymmetric/skew-symmetric | |
matrices of order n is a subspace of V.
(ii) W consisting of all upper/lowe
r triangular matrices of order n is a subspace of V.
(iii)
iii W consisting£ Of
of ; alln x py1 matri
matricces having real positveive elements is;
not a subspace of V
ay ig not closed under scalar multiplication (if
A is an element of W, then

4. Let Vbe the set of alln x n, complex matrices with usual matrix addition
and scalar multiplication.
Then
(i) W sale ihaof all Hermitian matrices of order n forms a vector space
when scalars are
real numbers and does not form a vector space when scalars are
complex numbers
(W is not closed under scalar multipl
ication).

Let Ar Ne gest
x-ly |

Let @= i. We get aa =iA =[ " i lew

xi+y bi
(ii) W consisting of all skew-Hermitian matrices of order n forms a vector space when
scalars are real numbers and does not form a vector space when scalars are complex
numbers.
K l x+y 1
= EW.
a —x+iy 21

—| ix-—y
Ket @=71. We get 1A — zW.
=—i1x—y —2

Example 3.15 Let F and G be subspaces of a vector space V such that F 7 G = {0}. The sum of F
and G is written as F + G and is defined by
F+G={f+g:fe F,ge G}.
Show that F + G is a subspace of V assuming the usual definition of vector addition and scalar
multiplication.

Solution Let W=F +Gandfe F, gé G. Since0 € F and Oe G, we have 0+ 0=0€ W Let

f+ g) and f, + g> belong to W where f, f) € F and g), 82 € G. Then
(f, + 21) + (f + 82) = (fi + fo) + (81 + 2) © P+ GW.
Also
forany scalar @, a(f+g)=af+age F+G=W.
Therefore, W = F + G is a subspace of V.
NOW present an important result on subspaces.
y, be any r elements of a vector space V under usual vector addition
ht Latvia
4nd scalar multiplication. Then, the se t of all linear
¥e combinations of these elements, that is the set of

‘ements of the form

 3.26 Engineering Mathematics
| | 7

is a subspace of V, where @, Q,..+, @ are scalars.

ons ofof Vv), V2, ye
‘nations sSe
ol UGE
Proof let W be the set of ; all linear combinati
r
-

W2= 2 bv;
f

Ww) = y avi and

j= =

be any two linear combinations (any two elements of W). Then,

W, + Wo = (a; + by)v, + (az + by)Va +... + (G, + OV © W
Ow, = (Ga))v, + (May)v2 +... + (Aa,)v, € W
Ow, = (Ab,)v,; + (@b2)v2 +... + (@b,)V, © W
and Q(W, + Wo) = Aw, + AW.
Taking @ = 0, we find that Ow, = 0 € W. This implies that w; + 0 =0+ w; = Ww;
Taking @ =— 1, we find that (— 1)w,; = (— w,) € W. This implies that w; + (— w,) = 0
Therefore, W is a subspace of V.
The elements v;, v3,..., Vv, are in the subspace W as
Vy,= lv, + Ov. +... + Ov,, v2 = Ov, + lvo+...4+Ov,,..
We say that the subspace W is spanned by the r elements Vj, V2, -..-, V,. Also, any subspace that
contains the elements v), V2, ... , ¥, must contain every linear combination of these
elements.
Spanning set Let S be a subset of a vector space V and suppose that every
element in V can be
obtained as a linear combination of the elements taken from S. Then
$ is said to be the spanning set
for V. We also say that § spans V.

Example 3.16 Let V be the vector space of all 2 x 2 real

matrices. Show that the sets

, SS Tacepleoms knee) ty...)

© le Me Ma} 3)
Solution Let X = i “|be an arbitrary element of V,
cd
(1) We write

a ob ie <C) Y «i 0 O
b : a|
; : F : 1 +b r
Since every element of V
: 4 ‘ al 4)
can be written as a linear
S spans the vector space V. combination of the elements of S, the set
(i) We need to determine the scalars
Qh, OX, Oi, Qs, SO that
 Matrices and Eigenvalue Problems R21

|}c 10
6dsje, : + & 2/9ae ed
Oo jb 1] a
q oad
:
Equating the corresponding elements. we obtain the system of equations
a Lt Oe + s+ Ape a, Oat Oat Gy Sd,
Q3,+ Ay=c, Qs
= d.
The solution of this system of equations is

Qs= d, A3=c-—d,Q,=b-c,a,=a-b.

Therefore, we can write

a d =(a-b)| r 6 |+o- 1 | a : |
ct A |.
; 0 0 eg | ee olln a
Since every element of V can be written as a linear combination of the elements of S, the set
S spans the vector space V.
Example 3.17 Let V be the vector space of all polynomials of degree < 3. Determine whether or not
the set
N={e7
+h are}

spans V?

Solution Let P(t) = ar + Bt? + yt+ 6 be an arbitrary element in V. We need to find whether or not
there exist scalars a), a>, a3 such that
ati+ Bt?+ yt+ d=aPt+ag(t> +) +a,(t?+t+ 1)
or at3+ + ay) f° + ant” + (ay +43) t+ a3.
Bt2 + yt + 5 = (a,
Comparing the coefficients of various powers of ft, we get
a, + a3 = @, a2 = B, an + a3 = Y, a3 = O.
‘The solution of the first three equations is given by
a,=a+B-%2=B,a=y-B.

tituting in the last equation, we obtain y— B= 6, which may not be true for all elements in V. For
stitutin
ple, thepolynomial + 20? + t + 3 does not satisfy this condition and therefore, it cannot be
= i ear combination 0 f the elements of S. Therefore, S does not span the vector space V.
as a lin
*3.2 Linear Independence of Vectors
vector space. A finite set {V}, V2, - ++ Yn} of the elements of V is said to be linearly
Pendent if there exist scalars %, 0,-- >> a, not all zero, such that
ars - QV; + QnV2z +... - nVn = 0. (3.20)

Peetsted only for & = 02 =---= % = 0, then the set of vectors is said to be linearly
ee ee
ion of linear dependence of V1, V2... ., ¥, can be written alternately as follows.
A175 FC
 3.28 Engineering Mathematics

Proof Let the elements v;, ¥2,.... V, be linearly dependent. Then, there exist Scalars
Q;, Ob, ..., Q@,, not all zero such that

OV, + OV Pr Cer vi + OV; + 3 Vi+1 Sa

O41 ee 2 OL, “=:
ny

Let a; # 0. Then, we can write

CL Qs OQ j-] Ofi+1 m e220 =

y.-=— = ¥j—| —/V¥2-..-- 15 fe Of: |Vitel — °° on ft
a Q; | \ a; Oj i 3

* __* .
= OV, + G2V.+...+ Aj 1Vi_y t+ GisiVisr t+ ---+ GnVn
where @;, @>,..., @;, are some scalars. Hence, the vector vy; is a linear combination of the vectors
Wis(Woae == » Wet Vite, s* «Ve
Now let v; be a linear combination of Vi; V2... == 5 Vy, Vig» - - - , V,- Pherefore, we have

Vj = Vj + GnV2 +... + G;-4Vj-1 + Gigi Viz] +... + 4,V,

where a;’s are scalars. Then

CWE FQN ahash Ap Wee (= DV; + Gp Vi Fo... FAV, = O.

Since the coefficient of v, is not zero, the elements are
linearly dependent.
Remark 9
Fq. (3.20) gives a homogeneous system
of algebraic equations. Non-trivia] solu
det (coefficient matrix) = 0, that is the vectors tions exist if
are linearly dependent in this case. If the det
matrix) # 0, then by Cramer’s rule, @ = (coefficient
02=...=aQ,=0 and the vectors are linearly independent.
Example 3.18 Let y, = (1, — I, 0),
v2 = (0, 1, - 1) and V3 = (0, 0, 1) be
the set of vectors {yj, V2, V3} is linearly elements of IR3. Show that
independent,
Solution We consider the vector equation

QV) + QV, + A3v, = 0.

Substituting for v,, v5, V3, we obtain

@%(1,— 1, 0) + @&(0, 1, - 1) + a3(0,0,1)=0

or
(Qi, — Of + ,—- + a3) = 0.
Comparing, we obtain a =0,-
a + @=Oand- + 03 =
Q = & = a; = 0. Therefore,
the given set of vectors is lin on of these equations 1s
ear] 0. The Soluti
Alternative y independent.

I o 0
det(v,, V2, V3)
= | —| ] 0 =1+0.
' 0 at a
 Matrices and Eigenvalue Problems 3.29
example 33.19 Let v, = (1,-1]. 0), v5
ir’. Show that the set of vectors iy, aems (0,(Oil 1,-1,v3=(0,2
ee
» <, 1) and vy= (1, 0, 3) be elements of
V3 V4} is linearly dependent.
Solution The given set of elements w; be linearly
SAtS will denendent if there ac: )
jot all zero, such that early dependent if there exists scalars OX, A, Ay, My,

COV) + Oav, + OV, + Ayvy =

0.
(3.21)
substituting for ¥), V2, V3, Vy and com
paring, we obtain
pig, A,==
O+ 0, _ 1 + QM.
ry + 203 =0,-
a.+ 05+ 3a,=0.
The solution of this system
of equations js
Q) = - Os, Qs = 3Q4/3, ly, =- 4a,4/3, 4 arbitrary.

Substituting in Eq. (3.21) and Cancelling @4, we obtain

r
5FV a
“Vit 4 ve+ vee0
Hence, there exist scalars not all ZCTO, such that Ea, (3. 21) IS satisfied. Therefore, the set oft vectors

is linearly dependent.

3.3.3. Dimension and Basis

Let V be a vector space. If for some positive integer n, there exists a set S of n linearly independent
elements of V and if every set of n + 1 or more elements in V is linearly dependent, then V is said to
have dimension n. Then, we write dim(V)=n. Thus, the maximum number of linearly independent
elements of V iis the dimension of V. The set S of.» linearly independent vectors is called the basis of
i Note that a vector space whose only element is zero has dimension zero.
Pieares 3.3 LetV be vector
a space.of dimension n. Let vj, V2,..., ¥, be the linearly independent
elements of V. Then, every other element of V can be written as a sea combination of these
elements, Further, this representation is unique.
Proof Let v be an element of V. Then, the set oatV},-++» V,} is linearly dependent as it has n + |
elements. Therefore, there exist scalars Q, @1,--++» Om; not all zero, such that

OgV + AV, +--+ + AnyVy = 9. (3.22)

e get Vv) +... + @,V, = 0 and since vj, Vo,..., v,, are vey
z dep€en = a, = 0. This implies that the set of n + 1 perches Vi, Wigan
ndent, v
we et —-= &2=
"nis linearly independent, which i is no t‘possible as the dimension of V is n.
Therefore, we obtain from Eq. (3.22)
v= > (—;/Qo )V;" (3.23)

hoe. viis linearcombination of n linearly independent vectors of V.

tia there be two representations of V given by

a - + avn + + AnVn and vy=b,v, + bov2 + - > * OY,
V=a,v, + 42.27 -- .
,Beats one
ane i
i. Subtracting
| these two equations, we get
 3.30 Engineering Mathematics

0 = (a, — b))v) + (az — 62)V2 +. 6 + (Qn — On)Vp

Since v), ¥5,..., ¥, are linearly independent, we g

a-b,=0 or a =6,i=1, 7i Reese)

Therefore, both the representations of v are same and the representation of v given by Eq. (3.23) jg
unique.

Remark 10

(a) A set of (mn + 1) vectors in IR" is linearly dependent.

SQ) A set of vectors containing 0 as one of its elements is linearly dependent as 0 is the linea,
combination of any set of vectors.
Theorem 3.4 Let V be an n-dimensional vector space. If Vy, Vo, . ~~» Ve K < im are linearly
independent elements of V, then there exist elements V,,), Vj.2, -. - » VY, such that {vj, Wane «2.5 Va} is
a basis of V.
Proof There exists an element v,,, such that v), V2,..-, Vis Vg, are linearly independent. Otherwis
e,
every element of V can be written as a linear combination of the vectors V,, V2,.-.,
Vy and therefore
V has dimension k < n. This argument can be continued. If n > k + 1, we keep adding element
s v,,,
Ve+2> +++» V, Such that {v), v2,..., v,} is a basis of V.
Since all the elements of a vector space V of dimension n can be
represented as linear combinations
of the n elements in the basis of V, the basis of V spans V.
However, there can be many basis for the
same vector space. For example, consider the vector space
IR?. Each of the followin g set of vectors
Qed, OF, 10. t= 19, (0.0 1)
Gi) [1 = 150], (0,0; A9s[19.39
(iii) [1, 0, 0}, [0, 1, 0], [0, 0, 1]
are linearly independent and therefore form
s a basis in IR?. Some of the standard basi
s are the following.
1. If V consists of n-tuples in IR”, then

ey = (TN, Or. £70), es = (0: EO sO), E TS GOO, 0 ?

1)
is called a standard basis in rR,
2. If V consists of all mx n matr
ices, then

Oo ne LO carat)

Bie SO 8, eA Oi rs 1h Oui ri and a

Oia 0
where lis located in the (r, s) location,
that is in the rth row and the sth col
’ standard basis. umn, is called its
i
For: example, if V consists a
of all 2 x 3 matrice S, then
any matrix d in V can be
written as
ey.
 Matrices and Eigenvalue Problems 3.31

X y Z 28 + Eig + chy + xB +yEy +2 Bs

_ If V consists of all polynomials | |

standard basis. poynomials P(t) of degree <n, then 1G at ce t"} is taken as its

fxample 3.20 Determine whether the foll

owing set of vectors {u, v, w} torms a basis in IR?, where
(i) ut (2, 2. Q), v= (Se Q, 2), i (2. = D 2)

(ii) U= (Oo, —1), v= (1,0. l), w= Gs 1, 3).

Solution If the set (U,V, w} forms a basis in IR3, then u, v, w must be linearly independent. Let
Ql, Q, O be scalars. Then, the only solution of the equation

QU + OV + Gw=0 (3.24)

must be @| = Q> = a = 0.
(i) Using Eq. (3.24), we obtain the system of equations

20, + 30%, + 20,=0,2a,-2a,;=0 and 2a,+2a,=0.

The solution of this system of equations is Q = Q) = @ = 0. Therefore, u, v, w are linearly

independent and they form a basis in IR®.
(ii) Using Eq. (3.24), we obtain the system of equations
— Q + 303 = 0, O, + 03 = 0, — &—- &% + 30, =0.

The solution of this system of equations is Q@ = Q, = a; = 0. Therefore, u, v, w are linearly

independent and they form a basis in IR.

Example 3.21 Find the dimension of the subspace of IR* spanned by the set {(1 0 0 0),
@1 0 0), (1 20 1),(0 0 O 1)}. Hence find its basis.
the vector equation
Solution The dimension of the set 1s $ 4 If it is 4, then the only solution of
260 1) 4 e010) Gel) =O (G.25a)
ad 0 0 0)+% (0 1 O 0) + & @
the system of equations
Should be = & On = Oy = 0. Comparing, we obtain
: ae = Mat Q,= 0.

_~*€ solution of this system of equations Is given by

5 i hae Ret at th, = 204, 03 =— where 4 is arbitrary.
A, = &4 .
Therefore,
is satisfied for non-zero values of @ 1, 2, @; and 4.
ee, the vector equation (3.258)
, r 7
| I ie iT coal ] “a ig - wie” oe , —_

; : h Sat a so A
j4 oat T
sf
} i Or ir
ot
dh .
be
‘e set
rb
18 less
th:
J an
=—T «

iat. a pt Sry - < >

 3,32 Engineering Mathematics

!I 2 O 1). Conside,
Now, consider any three elements of the set, say (1 0 0 0),(0 1 O O)and(
a
the vector equation

Comparing, we obtain the system of equations

0) + = 0, %+20,=0 and O; = 0.

The solution ofthis system of equations is 0 = Q) = 03 = 0.Hence, a — “shmeaty

independent. Therefore, the dimension of the given set is 3 and the basis 1s e Se Vectors
(1 0 0 0),(0 1 0 0),(1 2 0 1)}. We find that the fourth vector can be written as

(0 Ceha=a— 0 0 He20 1 0 OF 2 O ZF).

Example 3.22 Let u = {(a, b, c, d), such that a+c+d=0, b + d= 0} be a subspace of R*

Find the dimension and the basis of the subspace.
Solution uw satisfies the closure properties. From the given equations, we have
adrera=0 and baid=0. of a=—c—d and b=-d.
We have two free parameters, say, c and d. Therefore, the dimension of the given subspac
e is 2.
Choosing c= 0,d=1 andc=1,d=0, we may write a basis as {(—1 -1 O 1),(-1 0 1 Q)}.

3.3.4 Lnear Transformations

Let A and B be two arbitrary sets. A rule thatassigns to elements ofA exactly one element of B is
called a function or a mapping or a transformation. Thus,
a transformation maps the elements of A
into the elements of B. The setA is called
the domain of the transformation. We use capital
S etc. to denote a transformation. If T js letters 7,
a transformation from A into B, we write

For each element a A, we get a unique

element b € B. We write b = 7(a)
the image of a under the mapping or b = Ta and b is called
7. The collection of al] such ima
image set of the transformation T. ges in B js Called the range or
the
In this section, we shall discuss
mappings from a vector s
(wo vector spaces, both real or Pace into a vector space. let V and W be
complex, over the same field
V into W. The mapping T is said F of scalars, Let I be a mapping
to be a linear transfo rmation from
following two properties: or a linear mapping, if it satisfies the
(i) For every scalar w and
every element y in V

T(av) = aT(y), (3.27)

(1) For any two elements Vv}, Vo in V

T(v¥, + 3) = T(v\) + T(
: v5),
rT space, the product (3.29)
r defines a mapping fr wy and the s
om V into W. Since T( um'Yi + V2 are defined and are element s in V. Then
T(v\) + T(¥2) are in W. y,Y) ) and 7(v>) are in W. the product
The conditions given @7 (vy) and the sum
in Eqs, (3.27) and (3.28)
are equivalent to
Mav, + B Yo) = Tia v,) +
T(@ v5) = AT(v,) + BT(y>)
for y, and v» in V and any scalar
s @, B.
 Matrices and Eigenvalue Problems 3,33

Let V be a vector space

I of dimeng sak let the set {V}, V2, ..., ¥,}
Nension nm and , |
Jement Vin V can be written be its basis. Then, any
eleme as a iin... ee:
ritten as a linea combination of the elements
v,, ¥5,... , Vi
sretore, Rey
The vs OV) + MoVot+... + nV »
wh
ynere @,, @e,..., @ ane are scalare Shin zero,
Calars, not all If——- ;
T is a linear ; rr
transformation defined ;
in V. then
/ (¥V) _ T(@,v, + OV 5 a Vin)

= P(a)v,) + T(QoVo)+...+T7(a,V,)
= QL (vy) + QT (v5) +... 4 OT Xv;,).
Thus, a linear transformation is completely determind
by its action on the basis vectors of a vector space.
Letting &= 0 in Eq. (3.27), we find that for every element
v in V
T(Ov) = T(0) = OT(v) = 0.
Therefore, the zero element in V is mapped into zero element
in W by the linear transformation 7.
The collection of all elements w = T(¥) is called the range
of T and is written as ran (7). The set
of all elements of V that are mapped into the zero element by the linear transforma
tion T is called the
kernel or the null-space of T and is denoted by ker(T). Therefore, we
have
ker(T) = {v1 7(v)=0} and ran(T) = {T(v) lve V}.
Thus, the null space of T is a subspace of V and the range of T is a subspace of W.
The dimension of ran (T) is called the rank (J) and the dimension of ker (T’) is called the nullity
of T. We have the following result.
Theorem 3.5 If 7 has rank r and the dimension of V is n, then the nullity of Tis n — r, that iS,
rank (7) + nullity =» = dim(V).
We shall discuss the linear transformation only 1n the context of matrices.
Let A be an m Xn real (or complex) matrix. Let the rows of A represent the elements in IR”
(or C”) and the columns of A represent the elements in IR™ (or c”). Ifxisin IR", then Ax is in IR”.
Thus, an m X n matrix maps the elements in IR” into the elements in IR”. We write

T=A:IR"—> IR”, and 7x = Ax.

and
We now prove that the mapping A is a linear transformation. Let v,, ¥. be two elements in IR”
a, B be scalars. Then
T (a@V,) = A(Q@V,) = @Av,
T (B V2) = A(B V2) = BA
and Tr(a vy, + BY2) = A(a@v, + By.) = aAv, + BAvs,

The range is aa linear

o of T is line subsp
| ace of IR™ and the kernel of T is a linear subspace of IR”.
ue Tange

ear transformations
‘Zim and product of lin
+ T, to be the
nsficormations from V into W. We define the sum 7;
wo linearr tratrans

Sv =T7\v + T5V, VE Vv.

3.34 Engineering Mathematics

It can be easily verified that T; + T> is a linear transformation and T, + T> = Tz +7}.
Now, let U, V, W be three vector spaces, all real or all complex, on the same
field of Scalars. |o
T, and 7} be linear transformations such that

T,:U3V and 1,:V5W.

The product 7,7, is defined to be the transformation S from U into W such that

w=)5Su= T>(T,u), uc U.

The transformation 7 7, is also called a composite transformation (Fig. 3.1). The transformation
TT,
means applying first the transformation T, and then applying the transformation
T>. It can be easily
verified that TT, is a linear transformation.

U T,

TT;
Fig. 3.1. Composite transformation.
If T, : V— Vand T, : V—> Vare linear transformations,
then both 7,7, and TT, are defined and
map V into V. In general, T,7, # TT. For
example, let A and B be two nm Xn matrices
element in IR". Let 7, and T, be the transformations and x be any

T\(x) = Ax and T>(x) = Bx

from IR” into IR". Then

T,(T;(x)) = BAx and

T| (T>(x)) = ABx.
Therefore, TT, # T, T, unless
the matrices A and B commute.
Example 3.23 Let T be a linear transformation from IR?
into IR2 defined by the relations
| l
Tx = Ax, A= 2 3 A
45) 6
Find Tx when x is given by [3 4 5)".

Solution We have

: rl Sa |
Tx = Ax = : 2 3 Al = 26 |
ab * San G ah (62
 t Matrices and Bigey
wenvilie dnl
Mrobleny 4448-
Example 3.24 Let Tbe a ling A’ Cansfo

HG IHC OE} of:

rmation Hetine
d by
|
1/0 | |
| | |} | ie a i |- | 7[ | ei 7
: \ | ’ () | |

Fin 13 alt

Solution The matrices | ;|

2 1] OiTo 6
ave linearly independent and hence
SVD? Lea? i
form a basis in the space of 2 x 2 matrices,
We write for ANY SCAlANS Of), Q, Oy, Gy, Hot
all zero

od «|
wy
bl " «a
0 |
oe: 9 tig yk
+a
O10
|
Pe) dha
ae

X, XY) + Qs
i
(XX) + Mo + Qs Qi + AtAy t Ay

Comparing the elements and solving the resulting system of equati

ons, we get @, = 4, @» = |
Q@3;=—2, &, = 5. Since T is a linear transformation, we get

As oll droll droll Nee’ 9)

1) 1). f-1) (-2
7 = 4|2]+41/-2]-2/-2/+5] 2/=| 20],
: (3 3) or NeeShe ceele Speen bt
Example 3.25
* Ach. 0 | Let 7 be a linear transformation defined by
i er) ee ;
-d Tx = Ax, where Al {pee Gun?

Solution Let (9),y >)! be the point that is mapped into (3, 2). Therefore, we have

= > abalszh
a =»
: ,
| ae ing we obtaintthe ot
e system of equations y, + 2y) = 3, 3y, + 4y) = 2.
paring
, = _ — 7; ,
ana compal
= 7/2.
equations Is yy = — 4, yp
eee

th iS system of
-

0 ae as)
or
aa

x 4
3.36 Engineering Mathematics

7 r | rr _ l ‘ i / ‘ i I}2

Example 3.26 For the set of vectors {X), X2), where x) = (1, 3)’, X= (4, 6)", are in IR , find the
matrix of linear transformation 7; IR* — IR“, such that

TX ee z2
i 4 Jf and Tx, =(-2 4 1O)!.

Solution The transformation 7 maps column vectors in IR* into column vectors 1n IR“. Therefore
T must be a matrix A of order 3 x 2, Let

ay b,

A= ils b»

ad 4 b 1

Therefore, we have
ay b, , ee ay b, ie 2

a5 b 5) = and dy b? |= — 4 |:
3 6
dy b, _ ads h4 | _ 10

Multiplying and comparing the corresponding elements, we get

a, + 3b, = — 2, 4a, + 66, = — 2,

dy + 3b, = a 4a, + 6b, =- 4.

a, + 3b3 = —-7, 4a, + 6b, = — 10.

Solving these equations, we obtain

re Mey
A=|-4 2
ie hd

Example 3.27 Let Tbe

alinear transformation from IR? into IR*, where 7x = Ax. A = |me? '|
=, ar) :
x=(x y z)!. Find ker (7), ran(T) and their dimensions.

Solution To find ker(T), we need to determine all vy = (UV; v0

3° such that Tv = 0. Now
Ty = Av = 0 gives the equations

U,;+U,=0, -v,+v,=0
whose solution is v; =~ v5 = v4, Therefore y = Os et at
Therefore, dimension of ker(T) is 1,
Now, ran (T’) is defined as {T(v) | v
« V}. We have

Tv) = Av =| l :a U,
+ U2 |
ae 0 l r. —U,; + DU,
 Matrices and Eigenvalue Problems 3.37

= | ‘+ vag +
af oa)
| l) (1) (0 ;
| . hy

Since ¥ m 0 i the dimension of ran (T) is 2.

Example 3.28 Let T be a linear transformation Tx = Ax from

IR2 into IR?, where
2 |
A=/1 -1] and x=(?|
3 2 :
Find ker (7), ran (T) and their dimensio
ns.
Solution To find ker(T), we need to determine all v = (v, v2)" such that Tv = 0. Now,
Ty = Ay = 0 gives the equations
2U; +U2=0, v;-v,=0° and 3, + 2v,=0
whose solution is v; = v2 = 0. Therefore v = (0 0)’ and the dimension of ker (7) is zero.

2 ! 2 |
Nav V}
Now, ran = 7 Gy) Sill eI 5 =v,| 1}+0,}-1}-
Bie Dit 3 2
smcere 1 3)’, (1 —1 2)' ate linearly independent, the dimension of ran (T) is 2.
Example 3.29 Find the matrix of a linear transformation T from IR? into IR? such that

{1 6 by. (2 l 6
—4 - fh =) — 6 :
Ti l = 2 5 Ia —l/=

I 4 1 2 3 5

\he - Solution The transformation T maps elements in IR? into IR?. Therefore, the transformation is a
trix of order 3 x 3. Let this matrix be written as

Gj 2 G3]
=|B; B2 Bs |:
Vt Fo 3

B> B; —2|=/ 6).

| 15

Weioe 3 13/2 |.
| 1 2
 3.38 Engineering Mathematics
sar 3 in | defined by
Example 3.30 Let 7 be a transformation from IR° into IR
7 7 p)
wf ce i —. 7, Seay
T(x) \ PT) 5 *X4 ) = X | + x A, + 4

Show that 7 is not a linear transformation.

- ‘& |

any two
be any elements in IR~. Then
Solution Let x = (x,, x9, 3) and y = (1, Y2, 3)) be
K+ y = (X) + Vp X2 + V2, %3 F V3):
, 4

We have
, 4 : 2 2 2
PX) = co tat XS Tiyya=yy + Yo * ¥3

T(x + y) = (x, + y))° + Go + y>)° + (x3 + y3)° # T(x) + TQ).

Therefore, 7 is not a linear transformation.

Matrix representation of a linear transformation

We observe from the earlier discussion that a matrix A of order m Xn is a linear transformation whic
maps the elements in IR” into the elements in IR”. Now, let T be a linear transformation from
finite dimensional vector space into another finite dimensional vector space over the same field F. \
shall now show that with this linear transformation, we may associate a matrix A.
Let V and W be respectively, n-dimensional and m-dimensional vector spaces over the same fiel
F’. Let T be a linear transformation such that T: V > W. Let

x= {¥j, Vos 2 ey eae Y= {W), W2,---; Wr}

be the ordered basis of V and W
respectively. Let v be an arbitrary element in V and w be an arbitrar
element in W. Then, there exist scalars, a, 0), ... a, and B;, Bo, . . ., Bn» not
all zero, such that
V= QV, + OhV7+...+ GV, (3.29 1
w=f,w, + Bowit+...4 Bn Wm
(3.29 i
and w=Tv=T(aQ,v, + Wor =. .+ Oy.)
SSTWi On Ee ot a, T v,, (3.29 iil)
Since
Si @V ery element
) PV,iE
ES Lys - 118is in
7 W,eee it can
sea be written
. as a linear
; combination of the basis
oo Wir W2,- + +» Wm In W. That is, there exist scalars a; i= 1, 2 m= 1.5 wey
2 n not all
Zero, such that y ie ae > J — ee gy /

LV; t = ay:W,
ae + 3;a5; W aa me os Arn; .W
mM

= [Wy Wea
Wo. l :
(tps Aa... Ami], ) t=1,2,..., ee
0 (3.29 i)
Hence, we can write

30 Bie ae ip,

i . md 2
me Pe

a
 Matrices and Eigenvalue Problems 3.39
where A ts the m X n matrix

ay] a)2 ne Q) »,

Rew a>) a orev Arp,

el & (3.29 vi)

Ay | Am? tee Amn

The m Xn matrix A is called
the matrix representation of 7 or the matrix of T with respect to the
ordered basis X and Y. It may : X is a basis of the vector
: ae be observed that space V, on which ETacts and
Y S the basis of the vector Space W that contains the range of 7. Therefore, the matrix
representation
of P depends not only on T but also On the basis X and Y. For a given linear transformation T, the
elements aj of the matrix A = (@,;) are determined from (3.29
| v), using the given basis vectors in X
and Y. From (3.29 iii), we have (using (3.29 iv))
W = Gy (ay,W) + Q21)W2 +... + Qn Wm) + Op(ayow) +
Axx W2 +... + AnaWm)
+... t+ A, (G1 )W) Gg Water e AmnWm)
= (Q) ay, + Op a). +... + G&, ay,) W,
+ (Qi Az) + Ondx7 +. . . + A, Ar,)W2
Hs $ (QEyy + On Oy + os + OG, )Wr,
= Bw, + Pow. +... + Bw»,

where B; -_ Of, aj, = aj? fee OCC 1 Gins a Vs z. =e ey

Bp, Qy| a2 ses Gin | Q

B; a2) 492 -++ Ady | Qo
Hence, eo) |S=9! he |

; foam | Im Am? ese Amn X,

or B=Aa (3.29 vii)

where the matrix A is as defined in (3.29 vi) and
B = [Pi. Bs. cn Py. Bl’; a= [ Q, OQ, wos ey a)".

-a given ordered basis vectors X and Y of vector spaces V and W respectively, and a linear
formation T: V > W, the matrix A obtained from (3.29 v) is unique. We prove this result as

‘ ieA= (a,j) and B = (b,;) be two matries each of order m X n such that
—) TA=YA and TX= YB.

YA=YB

., Wm} is a given basis, we obtain a,; = b;; for all i and j and hence A = B.

a _ oT
 3.40 Engineering Mathematics

Example 3.31 Let T: IR? > IR? be a linear transformation defined by

Bh

yrtz
T| yi =
|
ioe

t to the ordered basis

Determine the matrix of the linear transformation 7, with respec

| Oil eesti
O/+inIR? and Y= } | in IR“
ko)’ (a)
|

(standard basis e;, @>,e; in IR? and @;, é>, in IR).

iNT EY C4
(ii) X= 4} 1},/0],| 1|/$ in IR? and Y= (obs) tim
1) \l Q WIM

((0) (1) (1
(ii) X= 4) 1],| 0},] 1]}in R$ and r= {('},Alas
Li} (1) lol] ake

Solution Let V= IR’, W= IR’. Let

X¥= {vj, vy, v3}, Y= {wy, w>}.
] Q
: l 0
(1) We have Vv,=| 0],v. =/ 1]|,v3= of.
wi=[o}wa=( 2]
0) O :
We obtain

{-6)-We-()°
i a (0) +()o
7

(al=(a}e+()}-0
0
T| 0 I|

heb
Tr ing the notation
given in (3.29 Vv)
, that is TX = YA, we
write
 Matrices and Eigenvalue Problems 3.4)

7a jj
T Iv, « ¥4q, V3
] = [w, Ww >|
MD ade»v &]

EEA
0

or T
tI
Therefore, the matrix of the linear transforma tion T with respect to the given basis vectors is
given by

Be l 0
(ii) We have Wy) 1),¥2=)0 |,v4 = ifm=[ohme=(1]

We obtain

()-(o-(im
(ede
' 1 "i ‘
()-G)e-()6
ry [ay ’ a
* 4 i i
‘ : a
“ alk ee x
u %
e
7 * —

il % ; \ t 7 f 1
Di | om lie)

|) | HEIN FON |F2 L 4

ah ia Ah i Li a, " j

| | } ) i ;
| | s i |

T aJ hil| a ia | a | ov
. | “ \ . ° } d i | j | : i

Di iT : . ~*~,
bal i . | ~ : | | ' | { i }
| Toy vt f Wt - 2 eo — a |
=» =! | Wi, BY ly Ky mee} ‘ue : —ae
3 z J 1 } ii ‘1 }
: ' Tt | i i | ' Nie
] ' 7 i i ‘ i 7
| ao 4 ll \igy A 4
; : oh ogee to, et

with respect
to the given basis vectors is
i ik
ri : a. 1 =r ; re + “VT = er ) 7 |

|
‘the i.
near
LeBee Lu
TAalloaliVilil
— ri t
ee a te — » ;
 3.42 Engineering Mathematics

=
a, WwW
2 -—
-
We have ace | Faas Ova = | bh
(iii) | |;
i
| | Q)

We obtain

fH OeCo-Ce
Using (3.29 v), that is 7X = YA, we write

AHH ROGUE ts
Haka ] ()

Therefore, the matrix of the linear transformation T with respect to the given basis vectors 1s
given by

A= |
fit
i Oe 1

Exercise 3,2
Discuss whether V defined in Problems 1 to 10
is a vector space or not. If V is not a vector space, state which
of the properties are not satisfied.
1, Let V be the set of the real
polynomials of degree< Fe pice : =
scalar multiplication, gree Sm and having 2 as a root with the usual addition and
Pe Let V be the set of all real polynomia
ls of degree 4 or 6 with the usual addition and scalar multiplication,
= Let V be the set of all rea] pol
ynomials of degree > 4
with the usual addition and scalar multiplication.
Raa Let V be the set
of all rational numbers wit
h the usual addition and scalar
wm Let V be the set of all Positive rea] multiplication.
numbers w; addition
mbers with We defined as x + y = xy and usual scalal
:
multiplication,
6, Let V be the set of all ordered pairs bx
(x
F
+u, y + v) and j scalar multiplication 4 y)
y. l
in IR? 2 with
]
vector; addition
wr
defined as (x, y) + (4, ¥)~ *
efined as A(x,
y) = (Bax, y).
 Matrices and Eigenvalue Problems 3.43
et V be the set of all ordered t+; | |
a wu eplets (XY, 2), x, y,Z€ IR, with vector addition defined as
(X, y, z) + (4, UV, w) = (3x + 4u,
y-2 U,Z+ Ww)
and scalar multiplication defineg as

A(x, y, z) = (ax, Oy, 2/3),

g. Let V be thehe
set of all positive rea] numbers with additi
on defined as x + y = xy and scalar multiplication

9, Let V be the set of all real valued continuous functions bh

fon {a, b} such that (i) | f(x)dx =0 and
b
(ii) | f(«)dx =2 with usual addition
4 and scalar multiplication.
10. Let V be the set of all solutions of the
(i) homogeneous linear differential equation y” — 3y’
+ 2y = 0.
(ii) non-homogeneous linear differential equation y” — 3y’ +
2y = x.
under the usual addition and scalar multiplication.
Is W a subspace of V in Problems 11 to 15? If not, state why?
ll. Let V be the set of all 3 x 1 real matrices with usual matrix addition and scalar multiplication and W
consisting of all 3 x 1 real matrices of the form

a a a a
fe |, (ii) Ja |, (ii) |b], (iv) |b}
a+b é | 2 0
12. Let V be the set of all 3 x 3 real matrices with the usual matrix addition and scalar multiplication and
W consisting of all 3 x 3 matrices A which
(1) have positive elements, (ii) are non-singular,
ae a
(iil) are symmetric, (iv) A°=A.
13. Let V be the set of all 2 x 2 complex matrices with the usual matrix addition and scalar multiplication
: Meas 1S X+1y
and W consisting of all matrices of the form ig 4 Iwhere x, y, z, & are real numbers and

q (i) scalars are real numbers, (ii) scalars are complex numbers.
| 14, Lat or att real polynomials of degree < 4 with the usual polynomial addition and scalar
.; duultiplication arid W consisting of polynomials of degree < 4 having
ties a (ji) coefficient of 1? as 0,
a (iv) only real roots.
i) coefficient of ¢° as 1,
5.Let V be the vector space of all triplets of the form (%), *2, x;) in IR? with the usual addition and scalar
triple lets of the foém (x, X2, X3) such that
multip W
and be the set of trip
(ii) x, = %2 =%3 + is

: 2 2 <A
3.44 Engineering Mathematics

Rie(4, 3, 10),| (ii) 3, 2, 5); (iii), (— 2, 1, — 9).

(i) (2, 3, 1, — 1). Determine whether or not x is a ling
i, 1) and w=
17. Let u = (1, - 2, { 3). Vail, 2=
given by
combination of u, v, W, Where x Is
(iti) (4, 3, 0, 3).
(i) (G, 0, 5,-— 1), (ii) (2,-7, 1, 11),
ination of P,(1), py,
18. Let P,(t) = —4t- 6, P2(t) = 21 2 — 8. Pi) = 2t = 3, Write P(t) as a linear comb
P(t), when
(ii) P(t) = 2t? - 3t — 25.
(i) P(t) =—-1t7? +1,
that the set
19. Let V be the set of all 3 x 1 real matrices. Show
1 1) (0)
S= 2) 1],| -L}],| O];7 Spans V.
| 0 0) \l

rf 3G HC HE hom
20. Let V be the set of all 2 x 2 real matrices. Show that the set

21. Examine whether the following vectors in IR?/C? are linearly independent.
(i): (2) 2 1h = 1, DD, COs), Gi) 92933), Sly (G, 7, 8),
(iii) (0, 0, 0), (1, 2, 3), G, 4, 5), Gy): 2, 1,—= DIX =3, 4); i, = 1,5);
(vy) C1, 3,4), G, 1,0), G, 45-2); d, = 2,1).
22. Examine whether the following vectors in IR* are linearly independent.
Wi Geely 2; =), C1 1,:05,3), G,— 1, 0,.2), (= 2),.1,.0, 3),
(i) i; 2,3, 1), (2) L, = 1,1); 4, 5,5, 3), G, 4, 1, 3),
Gi), (8 2-3,.4), (2:0) 1 = 2)Gy ara oy,
(iv) (, 1,0, 1), 1, 1, 1, 1), @1,-1, 1, 1), C, 0, 0, 1),
(v) SCl,.25:3,—. 1) (Os; = 1, 2), (hd; 1,8) Cl 78:
oy
23. If X, y, Z are linearly independent vectors in IR3, then
show that
(l) X+y,y+z,z2+x; (li) XX+y,x+y4z
are also linearly independent in IR3.
Write (— 4, 7,9) as a linear combination of the elements of the set S : {(1, 2, BNG115)-4); GB, 1. 2))-
Show that S is not a spanning set in
IR3.
25. Write 1° + ¢+ 1 as a linear combination
of th € elements of the set § : (32, 22-1, 22 + 24 + 2}. Show th
5 is the spanning set for all polynomials of
degree 2 and can be taken as its basis.
26. Let V be the set of all vectors
in IR4 and S be a subset of V consisting of all
vectors of the form
(i) @, y,-—y, — x),
(il) (, y, z, w) such that x+y +z-—-w=0,
(iii) (x, 0, z, w),
Gv) (@, x, x. x).
Find the dimension and the basis of 5.

Fotonbatwalues 9 4k do the following set of vectors form a basis in IR??

@ (& 1-4, ©, O, 3k~1, 2)
amar,
y ’

D
P se
(—£,.1, 0)},
ne) ‘ail
 Matrices and Eigenvalue Problems 3.45
(ii) {(K, 1, 1), 0, 1, 1), (&, 0, &),
(iii) ((k, kB, (0, k, kb), (k, 0, B),
(iv) {(1, &, 5), (1, —3, 2), (2, -1. 1)).
28. Find the dimension and the basis for th
(ii) symmetric matrices, (iii) € vector space V, when V is the set of all 2 x 2 (i) real matrices,
skew-symmetric matrices, (iv) skew-Hermitian matrices, (v) real matrices
A= (a) with ay, + a2, = 0,
(vi) real matrices A= (aj) with a,; + dj2= 0.

29. Find the dimension and the basis for the vector space V, when V is the set of all 3 x 3 (i) diagonal
matrices, (11) upper triangular matrices, (iii) lower triangular matrices.
30. Find the dimension of the vector space V, when V is the set of all n x n (i) real matrices, (ii) diagonal
matrices, (111) symmetric matrices, (iv) skew-symmetric matrices.
Examine whether the transformation T given in problems 31 to 35 is linear or not. If not linear, state why?
| x
a 7: IR? > R' : r(*)=xX+y+a,a#QO, a real constant.

oll X+Z
ra

2
33, T:IR' > R’*; T(x)= . }
x

0 x#0,y#0
c, a ey
34, T IR? sR'i7{*) = i2y, x=0
13x, y=0.
oN

— R';T|
35. T:IR? +x +z.
y| =xy
Z
36 to 42.
Find ker(T) and ran(T) and their dimensions in problems
fs 2x+y
37. r+ RT| (=| y— = |.
3x4 4y

Ty’.
39. rm? RT|") = x+y.

7o ie Jf x-y’)
x-y
a re 3. i

41. re?
, _
=
: [x
a |
:
= Te
“3 — Si juego I (a A By
pea: 71) vy) =x + y- AY
eae ly
; a a mei ES | a

| Nee ae
Me
 4.46 Engineering Mathematics

43. Let 7: IR’ — IR? be a linear transformation difined by

ar .
x+y)
Tr) y| = ,
es

|
Find the matrix representation 26
wi respect
of TT with to the ordered basis
he ordered

ih iy oy 1) (0).
(1) A ‘ 2+ Q | | in
i IR-4 “and Y —_ 0} : £ |] i

Was

a) 1) (2)
(11) X=;/0)|, | 11], | 1)}in IR° and Y= 3) 5 in
1} lo} U3 :
44. Let V and W be two vector spaces in IR’. Let 7; V-— W be a linear transformation difined by

x | ()

NL S| = yy

Wed X+y+z

Find the matrix representation of T with respect to the

ordered basis

Pay 160) 1) (0 oy
(i) A= 9/0), )1),|1]¢inv and y=4/0]/,/ 1] 10 in W
Wy Oy a 0} \o |

1) (0) (0 Le (Le 60
(ii) N= 5/0) /1),/O}pinv and y= dio} Jal |, in W.
0) \1 Ll) lo} (4
45. Let Vand W be two vector spaces in
IR3. LetT: V-+Whea
linear transformation def
ined by
x eiIF

Hy | = X+y

X X+y+z
Find the matrix represent
ation of 7 with respect
to the ordered basis
1 \
eT —|]|
X=
l/+inV and |a |eee Li.) -l}} inw
=I Seen
B
4
 Matrices and Eigenvalue Problems 3.47
7. IR° ne? >—s IR” 1R® he
46. Let T: a 1;
be a linear transformation defined by

- + J
; TZ
/ y — y

A+ Zz
ae |
x+y+z
Find the matrix representation of T with respect to the ordered basis

ay" Q | l |

: ci Pe 0 l ]
X= 4|0},) 1],]1]}in R? and y= aol, “W et, in IR
Veal 1/}1/ to
1 l 0

47, Let T: IR* — IR® be a linear transformation.

12
Let A=|2 3 | be the matrix representation of the linear transformation T with respect to the ordered

3 40
1, =i]?in
basis vectors v, = [1, 2], vp = [3, 4]” in IR? and w, = [-1, 1, 1)’, w2=[1,-1, 1”, w3 =U,
IR? . Then, determine the linear transformation T.
|| 2 A)
: |be the matrix representation of
48. Let T: IR? — IR’ be alinear transformation. Let A = ;
vectors v; = [1, -1, 1]’, v2 = [2, 3, -1]’,
the linear transformation with respect to the ordered basis
¥,= [1 0-1)"in IR? and w,=[1, 1)’, w2=[2, 3]’ in IR* . Then, determine the linear transformation T.
eg 52
A=| 2 3 be the matrix representation of the
49, Let T: P\(t) > P(t) be a linear tansformation. Let
=I 1
in P(0).
‘linear transformation with respect to the order ed basis [1 + ¢, ] in P,(t) and [1 —1, 27,2 + 3¢—-1°]
Then. determine the linear transformation 7.
vk. 4 = - | f (x;, X2, %3) in IR° satisfying
Se (i) x, I — 3x. 2 + 2x, aes = 0;
50. Let V be the set of all vectors of the form (1, *2, %3)
Sa s % = 0 and 4x, + 5x2 = (0. Find the dimension and basis for V.
(il) 3% — 2X2 + 43 =

uations
yeral Linear System of Eq
: P| |e
| ee, ee? he

method and the Cramer’s rule for solvi ng a system of

vehavee discussed the m atrix t matrix A is non-singular, that 1s
on unkee Ax = b. We assum ed that the coefficien
a tn A is n. The matrix method requires evaluation ofn> determinants
7.
Ys
1) + oenerate the cofactor matrix, and one determinant of order n, whereas the
A

a of (n+ 1) determinants each of order n. Since the evaluation of high

, }
ae* i
(ey
7 68 CUO
1
|

: mA ge
' é of

pe le e WeTV
tae consuming, these methods are not used for large values of n, say
“ emery”
hi A = ry

1 ey! yt

b-..
 |j }
( f t
| y J Z
) | : 2
’ “4 E
1 | |
|i >
; |

given by |
Ax =bD C3;

Xx]
aj, 412 ain
X49
Qn, 8422 Can b= Syl
where A= j
x

én a the — vect,
are respectively called the coefficient matrix, right hand side column
of the matrices A, b, x are respectively m x n, m X l and| n X 1. The maatrix
The orders
jer

ai 1}? ree Ain D;

a> 4) ce ayy b> 2 ;

(A |b)= oF
am| GAm2 sce Aimnn Dm,

is called the augmented matrix and has m rows and (n + 1) columns. The augmented matrix descrit
completely the system of equations. The solution vector of the system of equations (3.30) is
n-tuple (x), x2, . . . , X,) that satisfies all the equations. There are three possibilities:
(1) the system has a unique solution,
(ii) the system has no solution,
(iii) the system has infinite number of solutions.
The system of equations is said to be consistent, if it has atleast one solution and inconsistent,
if ith
no solution. Using the concepts of ranks and vector spaces, we now obtain the necessary
and sufficie
conditions for the existence and uniqueness of the solution
of the linear system of equations.

3.4.1 Existence and Uniqueness of the Solution

Let V,, be a vector space consisting of n-tuples in IR” (or
C”). The row vectors Bey» «natty; OF t
m Xn matrix A are n-tuples which belong to V,. Let S be the subs
pace of V,, generated by the ro\
of A. Then, S is called the row-Space of the matrix A
and its dimension is the row-rank of A and
denoted by rr(A). Therefore,

row-rank of A = rr (A) = dim (S). (3.3:

Similarly, we define the column-space of A and the colu
mn-rank of A denoted by cr (A).
ace of m Xn matrix A is generated by m row vectors of A, we have dim(5) SI
x Ser S isrow-sp
since the a subspace of V,, we have dim(S) S n. Therefore, we have
. rr (A) < min (m, n) and similarly cr (A) < min (m, n). (3.3:
SIPRAA, Let 4 = (4) be an m Xn matrix. Then the row-rank and the column-rank of A are sa
ig ah wi =a, Mee of A. The dimension of 5 is the number of linearly independent fo
AE OAS OF S be r. Therefore, r rows of the matrix A are linearly independent and tn
=
——
 Matrices and Eigenvalue Problems 3.49
remaining * — 7 TOWS can be written
the linearly independent rows of re
os a linear combination of these r rows. Let Ry, Ro, so, R, be
as " “ :

en, We can write

R 4) — rR, T M7412 R + oe ie e = 0 re ee ae
R Na
Crip F' Gago Rye...
+ 742,Ry

R,, = Qin Ry + On 2 Ro . ee Goa ak,

where @;,; are scalars,

AI:
he Jjth element of the: row R,4) iS
The refore, , the given by
qd —
ney Or ays Or+1,2 2; aoe Sare al 8 i ae a yj
Similarly, the jth elements of the rows R_,,Beals R.. can be written
seen
Hence, the jth column of the matrix
A can be written as

| 0 0
0 l 0
Q 0 0

0 os aj 0 + _ a,j 1

O41] : Or 41,2 Orel or

Om] Om ,2 Om r

Thus, every column of the matrix A can be written as a linear combination of r linearly independent
rows of A. Therefore, the dimension of the column-space cannot exceed r, which is the maximum
number of linearly independent rows of A, that is
cr (A) < r= row-rank of A.
Simi arly, by reversing the roles of rows and columns in the above discussion, we obtain

rr (A) < r= column-rank of A.

Combining the above results, we have

rank (A) = rr (A) = cr CA or.
"hg
of linear algebra.
Now,| we prove the important res ult which is known as the fundamental theorem
tem of equations Ax = b, where A is an m Xn matrix, has
*heorem 3.7 The non-homogeneous sys (A |b) have the same rank.
solutionifand only if the matrix A and the augmented matrix
s Ax = b as
‘ite the g iven system of equation
.+x,C,=b (3.34)
x, C, +%2C2a+--

Ax = b is equivalent to finding
ihete C. is the ith column of A. Thus, finding solution of the system
AS x), x, .. ., x, which satisfy the equation (3.54).
 3.50 Engineering Mathematics

ank of the matrix (A |b) is 7 and r Sn. Therefore

Let rank (A) =rank (A |b)=+r. Then, the column-r Then, the
ose thesethes are the first r columns,.
there are r linearly independent column vectors. Supp ination ; linearly
remaining columns C,41, Cp42. «+ C,,4) can be written
as | a lines= i sell combcieisapiaia t of these
of (A |b) is a linear combination of its firs,
independent column vectors. Thus, the (1 + {)th column
n columns, that is
OQ, C; + OQ Cy + er (X,, ¢ n
D=

which means that Aw = b, or Ax = b has a solution.

Conversely, let Ax = b have a solution, say x = a. Then, we can write
b=a,C,) + Co+...+ &%C,
sion, Since, these
Thus, the column-spaces of A and (A |b) are the same and have the same dimen
(A) = rank (A | b).
dimensions are rank (A) and rank (A |b) respectively, we obtain rank
Remark 11
A system oflinear equations Ax = b is consistent, if the vector b can be written as a linear combination
that is, b
of the columns C,, C;,...,C, of A. If b is not a linear combination of the columns of A,
is linearly independent of the columns C;, C2,..., C,, then no scalars can be determined that satisf\
Eq. (3.34) and the system is inconsistent in this case.
In section 3.2.3, we defined the rank of an m Xx n matrix A in terms of the determinants of the
submatrices ofA.An m Xn matrix has rank r if it has at least one square submatrix of order r which is
non-singular and all square submatrices of order greater than r are sin gular. This approach is
very time
consuming when n is large. Now, we discuss an alternative procedure to obtain
the rank of a matrix.

3.4.2 Elementary Row and Column Operations

The following three operations on a matrix A are calle
d the elementary rew operations:
(i) Interchange of any two rows (written as R;
~ R ios
(ii) Multiplication/division of any row
by a non-zero scalar (written as @R,)
(iii) Adding/subtracting a scalar multiple of any row to another row (written as R, <—
R. + aR AK)
that is & multiples of the elements of the jth row are added to the cortesn SNj — Kj +
the ith row.Ae The elements of the jt
j rowhremai onding elements 0!
ay tye ¢
the elements of the ih
These ope| rations cha: nge the form
orm of A but do
row-space of A. A matrix B is said to be row eqnot change tthe rc
A
|
as they do not change the
from the matrix A by a finite sequence of deineia toa matrix A, if the matrix B can be obtained
We observe that | » ATY TOW Operations. Then, we usually write B= 4

The above operatio quivalent to C, then A is row equivalen

ns performed on t to ©
columns (th at is column j eat
column in place of row) are called e/ementa’

ha?
 oe PO SS oF. PA)

La! Matrices. aitd Eigenvalue Problems 3.51

) ae
,4,3 Echelon Form of a Matrix
An m n matrix is called 4 ro . |
7 Ow echelon matrix or i row echelon form if the number of zeros
receeding the1e first non-ze“AOro entry | y of a rowW Increase
pre
s row by" row until| a row havi. ng all zero entr }
for noOo er eliminst ation
rollowing are satisfied.lane ys possible) i s ohiniwy
€) is obtained. Therefore Ore, a matrixeeis in row echelo,
ies
“fab the
n form if

(i) If the ith row contains al] Zeros,

it 1S true
vg Ov niachank
Tor all subsequent rows.
(ii) Ifa column contains a non-zero
ent ry of any row, then every subsequent entry
is zero, that is, if the in this column
ith and (i + l)th rows are both non-zero
entry of the (i + 1)th row appears rows, then the initial non-zero
In a later column than that of the ith row.
(iii) Rows containing all zeros occur
only after all non-zero rows.
For example, the following matrices
are in row echelon form.

Me 7 ull =f, 25 ‘ee

O § 1,;0 OO 3 : : : ;|
OOF 0 OI)0 oo @ 6 ’
ao Oo Om Oo
Let A = (aj) be a given m Xn matrix. Assume that a,,; #0. If a;, = 0, we interchange
the first row with
some other row to make the element in the (1, 1) position as non-zero, Using elementary row
operations, we reduce the matrix A to its row echelon form (elements of first column below a);
are
made zero, then elements in the second column below a>, are made zero and so on).
Similarly, we define the column echelon form of a matrix.
Rank of A |The number oon zero
Jowsin the row echelon form of a matrix A gives the rank of
the matrix A (that is, the dimension of the row-space of the matrix A) and the set of the non-zero rows
in the row echelon form gives the basis of the row-space.
Similar results hold for column echelon matrices.
Remark 12
(i) If A isa square matrix, then the row-echelon form is an upper triangular matrix and the
” column echelon form is a lower triangular matrix.
(li) Thisapproach can also be used to examine whether a given set of vectors are linearly independent
| OF -not.
Side in) form the
Weechelon form. giveneach
The with
matrix vector as its row (or column) and reduce it to the row
vectors are linearly independent, if the row echelon form

‘has no Tow with all its elements as zeros. The number of non-zero rows is the pai of
i ato f vectors and the set of vectors consisting of the non-zero rows is the basis.
» 3.32 Reduce the following matrices to row echelon form and find their ranks.
é 3.32 Reduce the following ™
| pe 3. A
a S| Wigs]. 405
ay (ii) ie Seln Be 7")
4 ide
3.52 Engineering Mathematics
We have
Solution Let the given matrix be denoted by A.

LP 3s pile ee a ad
: +2R, 2/0.a -7 -6|Rat2ks
larger ~|9; ae; 8.
(i)| A=| le 2 py ke om Ry
—2 8
|
the number the row echelon n fort
of non-zero rows in1n the fo
This is the row echelon form of A. Since
is 2, we get rank (A) = 2.
1 2 3 4
Ry-“Ri {5 3 -2 -3|Ra+Ro
ee ait. Se. 3 ee or are. 3|R4—5R2
i dae wl OC lo ts 10=C- 15

ee ee:
Jesu 2
ae. eT
6 6 6
Since the number of non-zero rows in the row echelon form of A is 2, we get rank (A) =2

Example 3.33 Reduce the following matrices to column echelon form and find their ranks.

reales ers 24 3h ma Sie 2

(1) aie iti> (ii) .

of te 1S tf bpm ah vay

ie 2 ie
4/C ,-Ce
,/3
e l a/3 3/3 es,
ee l 5 c, 0
ee Ws. 73) 3 C=) -7/3 0}
SERBS) 2 1/3
2 1/3
Since the column echelon for fs. 0
m of A has two non-zero col
umns, rank (A) =2
It Le = I
(ii) seLie lie 9 aa
-_ C c —] a
204e 9) 6,
Co e1 = £30
ote §4 |Ca C, : = 2 1} C,+C,
— 2 4 2
 Matrices and Eigenvalue Problems 3-53

| ~ ». Gg

me -2 O. O|
4 L =] O60 @6!
ae eZ iQ’ 0
Since the column echelon form of A has 2 non-zero columns, rank (A) = 2.

gxample 3.34 Examine whether the following set of vectors is linearly independent. Find the
dimension and the basis of the given set of vectors.
(i) (i, a, >) 4), (2, QO, F ei 2), (3, ‘La 4, Zz),

fia tis ts, 1), (1, 1, 1, 1), 1, I, 1, 1), 4,0, 0, 1),

(iti) (2, 3, 6, — 3, 4), (4, 2, 12, — 3, 6), (4, 10, 12, — 9, 10).
Solution Let each given vector represent a row of a matrix A. We reduce A to row echelon form.
If all the rows of the row echelon form have some non-zero elements, then the given set of vectors
are kinearly independent.

+2, 3 2 3
(i) Awia 0. 1 SS
a Pe

ee 4
=|0 -4 -5 -10}.
pores OO 0

eeallthe rows in the row echelon form of A are not non-zero, the given set of vectors are
e pe| apa Since the number of non-zero rows is 2, the dimension of thehie set of

pi 0,1, 0) (00,0,1)ote given staf fos

thbans
e
ons
 3.54 Engineering Mathematos
3
: Z 0 3 —Z R,+R,
FE -F 4 R,-2"1 _|9 -4
on
(ii) A=(4 2 12 = -2 6: “JR, = 2mOh : |g 4 p a 3 4
{ 10 1-9 ™
2 g 0G =p 4
il OeniiAewiOin me Saree
0 (mds @ 0
n-Zero,
ro, the
the given set of vecto,.
gi ~LOTS.. are
Since all the rows in | the row ecshelon form of A arerows
np is
He2, the dime
jimension Of: the SIVEN
given set
<a, .
:
linear Sj
ly dependent. Since the num ber of non-z ero | 4 0.3. —)
he set ((2y 3. © — 3, 4), (0, — 4, U, 3, — 2)}.
vectors is 2 and its basis can be taken as {he Set \\%

3.4.4 Gauss Elimination Method for Non-homogeneous Sy stems

;
Consider a non-homogeneous system7 of m: equations in 1nmunknowns
In

Ax=b (3.35)
ayy ay2 Gin by PED
Ay, 422 «++ Ga» b> | x2
where A=] . =e. ex |.

Sm Gad, ers Bane b,, x

We assume that at least one element of b is not
zero. We write the augmented matrix of
mxX(n+ 1) as order

a)| a1 <i al Bs
“2; a

ay Q)5

0 a4

where r < m and 411*

0,4,,%0
J.

We have the Vt) a

 Matrices and Eigenvalue Problems 3.5:
(a) Let r<m and one or more of the
| elements b r+is9r.2,. * * *

.., 6. are not zero. Then rank (A)

|

rank (A |b) and the system of equations ha

S no solution. ‘i
(b) Let m 2 n and r= n (the number of columns in A) and b7,,,b7,.,...,b
this case, rank (A) = rank (A |b) = p and the * are all zeros. In
solve the nth equat system of equations has a unique solution. We
ions for Xn, the (mn —
1)th equation for x,_; and so on. This procedure is
called the back substitution method
For example, if we have 10
| equations in § variables, then the augmented matrix
10 x 6. When rank (A) = rank (A |b) =5
is of order
1
he system has a unique solution.
(c) Letr<nand b *
r+l> b>. 5 ee b> are all zeros. In this case, unknowns, Eis Hey 00 ing Ky CON
be determined in terms of the remaining (7 — r) unknowns
Xrel> Xpy2s ++ +X, by solving the rth
equation for x,, (r — 1)th equation for x,; and so on.
In this case, we obtain an (a — r)
parameter family of solutions, that is infinitely
many solutions.
Remark 13

(a) We do not, normally use column elementary operations in solving the

linear system of equations.
When we interchange two columns, the order of the unknowns in the given
system of equations
is also changed. Keeping track of the order of unknowns is quite difficult.
(b) Gauss elimination method may be written as
(A |b) Elementary (B |c),
row operations

The matrix B is the row echelon form of the matrix A and ¢ is the new ri ght hand side column
vector. We obtain the solution vector (if it exists) using the back substitution method.

(c) If A is a square matrix of order n, then B is an upper triangular matrix of order n.

(d) Gauss elimination method can be used to solve p systems of the form Ax = b,, Ax=bs,.. .,
Ax=b ‘which have the same coefficient matrix but different right hand side column vectors.
‘We form the augmented matrix as (A |b;, bo, = sip bp), which has m rows and (n + p) columns.
Using theelementary row operations, we obtain the a ofwpe on CB} eyieac& ., e.);
yhere B is the row echelon form of A. Now, we solve the systems Bx = ¢,, Bx = ¢, . . a §

‘a

La

——
nentcu
= 7
si

oo
,

matrix and reduce it to row echelon form by applying elemen tary

a ,. ——_—
a

®
6 a
-2
 3.56 Engineering Mathematics

F3
jy 4
fl A Ry - Ry/2 _ ah. 5/2 _4|R3+
‘ 3 5R,/3
2
@) (A|b)=| 1 -1 2)-4lp ae
4 Ry/2 5/2 —3/2D,
24 SEAN IG

2 1 -l 4
210 =3/2 S/2| =a)
0 0 8/3} -8/3 |
‘ain the solution as
Using the back substitution method, we obtain
8 _ 2 OF g=—,
5 ee
Bec
Tig) igi 4 ee, or y=l,

x+y-z=4, or x= I.
uti
Therefore, the system of equations has the unique solution x = 1, y —= I,8 oe

: Ze 0 ] 3
a) (AID =]bttmt 1ee ees 1/2 |1/212 |r; - 2,
1R,—2R
42,
BB eee ees T| kg
2 0 ] 3
= 10 24 1/2 lage.
Over) 2G Ont)

be vie that rank (A) = 2 and rank (A |b) = 3. Therefore, the system of equations has no
sOlUTION.,

1 -1 1
(iii) (Alb)=/2 1 Ry = aR a= 1| 1
>). 2 5 R3- SR, -3]0 R;- R,
+ 4) 4310
et ae
=) () 3

0 0
The system is consistent and
has infini
is satisfied for all values of te number
: of so] utions. : oh the act equation
the first equation, we get ‘s
ec, pias rend ©quation le aii ee: From
y = Zand z is arbitrary, ie he etefore we, obtain the solution * =
 Matrices and Eigenvalue Problems 3.57
mple 3.46 Solve the follow;
example
‘ a ae OhOwing
ae systems» ofOF equati
equations 7 £ Gauss
using eliminati
Gauss elimination method.
(ii) xX 2y —2z =]
2x
+ 3y+3z+6w =6
4x -2ly -39z-6w =~ 24 2x-—3y+2z = ()
5x + y= 5z = |

3x + 1l4y — -
solution We have i iettigs hal
(i) (A |b)
& -3 -9 6
4 -3 -9 6] 0
no 3 5 66 | 0 9/2 15/2 3] 6/R3+4R,
fas 2l —96 _¢
|® =18 -30 —12]—24)

=9 6/0
L522 3:) Gil.
0 0/0

The system of equations is consistent and has infinite number of solutions. Choose w as
arbitrary. From the second equation, we obtain
9
7+
15
7 2=6— 3w, or
2
y= 3 (6-30-32)
Le = 44-52-20)
|

From the first equation, we obtain

4x = 3y + 9z —-6w = 4-—5z-2w+ 9z-6w=4+ 42 - 8w
or eae — 20
Rank (A/8) Sid
Thus, we obtain a two parameter family of solutions
Rank (Af) Sa
xe=l+z—2w and y=(4—-5z-2w)/3
“Wa @a)
where z and w are arbitrary.
lbs ee eT
Poe
2 ae)
|i 0
leOR,
R 5R na 0 =i 5 —2 R3- 9R./7

a ae 110 -9 5|-4]R4+ 8R2/7

fei4 -12)5 Rash : est. lee!
, 9 1] 2 -2 1
: ime 0 -7 5) 2
—7 > A Ry —IR3= QO -10/7 |—10/7
9 10/7 |-10/7
0 0 0
all values of x, y, z- From the third equation, we obtain
st equation
is satisfied for 1. Hence, the system of equations has a unique
Wa tion s
gives ¥ = x= — 7R, + R3)/5, the last equation is redundent.
ar 7.
a at
 3.58 Engineering Mathematics
oons
©4 ati
a Equ ,ati
Homogeneous System of Linear
|
3.4.45
@ quations
Consider the homogeneous system ot
. 3

AX = () }

x = 4 ?re
sys tem 1 is always consistent since
efore, for the homoge
; eneous Ig
‘is anan ymm XX nmn matrix
matrix. The homoget\®”
where A is vail (A) anan ek (A ) er<n. It rank (Ay =. Noy
nk9).(ATh
ca| r
solution) is always a solution. In this CA oa feanile
system to have a non-trivial solution, Sa aaae n e,
so ich form a vector space of dimensio (n
we obtain an (n — r) parameter family of
as: (n ——r)
r) parameters
par arbitrarily.
can be chosen arbitrar!
1s system is called | : _ oe.
the null space and its dimension jg calle
The solution space of the homogencot
the nullity of A. Therefore, we obtain the result
| =” , m 5) ) a
(see Theore
rank (A) + nullity (A)
Remark 14 Be ie
:
(a) If x, and x, are two solutions of alinear homo geneous1S: system, then i 1 4+ ee PXae
18 also
atsa solu,
tuo
of the homogeneous system for any scalars @, B. This result does not hold for non-homogenegy;
systems.
(b) A homogeneous system of m equations in n unknowns and m < n, always possesses
a non.
trivial solution.
Theorem 3.8 If a non-homogeneous system of linear equat
ions Ax = b has solutions, then all these
solutions are of the form x = xp + x; where Xo is any
fixed solution of Ax = b and X;, 1S any solution
of the corresponding homogeneous system.
Proof Let x be any solution and Xo be any
fixed solution of Ax = b. Therefore
, we have
Ax = b and AXo = pb.
Subtracting, we get

Ax-Ax)=0, or A(x-—x,) = 0.
Thus, the difference x
— Xo between any solu
tion x of Ax

Remark 15
If
u the non-homogeneous syst
ysterem Ax =h
=D Where
1s, rank (A) = n, then the corre IS anim x y ma
3
sponding homogeneo trix (mn) has a unique solution, (4!
1S X, = 0.
US syst
Systeem
rn Ax =
0 hac Only the trivial solution |
Example 3.37 , (4!
Solve the follow;
owin g hemogeneay
‘ System of e
: | 1 quations Ax = 0, where A is give" DY

3 2| (ii) | l 1 | ] —] |
Find
nd the rank(A)(A) andand nullity
the rank nat (A), a, | 2 —..
63 f 4i.
,

Beata We write the augmented

matr~~
iy (A“ST |0)0) ang reduce
It to TOW e
 Matrices and Eigenvalue Problems 3.59

2 0}, 4 ik ; gs
(i) (A]%=/ 1 -1/0 a= Ri /2 4
7|9 3/2) 01Ry4+- 2/3 = 1/0 3/2) 011,
i) |.
10) (Ry=—3R
f@79
ot 0 1/2] 0 0 0|0
Since, rank (A) = 2? = nhumbe
nullity (A) = 0, r of unknowns, the system
syste has only a trivial
ivi so lution. Hence 3

l 2 -310 + 3} 0
i)
(it) (A |
(AL%=]1
| |
1 5a
-1]0|*2>*
0 R,-R, - 0
healt
—| 2 0) R,-—3R3 = 0)
do ai
—|

Since rank (A) = 3 = number of

unknowns, the homogeneous system has only a trivial
solution, Therefore, nullity (A) = 0

tl
Ke On =)
Mf taste” eticg 116
Pet =P To
(iii) (A]0)=]2 3 AHOT sap, 8/0 Ted 21Q|RapBon|0, 293 2) 01
mane / 27 =f

3 IN,
ae 36° 216 OP S13). 2810 00 0 0]0
Therefore, rank (A)= 2 and the number of unknowns is 4. Hence, we obtain a two parameter
family of solutions as x, = - uae ~> Ae Xy = — 2X2 + X3 — Xy = 4x3 + xy, where x3 and x, are
arbitrary, Therefore, nullity (A)=

34.6 Gauss-Jordan Method to Find the Inverse of a Matrix

LetA be a non-singular matrix of order n. Therefore, its inverse B = A“! exists and AB =I. Let the
matrix B be written as B = [b,, b>, . . . , b,,], where b; is the ith column of B.
From AB = I, we obtain
AB = Alb), bo, ... -»» Dal = 1= Uh, ta. « . . Il. (3.37)
Where
I, isthe column vector with | in the ith position and zeros elsewhere. Using Gauss elimination
— = solving n systems with the same coefficient matrix (see Remark 13(d)), we form the

AVE, foes. <> bo which is same as (A |J,

Where I is the identity matrix of order n. Using elementary row operations, we obtain
Nere Tis theidentity m
i (A |D
Elementary (I |B)
(3.38)
row operations |
we B>
3 = A-! This method is called the Gauss-Jordan method. In the first step, all the elements
ome:the
pivo
pi Reet made zero. In the second step, all the elements above and below the
second
iéa. a AL te ith step,all the elements above
and below meoO aj
Ak are made

: icky snes-Jordan wr find the inverse of the matrix

cai aN

aye St #4
 3.60 Engineering Mathematics

Solution We have
—| ie) i 0 0
(AI=| 3 -1 19 1 4
The pivot element a,, is — 1. We make it 1 by multiplying the first row by — 1. Therefore,

i 1 =2/=) 0 0 R—3R)
5 | al 2h | Ms
oO
As Ste a ORIN O27) 3 1 OC RG/2
R3 +R, 1 0
a 3 a OO Q) 2s @RE ed 7

[== |= =O 46 2 i o 3/21 1/2 1/2, 0

= 1 232 12 alee = eliOe hate s200l2 0 (—R;)/5
a ai R;-—2R, —
2 BDila=is oO | ji 0 =—s1—4.—], 1

i 0 3/2)! 12° 1/2

| 0 R-3R,/2 | ior = 710 Pe.
2710 3/10
=(0° Lf F/213/2 12 0), ee me i Ol—13/10 —2/10 1110 |
ome ls sasils 1/5 e * 00 1 4/5 iin 1/5
Hence,

=a | es
A= 7, |-13 —2 Tike
Ds 2h 9

Exercise 3.3
Using the elementary row operations, determine the
ranks of the following matrices.
faa ip meals (=o 3
bas a
3 |. ee es eae "= L |,
i Ss i 3 fr =9
I =] eres Ne Me ,
4.12 3 Ades Pie “tae oe 2
= eee ge i" :
ee eee
=I)
S13 14
ATP) Ley si] MH a) 0
‘ rea:s ") a Stpt
| igg1
ae a ——
a | a. *
6)
ike es a ieee
3 x? ...6
Oy as
| i i
 Matrices and Eigenvalue Problems 3.61
mentary column
psing the elementary operat i
mn Operations, determine the ranks of the following matrices.
1 [=] |
6 1 ” ly 2 i te 3 38
1 ow aae -1 1]. 13: |10 FP Be Sil
3 4 53 4 +5
; LS3 i oN
14. F a 15. nee sh |
Ee a1 63.20) Spi
4 6-5 4 3 <1 5
Determine whether the following set of vectors js linearly independent. Find also its dimension.
16. {(3, 2, 4), (1, 0, 2), (1, - 1, - 1)}. Df 2G — te 1, ON).
18. {(2, 1, 0), (1,— 1, 1), (4, 1, 2), (2, -3,3)}. 19. (2, 2, 109@) £1) (le
=a, 2)
20.) {G, 1.1); 1,5); (1 + i, — 1-7, dD}. 21a LG ale 15 lode ye Oe Pk), Ue Ooh)
fom (ln 2.3, 1), 2, 1, —1, 1), G, 5, 5, 3), (4, 1,3)}.
gamle 2) 34), ©, 1, = 1, 2),.(1, 4, 1,8), G7, 8, 14)}.
pee 0) 1), 1, 1, 1), (4: 4; 1,1), G1, 0, 0, 19).
5 12. 2,0, 2), (4, 1, 4, 1), G, 0; 4, 0)}.
Determine which of the following systems are consistent and find all the solutions for the consistent systems.

hr~

bt
—

1
33. |1
1

35.
 3.62 Engineering Mathematics

Find all the solutions of the following homogeneous systems Ax = 0, where A is given as the fol
ap z oe tha following

lowing

3 2 | | 2 3 fi 5
46. | =2 Bi ay. |) 3 Pe il 38. |4 i =ag |
| 5S 4 ee ee ee 4 9 -6|

2 -!I —3 I ] l l a
jl 23 4 l | l ye = | ns
4% I FF 4 M01 teagan fee ee | |

oe ee 24 5 9 l I 4 1

on) ier: 1 lL = aT [ =< 4]

I Sea = |3 ] —9
“)
3
a

42. oe 8 43. 44. ; ie. 2 |

wee 17 = 83 | anon 57 =i | f
AAO CA) 3 See ieee 7 oe4 Sy e9 2 -1]

ee 9 hed
Pe 25a Loe?
witeage = =
4 3 2 =4

mnlet, ie Sahl '

46, 0 2 Bil AT 1 1 383 48. ; . |
i 3 4 e ie 2: |
: l l 4
1 1 | ] bE. 196 4
=i | l v1 r sie
49, . 50
=] l l Bes AI
l J *22] 1 i 310) 0 “a

3.5 Eigenvalue Problems

Let A = (a,) be a square matrix of order n. T he
matrix A may be singular or non-singular. Conside!
the homogeneou s system of equations

AX=Ax or (A-AI)x=0 (3.39)

where A is :scalar and I isan identity
matrix of order n. The homogeneous
always has a trivial solution. We need system of equations (3 39)
to lind values of A for which the homogeneous system (3.39)
8 of A, for which non-trivial solutions of the homogeneous sys'*"

he eigenvectors or the characteristic 5 : fea } es

vectors of A. If x is 4 ”
‘a 7" le

trivial solution of the homogeneous system (3.39),

: 1 5 ff a"

then ax, where is any constant is also a oo

of the homogeneous system. Henc
©, an eigenvector is unique only upto a constant
F | eI t ; 4 a

multiple. !”
2 % d ( C
 Matrices and Eigenvalue Problems 3.63
m of determining the elgenvalyec . |
roblem OF ra 2 i S€nvalues and the corresponding eigen tnt Sh;
ies meIBerVGIhe Prootem. genvectors of a square matrix A is
cal
45.1 Kigenvalues and Eigenyectors
,omogeneous system (3.39) h- >
solu
hos “as than a le ntaly “jon-trvial4.SOlution, then the rank of the coefficient matrix
Ais eavant
(r » we Coefficient matrix must be singular Therefore, #3
ai,-A a)\> een. Ain

det(A =) =| °2 Ao,—-—A oes Gy 0 eae

, = Vs 3.40)

Gy) Gn? in Gee A

Expanding the determinant given in Eq. (3.40), we obtain a polynomial of degree n in A, which is of
the form

Prd) =| A-AT| = 1)"[A" —c, A + cy A. + 1)"e,] = 0

fee cy tA. + (= 1)". = 0. (3.41)
where cj, C7,..., C, Can be expressed in terms of the elements a; of the matrix A. This equation is called
the characteristic equation of the matrix A. The polynomial equation P,,(A) = 0 has n roots which can
be real or complex, simple or repeated. The roots Aj, Ay, ..., A, of the polynomial equation P,,(A) = 0
are called the eigenvalues. By using the relation between the roots and the coeificients, we can write
Ki + A» +...¢A, =C; =a $49 TVwee
T Oye

A, A> a A, A3 + oe +A, An =C)

A) Ar vee Ap = Cy.

If we set A = 0 in Eq. (3.40), we get

A | = (— Bie n — Cy = A; Ay tear pe Rin: (3.42)

pa sum of eigenvalues = trace (A) and product of eigenvalues = | A |.

is called
of eigenvalues
__ The setee is calle d the spectrum of A and the largest eigenvalue inmagnitude
& and is denoted by p (A). If | A |= 0, that is the matrix is singular, then from
ulradius of A an ; be eigenvalues must be zero. Conversely, if one of the eigenvalues is

Da Aes ea >a sigdiagonal or an upper triangular or a lower triangular matrix,

then Bete 0.Nowe of the matrix A are the eigenvalues of A.

a A;’s, we solve the homogeneous system (A — A ;I)x = 0 for each

After determining the eigenvalues
sponding eigenvectors.
“fs A a. = .

sf eigenvalues and eigenvectors

es i. ire correspondin g eigenvector. Then we have the following results.
nvalue of A and x be its corresp y.!
a : | carresponding eigenvector Is X.
—. Ax = AX = GAX = (AA)x.
al
 ~Os Lngineering Mathematics

hs. A™ has eigenvalue A” and the corresponding eigenvector is x for any positive integer »,
Premultiplying both sides of Ax = Ax by A, we get
AAx = AAx = AAx=A(Ax) or A’x=A*x.
Therefore, A? has the eigenvalue A* and the corresponding eigenvector is x. Premultiplyin,
successively m times, we obtain the result. 7
3. A —&lI has the eigenvalue A — k, for any scalar k and the corresponding eigenvector js x.

Ax=Ax=> Ax-—kIx=Ax-kx
or (A —kI)x = (A-k)x.
4. A™ (if it exists) has the eigenvalue 1/A and the corresponding eigenvector iS X.
Premultiplying both sides of Ax = Ax by A7|, we get

AtAx=AA7"x or A x=(I/A)x.
5. (A — kD) has the eigenvalue 1/(A — &) and the corresponding eigenvector is X for ar
scalar k.
6. A and A’ have the same eigenvalues, since a determinant can be expanded by rows |
columns.
7. Fora real matrix A, if @+ iB is an eigenvalue, then its conjugate a — if is also an eigenval
(since the characteristic equation has real coefficients). When the matrix A is complex, th
property does not hold.

We now present an important result which gives the relationship of a matrix A and its characteris
equation.

Theorem 3.9 (Cayley-Hamilton theorem) Every square matrix A satisfies its own characteris’
equation, that is ee 7 :
A"—c,A"™'+...4+(C Dc,,A + (-1)"c,1 = 0. (3.4
Proof The cofactors of the elements of the determinant |A — AI |are polynomials in A of degr
(n — 1) or less. Therefore, the elements of the adjoint matrix (transpose of the cofactor matrix) are al
polynomials in A of degree (m — 1) or less. Hence, we can express the adjoint matrix as a polynom
in A whose coefficients B,, B2, . . . , B, are square matrices of order 7 having elements as functic
of the elements of the matrix A. Thus, we can write
- De l | 7%
adj (A — AT) = B, A" + BA"? +... + B,_,
A + B,,.
We also have
(A — AT) adj (A-AD =| A-AIII.
Therefore, we can write for any A
A Se ea 5. n—-2 | ; : | ro ao ‘ np
(A-AINB,AN + BAP +... +B A+ B= Al — cA 1s 4K etc,Al + (- 1)
iparing the coefficients of various powers of A, we obtain

i: B, =I

oon AB, -B;= oI

 Matrices and Eigenvalue Problems 3.65

| RG | B, (— 1) Pc ra ot

AB, — (— l)Ke, [.

premultiplying these equations by A”, Ami

An Ant oy AST respectively and adding, we get
a | + ‘ + (— 1)ni Pe Ae (— lYc. lt=e

which proves the theorem,

Remark 16a We have for any
non-zero vector x

Ix = |x
Ax
= Ax
A’x
= J?x

Atk = Ax,
Multiplying these equations by (— 1)"c,,, (—1)""'c,_4,...,(- 1)c,, 1 respectively and adding, we get
teh X th (1) Ax +. oe (— Lic, APs + APx
= (— 1)"c,x +(- 1)"'e,,Ax+...+(- le, At x + A"x
or [(A”—c, A" +...4+( 1) ¢,,A +C1)"c, 1x
= [A"-c,A"' +... + 1)"6,.44 + (C D"c,] x = Ox = 0.
incex + 0, there is a possibility that
Amc AM 4. FC Do APE De IEO
21ark 16b
A.
Bre can use Eq. (3.43) to find A” (if it exists) in terms of the powers of the matrix
get
Prremul tiplying both sides in Eq. (3.43) by Aq! we

tan AAT +. Den ATA + © Dep A= AO =0

*a aati - Ee. pete rane os OE 1 ogKT) (3.44)
er powers of A as
seEq. (3.43) to obtain A” in terms of low
Anac AT! -c,A™? +..-+C Deal. (3.45)
for the matrix
fy Cayley-Hamilton theorem
——

s
3.66 Engineering Mathematic

of A, (ii i) fi nd th e sp ec tral radiu sof A.

of those
of A is given by
Solution The characteristic equation
1-A 2 ()
2 Aya - 4) -4) - 21-0 - 4) - 2)
\A-al[s|-1 1-4 si
i 2-1-3 , we
+ 3A ~-A+3=0
= (1 - AYA? - 24 - 3) - 2A - 3) = - A"
ri 2 Of t 2 4
Ael-t 1 2-1 1.2/=} 4
Now
feo SL a ae 7 f

al 4 wiry 2 io «12
A=A7A=| 0 3 4]/-1 1 11 10
0 6 S51 2 16 17
We have /

| | leat wo

-l 16
 Matrices and Eigenvalue Problems 3,6/

A*- 347 +A-3=0 or (A=3)(A?+1)=0 or A=3,i,-7.

The characteristic equation of A’ is given by
[-l-A 4 4
0 3-A 4 |=(C1-A)[G-A)S-A)
- 24] =0
eee

| 0 hb. Sed
or (A+ 1)(A>-8A-9)=0 or (A+ 1)(A-9)(A+
i) =0.
The eigenvalues of A? are 9, — 1, — 1 which are the squares of the eigenvalues of A.

(iii) The spectral radius of A is given by

P(A) = largest eigenvalue in magnitude = max [Aral 3:

in Os 3G
1 O 1/|, then show that A” =A”? + A?2-I for n > 3. Hence, find A’.
Oo ft 6
Solution The characteristic equation of A is given by
fae Oe
PAH AZES| Yk 81 SO aaa Do wet = 22 44 1 0.
a) ee ee
Using Cayley-Hamilton theorem, we get
A?—A?-A+I=0, or A?-A?=A-I.
Premultiplying both sides successively by A, we obtain
A>-A*=A-]I
A*—A>=A*_-A
A! i Me — A™3_ ar

A"— A™! = A™? — Ard

Adding these equations, we get
A"—A*=A"*—I, or A"=A™?+4A2_] n>3.
Using this equation recursively, we obtain
A" = (A"~* + A*—1) + A*?-T=A™4 49/42 _1)
= (A"™* + A*-1) + 2(A?-D =A™ 4 3(A2—1)

id
— An—(n-2)
+9
44 (m—
= 2A? 2 1) = Mt2 a2_11 (yay,
3.68 Engineering Mathematics

ro oO i’ Oo 0 Q
~
247
A 95A2 =25/1 1 O|-24/0 1 O/}F= i Ol.
yay 2 aL! A. I po.) 3
Example 3.41 Find the eigenvalues and the corresponding eigenvectors of the following matrice,
8-9
(i) A= H (i), Ay= | (iii) A= foe FL
i. sl 2 OQ 3
Solution
(i) The characteristic equation ofA is given by

|A -AI|= ey , =-0 or A*-3A-10=0, or A=-2,5.

3 8) a

Corresponding to the eigenvalue A = — 2, we have

| 3 4 | O | o A |
(A+ 2x =(2 ile! -[5| OT 3x, +4x.=0 OF X,;=- aot:

Hence, the eigenvector x is given by

ea fe | a
~~ — = X97 | b
| X94 Ad |

Since an eigenvector is unique upto a constant multiple,

we can take the eigenvector as
[- 4, 3.

Corresponding to the eigenvalue A = 5. we have

—4
(A — 5]I)x = ¢
4
allel
Xx] 0
= ‘ OF A — 42 =0, Or x =x.
Therefore, the eigenvector is given by x =
(x), x2)? = x,(1, 1) or simply (1, 1)7.
(ii) The characteristic equation of A is
given by

| A -—AI| =
=a
a | ae A or A°-2A+2=0, or Aa14i.
Corresponding to the eigenvalue A
= 1 + i, we have

[A - (1 UE i)I)x = f ‘ia = is
| —] —{ | X4 0)

or — 1x, +x=0 and =a — tae = 0!

_ Both the equations reduce to — xy
eigenvector is x = [— i, 1]"
Corresponding to the eigenvalueA= | — i, we have
 Matrices and Eigenvalue Problems 3,69

LA - (1 -i)I)x = : 5 ie|= i
—l ii| x2 0
or IX) +xX,=0 and -—x,+ix, =0.
Both the equations reduce to — x, + ix2 = 0. Choosing x, = 1, we get x, = i. Therefore, the
eigenvector is x = [i, 1]’.
Remark 17
For a real matrix A, the eigenvalues and the corresponding eigenvectors can be complex.
(iii) The characteristic equation of A is given by
l-A 0 0
[A—AE|=|} 0 2-4 1 WsOicer (1 —4)G—A3:—A)'=0
| 2 0 3-A

ora = 1.2.3.

Corresponding to the eigenvalue A = 1, we have

0 O 0 XY | 0 |
i) + 43> 0
(A. — Bxe= 07 2 2 ies a0) 08
X,+X3z=0.
2 OF e2iilixs 0

We obtain two equations in three unknowns. One of the variables x,, x2, x3 can be chosen
arbitrarily. Taking x; = 1, we obtain the eigenvector as [— 1, — 1, 1]”.
Corresponding to the eigenvalue A = 2, we have

or x, = 0, x, = 0 and x, arbitrary. Taking x, = 1, we obtain the eigenvector as [0, 1, 0]’.

Corresponding to the eigenvalue A = 3, we have

— 2 (Q) QO xX 10)

Si OF Bile Six ze O| or - ‘ae

(A. - 3I)x = | - : X¥2+x3=0.
2 T° 6 | x3 0

Choosing x, = 1, we obtain the eigenvector as [0, — 1, 1]’.

xample 3.42 Find the eigenvalues and the corresponding eigenvectors of the following matrices
) bok 0 I~ Oo 9
Woe (ii) A=|O 1 O|, (iii) A=/O0 1 O|.
1 o 0 1 10. 0) 00
 3.70 Engineering Mathematics
= () - 3
ach of the above : .
problems, t characteris
w e obtainin the AEas (1 | — A) lly |
, wracteristic equation
a leet
Pa . ie envalues are A = 1, 1, 1, a repeated value. Since a 3 x 3 matrix : oe nvalues
: ere os “ Aaa whether the given matrix has 3 linearly independent eigenvectors or it ha,
it is importan |
lesser number of linearly independent eigenvectors.
tl followin
inin the ige nvecto rs.
| g eige
Corresponding to the eigenvalue J = 1, we obta
0 AF = 0,
0 0

(A=—T)x=|/0 OF 1)/x= 0} or 4x3;=0

(i)
OiesOrncD 0 x, arbitrary.

Choosing x, = 1, we obtain the solution as [1, 0, OJ’.

Hence, A has only one independent eigenvector.
0 l 0 x] 0 ae 0

QO 1] x2 )= Ol, or guek sittvacy.

(11) (A-—I)x=|0 O

0 0 0) X4 /O

Taking x, = 0, x3 = 1 and x, = 1, x3 = 0, we obtain two linearly independent solutions

X= (0/0) 1/7, x = [1, 0,0)".
In this case A has two linearly independent eigenvectors.

6 mor tolls, | 1
(iii) Qx 0) 6 0illesh= hol
Q 0 O ||x3 0

This system is satisfied for arbitrary values of all the

three variables. Hence, we obtain three
linear ly independent eigenvectors, which can be written
as
x; =[1, 0, 0], x, = [0, 1, 0)”, x, = [0, 0. br.
We now state some important results rega
rding the relationship between the ei gen
and the corresponding linearly independe values of a matrix
nt ei genvectors.
1. Eigenvectors corresponding
to distinct eigenvalues are lin
early independent.
2, )"is . eigenvalue of multiplicity
m of a square matrix A of order n, then
oF ‘nearly independent eig the number
envectors associated with
J is given by |
P=n-r, where r= rank (A — QJ), l<p<
Remark 18 m
In Example 3. I the eigenval wie |

s aaa ‘ample 3.42 the eigenvalue J = 1 is of multiplicity

(i) Example 3.42(i), th 3. We find that in
e rank of the matrix A
eigenvector.
tele i
— J jg 2 and we obtain one linearly independent ie = a a 1 4 [

ey (ii) Exam ple 3.42(ii), the rank of the matrix A —J[ js | a d

-—-« &lgenvectors. . | andent
nd we obtain two linearly indep en
< 2
_ wr

' 1; a
 Matrices and Eigenvalue Problems 3.71

(iii) Example 3.42(iii), the rank of the matrix A — I is 0 and we obtain three linearly independent
eigenvectors.

3.5.2 Similar and Diagonalizable Matrices

Let A and B be square matrices of the same order. The matrix A is said to be similar to the matrix B
if there exists an invertible matrix P such that

A=P'BP or PA=BP. (3.47)

Postmultiplying both sides in Eq. (3.47) by P~', we get
PAP =B.
Therefere, A is similar to B if and only if B is similar to A. The matrix P is called the similarity
matrix. We now prove a result regarding eigenvalues of similar matrices.
Theorem 3.10 Similar matrices have the same characteristic equation (and hence the same eigenvalues).
Further, if x is an eigenvector of A corresponding to the eigenvalue A, then P~'x is an eigenvector of
B corresponding to the eigenvalue A, where P is the similarity matrix.
Proof Let A be an eigenvalue and x be the corresponding eigenvector of A. That is
Ax = Ax.
Premultiplying both sides by an invertible matrix P~', we obtain
P Ax = AP “x.
Set x = Py. We get
P"APy=AP'Py, or (P''AP)y=Ay or By=Ay.
where B = P-'AP. Therefore, B has the same eigenvalues as A, that is the characteristic equation of
Biis same as the characteristic equation of A. Now, A and B are similar matrices. Therefore, similar
atrices have the same characteristic equation (and hence the same eigenvalues). Also x = Py, that
is eigenvectors ofA and B arerelated by x = >-Py or y=P™'x

oFTheorem 3.10 states that if two matrices are similar, then they have the same characteristic
tion and hence the same eigenvalues. However, the converse of this theorem is not true.
Two matrices which have the same characteristic equation need not always be similar.
() If Ai 7 similar to B and B is similar to C,
then A is similar to C. °
jas chai be two invertible matrices P and Q such that

A=P"'BP and B=Q™'CQ.

A=P'Q'CQP=R'CR, where R= QP.
‘Examine whether A is similar to B, where
‘
—

=
a
.
oars

a
a

y
.-
.
+
Ts:
:
j
_
;
=

;
|
os
I j
>
ancgia
na
= 4
4
as! %
ay

wie -
3.72 Engineering Mathematics

bao le ill

es are sim ila r ‘¢ the re exi sts an invertible matrix P such that
Solution The given matric
A=-P'BP or PA=BP.

| We shal l det erm ine a, b, c, d such that PA = BP and then check whethe,
Let P= °
S|
P is non-singular.
21a — B:| Sa-2b Sa =| a+2c eal
: a bli & <5 tb
d | 23 Sc— 2d de —~3at+4c —-3b+4d
” : | 2 fe Ae

em of equations
Equating the corresponding elements, we obtain the syst
5a -—-2b=a+ 2c, or An — 2b —2e = 0

5a = b + 2d, or —p—2a—0

5c —2d = — 3a + 4c, or aa 4+ ¢— 24a = 0

Sc = — 3b + 4d, or Babi Se— 4d = 0:

A solution to this system of equations isa = 1,b=1,c=1,d=2.

it
Therefore, we get P = b | which is a non-singular matrix. Hence the matrices A and E

are similar.

7 iH Dit. ..0 jam 1G) 7: Sa a ob a+c b+d

(il) = |, Or =
em Om | 0). 1 lhe sued Coed) C NAD
Equating the corresponding elements, we get
a=at+c,b=b+d or c=d=0.
ie GB ns
Therefore, P= nb |, which is a singular matrix.

Since an invertible matrix P does not exist, the matrices

A and B are not similar.
It can be verified that the eigenvalues of A are 1, 1 whereas the eigenvalues
of B are 0, 2.
In practice,
Pp fo a ee to obtain a non-singular matrix P which satisfies the equation A=
“ahB. However, it is possible to obtain the matrix P when A or B'S
agonal matrix. Thus, our interest
iagonal is to find a similarity matrix P such that for a given matrix 4s

D=P"'AP or PDP"'=A
:2 S a diagonal matrix. If suc A : 7
uch a matrix exists, then we say that the matrix A 1s
 Matrices and Figenvalue Problems 3.) .

Diagonalizable matrices

A matrix A is diagonalizable, if it is similar to a diagonal matrix, that is there exists an invertible

matrix P such that P~'AP = D, where D is a diagonal matrix, Since, similar matrices have the same
eigenvalues, the diagonal elements of D are the ei genvalues of A, A necessary and sufficient condition
for the existence of P is given in the following theorem,

independent eigenvectors.
Proof We shall prove the case that if A has n linearly independent eigenvectors, then A is diagonalizable,
Let X}, X, ..., X, ben linearly independent eigenvectors
corresponding to the eigenvalues
As, Any os « vAy (not necessarily distinct) of the matrix A in the same order, that is the eigenvect
or X,
corresponds to the eigenvalue A, j = 1,2,...,n. Let

P= [X), X2,...,X,] and D= diag (A;, Ay, ..., Ap)

be the diagonal matrix with eigenvalues of A as its diagonal elements. The matrix P is called the
modal matrix of A and D is called the spectral matrix of A. We have

AP = A[x), Xa. «<s Xp] @ (AR, AM . «25 AX)

= (A,X), AgXa, . .. , AnX,) = (Xj, Xo... X,)D = PD. (3.48)

Since the columns of P are linearly independent, the rank of P is n and therefore the matrix P is
invertible. Premultiplying both sides in Eq. (3.48) by P™', we obtain
P'AP=P'PD=D (3.49)
which implies that A is similar to D. Therefore, the matrix of eigenvectors P reduces a matrix A to its
diagonal form.
Postmultiplying both sides in Eq. (3.48) by P™', we obtain
A = PDP“. (3.50)
Remark 20
(a) A square matrix A of order has always n linearly indendent eigenvectors when its ei genvalues
are distinct. The matrix may also have n linearly independent eigenvectors even when some
eigenvalues are repeated (see Example 3.42(iii)). Therefore, there
is no restriction imposed on
the eigenvalues of the matrix A in Theorem 3.11,
(b) From Eq. (3.50), we obtain
A? = AA = (PDP™"')(PDP™') = PD2P™'.

Repeating the pre-multiplication (post-multiplication) m times, we get

A” = PDP"! for any positive integer m.

fore, if A is diagonalizable, so is A”,

(c) If D is a diagonal matrix of order n, and

3.74 Engineering Mathematics

O(A,)
Now, let a matrix A be diagonalizable. Then, we have
A=PDP" and A”=PD"P™
for any positive integer m. Hence, we obtain

Q(A) = PQ(D)P"
for any matrix polynomial Q(A).

Example 3.44 Show that the matrix

3 1 -l
A=|-2 1 2
QO 3 2
is diagonalizable. Hence, find P such that P~'AP is a diagonal matrix. Then, obtain the matrix
B=A*+5A +31.
Solution The characteristic equation of A is given by

3-A ] —]
ee ee Ae A 6A 1A
— 60, of A= 1.2.3.
0 I 2-A
Since the matrix A has three distinct eigenvalues, it has three linearly independent eigenvectors and
hence it is diagonalizable.
The eigenvector corresponding to the eigenvalue A = | is the
solution of the system
2) 1 =) | x 0 1]
(A -I)x=/-2 0 2 ||x2 |=| 0 |. The solution is <= |—)
|
0 1 l X43 0
1

1 | —1]| x, me 1
(A -2I)x=/-2 -] 2 ||x2 |=| 0 |. The solutionis x = |0 |.
0 l 0 || x3 0 | ]
 Mamces and Eigenvalue Probiems 3.73

| = 3 is the solution of the

system
0 ; !
(A —-3)z=[-2 —_»
0} | 0]
“tl
: . =O The solution
is x; = / 1 |

ral = | lol
: :
tii

i 1 Q) it - =
P= (e1,82."3) =) -1 0 lj and Pp ? ; at

ion: Oy i-1 0 1}
Itcan be verified that P~'AP = dias (1, 2, 3).
We have D = diag (1, 2. 3), D* = diag (1, 4, 9).
Therefore, A*+5A+31=P(D*?+5D+3nP™
6 oj] Is 6 of fa cae 6] [TS 8 O
Now, D+5D+31=/0 4 O}]+/0 10 O}e/0 3 O}=/0 IF OO
0 oo 9 0 oOo Bb] jo O 3] jo oO @

3 1 Of9 oOo OF-2- = 1 23 8 -8|

B2s5A23%=-|-1 0 10 17 OF 2 1 -t]=|-18 9 18}
ly 2 af O. WE). A.J 2. ho at

Examplei 3.45 _ Examine whether thematA, where Aisgivenby

rix
ld oma
E
as

fh 2 2 e
0 a GPa =(°SE"t 6
a 5 g *2g" gi
wir -
a * Ifso,obtain thematrix P suchthat PAP isa diagonal se

acteristic equation ofthematrix A isgiven by

— eS i >

; 2 2-A

— :
3.76 Engineering Mathematics

—| 2
(A — 21)x = QO Q

=-1 2

ndent eigenvector. We
of the coef fici ent matr ix is 2, it has one linearly indepe
Sinc e the rank
nt eige nve cto r cor
c o res pon din g to the eigenvalue A= 1. Since
obtain another linearly indepe nde
ix A has only two line arly ind epe nde nt eigenvector s, the matrix is not diagonalizable.
the matr
ix A is given by
(ii) The characteristic equation of the matr

fh t-2 «3
2 1-4 9, oF A=5,-—3,-3.
-6|=0 or A +/?-21A-=45
|A-AI|=|
—] -2 -A|
is the solution of the system
Eigenvector corresponding to the eigenvalue A =5

—7 2 —3 | X| 0

(A -S5IDx= 2 —4 —6 xu |= 0 |:

a 2? Sian! (0
A solution of this system is [1, 2, - he.
Eigenvectors corresponding to A = — 3 are the solutions of the system

Pa 23: Silex, 0
(A +3Dx=| 2 4 -6})/x2/=|0] or x, + 2x,-3x;,=0.
—| —2 3 X43 O

The rank of the coefficient matrix is 1. Therefore, the system has two linearly independent
solutions. We use the equation x, + 2x2 — 3x3 = 0 to find two linearly independent eigenvectors.
Taking x3 = 0, x2 = 1, we obtain the eigenvector [—2, 1, 0]? and taking x2 = 0, x; = 1, we obtain
the eigenvector [3, 0, 1]’. The given 3 x 3 matrix has three linearly independent eigenvectors.
Therefore, the matrix A is diagonalizable. The modal matrix P is given by

Pale ee edad PI4.2|-5 4° ¢

i 8
Wel a
It can be verified that P~'AP = diag (5, — 3, — 3)
Example 3.46 The eigenvectors of a 3 x 3 :ma trix A corresponding to the eigenvalues 1, 1, 3 4
[1, 0, — 1]7, [0, 1,- 1]? and (1, 1, 0)" respectively. Find the matrix A.
Solution We have

and the spectral whe wi

= hg) pectral D =|
matrix
=—
oo
©
 Matrices and Eigenvalue Problems 3.71

Lies a
We find that Pore5=| — ] LE =f}.
| ] l

Oi = Ber 20 =) =!
Therefore, A=Ppp('_1 5 0 |1 Wie 4 1 -l
—-l -1l QO ;,0 OQ I l

' I O 1] 1 --1 -1 2
=a 0 AN isl i Sih ie aa)
ot oi Ol 3 3 3 -
G JG 1.1

3.5.3 Special Matrices

In this section we define some special matrices and study the properties of the eigenvalues and
eigenvectors of these matrices. These matrices have applications in many areas. We first give some
definitions.
Let X =a (x), X2,...,X,
| :
) T and y = (yj, y2,..., y,)’ be two vectors of dimension -
n in IR” or C”.
Then we define the following:
Inner Product (dot product) of vectors Let x and y be two vectors in IR”. Then

K-y=x'y= 2 xy; (3.51)

is called the inner product of the vectors x and y and is a scalar. The inner product is also denoted by
<x,
y >. In this case x * y = y ° x. Note that x- x 2 0 and x - x = O if and only if x = 0.

If x and y are in C’, then the inner product of these vectors is defined as
fi

Y= 2xiy; and y -X=y'x=

K+y=xX » Vix;
t=]

where X and y are complex conjugate vectors of x and y respectively. Note that x . y=y-x. Itcan
be easily verified that
(ax + By)*z=a(x*z)
+ Bly-z)
for any vectors x, y, z and scalars a, B.
Length (norm of a vector) Let x be a vector in IR” or C”. Then
| x ||=vx-+x=

is called the length or the norm of the vector x.

Unit vector The vector x is called a unit vector if ||x||=1. If x #0, then the vector x/ | x || is
alwaysa unit vector.
Orthogonal vectors The vectors * and y for which x « y = 0 are said to be orthogonal vectors.
 3.78 Engineering Mathematics

Orthonormal vectors The vectors of x and y for which

llyl=!
x-y=0 and |[x||=1
are called orthonormal vectors. If x, y are any vectors and x * y = 0, then x/|| x |, y/|| y |
orthonormal.
For example, the set of vectors

Ly fO\ 0
(i) |0|,| 1],| 0] form an orthonormal set in IR?.
0 0) ]

3i 0 35/5) (—4i/5) 0
(ii) | —4i |, 0 | form an orthogonal set in C* and |—4i/5|,| 3/5], 0
0 ee 0, 0) (i + i)//2
form an orthonormal set in C2.

Orthonormal and unitary system of vectors Let x), x3, .. . , X,, be m vectors in IR”. Then, th
set of vectors forms an orthonormal system of vectors, if

; 0, its
Kj°Xj =X; Xj = ( :
: [=].

: X;, X>, . . » 5 X,, be m vectors in C”. Then, this set of vectors forms an unitary system
Let
of vector
i

On fs J
2x7 = | a
ie ee
i In section 3.2.2, we have defined symmetric, skew-symmetric,
Hermitian and skew-Hermitian matrices
We now define a few more special matrices.
Orthogonal matrices A real matrix A is orthogonal if A“ = AT.
A simple example is
\ Ae peGe =sin “;
sin @ cos @
Unitary matrices A complex matrix A is unitary if Aq
= (A)! or (A)! = AT. If A is real, ther
unitary matrix is same as orthogonal matrix. We note
the following.
| 1.: If A and B are Hermitian matri
matrices, then vr an for any real scalars &¥ B.
since OA + PB is also Hermiti |

(@A + BB)’ = (aA + BB)’ = GAT 4 BB

" = aA + BB.
2. Eigenvalues and eigenvectors of A are the
conjugates of the eigenvalues and eigenvectors
A, since

Ax=Ax gives Ax = Ax.

3. The inverse of a unitary (ort
hogonal ) matrix is unitary (orthogonal). We have A-!= A’. L*!

q
 Matrices and Eigenvalue Problems 3.79

Bo'= A =(AT)1=((A) "= [(A)]? = BT.

We now establish some important results.

Theorem 3.12 An orthogonal set tors-istinearty independent.

Proof Let X;, X2,..., X,, be an orthogonal set of vectors, that is x;° x; =0, 1 #/. Consider the vector
equation

X= QyX, + 2X, +... + An Xm=9 (3.52)

where 0%, Q2,-..-, Q, are scalars. Taking the inner product of the vector x in Eq. (3.52) with x,, we
get
K° x, = (0X + Oko +... + Ok) ey — O° x — 0
Or @(X,°x;)=0 or a ||x, ||*=0.
Since || x, ||* = 0, we get a, = 0. Similarly, taking the inner products of x with x2, X3,... 5 Xm
successively, we find that a = @,=...=@,,= 0. Therefore, the set of orthogonal vectors X), Xz, .
.. , Xm 1S linearly independent.
Theorem 3.13 \ The eigenvalues of
4t) an Hermitian matrix are real.
(1) a skew-Hermitian matrix are zero or pure imaginary.
(it) an unitary matrix are of magnitude 1.

Proof Let A be an eigenvalue and x be the corresponding eigenvector of the matrix A. We have
Ax = Ax. Premultiplying both sides by X’, we get
x Ax
x’Ax=Ax'x or A= (3.53)
XX

Note that x7Ax and x/x are scalars. Also, the denominator x/x is always real and positive.
Therefore, the behaviour of A is governed by the scalar xX’ Ax.

(i) Let A be an Hermitian matrix, that is A = A’. Now,

(RTAX) = x7 AK = xTATE = (XAT)? =FTAX
— J
since x’ A’ x isascalar. Therefore, x’Ax is real.
From Eq. (3.53), we conclude that A is real.
(ii) Let A be a skew-Hermitian matrix, that is A’ = — A. Now,

(%TAx) = x7 AX = —x"A'X = —(x7ATX)T = _XTAX

since x’A’x is a scalar. Therefore, X'Ax is zero or pure imaginary. From Eq. (3.53), we
conclude that A is zero or pure imaginary.

(iii) Let A be an unitary matrix, that is A~'=(A)". Now, from

Ax=Ax or AX=]1<X (3.54)
(AX)7=(A%)" or ¥7AT = J XT,
 3.80 Engineering Mathematics

or xTA = 1. (3.55)

Using Eqs. (3.54) and (3.55), we can write

(x7A-! (Ax) = (AX? (Ax) =[A|?x"x

or x'x=|A|*x’x

Since x 0, we have X7x # 0. Therefore, |A |* = 1, or A = + 1. Hence, the result.

Remark 21
From Theorem 3.13, we conclude that the eigenvalues of
(i) a symmetric matrix are real.
(ii) a skew-symmetric matrix are zero or pure imaginary.
(iii) an orthogonal matrix are of magnitude | and are real or complex conjugate pairs.

Theorem 3.14 The column vectors (and also row vectors) of an unitary matrix form an unitary
system of vectors.

Proof Let A be an unitary matrix of order n, with column vectors x,, X>,..., X,- Then

— 7 “=—T =i Hin

—7 | =—T =T
Wad. eg Te, =.| 2 XX K7%2
AT A=A‘A=|,* |[xi,X5,...,X%,]=].°

x T —
EX) BoM
—

¥ O, it*j
Therefore, a xj; = : > Lie
We = i= a.

Hence, the column vectors of A form an unitary system. Since the inverse of an unitary
matrix is also
an unitary matrix and the columns of A“! are the conjugate of the rows of A, we conclude that
the row
vectors of A also form an unitary system.
|
Remark 22

(a) From Theorem 3.14, we conclude that the colu

mn vectors (and also the row vectors) of 4”
orthogonal matrix form an orthonormal system of vectors
(b) Asymmetric matrix of order n has n linearly independen
t ei £envectors and hence is diagonalizable.
Example
ile 3.47 Show
on that the matrices A A and A° ‘ihave the same eigeny | a. wail les
the eigenvectors corresponding to A and AT are mutually oar and for distinct eigen
Solution We have
|A-AI 1A" ATT |= (fA? any] =| a7— Ar
pince A and A’ have the same characteristic equation,
they have the same eigenvalues.
Let A and yp be two distinct eigenvalues of
A. Let
X be the eigenvector corresponding ' the
 Matrices and Eigenvalue Problems 3.
ie: | in
eigenvalue A for A and y be the ej seaveets
Ax = Ax. Premultiplying by y”, we wet Or corresponding to the eigenvalue yz for A’. We hav

yiAx=Ay"x. (3.5¢
| . re | a
sepia A'Y=Hy, or (ATy)’=(uy)” or y7A= py’.
Postmultiplying by x, we get

y'Ax = py"x (3.57

Subtracting Eqs. (3.56) and (3.57), we obtain

(A— wy’x = 0.
Since A # HW, we obtain y’x = 0. Therefore, the vectors x and y are mutually orthogonal.

3.5.4 Quadratic Forms

eG Re 2 ee =e be an arbitrary vector in IR”. A real quadratic form is an homogeneou

expression of the form
it fl

O= 2 Ss Suir (3.98
in which the total power in each term is 2. Expanding, we can write

QO = ayx” + (ay2 + A291) XjXQ +--+ (Ayn + An X1Xy

+ a22x 5 + (3 + €32)XoX3 +... + (Gay + Anz) XX,

PE AE 77 ee
= x' Ax (3.59)
using the definition of matrix multiplication. Now, set 6; = (aj + a;)/2. The matrix B = (b;;) is
symmetric since bj; = bji. Further, 5; + 5); = aj + a); Hence, Eq. (3.59) can be written as
OQ =x' Bx

where B is a symmetric matrix and bj = (aj + @j;)/2.

For example, for n = 2, we have
by, = dj), bi = b>, = (a2 sf @z,)/2 and b> = Ap).

Example 3.48 Obtain the symmetric matrix B for the quadratic form

+ x3:
@) Q = 2x? + 3x,;X2 : ‘ :
; 5X5 4 4x3
+ : ) : Q=x}+2x 1x2 — axx4 + 6X 3
2X
‘

7
rr

‘ =

ee ra Tt af,
os
=

1 @

= 1. Therefore,
@y, = 2, 42 + 421 = 3 and a>2
r l 3 |
. j by = 4 = eg bi2 = bai = 2 (a\2 ne a2 )= F and boo =i0's3'= :,
3.82 Engineering Mathematics

2 3/2"
B= so :
Therefore,
5 33
1 = — 4, a93 + 43, = 9, a2
2421 = 2, a\3 + 43
lL, @j+
(11) ay, =
— 2
b= a= 1 big ba =
hay.
% (aia + 221)
= 1, 543
= b-3] =
5 (ay3 + 431) =

3 = 933
== — 9» 93— = 4,
= 5(a 23 + a2 ) = 3, dy = Aa,
b3 = b32
l t =2 |
B= 1; Pees a
Therefore,
—2 3 4

ned as
If A is a complex matrix, then the quadratic form is defi
|
non x

Ax (3.6]
Q= z 2X ay XiXj= x
i=] j=

form is usually define

where x = (x, X,.-., X,) is an arbitrary vector in C", However, this quadratic
for an Hermitian matrix A. Then, it is called a Hermitian form and is always real.
; ae . | Ll+i |
For example, consider the Hermitian matrix A = 4 }The quadratic form becomes
1-1

a ee . ee ] l+i X|
O'=x Ax= [x},%X2] eae D io

= |x, [* + (1 + 1)X\X2 + (1 — 1)X 1X4 +2 X92 ie

= | x, |?4 (X1X2+xX1X2) 1 i(X,X2 —X,X2) + 2 | x2 12

Now, x)%2 + 1X2 is real and X;x2— x, is imaginary. For example if X = P; + ig, X2 = P2 + '92
we obtain —
HiX2 + 1X2 = 2(PiP2+ 9192) and X\xq—x)%2=2i(piq> — poq)):
We can also write

(X1 X72 +X 2X1) + i(X,xX2—X,X>) = 2[(P1 Pr am 4192) Bs (pPiq2 — p~q))]

=2Re [C1 7 1)X,x>]

Therefore, Q =| x, |’+2Re[(1 + i)X,x.] + |x, 2.

Positive definite matrices

Let A = (a;;) be a square matrix. Then, the matr
ix A is said to be positive definite if
Q ee
= x TeAx > 0 for
= |
any vector x #0 and xx’Ax = 0, if and only if
x = 0.
 Matrices and Eigenvalue Problems 3.83

If A is real, then x can be taken as real.

Positive definite matrices have the
following properties.
1. The eigenvalues of a positive definite matrix are all
real and positive.
This is easily proved when A is a real
matrix. From Eq. (3.53), we have

A = (x"Ax)/(x"x)
Since x'x > 0 and x7Ax > 0, we obtain A > 0. If A is Hermitian, then
real (see Theorem 3.13). Therefore, if ¥7Ax is real and A is
the Hermitian form Q > 0, then the eigenvalues are real
and positive.
2. All the leading minors of
A are positive.
Remark 23

(a) If A is Hermitian and strictly diagonally domi

nant (| aj|> 2 |ajj|,i=1,2,...,2)
with
.
positive real elements on the diagonal, then A
met jJ=1,/#I

is positive definite.
(b) If x’Ax > 0, then the matrix A is called semi-pos
itive definite.
(c) A matrix A is called negative definite if (— A) is positive
definite. All the eigenvalues of a
negative definite matrix are real and negative.
Example 3.49 Examine which of the following matrices are positive definite.

ila
oie Me ae (c) A fi P5
as
=
of A
| = 5
i
Cc = :
| =
-I QO

Solution

7 3 il Xj 4
(a) (i) =x AX = [xX),.%51] 2 pubes = 3x; + 3x)x2+ 4x5
>

gles
= {x E 32)
“ime
se Te? >0O forall x0.

(ii) eigenvalues of A are 2 and 5 which are both positive.

(iii) leading minors | 3 |= 3, ae |

= 10 are both positive.
Hence, the matrix A is positive definite (it is not necessary to show all the three parts).
= ae 3 —2i X | ey 3x, — 2ix,
(b) QO = x'Ax = [x),X2] oi 4 || x =t1,%2] ositm
Bye — 21003 + Bix hy 4 4x52).
Taking x, = p, + iq, and x2 = p2 + iq2 and simplifying, we get

= 3(p? + 47) + 4(P5 + 93) + 4(p192— p24)

=p? +qj + 2p2+ 243 + 2 p2- G1)? + 2(p, + g2)2>0:
 4.84 Engineering Mathematics

Therefore, the given matrix is positive definite.

Note that A is Hermitian, strictly diagonally dominant (3 > |— 27 |, 4.>| 2i|) with positive re.
diagonal entries. Therefore, A is positive definite (see Remark 23(a).)

| 0 i Xy X1 + 1X3

(c) Q=x'Ax =[X,,%2,%3]} O 1 O|/x2|=[%1,%2,%3]] = x2

—I 0 3 x3 | —1X,+ 3x3

= Ny Xy + 1X,X34 + X9X7 — 1X1X3 + 323%;

= |xy |? + |xo |?+ 3 |x3 |? + iex3 =21%5)

Taking x; = py + ig, X2 = p2 + ign, x3 = p3 + ig; and simplifying, we obtain

QO = (pi + ai) + (p+ 93) + 3(p3+ 93) — 2(P193- P34)

= (Pi— 493)” + (pa + 91)" + (P35 + 95)+ 2(p3 + 93) > 0.
Therefore, the matrix A is positive definite. It can be verified that the eigenvalues of
are 1, 2, 2 which are all positive.
Example 3.50 Let A be a real square matrix. Show that the matrix A7A has real and positi\
eigenvalues.
Solution Since (A‘A)’= A7A, the matrix A‘A is symmetric. Therefore, the eigenvalues of A’A ai
all real. Now,

x"A™Ax = (Ax)’(Ax) = y’y, where Ax = y.

Since y’y > 0 for any vector y # 0, the matrix A7A is positive definite and hence al] the eigenvalue
of A‘A are positive. Therefore, all the eigenvalues of A‘A are real and
positive.

Exercise 3.4

Verify the Cayley-Hamilton theorem for the matrix A. Find A”, if it exists, where
A is as given in Problems
to 6.
1 O —=-4 Si) gees | cis =2 ]
Path Se 4: 7Heal tay Sued eae ame MI
—-4 4 3 Les 58 3 1 -l
he 2 oee2 l l an L chideyt
a 1 |. ~5 l 3 =] 1 Gy Tf zh,
et Li <2 ihe, 4 | ae
Find all the eigenvalues and the corresponding eigenvectors of th ven j
of the matrices are diagonalizable? ‘© OF to 18. Which
(he matrices given in Problems 7 to

ee ee
She| =0 |. oe Ts
ae. |

gk a =e. 04
 Matrices and Eigenvalue Problems 3.8:
i . =| «

Lad
ial
= —_* — cS =

gt i i # @|
Diet Ge 2 =a =O it = 7
ea | l ~

0 if| 14. hingi 8 ‘|! Ce ae |

=| J] 2 i | st. 22 i
Seid Il od nomeQ lL hee hie ui

iteop 8 « 0-0) 0 05 3 Oo +f} 6-"e" "a

+ Aha 4 0 "0" oF 1 6 i-th “rr oF a
S.A sida Le 0 "O TG ol is (0 0 O80 4 OL
a OO 10 "oO OG g "0 “so 8")
LL 6 60 eo ree “6 ™ Oa
Show that the matrices given in Problems 19 to 24 are diagonalizable
. Find the matrix P such that P~'AP is a
diagonal matrix.
Ci) et =3) oot y 1
19, @ 2 ol 20: ||2) “toma Sieh! ewan zt
fenDp3 = Gras =3).0 0)
eo: i Ss S6 6 ele de>)
22. Soe At 23. =] ae a oy in det
=) 37 So a (tS oa
Find the matrix A whose eigenvalues and the corresponding eigenvectors are as given
in Problems 25 to 30.
25. Eigenvalues: 2, 2, 4; Eigenvectors: (— 2, 1, 0)", (— 1, 0, 1)”, (1, 0, 1)”.
26. Eigenvalues: 1, — 1, 2; Eigenvectors: (1, 1, 0)’, (1, 0, 1)”, (3, 1, 1)7.
27. Eigenvalues: 1, 2, 3; Eigenvectors: (1, 2, 1)’, (2, 3, 4)", (1, 4, 9)".
28. Eigenvalues: 1, 1, 1; Eigenvectors: (- 1, 1, 1)’, (1,-1, 1)”, (1, 1, - 1)",
29, Eigenvalues: 0, — 1, 1; Eigenvectors: (— 1, 1, 0)", (1, 0, - 1)", (1, 1, 1).
30. Eigenvalues: 0, 0, 3; Eigenvectors: (1, 2, - 1)", (— 2, 1, 0)", (3, 0, 1)7.
= E Let a 4 x 4 matrix A have eigenvalues 1, — 1, 2, — 2. Find the value of the determinant of the
matrix
B=2A+A7!-I.
32. Let a 3 x 3 matrix A have eigenvalues 1, 2, — 1. Find the trace of the matrix B = A — A~! + A2.
33. Show that the matrices A and P~'AP have the same eigenvalues.
Let A and B be square matrices of the same order. Then, show that AB and BA have the
same eigenvalues
but different eigenvectors.
Show that the matrices A~'B and BA“ have the same eigenvalues but different eigenvectors.
35,

Ann Xn matrix A is nilpotent if for some positive integer

k, A‘ = 0. Show that all the eigenvalues of a
nil ootent matrix are zero.
, ~zAP,
If A is an n x n diagonalizable matrix and A’ = A, then
: a
show that each eigenvalue of A is 0 or 1.
e x _
3.86 Engineering Mathematics
|
.

= P ‘AP, where P ;
+ 1S fi dtoa diagonal matrix D
= .
38. Ww

a)
cos @ see and tan 2@= 2h
of the form P= a-0
sin@ cos@
B™ for any positiy.
simi lar to B. The n sho w that (i) A-' is similar to B™', (ii) A™ 1S similar to
39. Let A be
integer m, (iii) | A |=| B |.
if and only jg
B be symm etri c matr ices of the same order . The n, show that AB is symmetric
40. Let A and
AB = BA.
41. For any square matrix A, show that A’A is symmetric.
_
42. Let A be a non-singular matrix. Show that A’A™ is symmetric
if an d only if A? = (A’)’.
43. If A is a symmetric matrix and P-'AP = D, then show that P is an
orthogonal matnx.

. Show that the product of two orthogonal matrices of the same order is also an orthogonal matrix.

ly my tO
4S. Find the conditions that a matrix A=|/, m2 nz | is orthogonal.
I; m3 ny
If A is an orthogonal matrix, show that |A |= 1.
47. Prove that the eigenvectors of a symmetric matrix corresponding to distinct eigenvalues are orthogonal.
. A matrix A is called a normal matrix if AA’ = A’ A. Show that the Hermitian, skew-Hermitian and
unitary matrices are normal.
49. If a matrix A can be diagonalized using an orthogonal matrix then
show that A is Symmetric.
30 Suppose that a matrix A is both unitary and Hermitian. Then,
show that A = A7!.
aL, If A is a symmetric matrix and x7Ax > 0 for eve
3 ry real vecto xr that
# 0,, then Owshow that Az |is real—and
z*Zz’Az
positive for any complex vector z + 0.
52. Show that an unitary transformation y = Ax
= ’ wh tr ‘
product. ere A is an unitary matrix preserves the value of the inner

53 Prove that a real 2 x 2 symmetric matr

ix b ee
th | 18 Posiiti
tive definite
ite if
j and only ifa>0(1 x | leading
minor) and ac — b? > 0 (2x2 leading
minor),
2s al 3.
Show that the matrix |~3 4 —1|
54,

is positive definite.
|-l |] 2 |

io) ama |
Show that the matrix |—2
0 4] is not POsitive
definite.
=—G \ S93 as

By — 2x 1X2 + Axyxs — x2 + x2
57.
, 2 |

58. ae Si
x, + 2ix\x» = Bx x; T 4ix>x, + 4x?
 Matrices and Eigenvalue Problems 3.87
§Y, vy: = ’ + AON
4 | +
= (4 = Daa, 4 x2
ve . Ave.
ol), ‘x ty 7 (QO s i‘
:

3.6 Answers and Hints

Exercise 3.1
: | |
ey . 2 -3 4 5)
A™ = 5 =3 2 |. 4. A-! = aN 9 —| —4
|| 2 =«3o | 2 > ee ts eee |
a (i) |A aaj (A) |= diag ( |A|,| A avait’ | A|)=|A_|" (use property 10 of determinants). Theref
ore,
|Adj (A) |=| A |""!
(i) Let B = adj (A). Since B! = adj (B)/| B |,we have B adj (B) =| B | I.
Therefore,
adj (A) adj (adj (A)) = |adj (A) [[=| A |".
Premultiplying by A and using adj (A) = A™'| A |, we get
A[A™| A |] adj (adj (A)) =| AJ""' AI or adj (adj (A)) =| A |""? A.
|AA*|=|A [| A] =|] or] At] =1// A].
10. (BAB’)’ = BA’B? = BAB’.
13. AB=BA > B'AB=A => B"'A = AB“! Similarly, A“B = BA“.
(i) (AB“)’ = (B"')‘Al = (B’)'A! = BA = AB".
Gi) (AB)? = B7(A7!)! = BA")! = BAT = AB.
(ii) (A“B™)? = [((BA)"']"= [(ABy']"= (A7)""(B"y" = AB.
14. Premultiply both sides by (i) I-A, (i1) I+ A.
15. (PAQ)! = QAP! =I A'P'=Q> A" =QP.
16. Use (I-A)I+A+A’7+...J
=I
17. (ABC)(ABC)" = I. Premultiply successively by A”, B! and C"!.
21. Multiply C; by a, C2 by b, C3 by c and take out a from R,, 6 from R», c from R3.
cosa O]|| cosa cosb sine
sina@
snB cosB O sina sinb sinc|=0

lsiny cosy 0 0 0 0

|ap as = am, 1, 23 1: 30, 1, 1,2.

ee he@ 32. (i) k #2 andk 4-3, (i) k=2, ork =-3.

33. @=7/6, or O=sin '[9- J161)/4].

10.
35. (i) A¥3, p arbitrary, (ii) A = 3, # = 10, (iii) A= 3, w#
‘i hm 6 38. 2.
36. 2

Dm a ts
0 ichq+r#
[Al=0- 94-0ra TAPS AS
ndp+ 0;
— (i) 3, ifpeqe
bs

? o

at
 ASS Engineering Mathematics

(ii) Vifpegetrandp+g+tr=0,
(iii) 2, if exactly two of p, ¢g and Fr are identical;
(iv) lLip=q@qe=er

40. (a) 2: (b) | Al = (ayaa + yb) + Cyeg)*, rank (A) is

(i) 4, if ajay + Bpdy + eye #0;
value zero.
(ii) 2, if ajay + Bybs + Eye) = 0, since all determinants of third order have the
. Conder (1 + AU A + At... +(— 1 'A™) = 14 (- 1)"A". In the limit 2 > oe, A" > 0. Therefore.
(i+ A= A + At. Jen
" Ha

43, (i) Trace (@A + BB) = @ ~ a, + B ~ b,, = @ Trace (A) + B Trace(B),

(ii) Trace (AB) = oy * a,b; = 2 z bi »Qmi = Trace (BA),

(iit) If the result is true, then Trace (AB — BA) = Trace (I) which gives 0 = n which is not possible.
Result is true for p= 0 and 1, Let it be true for p = & and show that it is true for p= k + 1. Note that when
BC = CB and C* = 0, we have CB**! = B*'C and CB‘C = 0.
45. Apply the operation C) — C; + C) +... + C, and then the operation R; <— R; — iS) GP aS Se
46. None. 47. Symmetric. 48. Skew-symmetric.
49. Hermitian. 50. None 31. Skew-Hermitian.
52. None. 53. Skew-Hermitian. 54. Hermitian,
55. None.

Exercise 3.2
1. Yes. 2. No, 1, 4, 5, 6. ae No, L, 4.5: 6.
No, when the scalar is irrational, property 6 is not satisfied. If
the field of scalars is taken only as
rationals, then it defines a vector space.
.
Yes, since 1+x=lx=x=x | =x] =x=x. the ze
xX + (— x) = x(I/x) : | a = hats
ee . ciehae : —
_ No, 8, 10. FINO. 3,-8.0.
Yes (same arguments as in Problem 5.), (@ + B) x= xFFP = ~% 8.
po, 8 ax + Bx.
(i) Yes, (ii) No, 1, 6.
(i) Yes, (li) No, 1, 4, 6.
(i) Yes, (ii) No, whenx, ye Wx 4 ye W.
(iii) No, whenx,ye Wex+ye@W,
(iv) Yes.
12. (i) No, when Ae W, @A € W for @
negative,
(ii) No, sum of two non-singular
matrices need not be non-singular
(iii) Yes,
(iv) No, @A and A +B need not bel
ong to W, (A =I, A27=I=A but 2A
# (2A)?).
a . : (li) No; let @ =i. Then QA=iA€ W.
» (i)No;forP,QEW.P+QeW, (ii) Yes
A
 Matrices and Eigenvalue Problems 3.89

(ili) No; for P, @€ W, aP @ Wand also P + Q¢ W,

(iv) No, for P, Q € W having real roots, P + O need not have real roots. For example, take
P=27-1,0=-°43

15. (i) Yes,

(i) No, x,y € W,x+y¢ W. For example, if x = (x, x, x — 1), Y= On. Yr Yi — “YS
X+y=(x, + V1 X1+ yi, x, + y,-2) € W.
(iii) No,x€ W, @x¢€ W, fora negative,
(iv) No,x€ W,ax¢ wW, (v) No, xe W, ax € W, (for aa rational number).
16. (i) u+ 2v-w, (ii) 2u+v-w,
(iii) (— 33uu— ll v + 23 w)/16.
17. (i) u-—2v + 2w, (ii) 3u+v-w, (iii) not possible.
18. (i) 3P,(t) — 2P,(t) — P,(2), (ii) 4P,(t) — P2(t) + 3P3(t).
19, Let S = {u, v, w}. Then, x = (a, b,c)’ = au+ By + yw, where a@= (a + b)/2, B= (a
— b)/2 and y=c.
20. LetS| = {A, B,
| a
C, D}. Then, E = *
b)
4 = aA+ BB+ yC + dD, where a= (—a — 6 + 2c — 2d)/3,
c
B= (Sa
+ 2b — 4c + 4d)/3, y= (- 4a - b + 5c — 2d)/3 and 8= (-2a+
b+ c—-d)/3.
21. (i) independent, (ii) dependent, (i11) dependent,
(iv) independent, (v) dependent.
22. (i) independent, (i1) dependent, (ii) dependent,
(iv) independent, (v) independent.
(— 4, 7, 9) = (1, 2, 3) + 2-1, 3, 4) — (G3, 1, 2). The vectors in S are linearly dependent.
25. P+t+1=([-1+ (" -1) + 2¢7 + 2 + 2)]/3. The elements in S are linearly independent.
(i) dimension: 2, a basis : {(1, 0, 0, — 1), (0, 1, — 1, 0)},
(ii) dimension: 3, a basis: {(1, 0, 0, 1), (0, 1, 0, 1), (0, 0, 1, 1)},
(iii) dimension: 3, a basis: {(1, 0, 0, 0), (0, 0, 1, 0), (0, 0, 0, 1)},
(iv) dimension: 1, a basis: {(1, 1, 1, 1)}.
27. The given vectors must be linearly independent.
(i) k#0, 1, - 4/3, (iil) k #0, (ili) k #0, (iv) k¥-8. °
28. (i) dimension: 4, basis: {E,,, Ey2, E2;, E22} where E,, is the standard basis of order 2,

i NOV 60) 1) fo 6
(ii) dimension: 3, basis: + 6 olla “Org: Salt

0 1
(iii) dimension: 1, basis: (_ |

(iv) a2 x 2 skew-Hermitian matrix (diagonal elements are 0 or pure imaginary) is given by

id, b, RLS ib, | - 0 b, | ay bs

a= —b, + ibz lay | —b, 0 + b, a> =B+iC

whereB is a skew-symmetric and C is a symmetric matrix,

3.90 Engineering Mathematics

dimension: 4, basis:

(vy) dimension: 3, basis: |

ff =1) (0 -0\ (0 0
(vi) dimension: 3, basis:
0 lh dhe if
29. (i) dimension: 3, basis: {E,,, E22, E33},
(ii) dimension: 6, basis: {E,,, Ey2, E13, E22, Ex, E33},
(iii) dimension: 6, basis: {E,,, E21, E22, E3;, E32, E33}.
where E,, is the standard basis of order 3.
30. (i) n’, (ii) xn, (iii) n(n+ 1/2, (iv) n(n —1)/2.
31. Not linear, 7(x) + T(y) # T(x + y).
a2. Linear.
33. Not linear, 7 (x) + T(y) # T(x + y).
34. Not linear, 7(1,0))}=3, 7(0; 1)= 2, TU.1) =0 4 7(1,.0) + TO, 1).
35, Not linear, 7(x) + T(y) # T(x + y).
36. ker(T) = (0;0,0)!,.ran (F)= x(1, 0;,1)'+y, 0, = 1)7 +:<(0, 1, 0)¢.
dim (ker (T)) = 0, dim (ran (T)) = 3.
37. ker (T) = (0, 0)’, ran (T) = x(2, - 1, 3)’ + y(1, 1, 4)". dim (ker (T)) = 0, dim (ran (T)) =2.
38. ker (D).= we — 2,0; 1%.
ran (T) = x(1, 0, 0)"+y(1, 0, 1)? + 2(0, 1, 0)7 + w(1, 0, 2)7
= r(1, 0, 0)’ + s(1, 0, 1)" + 2(0, 1, 0),
where r=x-—w,s=y + 2w. dim (ker (T)) = 1, dim (ran (T)) = 3.
. ker (T) = x(—3, 1)’, ran (7) = real number. dim (ker
(T)) = 1, dim (ran (T)) =1
. ker (T) = x(1, — 3, 0)’+ z(0, 0, 1)", ran (7)
= real number.
dim (ker (T)) = 2, dim (ran (T)) = 1.
=&& .
- ker (T)=x(1, Ly", ran (T) = x(1, Neal 1)’ = r(1, i
where r= x — y.
dim (ker (T)) = 1, dim (ran Cea kk
-er(T)(T) = x(1, A DaanT £ |
42. “ 3) ’ — (T) = x(2, a)" +.y(- is QO)? + z(0, LF Or ran (T) — n= ] 0)? + 5(0, 1)/, where

r= y+ 2x,5= z+ 3x. dim (ker (T))= 1, dim (ran (T)) = 2. sli

ala
: L 1 oa:
43.
. , at (ii) a-| neget an |

0 ——,bh
SS
-_ > ll

bb
NM a

a =e 172"
 Matrices and Eigenvalue Problems 3.91

—1/2 m2 3/9
45.
mit
A =| =1/2 — 3/2 —1/2 |.
‘
0 —] —]
0 0)

—| l 1 i ee 4
47. We have Tlvi, V2] =([w,,w2, w,]A = a1 Pie Sh |2 Fp
I a tad 4 0

Now,
OW, a
any vector X ==i(x, x2)" T in
. IR? with
; respect . .
to the given basis can be written as

Pletalols]
We obtain a = (— 4x, + 3x,)/2, B= (2x; — x)/2. Hence, we have

4 5] [4a+5f —6x,
+ 7x» |
Tx=aTv, +BTv, =a|\2|+B)3|= 2a+3B| => —2x,+3x,
0 1 B 2X) —X4

—xX,;+2x,+ 8x
48. Tx = |
—2x; + 3x2 + 12x; |

49. T Py(t) = (4x2 — 5xy) + 7 (x2 — xy) £ + (2x, — x2) &*.

50. (i) Two degrees of freedom, dimension is 2, a basis is {[3, 1, 0], [- 2, 0, 1]}.
(ii) One degree of freedom, dimension is I, a basis is {(— 5, 4, 23)}.

Exercise 3.3

| Ome 2. 2 3. 3. 4. 2.
aay, 6. 2. 7. 2. 3

9. 4. 10. 2 ll. 2. 1243.

19. 3) 14. 2. 15. 2:

16. Independent, 3. 17. Independent, 3. 18. Dependent, 3.

19. Independent, 3. 20. Dependent, 2. 21. Dependent, 3.
22. Dependent, 2. 23. Dependent, 2. 24. Independent, 4.

25. Dependent, 2. 26. [1, 2, 2].

3.92 Engineering Mathematics

27. [1 + a —2«a, a], @ arbitrary. 28. Inconsistent.

an Th th a0, (1. 33.3); S31. [8/2, 3/2; 1. 32. [-— 1, — 1/2, 3/4)
33. [(5 + a-4B)/3, (1 + 2a+ B)/3, a, B], a, B arbitrary.
34. [2-a, 1, &, 1], @ arbitrary. 35. [— 1/4, 1/4, 1/4, 1/4].
36. [— & q, a], @ arbitrary. 37. [- 15a, 13a, a], & arbitrary.
38. [0, 0, 0}. 39. [— 2a/3, 7a/3, — 8a/3, a], & arbitrary.
40. [2(68— o)/3, — (58 + a)/3, B, a], a, B arbitrary.
41. [0, 0, 0, 0}. 42. [(2B-5a)/4, — (108 + a)/4, B, a], a, B arbitrary.
43. [(a@ + 5f)/3, (4B - 7a)/3, B, ai], o, B arbitrary.
44. [38 — 50/3, (38- 4a)/3, B, a], a, B arbitrary.
45. [a-38, 5B, B, al, a, B arbitrary.

A It aad ai. a2 ‘4 5),

1 l —| ] |
alt i -1 (|
ie 2 1 [a 2 (3) ds 1
i ah = fiat <18 1 1
49 1 |9 l 0 cn l 0 QO —-]
2 i i TO = llcall 1 0° 0
cd @ i Sie 18 “6

Exercise
3.4

i - 16 =90
1. P(A) = 13 -9A?~92
+ 81 =0: Att 1G 13 al.
= V=o Mg aes
6 <4 23
2. P(A) = A} - 8A? + 202 - 16
=0: A“t= 4 Gill eRe «os|
—2 =4 6
3. P(A) = A? — 3A? + 2A = 0; Inverse does not exist.
oh a
4. P(A) =A -2?-444+4=0: Aled ee
<2 1
2 gy
§. P(A) =A? - 52? + 94 - 13 = 0: Ants 4 Li S Sener
ee eee,

6. P(A) =A? - 3A + 6A-44+ 21 =0: Ant=— 143i

 Matrices and Eigenvalue Problems 39 La)
‘ald

PA EAA YA = 2, 25:2, 1
0)%) ne diagonalizable.
(ey1:0
8 A=- , ,-1ae ae ae
L D5 ASE 4; l, 1)"; A=—i: (1-3, 1, 1, diagonalizab
9. A=1, 1, 1: [0,.3, —2]7: le.
not diagonalizable.
= , 1: (0, 1,- —1}1];
10. A=1 '; A =7.
A=7: (6, 7, 5)’; not :
diagonalizable.
ph 0, Wy A= t € 3: (1, 45. fy |’
1

A=1- 3: [l, -(/3+ 1), -i]";

diagonalizable.
a= obi th ©, — 1", (f.— 1. 0)’; A = 2i: [1, 1, 1]"; diagonaliz
able.
13. A=0, 0, 0, 0: [1, 0, 0, OJ7: not
diagonalizable.
14. A=0, 0: [1, 0, 0, 1]7, (1, - 1,~
1, 0)’; A=2: [1, 1, 0, oj”:
=— 2: [1, 0, 1, 1]"; diagonalizable.
15. 4=— 1,-1,~1: [1,-1, 0, 0), [1, 0, -1, of", (1, 0,0,- 1)";
A = 3: [1, 1, 1, 1]; diagonalizable.
iG 4——4- (1,1, - 1,= 1)"; A = 10: (1, 1,1, 17 A= V2: (2-1, 1- 42.-L A,
A= —V2:[-(1 + 42), 1+ 42, -1, 1]; diagonalizable.
17, A=—1,-1:[1, 0, 0, 0, - 1]7, (0, 1,0,-1, 0J7; A=1, 1, 1:[1,0,0,0, 1}’, (0, 1, 0, 1,
OJ", [0, 0, 1, 0, 0)”:
diagonalizable.

18. A= 1, w, w’, w°, w*, w is fifth root of unity. Let €;= w’,7 =0, 1,2,3,
4.4= cf (1. &;, Es e Esl =
j = 0, 1, 2, 3, 4; diagonalizable.
19. A=2, 2: [1, 0, — 1)’, [—-2, 1, 0]; A=4: [1, 0, 17.
es I fo
=| 0? GErt=2}0 2 “ot
=e “fF ie eae |
20. A=1:[1,—2, 0)’; A=—- 1: [3, —2, 2]; A=2: --1, 3, 1)”.
i 3
eo =2
0 2
—47"
1-37
2 A=6 B, .— 21: A=2i: (3 +i, 1 + 3i,-4)';A=-2:: Bai

és 3+1 —8 16

P=/1 1+3i 3367 AS

2S : i-6 2+6: -8]

22. A=0-[1,0,- 1 4=1:(- 1,-1.1 4=2 011, oF

” == Tees 1 =], 0
, meen (Ee H|t. =i)
2 hei eae
Th 3 kl
Sided id, 3)";A=2,2:(2, 0,17. 2.1, oF.
a ae: = he Dalal
,> : | Peo i¢r*=| > = =5
3.94 Engineering Mathematics
vat
M4. A=1:(1,-1,-175 A=2: (0, 1,17, 42-2: (8-5, 7:
) ekg IZ BB
r-|- 1 -5hP=—/12 15 ~-3}
i ee eee Ole: teh hignt
=) «alm J CG me 3 :
P=)
| 1 0 OPapet tas
2.2 lady 3 = PDP-' =| 0ts
1|;A
oO” fe i .a 1
im tf 3 212 1 alae:
MH P=11 0 IFPSl-1 1) 2) A= PDP= Gi)
@ 1-4 1 -1 -! en mA
? 2. 4 =i 14 5) | mh “12 «(G
27. P=|2 3 4;bP4=2 | 14 -8 25A=PDP=--| 2 4 14
9 1 a oe u34 - 4) 38
pa: 2 21 i or <0
mea 2 ey pte 19) oA ppp "2 0 ||) a ol
a ed 1s 16 0 oO 1
a 4 =| 2 4 i On 3
P=| 1 0 bP t= 4 1-4 -2|,A = PDP"! = 4 aes
Post 1 ei, oi a 5:Ss Salis
28. 3 Peay = 3 9 18 45
30. P=/| 2 ] 0; P= 2 =) at 6); A = PDP-! = 2 0 0 0 |
iii 1 ie? 5 oe fs
31. Eigenvalues ofB are 2A, + (1/A;) - 1,j= 1, 2, 3, 4 or 2,
— 4, 7/2, — 11/2. |B |= product of eigenvalues
of B = 154.

32. Eigenvalues of B are A, + A‘

_ (1/A)), J = 1, 2,3 \.0r' 1, 11/2, 1.
Trace of B = sum of eigenvalues of
B= 15/2.
33. Premultiply Ax = Ax by P~! and sub
stitute x = Py.
34. Let A be an eigenvalue and
x be the cor
A“ and substitute x = Ay. We get BA res ponding eigenvector of AB, that is ABx = (x. Premultiply by
are related by x = Ay.
y = J y. Therefore, A is also an eigenvalue of BA and ¢i genveclor 5
35. Let A be an eigenvalue and
x be
by A and set x = A-!y. We obt the corres ponding eigenvector of A~'B, that is A~'Bx = Ax. Premultip?
ain BA" y = Ay. Therefore, A is also an
corresponding eigenvector ¥=
As.
eigenvalue of BA” with!
36. From Ax = Ax, we obtain A*
x = Jk x= 0. Theref
37. Since A is a diagonalizable matrix
ore, 74 = 0orA=0, sinc
xe+ 0.
: .
peggy of A and pD are » there
same. exists
We a have
non-singul
a ar2pmatri
: a =

_ ix P such that r ‘, . = DP ve
and $0! L J tr | ‘ i { d alt

PAP = D’.Therefore, we obtain D2

or 1. Thus D ()
. = D*. Since A?’ = A,
oD=I. Hence, the eigenvalues of A as
 Matrices and Eigenvalue Problems
3.95
38. Simplify the right hand
Side and set the off- diagonal
element to zero.
39. Since A and B are
Similar we have A = P'BP.
A = P'B- Pp and A”
From this equation, show that
= P'B P. | Also| A|=|P™'||B||P]=|B].
40). We have A =An
A‘ and B7 = BT Therefore, (AB)’ = B’A? = BA.
41. (ATA)? = ATA. |
: Ta-l
42. a A a be a symmetric3 matrix. We have (A7A™')? = (A“)7A = ATA)
or (A7)TA2 = AT or
A= = (A ae Now, let A? = (Al)? We have AA = ATAT =s A =-A'AZA!T = A(A? Il = Waa. or
ACA)! = CA=LA
(AM) = (ATA RT =
= Ata - Therefore, A7A™! is symmetric.
43. Since A is symmetric, we hav
e
[= AVA
=I =A =A7!AT ATL(PDP™!) = PDP")! = (PD'P"') [((P~')"DP
| "], since D? | = D. This result is true only
when P-'(P')’ = I, or P-! = p?.
44, ee and B be the orthogonal matrices, that is A~! = A? and
B-! = B7. Then (AB) =B7A"=B A =

45. A = AT gives AA’ = I. We obtain the conditions as

if tm tn; =), % = 1, 2, 3 and
lily + mymz + nn, = 0, LiL, + mym; + nn; = 0, Il; + MyM; + Aon; = 0.
46. Since A is an orthogonal matrix, we have A“! = A’. Hence, | A} |=|A?
|=|A|or1/| A |=
|A|=>|A|?=1or|A]=1.
47. Let A and p be two distinct eigenvalues and x, y be the corresponding eigenvectors. We have Ax = Ax
and Ay = jy. From the first equation, we get x’A’ =Ax!orx’A =A x". Postmultiplying by y, we obtain
x’ Ay = Ax’y. From the second equation, we get x’Ay = px’y. Subtracting the two results, we obtain
(A — p)x'y = 0, which gives x’y = 0 since A ¥ LL.
49, There exists an orthogonal matrix P such that P-'AP = D. Now, A = PDP™! = PDP’, since P is
orthogonal. We have Al= (PDP’)’ = PD’P’ = PDP’ =A, sincea diagonal matrix is always symmetric.
51. Let z= U+iV, where U #0, V #0 be real vectors. Then

Zz! Az = (U"AU + V7AV) + i(U7AV — V7AU) = UZAU + V7AV > 0

since UTAV = (UAV)! = V’A'U = V’AU.
52. Let the vectors a, b be transformed to vectors u, v respectively. Then

(u,v)=u'yv =?:v=(Aa)’ (Ab) =a'A’Ab=a"b=a-b.

53. xTAx= Lev nl alfeaSis) ee

= al(x, + bx,/a)” + x} (ac - b*)/a*] > 0 for all x, xp.

0.
Therefore, a > 0, ac - b* >
2 l 3x l

54. x’ Ax =[x,,%2>%3] a3). 4. =h*e

Si | ] 2 X43 |

= 2x? —2x,xX2 + 2%1%3 + 4x5 + 2x3 = (2X; —X2)° + (4, + 43)° + 3xz + x7 > 02
the eigenvalues are not positive.
55. All the leading minors are not positive. If can also be verified that all
mt 1
1+ 2i
0
=} =9}
|
-2<H
0)
=243i

0
i
60, Judi
()
tod
=3
24/
0)
2={

0)
'
 Chapter 3

Matrices and Eigenvalue Problems

3.1 Introduction

In modern mathematics, matrix theory occupies an important place and has applications in almost all
branches of engineering and physical sciences. Matrices of order m x n form a vector space and they
define linear transformations which map vector spaces consisting of vectors in IR” or C” into another
vector space consisting of vector in IR” or C” under a given set of rules of vector addition and scalar
multiplication. A matrix does not denote a number and no value can be assigned to it. The usual rules
of arithmetic operations do not hold for matrices. The rules defining the operations on matrices are
usually called its algebra. In this chapter we shall discuss the matrix algebra and its use in solving linear
system of algebraic equation Ax = b and solving the eigenvalue problem Ax = Ax.

3.2 Matrices
An m X n matrix is an arrangement of mn objects (not necessarily distinct) in m rows and nm columns
in the form |
4, A ** Ap
| ad a

A= — 22 7 ay, (3.1)

Any Gm2 "* Sn |

We say that the matrix is of order m xn (m by n). The objects a), 4,7, ..-,4m, are called the elements
of the matrix. Each element of the matrix can be a real or complex number or a function of one more
Variables or any other object. The element a, which is common to the ith row and the jth column is
Called its general element. The matrices are usually denoted by boldface uppercase letters A, B, C....
etc. When the order of the matrix is understood, we can simply write A = (a;;). If all the element of a
matrix are real, it is called a real matrix, whereas if one or more elements of a matrix are complex it
8 called a complex matrix. We define the following type of matrices.

Scanned by CamScanner
 ON
3.2. Engineering Mathematics
is, it has 0
ne row and 77 columns is called q ,, Wy"ct,
at
A matrix of order | *: 7 tha t is,
Row Vector
written 45
or a row matrix of order 7 and 1s ~» Of
[ay 42 - al
(ay, 412 °°" Ayn), OF
ent.
in which aj, (or a;) is the jth elem any m TOV d one column i
| . row s mn ts calledgCo|Hn
I, that is, it has 7
Column vector A matrix of order | m X d is written 4s
, or or a | column matrix of order mi an
vect
bij by
b, . b,
“" | or ;

Din Dm

|
ent.
in which b,, (or B,) is the jth elem ‘< C4 called i ordeJer.
r. The vectors are usually denote
as a in a row/ colu mn vect or IS led its
The number of elements elements arejey
case lett ers a, b, ¢, - __ etc. If a vector has 72 elements and all its
er
by boldface low=. 8 if one or more elements are comple,
d n-tuple in IR’, whereas
‘ i

then it is call ed an ord ere

numbers,
ered n-tuple.in C.
' il

num ber s, then it

*
is call ed an ord
matrix.
iS called a rectangular
Rectangular matrix <A matrix A of order m Xn, m#n
rows is equaltp
A mat rix A of orde r m xX in which m = 2”, that is number of
Square matrices @jjy. that is the elements
of col umn s Is call ed a squ are matrix of order n. The elements
the number these elements lie is called
+++ Ann AE call ed the dia gon al elements and the line on which
Ayy> Ao95 ts a,, whe n i #/ are calle
diagonal of the matrix. The elemen
the principal diagonal or the main called the trace
elem ents . The sum of the dia gon al elements of a square matrix is
the off-diagonal
of the matrix.
zero is called a ull mar
m x 7 in which all the elements are
Null matrix A matrix A of order
by 0.
or a zero matrix and is denoted
elements ay, FL zero|
Diagonal matri x A square matrix A in which all the off-diagonal
e
called a diagonal matrix. For exampl

I.
A= te is a diagonal matrix of order

0 ain

A diagonal matrix is denoted by D. It is also written as diag [a,) @o) --- Qml-
iX
that is a, = & for all /, then the
malt
If all the elements of a diagonal matrix of order n are equal,
‘s called a scalar matrix of order n. =
If all the elements of a diagonal matrix of order m are 1, then the matrix

cee di
Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.3

> I 1s called an unit matrix or an identity matrix of order

n.

An identity matrix is denoted by I.

Equal matrices Two matrices A = (4i))mxn and B = (b,),,, are said to be equal, when
(i) they are of the same order, that is m = P,n =q and
(ii) their corresponding elements are equal, that is a, = b, for all i, j.
Submatrix A matrix obtained by omitting some rows and or columns from a given matrix
A is
called a submartix of A. As a convention, the given matrix A is also taken as the submatri
x of A.

3.2.1 Matrix Algebra

The basic operations allowed on matrices are
(i) multiplication of a matrix by a scalar,
(ii) addition/subtraction of two matrices,
(iii) multiplication of two matrices.

Note that there is no concept of dividing a matrix by a matrix. Therefore, the operation A/B where
A and B are matrices is not defined.
Multiplication of a matrix by a scalar
Let a be a scalar (real or complex) and A = (a,,) be a given matrix of order m x n. Then
B= aA = Qa;) = (Qa;,;) for all 7 and /. (3.2)

The order of the new matrix B is same as that of the matrix A.

Addition/subtraction of two matrices

Let A = (a,) and B = (b,) be two matrices of the same order. Then
C=(¢,) =ats= (a;,) + (b;) = (a; + bi), for all i and/ (3.3a)

and D = (d;) = A- B = (a;) - (b;) = (ay — 5y), for all i and j. (3.3b)

The order of the new matrix C or D is the same as that of the matrices A and B. Matrices of the same
order are said to be conformable for addition/subtraction.
..., &, are any scalars,
IPA), Aa, ..., A,, are p matrices which are conformable for addition and @, 0,
then |
(3.4)
C= a,A, + @,A,+... + aA,
is called a linear combination of the matrices Aj, Ao, .... Ap. The order of the matrix C is same as
that FA P= TB. cesy D

Scanned by CamScanner
 os”
3.4 Engineering Mathematics

scalar multiplication
Properties of the matrix addition and

Let A, B, C be the matrices which are conformable for addition and a, B be SCalars Then

1 A+B=B+A. (commutative law)

2 (A +B)+C=A+(BtC) (associative law).
.
|

3 -At+O=A (0 is the null matrix of the same order as 4),

aa
a.
4. A+(-A)=0. 5. a(A + B) = aA + aB
6. (a + B)A = aA+ BA. 7. a (BA) = aBA.
=

8. 1xA=A and O0xA=0O.

Multiplication of two matrices

The product AB of two matrices A and B is defined only when the number of columns in 4 jg eau
to the number of rows in B. Such matrices are said to be conformable for multiplication. Let 4 = (a)y
be an m x n matrix and B = (5,) be ann x p matrix. Then the product matrix
——
ee
a

‘5 a
—=

2 ._=

np

a
ea
gs
an
aeeiat
m *n n * p

is given by
is a matrix of order m x p. The general element of the product matrix C
Nn
(3.5)
Cy = Qb,; + Qjyba; + ... + Aigdy = Dd Fix OK;
k=
t-multi plied- byatt
B..
In the product AB, B is said to be pre-multiplied by A or A is said to be pos
4S
7—————
+=:

If A is a row matrix of order 1 x n and B is a column matrix

of order n x 1, then AB Is?
Dag
ee
a
2S
a of order 1 x 1, that is a single element, and BA is a matrix of order n X n.
eee
Me
pl
~*~

Remark 1 -< defined bu! the

Is © Bis?
(a) It is possible that for two given matrices A and B, the product matrix AB
product matrix BA may not be defined. For example, 3 if A is 4 2 x Lea where®
3 x 4 matrix, then the product matrix AB is defined and is a matrix of order“ © ”
the product matrix BA is not defined. ii
(b) If both the product matrices AB and BA are defined, then both the matrice $e tative:
NE
a
Sa
=a
lle
ay
ee

square matrices. In general AB + BA. Thus, the matrix product is not commu
If_ AB = BA, then the matrices A and B are said to commute with each other ole. lel

(c) If AB = 0, then it_does not always imply that either A = 0 or B= 0. For exa™

A
Scanned by CamScanner
 "e ee
a
'
i, FF
ll

i -

f
j
Matrices and Ei . A
genvalue Proble I Ir KH
ms 3.5
x OQ] tT wale
}
a=|? and B~|° 4
ah

then AB =
0 0
“ ‘| and :
BA = Q 0

(d) If AB = AC ax + by “ me
, it does not
always imply
that B = C
(e) De fAi X= nA x A + XA (k times) Then, a matrj
integer & is saideto be nilpo ,
| .
a « ]
«

tent. The smallest

i ‘ TX A S rf

= 0 for so
me Positive
of nilpotency of the matrix A. alue of f k forsmo
val as g A k
which = 0 is called the in
(f) If A? = A, then A is called dex
an idempotent
Properties of matrix matrix.
multiplication
1. If A, B,

(AB)C = A(BC) (associative law)

is a matrix of order m x
q.
- If A is a matrix of
Order m x n and B, C
are matrices of order n
x p, then
A (B+ C)=AB+AC
(left distributive law).
- If A, B are matrices of
order m x n and C is a mat
rix of order n x p, then
(A+ B)C = AC + BC
(right distributive law).
- If A is a matrix of order m x
n and B is a matrix of order » x
DP, then
(AB) = A(a@B) = (@A)B
for any scalar a.

3.2.2 Some Special Matrices

We now define some special matric
es.
Transpose of a matrix The matrix obtained
by interchanging the correnponuiig rows and colu
mns
of agiven matrix A is called the transpose matrix of A and is denoted by
A’ or A’, that is, if

a); a\2 va Gy, 411 421 se Gaul

Pe a>, 4 sos My,
» : rs
Gas a
then A’ =],Al2 2 i

a Oy Giny dd) ay
In =! inyy

Scie
IfA is an m x m matrix, then A’ is an n x m matrix. Also, both the product matrices eeA7A and
AA T
“€ defined, and
A’A = (n X m\(mxXn)is ann xn square matrix

Scanned by CamScanner
: j

3.6 Engineering Mathematics

square matrix.
X m) 18 an m1 x mM
and AA!= (m X nt
sce Del's
Eg
= A column vector b can also be written as [b, 2
=3
===

The following results can be easily verified

ee
ee
ot

of a column Matty :
of a row matri x 1s a CO lumn matrix and the transpose
1. The trans pose
==
a row matrix.

:
ces for addition1.
A and B are con formable Seams
-a y™ *
B)’ = A’ + B’, when the matri
3. (A +
eee
6
ee 5=
©s=
—

B are conformable for multiplication.

oS
=ee
-P4, (AB)! = BA’, when the matrices A and
A, 1s defined, then
If the product A, A) ...
r gf T
see A}-
Ad Rey Ay
[A, A, as

Remark 2
The product of a row vector a; = (4;) 42 ---
a.) of order 1 x m and a column vector
by)" of order n x I is called the dot product.or the inner product of the vectors
a
:

b, = (by; bo ---
a; and b,, that 1s
nt

y=
i aj b= Dain bey
: k=l
t matrix C in Eq. (3.5) can be written as
which is a scalar. In terms of the inner products, the produc

a,b; aytby °° a,b,

36
a,b, a,b, st a,‘b,

‘b, a, "D> sila aD,

a i

6Se
ow,
ee
=—

Symmetric and skew-symmetric matrices A real square matrix A = (aj) is said to be

| . 2. oF
ifa,;=a, for all i andj, that is A= A
symmetric,
; , ‘ _ T
skew-symmetric, if a;=— a, for all i andj, thatis A=—A.

Remark 3
4t
———————

(a) In a skew-symmetric matrix A = (a,), all its diagonal elements are zero.
, f
|
i7 (b) The matrix which is both symmetric and skew-symmetric must be a null matrix.
| (c) For any real square matrix A, the matrix A + A’ isalways symmetric and the quer.
4
!
is always skew-symmetric. Therefore, a real square matrix A can be written as e

| a
symmetric matrix and a skew-symmetric matrix, That is

a AFA)
r,
+
1
S (A — A’).
T

1 = (),

A square matrix A = (a;,) is called a lower triangular matrix if dij

7

Triangular matrices angi”

whenever i <j, that is all elements above the principal diagonal are zero and an upper 7et0-
j a = a Tae = = : Fr

matrix if a,, = 0. whenever i > /, that is all the elements below the principal diagonal at

tl
Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.7

Conjugate matrix Let A = (a;;) be a complex matrix. Leta7,, denote the complex conjugate of ay.
Then, the matrix A= (a) 1s called the conjugate matrix of ‘A

Hermitian and skew-Hermitian matrices A complex matrix A is called an Hermitian matrix if

A== -(A)’, Sometimes, a
A= Mae eee or A*® n matrix if A = -- A’ or
(A)! and a skew-Hermitia
Hermitian matrix is denoted by A”

Remark 4
(a) If A is a real matrix, then an Hermitian matrix is same as a symmetric matrix and a
skew-Hermitian matrix is same as a skew-symmetric matrix.
Ab) Inan Hermitian matrix, all the diagonal elements arereal (let a, = x, + iy, then a, = a ,gives
x, + ty = Xj — ty; OF y; = 0 for all J).
9) In a kew-Hemitan matrix, all the diagonal elements are’ either 0 or pure imaginary
) (let a, = x; + ty,; then a, = — a i; givesx; + ty; = — (x; — iy,) or x; = 0 for all /).
(d) For any sina Square matrix A, the matrix A + A is always an Hermitian matrix and the
matrix A— A’ is always a skew-Hermitia
ee EADn matrix.
e e TAT Therefore, a complex square matrix A can
be written as the sum of an Hermitian matrix and a skew-Hermitian matrix, that is

Am (A+ A’)+ > (A— A’).

Example 3.1 Let A and B be two symmetric matrices of the same order. Show that the matrix AB
is symmetric if and only if AB = BA, that is the matrices A and B commute.
Solution Since the matrices A and B are symmetric, we have
A’=A and B’=B.
Let AB be symmetric. Then

(AB)’= AB, or B’A’=AB, or BA=AB.

Now, let AB = BA. Taking transpose on both sides, we get

(AB)’= (BA)’= A’B’ = AB

Hence, the result.

3.2.3 Determinants
With every square matrix A of order n, we associate a determinant of order n which is denoted by
det (A) or |A|. The determinant has a value and this value is real if the matrix A is real and may
be __
real orcomplex, if the matrix is complex. A determinant of order n is defined as

ayy Sy. "" Oy

| G2, 422 "* Gdn
det(A)=|A| =] -

Gn) 4n2 "" Gyy

Scanned by CamScanner
 3.8 Engineering Mathematics

n af : A
Y~ 1% a, My = L974
=| j=!
i] I]
i+) wi
dX (-1)"% a, My = * ai Aj
j=l /= (3.7)

where M,, and A, are the minors and cofactors of a,, respectively.
We give now some important properties of determinants.
1. If all the elements of a row (or column) are zero then the value of the determinant iS zen
2. |A| = |A”]. |
3. If any two rows (or columns) are interchanged, then the value of the determinan, i
——
ce.
oe
multiplied by (—1).
4. If the corresponding elements of two rows (or columns) are proportional
to each Other
then the value of the determinant is zero. |
5. If each element of a row (or column) is multiplied by a scalar @ then the value
Of the
determinant is multiplied by the scalar a. Therefore, if B is a factor of each element of a row
(or column), then this factor B can be taken out of the determinant.
Note that when we multiply a matrix by a scalar a, then every element of the matrix is
multiplied by a. Therefore, | @A| = a" |A| where A is a matrix of order n.
SL. If a non-zero constant multiple of the elements of some row (or column) is added to the
corresponding elements of some other row (or column), then the value of the determinant
remains unchanged.
4 |A + B| # |A] + |B], in general.

Remark 5
When the elements of the jth row are multiplied by a non-zero constant & and added to the
corresponding elements of the ith row, we denote this operation as R; — R, + kR,, where R, is the
ith row of |A|. The elements of the jth row remain unchanged whereas the elements of the ith row
get changed. This operation is called an elementary row operation. Similarly, the operation
C; — C; + kC, where C; is the ith column of |A|, is called the elementary column operation.

Se——
Ss
==
—=
Therefore, under elementary row (or column) operations, the value of a determinant is unchanged.
ie.
se
ee
a
Sg
a—
eee
Product of two determinants

If A and B are two square matrices of the same order, then

|AB| = |A| |BI.

Since |A| = |A’|, we can multiply two determinants in any one of the following ways
-—s
<a
,.

—.-.
.eiicdiiiesicce
==
—_—_. (1) row by row, (11) column by column,
(iii) row by column, (iv) column by row.
The value of the determinant is same in each case.

Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.9

gank of a matrix

The rank of a matrix A, denoted by r or r(A) is the order of the largest non-zero minor of |Al.
therefore. the rank of a mat is the largest value of r, for which there exists at least one r X r=
gubmartix of A whose determinant = not zero. Thus, for an m x n matrix r < min (m,n). For a square
atrix
e A 0f order
ae rank of matrix‘ae is 0, then it| must be a null matrix.
n, the rank r= n if |A| #0, otherwise r < y. The rank ofa null matrix is zero and

example 3.2 Find all values of 4 for which rank of the matrix

u-l O 0O
i 0 ue — -l = «(0
0 0 yp -l
“6 lil-=6 I

is equal to 3.
is at least
Solution Since the matrix A is of order 4, r(A) < 4. Now, r(A) = 3, if |A| = 0 and there
one submatrix of order 3 whose determinant is not zero. Expanding the determinant through the
elements of first row, we get
al 0 0 -l O
AJ=H}O pw -l}+) 0 Bw -I = u[utu-9 +11]-0
11 -6 1 -6 -6 1

= pp - 6 + Wp - 6 = (MIU - 2)(H - 3).

3, the determinant of the leading third order
Setting |A| = 0, we obtain pw = 1, 2, 3. For p= |, 2,
submatrix
pw -l O
JA|=]0 -l)=# #0
0 0 Ly
of pl, r(A) = 4.
Hence, r(A) = 3, when pf = 1 or 2 or 3. For other values

3.2.4 Inverse of a Square Matrix

Let A = (a;,) be a square matrix of order n. Then, A is called a

(i) singular matrix if |A| = 0,
il cals ; -
(") non-singular matrix if |A| # 0. ) <n and non-singular if
ea
its rank
In ot ; 2 ; as singul fj nk r(A ;
ar if its ra said to be invertible, if there exists a
mr words, a square matrix of seinrchearmiie 5 n is
order
nga n. A square non-singular matrix A of
“n-singular square matrix B of order ”” such that (3.8)
AB = BA =!

Scanned by CamScanner
 os”
3.10 Engineering Mathematics

where I is an identity matrix of order n. The matrix B is called the inverse matrix of 4 and
write B= A™' or A = B". Hence, we say that A’ is the inverse of the matrix A, if mt
A'A=AATEL
(3.9)
The inverse, A' of the matrix A is given by

l
A'= 1A adj(A) (3.10)
where adj(A) adjoint matrix of A
transpose of the matrix of cofactors of A.
Remark 6

(a) (AB)'= BA.

We have

(AB)(AB) |=I.
Pre-multiplying both sides first by A”! and then by B™! we obtain
B-'A~'(AB)(AB)! = B7'(A7'A) B(AB) | = B'A™ or (AB) = B'A™.
ee
Nt
———
ilial
In general, we have (A,A,... A,)' = A, Aj.) --- Al
(b) If A and B are non-singular matrices, then AB is also a non-singular matrix.
(c) If AB = 0 and A is a non-singular matrix, then B must be null matrix, since AB = 0 can be
pre-multiplied by A. If B is non-singular matrix, then A must be a‘null matrix, since AB = 0 can
be post-multiplied by B™’.
(d) If AB = AC and A is a non-singular matrix, then B = C (see Remark 1(d)).
(e) (A+By'#A' + B', in general.
Properties of inverse martices
Se
|
ce
a
ee

* 1. If A ‘exists, then it is unique.

2 (Ay = A
3. (A7)' = (A')". (From (AA™')’ = I’ = I, we get (A')'A’ = I. Hence, the result).
LE | dis) d,, = 0. Then, Dp" — diag (1/d,,, I/d55, one 1/d,,): '
wv. Let D = diag (di, ds,

5, The inverse of a non-singular upper or lower triangular matrix is respectively an UPPS o

lower triangular matrix.
6. The inverse of a non-singular symmetric matrix is a symmetric matrix.
ZZ 1. (A) = A" for any positive integer n.

A
Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.11

2 0 -

Example 3.3. Show that the matrix A=]|5 1 0] satisfies the matrix equation A® — 6A* +
0 | 4]

1A-1= 0 where I is an identity matrix of order 3. Hence, find the matrix (i) A’! and (ii) A.

solution We have
2 0 -1//2 oO -1 4 =] <5
A*=/15 1 OS 1 Ol=]15 1 -5
01 31/0 1 3 5 4 9

4 -1 -5][/2 0 -1 3 -6 -19
A?=A7A=|15 1-5/5 1 0/=|35 -4 -30].
5 4 9ffo 1 3{ [30 13° 22
— I, we get
Substituting in B = A® — 6A’ + 11A
3 -6 -19] [24 -6 =30] [22 Oo -I1] [! 9% 9
p=|36 -—4 =30/-1/90 6 =30]+/55 11 O/-]0 TF 0
30 13 22] |30 24 54 0 11 33] |o 0 1

0 0 0
=|0 0 0|/=0.
0 0 0

by A”, we get
(i) Premultiplying A? — 6A? + 11A - I = 6A0 7A
A7A3 — ? + 1LATA — AT = 0

or AT'= A?- 6A + 111

1 Oo 0 5 -| |
4, =f —5 12 0 -6
5 6 SI.
=|15 1 -5|/-|30 6 O;+/;0I o/=|-1
18 0 0 $ -2 2
5 4 9 06

4 wl l 3 «1 | eS of} 10
6-515 6 -5|=|-160 61 —55}.
(ii) A? = (ae =| -15
5 2 2) . 55 -21 19
5 2 2
;
We can also write
IA).
a2 =(A1) (A!) =A- 61+

Scanned by CamScanner
 3.12 Engincas
eering Mathema fics ‘
m of Equation s
3.2.5 Solution of m x Lin ear Sys te
*npe

: ‘no wns
unknown
aquations In
Consider the system of # equa @ hidie™_ b
ay;X, + 412*2 a 52 “3
+ GoyXn 72
ityXj Bate Te

,
+ 4,,9%2 + ,..
+ Ayaytn
nae an b,,. (3,] )
Ay |X

. of equa tions (3.11) as

In matrix form, we can write the system
~
Ax=b (3.12)
bh, | xy
ay 42 «+ An
x
where | A= 7) @n ‘ = 2. |? com |“

aiy| a2 Qn b, x

and A, b, x are respectively called the coefficient matrix, the right hand
side column vector and the
solution vector.If b # 0, that is, at least one of the elements b,, 55, ..., b,, is not
zero, then the system
of equations is called non-homogeneous. If b =
0, then the system of equations js called
homogeneous. The system of equations is called consis
tent if it has at least one solution and
inconsistent if it has no solution.
Non-homogeneous system of equa
tions
The non-homogeneous system of
equations Ax = b can be solved
by the following methods.
Matrix meth od
let A be non-singular
. Pre-multiplying Ax
=hb by ant we obt
ain
x=A'p,
The system of equations (3.13)
is
is the only solution,

Cramer’s ry le

Let A be non-si
ngular, The Cram
er’s rule for
the solution of
Ax = b js give
n by
Hera |A,l|laa
ee ony
Where |A_| is the
determinant of th alm
hand side column vector b.
e
We discuss the pee eained by replacing the ith
following Cases column of A by the ret

Scanned by CamScanner
 i f £ & & & OSS

Matrices and Eigenvalue Problems 3.13

Case 1 When |A| # 0, the system of equations is consistent and the unique solution is obtained by
ysing Eq. (3.14).
Case 2 When |A| = 0 and one or more of |A,|, i = 1, 2, ..., m, are not zero, then the system of
equations has no solution, that is the system is inconsistent.
Case 3 When |A| = 0 and all |A,| = 0, 7 = 1, 2, ..., n, then the system of equations is consistent
and has infinite number of solutions. The system of equations has at least a one-parameter family of
solutions.
ons
Homogeneous system of equati
of equations
Consider the homogeneous system
Ax = 0. (3.15)
|
Trivial solution x = 0 is always a solution of this system.
If A is non-singular, then again x = A 0 = 0 is the solution.

Therefore, a homogeneous system of equations is always consistent. We conclude that non-trivial

solutions for Ax = 0 exist if and only if A is singular. In this case, the homogeneous system of
equations has infinite number of solutions.
Example 3.4 Show that the system of equations
1 -l 1 ||x 4
2 1 -3]}¥y}= | 0}.
l ] l 2

has a unique solution. Solve this system using (i) matrix method, (ii) Cramer’s rule.
Solution We find that |
—_ ae
AJ=|2 1 3) = 101+ 3)-2C-1-1) + 1-1) =10 #0.
‘ct ot, 4
Therefore, the coefficient matrix A is non-singular and the given system of equations has a unique
solution. Let x = [x, y, z]’.

4 2 2 4
(i) We obtain At = —|-s 0 5| and
b=| 0}.
| —2 3 2

[4 2 2}4 2
Therefore, x=A™'b == -5 0 5) 0;/=)/-1
1 -—2 34/2

Hence, x=2,y=-l and

z = l.

Scanned by CamScanner
 3.14 Engineering Mathematics

4-1 , wa)
| edt #3) = 8 $28 =
_
Dee
“s
(ii) We1 have |A,| = 0
:
2 1 |

1% 2 + 1(-12
—- 110 + 6) — 2(4— 2) Ile -— -—0
0) = - |
A= [2 9 3] =104+ -% .
12

[ <4 4
IAJ=[2 1 0] = 12-0) - 2 2-4) + MO — 4) = 10.
| 2

Ay] es — WAglier yg 2 TAal

Therefore, pea ee AE Pe |A|

Example 3.5 Show that the system of equations

2 3 IWxl=]2
3 2 4] x, 2

has infinite number of solutions. Hence, find the solutions. |

Solutions We find that |

l= 3 3 -1 3
Al=(2 3 1/=0, jJAJ=|2. 3 = 0,
24 5 24

1 3 3 { Jy 3
|A,|
A, =
aon be=—*
|JAJ=]/2 3 2) =0.
2 4 325
Therefore, the systen nti .
ystem of equations has infinite number of
solutions.
| Using
. Using the
th first tw equation
+1 ~ t= 3~ By,

and solving, we obtain x, =(1] -

off the third equation,
satisfies 10x35 and y. = ae ‘ soluti?of
oe ys (Sx; — 4)/5 where x, is arbitrary. Thi

Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.15
Example 3.6 Show that the system of equations

4 9 31x é

2 3 Iiel=|2
2 6 2] 2 7

is inconsistent.

Solution We find that

69 3 . 3
12 2
Al=}2 63 zZ1/=0, IAl=|27632 1] =0, |Al=l2me7 1|=6
2

Since |A| = 0 and |A,| # 0, the system of equations is inconsistent.

Example 3.7 Solve the homogeneous system of equations

[| 2g 3s bet 0
2 3 -2})y)}=
10).
4 7 4 | 0

Solution We find that |A| = 0. Hence, the given system has infinite number of solutions. Solving
the first two equations

1 2/|x| |-32z
2 3ily 2z t
we obtain x = 13z, y = — 8z where z is arbitrary. This solution satisfies the third equation. +t

Exercise 3.1 \
I: 2 4 22 1
1. Given the matrices A=|-1 1 1], B=|3 0 —I], verify that
2 1 35 lL 1 =

(i) |AB|
=|A||B), (i1) |A+ B| #|A| + |B.
2. If A’=[1, -5, 7], B = [3, 1, 2], verify that (AB)!= B’A”
| 2 2Z
3. Show that the matrix A=|2 1 2| satisfies the matrix equation AY — 4A - 51 = 0. Hence, find AT,
z. 2 a

4. Show that the matrix A = | 2 —3) satisfies the matrix equation A>-6A?+5A + 111 =0. Hence,
tJ -| 3

find A’!.

Scanned by CamScanner
gt es

| 3.16 Engineering Mathematics

0 | -i
A=|4 -3 4], verify that
t 5. For the matrix
3 =a a
¥
(ii) {adj(AT —_ dj(A').
. (ii)adj (A!” =adj(A ‘hi
mn |
| ji
:
6. For the matrix A=|-2 3. 1), verify that
} 13
|
(A),
(i) (A=) Gi) (A) =
2-1 1 i #1
3 4 , verify that
1 7. For the matricesA= |—-1 2 —-1],B=|1
i « 2 209
i
(i) adj (AB) =adj(A)adj(B), (ii) (A+BY'# A+B".
non-singular matrix A=
8. For any nor (a;;) of order n, show that

(i))lady ((A)|=JAI",3 (ii)adj (adj(A)) =|Al"”

' = 1/|Al.
9. For any ‘non-singular matrix A, show that |A_
symmetric, wher e B is any matrix for which the
| 10. For any symmetric matrix A, show that BAB’ is
a | productmatrix BAB’ is defined.
11. If A is a symmetric matrix, prove that (B a-')7(A7
'B’)-! =I where B is any matrix for which th
product matrices are defined.
”=
Fe

12. If A and B are symmetric matrices, then prove that

(i) A+ Bis symmetric, (ii) AA” and A’A are both symmetric,
(iii) AB — BA is skew-symmetric. ae ee
ei
ee

then ie that
If A and B are non-singular commutative and symmetric matrices,
ee
of
—

| 13.
| (i) AB . (11) A~ 'B, (11) A’ IB!

| are symmetric.
14. Let A be a non-singular matrix. Show that
| (i) if1+A+A?+...+A"=0, then A'=A’,
jeu 2 2
"A"=0, then A™'=(-1)""'A" leh
| (ii) if I-A+A?-...+(-1)
gular square matrices of order n and PAQ = I, then show that
} 715. Let P,Q and A be non-sin
16. If 1— A is a non-singular _ then show that
A)'=I1+A+A?+
| assuming that the series on the age hand side converges. igiA:
—<—e
17. For any three non-singular matrices A, B, C, each of order n, show that (ABC)

Scanned by CamScanner
 ue fe"
Matrices and Eigenval

:
em of equation
th e following syst xr+y-z=0.
Solve x + 3 y - z=5,
s
18 yoy tar 92 = 14.
3x + 2y +
x+4yt+ z=7,
432 = 6
tz=3, -xt+3y+4z=6.
19, x2
20 _xytyt2z=2 Bx-y
yr+3z=5.
y. 2-zeh sxty=7,
which the system of equations
2 Determine the values of k for
0
x—ky+z=0, kx + 3y — kz = 0, 3x+y—-z=
.
al solution, (ii) non-trivial solution
has (i) only trivi
e of @ for which the sys
tem of equations
nd th e va lu ial solution.
5x+3y-z=O0hasa non-triv
_ Fi
3x+ 2 (cos 20)y +3z = 0,
a(sin Qx+Y—22=0, c, are non-zero
+y +bz=0,cx+cy+Z= 0, where a, b,
ons x + ay + az = 0, bx
If the system of equati t
ivial solution, then show tha
and non-unity, has a non-tr
a b Cc
+—— + am]
l-a 1-b Il-ce

ch the system of equations

Find the values of A and } for whi
x+ 4y + 3z= 10, x+4y+Az=H
x+2y+z=6,
number of solution, (iii) no solution.
has (i) a unique solution, (ii) infinite
A is given by
Find the rank of the matrix A, where

21 -l
7,|-1 -3 4 28.
xl? ** 6 8 i. od 2
35 7 1
5 4 -5

4 0 gq —- &
111 , 15
2 5 3/m] ° : by
~1p q ri s@f-1 C2
3 29 -1 |
pgr —Ay —b, —C45 0
q

is a Skew-Hermitian matrix and if A is a Skew-

31. Prove that if A is an Hermitian matrix then iA
Hermitian matrix, then iA is a Hermitian martix.
+ A is invertible.
32. Prove that if A is a real matrix and A” — 0 asn > ™, then I
33, Let A, B be n x n real matrices. Then, show that
for any scalars a and f.
(i) Trace (@A + BB) = a Trace (A) + B Trace (B)
(ii) Trace (AB) = Trace (BA), _(iii)(AB - BA = I is never true. _
34, IfB. C are n x n matrices, A = B + C, BC = CB and C’ = 0, then show that
p.
A’*! — BP [B + (p + 1) C] for any positive integer
35. Let A = (a;;) be a square matrix of order n, such that aj
=d,i#janda,;=c,i=jJ. Then, show that
Al = (c — d)""' [c + (n— 1) d].

Scanned by CamScanner
 “7
. i ne

3.18 Engineering Mathematics

rmitian, skew-Hermitian
skew-symmetric, Herm OF Rone of these
Identity the following matrices as symmetric, fm dt -o
welag 8
a be - ;

bk @ 2 :
\ 1a
al
i |
37,|h d e|-
| ee
36.}
| i

i A |-2 5 44.
id ae oy
J <4 & ee g
|
I-i 0 24a 1,
1 24+4i I-i | 244i
| | 41.)-2+47 0 3-5
ws laeae <S aaS|. Ma l-2eae “8 3)
6 -(-t 3-8 © J A-# 3-3 0
| I+; 3+5i
] = j
0 _j I4i
Ogi a l=i\|,
43.| -i -2% O |. 44.|-l
42. |i O07 —j l+r 2
_|+i 0 I
i i 0

:
; 1 2 -i
45. |-2i i |
i i 1 @

, 3.3. Vector Spaces

or some other
a non- empt y set of certa in objec ts, whic h may be vectors, matrices, functions
Let V be V are denoted by
: is an element of V and is called
a vecto r. The elements of
objects. Each object
aic operations
a, b, c, u, v, etc. Assume that the two algebr
tion
(i) vector addition and (ii) scalar multiplica
ea i.
“SF
oF
se
—_
are defined on elements of V.
then
| If the vector addition is defined as the usual addition of vectors,
a + b= (a, 4, «+55 a,,) + (b,, 55, ana 5.) = (a, + b,, a> Te bo, ciavg hy + b,,)-
' | |
e , o,thet
ee of a vector by the scaler
' If the scalar multiplication is defined as the usual scalar multiplication
|
aa = a(a), 49, ..-, 4,) = (Ha, az,’..., Aa,).
Jow ine
fol
ts a, b, c in V and any scalars a, B the
|

V define s a vector space

i
if for any elemen
|
The set
properties (axioms) are satisfied.
Properties (axioms) with respect to vector addition
| 1. at+bisin V.
:) 2 atb=bta. (commutative law)

| | 3. (a+b) +tc=art (b +c). (associative law)

4.0at+ 070+ ama. (existence of a unique zero element in V)

Fe
(existence
in V)
of additive inverse or negative veCt! in
5. a+ (—a) = 0.

- — a
Scanned by CamScanner
 SL) ~™]) EZ" °°
lV / l/l

Matrices and Eigenvalue Problems 3. 19

plication
properties (axioms) with respect to scalar multi
6, aa is in V.
7, (a+ B) a= oa + Ba. (left distributive law)

g. (aB)a = =(Ba).
9, a(a t+ b) = wat ab. (right distributive law)
10. 1a = (existence of multiplicative identity)
ies. When these two properties are
The properties defined in 1 and 6 are called the closure propert
scalar multiplication.
catisfi ed, we say
that the vector space is closed under the vector addition and
not always be the usual addition
The vector addition and scalar multiplication defined above need but also
and multiplication operators. Thus, thevector space depends not only on the set V of vectors,
on on V.
on the definition of vector addition and scalar multiplicati
B are real
if the elements 0 f V are real,
then it is called a real vector space when the scalars a,
numbers, whereas V is called a comple
x vector space, if the elements of V are complex and the scalars
x
a, B may be real o r complex numbe
rs or if the elements of V are real and the scalars a, B are comple
numbers.

Remark 7
ied, then V is not a vector space. We usually
(a) If even one of the above properties is not satisf
ng the other properties.
check the closure properties first before checki
oduct etc. are not part of the properties to be
(b) The concepts of length, dot product, vector pr
satisfied.
| (c) The set of real numbers and complex numbers are
called fields of scalars. We shall consider
course on linear algebra, vector
| vector space only on the fields of scalars. In an advanced
d.
| spaces over arbitrary fields are considere
l tor space.
(d) The vector space V = {0} is called a trivia vec
ions of vector addition and
The following are some examples of vector spaces under the usual operat
scalar multiplication.

1. The set V of real or complex numbers.

any closed interval [a, b]. The 0 vector
2. The set of real valued continuous functions f on
defined in property 4 is the zero function.
3. The set of polynomials P,, of degree less than or equal to n.
4. The set V of n-tuples in IR” or C’.
rty 4 is the null matrix of
5. The set V of all m x n matrices. The element 0 defined in prope
Order m x n.
usual operations of
™ following are some examples which are not vector spaces. Assume that
‘ctor addition and scalar multiplication are being used.
two polynomials of degree
l. The set V of-all polynomials of degree n. Let P, and Q, be
n and thus may not be in V.
nin V. Then, aP,, + BQ, need not be a polynomial of degree
xe + ax" +... + Gy, and QO, = en oF bx" © age © Bins
For example, if P.. =

then P.+@Q, isa polynomial of degree (n — 1).

Scanned by CamScanner
 3.20 Engineering Mathematics
continuous on he
atang of one variable x, defined and
function atc, 4 <c<b, 1s some non-zero COnstan
2. The set V of all real-v alued scones
val
interval [a, b] such that the ) =p. Since flo) v0.
let f (x) and § (x) ‘e ma elements in V. Now, f(c) = g(c e,
For example, spac
then V forms a vector
=Ss
—=—s
-a
—-=

V. Not e that if p = 0,
==
|a

2p, f(x) + g(x) is not in

=ee
pee
2.

‘th real
: coefficients, of : degree n, where ad dition
Example 3.8 Let V be the set f all olynomials, Wi
Show that V is not a vector spac e,
fi“ - a+ ‘=fab wnause vail scalar multiplication. (P,,)
i ‘d
18 delin
0,= (Q,,) is a polynomial of d— |
Now, P,+
Solution let P, and Q, be two elemen ts in V.
e a vector space.
which is not in V. Therefore, V does not defin

Example 3.9 be the set of all ordered pairs (x, y), Where x, y are real numbers.
Let V
ee
ee
ee
el
a
eae
a Let a = (x, y,) and b = (xp, yy) be two elements in V. Define the addition as
a + b = (x1, y1) + (X25 Y2) = (2%, — 3X2, Yi — V2)

=ocom
= ee
eee and the scalar multiplication as
(x1, ¥y) = (@x,/3, ay,/3).
Show that V is not a vector space.
ee
Toa
ee

Remark
Solution We illustrate the properties that are not satisfied.
To show
Gi) (Xp, Y2) + Oy) = (2x — 3x1, y— Yi) # Xs Yy) + Xp, yo).
Therefore, property 2 (commutative law) does as given
ee
a
a
not hold. and scal
(ii) (CX), yy) + (, y»)) + (X3,
3) = (2x, — 3X2, Yj — y2) +
(%3, 3) these pr.
We need
(1, Yi) + (Cea, Y2) + (3, 93) = Oy, Mverse ;
y,) + (2%, ~3x3, yp — ys)
| Conside,
7 (2%, — 6x, + 9x,, Yi — Y2
Therefore » property 3 (associa + ys).
tive law) does not
Sati sfie d be]
Hence, V is not a vector Space,
|
Example 3.10 Let V be the set of al]
Ordered pai
and b = (x5, y») be two elements in V De Pars (x, y),
fine the addition Where x, y are real number, Let a= (tv!
as
a+b = (x,
Vi) + (x, y,) +
and the scalar multip
lication as = (1%, y)y>)

|
is not a vector space
(1,4) = (ax,| ay).
Show that V
Solution Note that
}
(1, 1) is an ele
ment of y.
|
From the given definition of
vector addition, ¥° *
Thi s ed
isli
dtin trunpe enonly for the ele
icey 4 me
|
Rut (1, 1), Ther efore vethe
y
s eleCle
, ment (1, 1)
» 1) plays the role of 0¢ Jemen Ee

Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.21

Now, there exists the element (I/x,, 1/y,) such that (x, vy) + (l/x, + I/y,) = (1, 1).
The element
(1/x,, /y1) Plays the role of additive inverse.
;
Therefore, property 5 is satisfied.
Now, let a = 1, B = 2 be any two scalars. We have
(a + B)(x), y= 3(x,, y)) = (3x, 3y,)

and a(x), ¥)) + BCX, ¥1:) = 104, y) + 2(X1, V1) = (xy, ¥y) + (2x, 2yy) = (2x7, 2y%).
Therefore, (@ + B)(xX), ¥\) # A(x), ¥,) + Blx,, y,) and property 7 is not satisfied.

Similarly, it can be shown that property 9 is not satisfied. Hence, V is not a vector space.

3.3.1 Subspaces
Let V be an arbitrary vector space defined under a given vector addition and scalar multiplication.
A non-empty subset W of V, such that W is also a vector space under the same two operations of
vector addition and scalar multiplication, is called a subspace of V. Thus, W is also closed under the
two given algebraic operations on V. As a convention, the vector space V is also taken as a subspace
of V.
Remark 8

To show that W is a subspace of a vector space J, it is not necessary to verify all the 10 properties
as given in section 3.3. If it is shown that W is closed under the given definition of vector addition
and scalar multiplication, then the properties 2, 3, 7, 8,9 and 10 are automatically satisfied because
these properties are valid for all elements in V and hence are also valid for all elements in W. Thus,
we need to verify the remaining properties, that is, the existence of the zero element and the additive
1

i
|
|

inverse in W.
f
4

,|
Consider the following examples:
and scalar multiplication. \
1. Let V be the set of n-tuples (x), x2, ... x,) in IR” with usual addition
Then
0 is a subspace of V. |
(i) W consisting of n-tuples (x), X2, «++» X,) with x) =
V, since W is
ii) W consisting of n-tuples (x, 2, -- ., X,) with x, 20 is not a subspace of Sas it W).
(ii) g Pp (ax, when ais a negative real number, 1s not : in
not closed under scalar multi plica tion
of n-tu ples (x, X2, «++» Xn) With x, =x) + } is not a subspace of V, since W
(iii) W consisting
is not closed under addition.
y, =y,+ | be two
| and y =(), Ya sa,,)With
(Let x = (x), Xp, «++» Xn) with x, =x, +
elements in W. Then
x ty = (x) ty p%2 t Va ee Xr + y,)
+2#x,+y,+ 1).
is not in W as x5 + y2 =X TH and scalar
polyn omial s P of degre e < m with usual addition
2. Let V be the set of all real
multiplication. Then
of deg ree < m wit 0 is a subspace of V.
ith P(0) =01
ynomia ls
(i) W consisting of all real pol

Scanned by CamScanner
 ~
3.22 Engineering Mathematics

with P(0)= 1 is nota gy |

(ii) W consisting of all real polynomials of degree < m P+ W. SPace of
- ddition (if Pand : QO € W, thenisn
W is not closed under addaitro
O¢
V. since
Coefficients
(iii) W consisting of all polynomials of degree < m with real positive(if P is ap :
subspace of V since W is not closed under scalar
multiplication Clemen; ta
W,then—P¢€ W).
matrices with usual matrix addition and
3. Let V be the set of all m x m real square SCalar

multiplication. Then
metric matrices of order 7 is a subspace
, ofy
(i) W consisting of all symmetric/skew-sym
of order 7 is a subspace of y
(ii) W consisting of all upper/lower triangular matrices
(iii) W consisting ofalln xn matrices having real positive elements IS NOt a subspace ofy
since W is not closed under scalar multiplication (if A is an element of W
then—Aé WW).
4. Let V be the set of al 1 » x n complex matrices with usual matrix addition and scala
multiplication. Then
a vector space when scalar
(i) W consisting of all Hermitian matrices of order n forms
scalars are complex numbers
are real numbers and does not form a vector space when
(W is not closed under scalar multiplication).
a xX+I
Let A= ( | ‘i EW.
x-ty) b

; al xi-y
Let a=i. Weget aA=iA= ¢ W.
7 xi+y bi
r space whein
ces of order m forms a vecto
| (ii) W consisting of all skew-Hermitian matri
not form a vector space when scalars are com
p i

scalars are real numbers and does

numbers.
f of xt+iy
Let A= e W.
—x+iy i
-aAi

Let a=i. We get ia=( . i *)ew

The sum ©
that FO G = {0}.
Example 3.11 Let F and G be subspaces of a vector space V such
F and G is written as F + G and is defined by
F+trGe= {if+ge:fe Fr, ge G}.

definition of vector additi

on
of V ass umi ng the usu al r
space
Show that F + G is a sub
multiplication.
€ F, ge G. Since 0 € F, and0 € G we have 0+
Solution Let W=F +O and f ¢ F and g,, g, € G. Then
Let f, + g, and f, + 82 belong to W where f,, f,

Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.23

(f, + 2) + (f, + 82) = (f, + fh) + (g, + g,)6 F+G=wW.

a(ft+gy=-aftage F+G=wW.
scalar,
for 4 ny
Als0- w=F+Gisa subspace of V.
therefore.
nov state an important result on subspaces.
We now
rem 3.1 Let Vj, V2. +++» ¥, be any r elements of a vector space V under usual vector addition
ad scalar multiplication. Then, the set of all linear combinations of these elements, that is the set
of alls elemen ts of the form
and
av, + Q5V5 Tt vee a .Y, (3.16)

Spanning set Let S be a subset of a vector space V and suppose that every element in V can be
obtained as a linear combination of the elements taken from S. Then S is said to be the spanning set
for V. We also say that S spans V.
|
| Example 3.12 Let V be the vector space of all 2 x 2 real matrices. Show that the sets

sf le atl ole a)
‘ Ce) —)

spanV
a

| b
Solution Let x = ? i be an arbitrary element of /.
| c
(i) We write

edna eels elec aleele sh

Since every element of V can be written as a linear combination of the elements of S, the set
S spans the vector space V.
(i!) We need to determine the scalars Ql, ly, O13, 4 SO that

eared deaf olrel deeb J

Equating the corresponding elements, we obtain the system of equations

Q+a,=c, a =d.

—=
em
SS
Oa
Se
SS

Scanned by CamScanner
 3.24 Engineering Mathematics
jons 1S
The solution of this system of equal h=c, @ =a- b.
4 ee d.4
a,= a,=c7-4,
e
a os

=z
al
al'=wi
=a
Therefore, we can write
1 0 _ +e-d) | )
1 90 Lif
p *)=10H 5]£O- Ao 0
jc

ination of | the elements of Sth

very element| of V can bepwwritten as a linear comb
Sisince ever |
set S spans the vector space ¥
Example 3.13 Let V be the vector space of all polynomials of degree < 3. Determine whethero |

es
a
SS
oT
not the set -
s={o,P+,e ttt i} Lae
spans V ? ympart
Solution Let P(t) = at? + Br> + yt+ 6 be an arbitrary element in V. We need to find whether : OSE
b | tens
not there exist scalars a,, aj, a; such that oan ants

a+ Bet y+ b=a, + a(t? +) + a(t? +14 1) es:

ar + Br + yt + d= (a, + a;)t° + ayt? + (a, + a3) t + ay.
rc pate.. .
+s
Comparing the coefficients of various powers of t, we get
ee
ee
ee

a, +a,= 4, a, = B, a,+a,=¥, a,=0

The solution of the first three equations

is given by
_ a=-atBp-y, a, = Bp, a,=y-fp.

a
feel “oea Me potain Y~ B = 8, which may not be true for all elements in!
be written as a then salen ofiene not satisfy this condition and therefore, it cant”
space V, tements of §. Therefore, S does not span the veclor

3.3.2 Linear Independence

of Vectors
Let V be a vector Spa
ce. A fjnite set
dependent if there exist
ex; 5 IViy Vo,
scalars 1, Ob,| ..., 2 “» Yn} Of the elements of Vis said to be Hinea"?
Oy,
not all zero, such that
If Eq. (3.17) is Satisf} ed only
TV. + ay, 9 + Peo as | 1)
for cr, av, =U, 3
independent, = Q, = 1. = _
: ly
a, 0, then the set of vectors is said to be [ineah.
The above defi
nition of linear
de Pendence of y
Theorem 3,2 Th
e s et of ve
Clors {y,, Mae
I» Va, " ; Ws:
element of the set Yn Can be Written alternately as follo
Mbination of the engine dependent
if and only if at least
“ng element Ss.

Scanned by CamScanner
_ ae

Matrices and Eigenvalue Problems 3.25

7 | | | is | seit
remark
3.17) gives a homogeneous system of algebraic equations. Non-trivial solutions exist if
dependent in this case. If the det(coefficient
‘ efficient matrix) = 0, that is the vectors are linearly
@, = 0 and the vectors are linearly independent.
yer #0, then by Cramer's rule, @, = @ =... =
Let vy) ~ (1, -1,ant0), v; = (0, 1, -1) and v, = (0, 0, 1) be elements of IR>. Show that
Exam ple 3.14 :
{v,, V3, ¥3} is linearly independent.
the set of vectors

Solution We consider the vector equation

+ QV, + Vv; = 0.
OV,

V2, V3, We obtain

Substituting for V),
- 1) + a,(0, 0, 1) = 0
aj(1, -1, 0) + @,(0, 1,
— a, + &, — A+ a) = 0.
(Q,
or

-a@+ a, = 9. The solution of these equations

Comparing, We obtain a, = 0, — a, + a& = 0 and
independent.
is Q = O& = 03 = 0. Therefore, the given set of vectors is linearly

Alternative
1 0 0
det(v,, V.,V3)=|-1 1 OO) =1#0.
0 -1 1

Therefore, the given vectors are linearly independent.

3) be elements
Example 3.15 Let v, = (1, —1, 0), v2 = (0, 1, —1), ¥3 = (0, 2, 1) and v4 = (1, 0,
of IR?. Show that the set of vectors {v, V2 V3 V4} is linearly dependent.
@, Q, A, &%,
Solution The given set of elements will be linearly dependent if there exist scalars
not all zero, such that
+. O4V4 = 0. (3.18)
ayy) 1 OnV9 + O4V3

Substituting for v,, V>, V3, V4 and comparing, we obtain

The solution of this system of equations is

Q,=- A, = 5a,/3, A= - 4a,/3, @, arbitrary.

Substituting in Eq. (3.18) and cancelling 04, we obtain

5 4
ed “ mS v,+ v,=0.

e, the set of vectors

Hence, there exist scalars not all zero, such that Eq. (3.18) is satisfied. Therefor
'S linearly dependent.

Scanned by CamScanner
3.26 Engineering Mathematics

3.3.3. Dimension and Basis

Let V be a vector space. If for some positive integer n, there existsaset 5 ofn linearly Indep
elements of V and if every set of n + | or more elements in V’ Is linearly
dependent. then i
to have dimension n. Then, we write dim (V) =n. Thus, the maximum number
of linearly indep “
elements of V is the dimension of V’. The set S of # linearly independent vectors jg calle
basis of V. Note that a vector space whose only element is zero has dimension
zero. ™
Theorem 3.3 Let l’ be a vector space of dimension n. Let vj, V>, ..., V,, be the linearly inde
elements of V. Then, every other element of V can be written as a linear combination Penden
OF these
elements. Further, this representation is unique.
Proof Let v be an element of V. Then, the set {v, Vv), ..-. V, 1S linearly dependent as it has
elements. Therefore, there exist scalars Q, @), ..., @,, not all zero, such that

OV + av) nn QAnV yn = 0. (3.19)

Now, @, # 0. Because, if a, = 0, we get a,v, +... + @,,V,, = 0 and since v,, v,, ..., V,, are linearly
independent, we get @, = Q@, =... = a, = 0. This implies that the set of n + 1 elements y. Visca:
v, 1S linearly independent, which is not possible as the dimension of V is n.
Therefore, we obtain from Eg. (3.19)

v= > (— a,/ap)v,. (3.20)

i=l
Hence, v is a linear combination of n linearly independent vectors of V.
Now, let there be two representations of v given by

V=a,v, + anv, +... + a,v, and v= b,v, + bv, +... + 5,Y,

where 6, # a; for af least one i. Subtracting these two equations, we get
0 = (a, — d))v, i (4 — by)v. + ... + (a, — b,)v,,.
Since v,, Vz, .-. V,, are linearly independent, we get
aj-—6,=0 or a;= 6b, i= 1,2, ..., n.
Therefore, both the representations of v are same and the representation of v given by Eq. (3.20)
IS unique.
Remark 10
(a) A set of (m + 1) vectors in IR” is linearly dependent.
‘neal
(b) A set of vectors containing 0 as one of its eleme 4 og 5 the line
nts is linearly dependent as 9 'S th
combination of any set of vectors.
caserll
Theorem 3.4 Let V be an n-dimensional vector Space. If v,, Vo, lined"
...5 Ve kK <2 aS vj 5
independent elements of V, then there exist elements {¥); Vaan”?
a!

a basis of V. Vi+ts Verrs ---) V, Such that

| A
Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.27
proof ‘Thee exists an element V+; Such that v,, v,, ..., Ye, Ves, are linearly
Otherwise, every element of V can be written as a linear independent.
combination of the vectors Vv), Vo, 0: v;
and therefore V has dimension k < n. This argument can be continued. If >k+ 1,
we keep adding
elements Vy+i+ Vi+2» «+++ Vn such that {v,, V2, «.-, V,} iS a basis of V. |
since all the elements of a vector space V of dimension n can be represen
ted as linear combinations
of the n elements in the basis of V, the basis of V spans V. However
, there can be many basis for the
game vector space. For example, consider the vector space IR*.
Each of the following set of vectors
|
(i) (1, -1, 0}, (0, 1, -1], [0, 0, 1]
(i) (l, -l, 0}. (0, 0, 1], [1, 2; 3]

(ii) (1, 0, 0}, (0, 1, 0}, (0, 0, 1]

are linearly independent and therefore forms a basis in IR?. Some of the standard basis are the
following.
1. If V consists of n-tuples in IR”, then

&) = (1,0, 0, ...5 0), = O,.1,, v5 9); 0055, = (0, 0,-5.0, 1)

is called a standard basis in IR”.

2. If V consists of all m x n matrices, then

O ws «z. 8

Be= 10 wa 1 ws OF PETZ a m and s=1,2,...,7

GO ac @ 0

where | is located in the (r, s) location, that is the rth row and the sth column, is called its
standard basis.

aw Ol. a
For example, if V consists of all 2 x 3 matrices, then any matrix ° F |in V can be

written as

b
; G]
= aE + PEn + ey +E +yEn + 2B
x y z

1 0 0 01 0
where E,,
| = hia a}E)>, = » 6 4etc
tc.

3. If V consists of all polynomials P(t) of degree < n, then ‘1,4, 07, ..., ¢"} is taken as its
Standard basis.
Example 3.16 Determine whether the following set of vectors {u, v, w} forms a basis in IR°, where
(i) uw = (2, 2, 0), v = (3, 0, 2), w = (2, -2, 2)
(ii) u = (0, l, -1), v= (-1, 0, -1), w= G, l, 3).

Scanned by CamScanner
 i
k
i
r 1
1
®
. ' !
i}
! t

3.28 Engineering Mathematics

i

in IR?, then u, v, W must be linearly indep

Solution If the set {u, v, w} forms a basis
: . Then, the onlyly s solution of the equation PENdeny
at Let a, Q, @ be scalars,
ie au tf Qf3V + aw
3 = 0 (3)
Big

ae. | must be @, = Q@, = a = 0.

Hig
1iG gi
(i) Usinga Eq. (3.21), we obtain the system of equations .
ia 8 2a, + 3a, + 20,=9, 2a,-—20,=0 and 20, + 2a, = 0.
= = = 0.; There efore, u, v,, w Are linear
.
The solution of this system of equations is @ = 0 ~
iil independent and they form a basis in IR’.
em of equations
(ii) Using Eq. (3.21), we obtain the syst
a + % =9, and — @, — & + 3a,=0.
—~a,+30,=0,
The solution of this system of equations is Q = Q = a; = 0. Therefore, u, v, w are linearly
independent and they form a basis in IR?.

Example 3.17 Find the dimension of the subspace of IR* spanned by the set {(1 0 0 0), (0109)
(1 2 0 1), (0 0 O 1)}. Hence find its basis.
| |' Solution The dimension of the subspace is < 4. If it is 4, then the only solution of the vector
tay : equation
0)+a,(1201)+a,(00 00 1)=0 (3.22)
a(1000)+a,(010

should be @, = @ = @3 = @, = 0. Comparing, we obtain the system of equations

The solution of this system of equations is given by

O,= Ay, O% =204, A =—- a4, where 4, is arbitrary:

fore,
‘ ; of : 0¢),.Q5, %, and at. There
Hence, the vector equation (3.22) is satisfied for non-zero values
the dimension of the set is less than 4.
ide r the
==«
=ro
er
=—s_
: Now, consider any three elements of the set, say (1 0 0 0), (0 1 0 0) and (1 2 0 1). Cons
ee
Se
ee
==
Se
oe
_s
of
aie : vector equation
| (3.2
a1000)+a,0100)+a,(1 20 1)=8%.
Comparing, we obtain the system of equations
iad | a, + @,= 0, & + 2a@,=0 and a, = 0. ‘ois
are nee”
Ee The solution of this system of equations is a,= a, = a, = 0. Hence, these three € lements e4 act0t
i] !|
independent. Therefore, the dimension of the given subspace is 3 and the basis is the Se
(1 000),(0 100), (1 2 0 1)}. We find that the fourth vector can be written as
attr (000 1)=(1000)-2c010
1710)+
2 0 1).

a ag
Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.29

pxample 3.18 Let u = {(a, b, c, d), such that a +c +d=0, b + d= 0} be a subspace of IR‘. Find
- 4imension and the basis of the subspace.
solution U satisfies the closure properties. From the given equations, we have
a+c+d=QOandb+d=0 or a=-c-—dandb=-d.

we have two free parameters, say, c and d. Therefore, the dimension of the given subspace is 2. Choosing
-20,= 1 andc = 1,d=0 we may write a basis as {(-1 -1 0 1), (-1 0 1 0)}.

3.34 Linear transformations

Let A and B be two arbitrary sets. A rule that assigns to elements of A exactly one element of B is called
. function or a mapping or a transformation. Thus, a transformation maps the elements of A into the
elements of B. The set A is called the domain of the transformation. We use capital letters T, S etc. to
jemote a transformation. If 7 is a transformation from A into B, we write
T:A->B. (3.24)

For each element a € A, we get a unique element b € B. we write b = 7(a) or b = Ta and b Is called
the image of a under the mapping 7. The collection of all such images in B is called the range or the
image set of the transformation 7.
V and W be two
In this section, we shall discuss mapping from a vector space into a vector space. Let
V into
vector spaces, both real or complex, over the same field F of scalars. Let T be a mapping from
it satisfies the following
W. The mapping T is said to be a linear transformation or a linear mapping, if
two properties:
(i) For every scalar a and every element v in V
(3.25)
T(av) = aT(vy).

(ii) For any two elements v,, V2 in V

(3.26)
T (Vv, + V2) = T(¥,) + (¥2)-
ned and are elements in V. Then, T
Since V isa vector space, the product av and the sum v, + V2 are defi
the sum T (v,)
) and 7 (v,) are in W, the product a@7(v) and
defines a mapping from V into W. Since T(v,
are equivalent to
+(V,) are in W. The conditions given in Eqs. (3.25) and (3.26)
+ BTW)
T(av, + By) =T(@v;) + T(Bv2) = AT (¥,)
for v, and vy, in V and any scalars @, B.
V> ..» V,} be its basis. Then, any element
Let V be a vector space of dimension n and let the set {Vj},
of the elements Vj, V2; ++» Yn:
Vin Vcan be written as a linear combination

Remark 11
by its action on basis vectors of a vector space,
A linear transformation is completely determined
that for every element v in V
Letting a = 0 in Eq. (3.25), we find
T(Ov) = T(0) = OT(¥) = 0.

Scanned by CamScanner
 . hematics : inear
3.30 Engineering Mathem element in W by the linear transforma |
ceecvpenmed ma 252 is written as ra
Therefore, the zero element In ion ami called the rans© al fenem his Son “iq
all elements W = TV 9 element by the | eels Call
The collection of
of all elements of
V’ that are mappe
an is l-
a ker(T). Therefore, we have ‘a|
the kernel or the null-space of 7 and ran(T) = {T(y) | we: Fh.
T(¥) =
ker(T)= (¥ |
e of T is a subspace of W.
I’ and the rang
of T is a subs pace of
Thus, the null space ion of ker(T) 7 is called the nyjj;
lity of | UV
aenh
— -(T) and the dimension
)
The dimension ofran(T) !s called the rank(7’ . . ,
We, have the following result. | , ‘on of V is n, then the nullity of Tis n~> thy F
lis. |
Theorem 3.5 If 7 has rank r and the dimens
rank (7) + nullity = 7 = dim (V).
of matrices.
We shall discuss the linear transformation only in the context
A represent the elements in IR" (or Cy |
Let A be an m Xn real (or complex) matrix. Let the rows of “ . ] n , i .

and the columns of A represent the elements in IR” (or C"). If_x 1s in IR", then AX is in IR”, Thy
an m X n matrix maps the elements in IR” into the elements In IR”. We write
SS
——
=—
: , ) | Example .
=
=
T=A:1R" 5 IR”, and 7x = Ax. | ae
ae
ott
a
The mapping A is a linear transformation. The range of Tis a linear subspace of IR” and the kemdl FP
ce
es
=
ee
of 7 is a linear subspace of IR”. |
a
tiie
| Solution
Remark 12
ee

Let 7, and T, be linear transformations from V into W. We T must be

—_ define the sum 7, + 7; to be tk
—2-2.
Ee
transformation S§ such that
=
? ;
| Sv = Tv + Toy, ve V.
—
a
Then, 7,i+ + 7;7, is’a linear r tytransforma
matition and i, +t, = T, +
T,.
Example 3.19 Let T be a

TS CEC I-} te a-(

a linear transformati
a

on defined by
—
a. : _
—
,

(
a
a

ae=a
naeee
eee
Baia
tha,
titi
0 |
a
=z

eee
ct
aeg
i
a
-

ee
i
2
ee
a
eee
;
Find T
(3
4 §
oy
HE

Solutiion :
The matrices | |k |k O1To 9 |
Ake
2
ae

a
basis in the space of 2 x 2 matr; LI iflg 1 |“Te linearly independent and hence iF |
TERE. We Write for an ‘hig
ee
_
Seee
ge
——
=
=Soe
te

——
=
Y Scalars 1, O, Of, Oy, not all zero 8 th,

Scanned by CamScanner
 Pf fl LL LQ) —»>—_—
o o —

Matrices and Eigenvalue Problems 3.31

Seal; j)+ay i} 0 0 0 0
+
38) =
“Ui ba la a So
= sad a, +a,
+a, OQ, +@,+A,+ a, |
A, +A,

of equations, we get a, = 4, @ = I,
Comparing the elements and solving the resulting system
= 2. A, = 5. Since T is a linear transformation, we get

De et Doe heal -| =9
1
= 4/2] + 1/-2|-—21-2] +5} 2] =] 20].
3] 3 3 3 36
= (4, 6)", are in IR’, find the
Example 3.20 For the set of vectors {x,, X)}, where x, = (1, 3)", x»
matrix of linear transformation T : IR? — IR°, such that
Tx,=(-2 2 -7)" and Tx,=(-2 -4 -10)’.
,
Solution The transformation 7 maps column vector in IR2 into column vectors in IR*. Therefore
T must be a matrix A of order 3 x 2. Let
| a b,

| A = a> by :

| a, 5b,

| Therefore, we have

a, F
b|[3]- 2} and |a, b ||]
6, —4 |.
|
a, b, | —'] | a, b, 10 |

Multiplying and comparing the corresponding elements, we get

a, + 3b, = 2, 4a, + 6b, = — 4,

Solyj
ving these equations, we obtain

Scanned by CamScanner
 ~
3.32 Engineering Mathematics
— Five) 2.
Example 3.21 Let 7 be a linear transformation from IR° into IR“, where

J
K Tx = Ax, A= 7 “ x=(x y x)! Find ker(T), ran(T) and their dimensions.
-1 0 If} 7
; We soy, Tr
Solution To find ker(T), we need to determine all v = (UV, U2 03) such that Ty = 0, \

ee
Pll Tv = Av = 0 gives the equations oy,

v, + v, = 0, —~o, +0, =0
T
whose solution is v,; = — 02 = V3. Therefore v = v,;[1 —1 1].
Hence, dimension of ker(7) is I.
Now, ran(T) is defined as {T(v) | v € V}. We have

bd Ol,vj | |
a
eee he ge i —vU, +03
03

Ss en 0
ae
i,
pa
ele
2.—
eg

l l 0 .
Since ' = of=lip the dimension of ran(T) is 2.

{
: | Example 3.22 Find the matrix of a linear transformation T from IR? into IR? such that
i
, |

|
g
| 6 ll 6 Zz
Ti) 1) =)2), 4]|,
T)-l/=|-7] —2 |=] 6],
4 | 2 3) \5
: ; . 10nis?
Solution The transformation T maps elements in IR? into IR?. Therefore, the transforma
matrix of order 3 x 3. Let this matrix be written as

@, Q a;
f=-A=|8, Bs B, |.
Y1 Yo Yi

We determine the elements of the matrix A such that

a
=
a, a a@;|[l] [6] fa, a, a@,)f 1 2] fa, a, a;]| !
B, B; B; li 2 , B, B, B, }-l}=|-41, | B, B; B; —2 |=

Y, 2 344! 4) 1% Fo Yell 1] | 2 Y, 2 43 3

Equating the elements and solving the resulting equations, we obtain

Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.33

Zz 6g
A= |-15/2 3. 13/2).
1 2

Fxample 3.23 Let 7 be a transformation from IR° into IR! defined by

T(X\, X2, %3) =x} + x3 + x4.

Show that 7 is not a linear transformation.

Solution Let x = (Xj, %5, x3) and y = (yj, >, y3) be any two elements in IR>. Then

RA Y= (y+ Ys Xz + Ya, Xy + V3).

We have

M(x)= x, +x54+x4, Ny) =r t yo +3

T(x + y) = (x1 + yi)” + (ey + yn)? + (xs + 5)? # TOX) + TOY).
Therefore, 7 is not a linear transformation.

Matrix representation of a linear transformation

We observe from the earlier discussion that a matrix A of order m X n is a linear transformation which
maps the elements in IR” into the elements in IR”. Now, let T be a linear transformation from a finite
dimensional vector space into another finite dimensional vector space over the same field F. We shall
now show that with this linear transformation, we may associate a matrix A.
Let V and W be respectively, n-dimensional and m-dimensional vector spaces over the same field F.
Let T be a linear transformation such that 7: V — W. Let
x= {V), V>, er V,$5 y = {W), W>, esi Wit

be the ordered basis of V and W respectively. Let v be an arbitrary element in V and w be an arbitrary
element in W. Then, there exist scalars, 0), Q>, ..., &, and B), B, ..., 8, not all zero, such that

V = QV, + Gov, +... + QV, (3.271)

w= Bw, + Bow, t+... + BW (3.27 ii)

and w= Tv = 7T(Q@)V, + Q2v, +... + a,V,,)

= aly + Qa ¥g% 0+ 2,7, (3.27 iil)

Since every element Tv, i= 1,2, ..., is in W, it can be written as a linear combination of the basis
Vectors Wi, W>, w,, in W. That is, there exist scalars a, 1 = 1,2, ....9, f= 1, 2, .... m not all
ia, |

Z€T0, such that

Tv; = a; + Aa;W Ft ..6 F Ani Wm
Fy (3.27 iv)
lI [Wy, Woy ees Woy) [Qyjs Fajr veer Amilo E = 1, 25 ceey
H ;
“NCe, we can write

Scanned by CamScanner
 3.34 Engineering Mathematics

'
ay] ay Gy,
ee
ee
ae
a> 47 A>,
; 7 , aay W .
—=———
a
T Vy. Voy ee Vy) = LWis 2) nm] : (3,2) \)
eT Amy? Qin

or 7Tx=yA
ia
ee
where A is the wm X nm matrix

a, 42 - An
. :
a5, a9 sae A5y,

A= : , (3.27 vi)

Am) m2 &mn

respect to th
The m x n matrix A is called the matrix representation of 7 or the matrix of T with
ordered basis x and y. It may be observed that x is a basis of the vector space V, on which Tacy
and y is the basis of the vector space W that contains the range of-7. Therefore, the matrix
representation of 7 depends not only on 7 but also on the basis x and y. For a given linear
given
transformation 7, the elements a; of the matrix A = (a,;) are determined from (3.27 v), using the
basis vectors in x and y. From (3. 27 iii), we have (using 3.27iv)
=A

‘>
3

oe
mk

Ww OLj(a Wy + Gay Wo FH o-- Fy {Woy) + Ola(Qy2Wy + Ap2Wa + oF 2Mal

+ oon + (a), + az,W> + myer + AnnWm)

(ja, + QyQ,2 +... + A,Qy,) Wy + (Q) a>, + Ong, + o1. + Oj42u)™2

ices (@)4,,) - Oy Anat ove T On Ann) Wy.

Bw, + fiw, Tee mW m

ee
9a
°°. ee where B, Oa), F Agty +... Pa, P= 1, 2, «5!

B, Qjy yQ2 wee Ay | Of

=us.
=4
a
Hence, Po =| 921 922 + Gan 2
ee
ee
a
=.
ee
le
ee

Bm Any —Im Ginn | & n

or Bp=Aa
where the matrix A is as defined in (3.27 vi) and

B= [By Bay + yl", a = [O,, Ol, «++, a)". , linet

W respectively: os - yt ®
For a given ordered basis vectors x and y of vector spaces V and
We prove tn
transformation 7: V — W, the matrix A obtained from (3.27 v) is unique.
= - & |

follows:

a
Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.35

pet A = (a,) and B = (6,;) be two matrices each of order m x n such that
[x= yA and Tx = yB.
Therefore, we have

yA = yB
nn m

Dy Wa= DL woz, f= 1, 2, 0, 0.
or i=] j=|

since Y = {wW), Wo, .-+. Wm} 1S a given basis, we obtain a; = b, for all i and j and hence A = B.

Example 3.24 Let 7: IR? — IR? be a linear transformation defined by

: +
T\ y\= ? ‘|
yore
2

Determine the matrix of the linear transformation 7, with respect to the ordered basis

1) (0) (0
AL 4 ry Tet ‘
(i) x= s/O,/1),;0]7 in IRX and y= {o}(c} in IR
0) \O) \1)] - 7

(standard basis e,, €5, @3 in IR? and @), @> In IR’).

(O\ (1) (1
is l =. gr : 2
(ii) x= 9/1,/0,])1]7 m IR’ and y= ()("Iin IR”.
1) (1) (0 |

Solution Let V = IR°, W = IR’. Let x = {v,, V>, V3}, y = {W), Wo}-
| iB 0 0 0
(1) We have v,=/0], vy=lll, vw=alo, wir 6p W>= |
0 0 ]

| 0 | I
, 0
|
We obtain 710] = v = : (0) + | (OY. FLL] =). )= | Jo,
0 0 0 l 0 1} \0 |

Ee
Using the notation given in (3.27 v), that is Tx = yA, we write

Scanned by CamScanner
3.36 Engineering Mathematics

O11 |
T [Vj V2. Val
=[Wi Wally yo

«oft tol [hCl + -J

Therefore, the matrix of the linear transformation T with respect to the given
basis Vectors
by S Biv
he 01 1
10 1 -

0 I |
(ii) We have Yi=([l, vw=|/9. wll, w= " m=(')
0 =

| 1 |
We obtain - J] 1 -(3)- (os Jo. T\0 -( J-(Jo-{ Joa
: 0 | =x = l re)

[OOo
Using (3.27 v), that is Tx = yA, we write

MC OIE + g
0)

1)
/1

(1)
]

(0
I
,
]

s
Lt 4 |

.. givel
Therefore, the matrix. of the linear transformation T with respect to the given basis vectors » given
,

by |

Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.37

piscuss whether V defined in problems 1 to 10 is a vector space. If V’ is not a vector space, state
which of the properties are not satisfied.

Le of the real polynomials of degree < m and having 2 as a root with the usual addition
and scalar multiplication.
2, Let V be the set of all real polynomials of degree 4 or 6 with the usual addition and scalar
multiplication.
3. Let V be the set of all real polynomials of degree > 4 with the usual addition and scalar multiplica
tion.
4, Let V be the set of all rational numbers with the usual addition and scalar multiplication.
5, Let V’ be the set of all positive real numbers with addition defined as x + y = xy and usual scalar
multiplication.
6. Let }’ be the set of all ordered pairs (x, y) in IR? with vector addition defined as (x, vy) + (u, v) =
(x + u, y+ 0) and scalar multiplication defined as a(x, y) = (3ax, y).
7. Let V be the set of all ordered triplets (x, y, z), x, y, 2 € IR, with vector addition defined as

(x, ¥, z) + (u, v, w) = (3x + 4u, y — 20, z+ w)

and scalar multiplication defined as

Q(x, vy, z) = (ax, ay, Az/3).

8. Let V be the set of all positive real numbers with addition defined as x + y = xy and scalar
multiplication defined as ax = x*.
9. Let V be the set of all positive real valued continuous functions fon [a, 5] such that
b b 7
(i) | J (x)dx = 0 and (ii) J F(x) dx = 2 with usual addition and scalar multiplication.
a a

Let V be the set of all solutions of the

(i) homogeneous linear differential equation y” — 3)’ + 2y = 0.
(ii) non-homogeneous linear differential equation y” — 3y’ + 2y =x.
under the usual addition and scalar multiplication.
Is Wa subspace of V in problems 11 to 15? If not, state why?
11. Let V be the set of all 3 x 1 real matrices with usual matrix addition and scalar multiplication and W
consisting of all 3 x 1 real matrices of the from
a a a | a

(i)] b }, (ii) |a|, (iii) |b], (iv) |b].

a+b a’ 2 0
12. Let V be the set of all 3 x 3 real matrices with the usual matrix addition and scalar multiplication and
W consisting of all 3 x 3 matrices A which
(i) have positive elements, (ii) are non-singular,
(iii) are symmetric, (iv) A°=A.
13. Let V be the set of all 2 x 2 complex matrices with the usual matrix addition and scalar multiplication
and W consisting of all matrices with the usual addition and scalar multiplication and W consisting
Zz wre where x, y, z, u are real numbers and (i) scalars are real
of all martices of the form ;
b at u
numbers, (ii) scalars are complex numbers.

Scanned by CamScanner
 addition .
= 4 wi th th e on polynomial nd Gl,
degree ee S 4 having
n o n s yO a k |
of de gr
———consist of all rea: l poly | nomialpol -_
14. Let V , cons!sconsisting g of of
| multiplication an
d HW’ (ii) coefficient
(i) constant term I, (iv) only real roots.
| | iii
| 3
|
Te (iii) coefficient of ras |,
of the form (xy. *2 x) in IR° =h a i addition gNd se
all triplet Me
15. Let l’ be the v ec
tor space of
of plete of the form (*1, *2 x;) suc
Hb e the se t
multiplication and
(ii) y= 7 a
(1) vy = 2N; =
3X4.
! + txy Ss. ty) ays = IEEE .
itrary, (iv) x4
(iii) x, 2 0, .X9.-%3 arb a3), Determine whether or not X is a linear COmbina .
d w= (1, 2s
y= (2, 3,4) an
«i==
3s
=—*
—i———e
-: eee

16. Letu=(1,2, —|),

=--
ee

ft) 4 oteti bY
ti of uy Vy Wy Where x is given 7 2.5) (iii) (—2, 1, —5).
EY
| 8 (i) (4, 3, 10), 6 a8
Il , =
1j and w = (2, 3,1, -1)- ,
Determine whether or not x iS° a line

H 7%, Letwa{ly-t
WS Gy ee
a combination of u, Vv. W, where x Is given by
Pig | i) (3,0, 5-1 (ii) (2,-7, 1, 1D), i)
Ae (4.3.0.3
)
|
be 18 m,. o-F At—6, Pa(t) = 202 — Tt— 8 P(t) = 2t- 3, Write EG ESS ERNST
. | ~ i 7 =
a
P(t), P(t), when . hs ‘
Tt
: H (i) P()=-t'?+ l, (ii) P(t)= 21° —3t—-25. , =
| J Exe
Show that the set
| | 19. Let V be the set of all 3 x | real matrices.

bit 1) ( 1) (0
S=4] 1}, |—-1}, |0]? spans V,
|
| 0 0) \l

20. Let V be the set of all 2 x 2 real matrices. Show that the set

AGED G FC Zp
21. Examine whether the following vectors in
1R2/C? are linearly independent. .
(i) (2, 2, 1), (1,-1, 1), (1, 0, 1),
(iii) (0, 0, 0), (1, 2, 3), (3, 4,5}. (ii) (1,
2, 3), (3,4,5), (6, 7, 8), “Fin
(iv) (2, 7,-1), (1, -3, 4), (2i, -1, 5)
(v) (1,3, 4), (1, 1, 0), (1, 4,
2), (1,-2, 1).
22. Examine whether the
following vectors jn IR?
are linearly independent.
(1) (4, 1, 2, -6), (1, 1, 0,
3), a, ~|, 0,
2), (—2, bs
(1) Cl,2,3215,.02. 1,-l,1 0, 3),
)s (4,
(ii) (1, 2, 3, 4), (2,0, 1, -2), 3.95,5 , 3), (5,4 1, 3),
.4 2)
~— IC 1,1, 1), -1,=1,1, 1)ee
a4: "a ;‘eas it 7 (1, 5, 1,8), (-1,7 —
(i)x THY tx, — vectors in IR}, then i that
are also linearly independ in IR? (i) x, x + WwXty+y, 4B
24. Write (~ 4,7,9) as a - }
Show that S is
not 4 Spannin 8 set in IR? T the© elements
eler of, the set S: {(1, 2, 3), (- 1}; 3,4 yr

a
ee
x

Scanned by CamScanner
 Matrices and Eigenvalue Problems
3.39
_ WritFae e ©hk + ¢+ 1 as a linear com
bination of the elements of the set S: Bor
— i fe Dp. 2}. Show
that S7 is the spanning set ) for all poly
nomials of degree 2 and can be taken
as its basis.
26. Let V be the set of all vectors in IR‘ and S be a subse
t of V consisting of all vectors of the form
(i)(x,.¥, —V, —¥), (ii) (x, 7,2, w) such that x+ yt+z—w=0 4

(iii) (x, 0, 2, w), (iv) (Ww. x, x, x),

Find the dimension and the basis of S.

27. For what values of k do the following set of vectors form a basis in IR° ?
(i) {(A, 1 ~ A, A), (O, 3K — 1,2), (— k, 1, 0},
(ii) {(A, 1, 1), (0, 1, 1). (A, 0. 4D},
(iit) {(A, AL A). (O. AL A), (kK, 0, KY},
(iv) {(1, &, 5), (1, -3, 2), (2, -1, 1)}.
28. Find the dimension and the basis for the vector space V, when V is the set of all 2 x
2 (1) real matrices
(ii) symmetric matrices, (iii) skew-symmetric matrices, (iv) skew-Hermitian matrices. (v) real
matrices A = (a,,) with a,, + aj, = 0, (vi) real matrices A = (a;,) with a), + a,, = 0.
29. Find the dimension and the basis for the vector space V, when V is the set of all 3 x 3 (1) diagonal
matrices (11) upper triangular matrices, (iii) lower triangular matrices.
30. Find the dimension of the vector space V, when V is the set of all x 7 (i) real matrices, (ii) diagonal
matrices, (111) Symmetric matrices (iv) skew-symmetric matrices.
Examine whether the transformation 7 given in problems 31 to 35 is linear or not. If not linear, state why?
7: -
=
EER,
a
as
—
=

‘ x .
31. T:1IR- > IR': r{*) =x + y+ a, a #0, a real constant.

Xx
2
32. T: IR? 3 IR? :T vil= ‘ } 33. T: IR' > IR’; T(x) = . }
: ,ae FZ 3x fs

: 0 x«#0,y40 x {
34. T: IR? + IR! ; r{*) = 2y, x=0 35.T: IR? > IR'; Tl y |=xy tate. \
y 7
3x, y=0. z

Find ker(T) and ran(T) and their dimensions in problems 36 to 42.

x x+) 2x+y
r x
36. Ts IR? SIR Thy} =] oz |. a1. 7 IR? 9 IRs T||- y=
; i
- x-} 3x +4y

‘a xtyt+w :
3
38, T: IR? = IR?: T J
: x
ss 3 | 39. T: IR7° > IR': T ||=y+ 3y.
z oat Pee Py
iM yt+2w

x
r=—y
40. 7: IR? = ir! rl y| =x + 3y. 41. T: IR? 3 IR*: T *)|=
y i= y
ee, |

Scanned by CamScanner
 cs
t 3.40 Engineering Mathemati
Yea |
re

at * 2e—F }
be 42. TL R; GI RS T|) | * |ax4z.
i.) z =
x vey
| eee | ened by TL] = 4°
‘ea
Lil | on defined i~<«
z
he 43. Let T: IR? 2 IR? be a linear transformall
is
T with respect to the ordered bas
ent ati on of I \
Find the matrix repres
1) (1) (9 |
ra | ema | = Ss). & in IR*,
= dohl iby lly in IR ane y {3}?
,
\7\
ae . a line ar transformation d ned
ation defi by ee
ca
T: VO W be a linear
44. Let V and IW’ be two vector spaces in IR?, Let
ir
Pe,
* 0
| | | pe
Tly|=| x+y |-
| | See
z x+y+z
| }
ered basis |
|
|
Find the matrix representation of T with respect to the ord

1\ (0) (0 (1\ (1)(0 :

7 :
| | x=dohfikjolsin v and y= ] ]1]) in©
O]1)
| | 0) (0) (1 1) \O)} \1

|f | 45. Let V and W be two vector spaces in IR*?, Let 7: V 3 W be a linear transformation defined by F
|

[ | .
i |
x x+z
|| |
Tly| = x+y |.
||
* XE VEZ 1
|
i
Find the matrix representation of 7 with respect to the ordere
d basis
i
i}

rT ~
i
+

> l cia ; 1}} in V and y= ~] , 1k —| in IV" ee

tic 2 x+y i
46. Let T: IR? — IR* be a
© be a linear transformation defined
by T po
a
oeyi es
x+2
. rs
qT
Find : oe 0)
ind the matrix representation of T with respect
to th
© ordered basis a, | :
1) (1) fo |
| x= 10h}, 1\ (1

| YI Torta] in IR*

Scanned by CamScanner
 Matrices,and Eigenvalue Problems 3.41

-1R2 il 2
47. Let 7: IR" — IR° 3 be a linear transformation.
;
Let A - 2 3] be the matrix representation of the linear
3 4
transformation T with respect to the ordered basis vectors
v, = [1, 2]’, v, = [3, 4]’ in IR? and
w, ifs.=(-l, 1, Fiow, cee,”
= [l, -1, 1y"re.
W, =[1, 1,—1]" oein IR*. Then, determine the linear transformati
on T.
3 aA 1 2
48. Let 7: IR° — IR° be a linear transformation. Let A = ; 3 \be the matrix representation of

the linear transformation with respect to the ordered basis vectors ¥ = fl, © f2,3) 7"
3
v,=[I, 47s, in IR° 1p}
1,-I]) o Pes a :
and w, = [1, 1)’, w, = [2, 3]’ in IR’. Then, determine the linear
transformation T.
a.
Le
0S

1 2
49 - Let 7: P,\(’) — P(t) be a linear transformation. Let A=| 2 31] be the matrix representation of the
—| |
linear transformation with respect to the ordered basis [I +2, 4 in Py) and [1 — 2, 21,2 + 32- t*]
in P(t). Then, determine the linear transformation T.
50. Let V be the set of all vectors of the form (x), Xp, X;) in IR? satisfying (i) x, — 3x, + 2x, = 0:
(ii) 3x, — 2x, + x; = 0 and 4x, + 5x, = 0. Find the dimension and basis for V.

3.4 Solution of General linear System of Equations

In section 3.2.5, we have discussed the matrix method and the Cramer’s rule for solving a system
of n equations in 7 unknowns, Ax = b. We assumed that the coefficient matrix A is non-singular, that
is |A| # 0, or the rank of the matrix A is n. The matrix method requires evaluation of n* determinants
each of order (7 — 1), to generate the cofactor matrix, and one determinant of order mn, whereas the K
Cramer’s rule requires evaluation of (nm + 1) determinants each of order n. Since the evaluation of ’
high order determinants is very time consuming, these methods are not used for large values of n,
say n > 4. In this section, we discuss a method for solving a general system of m equations in n
unknowns, given by | ¥

Ax=b _ (3.28)
Qj; Ayo «++ Aly | b, x)

AS @21 Qx2_ «+s 2g b= K x= X92

where

Bj Gg tse Ay D,. Xi,

are respectively called the coefficient matrix, right hand side column vector and the solution vector.
The order of the matrices A, b, x are respectively m xn, m Xx | and n x I.
The matrix

a, a. .-» A, | O
Ay, Gy «+ Gy b, (3.29)
(A | 5) 2 ' ; ,

Ay) 49 nit Cig, bn

Scanned by CamScanner
 ; P ics F |
3.42 ) Engineering Mathemati
Eng + 1) columns. The augmented matrix desert,
@ d has > vet rows and (/ | of the system of equations (3,28) ;. be
en te d ma trix an
rx a
is called the augm a so lu tion vector pos sib ilities:
em of eq ua ti on s. er e ar e thr ee
completely the sy
st e equations. Th
——

, 2. + x, ) th at satisfies all |
n-tuple (X)
a u nique solution,
(i) the system has
————

Pa
=. -.=z

no solution,
(ii) the system has solutions.
infinite num ber of
——ee
——————
SS
=a—--
-—

f
—d inconsistent, vij¢;
(iii) the system has it ha s at le as t on e so lution an
sistent, if
on s is sa id to be con or spaces, We NOW obtain
the necessary an4
The system of equati of ra nk s an d ve ct
g the concepts ion of the linear system of
has no solution. Usin
a‘ ‘

d uniquene ss of the solut

ee
rr
SS
ee
aSpoo
‘ ee
.ae
—————————————
SSS

e an
a
—ls
"

for the ex is te nc
sufficient conditions
es

—=—————

equations.
Ton Joo C464 |
: ‘web
ne Existence and Uni que nes s of the Solution
aaa
s Ry, Ry, .... R,, of the
Let V7, be a vect or spac e cons isti ng of n-tuple s in IR"(or C”). The row vector
\> ce of0 V,, generated by the row,
s space
sub
matr ix A are n-tu ples whic h belong to V,. Let S be the
the
m xn ension is called the row-rank of 4
the mat rix A and its dim
of A. Then, S is called the row-space of
ee
a
<=.

=a‘
.aliases
esss
ee and is denoted by rr(A). Therefore,
row-rank of A= 1r(A) = dim (S).
(3.30)
of A denoted by cr(A).
—E
Similarly, we define the co/umn-space of A and the column-rank
Since the row-space of m Xx n matrix A is generated by m row vectors of A, we have dim (S) <m
and since S is a subspace of V,,i we have dim (S) <n. Therefore, we have
me
ee
2
oo
ee
ee
ee

similarly cr(A) < min (7m, 1). (3.31)

——-
-=_—
=

rr (A)< min (m,n) and

Theorem 3.6 Let A =(a,) be an m X n matrix. Then the row-rank and column-rank of A are same.
Now, we state an important result which is known as the fundamental theorem of linear algebra.
es
: Theorem 3.7 The non-homogencous system of equations Ax = b, where A is an m Xn matrix, has
a solution if and only if the matrix A and the augmented matrix (A | b) have the same rank.
a ee — pi defined the rank of m X n matrix A in terms of the determinants of the
ee
oe
a
a

ee
neon :: ' . :: <n matrix has rank r if it has at least one square submatrix of order r which
ingular and all square submatrices of order greater than r are singular. This approach is very
time consuming when 1nn isis large.
lar |
Now, we dis :
mately. ’ rank of a
cuss an alternative procedure to obtain the

=_
3.4.2 Elementary Row and Column Operation
«=
7=_
==—

The following three Operations on a- ,

:
On a matririx
4 A are called the elementary row operations:
—
.---

i) Interchange of any two row

s (written as R, ~ R).
(iil) Multiplication/division
of any row by “non-zero scalar (written as @R,)
(iii) Adding/subtracting
as calar multiple o
that is @ multiples of i ore another row (written as R, — R; + ak,
the elements of
of the ith row. The are added to the corresponding elemen"®
id
t- ‘ ee
ses.
ee
aipe
a
a
a
gape
a

ele ments of the jth aes

ith row get changed), emain unchanged, whereas, the elements ofth

Scanned by CamScanner
aan i
7

Matrices and Eigenvalue Problems 3.43

These operations change the form of A but do not change the row-rank of A as they do not change
the row-space of A. A matrix B is said to be row equivalent to a matrix A, if the matrix B
can be obtained from the matrix A by a finite sequence of elementary row operations. Then,
we usually write B = A. We observe that
(i) every matrix is row equivalent to itself.
(ii) if A is row equivalent to B, then B is row equivalent to A.
(iii) if A is row equivalent to B and B is row equivalent to C, then A is row equivalent to C.

called elementary
the above operations performed on columns (that is column in place of row) are
column operations.

3.4.3 Echelon Form of a Matrix

An m Xn matrix is called a row echelon matrix or in row echelon form if the number of zeros
preceeding the first non-zero entry of a row increases row by row until a row having all zero entries
(or no other elimination.is possible) is obtained. Therefore, a matrix is in row echelon form if the
following are satisfied.
for all subsequent rows.
(i) If the ith row contains all zeros, it is true
(ii) Ifa column contains a non-zero entry of any row, then every subsequent entry in this column
is zero, that is, if the ith and (i + 1)th rows are both non-zero rows, then the initial non-zero
entry of the (i + 1)th row appears in a later column than that of the ith row.
(iii) Rows containing all zeros occur only after all non-zero rows.
is

For example, the following matrices are in row echelon form. 1%

i

oe

i234
fd
——
1

nas :
iv

PaeMiatani
7

4 1,/6 © F SE. 4 al
.

Ns \y

0009) [0 00 0J|, 0 4 \

ange the first row

Let A = (a;,) be a given m x n matrix. Assume that a,, # 0. If a,, = 0, we interch
ary row
with some other row to make the element in the (1, 1) position as non-zero. Using element
ts of first column below a, are
operations, we reduce the matrix A to its row echelon form (elemen
are made zero and so on).
made zero, then elements in the second column below a2)
Similarly, we define the column echelon form of a matrix.
of
Rank of A| The number of non-zero rows in the row echelon form of a ma trix A gives the rank
and the set of the non-zero
the matrix A (that is, the dimension of the row-space of the matrix A)
tows in the row echelon form gives the basis of the row-space.
Similar results hold for column echelon matrices.
Remark 13
upper triangular matrix and the
(i) If A is a square matrix, then the row-echelon form is an
column echelon form is a lower triangular matrix.

7 —_
Scanned by CamScanner
y
 3.44 Engineering Mathematics
—~
(ii) This approach can be used to examine whether a given set of vectors are ling
arly ip C
or not. We form the matrix with each vector as its row (or column) and red UCe if to Pete,
(column) echelon form. The given vectors are linearly independent, if the , ow e
has no row with all its elements as zeros. The number of non-zero rows jg th, ie .
| the given set of vectors and the set of vectors consisting of the non-zero
TOWs 'SSeg
the 4.
of
Example 3.25 Reduce the following matrices to row echelon form and find thei, rankks
‘i
-_ 1 2 3 4
als .5 . {2 1 4 =5
( i ; —1ae 4), Hu
la ge 6
2
8 1 14 17

Solution Let the given matrix be denoted by A. We have

5
ji) A= 5 . si atins: ; 6|R.+2R 0 : :
| | Ree OR, ee Tal
3 @ 3 0 14 12 0 0 9g
This is the row echelon form of A. Since the number of non-zero rows in the row echeloy
form is 2, we get rank (A) = 2.

12 3 4], ~ 1 2 3 4 1 2 3 4
ws eel | + 3 gap wall “2 “2 “3 Reeee [a
" ig 6 7/2? “ "lo 3 &@ 31k -5R, “l0 0 0 OT
R, —8R,
——
ee
8 1 14 17 O 15 ~ib <15 , 0 0 0 0

Since the number of non-zero rows in the echelon form of A is 2, we get rank (A) wd
Example 3.26 Reduce the following matrices to column echelon form and find their ranks.

3 i ‘F | I -] l
~|1 2 4 . {el 1 -3 -3
(1) la 7P | (ii) ; oe 4 al
4 -1 § 1 -l 3 3

3 Ey 3 0 0] [3 0 0
| } 2 41G-G/3 11 S/3 §/3 lt 5/3. 9
(1) A = :
4 =] TIC,-7C,/3 (4 -3/3 =7/3 2 14 -—7/3 (0)
‘i C; = C.. —_

2 ¢ > Ws pa 27 1/3 0

Scanned by CamScanner
>
Matrices and Eigenvalue Problems 3.45

Since the column echelon form of A has two non-zero columns, rank (A) = 2.

1 1 -!1 y 1 0 0 9Q | 00 0
(ii)
a | [ow )
A= Cy +c, = =
1 O 1 —— lL -t 2 1| C,+C, 1 -1 0 O
;

{=f 3 3,* ~“ [ ZF 4 8 il} 2 DD

Since the column echelon form of A has 2 non-zero columns, rank (A) = 2.
Example 3.27 Examine whether the following set of vectors is linearly independent. Find the
dimension and the basis of the given set of vectors.

(1) (1, Zi 3, 4), (2, 0, l, = 2), (3; 2D: 4, 2);

(ii) (1, 1, 0, 1), (I, 1, 1, 1), G1, 1, 1, 1, G1, 0, 0, 1),

(iii) (2, 3, 6, —3, 4), (4, 2, 12, —3, 6), (4, 10, 12, —9, 10).
Solution Let each given vector represent a row of a matrix A. We reduce A to row echelon form.
If all the rows of the echelon form have some non-zero elements, then the given set of vectors are
linearly independent.

L123 4),.5, [1 2 3 4 1 2 3 4
(i) A=|2 0 1-2), 4, =|0 -4 -5 -10]/R;-R,=|0 -4 -5 -10].
924 3/7 * to 4 ~§ =p 00 0 0
Since all the rows in the row echelon form of A are not non-zero, the given set of vectors
are linearly dependent. Since the number of non-zero rows is 2, the dimension of the given
set of vectors is 2. The basis can be taken as the set of vectors {(1 2 3 4),
(0, —4, —3:5 —E0) }. é
\

PIG, » |f 104 i101

i a CP ENS oO o10), , ]9 2121,,,,
Se) ee 5iL a le 6 4 op
food * ™ (6 <i 0 9G 0-100

11 01] 1101
0212 Gea 2
= R,-R,/2 = |
rAin ” 0010
0012 1 000 1

Since all the rows in the row echelon form of A are non-zero, the given set of vectors are
linearly independent and the dimension of the given set of vectors is 4, The set of vectors
(1, 1, 0, 1), (0, 2, 1, 2), (0, 0, 1, 0), (0, 0, 0, 1)} or the given set itself forms the
basis.

Scanned by CamScanner
 3.46 Engineering Mathematics
™
2 3 6 2 ae
2 36-3 4: 2 a a

Gp as} FF 6) 1g Mee OF Ohoey

-9 0} © | [0 40-3 2
41012 0 00 9 WF:
Since all the rows in the echelon form of A are not non-zero, the given set of Ve
ClOrs ate |
linearly dependent. Since the number of non-zero rows Is 2, the dimension of the
of vectors is 2 and its basis can be taken as the set {(2, 3, 6, -3, 4), (0,- 4, 9. 4BIVEN
ey sey

3.4.4 Gauss Elimination Method for Non-homogeneous Systems

Consider a non-homogeneous system of m equations in n unknowns
Ax =b (33)

a) | a)\2 Qin by x)

(a A=| Qa,C2! a 422 a an |b= b»2) x= X>

, |
where
|
Qni Fm? Amn 5m x),

We assume that at least one element of b is not zero. We write the augmented matrix of order
m~
-_
~»-
m X (n+ 1) as

Qj, G2 «. Any |B
ee
ee
eee

(A | b) = aa a 22 .. a 2n Ms

Gu; Gin «ws Gy, 16,

and reduce it to the row echelon form by using elementary row operations. We need a maximum of
(m — 1) stages of eliminations to reduce the given augmented matrix to the equivalent row echelon
form. This process may terminate at an earlier stage. We then have an equivalent system of the form

Qj, Ap ** Ay ss Ay} 4
O Gy ++ Gy, ++ Ay] by
* ° : 3,33)
eet
-—
ee
_
(A | b) = 0 0 Se: S88 Bye b, l (

aS One Oem ® Ba
~—--
—
.

ee!
OUD
ee
fSe
0 O oO. 0| b,

ases:
where r S m and a,, # 0, @y, # 0, ... , a, # 0 are called pivots. We have the following ©

util
Scanned by CamScanner
 Pew 0 I I I

Matrices and Eigenvalue Problems

3.47
(a) Let r <m and one or mor : e of the elements fh"mits boxy ...,
Then, rank (A) # rank (A 5), are not zero.
| b) and the system Of equation
s has no solution.
(b)Let m 2 and r= 7n (the number of colu
mns in A) and A-r+ty 5,49 * *
, ..., 6, are all zeros.
In this case,
Se, rank (A) = rank (A | b) = 1 and d the
the system of equation
ati s has a unique solution
We solve the nth equation for X,,, the (7 — 1)th equation for
.
x,-, and so on. This procedure
is called the back substitution method
For example, if we have 10 equa
tions in 5 variables, then the augm
order 10 x 6. When rank (A) = rank ented matrix is of
(A | b) = 5, the system has a unique
‘(c) Let r <a and — a * solution.
r2+b +++5 6,, are all zeros, In this case, +
unknowns, x), x, ..., x. can be
determined in terms of the remaining (7 —
7) unknowns x r+]> Xpaas - --+ X, by solving the
equation for x,, (r — 1)th equation for rth
x,_; and so on. In this case, we obtai
(n — r) parameter family of solutions, n an
that is infinitely many solutions.
Remark 14
(a) We do not, normally use column eleme
ntary operations in solving the linear system of
equations. When we interchange two columns, the order
of the unknowns in the given system
of equations is also changed. Keeping track of the
order of unknowns is quite difficult.
Natrix of order (b) Gauss elimination method may be written as
Elementary
(A | b) —> (B |c).
row operations

The matrix B is the row echelon form of the matrix A and c

is the new right hand side column
vector. We obtain the solution vector (if it exists) using
the back substitution method.
(c) If A is a square matrix of order n, then B is an upper triangular matrix
of order n.
(d Gauss elimination method can be used to 4
'tena
on
—
solve p systems of the form Ax = b,, Ax =by,, ...,
Ax =b, which have the same coefficient matrix but different right hand
qa maximum af side column vectors.
We form the augmented matrix as (A |b), bs, ..., b,), which has
m rows and (71 + p) columns. ee

Using the elementary row operations, we obtain the row equivalent system Chan
t eee

(B | ¢, C5,
=171 of the form
teey c,), where B is the row echelon form of A. Now, we solve the systems Bx = c),
Bx =¢, ..., Bx = ¢,, using the back substitution method.
Remark 14

(a) If at any stage of elimination, the pivot element becomes zero, then we interchange this row
with any other row below it such that we obtain a non-zero pivot element. We
normally
choose the row such that the pivot element becomes largest in magnitude.
(b) For ann x 1 system, we require (#7 — |) stages of elimination. It is possible
to compute the
total number of additions, subtractions, multiplications and divisions.
This number is called
the operation count of the method. The operation count of the Gauss eliminatio
n method for
solving an n xn system is n(n” + 3n — 1)/3. For large n, the operati
3p
on count is approximately
Ww/3,

Scanned by CamScanner
 >

tics gj,
3.48 Engineering Math emal ons (if possible) using Gauss iming
— ; of equat! iy
tems
Example 3.28 Solve the following sys
2 G Th jx 3
method. 4
eal . a Nec
yt =I /x : sli
| ape
¥ = |-2},
(it)
4. ~Q’ S)1z 3
i
Gy] ;
tal
211
<i 5
-) 2 -ljL4J

pat Tile Ct
(ii) {2 1 -TYI2 =|2].
5 —2 2)/4 5
ying clementay
Ss
ee
ee
oor

it to ro w ec he lon form by appl

reduce
ented matrix and
Solution We write the augm
row operations. | =|) 4
4 2
2
2 |g
1-i|
-3 /2 - 5/ 2| —4 ) Rs + 582
2 R,-R l
jy (AL By™ =|4 1b -2geil24>?2 R,+R/2 1g 5/2 -3/2) 4

» 1 -i 4
u go -3/2 5/2) —4'.
0 0 8/3|-8/3
on as
bs ti tu ti on me th od , we obtain the soluti
Using the back su
8
8 g=-l,
=—-—, OF
3° 3

_— By 422 =-_4
—
or y=l,
2
xty-z=4, oO x= un
2.2),
1. = 1+
th e sy stem of equations has the unique solution x @ j
Theref or e, |
, : 2, O
a @ ys
1/2}—1/2| Ry-2R2=|° ~
—-1 VWI), =lo -1 0 ae

(i) (A[D)=]1
=
aes go 0 gst!
—2R,
—2 a
| gj
R,

4 -2 3)3
om
(A | b) = 3. Th er ef or e, the system © equati
= 2 and rank |
We find. that rank (A)
solution
| rat wi pat!
1-1 -2R, = [0 « <al6l R= Rew] 3 -3/°0
. , 2/2R, _ g p
y=|2
IDY
Gi(Ai 1- R, —5R,
+1 453 2

Scanned by CamScanner
A
 oUF aaa
‘ i ‘o~ Fe
( ,
wa

3.49
Matrices and Eigenvalue Problems

equation
te number of solutions. We find that the last
The system is co nsistent and has infini jy —_ 32 = 0, Or Bs = F,
z. From the second equation, we get
is satisfied for all values of x, y, in the solution
+z = 1, or x = 1. Therefore, we 0 bta
From the first equation, we getx —-y
x= 1 y=? and z is arbitrary.
s using Gauss elimination method.
ple 3.29 Solve the following system of equation
Exam
= 0 (ii) x+2y-2z=I1
(i) jy = Sy = 92 + Ow
ox + 3y + 32 + bw = 6 2x —3y +z =0
5x t+ y-—5z = 1
Ax —21y _ 397 — 6w = —24,
3x + 14y — 12z= 5.

Soluti on
We have
= _9 6| O
-3 -9 6 l,.6 2
4
: 2 7(° 92 8 3] 6|R, +42,
jy alb=|2 3 3 8 -24
ee ee ee
ee

4 -3 -9 6|0
0 9/2 15/2 3] 6}.
0 oO 0 0} 0|
Choose w as
sis ten t and has inf ini te number of solutions.
-
The system of equations
is con
¢
obtain
om the secon equation, we
arbitrary. Fr
d
| l -S2- 20) \3
9 15 Or Y= 2(¢—aw—B2}=5
3 ly
2p +—s7=6-3w, 2
ae.
obtain
From the first equation, we Sy = Sx— Dw + 92—-6wed + Az Bw i

Ax = 3y + 92 -

or x= 1t+z-2w.
ons
we obtain a two pa ra meter family of soluti
Thus,
2wy/3
x=l]+z-2w and y = (4-52 —
ary.
where z and w are arbitr
1 gg] |
» 2 -2(1 5|-2]R:-9R/7
we R, -2 R, |y ~7
5: Sh 51-4] Ry +8R/7
1 a =| ) ~9
MYAVE) Fle 3 3/ 2
-1 2| 5 Ry -? ™ 1g 8-6]
13 14

1 2 -2
1 2 -2 =2
s|
«4 0 =7
0 0 -10/7|- 10/7 |"
- 10/7) Ry~ Bie *
-\59 00 —10/7| 0 0 0 0
le 6 =2/7| 2h!

Scanned by CamScanner
 3.50 Engineerin g Mathematics

The last equation is satisfied for all values of x, y, z. From the third equation "
Back substitution gives y = 1, x = 1. Hence, the system of equation has a yp; ies fain 2
y= 1 and z= 1. Since R, = (24R, — 7R, + R3)/5, the last equation is
redundant rationa)
3.4.5 Gauss-Jordan Method
In this method. we perform elementary row trans
formations on the augmented Matrix [A | p
is A Square matrix, and reduce it to the form
: ete 4
A
oa

[A b] Elementary (I |c]
——
ie
=
——
row operations

Where I is the identity matrix and ¢ is the solution vector. This reduction is Equivalent to
el
==
_

the solution as x = A™' b. The first step is same as F.

in the Gauss elimination method. From tind
onwards, we make elements below and above the pivot as zeros,
using elementary row transformay,
Finally, we divide each row by its pivot to obtain the form [I | c]. Alternately, at every Step, the “a
can be made as | before elimination. Then, c is the solution vector. ‘
eg
ee
i
aege
This method is more expensive (larger operation count) than the Gauss elimination. Hence. We dy
not normally use the Gauss-Jordan method form finding the solution of a system. However, this meth
is very useful for finding the inverse (A~') of a matrix A. We consider the augmented matrix [Al I]and
reduce it to the form

[AI AY
using elementary row transformations. If we are solving the system of equations (3.28), then we have
x =A” b, and the matrix multiplication in the right hand side gives the solution vector.
Remark 16
we interchange rows as in the Gauss
ek
eye
oii
i.ee
oe
If any pivot element at any stage of elimination becomes zero, then
elimination method.

Using the Gauss-Jordan method, solve the system of equation A x = b, where

ih=—m+
2

Example 3.30
es.
eg:

Por
er
he)
we
=) 1 -l 1 0
A=/|2 1 -3]|, and b=/4
1 1 1
itt
Solution We perform elementary row transformations on the augmented matrix and f
form [I | C]. We get
1-1 10), 4, [1 -1 1/0
[Alb]=|2 1 314) 2 “'2lq 3 -sl4/R,/3
] 1 1}] 0 2 Ol

dl
Scanned by CamScanner
 CCl

Matrices and Eigenvalue Problems

3.51

=]
at
l 0) Rar, |! 9 -2/3| 4/3
=10 1 —5/3|
22 ={0 1 -5/3]
4/3
2 |_| 4/3] R,/(10/3)
0 0
i)
10/3|-$/3
1 0 -—2/3| 4/3] | 1 0 0 i
=|Y 1 S73) 473 ‘hie 0 1 0} 1/2
0 0 R, +5R,/3 |
1} -1/2 O 0 I) =178
Hence, the solution vector is / |
x=[l 1/2 -1/2]7.
.
—| 1 2]
Example 3.31 Using Gauss-Jordan method, find the inverse of the matix A = 3 -| |
yo .
-| 34
Solution We have

=] 1217 0 0
(A|I)=] 3 -1 1/0 1 Ol.
-l 3 4/0 0 1

|
|
Dee Bak a { =f =2/-1'0 0
AID=| 3-1 1] O 10/2 o'=]0 2 7) 3.1 OfRv
~—{ & ££) G0” * OB Bot @ 1

—————
=ee iot +2)+1 50), _. 1 0 3/2/1/2 1/20
=|0 1 7/2)3/2 1/2 0 iene =|0 1 7/2/3/2 1/2 O| (-R,)/S
0 2 #2/-1 o01}/% 7 |o 0 -s}-4 -1 1
ee
ee

1 0 3/2{1/2 1/2 |. sam | 0 0] —7/10 2/10 3/10

={0 1 7/2|3/2 3/2 0 4 ee 0 1 O|-13/10 -2/10 7/10}.
O60 tla ws =)" * 00 1, 4/5 WS -1/5

ay @ 3
Hence, i= Slay 2 7
Wl gg g =

Scanned by CamScanner
 |
ematics
| 3.52 Engineering Math
1 a Vit alti ‘quations
Rhu | 3.4.6 Homogeneous System of Linear Eq
, yc syste
m of equations
| Consider the homogencous | (3.34
|(4)
| nh Ax = 0 m is always consistent since x = 0(
5 syste ,
1 matrix. The homogenous Sy" ‘therefore, for the h
i wi i

where A is an mi *
j f
= rank (A Me If rank: (A) =r - ©<n,homog be |
We Neoy.
| | |
ays 4 sol uti on. In this case, rank (A)
(A) < 7.
soluti on) is alw that rank
-trivial solut! on,y we
bid tite SS orequire
e
nsion (1 — ) ag (ny
system to have a non form a ve ctor space of dime
Bigs family of sol uti ons
he , an (n — r) parameter . “1,
:eae
an parameters can be cho sen arbitrarily. —
‘< called the null space and its dimension is¢4
| eneous system 1S Calle leg

RS
. space of the homoge" "2 =
atigAsat
at
Tere, rankwe (A)oan+ nullity
thers (A) =» (seeThevem ore3-5).
35)m
an | solution
Thee solution space Of

ni
& | |
a
Tat Remark 17.
Pi |
1 (a) If x, and x, are two solutions of a linear homogeneous system; then ’ @&Xx, + Bx, is also,
! || solution of the homogenous system for any scalars @, B. This result does not hold for
:it non-homogenous systems.
)| (b) A homogeneous system of _m equations_in_m_ Owns always p $a (iii) (4

in i} 1 non-trivial solution,
: ,
aa
ah ' : 3.8 Ifanon-homogeneous system of linear equations Ax = b has solutions, then all these
im i; 5 « « . . :
7 ' solutions are of the form x = X + x, where XQ is any fixed solution of Ax = b and x, is any solution T
|: of the corresponding homogeneous system. a fe
hi - : ; | ; a
, | Proof Let x be any solution and xy be any fixed solution of Ax = b, Therefore, we have
iat |

||
| en aan Se Exercise
ial Subtracting, we get gh
Using the
| |
| | Ax — Ax, = 0, or A(x — Xp) = 0.
] | Thus, the difference x — x, betwe os ee |
|| | Ax = b is a solution of the hatin Be! solution x of Ax = b and any fixed solution %p of
ids | ous system Ax = 0, say x,. Hence, the result.
iat Remark 18 ,
\!
gat
ay If the non-homogeneor
! eneous system Ax = b wh ere ,
1a , , re
Et that is rank (A) =n, then the corresponding ho mogeneous
Als an m system
x n matrix (m > n) has a unique solulio!
Ax = 0 has only the trivial soluttor 4, |
| that is x, = 0. |

iit Examplple 3.32 |

i
- | | Solve the followi
OW
Ing homogeneous system 0

| Gil «< e -
| :
. ” 13
I,
?
(ii)}1 )
.
Find the rank (A) and null
ity (A)

Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.53

‘on We write the augmented matrix (A | 0) and reduce it to row echelon form.
golut!

: 1 R-Ri2 |* se 2 1/0
(i) (ALO Pe —3R,/2 ~ [9 ~3/2]0] R+R/3 =]0 -3/2/0]
3 2/0 QO 1/2 0 0 010

Since, rank (A) = 2 = number of unknowns, the system has only a trivial solution.
Hence, nullity (A) =

p22 -3/0), | 2 -3]0 i 2 +3/0

(ji)ALO=]E 1 -10F
0 | =/o -1 2/0/z,-32, =/0 -1 2/0).
‘2

1-1 0;* “' [0 -3 4lo 0 0 -2/0

Since rank (A) = 3 = number of— the homogeneous system has only a trivial
solution. Therefore, nullity (A) =

bid WO, os i 1-1 1/0 i i =3 a6

(iii) (A|0)=]2 3 Le ee, 0 1 3 2{0]R,+R, ~{0 1 3 2/0
32-6 flo|” ~* lo +1 =3 —Blho 00 O Oo

Therefore, rank (A) = 2 and the number of unknowns is 4. Hence, we obtain a two parameter
family of solutions as x, = —3x, —2x,4, x) =— xX, + x; — x4 = 4x, + x4, where x, and x, are
arbitrary. Therefore, nullity (A) = 2.
IY
fas
tthf
f

Exercise 3,3 \2
ly

Using the elementary row operations, determine the ranks of the following matrices. \

20 -!l 1 —-2 3 2 1 -2
js | 2% |2 LL ZL 3. |-l -1l IJ.
01 3 -5 II 3.06UC«ds =

) LE 2 J
; ~| 6- lL -2 l- 1 3 4
23 4 5). elt a 3 a |
!
4-13 =5
- ,
at Ls xz
8 13 14

[23 ¢ 2a 2 9 =! 0
71/2 3 4 5§ o jet ft == , [ee OS]
7 Ws jot ". Bat 013 6
> AS Qh a7 rio | 6 | -2 6

Scanned by CamScanner
 ree |

|
ematics
3.54 Engineering Math
: fia

3 ¥ O 4
7
| 3 KZ =I a
a is
at ee
of
1 ly af 9-2-2
,es. .
.tric
ing ma
i
rm a
in c ea
e thP r e © nk of the follow
ra
ations, deete
' 4

coe
ementary e
ele unmn operdt
lu
Using the E
| 3 -! I} 13. | 7 i}
fo -l] 2.
eS
er nr
wu. {1 oO |
1 -| 3

4 og 0 *
| 2 FRE -2 Ly
|) 9 oy 3-2
14. ‘ i
i i
5 4 3 a l 5
§ 4,3
4
ndent. Find also its dimension.
Determine whether the following set of vectors is linearly indepe
ee
=
Se
oo

16. {(3, 2, 4), (1, 0, 2), (1, -1, -)}- 17. {(2, 2, 1), C1, -I, 1), (10,1),
18. {(2, 1, 0), (1, -1, 1), (4, 1, 2), (2, -3, 3)}- 19. {(2, 2, 1), (2, i, -1), (1 + i, iM),
20. 10, 1,1, 49, (+ i -1 -i, 2}
21. {(1, 1, 1,1), Gl, 1, 1, -1), (1, 0-1, 1), (1, 1, 0,1)}.
ga
SS
me

]
@ {1.23.1 @.1.21-16.S, §3)3.:6..% 1, 3}.
23. {(1, 2, 3, 4), (0, 1, -1, 2), (1, 4, 1, 8), G, 7, 8, 14)}. 45, |?
3
24. {(1, 1,0, 1), (1, 1, 15 1), (4, 4, 1, 1), (1, 0, 0, 19}. 4
25. {(2, 2, 0, 2), (4, 1, 4, 1), (3, 0, 4, 0)}.
Using the G:
Determine which of the following systems are consistent and find all the solutions for the consistent sys
|

26 i ale 7 V4 ix) ft 1 -4 7][x]|8 ft

46, ||
"Ty. tealleh lal oT 12S 8ilyt=|2). 28. |3 8 -2/|»]=/°}
]
a
ee
ee

1 |
29. |3 -9 2 i a : 11 [x 7 > OQ —-3lIx
31 G4 =3//7/= 0 1
5 -3 4 i : a) ae 3 YI=] 16).
is | 3 4 Bs 22 | a | is | 49, |~1
5 3 14
fs
4 | ~|
32, (9 Ft tn |. =? 1 271/* al? ‘4 ]
| l ] ? |= 0 . 33 i | | , y | l I y . a\.

: =|2|. 3 = A 35.
rt Rig,
tA
WwW | —| ] -1] Ww

)
|
\

Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.55

aLm | —
a

‘—
tt
=
Ss
3°
o
—

ad all the solutions of the following homogeneous systems Ax = 0, where A is given as the following.
Fi
1 2 Qi » . @ 3 -I) 5

3} 37.) 3 4 -71. 38. 14. 1 -10

y. |! -2 6 6
1 § =4 -l -2 1] 4

- 4 =f sy i Pe?
ow f
i wo (io i ot y |i rit
39.|1 1 | j o =? = =i! oe =f } oF
126 te 6 8 1 iia 4

31 14 Ll t 4 2 i «BP ¥ ad
04 10 1 % .f -2 -3 i { =% 4
43 44. 4 4% 8
1) 7 17 3 3 0 -5 -1
22 43 q =f =f <4 5-7 2-1

i Pg
gfii-
"ty > =f 3
422-4
matrices.
Using the Gauss-Jordan method, find the inverses of the following

i {e239 1 1
1-1
4.11 2 3]. 47. |1 3 (re
ee
has a boa ode

Proud cias
git: i ks “. 1
“I-14 4 4411
1001

35 Eigenvalue Problems
or non-singular. Consider
Let 4 = (a,) be a square matrix of order n. The matrix A may be singular
th :
* homogeneous system of equations
(3.35)
Ax=Ax or (A- AI) x =0

Scanned by CamScanner
 ics
{
3.56 Engineering Mathemat of Sie
| of order n. The
a
homogeneous system
eix
© ; r which the homoge
I is an identity matrix
Where A is a scalar and
- in} olution. We need to find valuan ® fel solutions of theone =
| (3.35) always has a trivial so” The values of OFA the
for characteristic
which non-tr values of f AA and
and th the CorTespo
a
(3.35) has non-trivial anne’ ‘saevalues ning |
| system (3.35) exis t, are called the ef -d the eigenvalues oOr the characteristic values of A. IfX js
any constant 1s also a solution a
| a non-trivial solution vector x are nag?35), then OX, where @ is
| ) non-trival of the homogencous ee s veciat ig unique only upto a constant multiple. The Probe |
i : the homogeneous system. Hence, an ig ding eigenvector of a square matnx A is calleq an
fii of determining the eigenvalues and the correspon
r | eigenvalues problem.
=
_
:=
= =<
o
-*

tt tan 3.5.1 ‘i and

an Eigenvectors
Eigenvalues 4 , ,
r _ swial solution, then the rank of the coefficient mar
=.=a
ae
SS }
ee
ee | If the homogeneous system (3.35) has a non trivial solu . sinoular. Therefore =
| |i | (A — Al) is less than n, that is the coefficient matrix eigen
| |i | “ais | <) a ¥) ; ayy A a)4 Saty Qin

eat + IRIZ o ) »-A ss a

i fH
. det(A-a =| . ose
“ 1=0. aa
Ther
1 succt
if Expanding the determinant given in Eq. (3.36), we obtain a polynomial of degree n in A, which isof 3. A-
at the form
i P,(A) = 1A — ALL = (-1)" [A" cy, @"") + cA"? — ... + (-1)"c,] = 0
‘or
if | or HAM cA + 0A" - ... + C1)", =0. (3.31)
4 AT
| : Where c,, Cree Cy CaN be expressed in terms of the elements a;; of the matrix A. This equation is called
| ! “ineg equation of the matrix A. The polynomial equation P,, (A) = 0 has n roots W ae
|i | “a = ae ao . repeated. The roots Aj, Ay, ..., A, of the polynomial equation P,(A)=

wiB | genvalues.
| Dy using the relation between the roots and the coefficients, We cam ¥™*
| K MF Me ct WO AG tithe + ea,
-——
eee
-—— —=
-_——
*—
ee
z=we
eas
Ee
NN
=
h! : "]

, . + Miho + Aydy + 0+ Any Ay= C5

!
[} ae :
| Ay Ag+1 Ay = Cy gs
| If we set A= 0 in Eg. (3.36), then we get
|
lA] = (-1)"c, —

Therefore, we get n= AA)... A,

| ——-
cin : |Sum of eigenvalues — .
a} | .
: Se ene 7 Product of ej eigeny
alues = |Al. |

Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.57

t of the, eigenvalues _is called the spectrum of A and the largest eigenvalue in magnitude 1s
The set
the spectral radius of A and is denoted by p (A). If |A| = 0, that is the martix is singular, then
called the
“Eq. (3.38), we find that atleast one of the eigenvalues must be zero. Conversely, if one of the
from Eq.
is zero, then |A| = 0. Note that if A is a diagonal or an upper triangular or a lower
eigenvalues of A. mn nF
rriangular matrix, then the diagonal elements of the matrix A are the eigenvalues
cea
eigenvalues A,’s, we solve the homogeneous system (A— A, I)x = 0 for
Afte epee_n the
to obtain the corresponding eigenvectors.
ne Ls dy *

properties of eigenvalues and eigenvectors

Let A be an eigenvalue of A and x be its corresponding eigenvector. Then, we have the following
results.
1. aA has eigenvalue aA and the corresponding eigenvector is x.
Ax =Ax=> a Ax = (QA)x.
2. A” has eigenvalue 2” and the corresponding eigenvector is x for any positive interger m.
Pre-multiplying both sides of Ax = Ax by A, we get
A Ax = Adx = AAx = A(Ax) or A’x = A’x.
Therefore, A? has the eigenvalue A° and the corresponding eigenvector is x. Pre-multiplying
successively m times, we obtain the result.
X.
3. A— KI has the eigenvalue A — k, for any scalar k and the sefeianedlithy eigenvector is
Ax=Ax>Ax-—kIx=Ax-kx
or (A — k1)x= (A - k)x.
4. A''(if it exists) has the eigenvalue 1/A and the corresponding eigenvector is xX.
Pre-multiplying both sides of Ax = Ax by A”', we get
A'Ax =AA'x or A'x = (A)x.
is x for any
5. (A - kI)"' has the eigenvalue 1/(A — 4) and the corresponding eigenvector
scalar k.
rows or by
6. A and A’ have the same eigenvalues (since a determinant can be expanded by
columns) but different eigenvectors, (see Example 3.41).
is also an eigenvalue
7. For a real matrix A, if a+ iB is an eigenvalue, then its conjugate @ — iB
A is complex, this
(since the characteristic equation has real coefficients). When the 1matrix
property does not hold. AR snE
A and its characteristic
We now present an important result which gives the relationship of a matrix
€quation.

Theorem 3.9 (Cayley-Hamilton theorem) Every square matrix A satisfies its own characteristic
“uation
C, I = 0. (3.39)
= cA"! # - a | C,.1 A rs lV
A"

Scanned by CamScanner
 _,
3.58 Engineering Mathematics

Proof The cofactors of the elements of the determinant |A— AI| are polynomials ;In Ao
(1 — 1) or less. Therefore, the elements of the adjoint matrix (transpose of the Cofactor jsdeg,
also polynomials in A of degree (1 — 1) or less. Hence, we can express the adjoin Rieat
polynomial in A whose coefficients B,, B2, .... B,, are square matrices of order ny havin ‘&
as functions of the elements of the matrix A. Thus, we can write Se“lemon,

adj(A — Al)=B,A"!+B,A"? +... + B,_,A + B,,

We also have
(A
— Al) adj (A-AD=|A-AII
Therefore, we can write for any A
(A — Al) (B,A""' + BLA"? +... + B,_;A + B,)

ET
SS
lles
i =
=A -—¢ AT1+ ... + G1)" c,.; AL+ G1)" ¢,1
Comparing the coefficients of various powers of A, we obtain
— B, =

AB, — B,= cI
AB, — B, = c5l

AB,_, — B, = C1)" ¢,_11

AB, = (-1)" c, I.
Pre-multiplying these equations by A", A"~', ..., A, I respectively and adding, we get
A"—c, A"! +... + (-1)"'c,_, A + (-1)"c, I = 0
which proves the theorem.
Remark 19
of the matrix A
(a) We can use Eq. (3.39) to find A”! (if it exists) in terms of the powers
ee
=—

Pre-multiplying both sides in Eq. (3.39) by A7', we get

Oe

AA" —c, ATA™! +... + C1) 'c,_, ATA + (-1)"c, ATE = Ao'0 = 0
| n ; (3.40)

or A'=- Cy [A”'-—¢, A"? + as +(-1;""' c, 11]

Ch

(b) We can use Eq.(3.39) to obtain A” in terms of lower powers of A as gal

A" = c,A" | = eA" me oak vs (=])" lo I.

Example 3.33 Verify Cayley-Hamilton theorem for the martrix

1 2 0
A=i/-l I 2).
12 1]

A
Scanned by CamScanner
 Matrices and Eigenvalue Problems 3.59

_ pol 3 73: : :
A’, (ii) find eigen value s of A, A° and verify that eigenvalues of A’ are
ifj) obtai n A” and os
Also of those of A, (111) find the spectral radius of A.
squares
colution The characteristic equation of A is given by

1-A 2 0 |

JA-AN=| -1 I-A 2} =(1-a) (1 -AP- 45-2 EF -A)-2)

l 2 1-A

= (1 — A) (A? - 2A - 3) -2(- 3) = - V+ 30-14 3=0.

12 0lf 12 oO] [-1 4 4
1 2/=|
-1 0 3 4
|-1
=2/|
A?71
Now,
12 11] 12 1 0

14 417 12 oO| J-1 10 12

A? =A’A=| 0 J | =| 1 Ti 20}:
s|| 1 2 1 -1 16 17

We have
14 4 i 2 Qo [1 0 0
=] 10
1 11 10} +3) 0 s-a)—|-L 1 2/)+#3/0 1 8
_a343A27-A+31=-]| 001 I
-_1 16 17 0 6 5 1 2 4
ie |
00 0 (3.42) Ap
-190 0 O|=0. N
0 0 0
A+ 3 = 0.
ic equation — a4 3 a
Hence, A satisfies the characterist

(i) From Eq. (3.42), we get

jet A
‘5 4 4 3 6 0) (1 9 0
=4}
ge 3A+I]) s lig3 a gl4lo 4 Oll=3s| 3 tb
A te)
=—[A*- =< 0 3 4
| —j 0 3
3 ilo 6 5 36 3) \o O
:
From Eq. (3.42), we get
0] [3 0 0 | {6 42
% 42) i 1 2
? 1 2/+|0.3 O}=] } ?! 10
geap =| 0 9 12[ -}-
O~942 «|; 16. 4%
15] 12 1) |0 0 3
0 18

Scanned by CamScanner
me on 3.60 Engineering Mathematics

| (ii) Eigenvalues of A are the roots of

Theis a3 _ 9924-3 = 0 or A—3)4° + =O Or A= 3,5 -7

my ia! : The characteristic equation of A? is given by

HBL! -l-A 4 4 | |
4 =(-1-A [3-4 (5-4) - 24] = 0
ig Sigees 0 3-A
tg | to 0 6 5-A

a | or (A+ 1) (A2-8A-9)=0 or (At 1) (A-9) At I= 0.

The eigenvalues of A* are 9, — 1, — 1 which are the squares of the eigenvalues of 4

at (iii) The spectral radius of A is given by

ii p (A) = largest eigenvalue in magnitude = max |A,| = 3.

1 0 0
Example 3.34 If A=|]1 0 1], then show that A" = A"-? + A? —I for n = 3. Hence, find4®
tt 0 1 0 |
ai
a| Solution The characteristic equation of A is given by

_ 1-A O 0O
ay AK AT|= 1—-A 1i/=(-AGQ?2-1)=0, or V-A-At1=0
| 0 1-A

| Using Cayley-Hamilton theorem, we get

i 4

| A>-A?-A+I1=0, or AX-A?=A-T,
: Pre-multiplying both sides successively by A, we obtain

7 Ae AR]
oF | MASAt A
: i } AN! = A" - = A’-3 _ A"

A" be Al! = A"? _ A

Adding these equations, we get

A" ire A- = A’? ” L, or A” — A"? +- A; is I, il = 3.

Using this equation recursively, we get

Scanned by CamScanner
 ol Jf £f LF MOL &

Matrices and Eigenvalue Problems 3.61

A" = (A"4
+ A? -1) + A? - 1 = A + 2(A?
- 1)
5
1) - A” O 4 3(A2 — 1)
= (A" G4 A? -_ 1) + 2(A*

| i
= atl = =nN A ,
Cn — 2) ) ( (A? - 1)1s
4 25 (ni =5| (2 - 201.

cpstinuting # = 50, we Bet

1 0 0 1 0 0 1 0 0
A™ = 25A? -241 =25]1 1 O}-24/0 1 O}=|25 1 O}.
1 0 1 ) 2 1 25 0 1

example 3. 35 Find the eigenvalues and the corresponding eigenvectors of the following matrices.
4 7 1 0 0
l . aoe 2 I}.
(i) A -| | (1) A = ‘ (G)A=|}0
o 4] 2 0 3

Solution
(i) The characteristic equation of A is given by

|A-AI|= see
32-4 =0 or A
AeA“-3A-10 0 =0, or =— 2,5 5.
A=-2,

Corresponding to the eigenvalue A = —2, we have

3. 4)\/ x, 0 4
‘A + +21) x = i ‘i
(A = or x, =-
r 3x,x + 4x,x3 = 0 or x, —%X.
3 X>

Hence, the eigenvector x is given by

_ ial=7 ee _.~ f-473

x= ag |
X9 ey | ]

Since an eigenvector is unique upto a constant multiple, we can take the eigenvector as [- 4, 3]”
Corresponding to the eigenvalue A = 5, we have

(A —- 5] )x== i 3 =. | i
x) mi
\0 or |)
x, —X,
“9 = 0,, orx,
| =x“Fe

Therefore, the eigenvalue is given by x = (x), 42)’ = x,(1, 1)’ or (1, 1)’.
(li)
The Characteristic equation of A
is given by

ae | -A I=
0 or Aes 9442 = 6, or A=1 4i.
Aa | ~t La

Scanned by CamScanner
7
 cs
3.62 Engineering Mathemati
|]t+i, we ; we have
the ei- genvalue A
=

Corresponding to

rA-(1+)1)X=1_y ; |
=i]mJ Le
! | ; ‘ A
F:
>

and — x) ~ 2 = 9 tee ae
— 1x + xy =
or 1, we get x; = -t. Therefore he
: — x, — 1x 2 = (0. Choosing *2 ~
to — XY
Both the equations reduce
-
eigenvector Is X = [-i, I]
= | —i,-we have
alue A =
Corresponding to the eigenv
i fx] _]9]
[A — (] = HL x= * i ~ A
ial

or
— x, + IX,= 0.
ix; tx= 0 and soution

Both the equations reduce to —x, + 1X2== (), Choosing x, =

1, we getx, = i. Therefore, the eigenvecty
iitisim
mpka
is x = [i, 1)’.
Remark 20

For a real matrix A, the eigenvalues and the corresponding eigenvectors can be complex.

i (iii) The characteristic equation of A is given by

|9 | |A-AI]/=/|
I-AO
0 2-aA
oO
1 /=0 or (1-4) 2-AG=-A=0 or A=1,2,3.

| 2 0 3-A
ee
er
—=
aeoe

Corresponding to the eigenvalue 2 = 1, we-have

|
oe

,
'| (A-Ix=]0 1 Illy, |lol or f + x, =0

: —
oe
a
=
——
:

We obtain t
20 2 xX 0 Xx; =0°

| atiracil fas Equations in three unknowns, One of the variables

ty = l, we obtain the ei |
x,,
“Sia
x5,
<}34
x, can be chose"
- 3

:
y. a Ing
i
a
Eenvector as [-1, =|. iy
Corresponding to the eig
envalue 4 ==
le
es
i

0
————————
aes
=a
5
2, we have
—] Q) 0 | x
|
(A -21)y = 00 | 7 _
7
SS
SS
ee

(ij)
| =O 4 *3 0
or PAx, =0,Boks
%.=O0O 8 d x, arbi“i
gt
* Ocan
-=Sear

trary. Taking i. =
—————
ay. | Ti.
*2= 1, we obtain the eigenvector as [0, I
Th

Scanned by CamScanner
 Matrices and E igenvalue Pro
ON
blems 3.63
Corresponding to the eigenvalue Q = 3, we
have

=2' 0 0 x, ||0
(A —- 31)x= 0 1 hx ilo or a=
20 IIx, lo ie ies
Choosing x; = 1, we obtain the eigenvector as if 6.4 17

Example 3.36 Find the eigenvalues and the corresponding eigenvectors of the following matrices
1 | 0 1 1 0 100
(i)A={0 1 ] (ii)A=|0 1 Ol. (iii) A=|0 1 Ol.
0 0 ] 00 1 001
Solution In each of the above problems, we obtain the charac
teristic equation as (1 - A) = 0,
Therefore, the eigenvalues are A= 1, 1, 1, a repeated value. Since a 3 x 3 matrix has 3 eigenvalues
it is important toKnow, whether thegiven matrix has 3 linearly independent eigenvectors,orit has
“|esser number of linearly independent eigenvectors

Cormesponding to the eigenvalue A = 1, we obtain the following eigenvectors.

0 1 0 0 X» =0
(i) (A-I)x=|0 0 1|x=|0] or x,;=0 f
00 0) lo)’ |x,arbitrary. h
Choosing x, = 1, we obtain the solution as [1, 0, 0]7.
Hence, A has only one independent eigenvector. — |

0 ] 0 x) 0 | De Lf t
“* x, =0 ah eal AEA
") (A-I)x=]0 0 O|}x,}/=/0], or
x,, X, arbitrary.
0 0 O||x;} {0 —*
Taking x, = 0, x, = 1 and x, = 1, x; = 0, we obtain two linearly independent solutions

x, =[0, 0,1)", x, =[1, 0, o]’.

In this case A: has two linearly independent eigenvectors.

ai) (A-—1)x=|]0 0 O0||x, {=| 0}.

0 0 OF] %5 0]
This system is satisfied for arbitrary values of all the three variables. Hence, we obtain three
linearly independent eigenvectors, which can be taken as

Scanned by CamScanner
 3.64 Engineering Mathematics

x, =[1, 0, 0) ¥2 = (0, 1,0), x5 = [0, 0, 197

We now state some important re sults regarding the relationship between the
eigenvalues : f
Fo. He
linearly igenvectors.
independent e1g 4 atMay,
and the corresponding
A 1. Eigenvectors corresponding to distinct eigenvalues are linearly independent.
*2, If A is an eigenvalue of multiplicity 2 of a SUAKS mann A of order n, then the num
associate d with A is given by berg
_. linearly independent eigenvectors
=
e
p=n- I where r= rank (A - Al), I< PSm.
ne
=ee

Remark 21
In Example 3.35, all the eigenvalues are distinct and therefore, the corresponding Cigenvecton |
linearly indenpendent. In Example 3.36, the eigenvalue A = | is of multiplicity 3. We fing that
(i) Example 3.36(i), the rank of the matrix A — I is 2 and we obtain one linearly Independey
eigenvector. | | |
(ii) Example 3.36(ii), the rank of the matrix A — I is | and we obtain two linearly independs
=.
5i,
ee
a

eigenvectors. | |
linearly independ
(iii) Example 3.36(iii), the rank of the matrix A —I is 0 and we obtain three
eigenvectors.

3.5.2 Similar and Diagonalizable Matrices

Similar matrices
The matrix A is said to be similar to the mam
Let A and B be square matrices of the same order.
B if there exists an invertible matrix P such that
A=P'BP or PA=BP. m
Post-multiplying both sides in Eq. (3.43) by P-', we get
PAP '=B. : |
if B is similar to A. The matrix P is called the sia
Therefore, A is similar to B if and only tion. We now prove #
in Eq. (3.43) is called a similarity transforma
repeat.
.

matrix. The transformation

ay
oe
oe
SH
se
eee
ee
a2a
fee
ae
age

regarding eigenvalues of similar matrices. sit

enc e the
nen ‘1, ist
h

characteristic equation (and

= z

Similar matrices have the same

7 my a

Theorem 3.10
of A corresponding to the eigenvalue :: 2
"il
eigenvalues). Further, if x is an eigenvector
P is the similarity ma
er,

eigenvector of B corresponding to the eigenvalue A, where

t is
i nvalue and x be the correspondin: g eigenvector of 4. Tha
:
ee
—_
2S
ee
el
a
=eae
4 Sa
— =sa 4see
fad —*
= oy .———
-‘
Proof Let A: be an eige
e
=
Ax = Ax.
matrix P~', we obtain
Pre-multiplying both sides by an invertible
P 'Ax=AP 'x.

Set x = Py. We get

_—-
Scanned by CamScanner
rr

Matrices and Eigenvalue Problems 3.65

| Sia |
PUAPy= AP 'Py, or (P''AP)y
= Ay or By =
Ay.
-1 : :
where B= pAP Theslore, 8 has the same eigenvalues as A, that is, the characterstic equation of
, similar
is same as the characteristic equation of A. Now, A and B are similar matrices. Therefore
s have the same characteristic equation (and hence the same eigenvalues). Also, x = Py, that
matricice f A and B are related by x = Py or y =
eigenvectors 0 P™'>.

pemark 22
(a) Theorem 3.10 states that if two matrices are similar, then they have the same characterstic
equation and hence the same eigenvalues. However, the converse of this theorem is not true.
Two matrices which have the same characteristic equation need not always be similar.
(b) If A is similar to B and B is similar to C, then A is similar to C.
Let there be two invertible matrices P and Q such that
A=P'BP and B=Q''CQ.
Then A=P 'Q'CQP=R''CR, where R=QP.
eee
-

Example 3.37 Examine whether A is similar to B, where

“pe

yl tes 1 2] aydoe | 8) ort elt 2

Yl 9 "9 4 4 lg yl ™ 01
=—_ : an — . = = s

Solution The given matrices are similar if there exists an invertible matrix P such that
A=P 'BP or PA=BP.

Let P = ° ‘|We shall determine a, b, c and d such that PA = BP and then check whether P is
Cc
non-singular.

§ 5 " 1 2ila @ : 5a-—2b Sa . a+2c 6+2d |

‘i | a bl|
c dil-2 ol 1-3 alle al’ - |5e-2d Sc| |-3at+4ce -3b+4d
ns
Equating the corresponding elements, we obtain the system of equatio
Sa —-2b=at+ 2c, or 4a-—2b-2c=0
5a= bh + 2d, or Sa-b-—2d=0

Sc — 2d = - 3a + 4c, or Ja+c-—2d=0
5c (| | al > + & a or 36+ 5c —4d=0.

A solution to this system of equations is a = 1, 6 = lle=l,d=2.

A and B
Therefore, we get P = ; i which is a non-singular matrix. Hence, the matrices
i 1 3 4

are similar.

Scanned by CamScanner
3.66 Engineering Mathem
atics
_.
oo Pa Fad Panpa a ea
Equating the corresponding elements, we get
Q@=a+c, b=b+d or c=d=0.

Therefore, we get P = i | which is a singular matrix.

Sincan
e invertible matrix P does not exist, the
matrices A and B are not similar
It can be verified that the eigenvalues of A are 1, 1 wherea
s the eigenvalues of—_ 0
In practice, it is usually difficult to obtain a non-singular matrix P which
4 2.

satisfies the equation 4 — ps

for any two matrices A and B. however, it is possible to obtain the matrix
P when A or B is oa :
matnx. Thus, our interest is to find a similarity matrix P
such that for a given matrix 4 We - -
; ve
D=P!AP or PDP'=A
Where D is a diagonal matrix. If such a matrix exists, then we say that the matrix A js
diagonalizable.

Diagonalizable matrices
A matrix A is diagonalizable, if it is similar to a diagonal matrix, that is there exists an invertibk
matrix P such that P“' AP= D, where D is a diagonal matrix. Since, similar matrices have the same
eigenvalues,/the diagonal elements of D are the eigenvalues of A: A necessary and sufficient conditea
for the existence of P is given in the following theorem. —

Theorem 3.11 a square matrix A of order n is diagonalizable if and only if it has linearly indepen®™
eigenvectors.
A has n linearly independent ee
Proof We shall prove only the if part, that is the case that if eigenvectors seat :
ly independent
then A is diagonalizable. Let x,, x2, ..., x, be m linear at 15
order,
the eigenvalues A,, A, ..., 4, (not necessarily distinct) of the matrix A in the same
eigenvector x; corresponds to the eigen
;
values3 A, jae= 1, 2, ...,
fs \
nm. let vo - - 4
i

of< / P= [x,, gg cacy x,,] |and( D = diag (A,, Ax» re An) : odo
—h ae ee a jy ' ae att —_— d the

be the diagonal matrix with eigenvalues of A as its diagonal elements. The matrix P is calle
matrix of A and D is called the spectral matrix of A. We have
AP =A [X,, X>, anny x,,] = (Ax, AXx,, sen Ax,,) a

= (A,X), Ao&, ‘tek Agh,) @ Gigs Ras ging x,) D = PD. ag P

fore the mat

endent, the rank of P in n and there
Since the columns of P are linearly indep
both sides in Eq. (3.44) by P~', we obtain
(a45]
invertible. Pre-multiplying
P-'AP = P-'Pp =D

__—
Scanned by CamScanner
 f ff .t Ff € FF GE E£

Matrices ‘and Eigenvalue Problems 3.67

a matrix A to
mplies that A is similar to D. Therefore, the matrix of eigenvectors P reduces
hich 1
diagonal form. ; — , .
its -qultiplying both sides in Eq. (3.44) by P ’, we obtain
post- A= PDP". (3.46)

k 23
independent eigenvectors when its
'

at A square matrix A of order n has always n linearly

independent eigenvectors
even
9 .senvalues are distinct. The matrix may also have n linearly no
Example 3.36 (i11)). Therefore, there is
yn some eigenvalues are repeated (see
.
of the matrix A in Theorem 3.11.
restriction imposed on the eigenvalues
(b) From Eq. (3.46), we obtain
A? = AA = (PDP™') (PDP ') = PD’P’.
t-multiplication) m times, we get
Repeating the pre-multiplication (pos
A" = PD”P |! for any positive integer m.
so is A”.
Therefore, if A is diagonalizable,

(c) If Disa diagonal matrix of order n, and

TA, 0 a”

Dp = Ar ;. _ then D” =

A 0 qn
get
) is a polynomial in D, then we
TRAE
yp
4

int ege r mm. If O(D

es
=

for any pos iti ve =

——e

0
O(A,) i
——

O(A2) iy | at
a

Q(D) =
0 OA)
izable. Then, we have
Now, let a matrix A be diagonal
and A”=PD'P
m p-!
A =PDP-!
we obtain
for any positive integer Mm. Hence,
-1
Q(A) = PO(D)P
O(A).
for any matrix polynomial

Example 3.38 Show that the matrix

3 1 =!
A=|-122
O1 2

i
Scanned by CamScanner
 atics — 3
3.68 Engineering Mathem
i . that P'AP 15 2 diagonal matrix. Then, obtain the Matriy yer
\: is: diagonalizable
ale . Hence, find PB=suchA? ‘ 5A + 31 . |

| ‘< given DY
| Solution The characteristic equation of. AES
geht Sl ,
anaul=]2 14S 23 ga
23 6A-+ 1IA-6=0, Or Az] =
Lui 9 —

‘ccdise cil three linearly independent eigen, |

Since the matrix A has three distinct eigenvalues, it has SENVECtors and
hence it is diagonalizable. )
The eigenvector corresponding to the eigenvalue A= 1 isthe solution of the system 1é
~ | dia
2 1 -l/|% 0 I js dE
(A-I)x=|-2 0 2] {x |=] 0}. The solutionisx; = )-1], soluti¢

| The eigenvector corresponding to the eigenvalue A = 2 is the solution of the system

(A -—21)x=|-2 -1 2]|x,|=]0/)]. The solution is x, =]0|.

0 1 Of} x, 0 i
The eigenvector corresponding to the eigenvalue A = 3 is the solution of the system

| 9 Fly 0 0
(A-3I)x=|-2 -2 2 x2 |= |0]. The solution is x, =| 1}.
0 1 -I{lx,{ |0 I
Hence, the modal matrix iS given by

Xt, Xp, x3] = |=] 0 Il] and P!'=/]2 1 -l

| 1 | = 0 |

| Pee beverified that FAP x uss (1, 2.oe9, ciif

| We have D = dj |
ay
o “Therefore,
98 A?(1, 2, 3), D?= diag (1, 4, 9)
=
a + 5A + 31 = pip? +
5D + 31)p-!,

N Ow, 2 1=10
10 0
D'+SD+3 4 gl, : 0) 3 0 0 9 0 0

00 9 oo tle 3 Of=/0 17 9
5} [0 0 3] Jo o 27,

Scanned by CamScanner
aa
Matrices and Eigenvalue Problems 3.69

| ence, we obtain

, 1 1 o1/9 oO olf-1 -1 i] f 25 8 3
A-+5A+3I=/-1
0 1//0 17. O}] 2. 1 -1/=/-18 9 18].
1 1 is0 o 27/|-1 0 1 28 19

example 3.39 Examine whether the matrix A, where A is given by

A 122 — 2 «3
(JA=] 0 2 1, (ii) A=] 2 1 -6],
-1 2 2 -1 -2 0
is a diagonal matrix.
is diagonalizable. If so, obtain the matrix P such that P-'AP
Solution
by
(i) The characteristic equation of the matrix A is given
1-A 2 2
|A-AI| = 0 2-A l
—] 2 2-A.

=a) = 2-2-2 -H1-G-Be@- P-AW*.

=(1-A) a —
eigenvalue
A = 1, 2, 2. We first find the eige nvectors corresponding to the repeated
or
A = 2. We have the system

0 0 1]} x, |=|9}
ee

(A —21)x=|
as
i
ak
P=
‘eo
4
#

We
cie nt mat rix is 2, it has one lin early independent eigenvector.
Since the rank of the coeffi t™ing to the eigenvalue A = 1. Sin
ce
obtain another linearly indepe ndent eigenvector correspLn ond
two lin ear ly ind epe nde den t eigenvectors, the matrix 1s not
the matrix A has only
diagonalizable.
the matrix A is given by
(ii) The characteristic equation of

—2-A 2 =3
=0, or A=5, —3, -3.
_ 214-45
= 2 1-A -6 =QorA+A
|A-Al|
‘
| —2J =—A

system
to the ei ge nv al ue 4 = 5 is the solution of the
g
Eigenvector correspondin
Sa? | 2 —3 Xx 0

2 —4 -6 x, |=] 0}.
(A — 51)x = [¢
_1 =2 ~SilaJ

Scanned by CamScanner
 3.70 Engineering Mathematics
T
A solution of this system is [1. 2,1]
system
nd in g to 4A = -3 are the solutions of the
Eigenvectors corres po
x 0
2 —3
]
2X _ 3x3 = 0.
=|0| or Xx 2
2 4 —6 X3
(A + 3] )x =
xy 0
—2 3
—|

matrix is |. Therefore, the

system has two linearly jnde
The rank of the coef fici ent
to find two linearly INdepen
eigenvectors. We use the equation x, + 2X7 ~ 3x3 = 0
: ector
[~2, 1, O}” and takin B 4,54
eigenvectors. Taking x; = 0, %2 ~ |. we obtainr the eigenv
given 3 xX: 3 matrix has thre @ line.”
x, = 1, we obtain the eigenvector (3, 0, 1]. The “es modal matri,
the matrix A is diagonalizable. The
independent eigenvectors. Therefore, h
given by

29. 3 12 -3

and Pol = =|-2 4 6}.

pPp=|2 10 8
12 §
={ Q@ 1

— 3).
It can be verified that P"'AP = diag (5, — 3,
nvalues |, 1, 3a
Example 3.40 The eigenvectors of a3 x 3 matrix A corresponding to the eige
. Find the matrix A.
[1, 0, -1]”, [0, 1, -1]’ and [1, 1, 0)’ respectively
Solution We have

| 1 01 1 0 0
1] and the spectral matrix D=|0 1 O|.
modal matrix P =| 0 1
6 00.3
7 »s bef —

I -1 -]
p'= ri ft :
We find that
1 1 1

7 1 oui o of I -! -|
= o 1 ilo 1 off-1 1 7!
re
ee
ert
ee
ioo

Therefore, A=P DP"

-1 -1 olfo o 3}/ 1 !

Scanned by CamScanner
 ll Js li hlly, MmmgyyyLS--

Matrices and Eigenvalue Problems 3.7

53 special Matrices ©
dy the properties of the eigenvalues and
.< section, We define some special matrices and stu
lications in many areas. We first give some
, se matrices. These matrices have app
lal = i
= . *

ector: tors of the

:
1

¥*
genven
‘ |

eig
| | |
»fpnitions. ion
s of dimens 7 in IR” or C’. Then
-et (X), Xa. 28s “y and y = (Vj, V2. «++, y,)" be two vector
. define the following:
1 = IV |" “74

aed
uct) of vectors Let x and y be two vectors in IR”. Then
jnner product (dot prod
x,y; (3.47)
xiy= x'y =>
i=|
?
r. The inner product is also denoted
= ¥
ner product of the vectors x and y and is a scala
is called the inner thatx+x 20 and x - x = 0 if and only if x =
Note
9.
-> [In this case X * y=y-x.
by < %5 ¥~:
is d defined as
r product of these vectors Is
ifx and y are in C”. then the inne
5 n -
n

x * ¥ =y’ Y= dx, and ysx=yx™ Dix

=I
i
——
————— i=]
—
¥ Note that x» y = y° *-
ugate vectors of x and y respectively.
where x and y are complex conj
it can be easily verified that
(ax + By): z= a(x+ 2) + BY +2
for any vectors x, y, z and scalars a, B.
éFf
Se

Length (norm of a vector) Let x be a vector in IR” or C". Then

? 2
= yxex =yap $39 Fe tn
Ix]
is called the /ength or the norm of the vector X.
et
|T
eh
sky

|x || is always
x/||x||
t vec tor The vec tor x is cal led a uni t vector if ||x|| = 1. If x 4) 0, then vec tor |
ai
a unit vector.
hogonal vectors.
vect ors x and y for wh ich x» y = 0 are said to be ort
(Orthogonal vectors
The
which
Orthonormal vectors The vectors x and y for
||x/= L; tel = 7
x*y=0 and
are
and 7 6% 3 == Q, then x/||x|| ‘s y ly
vectors. If Xs y are any veclors
ae called orthonormal

*rthonormal. For example, the set of vectors

1} (0) (0 eal
set in IR’.
i) }O},) 1), |0] form an orthonormal
0) (0) \l

Scanned by CamScanner
a
 s
3.72 Engineering Mathematic
37/5) (-—4i/5 "
%
3)P
(-47) ( 0
.
sn orthogonal set in C3 and | 47
. 3 4i/5 . eH
3i/5 5

Gi) |4ih | i} Q | form 0 (+i 5

Oo} \ oO} (ltl

a
form | an orthonorma1a]
set in C. x, be n vectors in IR” The
m of vectors
Let XY. X>; eer a | N

» this Ai
Orthonormal an d unitary syste | system of v ectors, if
O7 thonuorma
set of vectors forms an
i

, 0, i#/J
Ee yr ery Vy, fay.
ary system of Vectors
x.MY ben vecto rs in C”. Then, this set of vectors forms an wrtit
Diet ity Ray nen 0

0, i#/J |
Tie eee Was ae |PR
In section 3.2.2, we have defined symmetric, skew-symmetric, Hermitian and Skew-Hemmitigy
matrices. We now define a few more special matrices. where
; | ceiged os a? follo¥
Orthogonal matrices A real matrix A Is orthogonal if A '= A’. A simple example is matri
cos@ -—sin@ requil
A=! . rows.
TS
ae,
SG
= sin@ cos@
n bet
A linear transformation in which the matrix of transformation is an orthogonal matrix is calleda subse
orthogonal transformation.

Unitary matrices A complex matrix A is unitary os satisf

A! = (A)’, or (A)! = A’. If A is real, then
unitary matrix is same as orthogonal matrix. A iN aT
A linear transformation ini which the matrix of transformation is a unitary matrix is called a unilay Cons
tranformation.
We note the following:
1. If A and B are Hermitian
a, B, since matrices, then @A + BB is also Hermitian for any real scalas Let ¢
(@A+ BB)’ = (@
+ABB)’ = aA’ + BB" =aA+
BB
2. Eigenvalues and elgenv
ecto rs : | )
of A, since of A are the conjugates of the eigenval
| ues and eigenvec'™
—P
AR = Ax gives AX = 7x.
3. The inverse ofa unit a oO
Let B= A-!. Then TY (orthogo nal) matrix: is unitary (orthogonal). We have | alesqr

Bo!
=A = (A7)-

“(AYP = (ast = BP

Scanned by CamScanner
>
Matrices and Eigenvalue Problems 3.73

piagonally dominant matrix A matrix A = (a,) is said to be diagonally dominant, if

n

la,;| 2 »? —|a,|, for all i.

J=hery

the system of equations Ax = b, is called a diagonally dominant system, if the above conditions are
nd the strict inequality is satisfied for at least one i. If the strict inequality is satisfied for
satisfied a ‘ ‘
all i, then it is called a strictly diagonally dominant system.

Pe rmutation
matrix A matrix P is called a permutation matrix if it has exactly one | in each row
and colum n and all other elements are 0.
Aof® matrix :Let
property Aofa 7 matrix. Then, the matrix B is said to satisfy the property
B be a sparse
s
4. if and only if there exist a permutation matrix P such that

PBP’ = ins =

Ay, Ar
interchanges
where A, and A. are diagonal matrices. The similarity transformation performs row
become diagonal
followed by corresponding column interchanges in B such that A,, and A,
to the
matrices. The following procedure is a simple way of testing whether B can be reduced
the interchanges of
required form. It finds the locations of the non-zero elements and tests whether
required form. Let
rows and corresponding interchanges of columns are possible to bring B to the
Let there exist disjoint
n be the order of the matrix B and b,, # 0. Denote the set U = {1, 2, 3, ..., n}.
off diagonal elements
subsets U, and U, such that U = U, U U2, where the suffixes of the non-zero
Then, the matrix B
b, 0, i#k, can be grouped as either (fe U;, ke U,) or (i € U;, k € U,).
satisfies property A.
—2 1 O
Consider, for example the matrix B=; 1-2 I}.
0 1 -2

1 0 Olf-2 1. Of}f! O 0
Then, PBP’=|00 I|| 1 -2 J{/0 0|
01 0]/ O 1 -2{,0 1 0
} So | tJ
So

-/oo 1/| 1
S

1 -2J/=]
—f 0 ff
0 -2
[te 4)
t 1] (-2) A>, Ar? |
So \ th
— S

Scanned by CamScanner
 E . } atics + | im '
| .
llowed by an interchange of il
|

‘ma tri c at,

c.
- ona l 3 fo
*

are di ag on al -m a a 3,
where A,, and Az intterchang Inyge rows: 2 and Ny)
equivalentto in
i)

transformation is — |
and ce | L/ 2 E U3: a31 ey | :

U,r = tl. 35. U2 = 12h exist Such

i E i = ‘ x : 2! CX Tr 4

Now, a, # 9. 1 =
(,, 2, 3 @* ii.‘ 2os U,. Subsets we
ae 1403 € Gs
= U; UW U3. Hence,
U = i, 2, 3}
. ‘ > import ant result
s.
a aT -
We now establish some '™P rly independent.
s 1S jinea
orthogonal setsf ¢O r vector aa
3.12 An
Therorem ider the Vey
set © that is x; + Xj = 9, ##J- Cons
Let X). *2 - x,,, be an orthogonal
Proof = ( ,
(3.44
ee
Oe

x = OX + O4X_ To + OinXm . | F E . | ;

—
=2ees
Stee a _ -_
, —

x tn Eq. (7.48) withy,

, ‘a g the inner product of the vector
Takin
M,, are scalars.
:
Oh. so 85
ean
where
» 3 = 0
|. =
xy 0 xy
ve gel + 2 aa On Xm)
xox) = (4X1
_
*

|:
-_— — O
X1/
& xii
a

6 (X| -x,)=0 or ner products of x with Xz, X3...01,

or
taking the in x,, 1,
= 0. Similarly, the set of orthogonal vectors
a

' te [2 > get @

Since Ix ill FY, WE E : = Hy ~ = 0, Therefor e,
find that Q
+ = Ay
successively, WE
pendent.
oi

-+a
.
X,, 15 linearly inde
e eigenvalues of
Theorem 3.13 Th
a

matrix are real.

(i) an Hermitian imaginary.
-H en mi iti
tl an mat rix are zero OF Ppure
(ii) a skew
I.
tt x are of magnitude A. We have
(iii) an unitary ma env ect or 0 f the matrix
din g eig
: d x be th e correspon
Fe
=

eigenvalue an
Proof Let 1 be an
, we get
lying both sides by x’
Se
hk
a
Ax = Ax. Pre-multip (3.43)
ia
pas! pied

x OF
i

Ax=Ax
Sr

x’
—

or x X ISs i way
scalars. Also. i the denominat
——
ce
TS
So
be
te

xX AX an d x’ r x are
that x7
Note = i
by the scalar x AX.
Therefore, the behavior of A is governed
_ Tr

A =A’. Now,
‘ .
7

that is
Pt

(i) Let A be an Hermitian matrix,

= x’ Ax=x"AT¥=(x'A rx)? =x" Ax

©—
=
;

(x’Ax)
is real. From Eq. (3-49)
since x’ A’ Xx is a scalar. Therefore, x’ Ax
real.

is A? = —A. Now,
(ii) Let A bea skew-Hermitian matrix, that

(x’Ax) = x? AX =—xA'X = -(x7A'X) = —X'AX

Scanned by CamScanner
 re e——e—e——————eee

3.75
Matrices and Eigenvalue Problems

| |
Thl= ics a scalar, =T a. : zero or pure imaginary. From Eq, (3.49), we
sinc e X A X 18 Ther efor e, x’ Ax is
ima ary.
gin
conclude that A is zero or pure
(A)’. Now, from
(ii) Let A be an unitary matrix, that is A! =
Ax =Ax or Ax =AX (3.50)

T sats =
or X7 AT = AX.
seen =
— (Ax)? =(4x7)
x7A| = Ax’. (3.51)
a
, we can write
Using Eqs. (3.50) and (3.51)

(x7 AT) (Ax) = (AX7)(Ax) = [AP ex

or xox = Al)? x’x.
result.
have x’x + 0. Ther efor e, |A|’ = |, or |A| = 1. Hence, the
Since x # 0, we

Remark 24 of
From Theore m 3.13, we
conclude that the eigenvalues
real.
2L(i) a symmetric matrix are
are zero or pure imaginary. iS
Adil) a skew-symmetric matrix pairs.
are of mag nit u de 1 and are real or complex conjugate
4 (iii) an orthogonal matrix unitary
unitary matrix form an
d also row vectors) of an
Theorem 3.14 The column vectors (an
| -
system of vectors. Phen \y
.
x of order n,
“
with colu mn vectors X), Ks,,.0c09 Kye \

Let A be an unitary matri

“

Proof \
- =v. —T a
= KX, XX X) Xn
“i —T —. =k 7
are X,X,,
rey X5X, X5X>5
: —T X - a
A'A= A'A= - [X), Xoo oes £,| =
af. oF <T
: x, X| X,X%2 ++ X, Xn
a

. 0, 1#J
Therefore, x) x,= } !
iT lh, b=J. o
e the inverse ofan
unitary matrix is als
sys t em . Si nc
s of A form an unitary s of A, we conc
lude that the
Hence, the column vector ate of the row
s of A-! are the conjug
an unitary matrix and the column ;
rOW vectors of A also form an unitary system
Rema
rk 25 of an
vectors (and also the row vectors)
that the column
3.14. we conclude
(a) From Theorem orm al system
of vectors.

y
orth ogon al matr ix form an ort hon

Scanned by CamScanner
 3.76 Engineering Mathematics
—
(b) A symmetric matrix of order 1 ha s n linearly independent eigenvector. and
n
> diagonlizable. Me
r ame elon
* Example 3.41 Show that the matrices A and A ae the sc “isenvalues and fa, d
eigenvalues the eigenvectors corresponding to A and A" are:mutuslly orthogonal, th
Solution We have
1A — AL = |(A2)? - AIT = |[AT— AN] = |AT- a0,
Since A and A’ have the same characteristic equation, they have the same eigenvalues
Let A and pu be two distinct eigenvalues of A. Let x be the eigenvector Corresponding b
(i:
ees
-oF
2

-
eigenvalue A for A and y be the eigenvector corresponding to the eigenvalue y for 47 We
ee
eee
ST

Ax = Ax. Pre-multiplying by y’, we get *

y' Ax= Ay'x. 85)

\T Te wget
We also have A’y = py, or (A’y)’ = (uy)’ or yA =HYy.
Post-multiplying by x, we get
uy'x (3.53)
y’Ax =

Subtracting Eqs. (3.52) and (3.53), we obtain

q (A - p)y’x=0.
¥,
yo.a
orthogonal.
~ Since A # HM, we obtain y’x =). Therefore, the vectors x and y are mutually

3.6 Quadratic Forms

ic form is an homogeneous
Let x = (X), X2) «++ x,)’ be an arbitrary vector in IR". A real quadrat
expression of the form
n nt =)
(3.54)
O=> 2 aj; X; xX;
o_o
eS
ee
i
i=| j=l
.

in which the total power in each term is 2, Expanding, we can write

ee y= ay\x° + (Ajo + @o3) Xy Xp +... + (Ayq + Bn) FU%

ee
-
on
ati

> x
+ 459 x5 7 (a>, + a>) X54 Xy T sca SF (43, + a,2) X34n
“
T wee Anny P Xp asi

ee
ae
—T = x’Ax
x B= (b,)
set by = (a4, + a,/2. The matri a
using the definition of matrix multiplication. Now,
+ a), Hence, Eq. (3.55) can be wtitte
i,
symmetric since 5, = b,. Further, b, + 6, = a;
O= x'Bx
et
Pe
as

+ a,,)/2.
where B is a symmetric matrix and 5; = (a,
we

A
Scanned by CamScanner
 Chapter 3

Matrices and Eigenvalue Problems

Re ge eer ail etl eee ee ee ee ee
ecu

31 Introduction
in modern mathematics, matnix theory occupies an important place and has applications in almost all
branches of engineering and physicai sciences. Matrices of order m x n form a vector space and they
define linear transformations which map vector spaces consisting of vectors in IR" or C" into another
vector space consisting of vectors in IR” or C” under a given set of rules of rules of vector addition and
scalar multiplication. A matrix does not denote a number and no value can be assigned to it. The usual
rules of arithmetic operations do not hold for matrices. The rules defining the operations on matrices
are usually called its algebra. In this chapter we shall discuss the matrix algebra and its use in solving
linear system of algebraic equation Ax = b and solving the eigenvalue problem Ax = Ax.

3.2 Matrices
Anim X n matrix is an arrangement of mn objects (not necessarily distinct) in m rows and n columns
in the form 4

he. Se
an, 4 On
ae (3.1)

Qn, Gn2 “" Qinn

We say that the matrix is of order m x n (m by n). The objects Qj), @y, «++» Amn Ate called the elements
or a function of one or more
‘the matrix. Each element of the matrix can.be a real or complex number
row and the jth column is
“atables or any other object. The element a, which is common to the ith
B, C, ...
ed its general element. The matrices are usually denoted by boldface uppercase letters A,
When the order of the matrix is understood, we can simply write A = (aj). If all the element of a
Matrix are real, it is called a real matrix, whereas if one or more elements of a matrix are complex it
S called a complex matrix. We define the following type of matrices.
 a
/ A eae Vhirtlncnetge 5 Matrices and Eigenvalue Problems 43
Raw Vector A cnnates otiorder |} <a, that ps: it dices crrer Tae una fia cilbatiins i itt |
/ 0
ora res matrex of order a aids written iis ie
; | 3 |
lary, if ify i 4] Jas i if |
A= is called an unit matrix or an identity matrix of order n.
mowhirch ct) fora pas the yth clement t} |
ver rans and one Columbus... CO
Colunin vector A mates ot order ma: » |, that as, 1 Das :
mati of order ar and is written ats
Warp py
an identity matrix 15 denoted by I.
vector or a cofmmn
said to be equal, when
j matrices Two matrices A = (4jj)mxn and B = (by),,4 are
fh q and
Eq"-i) they| are of the same order, thal is m= p,n=
Ay | i 4 é,
rt
are equal, that is a, = by for all i, j.
=
~
ts
fo,
(ii) their corresponding elemen
fy ' |
i ey
A is
A seatrix
obtained by omitting some rows and or columns from a given matrix
submatrix
fr
fr
rtix of A. As a convention,
the given matrix A is also taken as the submatrix of A.
calle d a subma
in which b (or h) is the sth element.
boldface

bra
The number of elements i a row column vector is called is onder The Matrix Alge
letters

3.2.1
cobors
nuntbers,

lower

whereas
endercd

ly
are
n-fagpfe

denoted ustial
case

ions allowed on matrices are

ete. [fa veetor has -: elements and all its elements ar |
called

u
The basic operat
b,
by

more
a,
¢,

tC Tes
IR’,
then

:
one
its

or
an

in

scalar,
¢

(i) multiplication of a matrix by a

lements
ru

arex
comple
numbers, then it is called an ordered a-aiple in ©
ces,
Rectangular matrix A matrix A of order at = n,m # nis called a rectanetlir mates (ii) addition/subtraction of two matri
ces.
(iii) multiplication of two matri
Square matrices A matrix A of order a x nin which m = 7, that is number of pows js equal to ation A/B where
a matrix by a matrix. Therefore, the oper
Note that there is no concept of dividing
the number of columns is called a square matrix of order n. The elements a,,. that 1s the elements
iy. . 4,, are called the diagonal elements and the line on which these elements lie 1s called
ilar, ne A and B are matrices is not defined.
the principal diagonal or the main diagonal of the matrix. The elements ihe, Clon: teh mare itin
Multiplication of a matrix by a scalar
the off-diagonal elements, The sum of the diagonal elements of a square matrix 1s called the race m X Nn. Then
and A = (a;,) be a given matrix of order
Let a be a scalar (real or complex)
(3.2)
the of
for all i and /.
matrix.
.
B= aA = Q@(a;) = (a@a;)
‘
in fi
j a ; elements zero are called
ts marrixaud!
a
I Nul which the all
same as that of the matrix A. |
matrix matrix
A order
of A in n x m
The order of the new matrix B is
a or
matrixZero ts and
denoted 0. by
Addition/subtraction of two matrices
A ji matrix agonal
matnix square Di in A
which the all :
off-diagonal
oa
elements a,, zero
iare
is
Then
two matrices of the same order.
a
matrix. diagonal called #/
Let A = (a,,) and B = (b,) be
For
example ;
for alli ili (3.3a)
| C = (cj) = At B = (a,) + (54) = (a,;, + 5,),
(3.3b)
lay,
5;,), for all i and /.
5 :
D= (di;) %y A — B= (a,) - (54) = (a, —
the same
e matrices A and B. Matrices of
and
he same as that ofth
a Is
diagonal matrix of
order Hn.
the is D or C matrixnew the of orderThe same *
A dia er
we
aaa
re aa ‘denoted by D. It is also written as diag
to blesaid are
confbeorma order se: a@,,d rs are
scalaany
[a,, a5) ...
nts of a diagonal matrix of order n are equal, that is a, =a a,,].
ormablewhich trices
confare additforion @, and Op ---»2.
for all 7, then the matrix
ma
If
aE
p
|A,
are
A),
A,
matrix sca ed
a lar calli r then | (3.4)
ordeof n
j
the
a e
ac all of eleme
a onal
4 matrix di diag
a | aA, C= to . OA,
+ order TheAyAn
4. 7 x of
matrithe is C
same as
ricesices 4,
r
A), A>.
mbini atio1 n of the mat
ordeof aren thenI, the
matrix oso? 7
is called a linear co
i
that of A. i = by 2s are p.
 Prihg
&§F wering:
Mathematics morn +4
Matrices and
Eigenvalue Problems 3.5
? IFtin
Properties i tion matrix
addi the of iti
catiaron

0O
multipliscal and
B=

ab
LA,Le phabt C he
the behit
matriaoe
ices B, and
aremablewhich
confor ‘ addi; tion
for and he ff oo

7
Sh § at
Then
I. =B+A.
A+B (commutative law)

0) 0
0
AB =

0
0
and BA
2.(A+B)+

=
€)ative
ax + by
(associ Jaw) AB.+
then

0
0
3.
A+O=A matrixnull the is (0 same the of cries ,
does not always imply that B=C.
=0.

nilpotent.
4,
A+(-A)
it
(d) If AB = AC, AxA...

Define
CHA 2. +B) Ga op
ve
* A (k times). Then, a matrix A such that A‘ = 0 for some positi
(@+6. BDA

integer
oA = BA4

x
7. i @
BA) —
pn
e smallest value of k for which A* = 0 is called the index

said
& A
=

Th
_ 8
A=A1x

be
and Ox

to
A= 0.

1s
©
|
rix A.
of nilpotency of the mat
Multiplica twoof
Matrices
prod
The product | of ee
AB = - ed an idempotent matrix,
two matrices A and B is defined only when the number. (f) If A? = A, then A Is call
to the nv
«¢ NuUMber FE Such
vem in B.
of rows : Colutams in
matnces
on
Properties 0 f matrix multiplicati
are said to be confarnygile tor rai tt
MMII MOT hocatpeyPT :
nm < m an be
p and p x gq respectively, then
matrix
r
and (4;)= B p x a an be
1.* If A, . B, C are matrices of order m xn, n x
matrix. Then the
product sie mats
(AB)C = A(BC) (associative law)
is a matrix of order m x q.
Bis, 2: 8 b. |
ces of order n x p, then
2. If A is a matrix of order m x n and B, C are matri
A (B + C)=AB+ AC (left distributive law).
of order n x p, then
5,2
3. If A, B are matrices of order m x n and C is a matrix
6, | bap |
=: my =< 7 yx (A + B)C = AC + BC (right distributive law).
of order n x p, then
1s ; - iJ
4. IfA is a matrix of order m x n and B is a matrix
1
a pa
matrix of
order mx Thep.
element gencral the of
product matrix is C
given by
Cc. = 4h =
a(AB) = A(a@B) = (a A)B
y ay + lj
aj2by,
for any scalar a
*
inn, «a T Dy. a;, = = (3.5)
= k=! :
theIn ct
produ AB, isB
Some Special Matrices
said
bet 3.2.2
a ied Arca
le ltipli
pre-mulupl saidis A or A by
be to lied
post-multip B. by
We now define some special matrices.
matrix column
a is B and7 * l <j a is that order
of then1, x n AB matrix
a is
order
Ciof element,Single ’ 1, x |
andcol
a is BA and
ws and columns
matnx
Th e matrix obtained by interchanging the corresponding rows
of
order n x n
T |
Remark |
or A’, that is, if
(a) possiis It : : ‘me ee
sts called the transpose matrix of A and is denoted by A’
product ss matrix BA 02 ah product
the B, andA matrix defined
is AB the but
x43
i
_" ) -
1
ae fines
om Ty
fae
a ay
fined. For
bw
example,
“ae a
if A
ay?
1s 2 a
dj)
3 x
matrix and B 5 a
T 2: 8x= gs:
is edAB
-- Os
defin and a js
matni x of
order 4, x s 2
wherea b
a, Ox ‘ then A =
A
aes Pa ae =
=
=
1 tera -/ an
are BAid
defined, the both then
matrices are BA and AB .
Qj, Fs Sum
em
_
ae
a ts 2. i
Ge
mata 7a oes the matrix product is not commutative.
are Sald to commute with each other.
; T; x m matrix. Also, both the product matrices A’A and AA”
ana
If A is an m xn matrix, then A is
{ are defined, and is ann X nm square matrix
ATA = (n x my X ”)

—_
Sa
so
‘n ——=
ee
 ‘|
ae
oo

a
zero.are
Ann ~" Gn2 Gy,
diagonal principa
te

belowthe l
elemthe
entsall ts that J,

—
a:
1
triangular
uppeanr and zero are
3

diagonal Principal
0, = aj; if
- det(A)=|A|=|
matrix triangular
fowe
a r
= 0, whenever
j>
calle
isd (ay)
4 a3,
e the
225 °"
= A matrix square
A atrices
= ier

Gin "" 12 ay,

<j, that is allelements abov
‘

a
a
‘eral?

=
A

ee

if
5I

is

as
A3).

A
or

if
of is

is
order

be
realdet

is has
matrix

7m the
A and
or (A)

and
The

this

real

real
|A|.

may

a
defined

is
value

n,
by
value

of
complex.

we
A(R

complex,

matrix
symmetric matrix and a skew-symmetric matrix. That is
Determinants

With
3.2.3

every
order
order
square
nwhich
matrix
determinant
A- A 1s always skew-symmetric, Therefore, a real 98 Is always symmetric and the matrix

denoted
the determinant
matrix.nulla be must
ee |
skew-symmetricsynumetric’a
ois2 nd ene= For (c)

associate
i : wal square matrix A, the matrix A + A’isz
'
result.
the Hence,

Adeterminant
zero.are elements diagonal
its all (a;),
= A matrix synmneinc
a ‘ a |
(AB)’
(BA)’
A’B’
=AB. J a a 1 1 = 5 L a

3 Remark
get we sides, bothon transposeTaking
BA.= AB let Now, A”— = A is that/, and
i all for 4,~ = ay,if skew-symmeiric
AB.= BA or B’A’=AB,
or (AB)'=AB, 3 A?= A is thatj, and
¢ all for a, ay=
if symmetric,
Then symmetric.
be AB Let be to said1s (a,,)
= A matrix Square
realA matrices skew-s
andymmetric Symmetric
B’=B.and A’=A
havewe symmetric,
are B andA matrices
the Since Solution ‘b,, a, aa a,,"b,7d, a, .

commute.
B andA matrices
the is that BA,= ABif only andif symmetric
is
(3.6) aaa
| il aa
a,b,reales a,b, a,"b,
AB matrix
the that Show order. samethe of matrices symmetric
two be B andA Let 3.1 Example
a,b,as a,b, a,b, ae

) eee i 5
as written
be can 45) (3 bq. in C matrix product
the products,
innerthe of terms
In scalar.
a is which

ts that matrix, skew-Hermitian

a and matrix Hermitian
an of sum the as written
be
canA matnix square complex
a Therefore,
matrix. skew-Hermitian
a always
is A’— A matrix
b= } ayly, c,=a
a

15 that b,, anda,

an
A,+
is
Vectorthey
s

(a)
Poul

A’
any
the
AFor
and
(5),
6,
b,
6 \" of order x x | is called the dot product or the inner prod
=...

square
matrix
always
ia2.

matrix
(let ay= x, + iy; then ay
= — @y gives x,+ iy, = — (x,— ty) or x; = 0 for all /). The product of a row vector a, = (a2,, a _a,,) of order | x 2 and a column vector

complex
2 Remark

Hermitian
(c) In a skew-Hermitian matrix, all the diagonal elements are either 0 or pure imaginary

Ad poe AG AG=
. x, + iy; = xX;— iy, OF yy, = 0 for all j).
(b) In an Hermitian matrix, all the diagonal elements are real (let a, = x, + iy, then a, = @ , gives r
Ag]? Age: [Ay
matrix is same as a skew-symmetric matrix. skew-Hermitian then 1s A, ...
defined, A, A, theIf
product
and a matrix as a symmetric matrix is same (a) If A is a real matrix, then an Hermitian for
mulipheation are B andA
contormable the
matrices when =
B’A’, 4.
(AB)
P Remark 4 tor
addition areB andA
conformable the
matrices whenB’,+ Al = By’+ (A 3.
. A. = 2.
(A’y"

rowa
malnx.
Hermitian matrix is denoted by A”
or A*.
1s matnyy column
a Of transpose
the and matrix column
a is matrixrowa of transpose
The I.
a )’. Sometimes,
A complex matrix A is called an Hermitian matrix if Hermitian and skew-Hermitian matrices
The following results can be easily verified
Then, the matrix A = (a,) is called the conjugate matrix of A. | h, Ayo.[A, as be
written also canb vector A
column
Let A = (a,,) be a complex matrix. Let a, denote the complex conjugate of a,. Conjugate matrix r
& Ar square = (m x a)(n *% nr) is ane AA and
Mathys

3.7 Matrices and Eigenvalue Problems Eneineerine Mathenatres 3.6

 AB = BA =I
(3.8)

case eachin Samise

non-singular square matrix B of order n such that

a
determ
theina
ofnt value
he

-_
row,by
r(A) = n. A square non-singular matrix A of order n is said to be invertible, if there exists a
column T;
(iv)
column
by , row (iti)
In other words, a square matrix of order n is singular if its rank r(A) <n and non-singular if its rank
coluby
mn, column
(i)
row,by row (1)
(ii) non-singular matrix if |A| # 9-
ways follothe
wing
of one anym fminadete
2 nts two multiply
can we
4
(i) singular matrix if |A| = 9,
[A7}
= :
|A] Since
|B]. =|A] [AB]
Let A= (a;,) be a square matrix of order n. Then,
A ts called a

then order, samethe of

Inverse of a Square Matrix 3.2.4
matrices Square
two areB andA If
determ
twin
oof
ants
Hence, r(A) = 3, when p = | or 2 or 3. For other values of p, r(A) = 4.
unchanged.
is determinant
a of valuethe roPduct
yp O 0
operations,column)
(or row erementary
operation column elementary
the called
ts [A],of
#0. -i)=e
wo IAJ=]0 column
‘th the is C, ere : maa Thoretecs.
operation
the Similarly, operation.
row elementary
an called
Is ae ie Cases
ol
0 row ith the of elements
the whereas unchanged remainrow the-/th
of ements a chenydead
St
rs ona kR,,
+ Ri — 2, as operation
this denote
Wwe row,sth the of aap ener
submatrix
; and& constant non-zero
a by multiplied :
are row j/th the of
elemen
thets es
When
4
Setting |A| = 0. we obtain pt = 1, 2, 3. For u = 1, 2, 3, the determinant of the leading third order
Remark
genera
in l. Bl,+ JA]
= Bl |A+7.
= pp — 6p + ip — 6 = (we 1p — 2)(H — 3).
seein ok
1 -6 .|-6OW -6 11 | unchanged.
remains
| column)
(or . row other some
of elements
the0 added
| is Corresponding
column
(or) row some
of element
the sof multiple constant non-zer
a If o 6.
| S| ire | 0 -l p n order
of matrix
a is A where a@"|A|
= |@A| Therefor
@. bye,
is Matrix,
the of
multiplied
elementevery then@, scalar
a vy matrix
a multiply
we when that Note
elements of first row, we
determina
get
row the ofnt.out taken
be can# factorthis then column),
(or
one submatrix

a
¢ of factor
le
a
age
fisif
© order 3 whose
Therefore,scalar
of
determinant is not zero. Expanding the determinant through the
the by multiplie
is d determinant
ach

the
then@ scalar
a by multiplie
is d column)
(or rowa of
the matrix A is of order 4, r(A) <4. Now, r(A) = 3, if [A] = 0 and there is 2t least is - clementeach
If 5.
clement

Other, each>
to -_ @ 3.
‘;
zero.
is determina
the of nt valuethe then
Proportion
are al| columns)
(or rows
1 two of elements correspon
the If ding
4.
1S determinan,
the OF (—1).by multiplied
oe oe e a
OF we St @ a
O —-! pw o
zenis TOINAN JA),
= JAl 2.
‘ deter;
the of valuethe then zero are column)
(or row
O 0 = jp " example a of elements
the all If I.

Find.all values ofp for which rank of the matrix if the determinants
of . properties important
some now giveWe
respectively.
a,, of
‘eh rank-
matris of matrix cofactars
and arinors
the are 4, and Af, where
(4.7)
~AO - Is
| red
«<Q. ¥. then it must be a null matrix.
a
matrix I) 7 saa «
A,, ay X = M,; a, (-1)"%
2=
The artix of A whose the d rank r= 7 if|A| + 0, otherwise r <n. The rank of
a null matrix Is zero and
rr ; re ir e
ames
min (m, n).. For
Fe a square the rank OF etermin
: ant Is 0 ot zero.
Thus, for an m x n mainx rs

pel
f a matrix is the largest value of r, for
j=l
exists at least one r x r which there
cagA,3 = M4,
q aank .
(1a,
2 =
-. A. denoted by r or rA ) is
the order of the largest
ud af
non-zero minor of |Al.
3.8

3.9 Matrices and Eigenvalue Problems

Mathematics
Enetneerine
 IMA).61+ A-= (A") 2p)
. ae writealsoSr
Weecan~—i
PY 2 155 goon
eusaeey ©
—55}- 61 -§|=|-1
6 60 -5-15
6 s(@pal-is
ae 5 * aL ty 1. integer positive
any for A™= (A”')"
7.
oe 3 matrix symmetri
a is c matrix symmetric non-singu
# of lar inverse
The 6.
. UPfot reele
AY dRE na
9 4 5 matrix. triangular
lower
¥ 18 6 0 | 9 ora ctive! j ix or upper
2 g~i on Dea“6 : triancullower ingularnon-sin
a : of inverse
The 5.
-5 6 ol=|-!18
uo
Wd).... Ida, (Iid)),diag= D“! Then,#0.dy dy),«... dew (diy. diag= D Ret 4
+lo a es result).
the Hence,
I. = (A™')"A
get we7I, = J? = (AA!) (From (A). (A $.
= 7; oO o i J = , . ; rs h-t
— “AO A Poe .
111+ 6A : A*_ = AT!. §. ee ies
(A's
2. .
| que.uni) is it then sts. A“'exists.
If I.
=0 at— ATA
+ GAA?
— acta? ‘. 4
| + 6A°— AX Premultipl(i)
ying ~ martices inverse
gel we, - -1=0Db ae
IA DyA (aero ee oe : ss of ae ) Proper
es
DN ' general,
2a Peay | In
ij,» B + 2A" BY+ (A (e)
0 0 0 —_ ) 1(d)). Remark
(seeC = B then matrix, non-singu
a is A larand AC= IAB(0)
610
2 © a2 =e can0 = AB matnix,
sinceima
cs iil ©
null4ap ane »
. then matrix, non-singu
is B IfB~!
larby
A".by post-mult
> ip
pre-mult
= si Ix. B theni matrix, singular
non-sin
a is A and0 = AB If (c)
1 0 0 33 Ul OG 4 240 30 «22 #13 30 es matrix non-singu
a alsois AB
lar then matrices, non-singu
areB larandA If (b)
0 1 O,-[0Th -30)+/5
6 a
5 -30]/—|
=4 90(35 we ae re pels:
3
AD At!= A,)' A)...(A, havewe general,
In
© oO [' 0-11]22 —-30 -6 |24 -19| -6 3 ae \ c. B'4-!
= (AB)!
o- B-'4”
= B(AB)" B-'(A'A)
= B-'A~'(A
. B)(AB)!
get ge. we —I, - LIA
%+ a 6A*— A> B=in gating
a Fe obtain
ice Ba! by then and A~! by first sides both Pre-multipl
nE ree
I, = (AB)(AB)'
= 31S | 1 910 4 § . “ .
| ase |
=I0t. —4 O,=)35
1 -5||/5 = «tA 3 ae
- have We
“wy
-19 -6 3 ‘)0 -5|)2-1 4 - a B7a-!
= (ABy' (a)
|
és 6 Remark
«9 4 5 3 t ane ax
A. of cofactors
of matrix
theof "anspose
~
5),a Offs
1 OFS 2 lS gz= oo) A of matrix adjoint
= adj(A) where
<6 ~j Ta <1]© =17
@ 2% _ si
have We a
OHA)ig, An
,=0w
| :
AW (it) and A’! (7) matrix
the find Hence,
3. order
of matrix identity
an 1is nere cn
- (3.9) by given
is A matrix
theof A“ inverse,
The
3 | ‘es Sho 32 ae
| We
- 6A-— if A, Matrix § =x AA”
= ATA
A’ equationmatrix
the satisfies
0) | |35 matrix
: = |A the thatw — sai
4 of mares
oh theOf
oe Nverse “ dries:
theis ov.A~ that
; say we =n Hence,
B".=BA=
zi) 0 © x
4 Ao
=B write
canes nate
The eles,
OE mainNY ee Mae
2.11 Protlems Eigenval
and ue “fatrices Mathematics Engineering
3.10
 l. ™ i and —| y= 2, r= Hence,

bd
in
=
Therefore, cases. followi
theng discuss
We

"
*

=
|
aA
or
b. vector columside
n hand

de
by

Pal

bea
obtai
A,ned matrix
the of deter
theminan
is t|A,| where
replacing the ith column of A by the right

|A|
obtain
We (i)
(3.14)
_ aa id Ad x;=
A.

by given
is b = Ax of solutio
then for rule
zy".y, ix,= x Cramer
The
's non-si
be ngular
A Let .
Let ‘Solution.
unique
a has equations
of system giventhe and non-singular
is A matrix coefficient
the Therefore, rule Cramer's

solution.
only theis

'
solution) (triv = x
0 ial then0, = b If solution.uniq

_
a uehas and consis
is tent equat
ofions systeThem

1
(3.13)

£0.
x=A'b.

i 1-l

|A}=|2
obtain
we A7',by b = Ax Pre-multiplyingnon-si
that find We Solution be ngular
A Let .
rule. Cramer's
(ii) method, matrix
(i) using system
this Solve solution.unique
a has method Matrix
methods. followi
thengby solved
be canb = Ax equat
ofions System non-homo
The gene
equat
of ions system homogene
Non-ous

and solution soluno

tion.hasit if inconsist
one leastat hasit if consistecall
nt ised
valle
iSd equaof
tions systemThe homogene
equaof
tions system
the then
t , 0. = b If
System
the then zero,notiS 6, -.-.92, }> 6 non-homoge
called
isneous. equat
ofions
elemen
thets
of one least
Example 3.4
enn the and vector t
at is, hat #0.
b If : vector.
Show that the system of equations
has
infinite columnside handght ri the solu tion
solutions.
of
number matrix, coeffic
theient called
ly
of system respecti
arex veb, A, and
homegencous
the case. thisIn singular.
is A if only andif exist
0 = Ax for ‘ons
x 5, a. aes a7 a.)

-3]=101+3)-26-1-1N+1G-1=10
qherefore, a homogeneous system of equations is always consistent We con :lude that non-trivial
P
o*
sie
7=>
- .¢ non-singular, then again x = A” 0 = 0 is the solution.
eal? b, “es ty s+ Gh | Aa
|
where
x, by, a), + M2 yy
trivial solution x = 0 is always a solution of this system.

(3.15) Ax = 0, oF (3.12)
Ax=b
plomoee
as (3.11) equati
of ons
ider the homogeneous system of equations
system neous system
the write can We orm.
equations
¢ of
soluti Jb
ss Matrix
fn
ns-
-
" G Bs = .X_7 2 +
has snd 22. F O,9%5
+ Q,4,) ‘ : :
of family one-parameter numberfinite
consistent
is equations
of system
the thenm, ..., 2, 1, = 0,1= |A,|all and0 = |Al when quan
bs = 43,,X), ie + Ua9X5
+ x, G5)
ns

i=1,2 on. are not zero, then the system o

nas 70 solution, that is the system is inconsistent.
when JAI = 0 and one or more of |A),
case 2
ysin
by = aigX,oF + M\3X5
+ ayyxy

3,14). cast I
by obtained
ts solution unique
the and consistent
is equations
of system
the 9, # When
\A|

3.13 Problems Eigenvalue

and Matrices
MathematicSi
s ne
Enerin
-g
 -3}=4(1

3.15
Eigenvalue Problems
3.14
Mathematics Engineering
[A|=|9

3)-0+

23-1)
-!}
, 4
1

20.
hav
c(ii)
We

=
if 1
ee

oy
|A,J=/2

ts

1-12
4

=~
1(0

2(4

0)

19

4
2)
-3|

6)

6 & 3
+
0

-
+

—
—
=

J |
12

3
JAl=|2
4 9
> 3 1) =0,
= ;
e
\2 3 21, = 0. IAn 162 2
JA3]=/2

yale 6
jr

FO] -1
4

2-2
122

1(0

s 15 inconsistent.
0)

4)

4)

10

+ 0, the system of equation

=

+
-

=
-

es . 3 Al = 0 and |A2l
Therefore,

tem of equations
2
r

| e the homogeneous sys

Rrampie 3-7 SOV
2y¥=

|Ay|
My |A,|
|A| 1A,

=—]

=]
r=

3) 0
r=

, 2
Pe
=

Hl

lz 3 -2¥ = |0}.
.
t)
Show that the system of equations 4 7 Az
Example 3.5 ons. Solve
nite number of coluti
given system has infi
that |A| = 0. Hence, the
1o-) 3a 3
Solution We find
uations
thefirst two eq
has infinite number of solutions. Hence, find
the solutions. : hee
= — &= where z 15S
arbitrary This solution eqtisfie
s the thir equati
a ae) a
on
¥
we obtain x = 132,
Solutions We find that
Exercise 3.1
=f 3

Ly

Seee
3 1/=0,

ll
JA]= |2

BD

=
|A,| so
Az=|-l 1
3 2 4| 1. Given the matrices

La

Le

tana
=—_
i)

LL
(iu) \A+ Bz\A\7 8B
1 3 3 jontna (i) |AB| =|A{|Bi.

UN w
|Aa|= {2 fi =0; [A= (2 3 2) *9 2]; werify that
(A Bi - BA
2. if A’ = (1, -5, 7]. B= |
3,1,
7°53 4 ;2°5
2| - oe
54 = 0. Hence, find
A

th
$A

wie =
equion A
4 | sausfies the mattis
Using the first two equations
* %
" a)?
s has infinite number of solutions. 3. Show that the matr
ix A =

—
Therefore, the system of equation N

2
xX, —~%_ = 3 — 3%;
1 |
111 = 0. Hence,
Dx, + 3x * 2 — Hy

=_
i GAS +
es equalion
solution 3) catisfies the mat
— 4)/5 where x, is arbitrary. This 1 2
(1! - 10.x,V5 and x, = Gx; Show that the matrix
A=
and solving, we obtain x, =
4. -i

Land
2
satisfies the third equation. find A’.
 dd].1) (n—+ [c — (c =
d)""" \Al
that show Then,
7. i= =c, a, and
j # i d, = a, that such
n, order
of matrix square
a be (a,)
= A Let
p. integer positive
any for (C] 1) p+
( + [B BP = APt! = c'pat that (ABC)! C, each of order m, show B, A, matrices
0,
CB
C*
zn
are
C,
+
B
=
A non-sithr
nguee
lar
any
that
CIf
nmatrices
show
then
and
=B,xBC
= , hand rightthe on series
side converges, iT. For
true. never
is =I BA— AB (ui) (BA), Trace
= (AB) Trace
(ii) f assuming that the .
=TP+Atar
(I-a
e.y!
B anda scalars
any for (B) Trace
8+ (A) Trace
@ = BB)+ (@A Trace
(i) that show then matrix,
that show Then, matrices.
real xnn be B A, Let
33. QP.= A’! that show thenI, = non-
4 sing
1S ular Ift-16.
PAQ andn ordeofr matricesSquare
invertible.
is A + I thence, —> n as 0 — A” and matrix
real
a is A if that Prove 32. non-besing
A ular
andQi P, Let IS.
martix, Hermitian
a is iA then matrix, Hermitian
Atl=(_syuta* then +(-1)"A"=0, |-A+A?_
if (ii)...
Skea is A if and matrix Skew-Hermitia
a is iA thenn matrix Hermitian
an is A if that Prove At A= then IPT+A+
CQ) AT+..+4"

31.
that Show
ay
matrix,
OQ —Cz —b,
non-singul
g be A ar Let I4.
symmetric,
3 ¢ p?
are
& | 0 a ae h Hl 4) 9 2-2|

q
A7'B™'(iii)

Pp3
5
2
A~'B,
(ii)

(b)
(a)
ay0
AB“,
(i)

b,
30.
-Il
29.
r—~e|
iollietietlieatiaiiemeememememeieed
that prove then matrices, symmetric
and commutative non-singul
areB arand4 If 13.
a -& « 0 5 24 | ces

——
skew-symm
is BA etric.
AB— (11)
symmetric
-§5 4 3
both,are A’A and AA’ (ii) symmetric,
Bis+ A (i)
that prove then matrices, symmetric
are B andA If 12.
,
4 3 a\-1 26. ” : |
defined.
are
by. 25. Biwi
matrices product
ll
the whichfor matrixanyis B where
I = (A°'B')~'(BA~')’
that prove matrix, symmetric
a is A If 11.
ele Pe 2 =a] 1 |
by given
is A where
A, matmx
the of rankthe Find ; 3 defined.
js BAB! matrix product
th whichfor matrixanyis B where symmetric,
is BAB’that showA, matrix symmetric
any For 10.
solution.
no (iii) solution,
of number infinite
(ii) solution,unique
a (i) has

x+4y+Az=u x+4y4+3z=10, x+2y+2=6,

non-singular
1/{/Aj.
any
|A-‘|
A,
matrix
show
that
For
9.
=

Find the values of A and yw for which the system of equations

(ii) adj (adj(A))=|A\"?A. (i) Jadj(A)| =|A]"~,
8. For any non-singular matrix A = (a,;) of order », show that

l-—e
(i) adj (AB) =adj(A)adj(B),

l—-b
l-—a

c
ab
(ii) (A+ BY! #A7'+B".

i
zz @.3 2 1 -I

—=-]|.
and non-unity, has a non-trivial solution, then show that
If the system of equations x + ay + az =0, bx+y+bz=0, cx +cy + 2z=0, where a, b, c, are non-zero 3.6«C4], verify that B=]! -I/, A=|-1 2 7. For the matrices

2 (sin &)x+y — 2z=0, be i! 2-!

3x +2 (cos 26) y+3z2=0, 5x +3y—2=has a non-trivial
0 solution.

yep
Find the value of @ for which the system of equations
(ii) (AT) =A. (i) (AY =(Ay,
has (i) only trivial solution, (11) non-trivial solution.

x-iy+2=0,
is LS
3x+y-z= =0, — kz 3y kx +

1], verify that 3 A=|-2 6. For the matrix

Determine the values of & for which the system of equations 22.
21. | } -2
y + 3z=5. St¥+y=7, Iy—-z=1l,
~—¥
3¥ +4226. +23, 3x-y —-x+y¢22=2, 20.
+ (ii)
(i) [ad (A)]" = adj (A’),
19. adj(A
di tA”).-) ij [ad ] (A)] mT
2c+4ytz=7, jx + 2y + 92z = 14. x+2y43z2=6,
y= yFESZ Wwe sy- t= 4 X+y-z=0. 18. 4 3-3
the following system of equation: Solve verify that 4], —3 A=|4 5. For the matrix
1-l 0
3.17 Matrices and Eigenvalue Problems
Engineering Mathematics +16

~:~: ee:
 . wae
skew-symmetric.
skew-Hermitisn

ee
Hermitian,
symmeinc,

3.18 Engineering Mathematics

P.following

Matrices and Eigenvalue Problems 3.59

matrices
identrty
the
as

WOM of theye | properties (axioms) with respect to scalar multiplication

2 3} fa b | 19 6 c] 6. aa isin V.
SB -2 S54 37.\6 4 Als mee ee | Ba
7. (a+ BP) a= aart (left distnbutive law)
{-3 -4 6] le e f) ec ~€ Gf | g. (aB)a = a(Ba).
9. a(a + b) = aa + ab. (right distributive law)
[ I 2+4) J-i | | I 2+4i %J-1 | ; 0 2+4) 124 10. la = a. (existence of multiplicative identity)
” oy
: -S 3-5}.' e.)-2 + 4 4j = $ 3-3
- | 4 ; —2 + :4j a al
The properties
5. :
defined .in 1 and 6 are called the closure properties. ' - ;
When these two properties are
titi 3+5i 6 [-1-# 3-4 64 “=f ~3=4 ri | and scalar multiplication.
satisfied, we say that the vector space is closed under the vector addition
: The vector addition and scalar multiplication defined above need not always be the usual addition
iO i @ 0 -i ter I -j ; | and multiplication operators. Thus, the vector space depends not only on the set V of vectors, but also
42. | mi 43.{| -§ -2r O |. 44.)-1 0 1-; on the definition of vector addition and scalar multiplication on V-
ff f Oy i-)+i O i | —i lee 2 if the elements of V are real, then it Is called a real vector space when the scalars a, B are real
numbers, whereas V is called a complex vector space, if the clements of V are complex and the scalars
[1 2 =i be real or complex numbers or if the elements of V are real and the scalars a, 6 are complex
may a,
-i
Zi
may
§
45. i-2i i 1}. numbers. |
| fo oF Sy Remark 7
(a) If even one of the above properties is not satisfied, then V is not a vector space. We usually
3.3 Vector Spaces check the closure properties first before checking the other properties.
; b) Theconcepts of length, dot product, to be
vector product etc. are not part of the properties
LetavV be a“en a of certain objects, which may be vectors, matrices, functions or some other = satisfied.
eas a — of F and ts called a vector. The elements of V are denoted by _| (c) The set of real numbers and complex numbers are called fields of scalars. We shall consider
Miti te — ¥ th ii egovesys operations vector space only on the fields of scalars. In an advanced course on linear algebra, vector
? ree dition and (ii) scalar multiplication | spaces over arbitrary fields are considered.
a feet eo oere of (d) The vector space VY = {0} is called a trivial vector space.
vector addition is defined
ned as
as -the usual
, {diti
addition ,
of vectors,. then :
The following are some examples of vector spaces under the usual operations of vector addition and
a+b=(a,, ay, ..., 2) + (by. bo. cis BL) (a; + bj, a5 + by 2. @. bd scalar multiplication.
aCe
If the scalar, multiplication "Eat
isee defined as the usual scalar muluplication of ee
ye a vector by the scaler a,then ,
1. The set V of real-or complex numbers.
fa Q(a;, a3, ..., a.) ™ (@a,. aa, wa.) 2. The set of real valued continuous functions f on any closed interval [a, b]. The 0 vector
The set V defines a vector s <5 = ee defined in property 4 is the zero function.
roperties (axioms) are aia as it for any clements a, b, ¢ in V and any scalars a, § the following 3. The set of polynomials P, of degree less than or equal to n.
tik | 4. The set V of n-tuples in IR” or C”.
roperties (axioms) w
Re ) with respect to vector addition 5. The set V of all m x n matrices. The element 0 defined in property 4 is the null matrix of
1. a+bisin V order m x n.
2, a+b=b+a. (commutative taw) The following are some examples which are not vector spaces. Assume that usual operations of
3. (a+ b)+e=a+ (b+e). | vector addition and scalar multiplication are being used.
associative | ; ,
4,.a+Q0=Q+a= a. i: am] | 1. The set V of all polynomials of degree n. Let P, and QO, be two polynomials of degree
Renee (existence of g unique zero element in V ) n in V. Then, aP, + BQ, need not be a polynomial of degree n and thus may not be in V.
» -a)=0, a ™ Sage
(€Xistence Se f£ Po =x" + ax"-] +... + a, and O, == — x"A + bx" nl +... + 5,,
°¢ of additive inverse or negative vector in V) ae OR : ; ;
then P, + Q, is a polynomial of degree (n — 1).
 Eigenvalue
Problems
Matrices
3.221
and
3.20 Engineering Mathematics
2. The set V of all real-valued functions of one vanible = magic mud cOMUNUOUS ON the close, + (i/x, + My,) = C1. 1). The element
(1/x,, /¥,) such that (x). ¥,)
interval [a, 6] such that the value of the function - has Ss cae CONstan; > Now, there exists the element
For example, let f(x) and g(x) be two elements in aa sas ie = 2(6)'= p Sian (1/x,. 1/y;) plays the role of additi
ve inverse.
fied.
Therefore, property 5 Is satis
two scalars. We have
fb) +
g(b) fix)2p,= +
g(x) that NoteV. in not is p
let a = 1, B = 2 be any
8
SPace Vector OFMS
coeficents,
real with degree
of Addition Where
n, Now,
(a + B(x. 4) = 3p HY) = BX 3¥D
Example 3.8 Let polynomials,
all of set the be
is
defined by scalar usual underand ab = b + a
multiplication. Show that q notis F
Vector SPace
¥y) + 2(X,, vy) = (Xy, ¥)) + (2x, 2y'\) 7 (2x4, 2y5)-
and a(x), ¥;) rT Bix, yy) — Idx),
V does not define a vector space
Solution
2n, which is not in ¥. Therefore,
two be Q, and P, Let
elements V. in
Now, ~ On+ P,
not
1s
7
(P,Q) a 1S
+
and
satisfied.
polynomial Of
degree
property
+
oa
¥)
Blx,,
¥))
a(x).
¥)
(a
BY(x,,
Therefore,
Example 3.9 Let V’ be the set of all ordered pairs (x,y), Where x, ¥ are real numbers, Similarly, it can be shown that property 9 is not satisfied. Hence, V is not a vector space.
Lét a = (x), y,) and b = (x3, 2) be two elements in F. Define the addition as
—=FY
+b=
(24,
Sy
YD)
ay)) 3.3.1 Subspaces
(tp)
(0).
Let V be an arbitrary vector space defined under a given vector addition and scalar multiplication.
and the scalar multiplication as
A non-empty subset of V, such that W’ is also a vector space under the same two operations of
ax, yv) = (@x,3, ay,/3). vector addition and scalar multiplication, 1s called a subspace of V. Thus. W is also closed
under the
Show that /’ is not a vector space. two given algebraic operations on V. As a convention, the vectcr =, 2.- + 1s also taken as a subspace
F.
of
Solution We illustrate the properties that are not satisfied.
Remark 8
(i) (pe ¥2) + Cy Yi) = (282 — Sep 2 -—YP PD * > YD)
To show that Ht’ is a subspace of a vector space V, it is not necessary to verify all the 10 properties
Therefore, property 2 (commutative law) does not bold. as given in section 3.3. If it is shown that W’ is closed under the given definition of vector addition

properties,
remaining
existence
(ii) (x), ¥)) + (Xp ¥2)) + Oy Fy) = (2x, — 3, ¥, — v2) + Oy y) and scalar multiplication, then the properties 2, 3, 7, 8, 9 and 10 are automatically satisfied because

additive
element
these properties are valid for all elements in V and hence are also valid for all elements in I¥. Thus,

verify
—=¥,

need
zero
Jr,

that
(4x,
¥>
¥;)
6x,

and
the
we
of
is,
to
(Cy. Hy) + (C26 2) + (Xy. D)= C. P) + (Dey — 3x5, > — 5)
= (2x, — 6x, + 9x;. ¥, — _¥> + Vs). Consider the
examples: following
Therefore, property 3 (associative law) is not satisfied. Let 1}. of set the be V
n-tuples usual with IR”in x,) ... x, (x,,
addition scalarand
multiplication.
Hence, V is not a vector space. Then
Example 3.10 Let V be the set of all ordered pairs (x, ¥), where +, y are real numbers. Let
V. of a is 0 = x, withx,) ..., x5, (x,,
subspace of
n-tuples (i)
consisting
a= (x), ¥,) and b = (x, ¥,) be two elements in FV. Define the addition as is W V, of
since a notis 0 2 x, Withx,,) ..., x), (x,,
subspace of
n-tuples W (ii)
consisting
¥).in notis real
number, a ais when (a@x,
negative multiplication not
scalar under closed
¥,) (x, b= a+ =
¥,¥5) (44, ¥2) (X>+
the and
scalar n-tuples
of consisting
W (iii) withx,) ..., x>, (x,, =x,x, subspace
a notis1 + since
V, of W
multiplication as notis
addition.under closed
=, v,) Otx),
@y)). (@x, (Let (X1, i= XN, withx,,) man =x5 l +
and = y
Oi, withYn) aren Vo y> Fi ™ + ]
twobe
Show that a r not is F elements
Then
W.
in
space, vecto
Solution that Note yyx FY (x,= 5X4 see V2,+
ee an isis 1)1) (1,(1,
element anV. of
From : given
definition the of
addition,vector find we
saeTy
isnotin =x,y, + x, as W
+y,+2#x, +y, 1). +
,¥) +, Let 2. real all of set the be V
polynomials of P
degree additionusual withm < scalarand
¥)). (x), b=
multiplication. Then
This
'). Therefore, the element (1, 1) plays the role of 0 element 3 (i) W consisting of all real polynomials of degree < m with P(0) = 0 is a subspace of M.
only trueis
defined in property 4,
thet for
elemen (1,
 d.= a, Cc, a= a,r
ThenG £,¢€ g,, andF © ,f,
f where
W to
b,= a, + +O; Gy belong
g,+ f, and g, + f, Le
4, A, + a, + & + a, VW. 0=0€
0+ havewe G € and0
F, € 0 Since
G. &€ F, andfe ¥=F+G
Let So

scalarand mul
Equating the corresponding clements, we obtain the system
additionvectoofr
of equations
definition
usualthe of
assum'ng subs
a pace
is G + F that Sh
G}.¢€ Fig F+G={f+pfe
E al-abe alsefo cre lt
and
G + F as written
is G andF
of sum The {0}.= G FA
is defined by
that such
V space vector
a of bspaces
su beG and
F Let 3.11
(ii) We need to determine the scalars @,, G@, 03, @ SO that
Exam
2
S spans the vector space V.
“IX-yp
Since every element of V can be written as a linear combination of the elements
: of 5S, the se “yew
i ia-[get
We @=i.Let
Se se ,

2 —K+iy
x+fr i
writeWe (i)
Eb IG dd
numbers.
complex
are
c
scalarswhen space vector
a formnot does and number
reals are scalars
| be an arbitrary element of V. Let x= Solution

ab
when space vector
a forms
7 order
of matrices skew-Hermitian
all of consisting
W (ii)
V. span
xg" bj x+y |
W.e aA=iA=Weget a=i. Let
1 oa). kk OF aoflo silo (ii) xxi-y ai
fol 2p ait OMA i i
ew. Nad A= Let
3 Oe 1)O \O0 ; oOo) 0 S=) (1) <x+iy a
0 ' Oo I1\)(0 (oO 0) l
e multiplica
scalar
tion).
under closed
not (#'is
ai numbers complex
are scalarswhen space vector
a formnot does and numbers
real are
se the that Show visttrices.
real2 x 2 all of space vector
the be V Let 3-12 ample SCalars when Space
- | ; serie
(i) W consisting of all Hermitian matrices of order n forms a vector
taken the of
elements a as
linear
combination alt obtained Then
for V. We also say that S spans V.
the be to
spanning saidis S S.
Then from multiplication.
set mg : - : usual with matrices complex
n x »
Spann ing set matrix
atrix addition anq Scalar all of set the be V Let 4.
|
PW).
Let S bea subset of a vector space V and suppose that
then—Ag¢
every elem
rery element in
in P’
¥ can be
scalars.
are @, -... @, @,, where
V, of subspace
a is w of element
an is A (if multiplication
scalar under closednot is W since
a ; | y of subspace
a notis elements positive
et a,V,
+ +... QzVz
+ QV, real having matrices
m x n all of consisting
W (iii)
Vv of subspa i
of all elements a is ce
7 order
of matrices triangular upper/ all of
lower consis (ii)
¥ ting
p of Subspa
a 1S nce order
of
of the form
that of
these
elements, all of set the
linear
combinations E ‘ .
Then, = mm
‘
°
‘algplieanonheaters matrices symmetric/skew-
all of consisting
symmetric
W (i)
set thei1s
s ae Tr eres Va0 Vip Ct 5: SCalar
additionvector usual under
V space vector
a of elements
r any be v. : Cs no We
— . ' seal and addition matrix usual with s
matricesquare realn x n all of
multiplication. Then
stateaw , j : : : set thebeV Let3.
There oF ST
an important result on subspaces.
| W). then—P
W, e¢
V. of subspac
a is eG + F = W sisi.
A
elemen P (if
an js '* multipliscalar
cationunder closed
not Wis since V of subspace
scala
any for
ra,a( ie
iso, f+gy=artager* positive
coefficients real withm < degree
of polynom
all ofials consist
W (iii)
ing
ih ece a 4 Be Get git + of Pace | Og W,thenP+ (ifPandQeadditionunder closednot Wis since
V,
i
Subs .
(ii) ¥ consisting ofall real polynomials of degree < m with P(O) = | is not g
3.23 Marrices and Eigenvalue Problems
Mathematics Engineeri
om)3.22ng
 , ae
Matrices and Eigenvalue Problems 3.25
3.3
54ng
Mathematics Enginecri
The
solution m of
Systethis
Remark 9 a homogeneous system of algebraic equations. Non-tnvial solutions exist if
ofions
equat is
: Ea. (3.17) gives
a, @=ca, = -d, G@,
=b-e, ‘ent matrix) = 0, that is the vectors are linearly dependent in this case. If the det(coefficient
on
det(coeftic c "s
rule, @ @ = --. = &, =
matrix) * 0, then by
Therefore. writecan we the and0 =
vectors are
linearly
independent.
IR“. Show that
Example 3.14 Letv, =, “1, 0), Vv, = (0, 1, —1) and v, = (0, 0, 1) be elements of
the set of vectors {v,. ¥z, V3} is linearly independent.
ab 10 Ei | 1 1] }
;]b (a = _ a a = (b +
lg +(c-—d
3 ( | 4), o| ie
We consider the vector equation
L

Solution
ntsS Of S. the = 0.
elementevery Since
a@,v, + @2,V, + a3v;
of be canV
written
set S spans the vector space
combinanoa n as
Jinear elemethe of
Substituting for v,, V2. V3, We obtain
See

ersten

a,(1, -1, 0) + a,(0, 1, -— 1) + a,(0, 0. 1) = 0

Whether m o,
t+age

set the not ila

<n as
7

(a, — & + &, — G+ 2) = 0.

or
of these equations
Comparing, we obtain a = 0, — a + a = 0 and — @ + a= 0. The solution
spans ?’
P(1) Let Solution an be 5 + yr + Br?+ at?=
element arbitrary is a, = = % =O. Therefore, the given set of vectors is linearly independent.
need We I. in findto
whether
Alternative
scalarsexist therenot that sucha, a5, a,, si
Br+ an? + a, 5= + yt +
ay(t? at+ 2) + 1) 2+ + J 0 0
ar Br + +a)(a, d= + yt+ ar+ (a,+ a3)+ ay.+ f det(¥,, V2, V3)= |-1 1 OJ =120.
Comparing the
coefficients of
powers various get we 1, of 1
0-1
a,
+a; ag+a,=y% a,
a,=B, a= Therefore, the given vectors are linearly independent.
Example 3.15 Let v, = (1, —1, 0), v2 = (0, 1, -1), v3 = (0, 2, 1) and v, = (1, 0. 3) be elements
The
solution three first the of
equations is
given by
: a,-at+tB-y of IR?. Show that the set of vectors {v, v, v3, V4} is linearly dependent.
a,=y-Ba,=f,
Solution The given set of elements will be linearly dependent if there exist scalars @,, G+, @,. Gy,
not all zero, such that
Fe ee
Substituting last thein
equation, 8, = B y—
which we
obtain all for truebe not may
elements FV. in
example,. oy*
For the
polynomial 3 + 4 + 217+ ¢°
not does
satisfy this
(3.18)
and condition
= 0.
therefore,
+r Q3V3 + a.¥3 + Ov,
it
ayy)
cannot
be
written a as
ion
linear
combinat the of
elements Th S. of
Substituting for v,, V2, V3, ¥4 and comparing, we obtain
erefore, 5S
does span not vectorthe
Space F.
ence
IndependLinear of
Vectors
The solution of this system of equations is

¥2-Fwtwn
r,é
a? Br
ict iate
EHOISP ee nite theof ¥,} -.., v2, {Vj,set
elements saidis V of be to
/inearly
a Oy o.. @,, not all zero, such that @=- a, @&= 5a,/3, a, = -—4a,/3, a, arbitrary.
Wit oho O's anyyet > aint 3 Substituting in Eq. (3.18) and cancelling a, we obtain

+2
vj

4
of set the then0,= % PK ss
vectors saidis be to
Jinearly

0
= Le +
Ear
Sete peek
: i ay “* | +
i 4 _
2
Poh. ae ek Sg, u ‘>
pe a Pipe acs bs Yar so Yq cam bewritten alternately as follows. =
: : ; | et "7 Te ek, ey ; - |
Hence, there exist scalars not all zero, such that Eq. (3.18) is satisfied. Therefore, the set of vectors

i
is linearly dependent.
18 nd ~» (2 t
Vn} en

¥
Tinearly

-INer
depe

lee
one leastatif onlyandif

nts.

gor
#4 rng

iS

2 a
a}
eaa
ae

LS

by
*-® -
..
z-

«4
a
*

"
 aa
lems 3.27
= a
: fa ‘ ; Matrices and Eigenvalue Prob
2.26)
Mathematics Engineering
r rly
BFE linea 1
inde pendent.
can Wey Vewy
ts an element
33.3
Dimension ‘
¢ There exis
Basis and vectors
the of ---» V2- V}. “k WV
: ent can be continued.
combination
prom” ___ewery element of V can
. i
Let F be a vector space. If for some positive integer m, there exists a set S of t linearly :
be
written 4 as
ension k <n. This argum
jinear lfin>k+
-
addingkeepwe 1,
ev ry
elements of F and and ifif every -
set of of n + | or more 5elements in Fis “TY Indepeng erwise:
set linearly dependent, then ae oth erefore V has dime pak. Egy ween Ma basis of V.
to have dimension n. Then. we write dim (V) =n. Thus, the maximum number of linearly inde 'S Saig gnd th aman Vee v,, such ip : + can be represented as linear combinations
elements of I ts the dimension of |}. The set S of n linearly independent Vectors ic i i eleme” —* of avector space V of amewae Slab ote ois for the
basis of F. Note that a vector space whose only clement is zero has dimension zero. Cd the since all seats in the basis of F, the basis of V spans Y. ae 5 i the following set of vectors
the n cle” sider the vector space IK . , :
Theorem 3.3 Let J’ be a vector space of dimension nm. Let v,. v.. .... v, be the linearly indepe mee yector space: For example, con
elements of V. Then. every other element of I’ can be written as a linear combination of ae sai 1. 0}, (0, 1. -1): (0, 0, 1)
slements. Further, this representation is unique. Miss Oey 1, —|,
oy,(0,0, 1) fl, 23)
+. 4 . id ¥ =
ai .
f Let v be an element of },. Then, the set {v. ¥). .... ¥,! is linearly dependent ag jt His
==
Gi) [ 0, 0}, (0. 1. 0}, [0, 0, 1)
elements. Therefore. there exist scalars a, @,, .... @, not all zero. such that id Gi. Eee basis in IR°. Some of the standard basis are tne
a basis .
jnearly independent and therefore forms
anv tal + ... + . av,
0 + a,v, nYn = 0.
(3.19)
linest
nee
uples in IR", then
1. 1f V consists of n-t 0; 7)
1, ---. Oh... 8, =U, 0: =,
. 7 _ a = : . ; ;: = 1 Pa t A ar \
e, = (1, 9, 0, .--. 0), e, = (0,
=@,
1ofent,
=...
This
0.
get
n+that
the
set
a,
yimplies
we
independ
elements
ae
independent.
linearly
1s v, which possible
notis dimension
the as of n. is V Es
in IR”.
is called a standard basis
Therefor we
obtain from (3.19)Eg.
n matrices, then
2. If V consists of all m *
v= 2 © aloo, (3.20)
i=]

Hence, a is v
linear
1,2, ---."
combination 7 of
linearly
and s=
independent vectors F. of
Now, let
two be onsthere
representati v of
given by B= [0 wa 1h sm o|, r=1,2, --..m

a
2,V, V= av,+ t+ ...+
av, and bv, v= +
bv. by. ..++ *

°
oo

°
where i. one g leastat for a; # 5;
1 called its
1
Subtractin
i the rthrow and the sth 5column, is
two these
equations, get we
that is
= (a, =
Ay, —
b3)v¥> (a3‘7 eerie af: ee = where 1 islocated in the(r, s) location,
standard basis.
©\in ¥ can be
Smnce
©
t, are V, +++ ¥>, V,,
x \°
| le, if ¥ consists of all 2 x 3 matrices, then any matrixx
linearly
independen get we
oo
a,-—6,=0 or
i= 6, a= 2, 1,
....7 For examp
Theref represent both
the ations and samearev of
the ation
represent v of
given (3.20)Eq. by
(8) containingof setA
vectors itsof oneas0
elements is
linearly
dependent theis 0 as
linear ito 19). Gl. 'o
, where En“lo o 2? of 1o o|0
al_vector space. If v,, v>, -.., vj, k <n are linearly
elements ¥,. 15 ¥i5.35 --+) V, Such that {V,, V2, ---» Yah isy) 3. If V consists of all polynomials P(r) of
degree = n, then {l, r, f°, --., "} 1s taken
as US
pe 2 «
standard basis.
v, W} forms a basis in IR”, where
{u,
Example 3.16 Determine whether the following set o f vectors
= (2, -2, 2)
(i) u = (2, 2, 0), v = G, 0. 2), W
3).
Gi) u=(0, 1.-, v= ©). 0-1, w= G, 1
 ==
0.
1), 20 1(1+ 0|0)

T(O)
F in v elementeveryfor that find we (3.25), 1)=(10
00

OF(v)
Eq. in 0 = @ Letting 00)-2¢

T(Ov)
space. vector
a of vectorsbasison action
its by determined completely
is transformat
as written be can vector
linear
A ion the fourth that We find J}. 2 0 7 (

11 Remark linearly
are elements
three these Hence,
0. G@;=
~ 2 ~ wrlasaghod
= es Thaw, ‘h
100), (19Grr the given subspace is 3 and the basis is the set of vector (1.0.00),
Mig nc0a Vig Min elements
the of combinat
linear
ion
a as written
be canV in v element 0. = a, and (0 =
any Then, basis.
its be ¥,,} -... ¥y. iV). set the let and» dimension
of space a.vector
be V
2a,+ a2, 0, @,~
+ @
Let
B. a, scalars
any andV in v, equat m
of ions systethe obtain
We
andv, for Compa
(3.23) 0. = 1) 20
BTC,
+ @7(v,)
= Tiav.)
+ Tlav,)
= By) + Mav, @(1000)+a,01006)+a
to equivalen
are t (3.26)and (3.25)Eqs.in given condition
TheW.s in are T{v2)
+ 7(¥,) sum the the 1). O 2
Consider 0) 0 1
and(1 ),(0 000 (y equationvector
and 7(v)
a product
the W, in are T(v.)and P(v,) Since
H’.
t,say(1
se the of elements
intoF from mapping
a defines
7 Then, threeany consider
Now,
F. in elements
are and defined
are ¥, * v, sum the andv @ product
the space, yector
a is V Since
Oy
reforeTherefore,
a. and
and @;,
OQ}, GQ),
9, 11@),f o values 4. than less
is set the of : the
; dimension
(3.26) T(v,).
+ T(v,) v.)=
+ non-z
forero satisfi
is ed (3.22) uation
eq
Mv, vector
the Hence,
arbitrary
1s @,. where
a,, =— OO 2a,
= 0 G4, =
V in ¥> v;, elemetwo
nts any For (ii)
(3.25) aT(v). Tiav)= by given
is equation
of s system
thisof solution
The
F in v elementeveryand@ scalar everyFor (i) 0. = a, + a, a&+2a,=0, @+a,=0,
properties:
two following equation
of s system
the obtain
we Comparin
0. = @,
g, = @; = @, = a, be should
the satisfies
it if mapping,
linear
a or transformation
/inear
a be to saidis T mapping
The |, into
V (3.22)
from mapping
a be T Let
1)=0
scalars.
of F field samethe over complex,
or real both spaces, vectortwo @(1000)+a,(01
be WH’ andV Let space. vector
4 into space vector
a from mapping discussshallwe section,
thisIn
vectothe
r of equation
solution
only the then4, is it If 4. < is subspace
the of
Tf. the image set of the transformation dimension
The Solution
- the image of a under the mapping T. The collection of all such images im B is called the range or . basis.
its find Hence1)}.O 0 (0 1), 0 2 (1
called
is b and Ta b= or 7(a)
= b wnteWe 8B. be element unique
a get we 4, € a element
eachFor 0) (019 (1000),
sct the by spanned
IR*of subspace
theof dimension
the Find 3.17 Example
(3.24) B. T:A—> y IR?.in basisa form they and independent
If 7 is a transformation from A into B, we write rs etc. to denote a transformation. linearl
are W V, u, Therefore,
0. = a = Q = a, is equations
of systemthis
of solution
The
called the domain of the transformation. We use capital letters + 3a, = 0. @ — @
cnn
es da function
clensentz orof aB.mapping — and + @, = 0, a, + ja, = 0, & ‘—
The set or4 ais transformation. Thus, a transformation maps the elements of 4
A and B be two arbitrary sets. A rule that assigns to elements of 4 exactly one element of B is
(ii) Using Eg. (3.21), we obtain the system of equations
y IR’.in a
basis form they and independent
3.3.4 Linear Tranformations
linearl,
are is
ps
hoosae « =U,
= 0.d=1 and c = 1, d = 0, we may write a basis
as {(-1 -1 O 1), (-1 0 1 O)}.
‘e two free parameters, say, ¢ and d. Therefore, the dimension of the given subspace 2a@,-2a@,=0 2a, +3a,+2a,=0,
is 2. 2a, + 2a, = 0, and
andb+d=0 a+ec+d=0
- b= and ¢c=—d a=-— or
Jd.
(i) Using Eq. (3.21), we obtain the system of equations
—E

Solut
lution
a; = 0. a, = must be @, =
uw satisfies the closure properties. From the given equations
, we have
ae the dimension and the basis of the
subspace. G2) au + av + aw = 0
.
Exam ple 3.18
the of
equation onlythe
solution be @ >, @, Let
Then, scalars.
Letu~ \(a + b. Oe c, d), such that al a + ¢ + d =
At. indenende SMR ;
0. b + d = 0} be » subspace of IR*.

b W Vv, u,
must then IR*,in a
basis formsw} v, {u, set the If Solution’
Ne

3.29 Prohlems Eigenvalue

and Matrices

Mathematics Engineering
3.28
EE
 2-3

2 1-l
zeroall not Q,, Q, @, @,, scalars
any for writeWe matrices.
2 x 2 of spacethe in

A=|-4
basis
1.° 210 LEP.

we
BPs
alli’ dei Sale ol matrices
The

these
Solution
are linearly independent and hence form 4

Solving
oa ,

equations,
obtain =
+
@,

2,

4,
36,

= =
=
“-
a,

6b, 6b,
3b,

— —2,
—2,

4a, 4a,
4
43.8
Find

+ = we
get
the
Ik

and
7}{

3 {5 I. , : 3

5
3

ay

a,
b,

—7

b,
9-(3)ae -[al de

elements,

—10
J-faAeb DeelA

comparing

2|
| 0 |

Multiplying
—l

and
] }

_
]

1
a a

corresponding

@ ay
hil
6,

—4}-
—2

Dt

<|=|
[4]

rs
by defined transforlinear
mation
a a be T Let 3.19 Example

>2||;|-
we
+7).

have
7) +=
and transform
linear
ation
a is 7} + 7, Then,

a,
F. ve Toy,
+ Tv= Sv

b, 4
Therefore,

a,
5,
that such S transformatio

A=|a,
the be to 7; + 7; sum the defineWe W. intoV from
T must be a matrix A of order 3 x 2, Let transforlinear
mations
be 7; and 7, Let
12 Remark
Solution
P a , .
The transformation T maps column vector in IR? into column vectors in IR“. Therefore,
S 4 ‘ *

IR". of subspace
linear
a is 7 of
kere]the and IR”of subspace
linear
Tx,=(-2.2
a is 7 of
—4 -10)’. Tx,=(-2 -7)" and
rangeThe transforma
linear
a tion.
is A mapping
The
Ax.= Tx and IR", IR"+>A: T=
matrix of linear transformation T : IR? — IR’, such that
For the set of vectors {x,, x,}, where x, = (1, 3)", x, = (4, 6)’, Example 3.20
2 find the T are in IR’, spi writeWe IR™.in clemen
thets into IR" in
a IR™ in is elemen
thets maps matnx
m mx an
36 3 -3 3 3 Ax then IR”,in is x If C™).(or IR™in elemen
thets represe
A ofnt column
thes and
-2]1] + 4/2]= (or [R"in elemen
thets represe
A ofnt
20). =] 2] 5|+ —2/-2] rowsthe Let matrix. comple
(or x)real7 x m an be A Let
—2 —1 j i ] matrices
of . context
the in only transfolinear
rmation
the discussshallWe

Ay DANO ) ( ee4 r weei ( ils: “(; (V). dim= » = nullit

+ y (7) rank
\ a +
r\( | " ee| = 5 4 is thatr, — » Tis of nullitthe
y thenn, is F
0 0 0 0 1 i of dimen
the
sionandr rank has7 If 3.5 Theorem
4 aaa 9 T. of ee
ny};the Called result. follothe
wing have We
get we transformation,
linear
a is T Since
§. = companne
Jytd,
of subspa
a isce
the elements and solving the resulting system of equations, we get a, = 4, a, = I,
T of rangethe andFV of : subspa
a isce
7 of Spacenull the Thus,
aQ,+0,+@,
+a, +a,
+a,a, Vy. ve| iTiv) ran(Tand
)= 0!= f(v
| (¥
)
‘ = ker(T)=
| +a, a, a,
Calleg
'S havewe Therefore,ker(7T).
by denoted
is andT of null-space
the or erne!
ak ht ul” af + (| s\=(3 The».
Set ae ee ran(T
5a transformation element
zero the into mapped
are that
he OM wriis
ttenlinear
andTthe by
of rangthe
e calis , ¥ of : el ersmentsie
all of
asi" : led (v)= w ents
elemall of
0j
0Oo rom SEE idl IA transfo lineathe
r by W’ in collection
The
element
zero into mapp
iS ed in elemenzero
t | the Therefore.
3.3\ Problems Eivenvalue
and Matrices
Mathem
& atics Engine30
erin
Le)
 writecan we Hence,
obtain
we equations, resultin
iv) (3.27 000%
2, 1, f= Qusl's
---3 Gap [Gin Wel noe Wye [Why
= the g folvinat
g an :
Wy yj FF ese >;Wy>
[5 3) Y3JL ¥s 1%.
2 a,;W, Tv,;= 4h | ji Y3 2 (ae
ll-olelel ppla ONS 13
that such zero, Bll-il
A -la
16,l i2h
all not m ..., 2, 1, j= ....1,
2, 1, i= a,, scalarsexist there
is, ThatW. in W,, --., W2, W,, vectors |6 I! ai] ee
i= [Pe
em ON co a, ja,
basisthe of combination
linear
a as written
be canit W, in is .... 1,2,i= Tv,, elementevery Since 6] a,/fl@ @
iii) (3.27 Ty,
+ ... + a@,Tv,
+ a,Tv,
= that suchA matrix
the of elemen
thets determ
Weine
@,v,)
+ ... + aov,
+ T(a,v,
= Tv w= Fs Fo Ye
and
By}. B,; B, r=A=
ii) (3.27 Bw,,
+ +... Bw,
+ Bw, w=
a, @, a
(3.271) @,V,
+ ... + @yv,
+ @v, V=
as writt
been matrix
this Let 3. x 3
that such zero,all not B,,, ..., Bs, B,, and’a, ..., a. a@,, scalars,
exist there Then,
W. in element a is transfor
the mation order
of matnix
Therefore,
IR?. into IR? in eleme
psnts
arbitrary
an be w andV in element arbitrary
an bev Let ma T transfor
respectively.
W and V of basis ordered
the be Themati| on Solutio
(5 3) | 2 4

yy
es
Vie
acne
|,6 |=

Whe
zl—o

LWSE
WH)
¥(Vey
Vel
Fp-1l26
=;-+4a] Petl=

—
a

V
T
(6 l\ 2

W.
\

be

T:
6 |

Let
Let

that
such

W
F.
linear

be
the
that

Let
and
Vand
such IR° into IR?

over
from
T

field
same
transforma
linear
a tion
of matrithe
x Find

spaces
3.22

vector
Example
now show that with this linear transformation, we may associate a matnix A.
dimensional vector space into another finite dimensional vector space over the same field F. We shall

transformation
2, 1s i)

respectively,
ran(
of dimens
theioni} 6)”= = Since
maps the elements in IR" into the elements in IR”. Now, let T be a linear transformation from a finite

n-dimensional
0)

m-dimensional
We observe from the earlier discussion that a matrix A of order m xn is a linear transformation which I | :
Matrix representation of a linear transformation

T is not a linear transformation. Therefore,

ily
its + =i) q l 0 -!|
‘
T(x + y)= (xy +») + (2 + ya) + Os + ys) # T(x) + TY).
i's = AV= Tiv)
+25i 8 Jt |
Dy
My) =i tyrt+ ys Tix)=xhtxhtxy.
We have
,
x + y= (xy + Yy- X2 + V2, ¥3 + V3): have We Fi. ve| {7(vas) defiisned ran(7)Now,
1. is kKerof
(T) dimensioHenc
n e,
ie—1
Solution
[1 =i, v Ther0,
efor
= e Vv, — = 0, is
Let x = (%), 42 x) and y = (jj. ¥2, ¥3) be any two elements
in
solutionWhose
: IR“.3 Then
ay Dy? 0, = +7,Vv,
Show that 7 is not a linear transformation.

equation
the s Bives
T(x), Xz, X3) = xi + ce + x3.
Now
) = ry =Av=
Tv 0
thay such (v) v= all
i O7 Y tocdeiennin
‘need wee 7), ker find “To Seludin
"
;
Let Tbe a transformation from IR* into IR' defined by Examp ple 3.23

dimension
theirs,
and ran(T) ker(T),
Find =)". Fa Raa , rk 9 a)
1 | 2
Ax

re
A=
x=

A=|-15/2 3 13/2\.
whereIR->, into IR° from transtulinear
nnation beaLett $21
(2 2
z = i> =
Example
; - “ i

3.33 Prohlems Eigenvalue

and Matrices
Mathem
nme ares Enginees3.32
i = +
 eeer slain
SESE Se SE

writewe yA,= Tx is at
Using the notation given in (3.27 v). th
fol
as resultthis proveWe unique.
is v) (3.27 from btained
o A matrix
the W, — V 7: tra
!) ay+(}‘|=|. rlol=( linear
a and respectivély,
W andV spaces vector
of y andx vectorsbasis orderedgiven
a For
0 ml".OR 7 ,
[@,, &= B.] arr A, [B,, &
| RY T =

| 0 XS ap 0 ! and vi) (3.27

in defined
as 1s A matrix en
T)1)|= + (0) : =
(ol T|0| We
obtain : c
=Aa sedate sliced|
(1), | N+ (H(t
’ , (0), ("| , or

1) 0, (0 :
as >, «+. 43> #21 fs B: Hence,
wa l v,= 0}, = have We (i)
=
(°} Ws =
(o} w, |, 0
l a, p 6&1 cee) M3 ay) B,
0 1 0 0}

Wal {Wis= ¥ V3}o Yo. {¥j.= x Let IR.= W IR?,= V Let Solution| eaeEg ER
+ +...
REE@5a,,+ EEaa),= B; where

+ i. -
B.Win +
2iw, =
Bw,
ta ot
IR.in }-l :] (ih y= and IR?im 1p0 | a 4)1/Oh|
sa x= Giy Mm Byam)
+ .-. } AqgtOz+ a_1 (Gy+ cs +
a + ... +
a,,)w, +
ay + W,
(O42, +
G,4j,) +... +
Qa), =
(Ga,
+ ... tT
2anWm) @2,Wz nly.) ou,
€), and IR? in €3 €2, €,, basis (standard
in €3
IR*). + dr SP
ote : :
1} \o 0) +
a2W,) +... + W,
aw, +
Aa We) ag + +... 21W Wy pay)
‘ eo ale i
Mesa
| 0 ¥ an in 0 ] 0 x= (i)
IR*in "| ee gee ae by A a ‘
near is
usinv), (3.27 from 3.27iv)
are (using
determined iii),the
A we matrix
(a)= have and
thex 7,in
27of(3.a, Fromy.elements vectors
is basi
rmation
0
De a
matrix es For y. andx basisthe on also but7 on only not 7 of
depends i
aa
acts
7 la Thetefiire
7. of rangethe contains
that#’ space vector
the of basis the ys
basis the to
ordered respectwith7, transformati
linear on termine the matrix of the
whi .vy, space vector
the of basis
a is x that observed
be mayIt y. ang
. basis at
the to TeSpect
wit T of matrix
= (; Zzal si
§} the or 7 of i representatio
Tix matrixn
the called
isA mati
matrix
n x m The
yi ms mn m2 Imi
3.2 example vi) (3.27
ee Y 721 A=
a
linear
transformati be IR = IR? T: Let 4 ig oe
by defined on i
: | Wee[Ws vim since
we
obtain a is
basis, given w,.)
j i all for 5;=
and a; or
B. = | A henceand ee = : -
i=l :
yA x=
— ad mr or
gherefo Baap pz Ip,
yB = yA
re,
havewe v) 72

=f 2ma @ 2)a3). Wall <ver Woy [Way= Vp] ss Vag (V1,g

and rx=yA pet
y5.= Tx - A
ad), a> a)
| Rot an (ay)
B _
of each wwo be
matrices (bi)= B 4
that such xn a order
Mathematics Engineering
3.34
3.35 Marrices and Eigenvalue Prablems
 = ng
36eri
j atii cs Engine
Mathem
Afatrices and Eigenvalue Problems 3.37
T [v,, V2, V3] = [Wy, Wo] hi I ]
Exercise 3.2
ms 1 to 10 is a vector space. If Fis nol a vector space, state
defined in proble
0 -I |]
Discuss whether — | 7
i are not 5a tisfied.
ich of the properties on
S m and having ? as a root with the usual additi
1. Let V be the set of the real polynomials of degree
If1) (0) fo |
- | ”
and scalar multiplication.
4 or 6 with the usual addition and scalar
ho |
Let FV be the set of all real polynomials of degree
lol= o
[ok oy ist] +] -I}
4.
07 LO} (I ees
Th
5
erefore, the:
matrix the of
. linear
mation multiplication. a3 bene ae,
transfor T
respectwith giventhe to multiplication.
scalar
and
basis ve ; polynomials
real all of set the be V degree
of with
4 2 addition
usualthe
with the usual addition and scalar muluplication. |
Clors Let 3.
4. Let F be the set of all rational numbers
js
Biven
and usual scalar
5. Let V be the set of all positive real numbers with addition defined as x + y = xy
multiplication.
a-|f ] I
6. Let V be the set of all ordered pairs (x, y) in IR? with vector addition defined as (x, y) + (u, v) =
ot -—-
(x + u, y+) and scalar multiplication defined as a(x, y) = ax, yp).
7. Let V be the set of all ordered triplets (x, y, z), x, », z © IR, with vector addition defined as
. 0’ ] I
py — 20, 2 + w)
(x, ¥, Z) + (u, v, w) = (3x + 4u,
I l 0 : Af scalar
and
as
defined
multiplication
=
z)
(ax,
y,
a(x,
az/3).
ay,
x + y = xy and scalar
0
numbers with addition defined as
=
8. Let V be the set of all positive real
3 I ,
multiplication defined as ax = x*.
We
obrain ryilj={
toy ; OFJ=
=i
laf' [|
feos! é Fl
gle
op
9. Let F be the set of all positive real valued continuous functions fon [a, 6) such that
le
mn!
(i) [.J (x)dx = 0 and (ii) i f(x) dx = 2 with usual addition and scalar multiplication.
10. Let V be the set of all solutions of the
I l l
‘
(i) homogeneous linear differential equation y” — 3)" + 2y=0.
a
(11) non-homogencous linear differential equation py” — 3y’ + 2y = x.
under the usual addition and scalar multiplication.
Using (3.27 that v), Tx is YA,= writewe
Is W a subspace of V in problems 11 to 15? If not, state why?
O)/1)f1
| }fl l Let 11. | x 3 all of set the bey
real
matrices additionmatrix usual with scalarand
multiplication and W
| consisting 1 x 3 all of
real
matrices from the of
-I/]j1
ql P/ 0) | me a ‘ a a a
re
Ther matrixthe transforlinear
mation ofx
of the
(i)} 6 |, (ii) |, a | (iii) |b], (iv) ||.
: T
itl with
respect vectorsbasis giventheto is
given a+b a” 2} 0
with real3 x 3 all of set the be V Let. 12
matrices additionmatrix usualthe scalarand
multiplication and
ee 0 I | A
which 3 x 3 all of
matrices W
consisting
cE have (1)
elements,positive are (ul)
non-singular,
OF
are (iii)
symmetric, A7=A.(iv)
additionmatrix usualthe with 2 x 2 all of set the be VF Let 13.
matrices complex scalarand
multiplication
scalarand
multiplication additionusualthe with all of
matrices W and
consisting W and
consisting

numbers.
numbers,
z iy + x

complex
scalars
gs |

(ii)
are
 =bl
iR?,in set spanning
a not1s 5 that Show
3-7 =x
T)y| IR's
a T:1R?
40. 2)}.I. (3, 4), 3, 1, (- 3), 2, {(1, S: set the of elements
the of
41. T: IR? 3 IR’; r( combination
linear
a as 9) 7, 4, (— Write 24.
x IR?.in independen
linearly
t also are
xX,xt+y
(ii) ,x+y+z (QQxtyy+2
yt2w
w
| that show then IR?,in vectors independelinearly
nt arez y, x, If 23.
i. Zz T: IR‘ > IR’; T : - 38. 3). 8, 7, (-1,8),1, 5, (1, 2), -1,1, (0, -1),3, 2, (1,(v)
1), 0, 0, (1, D, 1. 1,-1,
39.
D, 1,1, (1, 1, 1,0,(1, (iv)
T:: 1R°IR? > IR;i. r(*) =x y
+ 3y.
2),4, 2, (3, -2),1, 0, (2, 4),3, 2, (1, (iit)
: x+y+w ‘
3),1, 4, (5, 3), 5, 5, (4, 1), 1,-1,(2, 3,1), 2, (1, (ii)

1,
0,
0,
0,
2,

3),
x-y
3),

-1,
(1)
(4,

3x+4y

(-2,
-6),

|- =z =| T: IR GIR? Tl yl 36.
;
37T:
3 * a
. IR" - IR
. ; =
; .

are
IR‘
22.

x+y x
(1, the 1,

4,

b
2), in 9).
3,

(1,

: ;
-2,
2),
(1,
0),
(1,
(1,

4),
(v)

2x+ ;
(1, vectors

linearly
5), 4, (3, 3), 2, (1, 0), 0, (0, (111)
1, whether

I,
).
Examine

t),
—|
5),
(I,
=,
(2,
=I,
1, following

(2i,
(iv)
8),7, (6, 5), 4, (3, 3), 2, (I, (ii) 0), 0, (1, 1), (1,1,1), 2, (2, (1)
Find ker({T) and ran(7) and their dimensions in problems 36
to 42.

independent.
z
y= ()
y=0. 3 x, e
independent.
linearly
are IR*/C?
in vectors following
the whether Examine
21.
ee lo erry =<
hi
35. T: IR? 3 IR': Tly| = xv tx +z x=0 34. T:1R
3 1R':2TI" | =42y,
0
¥
x#0,~40
atlra
ohh
tleec soe
(x

Iz set the that Show matrices.

real2 x 2 all of set the be V Let 20.
33. T:,IR! > 1R?; 32. T:1R°? 3 1R?; Tly -(.%,}
i: TO)
I
‘ 4
Jr
aa O | 0);
F, spans O/;
31
fo 1)
oR!
1
real
constant # 0, +y + aa 7)
; =a _T:rim?

set the that Show matrices.

real
| x 3
Examine whether the transformation
all of set thebeV Let
19.
T given in problems 31 to 35 is linear
* =
or not. If not linear,
r
state
. .
why?
wh a
34-25.
— 247= P(t) (ii) 1, + —17
= P(t)(i)
matrices, (iii) symmetric matrices (iv) skew-symmetr
ic matrices. 30.
Find
the dimension of the vector space V, when
ays when
V is the set of all nm x mn (i) real matnees,
Piaof combinalinear P(t), Pt),
tion
a as P(r) Write
3, — 21 = P3(t)
8, ~ 77 or?=
29.
9g. matrice
Find thes (ii) upper
dimens ion triangu
and the matrices, (iii) lower triangular ise2.0«¢s.
larbasis clara Pi)6, — 47 ~ 27 = P,(s)Let 18.
11),1, -7, (2, (ii)
for the vector space ¥. when F is the set
-1),5, 0, (3, (1)
0,

of all 3x 3 (1)
3,

diag ,
3).
(4,
(ut)

eee A = (a,) with a,, + ay; = 0, (vi)

sss by given
is x ¢
real matrices A = (a,,) with ay, + 4)2
:
wherw,. v u, of
Ase B58
|ir,a is x not or whether combination
Determine
—1).
» | 3, (2,
ii) symmet matrice
w= and1)
skew-s
2,-1,
28. at the dimens
ric ion and s,the (li) ic matnces, (iv) skew-Hermitian —
basis for theymmetr
(1,= v 3), 1, -2, (1, w= Let 17.
iv)
-5).
1,
vector space V, when V is the set of all
iq,
3, 2), (2, k, 5), (i,
(—2, (iii) 5) 2, (3, (ii)
2x2 (1)— eS wl
+
10),3, (4, (1)
\ =
jatiteg |
-1, I}-
by given
is x where
W, vy a, of
(iii) (Ck, & 4), (0, &, 1, (A 9: k)}.
combinalinear
tion
a is x not or whether Determine
—3).5, w=(1,
and 4) 4, v=(2,
-1),
2, =(1
a, Let 16.
(i) {(& 1, 1), @, 1.1, & 2, k)}.
(i) {(k, 1 —k, k), (0, 3k
-—1,2), integer.
an is x, (v) <4. txatxi
x, (iv) arbiX;trarXz,
1, 0)}. CA
y,0, > x, (ii)
l, + aS a Fy) (11) 3x;,
2x,
()x
For ™
4. em
= ,=
values of k do the following set of
vectors form a basis in IK” ?
scalar "and that such x;) x3, (x,, form the of tiplet
i
of s set the be W and
he dimension and the basis of
multiplication
5.
itio add usualthe with IR?in
..
2;) 45, (=). formthe of
:
triple
alltsof space vectothe
r beF <e
(iii) (x, 0, 2.)
1S.
— ag (x,*,x, X)-
x.y: —y, ak (1)
: ;- aw ae Pig EE Be. roots.real only (iv) 1, as r} of coefficien
(11) t
(ii) (x. ¥. z. Ww) such that x+y+z-w=0,
0, as £7 of coeffi
(ii)
cient term bi
t of all vectors in ip? and § be a subset
cons(1)
tant
:
of I’ consisting of all vectors of the form that + :
ar having
4 < degree
of
15.
polyn
der! . “ke spanning
Sea} and
of omials consisting icatio
is
set for all polynomials of degree 2 and can be
taken a5 1
addition
;
polynomial
usualthe
> 4,4
with4 < degree
of 7 Polynreal
omial
| as a linear combination of the clements of
the set 5:
as i ¥ pali
Let 14.iene
S: oS 3 . etteSheDes 2}. By. Show

Mathematics Engineer
\awices and Figenvalue Problems
3.38
ing
3.39
 ee

Ge | Ge +> he te
: lo iisttil
(3.29)
2 Aby «+» Gy “a (A\b)= ar!in LET OL is TeedSacer
02 da bd oie
4| a --- 4) a, 1}) f1 f1) OV | fo (1 1\

matrixThe ' ordered basis

nx1|m
Find the matrix representation of T with respect tothe

b,
of
A,
mn,
are
the
The
and
;

order
=
4

=.
are respectively called the coefficient matrix, right hand side column vector and the solution vector

matrices
==
X+ytz

*n
a

T
—

On
7

Fn
Fm
Fmt
IR*

by

respectively
IR’

be
Let
46.

:
+z ae
yl= he
linear

defined

|
x+y

Aw
Bee

where

eee
---
1)

72! Q,

x
[2 42

ad By
-I}

Aly
ee
transformation

(-1)
inw IL]-1]} 4)-1),/ y= andV im Up IL/-Ih]
4) x=
Ax=b (-1 (-1) 1) ne
{-1
by given unknowns, |

(3.28)
n in equations
m of system general
a solving
for method
a discuss
we section,
thisIn 4.> n say basis ordered
the to respectwith
T of representati
matrix
the
on Find
n, of values largefor used not are methodsthese consuming,
time veryis determinants
order high
of evaluation
the Since
n. order
of each determinants
1) + (n of
x+yr+z x
evaluationrequires
rule Cramer's Thy
x+yl.
the whereas
n, order
of determinant
one and matrix, cofactor
the generate
to 1), — (n order
of each =
determinants
n- of evaluationrequires method matrixThenm. is A atz =
matrix
theof ranktheor 0, # |A]is
that non-singular,
isA matrix coefficient
the that assumed
We b. = Ax unknowns,
7 in equations
n of
system
a solving
for rule Cramer's
the and method matrix
the discussed
havewe 3.2.5, section
In
45

1
1}
\!
(0)

Equations
of System linear General
of Solution
3.4
\O0)
¥
=

y=
in

ine
and

I\

(1)
fo
fo
fo)

V for basisand dimension

the Find0. = 5x,+ 4x, and0 = x, + 2x,— 3x, (ii)
|

0: = 2x,+ 3x,— x, (i) satisfying

IR> in x;) x2, (x), form the of vectors
all of set the be V Let
1} T

to
of

the

T. transformation
linearthe determine
Then, P(r).
in
the

0}; OL/1,/0}7

basis
Find

rj- 3r = 2 2s. t, — [1 and P,(r)

in 1] 1, + (1 basis orcsred
the io respectwith transformation
linear
qOLJ1L}1])
matrix

respect

ordered
e+
(Zz

14
with y¥+z,

|F. =|
x+y

theof representation 3) 2
.0

A 1+
+ linear
transformati be 4
Pd) Py(f)T: Let 49.
representation

matrixthebe| a on
2 I by defined transform
linear
aation
be ™ Vi7: Let IR?.in spaces vector
two be HW andF Let 44.
T transformat
linear
the
ion determine
Then,IR*.in 3)’ (2,= w, 1)’. (1.= w, and IR?in -1}' ={l,1
-17".
3. (2,= v, r 1 -1. [l, = v, vectorsbasis ordered
the to respectwith transformation
linearthe

of representation
matrix
the be s| —_
Let T: IR’ —> IR? be a linear transformation. Let A = \ 2
I
T

to

{E(w er CG) oe
of

=
the

T.
1,

in
1.
the

with

basis

1)

=
Find

1)’,

die
.

v,=
-1}'

v,

in
7

to
respect

(3.
[1,
Then,
=E-1,

4)’
w, the
,

IR?
tnear
2)’.

and
7

~_
ae
=

with
T

*
.
a
-

ordered by
1R3

IR>*.vectors
x¥-r
—,

}
be

(1, ordered
matrix IR?
T:

jp?
Let

(:

w.=[1,-1,
43.

be
linear

—
representation

of
(@eerniie the
be

Let
T:
defined

the
renstornation
IR?
IR’
Let
47.

A=|2

3] i2

linear
matrix
respect linear
w, transformation
J

transformation
=
r

y|

representation
7:

—,

Marrices and Eigenvalue Problems

42.

IR?

3.4)
transformation.
1R?.

3x+2zj)
2r—j

Mathematics Engi
: ne
aUering
>
 lems 3.43
Figenvalu e Prob
Matrices and
not change
A as; t hey do
row-rank of B
| do not chang e the ‘fFthe matrix
3.42
Mathematics Engineering
nge the form of. A but Jeni to a matrix A,
ee
WIGLFEX and has wn rows and (7 + 1) columns. The augmented matrj se operation s cha 5 aid to be row equiva tions. Then,
row ° pera
1s
ies ina”“ssn A. A matrix Bis
pletely the system of equations. The solution vector of the system
of equations 3 settles the row-space of te sequ ence Oo f elementary
a
the matrix A by a fini
°
n-tuple (x), x, ..., x,) that satisfies all the equations. There are three possibilities:
) 15 a n can be obtained from that
we usually write B = A. We observe
f. |
is row equi valent to itsel : 7
(i) every matrix
the (i)
system
,
a
solution,unique has
equivalent to A.
nt to B, then B is row ¢q uivalent to C-
(ii) if A ‘s row equivale to C, then A is row
the (i)
system no has
solution,
B is row equivalent
equivalent to B and elementary
(iii) if A is F ow e of row) are called
(iii) system
the number infinite
has solutions.
of
and incons;
wea aaa equations is said to be consistent, if it has atleast one solution (that is column in plac
= 7
-
7
performed on columns
i 7
The 5 fem
spaces, we now obtain the me if it ns
Se a pt Using the concepts of ranks and vector linear — and The above operatio
existence and uniqueness of the solution of the
for the ns.
nditions column operatio
of a Matrix
equations,
3.4.3 Echelon Form Zeros
m if the number of
or 1" row echelon for
ed a row ec helon matrix ing all zero entries
Anmxn7 matrix is call by row until a row hav
3.4.1
Existence and ss
Uniquene of
‘the
Solution
entry of a row incr eases row row echelon form if the
preceeding the first non-zero refore, a matrix is in
L . i a +
is possible) is obtained. The
et
vectors
The
row
&,,
avector
be
in
of
theOF RyRIR"(or
C").
VY, vos Royg I»
consistin
-tuples
space
(or no other elimination
d.
following are satisfie
: - 1 2 1 ® . i i I ;
subseque nt rows.
all zeros, it is true for all
(i) If the ith row contains every subsequent entry in
this column
is and
denoted by
rr(A).
Therefore,
of A = rr(A) = dim (5). G 30 non-zero entry of any row, then
rows, then the initial non-zero
row-rank
zs. | (ii) If a column contains a
+ 1)th rows are bo th non-zero
is zero, that is, if the ith and (i of the ith row.
rs in a later colu mn than that
entry of the (i + 1)th row appea
me the we
define
co/umn-space the andA of
column-rank A of
denoted by
er({A)
0 nly after all non-zero rows.
(iii) Rows containing all zeros occur
= _
padi 1 x m of
matrix is A
generated by rowm
vectors havewe A, of di .
since subspace
a is S havewe VY, of dim n. Ss (S)
Therefore, havewe ; rage
7 x
are in row echelon form.
re(A)S min (#t,) and = similarly er(A) = min (m, 7). (3.31) For example, the following matrices

-l

l
3
ye

S552
Le
Theo
2 Let 6]
3.6 rem == A
(a,,) an.mbe # x
matrix. the
row-rank i
Then

1
and
column-rank

0
sameareA of

OO
Oo
oO 5 4

¢
ow,
| an € stat we
|important which result the as
fundamental j is ich
known theorem of |
algebra linear

0
0 0 O

0
0
0
0
ae!The : 3.7 siecle Theorem
geneous non-homo =o. of
equations system Is I A . b. = x Ax
where 1

0
has+.
ix,
theif
matrix only andif ution the andA
augmented matrix
interchange the first row
barn ieb)| (A na

.
abet, bswe hep the rank: of m x m matrix A in terms of the determinants of the n matrix. Assume that a,, # 0. If a,, = 0, we

(a,)
section 3.2.3,
In Waar
ped be a given m x

Let
‘e

other
ro. Using elementary row

some
A
with
ake the element in the (1, 1) position as non-ze

row
n ees has rank r if it has at least one square submatrix of order r which

*
is nonsingulc wee at
n form (elements of first column below a), are

to
m
lisse comsamiee sahias ei: oo of order greater than 7 are singular. This approach ts very operations, we reduce the matrix A to its row echelo
made zero and so on).
winaiie: rge. Now, we discuss an alternative procedure to obtain the rank of.a zero, then elements in the second column below a are
made
Similarly, we define the column echelon form of a matrix.
echelon form of a matrix A gives the rank of
Rank of A The number of non-zero rows in the row
matrix A) and the set of the non-zero
4 3.4.2
€. Elementary Columnand ow Row
Operati Column
the matrix A (that is, the dimension of the row-space of the
Operation
rows in the row echelon form gives the basis of the row-space.
The
following operati three
operations£ a4 on
matrix are A i
called
the
elementary row
operation rons: é
Similar results hold for column echelon matrices.
bi
Interchange rows two anyof
(written R,)~ R, as
eMe row any of ra
a by scalar (wri
written ak,as
Remark 13
ing/subtracti '
r bes and the
(i) If A is a square matrix, then the row-echelon form is an upper triangula
( row
ae to row anyof
another multiple eee mucke ne that aR,+ R, — R,
elements
column echelon form is a lower triangular matrix.
conreapondin
the to added
are row jth the of ne ty ise
The row. ith the of
unchar remain
whereas,anged, row jth the of the|
elements ~ the of
“changed).
get row ith
 :

basis. 0 I/3 2 ua ae ’
the forms itselfset giventhe oF 1)) 0, 0, (Q, 0), 1, O. (0, 2), 1, 2, (0, 1), 0, 1, {(1,
© se =i
vectors
of set The4. is vectors
of set giventhe of dimension
the and independent
linearly |
vectors-are
of set giventhe non-zero,
are A of form echelon
row the in rows the all Since 0 ee p = Cy)-C, 7/3 -7/314 G-7C,/— 3 oe
I | j Et “ 7
0 0 3 Se 5/3 5/3, Typa C73a- 4/¢2 1.ey A= (i)
fe at 1 0D
Pe Ft ee rer a i
| = R,—R,/2 m €
2 & Fil
0.27 3 4 O-2 W*

a a oe a @ Lt
ae ee ee
(ii)
O-6 .F G 7I
-]
—>
3
4
‘|
6 lptteae eh O 6 i —I

) ee tet
Le: of os Ale- 7 2
esalter
ek 6 Pia ae Be = (ii)
A=| ii) 2
fig ae tee
® alter (i ranks their find and a
form echelon coltoumn
0 me t 1101 matrices follow
theinguce Red
2. = (A) rank get we 2, is A of 3,26 Exa
form echethe
lonin rows non-ofzero num
€ ber
th
<103%.
25, 4) (O, Since
4),3 2 {(1 vectors
of set the as takenbe can basis The 2. is
4 0 66 ~15 -10 -15
vectors
of set
giventhe of dimension
the 2, is rows non-zero
of number
the Since dependent. linearly
are
0 0 0 “]0 3/R®s-5R,
2 : 17/%s~8R,
14 1 8
vectors
of set giventhe non-zero,
not are A of form echelon
row the in rowsthe all Since
| <3 Sa U8)of ot eee,ee 4 %.ey tere
/7 4 | LS a
2-0+. i 6.10 16-4 Be’ «# 3 i..2. a 3 2 1 |R2-2R,
5 4 21 :
S28 4 2 12
R,—2R, ;
4 Bg i za "fb...
« iso echelon
row the In roWs non-ofzero number
, uf the Since CA
Si- A. ofm form rai ih we i62, farm
linearly independent. echelon
row the: is This
y OOMAy 12 14 0 82
If all the rows of the echelon form have some non-zero elements, then the given set of vectors are
-2
6. ~7 8/16 f,+2
-6R,-7 0
Solution
>
Let each given vector represent a row of a matrix A. We reduce A to row echelon form.
Ryton,
oe l| § Ps -2R,
Ry : 2a 1ae ee
(iii) (2, 3, 6, —3, 4), (4, 2, 12, —3, 6), (4, 10, 12, —9, 10).
iy Ci, 1,0, 1,1,
have We A. by denoted
be
1.1) G1, 1. 1.4), G,0,.0, 1),
(i) C1, 2, 3, 4), (2, 0, 1, — 2), G, 2, 4, 2), matrix giventhe Let Solution
dimension and the basis of the given set of vectors. 17 14 | 6
F 2 |S 2 8
+2 l2>
Example 3.27. Examine whether the following set of vectors is linearly independent
Find the .
Since the column echelon form of A has 2 non-zero columns, rank (A) apt) al, at 2 OF}
= 2.
a2a
[i s 5 1-3
64 Ls 2 4 1-2 32 2
al - i NKs.
oO; 0 -1 1 €,+€,
4) 2 -1 i Te GsB tT ge ky iA basis ra their find and form echelorow
n to matrices followi
the ng Reduce 3.25 Examp
the 515 :
0 20 -l ” —-4 2
-2/C,+2C, | ms jot —| of 'MEnsiorows
n NON-Ze
theroof vectors
consisting of set the and vectors
of set giventhe
C5-C form itheis rows
eas
i oI OO 0 rt mel te g nOn-Ze
of ro number The zeros,
as elemen
its ts
all with row no has
row thetO It row thejf independent,
linearl
arey vectorsgivenThe form. echelon (column)
'ndependens
reduce
and column)
(or row its as
I cr ,
vectoreach with
linear] re
matrix
the form We not.or = ci
are vectorsof set given
a whether examine
to
| Since the column echelon form ofA has two non-zero columns
usedbe can
, rank (A) = 2.
a
approac
This
h (ii)
:
. il
3.45 Problems Eigenvalue
and Matrices
Mathematics Engineeri
3.4ng
4
 46g
3.in
Enigineer
Mathematics Eng
VYatrices and Ergenvalue Problems 3.47
2 49 Brg R, —2R, ‘
(iii) A=|4
9 i R,-2R, =/0 -4 0 3 -2 Ry +R, ~ , 3 6 “3, (a) Let 7 < m and one or more of the elements b..). 6 ,..2, .... hb. are not zero.
4°10 12
solution.
2 0 “tO” “9 Then, rank (A) # rank (A | b) and the system of equations has no
in A) and b, 1. D, -2. «--+ by are all zeros.
Since all th Y £6 -3
A are not non-zer f a
echelon formof ofnon-zero "= @ 4] (b) Let m 2 mand r= n (the number of= columns
Ws tn the the
linearly d a; ent. number rows is 2 Siven set of vect | 1 and the system of equations has a unique solution.
In this case, rank (A) ¥ rank (A | b)
of eas Since
! > < and its basis can be taken as the set {(2, 3 + ine dimension of the “OTS are We solve the nth equation for x,,, the (m — |)th equation for x,_, and so on. This procedure
9 -_ Set is called the back substitution method.
Gauss Fj
> 3, 6, -3, 4), (0,~4
re Uy 5, =I , a? oe
3.4.4 Elimination Meth . is of
ae S
: For example, if we have 10 equations in 5 variables, then the augmented matrix
Consider a non-ho mogeneous system of Non-ho
od for m equation s inous
mogene Systems
n unknow ns
Me
order 10 x 6. When rank (A) = rank (A | b) = 5, the system has a unique solution.
-.., x, can be
(c) Letr<n and b..,, b,.>. ---, 6, are all zeros. In this case, r unknowns, x;, X>,
seins the rth
| determined in terms of the remaining (n - r) unknowns x,.;, x... ---, x, by solving
this case, we obtain an
| : (3.32) equation for x,, (r — 1th equation for x,_, and so on. In
Qi Wa. xa: a (n — r) parameter family of solutions, that is infinitely many solutions.
a - Oty 5, x,
45> ae a>, b
A wig a>;
néere
: ,b=!| 7 ll x4 | Remark 14
ns in solving the linear system of
| . (a) We do not, normally use column elementary operatio
Bn Ano ” equations. When we interchange two columns, the order of the unknowns in the
given system
= ie Day x,
is quite difficult.
of equations is also changed. Keeping track of the order of unknowns
«i
th
ana ies that at least one element of b is not zero. We write
€ augmented matrix of order (b) Gauss elimination method may be written as
+ 1) as
Elementary .
(A | b) > (B ‘c).
Q5) Giz «es App by row operations
M20}
: ;
. .
(A[b)=| 2! 92 the matrix A and ¢ is the new right hand side column
The matrix B is the row echelon form of
: the back substitution method.
vector. We obtain the solution vector (if it exists) using
eg Baw ass, a 1d an upper triangular matrix of order n.
(c) If A is a square matrix of order n, then B is
and reduce :
leith 9) eo ion method can be used) to solve p systems of the form AX Z= b,, Ax “A
(d)7 Gauss eliminat
= by, ..-,
It to the row : elementary row operations. We need a maximum of
— echelon form by using column vectors.
es
evan | i: Ax =b. which have the same coefficient matrix but different ngnt hand side
stages of eliminations to reduce the give P ; . (7 + p) columns.
form. This given augmented matrix to theival
equivale nt row echelon
f the fi ! We form the augmented matrix as (A |b,, bp, ---- b.), which has m rows and
‘ Process May terminate at an earlier stage. We then have a n equivale nt system Of the form Using the elementary row operations, we obtain the row equivalent systern
(Bj ¢;,
Bx = ¢;,
form of A. Now, we solve the systems
| -«+y €,), where B is the row echelon
.., BX = Ci. using the back substitution method.
: Bx =
ay) Gyo ee ay, - Ay b,
| Remark 15
° d55 rr a, ne Ay hb,
becomes Zer0, then we nips this row
(a) If at any stage of elimination, the pivot clement We normally
with any other row below it such that we obtain a non-zero pivot clement.
: |
2
(A
0 Oe ay Om | & (3.33) asi
TO 0 Bis | choose the row such that the pivot element becomes
largest in ee
compa?
e F QQ . n x nsysiem, we» require
Tec (n - 1) stages of ecliminab
is possi
HITLt on. _ 1ft 15 fo
pow” ble :
Hy)
tions, multiplications and divisions. This number is called
|
‘ ” (b) For ann x 7” Sy
total number of addi pons, subtrac Gauss elimin ation method for
the method. The operation count of the count is approximately
the operation count of
0) ene 0 b,,
f) 0 ore
! nn 2 Fn $3, For large 4. the operation
m 15
solving ann xn syste
where r Sum and 7. ‘ ! n'/3.
45, #9, az, #9,..., 4, #0 are called pivots. We have the following cases: |
 0 0 0 0 -2/7 -2/7| O 0
-10/7/ -10/7)
0 0 sales ik -10/7 -10/7|
0 “19

Et
ae | 10 -7 eee 0 -7
aha

ta
=2 t 2 --2 | 1 2 eo
no has equations
of system
the Therefore,
3.= b)| (A rank and =: (A)
te 2 ue) 2 tsi" ¥ a
jo -> $141k, 48/7 sSiy? (* S$ 4
~) %
= zs (ii)
(A |b)
3/3}-2 {4
_|9 -7 ute WO
|

3
3) -2]R8)-98,/7
4

Mio
“9F 2 t' oe

=z
Fae -2 2 1

ses |
Be ee a-1

Zin1
(Ajb)
(ii)=|1

-1
3 ] O }2

z013
arbitrary.
are w andz where JanO. a ”

6.0.01)
2wy3
5:-- (4 y= and x=1+2z-2w y=1,z
1, = =-1.
x

1/2|}-1/2|
solution
unique

1/2)-1/2}.
the has equatio
of ns system
the Therefore,
solutions
of family parameter
twoa obtain

R,;-2R,=|Q0
we Thus,
x=]or 2x+y-z=4,
2w. x=1+2z—~ or
Bw - dz 4+= Gw— 92 + 2w -— 5z — 4 = 6w — 9z + 3y = 4x j, yo Or S48; "5" =57
obtain
we equation
first, the From 5 Fn :

I, z7z=—or F 37;
8 8
obtain
we equation,second
the From arbitrary. as solutio
then obtain
we method,
as w Choose solutions.
of substitback
utionthe Using
number infinite
has and consistent
is equations
of systemThe
0|00 0 8/3|-8
0 /3 0

o
6). 3} 15/2 9/2 ~4|, 5/2] =-3/2 =/0

oO
I
6/0 =-9 -3 4 —| I 2

fei
-12\-2
-30 -180 -6|-24
-39 -21 —aal 5/2 O° 2 2-3 “1
srg+ Ry 5/2|~4) ~3/2 i219 Re+R

BE
+4R,
R, 6) 3 15/2 9/2 ,
=|0 6
bie 6 3 3 2 =] by (A
| i — “elo 2? 2f-2) 1-1 G@Alb=|
Oo 6] -9 1R,-R,/2
[4-3 |
(i)
| ee i ae
06). 6] = =3

=
have We Solution ie
by form
Operations
TOW .
5. = 122
- l4y+ 3x echel
w ro
on to it reduce
and matrix augme
thentedwriteWe Solution
=] +y-—5z
Sx -24,
= 6w -— —39z —2ly4x

il
-3y+2
2x z=0 6 5 2I|- 5-2
6w+ 32 + Jy + dx

II
=1 2y-2
+ x (ii) =0 9z+6w
—- 3y - 4x (i) =]2 yl Me aan
fe Ur)
method. elimination
Gauss using equations
of system following
the Solve 3.29 Example l Tx -] 1

arbitrary.
is z andz = y 1, x= {3} l= 3) =2 2
solution
the obtain
we Therefore,
|. = ¢ or 1, z=
? -}]}2- -l
getxr-y*+
we equation,
first the From | Itty =t J)
=. = » or0, = 32 - 3y get we equation,second
the Fromz. y, x, of (i) |, —2 ce
values
all for satisfied
is =F Lp)
equation
last the that find We solutions.
of number infinite
has and consistent
is
fy Dfx]
O 2 4
systemThe =iff
£s >
Wie,

chimp Crdus« Usin

} g posst
(it
hic equat
ofions systems

and
3.49
follothe
wing Solve 3.28 Exam
si eple

Matrices
Problems
Mathematics Engineer

Eigenvalue
2-48
ing
 2 2 :. : on]2 0 eT Fgh
/3 RR, ~§l4]3 aoe| -3}4]
A Hence,
[Alb]J=/2
7 ie —2 oe)-13

3 2° =f. 7 0j =] i J] $ 16 2 sx]lL

W/5 -1/5 Te-o I 4/5 "" 16 -15)" 90

11475
sea" 3A." 0} wi 1. W2,3l2 form {11 C]. We pet entary row transformations on the augmented matrix and reduce it the
—2/10 7/10}. 0 1 O}-—13/10 Solution We perform element van
l l I l
2/10 3/10) 1 0 Oo] —7/10 Salis Ww OT it gn
h=/|4and -3!/,
1 A=/2
-s[--—1
4 7 (6.6 * ear g-1 2. Oo 0 |] -] i
1 7/2/3/2 =!0
where
b, = x A
1/2
equation
of system
the solve
Oo}
method, Gauss-Jordan
the Using 3.30 Exampl
1/2 O| (-R;)/5 1 7/2})3/2 on =10
1/2 0 1 0 3/2/1/2 Ni ie: ek

method. eliminati
ott. 2-2/8 1 at 4 Cire e ae Gaussthein as rows we then
interchang e zero, becomes of stageanyat
elimination element If
pivot any
=o 2 71 32 1 O} RR 3-1 1 2A DB oe (A|D=]
1 =-1 -2/-1 0 0 1 -1 -2]-1 0 0 16 Remark
vector. solution
the gives side hand rightthe in multiplication
matrix
the andb, A x=
The pivot element a,, is —1. We make it 1 by multiplying the first row by —1. Therefore, havewe then (3.28), equations
of system
the solving
are we If transformations
row . Bementny
using

D 1 =| 34/0 A7](115 [A]

(A| =| 3 -1 1/0
formthe to it
1 Of.
reduce
andT] [Al matrix augmented
the consider
We A. matrix
a of (A~') inverse
the finding
for usefulvery
is
09 — £2[1
We have Solution method
this However,system.
a of solution
the findingform method Gauss-Jordan
the use normally
not
do we Hence, elimination.
Gauss
the than count) operation
(larger expensive
more
is methodThis
vector. solutiontheis c Then, elimination
before
| , as made
be can
3.4 —1
panethe
Example 3.31
step, every
at Alternatel
c].| [I y, formthe obtain
to pivot
its by row each divide we Finally,
A=] Using Gauss-Jordan method, find the inverse of the matix
1 —I 3
12 —1 trans row
Lenaigt using
elementar y as
zeros, the
pivot and
above below we
make
elements onwards,
afindi method. eliminatioGauss
nthe in as is step first Theb. A™'= x as
same the
solution
to second From equivalent
is reduction
This vector. solution
theis c and matrix identity
theis I Where
-1/2]". 2 x=[1
Hence, the solution vector is

0 0 1] —1/2 0 0 1|-1/2 [Alb] Se

» | Ry +5R,/3 | le]
=/0 1 —5/3| 4/3 | =/0
ve
1 O| 1/2).
formthe to it reduceand matrix, Square
a iS
re | R, +2R,/3
wh b),| [A matrix
—~2/3| 4/3
augmented
the on
t o' 6 1
transformati
row ons elementary
perform
we method,
thisIn
2 o| -if? oO * to 0 BORST-SkS 0 | Method Gauss-Jord3.4.5an
4-5/4) 4/35) eae4g =o TF ~5/3|3 4/3] R;/(10/3) =(0| |
_
N=] redunda + 7R,—
R,WVS5, = R,
(24R, 1. z=
Since 1
and y=
10/3
lastn the
is nt, equatio
|
“mg
2a
a has
schiis.unique
i
equatio
|
of n system the Hence,
1. = x 1, = y gives
fo 2273| 4/3
obtajWe substitBack
ution
equatiothird
n, the From
z- y, x, of values
all for
| satisfie
is d equati
last
on The
3.51 Matrices and Eigenvalue Problems

Mathemati
eerics
nge Engin 3.50
 (A),

9
nullity
and (A)

1
15

21

6
rank the

6
27
Find

-2
QJ =6 3.2

2
1

38
Yond 2

babe

or
4] 1

36
3

ult7
es

ae

=
(iii)}/2 g Gees Rs

‘
EQ.
8 CV

@3-l
|
3.
=F,

5}

4ilt
'y
1 List & Zi | F =i" ft i. yb
~3 2 ac
0 -1 20
|3 AI :
by giveisn A wher0,e= Ax equaoftion system homoi geneous
followin
the g Solve 3.323.32 ple Example
"8 Sse 8 -5 1 4 1 4 ~ -13 = -5
Se
solution,trivial
4 [2 3 4.
the onlyhas0 =
5
Ax system
3h. 1-2 5. | 1
homogeneousresponding
: oa 6.
J sy Eo 1 -! 12 0, = x, is that
solution, unique
: : ;l
a has n) 2 (m matrix
n x m an is A where
b = Ax system non-h
theomoge
ee we, If ncou
Remark
5-5
18
1
Ax ysiem
|
x,. say0, =
t=
ae ae Me the
result, Hence,
of Xp,
#41 2b 212
solutionfixed any andb = Ax of7 Lema
e
the of
les
solution
a is b = Ax
ae
2 or Y wu -1 20

matrices. follow of
theing ranksthe determine operatrow
ions, elemen
thetary Using A(x — x) = 0. or - ~ AXxy = 0, Thus, the difference x — x

3.3 Exercise

=b. Ax, and Ax=b get we Subtracting,

2. = (A) nullity Therefore, arbitrary.
are x, andx, wherexy,+ 4x,= x, — x; + x, =— havewe Therb.efor
= e, Ax of
x, —2r4, —3x,
= x, as solutions
of family solutionfixed any be x, and solution
any be x Let
parameter
two
a obtain
we Hence,
4. is unknowns
of number
the and2 = (A) rank Therefore, Proof
any1S X,
solution andb = Ax of Solution c system. espocorr
; AY i fixedanyIS Xo re
homogencous ndinthe
g of
OO O 00 S30 =3 =f OQ -
h en !
; pore 426 whe +x, X9 = xr OrmM
fj | the of are solutions
2/0 3 1 =|0 2)/0\/R;+R
3 I 0 x 4/0).
l 3 O)=|]2
(AJ (iii)
; 2K,
— R, esis - EE
176 =—f rt 1'o a : 1/0 =—1 t 1 a possesses always ee =
wns solution. non-trivial
in 1Ons equati m of system €neous
no : homAog (b)
0. = (A) nullity Therefore,solution. for q
hold not C&S result ThisP. %
trivial
a only has system homogeneous
the unknowns,
of number
= 3 = (A) rank Since a also18 Bx, Tl
systems. on-homogenous
ox | then em,yssyst scal any homoge of
the nous
lOogeneo7 us for
hom ) system _
-2{00 0 linear
a of solution a:
40 -3 Jo " * 10} -! 1 . twos 4T€x Xz and
! x, If (a)
2Iol. -1 =|0 -1
2)0|R,-3R, =|0 ais -1/01 (i)
(A|0)=]1 17
SS), Theore Remark
-3/02 I -3{o2 fi _, 2-3/0],
1 (scm7e = (A) nul+lity (A) rank
callISeg
dimejtns
s ioan
nd aes
resultthe obtain
we Therefore,
A. of
Space i: system nullity
he
Beneous (
P homo
the of Space
Hence, nullity (A) = 0.
solution.trivial
a only has system
the unknowns,
of number
= 2 = (A) rank Since, “nace Solution
The
: ee
(n a8 7)“~: tr dimeofnsionspace vecto
a r form whichone arbitrarilchos
y, been can
0 0; 0 0 1/2) 0)
ss ; 3 j solutio
Of Parameters
| 2/0 3 homo~genecng nb
If y.s< (A) rank that
; —3R,/2
R, requiwe
re ree mile para
r)mete
~ (9
-—3/2]0).
=|0 R3+R/3 —3/2|0)=/0 -1/0 (AJO)=]1
(i) r a0
(trivio= x since Therefo
0).| (A rank
= (A) rank case, this
n . non-triy;
a have
to System
0 1; 2 0 1] 2 0 LV 2 (3.34) sj CONsiSstentalways
is system homogeno
The=f
us fests
a always
ig solution)
:

A
x
A
ig
anom

matrix,

reduce it to row echelon form. We write the augmented matrix (A | 0) and solution
Where

oeequatiof ons system

homogen
the eous Consider
Ax=9

Equati
ar Lineof Peat
System ,

and
Homo6 gene

3.53
3.4,
ous

Matrices
Problems
yy Mathematics
EnEngigine er
2neer$52ing

Eigenvalue
(3.35) (A-AI)x=0 or Ax=Ax
homogeneous
system
equations
of
the
a :
fe w
afl") Eid s|
Let A = (a,) be a square matrix oforder n. The matrix A may be singular or non-singular. Consider
a
Eigenvalue Problems 3.5 “bt,a on dy aa y=f2
Ni to
[4] aed tora ' ha a
1001 a oe a
| 44-44 92% cpr
A ME [a afte] -1 fr
J 7 ¢
50. j22) 4}fz] 13
ee =| “ 7 af |.
-3 2 3.1/0 16]. 3/lvj
i686
r
2 =!{1 30. /y}/=]
2} -4}.
-9 |3
4 = fof 0 =3ITs
®
29.
2 7 Ullx| 1 3 dais
Ey
yxy
3 PA" |3
2 9.1 134 26/{z}-8 7 1 ate] 12
a ae (es ica 31.3 }1 47. |-2 -3){z]
1 2
31.2 {1 46. -2]//r
8 /=[6].
(3 28. Silyl{
eo 5 =f2]. 27.12 3}. 2i[/¥J}-!l
=] [I
1 23 8 Flix
26.
ae a 1-4 | 7ifx4 [
ar |2 = «ffir 2-3
system consis
thetent
for soluti
theonsall
matrices. following find and consis
aretent systems follow
the of inverses
the find method, Gauss-Jord
the an
Using theing
of which Determi
Oy).4, 0. (3, 1). 4, 1. (4, 2), 0. 2,
-4 2 42 {(2, 45.
1). 0. 0, 1, 4.1)
(4..Ne hb
2g 29 3.2 4.01;
1.0, iG, at
| ee: 45.

| —
14)B. ERE
(14,
2 HL
4), )
3, 2. {(t. 23.
*S Se Lop 34,1,
rT

4. (3. B35). Bote Teoh(26 223.9(01,

), 22.
OSETe, BG (1,8) 1),
Bay 3 ee ey Ik, RAMEE(dy 2by
oe 3| 4 to
8 iy} ~f, Cr
1-5
4 4 os oe mis 273 io ee OE 1h D3. 1, (hb. 20,
, 44, ; . |
3 a7Ll 42. 42 19. ays
ro

eesee ee 1 10 04 1 ei, Spa, =F) 0)..(20,

. BS.
led

=f W Bo
a mn . Pa.
Bo
(=> @ f= i WS vs
|0 D4 ee =,
é

2)(s (4,6):
4). ((3..2.16;
dimen
ts siondis0 Find independent.
linear
is ly vector wing
| oe ae 2 1 P of s set | follothe hether
9 5 -l w Detennin
= — RB 126
4 j ~ 41. ‘acasplt
: ; 40. ft tt Tr) * ol ap oe iS ~§ 4 5
ae ee Pee.ie 4 123 3 SS & mal
=
2 4 bE
4 —
Pap
=
eop | .. 14.
, 2
4] (} | 2 —3 2 |
-6 9 4 il +7 -l «4 4 1
-101 |4 38. a ee eae <a BF =e (1 $6 ‘ LS sf q |
at q —1 |
3
5 =]1 4 Ean ian = ao x =] | fe oh cE | ea ’ )
following
the as. given
is A where
0, = Ax systems homogeneous
following
the of solutions
the all Find
= !
oe ErteMS bevy rerf the of rank the deler
| mine
ACMOTYS
Pp
reat oper: column
¥
voli e eleme
(he the
ntaryNEUsing

5 ay 3 £€ *
ft
Psd
oe
-_
al

P= J
—
“—
i

3.55 Matrices and Eigenvalue Prohlems

.
VMeithau
atnc
s stye COP
al
C
cy
 a
(3.39) EO ec,At eo (1%
+ +... oA AY
|AI.= eigen
ofvalues productand }, (A
equation trace
= eigen
ofvalue
sums
characteristic
ownits satisfies
A matrix square Every theorem) (Cayley-Ham
3.9 Theorem
ilton
get we Ther
€quation, Ay ... AA,
= c, = (-1)""
= c, |A|
characteristic
its andA matrix
a of relationship
the gives which result important
an presentnow We
get we then(3.36),
Eq.in0 = A setwe If
hold.not does property
this complex,
is A matrix
the When coefficients).
real has (3.38) :
equation characteristic
the (since ze = A, ... A,A,
eigenvalue
an alsois i — a conjugate
its then eigenvalue,
an is iB+ @ if A, matrix
real
a For 7.
3.41), Example
(see eigenvectors,
different
but columns)
by or rows
Co n=
4; +A, +... AAs
+ Ayan
by expanded
be can deter
a (sincemiman eigenvalues
same
t the haveA’ and A 6, 4a, +
k.
scalar
+n ay c= d= t+ tagAy
any forx is eigenvector correspondi
the andngk) — |/(A eigenvalue
the has kI)"'
— (A 5.
coeffthe and
icients, rootsthe write can we
(VA)x.
= A''x or AA's
= A‘Ax between relatthe
ion usinBy
g eigenval
equation polythe the ues.calleare
d0 = nA)
nomiofalA, ..., Az, A,, roots The repeorated. simple
get we Av',by Ax = Ax of sides both Pre-multiplying roots
vn has0 = comp
orlex,
realbe can .
x. is P(A) equation polynomi
The A.
al
cigenvector correspondi
the andng 1/A eigenvalue
the has exists)
it Av'(if
4. is equation
matrix
the of equation characte
the ristee
ThisA. matrix
the of a, elemthe
entsof
ic
term
ins exprbe
esseoan
d Cy -.-5 £2, Cj»
&)x.
= (A x= kI)
- (A or (3.37) were
0. = (-1)"
+ c, — RP?©, +
Ax=Ax=>Ax-kix=Ax-—kx oA?
— A” or
=0.: (-1)"'¢.] Aitee
x. is ¢igenvector aeOy - [a” (-1)= " AI]
correspondin
the and& g scalarany for k, — A eigenvalue
the has —&kI
A 3. =|A- PAA)
. . :
result.
the obtain
we times,
m successively formthe of
Pre-multi
is which
J, in n degree
of polynomi al
x. is plying eigenvector cofrespo
the and
nding
J? eigenvalu
the has
e A* Therefore,
a . obtain
we . ,
(3.36),
Eq.#in given determin
the ant panding
= *
: Ex
2x.= Au or A(Ax)
= AAx= AAx= AAx nr 7
| ae an» Bay|
get we A, by Ax = Ax of sides both Pre-multiplying (3.36) ‘
3 ;
whi :
mm. interger positive
amy forx is eigenvector , correspon ee ay In
the and
ding
A” eigenvalu
has A®e 2. : = AI)— (A det
g --- Gy,—-A a>, |
(@A)x.
= Ax @ => Ax = Ax
a, a, a4,,-A
x. is eigenvector correspor
the and
d*=c@A eigenvalu
hasA e@ 1.
, “
results. mar: Matrix be must
— :
coefficien
the OF t Therefo singular. :; matnx: coefficien
|theic tis
t that fi,
7. than lessis AI)— (A
ollowing havewe of rank the then solution non-trivia
a hasl
Then, eigenvector. correspon
its be x
dingandA of eigenvalu
an be Ae (3.35) System homogen
‘ eous
followi
the |
= S Prope th If
eigenvect
and ors eigenvalu
of rties
es Eigenvec
and tors
‘ 2
ues
Eigenval35.1
Matrix square
sponding
¢1g eigenvectors. correspond
the ingobtain
to 7 se s ..., 2, pee 3 Ric *. a of problem.
es
j=], ultiple ponding
eigenvectors eigenvaly
an called
is A
correspon d
the and eigenvalu
¢ es
air | :
| : .
A. of gen
eigenvalu
the are
esA 3 matrix
the of elements diagonal
the then 8 _ co anyis @ where
X, @ ten + |
lower
a or - matrix, triangular
S h ther
hom theof solution
4
triangula
upper
r an or diagonal
a is A if that Note0. =
; ngy If A.
of Véciors CUAEACIOR ISNG
= (3.35), system mogeneous ,
|Aj then seit,
is tiel enw homoow
the of solution non-trivia
# l
the of oneif Conversel be
zero.y, must comrespong; ete
oopd
- tow: sleet
eigenvalu
the of esone atleastthat find we 3.38),
( Eq os homogene
theOF the called
arex vectors solution
Sauions
ye ae4 ch the or eigenvalu
the escalled
are non-tnivial
Exist, 33>) eta
then singular,
is martix
theis that0, = [AlIf (A).
p by demeacd
16 aadA OF
is magnitud
Féclliis suectral
dis4 bie Ongce homogene
the ous
which
SOF
: in
- e eigenvalu
largest
e the andA of spectrum
the called
is : whifor2 of valuesThe solutions. non-trivia
has l-35)
eigenvalu
the ofesset Cyuat;
Ge wenn ‘u woe
es ae
The ale
ordeoffindto matrix trivial
a has
needWe identity
N a solution. always oO
|
js I and sca4laris J
Where
3.57 VWarrices and Eigenvalue Problems

Mathematics En
gineer
3.in
56g
 1D
1
martrix

1
0 O
e th for theorem

=|
Cayley-HVeri ton
amilfy
3,33 Example

8 3 0
=

3) Of=]
22
18 9 AZ

16 Il 10
WW 12
(3.41)

[3
0]
ite

Oo}
(- ,

2h 2}4+]0
+ ws “24 cA"
~!

cE f-1
oar
= ! A"

15 12}-]-1

L-2) 12
as A of powers lower
of terms
in A” obtain
to

we
Eq.(3use

get
4
.39)can We (b)
"
'b

From

1
3

3
O
6

o
Eq. 3

0
(3.42),

3)

lo

af
(3.40) er (ryie’#

-
Al?
¢;

0
:a as

1
[A" 1)"
: ly ae A| or

A5=3a7-A+3At

=|
=>

3
3 6

0 0
1 0
4
65 3 4
-1

|
0)
'D=0=A AU

(1
6l+lo
(-1)"c,
+ ‘A C,\A te)OP se PG AA )

4|-|-3
i n-}

712
; an! =I c A® A

=
get we A-!.by (3.39)Eq.in sides both Pre-multiplying

get
[a?-3a+I]
A. matrix
the of powers

From
the of termsin exists)

—
it (if Aw!

+
findto

(3.42),
e ae ee a (3(3.39)

A>
9. Eg, use can We (a)

the

347
(1) Hence,

0
Eq. satisfies
19

equation
Remark

00

0
eee - theorem.

-A+3=0.
the

we characteristic
proves which

=|0
= fl — + A tc (-—1)
+

0
eon ] ee

(3.42)
| i

O|=0.
; ae A" 2 — A”
get we adding,
and respectively
| ae A" A”, by
n equations plying
-quati these

1
Pre-multiplyi

16
17
~1
LT
Ss)
24
c,1 (-1)"
= AB,

G09 10
={
1sh
11 ca
I | Ch| ~ y" (— = B, = { AB,

10/+3] 12
10341-|-1
a 2}+3]/0 0100
We
have
cl = B, == AB,

St

1
6 3
G O

12

7 10
c,!= B, rs

15 OW
tO
4]/-1
AB,

1 2]=]
Ate

4
af.

12—j
Ol
4]]
f-t
I = B, =

A"A@=|

-A>+3A°-A+31=-|
ain

5
obta;we A, of POWers various
of coefficients
the Comparing

ia
| poe nel _

1
;

4
3
06 0
” ( ‘ Al a +1)"
7 4 a4 <2 =A

|
Now,

2
Gy) * a—[“>

oli
‘

0]
2|=1

f-t
2/|-1
A+R +Bnit

12
= B.Al
+

#4
(B,A""!
AI)—

iht24
(A

A7=/-1

-
A)
A any for writecan we Therefore,

(A?

=-¥

+30
=(1-
I All (A-AI)=|A
aly

2
(A-ATD

24-3)

-
-
B..4 GA

-A+3=[-
22-3)TZ 3
TR Pree

4)
=

-2
-l

401
have also We

Ay?

-2}
1-A
2 qu K

2/=(1-a) 0
2
eee BA"
= Al)- A fl ad

(1-4)
Clemeny,rane write Can WE wy

1—A
a a. Matrix:‘ie
” order
of matrices SS

|A-AIj/=|
an : ae
Sol

* Sic
reue Mose Sri
jution
dre| x : l* elements
the of
The characteristic equation of A is given by
TVAT idjisjthe cApPress. fiinctions
as
can We eMee. B. fs 3 clei SREB
degre.Auf white
JL in polynomial
Giiern
the of eerueseacice
Wry a a i hee Poe deeree
of A in polynomials
also
<quares of those of A, (ili)find the spectral radius of A.
jp ‘ ni: : adjoint
the of
MaksThey phan] ments
ele the Therefor
lesse,
.or 1) (—
of A? are (i) obtain A”’ and A°, (ii) find eigenvalues of A, A’ and verify that eigenvalues
. achtyt Pt H deters
the ;Of nts elemthe
e of cofac
| tors
The= Proof
3.59 Matrices and Eigenvalue Problems
Mathemati
rine
cs Engines 4.38
: 3
 Aw12e of a°-24+2=0, o ; al" | IA ~ ati

1]
1-A
The characteristic equation of A is given by (ii) get we recursively,
equati
this
on Us
3. n2 —-I, A? 4 A"-2
= A” or —~|l, At
—n
Therefore, the eigenvalue is given by x = (x, x)" = x,(1. 1 or (1, 1Y
AZ AN
2 |
x}.= x, orQ, = x;— x, or SE}-(0)
¢ x= 51) (A— get we these
equations, Ad
A?
i A’fare An!
— A"
havewe 5, = A eigenvalue
the to Corresponding
At
— A’?
= Af®--
ent AD!
3]4, [- as eigenvector
the take can we multiple, constant
a upto unique
is eigenvector
an Since
1} x2 x2 A =A?
A} A‘-
4/3ss |-=
2 r=
—4x,/3 x [*]-|
A’-A?’=A-]
by given
is x eigenvector
the Hence, obtain
we A, by successi
sides
velyboth Pre
8
(2)
(3
=e
+
=
4x,
x,
)(0or
3x, a Sar e aa " A+2Dx-= A?=A-
A? Lor A>-A?-A+1=0
4 0 4\(x,
3
get we theorem, Cayley-H
Usi
a
havewe -2,= A eigenvalue
the to Corresponding
1-a 0
1-0, B-R
o -as -Aa(
Lj H-D=1A0,
A=-2,5. or =O, A>-34-10 or 3 »-a(7 |A-AI|=
|A-All=

«a
ae
| or

i
The characteristic equation of A is given by (i)
by given
is A of equation characte
Theristic Sol
Solution
-

|
>7

f
+

3.
.

= A If

3
3.34 Ex
A

A°

A»
A™~
&

find
—I

-I
-.
oOo =—

iia
al,

92
A
that

forn
how

(ij
A=

2:
(un)
Hence,

A=
:

|
|. l

2|
©
©

(i)
then's

1 i}

3 1}.- 0
3 = |A,| max= magnitu
in de
Example 3.35
eigenvalue
largest
= (A)P
Find the eigenvalues and the corresponding eigenvectors of the following matrices.

: by given
is A of radius The
spectral (in

—_—
|] 0 | 0 1 0
A of eigenthe
value
of s squarthe
es are which

=
1 O|-24)0 1 — 1, — 9, are A? of eigenThe
values
1 O}. 1 O}=|25 A”? = 25A7 -241 = 25

—
0,= 1) (A+ (A-9)
I
0 OF; 0 0 0 0
8A-9)
- (A°
=O0or
1) (A+
(A+
Or

Substituting m = 50, we get

6
= 1) - (A?2) - (n ; 4 Arta-2b
=
S-A 0
0 = 24] A) - A) (S - A) [G3 - — (-1 = 4 3-A0
gg “I-A
as
by give
isn A? of

i
equation characte
Theristi
A=3, +
or 4,-;1)=0 3)(A* (A=or

li
ee ae

A"
oak
(A"4 + At -1) + A? —-1 = Att + 2(A7 - 1)
of rootsthe areA of
Eigenv(ii
a
3.61 Matrices and Eigenvalue Problems

Mathematics Engi
.
ne60er
5:
 > ran
oe e
—e_ oe fe Sos
Saf
ata
" aS ee] oh te res Por
r tas

ik
Seti as Fars Tp Me
Tas 4 MOE
ig heoe ha
LMge ae ea at
oeis

F
oe

-
ne SS, , [eee 3
isiee2 ee Fee
6 pyr tay
te sgedite
Bhi

a
tae bg wet scuualig

Pe

=i

ry
’
Se
et

a
aos pe

*)
he

;
et
8

ele ian
Ti

YP
as taken

es
be can which eigenvectors, independent
linearly

;
three obtain
we Hence, variables.
threethe all of values arbitrary
for satisfied
is systemThis 0]’.1, [0, as : | ,Be idle
eigethe
nvector obtain
we 1, = x, Taking arbitx,rary.and0 = ¥30,=oat
0 X5 0
x or
O 0
ts,

0}. |= ty 0 O 0 -I)x=
(A 0 | x; | Fie Poo
(iii) | a tg
1||x,|
0 0=|0 (A-21x=|
0 x;| 0 0 —|
eigenvectors. independent
linearly
two hasA case thisIn
o]7.0, =[1,x, have we 2, = A eigenv
thealue
to
1)’, (0,0, x,= Corresponding
solutions independent
linearly 1]. -1, [—1,as eigenve
the ctor obtain
we 1, = x,
two obtain
we 0, = x3 1, = x, and| = x, 0, = x, Taking Taking arbitrarily.
chosen
be can x, x5, x,, variable
the of
s One unknownthree
s.in equatio
twons obtain
We
= Jo O|}x,|
0 0 {/0 x, 2]|/ 0 2
X,,Xy
arbitrary. +x,=0
xX,
i or x,|=|0
0 ]|,
0 (A-I)x=|
(ii) 0 or

he.
= Xy ]}0] x, Ill 5 (A-I)x=]0
Te
0 Ole ot [}0 x, Off 0 0
eigenvector. indepe
onendent
only hasA Hence, havewe 1, = A eigenv
thealue
to
0)".0, [1, as solutio Corresponding
then obtain
we 1, = x, Choosing
3-A O 2
arbitra
x, ry. 0) 0 00 3. 1,2. A= ox 0: 2)'=
= 3' =a): A)(2 (l= -or = 1
=! x, or 1}x=l|0
0 )|
2-4 0 IA-Alj=/|
(A-I)x=(0 (i)
0 = At; 0
QQ O32 Id
010

by given
is A of equation
eigenvectors.
following
the obtain
we 1, = A eigenvalue
the to Corresponding characteri
The stic
(111)
compl
beex,can eigenvectors corre
the
spond
and
ing eigenthe
value
A, s
eigenvectors. independelinearly
nt of matrireal
x a For
numberlesser
hasit or eigenvectors, independelinearly
nt3 has matrix giventhe whetherknow,to important
is it 20 Remark
eigenvalue
3 has s, matrix3 x 3 a Since value. repeated
a 1, |, 1, A= are eigenvalue
the s Therefore,
0. = A)’— (1 as equation characteris
the ticobtain
we problems,
abovethe of eachIn Cigeny 1)’.[i, = isx
Solution the Therefor
+. — | xe, get we 1, = x, Choosin
0. = gix,+ ~*);to reduce equation
the s Both
1 00 0 1 0 0. = 1X5r ty = and
0) — A + ix,
O}. 1 Or
A=|0Gn) 0}, |0 A= (i) ES
100 110 6 “ih fs
0) | -aI]x-=
-(1l[A
x, iol ;
matrices. following
the of eigenvectors correspond
the anding eigenvalue
the Find
s 3.36 Example
havewe i, — 1 = A eigenval
1)’. -1,0,[ as the toue Corresponding
eigenv
theector obtain
we 1, = x3; Choosing the» Weg [—j,= x is
Uher|etor=e
=

Bux), 1,0e &5.= Choosing eigenvector

}{0 Ijlx,0 2 =O,: Ny — Sy = 7 to reduce equation
the s Both
eet 4 or |;O| 1x,| 1 0 0, = ix,— xX; - and0
(A-31I)x=| x)=+ ix;— oF
(a7
10 xj 01 0) ~=2
() Ay aoa —|
[0 | May ~~ x= 1] +i) [A-(1
havewe 3, = A eigenvalue
the to Corresponding

havewe i, | = A
3.63 Matrices and Eigenvalue Problems eigenval
the toue Corresponding

Mat& hematics Engineeri

3.6n2
 are similar.

Therefore, we get P = } | which is a non-singular matrix. Hence, the matrices A and B get We Py.= x Set
APo!y.
= P'Ax :
obtaiwen P~'| matrix inverantible
by
A solution to this system of equations is a = 1,4 = 1,c¢ = 1, a= 2.
sides both Pre-m
0. 4d= Sc—- 36+ or 4d,+ 35 — ™= Sc Ax,= Ax
+c-—2d=0
3a
is ThatA. of eigenvector corres
thepondin
or de,+ 3a -— = 2d— Se be x g and eigenva
an belueA Let Proof
-6b-2d=0
Sa or 2d,+ 6 = Sa matrix. similar
theity
is P where
A, eigenvalue
theto corresp
B ofonding
4a-24b-2
or c=0 Je,+ a
eigenv
2b - Sa an is P~'x thenA, eigenva
the to
lue corres
A of ponding eigenve
an isctor
x if
same the hence (and Further, eigenv
equatio
of ns system
the obtain
equation characteristic
samethe have
we elements, corres
theponding Equating matrices Similar3.10 Theor
-3b+4d |-3a+d e Se} |se-2u. matrices.similar
of eigenvaluesregard
é =
™0 4}fed}[-3' = 0}. d}|-2
ce result
a prove now We transformation
similar
a . itycalled
is (3.43)
b+2d a+2c Sa] Sa-2b Eq.in transfo
Thermationmatrix
& 2fe [1 S| 4 Alf |? Gy similarit
the y called
is P matrixThe A. to similar
is B if only andif B to similar
is A Theref
non-singular. B. = PAP™'
is P whethercheck then and BP = 4 c get we P~',by (3.43)Eq.in sides both
PA that such
d andc b, a, determine
shallWe b Post-mu
(3.43) PA=BorP. A=P'BP
BP.= PA or A=P"'BP that suchP matrix inverti
an ble exists there
matrix
the to similar
be to if B
that such
P matrix saidis A matrix
The order, samethe of
invertible
an exists there
if similar
are matrices
givenThe Solution matricessquare
be B andA Let
matrices Similar
4) n=l!ans J [t ae 2) Jeen-[!oa=[E Matrices Diagoand
nalizableSimilar 3.5.2
|
|
where
B, to similar
is A whether Examine3.37 “xample independenlinea
t rlythree obtain eigenvectors.
we and0 is 1 — A matrthe
ix of rankthe (iil)
3.36(iii), Example
QP. R= where A=P"'Q''CQP=R''CR,
Then Pendent
inde two
linearly obtain
we and1 is I — A eigenvectors.
B=Q°'CQ.
and A=P"'BP matrix
the of rank the 3.36(ii), Exam(ii)
ple
that suchQ andP matrices invertible
two be thereLet t eigenvector,
independenlinearone
ly obtainwe and2 is I — A
C. to matrixthe of rank the 3,36(i), Examp
(i)le
similar
is A thenC, to similar
is B andB to similar
is A If (b) in | ihe fedWe
F 3 # multiofplici
is |ty= A eigenval
the ue3.36, Exam
Inple indenpendent
linea. rly
e
eigenv ec
Ing Tespon
corrding
the‘ therefand
ore, distinare
ct
Two matrices which have the same characteristic equation need not always be similar.
| eigenthe
value
all' s 3.35, L
equation and hence the same eigenvalues. However, the converse of this theorem Is not true. Examp
: In le
| |
(a) Theorem 3.10 states that if two matrices are similar, then they have the same characterstic
21 Remark
Sps
|! m.Al),
— (A rank r=
Remark 22 where p=n-r,
is
eigenvectors of number by given
is A
= y or Py= x
P7'x. by are B andA of
related with associated
the thenn, ordeofr A eigenvectors independen linea
t rly
that Py,= x Also, samethe
eigenvalues). hence (and equation samethe
characteristic have matrices matix squa
areofm multof
iplicity
the as is B
same eigean
nval
is ue
A If 2.
similar areB
Therefore, matrices,similar andA Now,A. of equation characteristic independen
linear
t are
ly eigenvalue
distin
s to ct corresponding Eigen
1. vectors
of equation characterstic
the ts, thatA, as eigenvalues
samethe hasB Therefore,
P-'AP.
B= where
of eigenv eigenvectors. independen
linearly
t correspon
the and
ding
Ay.= By or =Ay (P°'AP)y
or P'APy=AP'Py, thealues between relation
the ship regarding results important
some state now We
1)’.0, [0,* x,
3.65 Problems Figenvalue
and 0)”l,
, [0,= xy
Matrices 0)’,
0, [l,= x,

Mathematics Engineeri
3.6ng
4
 ie
oo
A®wil=

=
(3.45);
P'AP=P'ppep
obtain

=
we

‘aed
P™',by (3.44)
Eq.in sides both tiplying invertib
the that
matrix Show 3.38 Example is P matrix
the therefor
andne is P of rank the independent,
linearly
are ofP columns sm
Q(A). polynomial any for
matrix (3.44) PD.= D X,) «+5 X35 (Xj)= q) AX«.., A2xy, (Aix
= ,
= QA)
PQO(D)P"! Ax,)..., AX, (Ax)
= , My] -+-9 X20 [x,,A AP@=
we
obtain m.
Hence, integer any for
positive have We A. of matrix spectthe
ral called
is D andA of matrix modal
the called
is P matrixThe elements. diago
P-! A™=PD"™
and A=PDP" itsnal
as A of eigenvalue
withs matrix diagonal
aebe
A,) say An, (A,, diag D=
havewe Then, diagonalizable.
be A matrix
a let Now, and x,,] seey Moy [x,, P=

| O(A,) 0 Let ...,", 1,2,/= A,, eigenvalu

the to e correspo
x, nds genvector
the is that order, samethe in A matrix

——S=——_E——
the of distinct) necessari
(notA,
ly --+y A2y1s stiseasaes
z a) = Q(D) the to corresponding eigenvectors independent
linearl
n ybe X, -.., Xz, Xi, Let
is A then eigenvectors, independen a
diagonali
i linear
t » ly has A if that case the prove shall We
0 Q(A,) of Proof

get we thenD, in
linearly
m hasit if only andif diagonalizable
7118 reer eigenvectors. independent
polynomial
a is Q(D)
If m. integer positive
any for — . ord of A matnx
as eases Square
A “ 3.11 ;Therorem
. theorem. followi
the ng
in given
js P of existen
sufficie
andnt necessa
A ry A. of eigenv
thealues thecefor Jon
are D of elements lagonal
ae es count
samethe have matrices similar since, matrix. diagon
a is alD where
D, = AP a aa.
ible invert
anib]
ee eeCXISTS ar ; = P~ that such
P oes
matix
there
is that matrix, diago
a nal
to simila
is itr if ;e
; eae di ois A sematrix
A

.
matrices
andn, order
of matrix diagonal
a is D If (c) Ee SLO
druvon
IS 4alizad
T | Diagonalizab
matrix
the that say we cxists,
then Ox x matrix
a such
If matrix. gonal
A”.is so diagonalizable,
is A if
diagon
a is D Where
Therefore,
'=aA PDP or D=P''AP
m. integer positive
any for PD"P™
= A™
2iven
4
get we iT gob ll .
' have
. we A, X Matrix
times,
m (post-multiplication)pre-multiplication
the Repeating matrix P such th When A
ooest matnix. Thus, our interest is to find a similarity to obtain the Matrix P ae
Or Bis a diagonal os 2 Ses A and B. However, it js Possible
the

=
TES ALIS which
Po matrix
E ays
non-singul
falta te

A?
OT any two
- a ob to difficult SUAHY|
A=P 'pp for
:a ; fainar dif% usually
1s 1 ce, practi
In
4 {) are | OT
RB af 1 i d : ;

we
(b)
Sirenv
theilues

Eq.
where
| I,

PD’P™.
as are A of

(PDP')
eigen
eethe
value
at th:s

From
sity]; thot are venti
verifj
e beed canIt

obtain
restriction imposed on the cigenvalues of the matrix A im Theorem 3.11.

(3.46),
| py Bare andA matrices
the exist,not does
P matrix
ble
% ; 4 mveny
an ££
when some cigenvalues are repeated (see Example 3.34(iii)). Therefore, there is no
eigenvalues are distinct. The matrix may also have a linearly independent eigenvectors even T * at
Since

=AA=(PDP°')
matrix Stnwular
a is Which
}.
(a) A square matrix A of order m has always n linearly independent eigenvectors when its
= P Weigel Thetetore,
Remark 23
(3.46) PDP.A=
aa

iad
1T—

obtain
we P”',by (3.44)Eq. in sides both Post-multiplying start we elements. “Ponding
its diagonal form. Corethe Equating
t | | ie i Bil '
i} L :
which implies that A is similar to D. Therefore, the matrix of eigenvectors P reduces a matrix A to
| il [«
|
|Oi
4s

! #2 its ir 0] |4 - (11)
,
l

f
1

3.67 Matrices and Eigenvalue Problems

SU

Fh,
*,2
fe

MEE

.FITTS
tots

Miatheoniar
 10 -Silay]
=-2 ~{
0}, * | fywo od +
(A - 5I)s *
0) ei ~3 2 -]

27 9 lo 3] loo 15] 0 10 909)

9 17
Rigenvector corresponding 16 the eigenvalue A = 5 is the solution of the system
Olei
40 9141jg
0 Oisl
4o
) *+3s Sle

i
)
«4 a2DBD+3l-10
~ J, = 72 =—|
é( A
\A~ AT =| pa
2 {-A ~6|=0 oF 2? + 47-24
= 0,
-45
or A= §, -3, -3.
~~} 2 -3} ar’,
4 6p + PCD 31= $44 + A’ Ther
(ii) The characteristic equation of the matrix A is given by
9), 4, (1, diag
= D* 3), 2, (1, diag
~ D haveWe
3) 2, (1, diag
= PAP
thas verif
diagonalizable. beied can
it
the matrix A has only two linearly independedent eigenvectors, the matrix 1s not |] 9 ~~ Lirgy
5]4 2 Pim} and Ot} laa Si
obtain another linearly independent eigenvector corresponding to the eigenvalue A = 1. Since
Since the rank of the coefficient matrix is 2, it has one linearly mdependent eigenvector. We 4 -1 -l 61 tid [
|0 Oljx,|
2 j-} |
by give
isn matrix modal
the
101 j= Jlix,0 0 (A-231)x=| | Hence
0 2
2Zilx, —i
Lo} dy -J Jj So
11 =}x,is : uti
solThe |on
0}io!= 2/|22|2 - |=|-2ixjx — (A
system
the 2.
haveWeA= 0 fol falmd 3 fo
eigenvaluerepeated
the to corresponding eigenvectors
the find first We 2. 2, 1, A=
: syste
the
m of

or
soluti ution theis3 = Aj falocgeri
¢7 the to
a=0, -[22-222- correspondingOSSWertor
(2-4-2aJ)=0-AO-
=(1-A)[(2-4)
A0C- If jo! Ojfx} oF Ma
2-4}
2 - |}
x is solution
The |0\|= . oft._1 2I)
— x=[
(A
1 2-40 -2
=| IA-AI| fo fx, —I}y Pe.
2 2 }1-a
by
given
(i)
equation
is
matrix
the
Aof
The
characteris tic
Solution

is diagonalizable. If so, obtain the matrix P such that P"'AP is a ae

er
06 2 -1 2 2. =I
1-64.2. A=1Gi) 1 =~ 2 the-dhpei
to: vetees
fi)A=|)02 cXMRCEpON
NNAN
NNIling
~
VECIogs
SV eizen independelinearly
nt threehasit S°
Vales, 0= disgonali nce
is it cizable.
=f 2 2 3 12
d three
128 rad
by given
is A where
A, matrix
the whether Examine3.39 Example © +
=e. | ae a A
has the Since
matrix
& =
2-627
+112-6=0, «2 ~ |
19) § -~2 1 oO 27//-1
oO ouyo ft z | os > =| IA~AN
181. 9 -1/=l-18
1 2) of] 17 i]00 +5A+31=|-1
Av [3-Z
-8 8 25 [ 1a] -1 olf-1Oo O}19
1 1 :

matrix
the obtain Then,
, matnx. 5A 425 B=
diagonal} adaam Saree.
Hence. WE obtain
4 Is 31 +
P'apthat
: suchP find
-
3.69 Waetrices and Eigenvalue Problems Hence Zable.
hemati
Matcs

Er
 l
(0
O}

in
an
10]

set

iR°,
(i)

form
}

1}{O)
(0)
f
AEN SF SOO GANAS TE ICL SNICEDER 4
are yijy|!
¢14 all o ONGot “t-¥
x/fixi.then
0, = vy - 1 and vectors
any are
¥ x, if vectors. orthonormal
called

orthonormal
are -l 1 Ol—1
| 1/0I A=ppp'-!
llLiwi=1
izi= and =] <i The
x-¥=0 GIT FEO Lo.
|
which
forW andx vectors
The vectors Orthonormal
vectors. orthog
betoonal saidare
0 = y - x which 1 4]
for
y and
x vectors
The vectors Orthogonal I
| i= sin] _ that findWe
vector.Unit
@ |
always
is x/!ix\\ vectorthen
@,+ x If 1. = ix}
if i -1 1 i
vecto
unit
ra called
is x vector
The vectorUnit | -
x. vector
theof normtheor /ength
the called
is 0 9 -1 -I

oS
2 +22 Yap =
+ x5+ 22 Yx-e = HY
1 m9 en 5l

=—_
r-le
matrix Mmxia]
l

&
ThenC*. or IR™ in vector
a bex Let vector) Of :
a of (norm Length
B.a. scalars
and
z y, x, vectors
any for
2) Bly- -2z)~+a(x z= By)-
+ A. matrix
the Find respectively.
(ax | are 1.31. 0771. [1. and —1]
3. }’
fo, -1}",
©. [1.
eigenvalues
theto correspondi
A matrix
ng3 x 3 a
that verified
easily
be canIt of eigenvec
Thetors3.40 Exam
=. =¥- -yx that Note respectively.
y and x cf vectors conjugatecomplex
are
y and
x 3).— 3, — (5. diag
= P'AP that
where verif
beiedcanit
et =!
(ioe |} @ =|
xF- a+ y-x=y
ae Fur¥
= 3 x-y=x'
6]. 4 <|-2
= p-'! and O I Pu el
as defined
is vectorsthese
of product
innerthe then C*,in arey
; and
x If | L233 3 —? “4 I]
08. x= if ooh andif =0 1-1 and thatx-x20
Note
« y- = y x- case thisIn y>. x,< by | rk
denowd
also
is product
mmerThe scalar.
a is and
y and
x vectors
theof product
inner
the called
is | ; p aoe
matrix modal The by given
| ea. threehas diaco
is naliz
A able.
matrithe
x Therefore,
matr
3 ix
x 3 givenThe 1]’. 0. [3. eigenvectors. independent
eke2 ES
eigenthe
vector ob 1.
we ta *3
in
i247) SSPE
: - = | telasad
ne 0" (2:1.
ind arly eigethe
avector
rnpe Tr
y :
= =
Then IR”.in vector
nwo be
sw andx Let Mdependen:
inearly
vecters
ef product
(dot
) ProductInner : |. is Tmatrix :
coeffici
theof
ent rank The
following:
the define
we
ThenC’. or IR*m « == - a Iv x Let |0 x, 3 ~2 {-1
dimension
of vectors
reobe ¥,)' ---. ¥> (3).= ¥ and x,)' x (x. =
ss
35,58
- Sy * ie i ee 2s 7 =| bait
.
definitions.
: 7 =
some :ive first We areas. many
im applications
have matricesThese oe
matnces_
these
of eigenvectors 0 7 Ee3
7 H
= —theeof 2-.— if
properties
the studyand matmees special some define
we section.
thisIn \
svstem
theof solution
the sare ~3 = Z to
Matrices Special 3.53 cOtresponding Eigenvectors

3.7! Marricet eed Ereeeveter Profiems 7 . |

A sOlunon of this system is {l. 2, 157,

Mathematics
W2£ineerine
= 2/ 0
= |
 B = yy" [((A7!
nee ayy
<
0 (ATy!
= =A B™!

B=
“1 A>, ta -2 0 —2?|=}
matrix
;

=
(orthogona
unitar
l)
Let

a ofy

A
inversThe
e 3.

is
A~'.

We
aR:

Then
>

AW

at
P| K _\An 0 -2 ant

have
AX,= AX gives =Ax Ax

. 2Rare!
unitary
ae“| : : | 0 ppp’=|0 Thea:
1 0 0 > eigenvec
andtors eigenva
the lues
of njugates

(orthogonal).
: co the are A of eigenvectors
and igi
0 0 Oojf1 1 2 O||-2
0 1 : ; Eigenvalues
2.
BB. =aA+ BB’
+ aA’
= BB)’ (a@A+
= BB)’ (@A+
B,
a,
since
scalarsreal any for Hermitian
alsois GB+ @A then matrices, Hermitian
areB A-and
If 1.
following
the :note We
“ 0 tranformation
unitary
a called
is matrix unitary
a is transform
of ation
matrix
the which
in transform
linear
ation
A
‘ ii 3,1 B=| matrix
the example
for Consider,
; 1 i-2 matrix. orthogona
as lsame
is matrix unitary
then real,is A If A”= (A)!or (A)’,
= A”!if unitary
is A matrix
A. propertysatisfies complex
A matrices Unitary
Tk
” KS Up ke U, € (i
U,or(ie as
either 2!
7 i
be can #k
grouped #0,b,
1a:

B ix
mamxthe Then,
U,). € k U, je pa of matrix
orthogonal transformation.
subsets U, U, and
and U,U> such that U = U, U U;, where the su
ff of ix
the non-zer
transformation is an orthogonal matrix is called an the which
in transformati
linear
A on
eso off ff diagona
cos@|
ciago l elements
sin@
itbe the order of the matrix B and b, #0. Denote the
eA
set U == (1,2, 3,.-...7} _ Let there exist disjoint
rows and correspon
, . wire —-sin@ cos@ ‘hee
7 ding
ing interchanges
inter of co jumns are possible
i to bring
i B te the required form. Let
an elements non-zero
the of locations
the finds
It form. pale, is examplesimple
A A’.=
al cistoie-s
ine ie caelhan
7 testing
of way simple
a 1s procedure followin
column
interchanges The
correspo
by gedndingfollow P
ni = sat
Se that suchB in
A real matrix A is orthogonal if AW Orthogonal matrices
diagonal ; matrices.Specialmore fewa
where A j|
skew-Hen
and nitian Hermitian definenow We matrices.
A,, symmetri
skew
c,- symmetric.
definedhave we 3.2.2, sectIn
ion
and Az» are diagonal matrices. The similarity transformation performs row
interchanges
An Az, P=; LL
ae
= X, x} X= Xj"
= a= opr’
an if A, jf vectof
ors, systemj uNit
N aary form
, s vectof
ors set this
that suchP Then,0”.in vect
7 orsbe x, ..., X, X), Let
propthe
erty satis
to fy saidis B matrthe
ix
matr
Then, matrixix. spar perm
a exis
ion e pro
# 4
a sebe B utat
Let ts ther
ife faee
matrix = per)
= . * isy
: ti lLI= xg xi :
oO =
Moxy
0. are elements
otheall
r and jumn Perm - =
0 malr utation
row each
in | one exachas
tlyit if matrix this INEn, * if
perm
a utation
callisedP matr
A ix xtrix vectof
ors, system orthen forms
an ormal vectof
thenth ayallt Th IR".in vect
a orsbe X,, ..., Xo, x,, Let ors set
system. vectof
ors and
unitary
dominant diagonally stri
a ctly callsied ae? satisfied | system Orthonormal
for satis
is fied stric
inequali ty ttheIf i. one least at for satis isfied stric
inequali ty t ihe eenThe C3.in set orthanonororm
are conditioabov
ns the
e if system, dominant | (l+i)/J2 mal
diag
a onallycallisedb, = Ax equa
f tions — ‘
ae : i) 0 0 ll4i O} 0)
w/5 5 4;/| and C* in set orthan
ogonalform
| 0 3i},] |, 4i| (ii)
n al? piogon
if dominant, diagbeonal
to ly said1S (@,;)= A
ote matrAix matrix gominant
BE ‘
3.73 Problems Eivenval
and ue ie i
Mathematics"gineering
Matrices “Enpipea372
rtnn
_/e
.
ih:
 Ay x7 =~ -(x’
= A’d)
rT 7 AT=
3°-x’
= A'3 (¥°Ax)
vectors.
of system orthonormal
an form matrix orthogonal oro e ay
vectors) of an (a) From Theorem 3.14, we conclude that the column vectors (and also the «ow Now. —A.
= A’ is that matrix, skew-Hermiti
a be A Let
an (u
25 Remark
real.
system. unitary
an form alsoA of vectors
rOoW is A that conclude
we (3.49),
Eq.
the that conclude
we A. of From real.
is AX x Therefore.
scalar.
isaT A’
rowsthe of conjugate
the are A’ of columns and.the matrix unitary j 7 = x’ since
aN
also
ts matmx unitary
an of inverse
the Since s¥stem. unitary 4, —57 A*
ix"
an form
A of vectors column
the Hence, alee
s\7 ARayy
x'= (TAS)
, eeee
#=7. Now.AT. A=is that
aaa ‘= 1, “eH Therefore, matrix. Hermanitia
ben A Let (i)
zx; O, =
Ax.x’ scalar
the by £ove
isrmme
2 Adof ‘t behavi
theor
positive.
and real alwa isysx x7 T
tx ale. Fi x! ; = 2
denothe
minaSO,
torAl scalars.
are
4 x x"
x and Ax ¥’x7 thate N
I i. pie = X,,] reo Xs [X), = = AA
i A (3.49)
X vo= “x Xs%, x, = r
ae re.
aT 2 pena tH
ae ii 85 X, X, Ry, x}
x by sides both
Then». x...
1,. vectors columnwithmn. order
of matrix unitary
an be A Let Proof Pre-mu
Ay
lti in
= plyAx
be x and cigenv
an alu
be eA Le
vectors.
of system |, Pro
unitan
an form matms unitary
an of vectors)
row also (and vectors column
The 3.14. Theorem
Magn
ofitud
aree imawix
i i unitan
ary (iii)
or zero are Matrix
skew
a -Her
(iimi

|
) tian

or
are
are
real. are riX mat

and

real
Herman
itia(1)

pairs
n

or
of

matnx

of are
(iii)(ii)
cigenv

zero

pure
Ad,XK Ir Thaleue13
s 3.

an a a
x

complex
Vectors Theo

matnx

conjugate
(i)

are
at Orithof ;

magnitude
vonal ogon
Set
al© th

orthogonal

real.
: * Ther0.efor independen
; == e,@,,= = ‘ae
lin
t.'S
earmlyX

matrix
of waey

imawinary

of
a,=
=

we
Products a,

the
24 Remark innerthe

that
that find we

From
taking larly, Ivery,

3.13.
@ SUCCEs

symmetric
get we 0. *
=o
=

the
result. |. =
Hence, [A] 1.or P= 0. = x X¥’
Therefore. havewe 0. + x Since xf?a or O ||? ix, Since

Theorem
conclude
skew-symmetnc
= x,). (x, cry
0 => xy — on)
X fg — a

eigenvalues
Ss. [AP= x or Xt, + =. .%,
+ (a,x, xy
With = or
= . %
8)= 48)G Eq in x Yector
the of
(3.4 Pp
x xl Ay= tWAx)
(A = "ytAxD
A (x heuct rod innerthe Taking SCalars.
are Q@ s+. a+ get We

e thSIder 4
tei
: mp = [=
EO
Using Eqs. (3.50) and (3.51). we can write
» |
(3.351) RST SAC or i thar Vectors,
Of S¢l
€pendent.
in - | an De Xn,
Orthogona Cquation
. neay TP :
ing linearly
is ¥€ctors
of set £0nal a Proof
= r f Ortho“Wn
=AX A’ or = (437) x)
(A x) get we
4
= =F &ee erie sz =r\l
th ch a4 results 3.12 orem Therore
=, sigh CXist
rf ¥ a> A Property

—
145

i
=)
important

Ved
Die aear = me Bh matrix SACE,~ “Omeish
or =Ax Ax
establish
AN = Ax
now We
i u = 3 0). = tf+) a4 Subse
iil
uy...= 2 2 He U,. Ls
3 “elumns c" = 3} we Tre
EG 8 Sy
! yt, ae FO Me 0 se, Pie's
A! = (A)! Now, from Giiy Let A be an unitary matrix. that is
) of Be @in#Q . £} Je>
SIMilari,
Cc Mlerchan
an ¥ 3 2. l. ‘
conclude
eg?
that A is zero OF pure imaginary.
te ea $l! . o ab,§Bowethe that
; |
an Note
by followed
'OWed. 3 and2 POWs
©
we Eq.
(3.49), From pure
imaginar ZeroIS AX \ Therefore. = fs 0. gt ad, Yo Noy
i 7 1 a
y.or scalar
a is ~'A'’xae
: 4 proper,
aS :
:
Changin sae
Bh Hence ces TESto ‘este er equivalent
Sjx atid
Se Propicns Eigenvalue
aid Matrices Matric. format
Iran
ions
dy are Ass
and Ay, Where
iy Methes
. ye
erry
in;g Wie
LIr;NC say
3,
 2 2faxyf,
+ ~ HE,
+ 4)
+ (xp
+ |x=
= 4%) a,,)/2.
+ (a,= },, and
|xp/?
matrix symme
a istric
B whe
2 .+ eyez
= (1 Xt +a
7
=|yPra ‘i x'"BxO=
as wtitte
be n can (3.55)
ty 2 |-j is Eq. Hence,
a,,.+ a, = b, +
(by=) B matrixThe b, Furthe
6,.
r,= hy,
=|x,,x
Ax ,xX’ O= ay)+/2. (ay= by set since sym
x, L+i]] I Now, multiplicmat
(3.55) of
atirix
on. defini
the
tion usin
becomeform
xA=x
s quadraThe
tic
. cy A=
l+7 |
matrix Hermiti
the an consider exampl
Fore, Fee +.
‘i, Vs a,,>+) (d5,
real. Ft, i. + +; Xv;
alwa
is ys and form Herm
a itiancalle >)+ (a5,
= VV Was+
is dit ThenA., matrix Herman
itiafor xy x »
defined usual
is ly form quadrati n a,,)
ue) (a,,T, o-
this c HoweC".
ver,
in + As vy)
vector arbit és,+ (ay+
anrary
is ...,xx5, > yx a) =
,) (x,,= x where ie,

f=) fel
write can we Exp2.
(3.56) Ax x =
andising, term eacin
h
X;X;
ayy F: > = power total the whiin
ch
n n
(3.54) fol
ayaye}
a LL Oe
as defined
is form quadratic
the then matrix, complex
a is A If
homoge
an is
neous Jorm fonnthe of
4/| 3 —2 quadra
real
tic
A IR” in expres
vector arbitr
an arybe x,)"von, Kay (Xb,= x
3}. —§ 1 B= Let
Therefore,
—2 | J Forms Quadra3.6
Orthogonal.mutual"ar tic
ly
y e andx vector
the
s Theref
0. =ore. y’y
ayy= 63; 9, ay= . obtain
we HM,
J = = — = =
Ox 3, = a3,)
+ (@23> = by= by, c. A Since
=
9. = p)y"
- x(A
2, ay)+ (ay,> = by,= by, 1, = a3,;)
+ (a\2> = (3.53) obtain
we
_ | hy,= by 1, = ay = by ,
(3.53)and
seal sot. _ (3.52)Eqs.
Subtra
Therefore,
4. = ay, -5,=
py=" yAx
ay 6, = dy,+ dx, 4, - = ay,+ aj, 2, = ay,+ i
ay2 1, = ay) (it) "
uya? ya or yet we x, by
| 4 aia B= Therefore, ae,
(3 (Hy= )” (A’y)! Post-mu
3/2 2 or HY.= Aly have also We
Ay’x. y"Ax=
1. = ay = 6 and > = @3,)+ (a).> = have We
~ a 6, b= My=ay=2, wee
3 l e; get we y”, by
the to ltiplying
©Oresponding Correspo
ecto
ndringeigeny
Therefore,
1. = fia5 and =4J4 tly,+ ewenve
thectbe the be y
ayj> 2. = Ty) orx andA
(i) Let A. of
NValueeip
s,e eigenvaluedis s tinct
Solution samethee hay they (Wo
equation,
2.4 4 x3 5 — Gary
+ 4xyxy
- 2eyxy
+ x) = O (iil) All_ a7 characte“a
rime
stthe
ic
5 3 se 2 ae
=) AN-T) [A T= have
x4.+ x5 3x +
Aq?)
— (A‘)?
2x4O= (i) IA~Aj}=
form quadrati
the cfor B matrix symmetri
the c Obtain 3.42 Fisting, Zonal, ;
Example pay ortho
An.= Dy and ay,/2
+
mutually
are
A‘ and4 to have We Solutio
(ayy= by= by ay= by eigenval “rresponding
same
ue the have A’ eenyvectons
and A Matnce
thes at
havewe 2, = n for i — 7 thisieinn
exampl
Fore, tha W =
Lem 3 Sho ”
:
cigenval
aD %
3.4]
e . Examp!
3.77 Problems Civenyve independent
Eigenv
andalue Matrices line»arly has,) r
ordeOf “MX
diagontizatie
syinA (b)
cw
Mathemeyj
SEEKING
 3.7
8 e;
meesr]Engine
NEiuetic
Mathem
Matrices and Eieenvalue Problems 3.79
8 ,
Xsoe Now V)¥+> and real18 Fs7,7 ¥,1,— > y.
Vy) ATT nari isi Va
afiIma : Fore aT
exampl

_2
ly = =
un
= PEt ig.
if
vy ifs, py == ig,,F
=3{x,+4,,| + >0 2 all for 0. # x
ax3 = EsF,x,+
22s A ak oe
= ,+ .
gga) :M2
Eigenvalues of A are 2 and 5 which are both positive.
and XX—
(il)
write also can We
22 =
P\q -)
prg,
(1p. 3 | =
=
Leading minors |3| |= 10 are both positive.
[3 ¥547,+) i) a +.T>A) ix, ~
(iil)
JV
1X5) = ,p,
2[(p
to

+)
Gyqs
a ,)]ef (j? =
Poay Re[(l= x5]. )X +
erefore. fe
the matrix A is positive definite (it is not necessary to show all the three parts).
IxP O=
tive
F19+ 5] [(1 2Re+ > [vo+ l Hence,
matrices definite Posi .
A = (a,,)(a..)bebe a a square sais © i
Let . .
matrix. Then, the matrix A is said
to be Positive definite if (b) = 2 =
¥2] (x1. 2i x, %]4||
=[5.
=F —_
2ix,; 4x4+
=
Ax X" = QO r
vectorany for0 > and0 + x Ax x° only andif 0, =
= 3x, X; = 2iX,X + 21x, X>5 + 4x. .
= x if
is A If
then real, canx be
5
taken real.as
Taking x = p, + iq; and x, = p+ ig2 and simplifying, we get
matrices definite Positiv
- PW)
O= 3 (pi + ai) + 4 (pat G2)+4 Wid
wing have
properties. follothe
alues
eigenvThe ve
a of
definite Positi
= pi + gi + Ips + 2g3 + Up2- GY +2. > 0.
t+42¥
matrix e. real all are
and
positiv
This is
proved easily when reala is A
matrix. Eq.
(3.49), From
Therefore, the given matrix is positive definite.
havewe
= A /(x"x).
(x"Ax)
positive real
Note that A is Hermitian, strictly diagonally dominant (3 > |- 2, 4> |2i|) with
26(a).)
diagonal entries. Therefore, A is positive definite (see Remark
we O. >
obtain x’Ax and x
™>0O Since
If0. > A
is A an,
Hermeti Ax XK then and realISi 2.
13).3.e,
Therefor (see realis
Theorem form theif n
Hermitia eigenval then0, > O
the ues are
and. real
positive
(c) O= X'Ax= = [*.*3, x;] Xy
0 ] 0 xy
[X,,%> et |
the All 2.
leading minors are A of
positive.
4
0 4 Xy —iX, + 3x,
Rem 26 —j
is A Ifn (a)
Hermitia
+ 3X5;
ly and
dominant diagonalstrictly
= x, + iX,X3 + X2¥q — xX
with
positive real
elements the on
diagonal, then A
IS
definite. positive ,
\x,/° * \x3I° 7 3)x;1° + i(XyX3 — 455)

Taking
(b) Ax x’ If then0, 2 the
matrix itive
is A
semi-poscalled definite.
= p, + ig, and simplifying, we obtain
= py + iy, Xz = P2 t 14a %3
A (c)
matrix negativecalled
is A definite is A) (— if
definite. positive the ues

x,
eigenval All 2 of
(p? + 2) + (ph + g3) + 30p3 + 93) — 2P 43 — P34)
negative definite matrix and. 1cal are
negative

———
QO
(b)An/|"
+ (Ps + a + (Ps + 93) + 2CP3 + 95).> 9
Ex
(p, — 43)
Examine3.43 which the of
following matrices are
definite. positive
‘ be cA
¢
. It can be verified that the eigenvalues of

—2i

10
A=]

j
Therefore, the m atrix A is positive definite
(a)

(c)
s
F

"|
2i0 3|

O}.
are 1, 2, 2 which are all positive.
A
Solution

0
1
,
2
4

4
. r Te
that the matrix A’A has real and positiv

-i
Let A be a real square matrix. Show

03
Example 3.44

Ife,
eigenvalues.
luess 0of A’A
. value
tric. Therefore, the eigen
(a) (i) O= x"Ax = [x x] Since (aT A) = A'A, the matrix A Ta is symme

3
7
= 3x; + 3xjx, + 4x4 S
T Ty
Solution
2
4
=~
xy
yy, where Ax = ¥.
Now, real.all are
x ATAX = (A x)(A x) =
 by
Y3 ay + Vi a, O=

is,
oe 2 The

a,0,>Where
Vi +a, 5G Of umn

form
* 41 G4)—

is
0, =
given

a,
say---
is,
We
form.

may are
te mixedthe all if form

that
canonical
in be to sai!is Ax x’ = Q form quadratic
A

also
5that
form
squaresvf such rms
XyXyas

a/#/.
¥,X3,
etms with negative coe! = ets Vee

therefore,
canonical

aabsent,
a positive
with termsthe nana
0 sum the in coeffiecients
{tk Form Quadratic
a of Form Canonical
That is. Eq. (:
3.6.1
thes

canonical
3.60) Arrange £ the
t terms in
in Eq. ecieiite
Eq. ((3.57) suc $7) is
:s arranged as
. thenand first appear fficients
ane be (3.57)Eq,in
positivs aie ‘mca
i aes ' A,. get we =-= rank
If transformation.
linear non-singular
a under form quadratic
a of invariance
the provesThis
form. quadratic
alive.
a
:
,~ k coefficients be neg
An Vir + ... ys + A 2 +
alsois By y" and symmetric
is B Hence,
P’AP=B.
= p'a’p
= AP)’ B’=(P’
0
Now, P’AP.
= B where
Va ‘ 0 0 =
(Py)’A(Py)
y’
APy
=y’
By
P’
=x7Ax
ia | h, | Val 4 [v), V3 =

Gen se t Ae Ye yi ra2 v3 =A, Y2) " Phen,¥,. cosy Bae My LO Ky seu Koy Hie
the from form
variables the
quadratic which
transforms linear
transformation, a be
non-singular
Py a Let xy --- X%, (X,,= xX and symmetric
is A where form, quadratic
bea x’AxLet Proof
¥| X>
‘

0 2,
a

a
form.
3.15

form

a
linear
Under
quadratic

x7Ax,

remains
of
We
the
ae: :
Signature
integer.
negative
i

now,
Tt = ;

form
¥
acan
be
Theorem

prove
quadratic

under
linear
non-singular

quadratic
Under the orthogonal transformation X
r.= 2k = —#)(r - k
5d y

invariance
i

: : =Py, weo
transformation,

non-singular
Ifr=n, then P’AP = D, where D
=n r
diag (4; dad
2 Any: oes A, * A ig =
eigenvalues)
negative
of (Number
— eigenvalues)
positive
of (Number
= Signature

transformations.
define
We form quadratic
a of Signature
are
p-' A P= _ PAP
pap = lo | _ wher
k. by denoted
D=
is and form
er ow diag
dieg e(4),
quadratic
Cy ta42 +
the of
Ad:“r
indexthe called
is eigenvalue
positive
s of number
The form quadratic
a of Index
the no
of A. Therefore matrix modal seas

is, ’ . eigenvalues. of
non-zero the
number
t the ran k of thea quadratic ? form 3 , se
; thay form, quadratic
the of rank the called
QO = x/Ax be r. Now. A is 2 Symme
is A ofr rank The form quadratic
the of Rank
.
tric matrix. Let P be x, D = diag (Aye Azy seve Ae)
—_ 1 fm -
form. quadratic
the of nature
the definesThis
yadratic fo! !
“ = x™Ax to the sum of squares form Q = y D y where D is the
diagon theorem 3.16
negative.
are some and positive
are A of eigenvalues
some
if indefinite
orthogonal transformation x = P y, where P ts an orthogonal matrix, seme a
;
of
square The
number form.
5
of
squares
a sum
the gince
matrix non-positive.
and real are A of eigenvalues
the all if definite negative
Semi
r. rank the to is
equal terms
Hence, is it
diagonalizable. is A -~
symmetric, ? gemari non-nevative,
and real are A of eigenvalues
the all if definite positive
—
be can Axx =Pe O form every
quadratic
he Fe 7 negative,
and real are A. of eigenval
the all
uesif definite negative
For ex
positive,
and real areA of eigenvalues
the all if definite POSi
five
ample. =
are canonical forms. OF= 3x7 — 4x3 6x2 + 5x3, _ fank (A) Adratic
qu The
et |i, w = jad
real numbers. any 2,, AFC Gia, «+++ @ here Ax, where A is a symmetric matrix. quadratic form Q=x' form Q is said to be
form quadr
a atic
of ture
inde
iis signa andxX )
Tank
>|
TAY y QO= 4, ¥, 7 a2, 7. yank i
Nature
(3.58)
Values "igen Positive.
and real are A™A of eigenvalues
the all
the all
yee"
henceand
. ;
definite positive
is A7A matrix
he |
Therefore Posi
© tive.
£ . a's ‘ rt 1 :
t =O.
y ar A'A
of
y vectorany
any orfor0 > y'y Since
(3.57) tay Vi tats. Oa
Mathematic
3.8) ‘Matrices and Eigenvalue Problems sRM¢ eEn
ergi
inne
gey3.
mF;;80
 3 ren
Matrices and Eigenvalue Problems
‘1gnaty betw“© dratic

Lal

‘ad
fo , MC

a
re ee
i \A the ud
Quadra the
Number eg The
Number
YiVeste, Io S
is x = Py. and
'C
form POSitive
nsformation ;
real, non DF aw
inertia T
OF
Positive t
onal tra
and
7 + 393 + 393 = 392 + 313.
Ne
Bative a“ emis 1s k
Called tl
transform.”
~mBulare€duc
Red any 4nd Cx ing eS ck» ae - auonical| form = x/AX- = y’PTAPy = y’Dy = (0)
~ 2.
ations
signature
ofk Calleg
os =.
mo8gi We quadrOF Atic fo Teal*
quadratic form Q ™ in
ix
Condition Number of a Matr
A}),
reducing
the norm of amatnx denoted by ||
Let A bea real or a complex matrix. Then.
ac
J&ati
Let
4ra
Sing:
quad
i ma tix
formi
norm of 3
a nitocal
nical Cano
"defined as follows:
rmfo
We wr
*y. diais
1
ijAll =
|* Ds
m,
te themL) (i) Euclidean nor |
Define |
Compute A’A=(A) A.
} norm or Hilbert norm:
fii) Spectre alue in magnitude) of A A.
—-
4. = spectral radius (largest eigenv
eaten
rm
3:= ee te x4, 0 0 r\X,o
| ry=p
= X, (3.6)
tit Al] = VA.
L ,
lerms

quares form be gq
bys

matrix. then
with

rix (A = A) or A is a real symmetnec

+2
Yi

if A is an Hermitian mat
aned
compar
- | + by?
ave
= (spectral radius of A)”.
|

:
cv3.+
Substitute
+ cy?
the
of A A= spec tral radius of A>
(3.61) in 3
“hy Simpy
termsthe
lve for a, plify 4 = spectral radius
| in
€ th nt
quadraticgiven fj
Macratic Py q, br ande
snectral radius of A.
Orth2.ogona
se ilSatio . , form. So
Therefore, ||A|| = JA =
a the uesFind
eigenval
the mati
transfor on-
Under vectors
eigenand A. of
normaltheizedtainObta;a form
.
ions, we use the spectral norm
aticPy,= x
quadrthe
eigenvalues
: mr
of A.).¥ | as A > T «ceo £
ay,
V - depend} Ay + Asvs +
For most engineering applicat
duces
G€pending on the rank of A, where A's are the
to
aii
Condition number of a matrix
Condition number of a matrix 1s an important concept in the si Aestapoee = = se
Exe
| ampl I
Reduce |3.45
Gneniiee avhods
¢
th thetic
quadra
a aes
2x5+ +2x7 formic
to solve a
equations. In engineering applications. we often require
2x4+ — (Xy 2x =
2X2%, canonic —
to al2xyx, form
throu angonal
ortho
transformation.
equations. Since the system is large, we solve it by Heralive Dist 0 Condition number
Find indexthe si and
gnature. 7 ™
produce round off errors. The round off errors should not magnify during wick of the sensitivity of
Solut matrixThe c
the of
quadrati form is
of the coefficient matrix A of the system of equations Ax = b,, gives a MEAS =
x mf J]
the system to round off errors.
Ama} 2 —J] | r of a matrix A as
the condit‘Tiion numbe
Using the spectral norm, we define
= 9
| a [2 (3.62)
Eigenvalues ofA are 0, 3, 3. cond(A) = K(A) = HAI ATE = Vi
1]- | _
Eigenvector corresponding to A=0,isv, = [1 1
: ; Tr
.
the © aa where A = largest eigenvalue in magnitude of A a
ue A = 3, w e have the equation for a finding
Corresponding to the repeated eigenval
as x, + x2 +x; = 0. One eigenvec
tor can be taken a s v= [1 0-1)’, Since the model marie and = = smallest eigenvalue in magnitud
e of A A.
matrix. then ( (3.62) ) simplsimpifies to
a real and symmetric.- matrix, then
e above equation and also be wit ai If A is an Hermitian matrix or A is
and v>, we obtain ¥3 ~ es fo (3.63)
cond(A) = K¢(A) a=‘i
= AK 1A
matrix is given by
3
;
The normalized modal
—_=
I
magnitude of A,
vJl3 yl2 V6
Where = A,
largest
eigenvalue in
magnitude of A.
Uy, = smallest cigenvalue m
p=|w3 0-26 and
ys -vv2 6
— “WE Meering Mf j aa
bi larger the valine
e MOM erry ty
Oy CXAMple. wh ry Of the Condit
Matrices and Eigenvalue Problems 3.85
Places Arithmer, “Re cong 'On number
arithmetic Metic may diffe Illon number ‘Sin IS the Sensitivit
Sale, ix. We have
" Completety ¢ 78% mane
the solutsohas ) oFthe sy t omplex (Hermitian) matr
We can use oth YF TOM the Solution _ o¥Stem
ion obtained
|
Using 2Found off
er norms | . om { 2. Sen 2 BSrarp_| 29 oe
Obtained b Y. four
Example 3 das = 9 “In
S also tg compute th ¥ Say, six decima
J ¢ c *
the Condition umber ion: number. is given by
3 Of the fol - haracteristic equation of A’A
matrices, The ©
(1) : | if =—2 @ nae 29-A 12+16i
> (li) |» 63 |=«9 - a7" 400-0
; 29-a
0 5 (iii) = 344;
JA A-AIl= |15 16;
Solution z oe og given by A= 49, 9.
The eigenvalues are
= A, = 49.
(1) A is real, We have in magnitude of A A
Largest eigenvalue A = Ht; = 9.
in magnitude of A
Smallest eigenvalue
(A)= 9 =4
i, - [%
2 3 |
: e,
Ik Ficmesactics
cond
* 2313 2] fio 29}
Henc
Facteristic equation
of A’A is given rmula
by
\
i 1
0
e of AL 7
Largest eigenvalue in magnitud
[ATA = aa = lO—A 10 > magnitude of A 3
il Smallest eigenvalue in
é
ca
The eigen : , | tion
‘<Biak in are given by 2 = [15 + 575}, [15 — 5V5], Singular Value Decomposi |
Largest : eigenvalue
3.8 re .
in many areas ©
{
magnitude of A7A = A, =[I5 + 5V5]. a , ie
cept which has app lications ci
: ecomposition is an important con : ‘soonalization of asquare
—. Te ee etc. In an earlier section, We emgage am real or a complex
Smallest eigenvalue in Magnitude of A7A = M, = [15
- 55}. » i
matrix. We ‘now scust
chee : diagonalization
: of ar rectangular matrix
| matrix.
A Is a rea
matrix, We shall discuss in detail the case when = D is a
w herey ant
decom| posed as A =; PDQ,r ordet
Hence, : cond(A) = fal = HS+5v8 = 2.618. 1 x mnreal matrix A can be
matrices OF ©
Hy = lized iadiag te
general
oo eather m x n and P and Q are orthogonal
sti + defraine rnatrix
and n x n respectively.
(11) A is real and symmetric. The characteristic equation ofA is given by
a symmetric and positive semi-detin
Proof Consi“ he ”2Xn matrix B= A’ A. Now, B is
der the
5-A_ -2 0: since ; (3.64)
4 te ai
IA-AN=| -2 6-A 2 | =A?- 180? + 990- 162 =0. pr=(A’a)’ = A'A=B 2 ;
2 “fee
0 x™Bx = x’ A’ AX = (Ax)’ (Ax) = AXI @ : : itive or zero
zero.
mx anita vector, Therefore; the cigenvals oF 82"VT ae we
and
The eigenvalues are given by A = 3, 6, 9. seam are ZETO, Wwe order thek eigenv alues
(B) = 7. then We nee
such that
wa, Denoteathe eigen values of if Bas
some of the
4, =H i =: oon a
Haalues
eigenv ‘cran
nitude of A ac oi ae
Largest eigenvalue in mag and then the zero cigenvalucs.
n
itude of A 3 ee taken first dere wena Hy») Define
Hence, conten” Smallest eigenvalue in magn Se M Te
€ non-z ero eigenvalu | , r o eigenvalues aS Hy +1> fT
diagonal and zeros
the r positive eigenvalu
es a5 Pty
e erSm * 1,aea ply, Hay +> muHr on its
witith m4
ix a D of= ord
the generalized diagonal matr
elsewhere.
 25878
0 = pp, 3.8643
= 3,4, 0.0669
= 42 66,= 2, 14.933
= 03,
yr,= a, obtain
We
0. = +1 154- A> by givenare B of Eigenvalues
matrices. Orthog
areona
Q l andP
3m) sinc
137 5 3 3ij1 2 PDQ, A= or PDQ=Q, P(P’AQ’)
r 7
obtain
(i) We haveB= ATA =|
we Q, by
a al 373
multiply
postingandP by
Solution D. = Pre-m
P’AQ’
is that D, matrix : diagon. al genera
n lized:mx the es
1 I P74 Theref
j.# i forO andF; ---, 2, 1, = =;i for Mp = ¥; V; H; Au,j
— Ti ‘haan
= An =vcy (3.69),
and (3.68) From
....m
m i-
l,r+2,
Git) |1 1]. ) |; st
fory=
0, = r+ce, Theref
A. ore,
--+y : =rt
j +lr
.
for 0,
fl 2
t2.
=
pra 1s
matrices. following
Au,
the of decomposi
value
tionssingular
Find 3.47 Au.v) =
(3.67),
Example
From
cy,
are
CSMIENS
m are elements whose matnx

respectively.
defines an m x n
xnn vi Vn
andm =< m orders
of matrices unitary
areQ andP and
n x m order
of matrix diagonal
#
,
_-
P7aqQ’™

generalized
a is D where PDQ,
= A as decomposed
be canA matix complex
nmx m arbitrary
An (c.)
= Au,]ana Ally, [Au,, Fa
¥2 =| al
uJ7 ..., !
uw, Afu,,|"? os = PPAQ’
Therefore,

result: following
the havewe matrix, complex
a of case the In r Tanr_|ve ow,
30
vi
Remark s Vv; =
is given by A = PDP’. P’ = Q, or P = Q’. In this case, singular value decomposition
columns
its formYop so V2 *¥te matrix orthogo P
an isnal that Note Vn]. +--+) "2 1s
ore, There identical. are Ve oa Wie Vi> and | niini U5, U); eigenvectors The A. of eigenvalues of choiceThis¥~ -..5 Var Vie sik»
P matrix
¥ Vesey
m x m the Define arbitrary
= . diei
vectors
© Weeks
the with system normal
3 oe th E ae
if A is an xn real symmetric matrix, we have B= A’
Select a
A = A’. The eigenvalues of B are a
29 Remark we m. < r Since System. pacersHai:
Sey tha SeahVy S55
above. as described | oli 0. — ¥ ¥, an fo - V> v;, Therefore,

W,j pe, Mi; f = 97 fort=;

1,= = sf SY Sibi
V,.\, V-+2 determine
determine v; by solving (3.71). If
rank (B) = r. we
thats aa
By -
NOP a
nye —!
= (Au)
OOD Hig
also uw; are 4 :
fore,
(Au,)' | i
s of AA’ |
r
eigenvalue
se eye
and v; are the°:normalised
a eigenvecto| rs
or
! r
of AA’.
Now,
r A(A’ v) = w(Au).
: Hi= *r (AA’)v; = OF
f3.71 )

|
AT Au, = Hj Uy OF Alu,
rs 2, l, i=
v) = Hi Ue
I no
= Fi¥,
But from (3.66)ye,
we & et
; Wie ca n determine (3.67
ae sie
v; in an alternate
that SUCH¥, «++. V2v Viv
way.
vectors
the
From
Tomes | Define
ha (3.68), we 28 Remark
Leese
PY ire? Fe ees i
a5 Au, that implies
This
where D is sui
i _ for 0, = ll Aw,|}? and
eww, tably defined.
A we

Tm Si 0, > ||? Au,

II Therefore,
| or
HG (a; Ww)
= pe a
3 " 23
A’ u) = (Au) (Au,)"
=
g7;,7 Au, =
O33 D..5
we.
P32 ea foll the have we ro aT
For example,
owing its os u’ fay

matri
decompositions: Now, |]Au,|/?
Orthog
* 2 with af, af...
decomposition
fa matrix is not
unique.
o
uige eigen (3.65 theonalform Now,
~Hesy,
ee Singular values are Pe SS _® pene yaa
_ v,, is arbitrary. singula Eee eR.
u,. Hy A u; 7 A ‘ and
The deco
‘
h sitive square Toots O° Ft _ Since,
==
ee
.
r value
f Lt}. jo. = |
r = igen
the non -Zero z a’ Bu. =
ul Tene Us, Uy,
The TOMMAlised a are SOlutions ar
i
= PDQ, is called a singular aH we are called the sing of
| ae uy,
oe u,, a5 is That System.
decomposition. lar values of
them
A.
denote rthonormal
oat
andB of
eigenvectors
alised A
and the number
fhires-an
s Hi, Hz
Eigenvalue
Afai
a rices: and normthe Find
Problems
Mathematics Eng
3.87
j ineering
3.86
 |] I[x2
0.57605)".
- (0.81741,
= ~2= v, 0.70472)";
— [I, = x 0.066966:
= Hz= Ay
by given
is D matrixThe ____
ce
| IIx, i
= 3 3 ea

3-A pte ae 2=
0.81741)" (0.57605, v, 1.41900)":x,
=[l, = ye
[4,.93303: A,=
=H Ay 0. = 6A — Az = 1 AS.
U HS WO.= :
ah r=ifs A>
Je 6, = Be =; ns 3 = | :

_ by givenare B of wah
Eigenvalues following.
the as obtained
are eigenvect
The ors
10 7 3 3/2J |
4 fe } | B=A‘A= haveWe 7
“t ay
; = |If= }
37 is 1] Hk haveWe v,. [ee= ¥; (AA’)
of solutions
are ¥, vectors
The

acl
=~ A 1s decompos
value The
ition singular niatrix. orthogona
an also
l is P that Note
= -0.57609]}0.81741 |
0.81737 0.57605
= v5) [y, P=
| VIO—¥ vv10

have We by given
is P matrixThe
matrix. orthanogon Q
is al that Note |-0.57609 }|-0.3
3 6060]"
2_|2987
,‘mi 8 0.817
[ 37 0.932 2/} 172] == Any “ al
‘ “s
| vio aio
| orl =}
| 3/10 wio |_|u7 0.817
|" 4)|" 0.933]272 3-86433/1 My
— DY giveisn Q matrixThe [0.57_ 605 0.362][060_[! =! au, v,=—
= _ follthe
owinasg v> v,, vectothe
rs cline
/vlol (3/1 = u;
= 0. sat 1": oe matrix. orthanogon Q
is al that es
oe ; |" ~0.36060 0.93272 uy —_
ae a ele fo . 0.93272 [935=060| Q=|"
3x10] =[I/—L=
vl0, Speeh,
e= =e ope r

follothe as
wing: are
obtained ectors by give
isn Q matn
e xHa
Now, 0.36
~ 060)" (093272,[xj a4 e
[30 9] ollals BHAA 4 _ 2 = y, 0.38
~ 6607
(1,=)":2 0.06
= 6966
A, :
!
3 3]_pe 3}frfr .. r
272)’.
0,93
* : =
| |x
0 Q) ~ D 3606 =
[0 0 x) = u, 1]’: (0.3
= 8661
xX, , 14.9 A,-
= 3303

0 10 diag?The lowi
fo]
ng:the as ined
™m .,co nal= : a obtaare eige
€ nvectors
by giveisn D x ee0- 10, gsm 0
0.25878} :
0. 100, areB af V The 1H
F sg 3864
| =33p
¢ decompositio valu 1 singular piv en 4
n
acre he ry
. ed er
Is given by A = PDP’ = Q'DQ.
by give
isn D
The eigenvaises o° A
eigenvalues The Ic
al and symmetr
matrix
A of B =A
diagonal
are Sq Uae
The
citi . OF Sige}

=)
\
decompos!t!
PDQ. A= 1s on
Mathematics Engine
88er
3.ing
Matrices and Eicesvalue Proice
90) 4.
: fINCEri
TS ng
fa|them etic,
Matrices and Eigenvalue Probiems 3.91
woh tpl Exercise 3.4
D=]0 9 iven in Problems
Verify the Cayley-Hamilton theorem for the matrix A. Find Av" if it exists, where A is as g
O J to 6.
1 —2 |!
Thse
re a
cleor
10 -4
cig
CNVe med
obare taWe : as
theing
3 2
follow
1. | O95 4 2.10 2 0 3:\2
A Oy
“SUL } ay TP -4 4 3 1 2? 3 3 | =I
= “ ayiafia =
x, i
12 —2 E = 2 lii
; | m= AL= up i]% (ho 82> O:
ery.ibs] 4. /1 1 If. 5 1 3 =11. 6 |i 1 i
13 =-l -—2 =i 1 iil
2]
matrhe
ix jsn Q
give
i
by
Find all the eigenvalues and the corresponding eigenvectors of the matrices given in Problems 7 to 18.
Which of the matrices are diagonalizable?
ea ak
ft i I
14 3 ee ee I ee ae
Ott| wy =I
a. 11-1 ol. 6.) = 1 OL
Note
* bore +.
that is alQ
ogon
orthan matrix
Define the
vector g. Vv,
the as
followin a1 32 11 O -1 =? 0 1
yt La a. wd Oi i
v0;.) “2 4 St. Hi. | 1 OO i]. 12. |i o il.
i
1 is
am Au, = = i 1 =| : = :
—-] 3 4 -—j —j | ‘fF ¢€ 0
Hy l
V6 V2U1 }_ V3
qe [1
fo 100 02-2 Oo Toe a
vectors
The are y,v, v.,
arbitrar but ¥y, ¥2.1: v
Should form
an ormal
Choose

orthon system.
0010 ii =) ob 3
. f : 15. b
X45

i it o: 4
a

as 0001 i =f {ot
and

og

ead
Vo.

al
ld
i

0000 = oe ee a | Et 8
4

0 i
ia (0 .
fg gs 00001 01000
0. The solution is
Making
a: = b, c = —2b. We obtain a — b=
orthog v3, v,,
x, onal, theonsget we
equati a+
h+¢=
00010 00100
a)1.Abo
16. oe | 17.)/0 0 1 0 OJ. is: 160 01 05
cos us 01000 00001)
¥1000
010000
matrixThe Show that the matrices given in Problems 19 to 24 are diagonalizable. Find the matrix
adiagonal
is
matrix.
is P
given by
wv2 W3 wie 3 2 | ~a =e | 0 Zi
P=[v, vy)vy -vv¥2 ={UV3 wel. 19. }0 2 O}. 20. |-2 #0 4). Zi. |}2 OO 3).
W3 IL 2 3 -6 -3 3 I =3 0
~260
ee Cae| | 5 -—6 -6 |

24.
zz. |] 1 Jj. 23. |-1 4 2}.

|1
 =o
ets
7
“

aS
#5
Saat
Bee

=
+
"

ae
at

3
3
=
pid

64
a

0
A

=
is

4
5
|

Ty.

5+
67.
both

Al, A
is
that

xi
66.

24

63.1
and an

3
matrix
A

that
unitary

oe
be

showthen
Suppose
50. 49.

=
matrix

ciliate
A’
if

and
A.
using a
is

=
are e

ne
A

the
A

symmetric.
that
orthogonal
48.

ol
4
Then,matrix, Show

—
normal.

Hermitian. ae

~
the ie
the a
‘Find

+
Ifa unitary

matrices

diagonalized

232%,
a

to

are

numbers
of

mairices.
*2 ~

4x,
ig

2rir,
following
a?
3x: +
condition

e
=

+
that show Hermitian,
J,

6x; xj
matrix A

+ 3x5
65. 64,
AA*os matrix

+ 3x3
ae that an
is ‘
Prove

distinct
If

thar
47. 46.

show

.
|

;
skew-Hermitian

to
orthogonal.

64
matrixsymmetric

eigenvalues

and
the

form
—

using
=
can called eigenvectors

=
m +
=

corresponding

reducti
15

+
a

+
A
*

4x
-

\x3
normal matrix, that

Reduce

ve
63.
Find

reduction.
the

Ix:

2f xix2xpx). Problems

canonical
quadratic
aril

65 22,7.
Og ms] AY
'

orthogonal!
form 26x43
My Ms ma,
-

[AJ 3°a |/ fi 7

22r57,
matrix
tis

ahs

usimg form
reduction. Lagrange canonical
to 63 and 62 Problems
in form quadratic
the Reduce | matrix orthog
an. onal
also
, is order Samethe of matrices orthog
two
onal
of

orthogonal.
produc
thet that Show 44.
a the orthogonalconditions

| | ‘ use:wiah orthoan is P that show thenD, = AP and matrix symmetri

a IS - c 3.
x3.+2a 6i)xpx,
A- (4 - EM
x,x, 4i)
é + ne
(2 - asd (4%)
= AT if
xj be
60. only andif symmetri
is ° A°Ac that ane matrix, ila
a be hei:
symmetri
js c.
A'A that showA, matrix squareany Lor ay
4x2.+ 4ix,x,
+ 8x,x5
— 2ixjx,
+ x} 59.
2x}.+ 8x20; 4x,x5.+
-— 2e,x2
+ 3x) 3B. andif symmetri
is ABc that
BA= AB if only
show Then, order samethe of matrices
— ;symmetri
~ be B c and
an A kbLet: 40.
A ; -
Xx}.+ xj -— px, 4x + 2x\xz
= 7 |. B Al= (in) i, mteeper
any for B™ ts-sim
A”itarts Positive
(ir)', ta'B simila
js ! r A (i) that
61. 61 toto7 57 given forms show ThenB. to simila
be r A Let
Problem
m s quadrat
the icfor A . matrix=. Hermiti or
the an symmet
thericFind 39.
4 a-h | cOs0 @ sin|
—
2h = 2?@ tan and
5 -3 -6 —sin@ |"
=p form
form Cc the of isi: P
: definite noteis
Ht. positiv a 4] 0 |—-2 matrix
the that Show56.
| —2 -—3 : where4p -»Pp - D matrix
matee. 4tonal
- 4diavon
ta al
ne transf
15 ormed
FHL ~-4 el hip= A TTVQITEsS
ima He
the that -
Show 38.
:
|
4 y =f - atk alee BRN
5 vac that shaw thenA. = A- and migirix daguna
f lizabl
as an e Ais16 37.
a defimte, positiv Air
1 itt is
7 e—Il] 4 |-3 matmx
the
* that Show55. zere are Matrix nilpote
a of nt
3 21 Ke ada tat SRVy = AS &F inlever PUSsitiv
’ sume
e tor= i
bE W —
cecal ek rreleesgy
ees me! Ais
y “Rt AMON
GF otasTT
a hat errenylsame An
’ 36.
aiesthe Wave BA andB A’ | a

B! (2 0 j
Matrice
the s thar Show
6 — ac and minor) ‘ re er 5 F Lh _
minor). leading 35,
2 x 25 . Tey eid
- erat AR Ts craenvecy diferen
Aut t Gives SHECNV
thar show Then, order “ame the af
leading
|x : matrices S4Ure SBE
a>0
a ifif bY
only (1
only and
ifj definiteive 1S
definite itiveposit
is 6 oe
a| i matrix
- symmetr
x 2ic real «. hea B andA Let 34.
2—_ a that Prove54.
a ay Bes teote OOS RSRepenvitues
SNe the
Pi MEMES have PAP
: andA matrrees
{Toe
the< th [Nat 33.
i Show33.
. se tis Ein| 19. eigenva lues
have
A
—4)
l
Marin
aoe~S g3.2 Let Let £. 32..
0 l
]
yee sy li f
i ott a e
-
‘lig il
E em il oie
i 4 = es epee | ryt eh RE ;
PTA a" 4, . R Matriy
i fl F iy 4 RaisRw a
0 l -4 1 (ii)
| i
Leta as
i 1 f | f
1 l 2 : i
; Tay a I
] 0 ‘ 4
| -4 Fy Lhe Eigenv
| i |
40.
alues
,'f fi My | I | t wae Pe
- ‘ | | i} Eigenv
=.alues
3 A’! propert
roperty
f 1 ts satisfy matrice i | | i 7 f j
1 1 s followi | i | | |. i
the. ngDo §3. Poi
L ba
ree
| ,‘
‘ H | i | + p ea || Eigenv
productinner
. ld s | SSR .
28.alues
of
the
value
the
preserv es
matrix unitary
an is A whereAx,= y ce pagpasae
, unitary
transformation
an that 3
Show52. Ae sD Eigenv
27.
alues
SR VOGT comple any
x for , i il
positive I ey 4 &
da teeth OP. We | a th |
Dien 2. | Eigenva
“6.lues:
: = ye OTS:EPe SR
4 Seats —
f) , Pa
and realis Az Z’z at that show then0, # x vectorreal every
for0 > x"Axand matrix asymme If 5],
is A tric Lf =
4 7 7| IPenwal wes:
=),
F a3 Fy 36
4 zy.4dLp | ed a

3.93 Atoee
AU atriR
may the Find
Problems Eigenval
and ue Matrices
’
ss Feathrenrertig Engineer
' ing
pp 3.973
=
 Hs a

§).€ 5
andS, € s, where
s,,+ s, = xX as written uniquely
be canV in x vectoreach vectors
of set the Therefore,
$5. of
Therefore,
55,= s, and Zpue--eZeh
Zyy Mgr Mpyeeee(Xp. basis
a to extended
be canx Similarly,
S,. Of ¥,} ¥o.---.
Yi+ Nov Xge---s(Xj.
s| = s, gel we {0},= Sy A S; Since,
S, 7 ES;S; — S2 = s} - s; Therefore,
€S, 8) — (82 basis
a to extended
be canx 5,,¢ S,)7 (3,

a
and €S, sj)— (s, is, That €S;.8> 52,€ S, and S,,€ s| ES.s, But,>. — 8} = S; — S, OF 8, Since,
S,.7 S$; of basis
a be Xp} +--+ X2y Letx={x,,
Proof
+ s, = s;+ s, Therefore,
55. + s) = x ands,+ s, = x asV inx vector arbitrary (3.72) $3), (5, dim
— dim(S5)
+ dim(S,)
= 5)+ dim(S,
an write

ae
can we that Suppose
Sp.
€ Sz 5;,
€ S, and €S,,
8; S,. « Lets,{0}.= S, a 5; let Now, thenV, space vector
a of subspaces
are 5; 5}.If 3.18 Theorem
{0}.= S> 7 5, get we arbitrary,
is
x Since
x.= 0 and
0 = x Hence,
53.in vector
a and5, in vector
a of sum theasx write
to way
one exactly
is there (i). By €S,.
x 0€S,,and asx=x+0=0+x,
ways two asxeS,,0eS,, V. of subspace
UW
in x vector
the writecan We €S,. a where +5,), +... +S) {S|= § 3,458,
thenV.404
x and 5, € x Then,
S>.M 5;€ x that such
x vector arbitrary of subspace
ire es “a S5, PS).
an exist there Let,Sp. A ES,
0 is that €S>,
0 andS, € 0 V, of subspaces
are S, and 5, Since(ii) RY in planex-yis 55 + 5) Then, in. y-axis
the be Sy
: S.+ S,= V if onlyandif 5, + s, formthein Rt.
and x-axis
the be 5; Let
5,0 S++ ThenR°.
, in be &) (a)
expressed
be canV in vector every Therefore,
S}.€ 8, S,,€ lye (0. S5= and R}
8,18, + {s,= S, + 5, definition
By (i) |e0) ((x,= $, Let QJ
| Proof followin
the g.
are examples
simpleTwo
V of subs
a pace
iS 55 + ‘| = §
Sp.® S, = V by denoted
is sum directThe {0}.= S,% S$, (ii) and S,,+ S, V= (i) Therefor
S>. e, ax, and5, € @x, sinc§,e of elemanent
if only and
if S;, and S, subspaces
its of sum direct
the calledis
V space vector
A 3.19 Theorem is
| Subspaces
of Sum Direct 3.9.2 _ Also,
¢ ype
+ (x. and 3, VIE
4 (x, since
S, of elemen
an ist
ws ED egy Rilte EE
A= | a 1
= dim (§,) + dim(4,) — dim(5, ™ 53).
| ' Now.
yan andx elements
are© —
=e
We.obtain
: a
dim($;
are
a ther Chen, y
+5) =p+qt+r=(p+qg)+(P+n-p
§, 1n $ and +X,*, = x
clemen
wot be ¥y
¥, a xs, that sucVh
+ 5. and
1 § in
;
elemen
two
tsbe_ y, x,, on
| of “aMpspace Lew
aeSo at ‘ 1s By Tr
fonvarc SIME1s proofPhe
= sara ji. Therefore, the set of vectors [X,, Xz,---:Mp, Vy, Y2r---:Ygr Z1+ Zze---»Z,) forms a basts o F
r -
ince 1
a
-s
ince :
the vectors X), X, Xq- Vy» ¥2- --» Yq are also linearly independent, it follows that pm 0, is i Fe ae a bi : S=5 Also $5].in (53 9S}KS) all of spaceis, hat (is ot

; oe ee 4 CML
| a eral gk Bete Sr
0. = +
BaYq +
+... ByY2 +
Bry, +
&X, +
+... ODX2 GX, = y +
\
SUTI
The
N
We
define
i mt Duc“P
e aiSUsDStye Se Si.
108 and
get we (3.74), From
i. forall
0= 500, Fe space veVect
Aa orhe WV Let
that it
follows
i t, iindependent, are
linearly Zj-
Zzy--.2, «Xp:Xp, X;, the
vectors 6,.
Since me
scalars Subs
OFpaceSu
s m
a
ry ny
Ph gt
3.9.1
_ iTei es ra i ad
atATe am| MO
eROnN Vai spa
spacce
e the PAPE MT
OX,
+ ... + 55X3
+ O,X, ¥2,=
Hace SUBS
+ ... + FZ,+ Viz) im EE: sttAbsrrac=efren
mt -
th pe| La
ei = rei; TdCT Of
Pep ih 1 oF ee MMULICes
Therefore.
S,- of rs I Le MTLy
D | 4a
SU4DADas elt
peal | | . 6 ;
ecm | idee sci R-
itself ina
Hence,
53.™ 5, to belongmust them
of both some
for pie SPS gin ONi ie f throuet
Re i]
beUTOE aw
Wh Ee . d l
s
itee
fir Pie scalars like é,
ilso Wi
an is
element side hand left The ¥- Bj»(4), Q;, calars ‘y Dc verVector.ZETO
an is
elemen side hand rightthe andS, of * sikhWo
i “Ty — = . 3
defined Cu
rler hive
i t
‘om .
| Vie

HZ)+ --- + B22+ (N21 =— By, +

+--+ By2 + : OpXp
By, a
t+ + OyX2 O6Xz+ phi
+ +» coSpaces€CVo
(3.74) st On
lootra.
‘4
‘10TO@ 3.
. 9
|
]
7" |
ht +
++ t+.
YZ, 2, + By, ...++ +
Poy, +
By; &X,+ +... OX
. ‘
. Z a
if
9.=
-_— -
|
. | nn
then independent,
not are they
If we spans ij i! l
~ !
a
tor s
pace $= 5, + S,. We shall show that the set of p + g + r vectors (3.73) is linearly
A
Eg
(3.73) Zod ZgoeeryZys Ngo nee Vow Vyr Kyo soy Xgo {jp 7]
1 1

3.95 Matrices and Eigenvalue Problems

le

—_—_—_—-_
a.
 Matrices and Eigenvalue Problems 4.97
3.96 Envineerine Mathematics |
ime iGill WIE aan Thus R? = S$, + S,. Now, we shall show that the subspaces 5,, 5, are independent. That is, no non-
; | ' , ! : 2 ca
xamatrices A, Let $, be the cubspace of all syininetr; the other subspaces.
vector space
For example, let| V be a Fie :
of ealmS) We ihe subspace of all skew -SYINMCiC that ics zero vector from S, or 5, can be written as a linear combination of vectors from
a, De TN Suppose that the subspaces are dependent. Then, we can write.
matrices B, = [{A, + (A,) }/2] of order mn, anu . Pa aE
@ dp.
:
of 5, and S>, that Is V=S,
i
n. Then, Vs the direct sum
C,=[lA;- (A,)")/2) of order 0 1
¢,0+2b=0.
Remark 3.31 a\-1|+b|1|=c|3|, ora=c,-a+b=3
.
When a vector spa ce Vis the direct
sum of its subspaces 5, and §,. then l 2 j
) = basis |W).
basis (4)) + basis (53 endent,
dim(S,) + dim(S.) = dim(V). a= b=c =0, We conclude that the subspaces S,, S, are indep
and
The solution of this system is
that is 5; 0 52 = {0}. Therefore, R’ = S, @ S,.
Remark 3.32 |
its subspaces S), 5p, ---+ Sp if and only if
The vector space V is called the direct sum of Example 3.49 Let V=R*, and
|
: |
1,2, peg (3.74)
[ E5;]=1 1= 0 (! 1
and an. l
(1) V=5,+5,+...+5, e
j*s
2 1 |} 7
. 5, = span :| .|p and 5S,= span 4
It is writen as V= 5S, @S,®... @ S, Further, 5, = span
3 —| I F 8
oF
9 }\8)] 3
Fr | 3]
r
basis (V)= > basis (§,). (3.77) '
dim(V)-= ¥ dim (S), and
Ls.
Show that V= 5S, ® S; @ S;,
i=]
i=l
¥ 1,-1 Hoe 5 1h 8)’,
iaiaaaiake written Solution
, 3,-i),
It is easy to verify that the vectors x = (12,
=10,
2,w] #0), and hence form a basis forR’.Every
endent if no non-zero vector from any S. can be w =(l,-7, 8, 3]', are linearly independent (det{x, y,
A collection of subspaces {S,, 5,,....5,} is indep aces S,, Sy. ---. 5j_1» Sis ---S-
as a linear combination of vectors from the other subsp vector in R* can be uniquely written as
Remark 3.34 l
: 0 1\
sum ofits subspaces S,, Sy...5, if V= 5, +5, + ...+ 5, and
the subspaces , i
A vector space V is the direct
| y ls. |=s,
+5|_ 1 HS ES,t + StSt! Sy
5E5
+5+tH
S,, 5,...5, are independent. {+} F] bit
a) 2)
|
| Example 3.48 Let V=R°, and -] 2 g 3
1 independent, IS
are
s S,. Sa,Sy 1
1 \/0
Taus, RA =S;+S, +Sy:Now, we shallshowthatthesubspace other subspaces.
3 | >. ination of vectors from the
5, = span 4|—1|,| 1 |, and 5, = span<| can be written as a linear comb N
non-zero vector from any S, Letus trytoexpressthevectorinSy in ems ofS YS
sarvosethatthesubspaces aredependent.
|
2 1
:
Show that V = S, @ S,.
— inS, andS,,Then, we canWale
toverifythatthevectors x = * : ; ,y= aoa z= [, [1,~1 1]’, y= (0,1, 2”, z= 0 1) ;
- P
iseasy ,
Solution it(det[ ce form a basis for R°. Every vector in R° can be uniq
fi uely
4, 1] are linearly
] # 0),
and hen
_|-7
x, y, z] # 0)
Te SEpa
independent
2 1
al ttl _alt alin} s | .
7Z | 7: | 14
| |
0
cok 1
4) 4 (2,8 3
Bl1||+y||3|| =s, +s, €S, +S, . |
Ocoee Ee an a} —-1}+ ; : it “. = 3d. m is inconsistent
] 2 1 ca we find thal
weuese=- 1 Sa-b+e“™ avahene .
; See | tiriggacad « pac=d=0. Hence, Sy (S;| + 5:)= (0). ys ;
‘ 18@~
pie
and the solution of the system
 value Problems 3.99
Matrices and Eigen
598 OEATONE? rie Mariernahos
| one:
pleme ntary subspaces.
im terms of weelors in 3) OS We cere Lute We may also obtain other com If it is not a basis, then
e, 154 ae
S, (8, + 8.) = {0}. The vectors ind, cannot be written the b asis B, as B = {x. y. e,}- = &- Comparing
8 4, Now, consider the extension of
e€
that the subspaces S,, §>, $; are independent. Therefore, V = 5, PS, d: B such that ox + Py
|
and y. Then, there exists sc , and hence7
combination of x syst e m of equ ations 1s inco nsistent
—~2a= 1, and a— B=0 ah. Theno of Bis
3.9.3 Complementary Subspaces elements, we get & + Bp=0, is a basis. The size
y. We conclude that B = {x, y, €2}
Hf the vector space Vis the direct sum of its subspaces 3, and S,, that is
B= 3) oe Yo. Chen 8) cmd 4 e, is nota linear combination of x and ar combinations of
X,
and e; = [(x — y)/2] + & are line
space can have more than Orne: dim(R°) = 3- We find that e, = [(x + y)/2] + e>, pace of S$ is given by
are called complementary subspaces of each other. We note that a vector V= R°. The comple mentary subs
subspaces, that 1s, a complement of a subspace
may not be umieyue y and €>. Therefore, B is a basis of
pair of complementary
where 7)€ RK.
Consider the following examples 5 = span(B — B,) = (e2) = (©, 7, 9)).
ntary
ces. (b) Any pair of straeht lines three R’. In this case, the compleme
= (x, y, €3) is a basis of V =
2
(i) In A*: (a) x-axis and y-axis are complementary = subspa
4 4
Similarly, we can show that B
a}
ongin are also complementary subspaces
e
;
of yaxie subspace of S is given by
ementary subspaces (b) The complement where WER.
(ii) In AO: (a) x-axis, y-axis and z-axis are Compl S = span(B — B,) = (es) = (0, 0, m)}.
The complement of 2-aats ts ty plane. op
is yz-plane, or the complement of vz-plane is x-axis. (¢)
and any plane Contuinine of a subspace may not be unique.
the complement of xv-plane is 2-aars. (d) Any line through the origin
subspaces.
This verifies the remark that a complement
the ongin but not contamning the above line are complementary
Orthogonalization Process
(iii) The space of all real 2 = 2 matrices can be written as a direct
sum as 3.9.4 Inner Product Spaces and Gram-Schmidt
inner product denoted by (,) on V satisfies the following
Consider a real vector space V. An
properties:
by * aleIfeA foe) le {a |
(i) (ax + By, z) = a (x, z) + BY, 2)-
a}f~(ce of be| ol] Tho~ 0) |lo |
(ii) <z, ax + By) = a (@, x) + BG. y).
thiscomplement.

4
= 5, @S, 8S, 8 Sy,
(iit) {x, y) Ss Cy, X).
Theorem
subspace

= 0.
The subspaces S5,, 5,, 5; and S, are complementary to each other. (iv) (x, x) =Ilxil? 20, x #0. (x, x) = 0, if and only if x
is B,).combination
vector
thatS, subspace

together with a specified inner product on the

Every
space
3.20

An inner product space is a real or a complex space

fromis Extend

Vhas
Proof

aof
combination
subspace

basis
basis

basis
vectors
B,,Define

obtain (8 and

space.
Let

5,.
independent,
S$,

B,).let

+ (8 linear
B,
be

8,
is spanof

be

of

(0).+ fromof
V.
5}- Suppose

to
linear V

B
a

€ a

8, thatof a
S, Therefore,that
vectors

The following are some inner product spaces.

Let
the

S,,
is $;

complement
as

is
linearly

x
y =

x
a
—
and
Since

§;,
U is Jet
generates

of direct
of

y.
standard inner product on R° is
y

S$, 0
=

= x
—
(i) Consider R", Let u = (x), X2, --» X,) and W = (, Yar ---1 Yn). The

=
and
Hence,

$5
we

5,
Since

0.
B, #

y
S,. x

4
=
=
get
V=
S,
we
V,

is
B

defined by (u, w) = Va
a
Example 3.50 Consider the vector space V = R? . Let mmfS= {(x, y, z) 1x + y+ z= 0). Determine the i=!

|f(z)gde.
complement ofS. P of two functions f and g is defined
(ii) Let the vector space be V = C [a, 5]. Then, the inner product

(f.8)=
ae It is easy to verify that the vectors x = [1, -2, 1]", y =[1, 0, -1]’ form a basis to S, that is,

product.
standard
by
ida ~ es
— the basis B, to a basis of R°. Consider the standard basis of R*, that is {@,, &2, €3).

consider

product

spaces

inner
inner
seach a “as of the basis B, as B = [x, y, e,}. If it is not a basis, then e, is a linear

only
shall

and
real

the
the
We
Reality decease af ai irsexist non-zero scalars a and B such that ox + fy = e,. Comparing
and Remark 3.35
hence e, is not a linear shi ; ia = 0, and aaa B= 0. The system of equations is inconsistent,
B is dim(R*) = 3, We find co fe and y. We conclude that B = {x, y, e,} is a basis. The size of Every orthogonal set of vectors {X,, Xp, ..-. X,} in an inner product
space 1s linearly independent. If they
and e,. Therefore, R ei + €, and e; = —y + e, are linear combinations of x, y
B is a basis. ae=. are not linearly independent, then for some a, # 0, a,X, + a2X + ... + a,x, =
0. Taking the inner product
the orthogonal property, that is, (x; x) = (0, /#/
rebar e complementary subspace of5 is given by
7
*
i
.
ia
on both sides with x,, X2, -... X, successively and using set of vectors is linearly
i. Hence, the given orthogonal
and <x,, x,) > 0, for all i, we get a; = 0 for all
-«*S _
=span(B-B,) {e;}= ((7,= where0)),0, ne R.
independent.
|
ss ae Oia a
 2

4
+" 3-4)
2+ span{(1.
= i5,
{((l.—> span
= g, “1,R= whether
Test
7. Requiring w, to be orthogonal to U,,_;> U,_2, --, U, and simplifying, we get
b)+%
0, ((a,= S, 6. ~ Yn = U,_)/llu,_jll,where y,, y>, .... ¥,—) are orthonormal vectors.
4 0): b, =
Ua, 5, 5. ¥n—2- ¥n—2)
-il (x o ~ Y¥2)¥2 Myads
— YY;
=|? (x aA-1 = u,-1
6,9: ((a,
= S, 4.
} d,:
© + 4), ((0, test
6, to 4 problems
In get we procedure,
abovethe Repeating induction.
by resultthe proveWe
R}- de .¢. 2 ie |
vectors. orthonormal
are Y3; ¥>, y,, Hence, u;/iluil.
= y;

92 5) for basis
a Find

—
—
=
—
—

——
=

X3
yix-¥1 522 0},= 7 ed 212

Hut

(Xa.
y» (C4= Sy Let

iuiu
2.

(X3,
thuyllu
%;-

YY;

YD¥>
Ms
-yt a

HustX,,U,)
:

‘X2,U,)
Sa and
Sy 7 Sy 7 5, Determine
i

-
-
>

=
5,=

uIP
u,
ut?
a3sy

U,

x;
8)

uy,
1,
Te

ath Gg
{(a,a, b c)la

——
| 2 + b +
- 2c
clab, {(a,
= 5, i

u,)
u,)

———

(X;3,
u;)

(X4s
2 == 0,5 4 4c
(X3,

———
(X3,
4¢ == 0}, a,
=O),
b,c € R- Ga.
5s Sa St subspaces
theR’,in1.
=X;-—

=—- 4.) 0, = U2) (x3,

+ Uy) f32(a,
+ Uy) (uy,f5)= U2) (U3,

us)
U>) (Uy,
u>) (X3,
5,={(a, b, la = Ph
3.5 Exercise
u,) (uy, i
ea:i
Oe
ar (x+ *) U,) a2
ar (Us, ™y
f52+ U,)
ye (u 4] =n,
0,)ye{u;,
_— Ys) Yo (¥1» set The
get we u,, and | u to orthog
be to
onalwu,
th onormal basis for R° _ Requiring
vectors. ortho
arenorma
y> y,,
l Hence,Ilu,ll.
/ U, = ¥2
il 2 %5*jia,l »:- ¥1) >- 22= ag ha 2 ay
17 2» aaa “a u) Uy) {X>, u,) (X2,
Tt
pe Oe
u,) (uy,
=,
(3 ——=
uy) (2, =- ty) 0, = u,) (x2,+ u,) (uy,fay= Uy) (Uz,
}
=——_——
(4,
17
> —6) —4,
1 @
= (2,-2.2 —Fa\9" 2")
T j get we n,, to orthbeogon
to al
u, Requiring
‘ ¥2) (X3 yy (X3, %3 = u; orthogonal.mutual
are@ly “+
¥2 (5, — such detebermin
to ed scalars
are$ fi; where
t, lu, u,/= Sas X,»+ 1 _ ju,
2 Veta
+ 0 + f,3U4
+ f2Us
+ *
2 LE MEP hu,
* y2
10 =
iB 92) (xs,9 = 9) xy, -2.2 2
(|3 U2
: Eas - T ll,u, u,/I
= Ys +X3, Bot)
+ %3)U,
= M3
7
Wi Op X,—= vp,
Be(3 0, 2 =< (6/8)(2, 2, 0) = ee uy = sine golutio®
6% th 2, OF 1S pI Fw of toe ance det (ay 2X3) # OF y,=u,/iuyll,
Mm=X,
ee
rz — oy the sive Gectore ate linearly independen example 3 Kok linearly indepe
|| i estas
; a YX to X from
T matrix matrix.
wane
translat
Theion
V. Of triangula
lorawer
enendent vectors
-»Y,}
Using
there Thy
=(3,0,2)'.x;=(2,-2,2)
peg -
the Gram-Schmidt
.¢ the 0 X, = (2, 2,0), Tyx»
space
ae
ni8 a 51
Productinner
an to basis
idt orthogonalization procedure, obtain reer a
arbitrary . ty,= Y basis orthon Exists
an ormal]
i
—
.
basis for
7 an be 5X X35... {x,,
—
am-Schmid; & Let 3.21 ,
Theorem
js it Gy, the using
SPace, by space productinnerthe to basis rmal
pence Yu my The procedure
of construction implies that the matrix oy Vn
Productinner
an In vectors) a et hac: l Orthono ‘ich orthogonalizat
independ
carly
ent ; an derive
to possible always
Sy Se trix 0 of transformation is lower
| lin of set (a
are 0 asain
vectors. The set of orthonormal vectors
y = {¥\. You om Ye
| - -
Xp — Ka Yaa — Kye VD? — 0 — Sn Ya om u,/ liu, |. =
n=

3.101 Matrices and Eigenvalue Problems

 3.1 . Matrices and Eigervalue Problems 43.193
In Problems dptqtrel :
. the c | (i) 3, ipeqeran 0,
tary subspaces of S, in R’- (ii) 2. tpequrandptgtr= See
a 5) = (Gy, Dl2x~y 4 ae 95, = {Ge y¥,Dixt+y+7z=0O). AD en
Using the G aivice z=0}. identical:
'y (iii) 2, if exactly two of p. g and r are
izati
gonalin on procedur al for R° for the set
inde endent
linearly indep vectoortho
rs given obtain an orthonormbasis of awed. tp ag = @
=(1,-1. 9)7
10. x, problems 10 to 13. —— 3 ;
,
+ €yc2). tank (A) 1s
1 —- »—4, OY. =, 1,-1)’, x; = (0, 2, 1)". 30. (a) 2; (b) JA] = (4,42 * bb,
0:
(i) 4, if aja, + 5,5, * €1e2 * the value zero.
sae -2)7-4, 2,= x, 1)7,-1, (1.= x) 1)’,: have
since al | determmants of third order
(ii) 2, if aya, + 6,52 + C12 = %-
page C1,= -~1Y,
=.
0-177,
Hmit x
371, G,= x,
* A - iy ay} =F + (-1)"'A*. In the
32. Consider (1 + A) (I-A Ee
( ei Ae = a
Shin
eta Setndati A™ —> 0. Therefore, (I + A)
eas
bs= & Trace (A) * B Trace (B).
= @
Trace (2A + BB) = a>.a, + BY
—ae 15.to 14 ee i amiLa
14.

33.
(1)
1”.0, 0, (1,= x, 1)".1, 4, (4,= x, 1)", 1.1, -¥;=(1, eng pe

=I
0,, re |
= x, 2.1, 0, (1,= x, 1)’,
(0,2,2, D7. 0,2,0,= x, 1).
3.10
Answers Hintsand Gi) Trace (AB) = 55, Gaby = SD bina: = Trace (BA).
r=| i= i=l ai
(AB - BA) = Trace (I) which
gives 0 = 2 which is not possible. :
(iit) Hf the result is true, then Trace
Exe 3.1
itis ruc for p= & + 1. Nowe
Let it betue for p = k and show that
:
have CB” | = B'*' C and CBC = @
322 as Result34. for trueis |. and
0 = p
I I a = we 0, = C
= when that BC and CB= ™
At=<
” 2-3 2] Ket By Bye Se
=
satin eat 2 | =i =H
esYe Ce Canin
= Gtx.G+ Seeger
f =wZ 2 thamene
Apaly
i> 3 -1 Symmetric.
37. 38. metic
Skew-sym
RG tag None. 36.
eo
Gi) ee ee eee i None. ,49. = 41. itien.
t3 Skew-Herm
Hemniti
- 39.
Hermitian.
Simce (A). = adj
adj =
(BABI, BO - — . 43. itien.
Skew-Herm Hermie
44. -
, Thacetow,BE. 8) None. 42.
(AT adj
= A)) (adadj (A) 0G tar?
= 1 Bad Ratewe
A~*= None. 45.
aplying
-i
usingand
z= A by eet we IAL =A™~ (A) adj
9. IAA“) = ABA“) = © oe LAY = HI
(A)) (adjadjoF AI AP” A. 1A= ~ 3.2
(A))(adjad 149) AUA = Exercise
6. 5. 4. I, No, 3. 6-. 5, 4, 1,
Yes.1. No,
2.
6 is not
No, when the scalar q is irrational, Property
taken
1s
A= B* . Simileriv.
scalars
of
A sa=B'A=a
field theIf
13. AB= BA — B’AB=

4.
satisfied.
A“B = RA-
ee B=BA™. only as rationals, then it defines & veactor space.
@ (AB *y = By a’ = By a’ - BAH x, the zero vector @ is 1 = I. Define
= Yes, since 1 + x= lx =x=xandx+I=« | =x —
= 1 = 1= ©. Therefore. negative vector
ts its reciprocal.
_y = I/x. Then, x + (— 3) = x(x)
=
Ba y Ga)Y
(A“B B’=
(a7y* =
tp Bata
7. No, 2 3. 3 10.
No, 8, 10.
aae =(@ay"y =
+ Br.
MABy'Y
(a * B)x = x78 = x* x? = x* +3? = ox
By =
0477? =
ate
Yes (same arguments as in Problem 5).
-
Pre-con siies botk A. T+ Gi)A,— I ) by
No, |, 6.
:
(ii)
(PAQ;
Yes,
=
=I *=Q QAP?
(i)
=A4"P 547 =
OP
(ii) No, 1, 4, 6
:
(i) Yes,
Use=
(1+A+(I-A)
A7+_j
x, ¥ € Wirtye FF,
(ii) No, when
: L 7. andBO A” by : 10.
=
Saccessively Pre-oanitipiy (ABC; (ABC)
(i) Ves.
ye Wi x+y e FM, (iv) Yes.
11. }
(iii) No, when x,
Lit19. ee ; : *
[(9 — 461 ys] (i) No. when Ae W.aA€ W for @ negative,
22.) k and £42
6= £6, oc = sm
2-3, k=2Gi) : -3 k= of
=
y- : ices need not be non-singular,
(ii¥*No, sum of two non-singular matr
12.
3
pgwber
25. GAx wy,.G) 1-3
cag. ;
as 2 27. 1. aoe (iii) Yes. # (2A)°)-
not belon gto W (A =1LA°=1= A but 2A
(iv) No, aA and A + B need
9. IAS= (ah gr — PhMp+ 9 + ry,sank (A)is . (i) Yes,
(ii) No; let @ = i, Then, @A
= 1A € Ww
 —1 0)"
s(0, 1), where r = y + 2x, $ = 2 + 3x, dim(ker(7)) = 1, dim(ran(T)) = 2.
42. ker(T) = x(1, 2, -3)", ran(T) = x@, 3)" + y(-1, 0)" + 2(0, 1) oF ran(T)
T : r ny? + =(0, 1)" T) = r(-] , oy! + : o!
= }. dim(ker(T)) = 1, dim(ran{T))

= x — ¥.
TY,re.where r eg 41. ker(T) = x(l, 1)’, ran(T) = xQ1, 1° - vy, TY T = rl, lo 1) \1 OJ 0 O/
T : - T l?. ! 2
dim(ran(T)) = 1. 0 0} (0 1 (t 0) eo
E,,
Ex,
2,
E,,}
order
standard
basis
of
the
is
E,.
ker(T)
{E,,.
x(1,
where
0)?
-3,
40.
ran(T)
2(0,
1)’,
0,
dim(ker(T))
4,
basis:
number.
real
(i)
2,
+dimension:
=28.
=1. |
= 1, dim(ran(T)) ker (T) = x(- 3, 1}', ran(T) = real number. dim(ker(T)) 39,
a.
k#—-8—
(iv)1 +
k# 01
(iii) #0.
k (ii) 42,
- 1 #0, (i)
independent.
linearly
be must vectors givenThe 27.
= 3. r=x-w, s = y + 2w. dim(ker(T)) = 1, dim(ran(T)) where

0)’,1, z(0,+ 1)” 0, s(1,+ 0)"0, r(1,= | 1}, 0, 0, (0, 0), 1, 0, (0, 0), 0, 0, {(1,
1, a basis: {(1, 1, 1, 1)}. (iv) dimension:
basis:
(1,
a 3
,
(ti1)dimension:
2)7 0, w(1,
+ 0)?1, 260,+ 1)7 0, y(1,
+ 0)"0, x(1,
= ran(T) 1D},1, 0, (0, I), 0, 1, (0. 1), 0, 0, {(1, basis:
a 3, dimension:
(11)
ii)
0)},1, - 1, (0, 1), — 0, 0, {(1,
dimensi
: basis
a 2, dimension:
(i) <6.
38. ker(T) = w(1, -2, 0, 1)’,
independent
linearly
. are§ in elements
The 2))/3.
+ 24 + 2¢r+ 1) = (f° +r+1=[-re+
25.
3. = 0, = (T)) (ker dim
dim(ran(T)) are§ in
dependent.linearly The 2). 1,
vectors (3, +
9-4) 2(-1, 2..5)(1, = 4, 2 ‘
37. ker (T) = (0, 0)", ran(T) = x(2, — 1, 3)’ + y(l, 1, 4)’. dim(ker (T)) = 0, dim(ran(T)) = 2.

x 7 7 = ‘ - . Z

0)’.1, 2(0,+ -1)70, y(1,

+ 1)? 0, x(1,= ran(T)
0)’,0, (0,= ker(T)
36. independent,
(v) independent,
(1v)
y). * T(x# T{y)
+ 7(x) linear,
Not 35. dependent,
(111) dependent,
(11) Independent,
() 22.
1). TO,+ 0) TQ,# 0 = 1) TU,2, = 1) 7(0,3, = 0) 7T(1, linear,Not 34, dependent,
(¥) independent,
(iv)
' ‘ ) a dependent,
(in)
y). + T(x# T(y)+ 7(x) linear,
Not 33. dependent, independent,
(i) 21.
wd— +e + 2a (- $= and
2 (11) | ae
2/3
= 5c b+~ 40 (- y= 4dy3,
+ 4c — 2b+ (Sa P=
t

31. Not linear, T(x) + T(y) # T(x + y).

2a)/3, 2¢-~ b — a (- = a where5D,+
32. Linear.
yC BB+
+ @A= r '= Ek
30. (i) nr’, Then.D), B,C,(A, S= Let 20.
5
(iv) n(n — 1)/2. (iii) n(m + 19/2, (ii) 7,

ae et
py-
where E_, is the standard basis of order 3.
ja
(a —= : 8
“
62,
+ (a @&
=ja-+ cere w¥ By B + au = oc)’ 6b,
wh; - (a. = 5 ‘}
Then. wi} ¥,
¢ A 10, 5S Ler
= ae = 19.
(iii)dimension: 6, basis: {E,,, E>), E, E3;, E32,
Es}:
basis: 6, dimension: (11) .C8).
SP + Petr)
- SPC)(mn) P3(t)
{E,)>
E43}. E33, E33, Ey; E,3;
— 2Px(t)
= ary -3P ()) ia.
possib
NOt
le (itt) Ww, l
29.
(uj3u+yv— 2w,+ By (i)
dy = ou 7.
(i) dimension: 3, basis: {E,,, E22. E;;}.

li of of e (vi) dimension: 3, basts: 23wi'

+ leIw — 3m (—3 Car)
lo
w
: J|
¥— + (ujzu Ww, 2y w+ (i)
1 -1\
16.
(0 0)| (0 0)
number} ration@a
al (for W, f Wlaxxe
e No. iv)
(v) W. Warrte No,
o dimension: 3, basis: (v) e (iv)
of
negati
@ ve,tor Wy 2
li of lo -1) :
ay ,
c
ze No, (im)
fo o\/! oO 1\ o

Tye ye pcan
+ eae RF eee
I
i}, ' 4 = a. Dhow
-— a) my. (x).= ma if example, Bore
dimension: 4, basis:
bpr
of oflo _1
Boe -¢E Rye
lu oflo
Bi WN (ar)
' 1 0
9 1\(0 0\(0 1\f1

Ves, (i) 15.

os
matnx, and C is a symmetnc ee skew-symmetric re
4 ug? = O ae ee
yr fa] 1 F
+

“bh, ia —b, +ibz i

&
pe 9)
i) P.
te
P

for
I

So,
=57|

(/

ic =B+ Iv)
not

real

W, Oe P+ also Wand
real

have

roots
need

|
"2 9 Ai 7
pias

Oe FP,
b,+ib2\) _(
for
ia,
No,
roots,

(tn)
Yes (i)
having

i, =
_

HP
(iv)
ore
WwW,

i)

MEePRE:
oOe 4)
a2 x 2 skew-Hermitian matrix (diagonal elements are © OF pure unaginary) 1s given by
i:

..
3.105 Matrices and Eigenvalue Problems
:

5104

Mathematics
Engineering
 Matrices aed Eigenvaiue Problem 3.407
44.{2 - @ 1, @ 3). @ arburary 35, f-14, 4, ba, ay
36. { - @ @ al. o@arbetrary.
37. 1 }-
iS. San, @ ah.
arbirary.
38. (0, 0, 0).
9.
[-lal3, Ja}. -
Bed, a ao).
eberary
B aj. a B arbetary.
40. (2 (P - av3. - OB * ay3.
41. [0. 0. 0, 0}.
al. a B arbitrary.
42. (28 - Says. - (108 + a4. 8,
B. al. a B arbitrary
43. [(a > S5BW3, (48 - Fav3,
3,B al.2 B arbitrary.
44. (38 -— Says, (36 - 4ay
arbitrary.
45. [a — 36. 58 B a) a, B
nf 2) [4 - f-) § 3 i
tah

s
lbal
ae ee b 4
Fug 9
have‘We 47.
12 3/=/2 Sid esa — Pe)
hl

v.) fv,7
a if
a=)wijws fw,= Z| '
at
1-1
2 47. =?
46. 1=—3
' ‘4 al
‘om

1 —2Z 3|
Le 4
Now = 2 =I
¥ any
WEStOr withIR? im =.) (2, X=
respect be can basis giventheto
written as
-| -3 WwW i
ix]
[ =| -1 -!
eT 3}
= = t
+ 0 1 = | 0 0 -1|
1390
72i Pla] it] 0
|
0 0-1 — yy 6 8
)2. Hence. we have 711
1 iG -W3 a)
4] io Oo | |
bs 4a+5f j— 6x, + 7x,
Tx= aTv, + BTv,~ al2|+p [|= ]2a+38 =5 «Rie gies
0} i! B 24, — 35 Exercise
3.4
—

|
j ot 16 20]
48. fx =} —%, + 22, +8,
sid;

;
iZr;

. 1
1 PUY = 2 - 98% -944+81~0;A'=—| 16 1 4S)
irs

Dx.
a |

TPs)49. (4x,= Sx,)- T(x,+ x,)¢- -—

x)e?. (2x,+
+
*
_

Two “) 50.
degrees of
dimension freedom. 2 2. is
basis ({3.is
;
[-2, O},1,
wal
1p}.O,
6 —4 -2]
|
degreeOme {ti) of
dimension freedom. a i. is
basis is
{(-5, 4,
23)}.
2. P(Ay= 2-827 +200-16=0,A'=—| O 8 4
7 =a
Exercis 3.3
no? z. 3. 3. 4.
Z

bh
6. ers exist.
not e
5. 2. a. 2. 8 3. 3. P(Ay = B- 38 + 2A = 0, Invdoes
independent,2.
a. 19. z: it. 2. 12. 3. [4 4 -4]
14.
13. 2.
15.
2.
rs
Dependent,
4. pay B-2- ads 4 AT = 21-2 =
16. independent. 3. 17.
18. Dependent, 3. [-2 1 |
{J. Dependent.

3.
19. independent, 7. 26.
21. Dependent, 3.

2. 2
22. Dependent, 2. 23. 1A. independent, 4. 2-3 -7|
25. Dependent, 7. 26. i
~
.
- 5k 7 W-LEDA
a! | 5 3
5, PAY2=

2]
= | = “ls

2,
} ae
27.{! + a - 2a. of. o arbitrary 28.
29. {1, 3, i} 1.

gb
(3, 3, 3]

eee
ee
oe
1) 3/2, (3/2,34. 3/4]. -1/2, [-1, 32.

:
mG
(5 33.
ae Rs
— a +
jé Aaya? - Fst
46/3.

ae
+ 2a+ (1
B35, B a. Bi.«
atoitrary —

ee
 +p
33. Premultiply Ax = Ax by p' and substitute x = Py.

= B of
15/2. eigenvalues 2 Min ; | '
: Eigenvalue
32. s oF. 17.1.12.0, ee, 3 1,- 2154
23.
of sum = B of 1.
Trace 11/2,1, or 3, 2, 1, = 7 - Ae.+
(1A), A; are B of
154.= B of eigenvalues 01 r @ ¢€ =)
of = |B]
product -11/2.7/2,4, - 2, or 4 2,3,1, = 1,7- + 24, areB of
(1/A,) Eigenvalue
31. s 1 -1 -1 0
IfP'=|1 P=|
0 =! I J -! 1
15 6 3 | 5 2 1 Oil -!|
0 0 =--/0 6}, 4
A=PDP™ -2 1
aa 2. Pe] 40. 1 he A= I
OF 1K =?2 LAYS 6 A=
flO, 22,
atiead iz
45 18 9 =-3 12 =2.3
1, -8 7) 2467 i l=2'-6 —4 — -2
32
-8 2-6 2i-6 3 PP 143i1 P=|
-1 22 tC | 0-1 «16 -8«= 24 ih 3-i 3+i 3
=3
—
1j|;P
-—2|,A4=P
ae
29.
1|O
lI.
P=; DP
Ue Eee a : ie = 4" 3, - | (3-7, egy
A
|
afow eo si, ae Aleh hee)[Sp A 2h
= =A l l mf = ; % 7 J i a

“i
4] =2 “fat 1 2 Oo 1
sete
|
2
23
2
P=
2
0
z
=-1
|
l
1

OJ. 1 2|,A=PDP
0 ™'=|0 I;P=2|
-1 12 P=| 28, a) [-s 24) # +
100 22 fo “ey Bi be SB ED
feb es , 2S
(+ 1 Oaay [= = a

38 4 34 }— l 2 =) 9 4 1 1 7 “| |L iu |- |

34), i 2 |,2
A=PDP'= -8 14
—— 3
4|P'=— P=|2.27, 0 2 al |
pPis=tig 0 P=/|
46 -12 4 =-5] 14 —-l)l |] 12 . ryt 4) + i

1)’ Oh fds de ee Bea pea dy oe oe A if.

-4 -3 3 —! —| 1 J | 0 diavenaliza
4, 2.9,ble (ood, on EPSGe bike Ser A
£1, M 1 = -A=PpPr*
2 1 -pol=|—|
1 0 P=|1 26. ie ae ae os Let uninyof reat fitth HT, SE gi, A Ma,
-7 -5 6 I 2 -1 3 i I diapaanltzub
OOP", le O20, MO.
i
ate
Be
do
Eps.
|Vg=ee
TG0
tia
Bo
AFO,
7.
2 La oak : _ 0 diagenatiza
bp. 1, ble 2 I v2), WP Ga a = A
= “i ase " = , ESL ae A a EE = a “Ss No, se mE A 1G.
PDP~'7 | ”) \ | -. po! 0 0) diavoanwhva
|| i Tg
bleI,|

32) 6 + 0 4 oi | Bl 0; tk i. Ce gy", oy ade OP

[Vy NB otk
| ! ‘a I. i. : A: Is.
daawomuliza
Mo. od ble [ee ft A
I -! Oo 2 7 | -i TY 0, 14 ' A tobe || ony otk Yd. OU A I4.
-3 15 aapo!
. —5 | P=|-I divarendivali
et leOS PO ad moo 4 Is.
—6b & 2 #& oO: “7 diayenalica
1p. fio,
ble A Bette pth DP EM Be A I2.
diaponalwabl
‘)o. e1) wi ft vit i
ae5, — [8 isA=-2:
iL. [0, 2: a= 1}; l=- fl. A=] 24. | | ik Ll. ‘ee | || 1), ly t A
; a a able nol
diaconabia4)" (oy 4 |I a | | | ii.

. 7 dwenhieable
mL “ 3 io
-5 -6 3 0
1|:p"'=| P=|-1 ony: i a a a ae
ia a 322 divvzanteeat
fp. 1 h tl rt | Be ee! ae il 4 dig By he Sa ,
4 5
cmalicable
obey int Wy Pe fee A | 1 | | ‘
| 7 uy
3,109 Problems to
Eigenv
and
alue Matrices tdEny PPM Miarhe
rn Peer
sie LUN
 273 23 —i3 009
b, mas Gb ATABTD™= (Ab) (Al= )’ *¥ i
273), -1/3 2/3 0|,Q6 =| Then resp
(uveu
P=O723.
',d=|0 v ecti
u, vely. Vecttoors
transf ed
be borma, vectors
1/37 =2/3 [2/3 0 30 the Let 52,
V"A« U, yTaw ty (UT= AV)? UTAV since
| 9 > Vla
+y UTA« U ¥7A
~ U) MUT
4 AV VAY
+ )
we -2wWe6e We (UT
= AUAz 3"
00 o Then vectors,
rea}be
|.
-1 wv2\t #0V 0, ¢ U wherefV,+ U
vV2 0 —~VJ2 Ge
z= Let SI.
DON
6 "ga
I | P= 12.
we ows W3 is matrix diag
a onal sincA, symmetricalw
fe e .. ppp
»r PT PH
= . ays
since PDP=’, PDP=“!A (PDP
» ’)?AT have We
orthisogon
NowD.
. = PAPthat such
P P al.
matrix orthan
ogonal exists
Me, A There 49,
|' 0.08981 since
0 = r'y whic0,h =
0.99596 |’ 138197 0 px'y xy- (A obtain
we resulls,
-052573\" ~087065 : = , Ay x’ get we two the Subtract ing
0.99596 nd
equation,secothe From Ax'y= . Ay x’
0
—0.0898| 0 3.61803D O.87065 Axl,= xTA or Ay= ! obtawe
in y, by
052573 P "1. x7A7 get we Postmultiplyi
equation
first
, the From My.« Ay
have.We and Ax ~

—_—
eigenvectors. corrthe
espobendin
y x,g and Ax
cigenvalues
distinct
two be yt

-—
-094237\0.33458 |" 0.620060 andA Let 47.
019795] 0.98022 .
° 033458] 0.94237Si 0 = I/|A|or |A|= 4. 4 = JA] of 1 = JA]?= [A]
04510
| = 1p [A’|
= [Avy Hence,
Al=
-—0.98021 0.19795 “a Av! havewe matrix, Orthanogon
is al
A Since46,
mym,
+ ily0, = nyny+ mymy
+ Mls0, = nny+ mymy
+ Il,
5.312,69. and 1d 18 We 7 +m
as conditions
ObtaiWe
n I, = AA’ givesal
2. 68. A= 45,
5.59.67.
6.171366,
. BAT
= (AB)?ThenB’,= (ap
= y, BUA
=
2).+ 2y4+ B-! and AT = Av!is that matrices,
8y) 65. 4y2.+ yh+ yj 64. orththe
ogonbealB andA Let 44,
297.- 3y}+ yj 63. p’,a po! oe | mw "')T PP when
‘ result Thisp. = D? since "DP7), ((P-') only trucis
oO 2+: 0 (PD™P
= 7 -') (PDP) (Ppp = y* ATA
= T ACAP=
06 -2-33 Oo
Sy.- Sy,+ y\ 62. 2-j -3 havewe
61.|3-4i ~24y|.
| |-!4+2i 60. symm
isetri
A c, Since43,
0 3445 2 9g
symmet
is ric.A’A™' Therefore,
ql?ow T (AAT)
mm )T
-1-2i 1 AAT.
= A(AT
=>)! ATAA=
TAT=> ACA
Or
of AT=
ATAT
= AA haveWe (A7)?,
= A? let Now, (A")®,
@ A*
| (A"'or)7A? AA, (A-'
= )7A (A7A7')!
-2j =—4 4TH as haveWe matrix, symmetri
hea c
| Ss & A7A™Ler
' 42,
21. 0 |-i 59. 4 1 1 58.| ) 2 =f is] Sh. ATA,
= (ATA)T41,
~4|i 1 3} i 5S BA.@ BTA
=T (AB)’
0 -1 1 ThereBY.
fore,
= B and
A?= A have We 40,
positive.
NOL até eigenval
the all
ues that |BI.= |P""|
= |BI|P
[Al| Also
verified
be also canIt positive.
not are minors leading p-'p-
= ipA! that show equatthis
P"'B
= "P.A™ and
the All 56. ion, From P-'BP
= A. havewe
9, > x2+ 3x3+ xs)+ (x,+ x,)?~ (x,= 2x}+ 4x3+ simila
are
r,B andA Since
39,
Qxyx,
+ 2xyx,
- 2x}= zero.
to clement off-dia
the gonal
set and side hand rightthe Simpli fy
jg,
r.
Xy 2 l —| J. or 0
eigenva
the lues Hence, are A Of
X3 ot A I. = D or 0 = D Thus0. = D ~ D? obtain
we
13 00) 5, [x= SAX 5S. get We A, = A‘ SinceD*. = Therefo
D?.=re, P’'AP
PUA‘Phave We same.are D and A of
x 3| 1 2 =p PAPta such cigenva
the lues
and
P matrix non-sin
a gular
exists there matrix, diagona
a is lizable
A Since37
0, > b’ — ac 0, > a Therefore, ), # x since
0, = A or 0 = #4 Therefo
0. = re, A“x= A‘x obtain
we Ax,= Ax From 36,
X3. Ky, all for 0, > h*yia Ax.= y
- "| (ac x + bx./
+ ay a(x, eigenvector
= correspo
the ndin
with
g BA!of
eigenv
an alue
alsois A Theref
Ay.ore,
= BA“'y obtain
We A“'y,
= x set andA by Premultiply
XZ c h Ax.= A~'By
is thal
ex}+ 2bx\x, x3] [x,,
= Ax — =| ro
A’'B,
of eigenvector corres
thepondin
be x g and
3 h
+ elaax)=
_ r
x 54. -1
: :
eigenv
an alue
be A Let 35,
x, b\\ a ~
Ay.= x by related
are eigenv
and
ectors
BA of
4}. 12, = u,P
an alsoIS Therefore,Ay.= eigenvalue
3}, and2 3}. (i, = U, BAy get We Ay.= x
colu
ofmns inteby Ax.= AxrE substitute
and Av’ by Pre-multiply
rchange foll
3 owedand2 rows = ‘oe eigenvector correspo
the bending
x and figenval
(interch
Yes.ange
(ii) No. (1) an beue4 Let
53.
SAV Problems Eigenv
andalue Matrices Mathematics Engi
neering
112 EnginMeer;in
§ Mathemar
i es

‘
; ‘ars 0
*FeQ ips ] (0)
| 2 0,Q=/0 0
1/5 1/8
004
10 1/9 ~1/,/2
75 P= eI i} “fF 0 0

2 Pp
1/14 3/J14
Lae Ee gh O= 40
2/14

aap a
1/35 3/35 ~5/ 35
Exercise 3,5

l. {@,b,clia=ph
==) (lad, e) last
en 0,5 +4c=0)},
2. {(-1, 3, 5)"}, {(0, 0, 0)}.
3. Basis of (5, 4 So) =
(1, 5, 37}: 1. 3.
4. Reg 4 & ‘S11 S)
= ((0, & 0)7}. Not a direct
5. (SAS) = ((0, 0, 0)") sum.
.R=s5.@ Sy
6 B=5, 45, (s, MY S>) =
{(0, 0, a)"}. Not a direct
sum.

iti
8. A basis of S, : {(x
, y)} = {(1, 1,-1), (0,
is S = span {e€;} = span 1, 1)}. A basis of R°
{(qr, 0, O)}, & arbitrary, : {(x, y, e,)}. Compleme
We can also take § = spa tary subspace
complementary subspace n {e,}, or § = Span{e,}
s. as
9. A basis S, : {(x, y)}
= {(1, 0, -1), (0, -1. 1)}
is S= span{e,} = spa . A basis of R3 - {(X, y,
n {(a, 0,0)}, a arbitr e,)}. Complementary sub
complementary subspaces. ary . We can also take § space
= span{e,}, or § = spa
n{e ,} as
10, (IN2)(1, -1, 09"), (ING)(A,
1, -2)7 (IV3)(1, 1, 1)7.
J

11. (1/N3) (1, 1, 1)”, (1/V6)(1, -2, 1)

(AV2)(1, 0, =1)7.
12. (1/V2)(0, 1, -1)". (1/N6)(-
2,
-1, -1)7 (1N3)(-1, 1, 1)7
13. (1N2)(1, 0, 1)", (1V6)(1, 2,~
1)’, (IN3)(1, =1, -1)7,
=> (WN3)(1, 1, 0, 197, (0, 0, 1, 0)", NB)CA, 1,
0, -2)', (WN2)\(1, =1, 0, 07
15. (1N6)(2, 1, 0, 1)", (1N30)(-1, -2,
3. 4)", (1N630)(-8, 19. 14,
~3)', (1/V21294)(65, —52. 91. 78)"
1
 ant Peaabehonas
Afhaterees ctl Page vecti
il
idiot,
be clianyronvally
tdfable 1 tar
Be
io
;t
. (ii, } yo
yey
“
ehni wr
EE ia
miairis
reyoe
A
Math
\*
wiall
‘
nant
” f-4i\ V\ o 7
ty dom! 3 1 tar all
pelaee
Comliliiy are
Me the abowe
1
| i)
n paverear
y
w diagonally domina
fanny
= ty, 1s called imequalily bss abistied fart
ard ~
fF. iW the strict
Oo; N al 4/|4
é ax

\ @ ra1l POon
Orthog weg d
wi
and C" in
of equate
“ |
ial jcoast ane
ir
1%
ts wat afied fire
;
+ qnequality
ly dominant avefem
!l
orthanonorma
“T H
ee
jetty diagonal exactly one |
in enulli Ptiw
set (' in
teo Veal has
Orthonor
ation ma
Let, w.. a), called a permut
ane
unitary
, normal ya) a. he
system
for ms aN Ortha
4
A matris pis
©
vor
af
Lt |
ors
vectof
) :
if ' a tT VA Tipsy iti in"
rey rif veuloy ‘.
Then, this @,
other clements are prapert
is sand to satisty the
matrix. Then, the matrix
Aofe matris Let B be a sparse P such that
i, i, 7 i i. r
mutation matrrs
: :
re c¥tS ts a per
and only if the
| Lin: J he i i rf |
‘_
| *|
| |
ifoT ppp! ~ Ay An
Ay An
hanges
on performs Faw interc
7
ilarity transformati
al mainces. The sim
we }
matrices. diagonal
i.e,
W Wve
Im
defined
«
symmetric
A,, amd Ay, become
Clin 19
and A») are diagon
show symmetry
where Au ponding column interc
hanges i B such that
can be reduced to the
e
defdefinenow
followed by corres Of testing whether B
special more fewa tre
cedure §s a simple way her the interchanges of
The following pro
Orthogon
ments and tes ie whet
matrices
matrices. ons of the non-ze ro ele
rr n realA
A omalnati
ortis hap
loc ati required form [ct
A’ if
the
required form. It find:
A" = ple “me
le to brenip Bote the
cxamsimp) A is
of columns are pe maib
iu in al }
fet there exist disyoint
Loe
A= ding interchanges
‘in® con rows and correspon
matrix B and +, # 0.
Denote the set E Th ie ae es
re ro off dmiponal clement,
a be the order of the suflines of the momn-
h that U- Uw U,, where the U',) Then, the matria
#
Uy, 4
A on
the matrix of transformation "an orthogonal matrix is called an subsets U, and UL, suc
mati
line
Cjpartee
sfor
tranaf
orthogenal transformation U,, K
ch
whiin
d as either (1
by # 0, i # &, can be groupe
satisfies property A
A "ITA )
Vifary iMate “wt If A is real, then -2 ]
Wy
trix ts same as orthogo nal matna
’ ) i i ) ;'
Be| 1-2
Consider, for cxample the matrix
irenfornanio n
trans format pon mh @ uNitary matrix is called a unitary 0 1-2]
the g:note We
followin
pou 0 |
jo oI |
ar fi, since Let the permutation matrix be taken as Pe
7 ee |
PR) A+ (ar
iB)’ +f)’iad
« aa’ | eaAa+pRB+ fp
~
since

2.
Fogenvalues and
genvectors
cigenvectors the are AA of of
conjugates the of
crygenvalues and
eigenvecion
(orthogonal)

have
(orthogonal)
of
A,

pives

AiG
At
unit Ay

Aw
wveree

writary

unitary
=
~

matrt

A’!

A’
vet!
teThe

Z fy
Then
B<

=
{)

We
A". | of
O)l<-2

=
Oo

is
\,

a 1

,
7?
)

i)

atrix
-2

:
aif? Oo 1 } 1
oO 1 4 O —2 \ (! i]
B | ' = A= = (A) A’ (Ay)
4 «tay =B "