YASHWANTRAO CHAVAN MAHARASHTRA OPEN UNIVERSITY

SCHOOL OF SCIENCES
(FORMERLY SCHOOL OF ARCHITECTURE, SCIENCE & TECHNOLOGY)

V151: M.Sc. Mathematics

2023 {As per NEP 2020}
Pattern

MAT501
REAL ANALYSIS
(4 Credits)

Semester - I

Email: [email protected]
Website: www.ycmou.ac.in
Phone: +91-253-2231473
 Yashwantrao Chavan S25011
Maharashtra Open University Real Analysis

Brief Contents
Vice Chancellor’s Message iii

Foreword by the Director v

Credit 01 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Unit 01-01: Metric Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Unit 01-02: Open and Closed Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Unit 01-03: Sequences in Metric Spaces. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Unit 01-04: Continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

Credit 02 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Unit 02-01: Connected Metric Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Unit 02-02: Complete Metric Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Unit 02-03: Totally bounded subsets of metric spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Unit 02-04: Compact Metric Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73

Credit 03 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Unit 03-01: Riemann Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
Unit 03-02: Necessary and Sufficient Conditions for Riemann Integrability . . . . . . 94
Unit 03-03: Properties of Riemann Integrals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
Unit 03-04: Mean Value Theorems and Fundamental Theorems of Calculus . . . . . 115

Credit 04 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Unit 04-01: Convergence and Uniform Convergence of Sequence of Functions . . . 126
Unit 04-02: Properties of functions preserved under uniform convergence . . . . . . . 138
Unit 04-03: Convergence and Uniform Convergence of Series of Functions . . . . . . . 150
Unit 04-04: Power Series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159

Answers to Self-Tests and End of Unit Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169

Appendix 1. Foundations of Mathematics 195

Appendix 2. Sequences of Real Numbers 205

Appendix 3. Series of Real Numbers 211

Appendix 4. Geometric series and p series 219

i
 Yashwantrao Chavan Maharashtra Open University
Vice-Chancellor: Prof. Dr. E. Vayunandan

School of Architecture, Science and Technology

Director (I/C) of the School: Dr. Sunanda More

Programme Advisory Committee (PAC)

Dr Sunanda More Dr Manoj Killedar Dr. Chetana Kamlaskar
Director(I/c) & Associate Associate Professor, School of Assistance Professor, School of
Professor, School of Architecture, Science & Architecture, Science &
Architecture, Science & Technology, YCMOU, Nashik Technology, YCMOU, Nashik
Technology, YCMOU, Nashik
Dr. T.M. Karade Prof. Dr Shivdas D Katore Prof. Dr J N Salunke
Retired Professor, Professor, Professor,
R.T.M. Nagpur, Sant Gadge Baba, Swami Ramanand Teerth
University, Nagpur Amravati University Marathwada University,
Nanded
Prof. Dr Meenakshi Prof. Dr S R Chaudhari -- -
Wasadikar, Professor and HOD, KBC
Professor and HOD, North Maharashtra University,
Dr Babasaheb Ambedkar Jalgaon
Marathwada University,
Aurangabad

Development Team

Course Coordinator and Book Writer Book Editor

Instructional Technology
Editor
Dr. Chetana Kamlaskar 01. Dr.Subhash G. Pawar Dr. D. D. Pawar
B.E., M.Tech., Ph.D., M.Sc., M.Phil., PhD(Maths), M. Sc. Ph.D.,
Assistance Professor, School of Head, Dept. of Mathematics, Professor,
Architecture, Science & K.T.H.M. College, Nashik School of Mathematical
Technology, YCMOU, Nashik, Experience: 23+ Yrs, Sciences, Swami Ramanand
Experience @YCMOU: 20+ Credits written: 01 and 02 Teerth Marathwada
Yrs University, Nanded
02. Mr. Chetan Shirore, Experience: 20+ Yrs
B.Sc,M.Sc(Maths),SET,NET,
Assistant Professor,
Dept. of Mathematics
K.T.H.M. College, Nashik
Experience: 8+ Yrs,
Credits written: 03 and 04

This work by YCMOU is licensed under a Creative Commons

Attribution-NonCommercial-ShareAlike 4.0 International License.

✠ First Book Publication : 19 March 2021, Publication Number: 2364

✠ Publisher : Dr. Dinesh Bhonde, Registrar, YCMOU, Nashik - 422 222, MH, India

✠ ISBN Number : 978-93-91514-46-4

This SLM Book V130: M.Sc. (Mathematics) {2021 Pattern}, dtd. 19/03/2021
used in V151: M.Sc. (Mathematics) {2023 Pattern}, dtd. 31/08/2023
 VICE CHANCELLOR’S MESSAGE
Dear Students, Greetings!!!

I offer cordial welcome to all of you for the Master’s degree programme of Yash-
wantrao Chavan Maharashtra Open University. As a post graduate student, you must
have autonomy to learn, have information and knowledge regarding different dimensions
in the field of Mathematics and at the same time intellectual development is necessary
for application of knowledge wisely. The process of learning includes appropriate think-
ing, understanding important points, describing these points on the basis of experience
and observation, explaining them to others by speaking or writing about them. The
science of education today accepts the principle that it is possible to achieve excellence
and knowledge in this regard.

The syllabus of this course has been structured in this book in such a way, to give you
autonomy to study easily without stirring from home. During the counseling sessions,
scheduled at your respective study center, all your doubts will be clarified about the course
and you will get guidance from some qualified and experienced counselors / professors.
This guidance will not only be based on lectures, but it will also include various techniques
such as question-answers, doubt clarification. We expect your active participation in the
contact sessions at the study center. Our emphasis is on ‘self study’. If a student learns
how to study, he will become independent in learning throughout life. This course book
has been written with the objective of helping in self-study and giving you autonomy to
learn at your convenience.

During this academic year, you have to give assignments, complete laboratory ac-
tivities, field visits and the Project work wherever required. You may have to opt for
specialization as per programme structure. You will get experience and joy in personally
doing above activities. This will enable you to assess your own progress and there by
achieve a larger educational objective.

We wish that you will enjoy the courses of Yashwantrao Chavan Maharashtra Open
University, emerge successful and very soon become a knowledgeable and honorable Mas-
ters degree holder of this university. I congratulate “Development Team” for the devel-
opment of this excellent high quality “Self- Learning Material (SLM)” for the students. I
hope and believe that this SLM will be immensely useful for all students of this program.
Best Wishes!

- Prof. Dr. E. Vayunandan

Vice-Chancellor, YCMOU
iii
 FOREWORD BY THE DIRECTOR

This book aims at acquainting the students with advance Mathematics required at
post graduate degree level.

The book has been specially designed for Science students. It has a comprehensive
coverage of Mathematical concepts and its application in practical life. The book contains
numerous mathematical examples to build understanding and skills.

The book is written with self- instructional format. Each chapter is prepared with
articulated structure to make the contents not only easy to understand but also interesting
to learn.

Each chapter begins with learning objectives which are stated using Action Verbs as
per the Bloom’s Taxonomy. Each Unit is started with introduction to arouse or stimulate
curiosity of learner about the content/ topic. Thereafter the unit contains explanation of
concepts supported by tables, figures, exhibits and solved illustrations wherever necessary
for better effectiveness and understanding. This book is written in simple language, using
spoken style and short sentences. Topics of each unit of the book presents from simple
to complex in logical sequence. This book is appropriate for low achiever students with
lower intellectual capacity and covers the syllabus of the course.

Exercises given in the chapter include MCQs, conceptual questions and practical
questions so as to create a ladder in the minds of students to grasp each and every aspect
of a particular concept. I thank the students who have been a constant motivation for
us. I am grateful to the writers, editors and the School faculty associated in this SLM
development of the Programme.

- Dr. Sunanda More

Director (I/C) & Associate Professor,
School of Architecture, Science and Technology

v
CREDIT 01

1
 Unit 01-01: Metric Spaces

Learning Objectives:
After successful completion of this unit, you will be able to

i. Abstract the concept of distance and generalise it to the metric.

ii. Establish relation between three standard metrics on Rn .

iii. Define discrete metric on any set.

Introduction: In this unit, we shall abstract the concept of distance and generalise it
to the metric. The metric structure will be defined with suitable examples. The different
metrics can be defined on Rn .

Definition 1.1.1 Let X be a set. A function d : X × X → R is called metric on X if

i. d(x, y) ≥ 0 for all x, y ∈ X

ii. d(x, y) = 0 ⇐⇒ x = y for all x, y ∈ X

iii. d(x, y) = d(y, x) for all x, y ∈ X

iv. d(x, y) ≤ d(x, z) + d(z, y) for all x, y, z ∈ X

The last property is often called as triangle inequality. The set X together with metric d
is called metric space. A metric space is a structure < X, d > where d is a metric on X.
We shall often write that X is a metric space meaning that X is a set with some metric
d defined on it.

Problem 1.1.2 < R, d > where R is set of real numbers and the function d : R × R → R
is defined as d(x, y) = |x − y|. This distance function is called Euclidean or usual metric
on R.

Problem 1.1.3 < R2 , d > where R is set of real numbers and the function d : R2 ×R2 → R
p
is defined as d((x1 , y1 ), (x2 , y2 )) = (x2 − x1 )2 + (y2 − y1 )2 . It is the Euclidean metric
on R2 .

Problem 1.1.4 < C, d > where C is set of complex numbers and the function d : C×C →
R is defined as d(z1 , z2 ) = |z1 − z2 |.
Some basic properties of complex numbers are given below. Here z,z1 and z2 are complex
numbers.
p
i. |z| = x2 + y 2 . Here z = x + iy and x, y ∈ R.

ii. zz = |z|2

iii. (z1 + z2 ) = z1 + z2

S25011: REAL ANALYSIS Page 2

 iv. (z1 · z2 ) = z1 · z2

v. z = z

vi. Re(z) = Re(z)

vii. Re(z) ≤ |z| and Im(z) ≤ |z|

viii. z + z = 2Re(z)

ix. z − z = 2iIm(z)

x. |z| = |z|

Definition 1.1.5 Let L be a linear space (Vector space) over R or C. The elements of
L are vectors and the elements of R or C are scalars. A function ∥ · ∥: L → R is called
norm on L if

i. ∥ x ∥≥ 0 for every x ∈ L

ii. ∥ x ∥= 0 ⇐⇒ x = 0

iii. ∥ x + y ∥≤∥ x ∥ + ∥ y ∥ for all x, y ∈ L

iv. ∥ αx ∥= |α| ∥ x ∥ for every x ∈ L and scalar α.

Three standard norms on Rn : Rn is a linear space over R. There are three standard
norms on Rn .
a) Euclidean Norm :
p
∥ (x1 , x2 , x3 , · · · , xn ) ∥= x21 + x22 + x23 + · · · + x2n
b) Sum Norm :
∥ (x1 , x2 , x3 , · · · , xn ) ∥= |x1 | + |x2 | + |x3 | + · · · + |xn |
c) Sup Norm :
∥ (x1 , x2 , x3 , · · · , xn ) ∥= max{|x1 |, |x2 |, |x3 |, . . . , |xn |}

Theorem 1.1.6 Every normed linear space is a metric space.

Proof : Let < L, ∥ · ∥> be a normed linear space. We define d : L × L → R as
d(x, y) =∥ x − y ∥. For x, y, z ∈ L,
i) d(x, y) =∥ x − y ∥≥ 0
ii)
d(x, y) = 0

⇐⇒ ∥ x − y ∥ = 0

⇐⇒ x−y =0

⇐⇒ x=y

S25011: REAL ANALYSIS Page 3

iii)
d(x, y) =∥ x − y ∥

=∥ (−1)(y − x) ∥

= | − 1| ∥ (y − x) ∥

=∥ (y − x) ∥

= d(y, x)
iv)
d(x, y) =∥ x − y ∥

=∥ x − z + z − y ∥

≤∥ x − z ∥ + ∥ z − y ∥

= d(x, z) + d(z, y)
Three standard metrics on Rn : Let x = (x1 , x2 , x3 , · · · , xn ) and y = (y1 , y2 , y3 , · · · , yn ); in Rn
a) Euclidean metric:
d(x, y) =∥ x − y ∥
=∥ (x1 − y1 , x2 − y2 , · · · , xn − yn ) ∥
p
= (x1 − y1 )2 + (x2 − y2 )2 + · · · + (xn − yn )2
b) Sum metric:
d(x, y) = |x1 − y1 | + |x2 − y2 | + · · · + |xn − yn |
c) Sup metric:
d(x, y) = max{|x1 − y1 |, |x2 − y2 |, · · · , |xn − yn |}
We verify that each of these functions is a metric on Rn .
a) Euclidean metric on Rn
Let x = (x1 , x2 , x3 , · · · , xn , y = (y1 , y2 , y3 , · · · , yn ) and z = (z1 , z2 , z3 , · · · , zn ) be in Rn .
p
i) d(x, y) = (x1 − y1 )2 + (x2 − y2 )2 + · · · + (xn − yn )2 ≥ 0
ii)
d(x, y) = 0
p
⇐⇒ (x1 − y1 )2 + (x2 − y2 )2 + · · · + (xn − yn )2 = 0

⇐⇒ (x1 − y1 )2 + (x2 − y2 )2 + · · · + (xn − yn )2 = 0

⇐⇒ x1 − y1 = x2 − y2 = · · · = xn − yn = 0

⇐⇒ x1 = y1 , x2 = y2 , . . . , xn = yn

⇐⇒ x = y
iii)
p
d(x, y) = (x1 − y1 )2 + (x2 − y2 )2 + · · · + (xn − yn )2

S25011: REAL ANALYSIS Page 4

 p
= (y1 − x1 )2 + (y2 − x2 )2 + · · · + (yn − xn )2

= d(y, x)
iv) We shall use the following inequalities for two n tuples in Rn . (a1 , a2 , . . . , an ) and
(b1 , b2 , . . . , bn ) in Rn
n n n
X X  2 1/2 X  2 1/2
|ai bi | ≤ |ai | · |bi | . . . (∗)
i=1 i=1 i=1
Consider
2
d(x, y) = (x1 − y1 )2 + (x2 − y2 )2 + · · · + (xn − yn )2


n
X
= (xi − yi )2
i=1
n
X
= |xi − yi |2
i=1
n
2
X
= (xi − zi ) + (zi − yi )
i=1
n
X
2
≤ |xi − zi | + |zi − yi |
i=1
Xn h i
2 2
= |xi − zi | + 2 (xi − zi ) · (zi − yi ) + |zi − yi |
i=1
Xn n
X n
X
2
= |xi − zi | + 2 (xi − zi ) · |(zi − yi )| + |zi − yi |2
i=1 i=1 i=1
 1/2  1/2
n
X Xn Xn n
X
2 2 2
≤ |xi − zi | + 2  |xi − zi |  |zi − yi | + |zi − yi |2 ( by (∗))
i=1 i=1 i=1 i=1
 1/2  1/2 2
n n
 X X
≤  |xi − zi |2  +  |zi − yi |2  

i=1 i=1

2
= d(x, z) + d(z, y)
So
 2  2
d(x, y) ≤ d(x, z) + d(z, y)
Taking the non negative square root of both sides,
d(x, y) ≤ d(x, z) + d(z, y)
b) Sum metric on Rn :
Let x = (x1 , x2 , x3 , · · · , xn ), y = (y1 , y2 , y3 , · · · , yn ) and z = (z1 , z2 , z3 , · · · , zn ) be in Rn
i) d(x, y) = |x1 − y1 | + |x2 − y2 | + · · · + |xn − yn | ≥ 0
ii)
d(x, y) = 0

S25011: REAL ANALYSIS Page 5

 ⇐⇒ |x1 − y1 | + |x2 − y2 | + · · · + |xn − yn | = 0

⇐⇒ |x1 − y1 | = |x2 − y2 | = · · · = |xn − yn | = 0

⇐⇒ x1 = y1 , x2 = y2 , . . . , xn = yn

⇐⇒ x = y
iii)
d(x, y) = |x1 − y1 | + |x2 − y2 | + · · · + |xn − yn |

= |y1 − x1 | + |y2 − x2 | + · · · + |yn − xn |

= d(y, x)
iv)
d(x, y) = |x1 − y1 | + |x2 − y2 | + · · · + |xn − zn |

= |(x1 − z1 ) + (z1 − y1 )| + |(x2 − z2 ) + (z2 − y2 )| + · · · + |(xn − zn ) + (zn − yn )|

≤ |x1 − z1 | + |z1 − y1 | + |x2 − y2 | + |z2 − y2 | + · · · + |xn − zn | + |zn − yn |

= (|x1 − z1 | + |x2 − z2 | + · · · + |xn − zn |) + (|z1 − y1 | + |z2 − y2 | + · · · + |zn − yn |)

= d(x, z) + d(z, y)
c) Sup metric on Rn :
Let x = (x1 , x2 , x3 , · · · , xn ), y = (y1 , y2 , y3 , · · · , yn ) and z = (z1 , z2 , z3 , · · · , zn ) be in Rn .
i)
d(x, y) = max{|x1 − y1 |, |x2 − y2 |, · · · , |xn − yn |}

≥ |x1 − y1 |

≥0
ii)
d(x, y) = 0

⇐⇒ max{|x1 − y1 |, |x2 − y2 |, · · · , |xn − yn |} = 0

⇐⇒ (|x1 − y1 |, |x2 − y2 |, · · · , |xn − yn |) ≤ max{|x1 − y1 |, |x2 − y2 |, · · · , |xn − yn |} = 0

⇐⇒ |x1 − y1 |, |x2 − y2 |, · · · , |xn − yn | = 0 (∵ |x| ≤ 0 and |x| ≥ 0 =⇒ |x| = 0 =⇒ x = 0)

⇐⇒ x1 = y1 , x2 = y2 , . . . , xn = yn

⇐⇒ x = y
iii)
d(x, y) = max{|x1 − y1 |, |x2 − y2 |, . . . , |xn − yn |}

= max{|y1 − x1 |, |y2 − x2 |, . . . , |yn − xn |}

= d(y, x)

S25011: REAL ANALYSIS Page 6

iv)
d(x, y) = max{|x1 − y1 |, |x2 − y2 |, . . . , |xn − yn |}

= max{|(x1 − z1 ) + (z1 − y1 )|, |(x2 − z2 ) + (z2 − y2 )|, . . . , |(xn − zn ) + (zn − yn )|}

≤ max{|x1 − z1 | + |z1 − y1 |, |x2 − y2 | + |z2 − y2 |, . . . , |xn − zn | + |zn − yn |}

≤ max{|x1 − z1 |, |x2 − z2 |, . . . , |xn − zn |} + max{|z1 − y1 |, |z2 − y2 |, . . . , |zn − yn |}

= d(x, z) + d(y, z)

Problem 1.1.7 (Discrete metric space)Let X be any set. The discrete metric d on
X is defined as 
 0 if x = y
d(x, y) =
 1 if x ̸= y
We verify that d is a metric on X. Let x, y, z ∈ X.
i) d(x, y) = 0 or d(x, y) = 1. Hence d(x, y) ≥ 0 for all x, y ∈ X.
ii) By definition of d, d(x, y) = 0 if and only if x = y.
iii) We consider the following cases.
I Case: x = y.
In this case d(x, y) = 0 = d(y, x).
II Case: x ̸= y.
In this case d(x, y) = 1 = d(y, x).
Thus d(x, y) = d(y, x) in all cases.
iv) If x = y then clearly d(x, y) = 0 ≤ d(x, z) + d(z, y). Let x ̸= y. Then d(x, y) = 1. We
consider the following cases.
I Case: x = z and z ̸= y.
In this case
d(x, z) + d(z, y) = 0 + 1 = 1 = d(x, y)
II Case: x ̸= z and z = y.
In this case
d(x, z) + d(z, y) = 1 + 0 = 1 = d(x, y)
III Case: x ̸= z and z ̸= y.
In this case,
d(x, z) + d(z, y) = 1 + 1 = 2 ≥ 1 = d(x, y)
Thus d(x, y) ≤ d(x, z) + d(z, y) in all possibilities.

Theorem 1.1.8 Let f be continuous real valued function defined on closed and bounded
R1
interval [0, 1] with f (t) ≥ 0 for all t ∈ [0, 1]. Then f (x) dx = 0 if and only if f (t) = 0
0
for every t ∈ [0, 1].

S25011: REAL ANALYSIS Page 7

 R1
Proof: If f (t) = 0 for every t ∈ [0, 1] then clearly f (x) dx = 0.
0
R1
Conversely let f (x) dx = 0. Suppose on contrary that f (t) ̸= 0. Then f (t0 ) > 0 for
0
some t0 ∈ [0, 1]. Since f is continuous at t = t0 , for ϵ = f (t20 ) > 0 there is δ > 0 such that
ϵ
|t − t0 | < δ =⇒ |f (t) − f (t0 )| <
2
This implies that  
ϵ ϵ
t ∈ (t0 − δ, t0 + δ) =⇒ f (t) ∈ f (t0 ) − , f (t0 ) +
2 2
 !  
3 5 f (t0 )
=⇒ t ∈ (t0 − δ, t0 + δ) =⇒ f (t) ∈ f (t0 ), f (t0 ) ∵ϵ=
4 4 2
!
3 5
=⇒ t ∈ (t0 − δ, t0 + δ) =⇒ f (t0 ) < f (t) < f (t0 )
4 4
Now,
Z1 tZ0 +δ tZ0 +δ
3 3
f (t)dt ≥ f (t)dt ≥ f (t0 )dt = f (t0 ) · 2δ > 0
4 4
0 t0 −δ t0 −δ
R1
This is a contradiction. Hence f (t)dt = 0 only when f (t) = 0 for every t ∈ [0, 1].
0


Problem 1.1.9 Let C [0, 1] denote the set of all continuous real valued functions defined


on closed and bounded interval [0, 1]. Show that C [0, 1] is metric space under the metric
d given by
  21
Z1
2
d(f, g) =  f (t) − g(t) dt
 

0
Solution: C [0, 1] = {f |f : [0, 1] → R is continuous}


Let f, g and h be in C [0, 1] .
i)
  12
Z1
2
d(f, g) =  f (t) − g(t) dt ≥ 0
 

0
ii)
d(f, g) = 0
  21
Z1
2
⇐⇒  f (t) − g(t) dt = 0
 

Z1
2
⇐⇒ f (t) − g(t) dt = 0
0
2
⇐⇒ f (t) − g(t) = 0 for every t ∈ [0, 1]

S25011: REAL ANALYSIS Page 8

 ⇐⇒ f (t) − g(t) = 0 for every t ∈ [0, 1]

⇐⇒ f (t) = g(t) for every t ∈ [0, 1]

⇐⇒ f = g
iii) Consider
  12
Z1
2
d(f, g) =  f (t) − g(t) dt
 

0
  12
Z1
2
=  g(t) − f (t) dt
 

= d(g, f )
iv)We shall use Cauchy Schwartz inequality

  21   12
Z1 Z1 Z1
2 2
f (t) · g(t) dt ≤  f (t) dx ·  g(t) dx
   

0 0 0
Consider
d(f, g)
  21
Z1
2
=  f (t) − g(t) dt
 

0
  12
Z1
2
=  f (t) − h(t) + h(t) − g(t) dt
 

0
  12
Z1 Z1 Z1
2 2
=  f (t) − h(t) dt + 2 (f (t) − h(t)) · (h(t) − g(t)) dt + h(t) − g(t) dt
 

0 0 0
  12
  12   21
Z1 Z1 Z1 Z1 
2 2 2 2
≤  f (t) − h(t) dt + 2  f (t) − h(t) dx ·  h(t) − g(t) dx + h(t) − g(t) dt
     
 
0 0 0 0

  21   12
Z1 Z1
2 2
≤  f (t) − h(t) dt +  h(t) − g(t) dt
   

0 0

= d(f, h) + d(h, g)

S25011: REAL ANALYSIS Page 9

Definition 1.1.10 Open ball: Let X be metric space x ∈ X and r > 0. The open ball
B(x, r) with centre at x and radius r > 0 is defined as
B(x, r) = {y ∈ X : d(y, x) < r}
An open ball B(x, r) is a set of those points of X that are at a distance less than r from
x.

Problem 1.1.11 Describe open balls in the usual metric space R.

Solution: Let x ∈ R and r > 0.

B(x, r) = {y ∈ R : d(y, x) < r}

= {y ∈ R : |y − x| < r}

= {y ∈ R : −r < y − x < r}

= {y ∈ R : x − r < y < x + r}

= (x − r, x + r)

Every open ball in the usual metric space R is an open interval.

Problem 1.1.12 Show that every open interval in usual metric space R is an open ball
in R.
a+b b−a
Solution: Let (a, b) be an open interval in R. Take x = 2
and r = 2
. We claim
that B(x, r) = (a, b).
Consider
B(x, r) = {y ∈ R : d(y, x) < r}

= {y ∈ R : |y − x| < r}

= {y ∈ R : −r < y − x < r}

= {y ∈ R : x − r < y < x + r}
a+b b−a a+b b−a
= {y ∈ R : − <y< + }
2 2 2 2
= {y ∈ R : a < y < b}

= (a, b)

Every open interval in R is an open ball in the usual metric space R.

Problem 1.1.13 Describe open balls in the Euclidean metric space R2 .

Solution: Let x ∈ R2 and r > 0.
Consider
B (x, y), r = {(u, v) ∈ R2 |d (u, v), (x, y) < r}



p
= {(u, v) ∈ R2 | (x − u)2 + (y − v)2 < r}

S25011: REAL ANALYSIS Page 10

 = Interior of circle with centre at (x, y) and radius r.

Problem 1.1.14 Describe the following open balls in discrete metric space X.

i) B(x, 12 )

ii) B(x, 43 )

iii) B(x, 1)

iv) B(x, 3)

v) B(x, r) where 0 < r ≤ 1

vi) B(x, r) where r > 1

Solution: A discrete metric space d on 
X is defined as ;
 0 if x = y
d(x, y) =
 1 if x ̸= y

i)  
1 1
B x, = {y ∈ X : d(y, x) < }
2 2
= {y ∈ X : d(x, y) = 0}

= {y ∈ X : y = x}

= {x}

ii)  
3 3
B x, = {y ∈ X : d(y, x) < }
4 4
= {y ∈ X : d(x, y) = 0}

= {y ∈ X : y = x}

= {x}

iii)
B(x, 1) = {y ∈ X : d(y, x) < 1}

= {y ∈ X : d(y, x) = 0}

= {y ∈ X : y = x}

= {x}

iv)
B(x, 3) = {y ∈ X : d(y, x) < 3}

= {y ∈ X : d(y, x) = 1 or d(y, x) = 0}

S25011: REAL ANALYSIS Page 11

 =X

v)
B(x, r) = {y ∈ X : d(y, x) < r}

= {y ∈ X : d(y, x) = 0} (∵ r ≤ 1)

= {x}

vi)
B(x, r) = {y ∈ X : d(y, x) < r}

=X (∵ d(y, x) ≤ 1 for all x, y as r > 1)

Definition 1.1.15 Closed ball: Let X be metric space, x ∈ X and r > 0. The closed
ball B[x, r] is defined as
B[x, r] = {y ∈ X : d(y, x) ≤ r}
A closed ball B[x, r] is a set of those points of X that are at a distance less than or equal
to r from x.

Problem 1.1.16 Describe closed balls in the usual metric space R.

Solution: Let x ∈ R and r > 0.

B[x, r] = {y ∈ R : d(y, x) ≤ r}

= {y ∈ R : |y − x| ≤ r}

= {y ∈ R : −r ≤ y − x ≤ r}

= {y ∈ R : x − r ≤ y ≤ x + r}

= [x − r, x + r]

Every closed ball in the usual metric space R is a closed interval.

Problem 1.1.17 Show that every closed interval in usual metric space R is closed ball in
R.
a+b b−a
Solution: Let [a, b] be a closed interval in R. Take x = 2
and r = 2
. We claim
that B[x, r] = [a, b].
Consider
B[x, r] = {y ∈ R : d(y, x) ≤ r}

= {y ∈ R : |y − x| ≤ r}

= {y ∈ R : −r ≤ y − x ≤ r}

= {y ∈ R : x − r ≤ y ≤ x + r}

S25011: REAL ANALYSIS Page 12

 a+b b−a a+b b−a
= {y ∈ R : − ≤y≤ + }
2 2 2 2
= {y ∈ R : a ≤ y ≤ b}

= [a, b]

Every closed interval in R is a closed ball in the usual metric space R.

Problem 1.1.18 Describe closed balls in the Euclidean metric spaces R2 and R3 .
Solution:
i. Let (x, y) ∈ R2 and r > 0.

B[(x, y), r] = {(u, v) : d((u, v), (x, y)) ≤ r}

p
= {(u, v) : (x − u)2 + (y − v)2 ≤ r}

= closed disc with centre at (x, y) and radius r

ii. Let (x, y, z) ∈ R3 and r > 0.

B[(x, y, z), r] = {(u, v, w) : d((u, v, w), (x, y, z)) ≤ r}

p
= {(u, v) : (x − u)2 + (y − v)2 + (z − w)2 ≤ r}

= closed sphere with centre at (x, y, z) and radius r

Solved Problems 01

Problem 1.1.19 Prove that < C, d > is a metric space.

Solution: Let z1 = x1 + iy1 , z2 = x2 + iy2 and z3 = x3 + iy3 be in C where
x1 , y1 , x2 , y2 , x3 , y3 ∈ R.
i)
d(z1 , z2 ) = |z1 − z2 |

= (x1 − x2 ) + i(y1 − y2 )
p
= (x1 − x2 )2 + (y1 − y2 )2 ≥ 0
ii)
d(z1 , z2 ) = 0
p
⇐⇒ (x1 − x2 )2 + (y1 − y2 )2 = 0

⇐⇒ (x1 − x2 )2 + (y1 − y2 )2 = 0

⇐⇒ x1 − x2 = y1 − y2 = 0

⇐⇒ x1 = x2 and y1 = y2

⇐⇒ z1 = z2

S25011: REAL ANALYSIS Page 13

iii)
d(z1 , z2 ) = |z1 − z2 |

= (x1 − x2 ) + i(y1 − y2 )
p
= (x1 − x2 )2 + (y1 − y2 )2
p
= (x2 − x1 )2 + (y2 − y1 )2

= |z2 − z1 |

= d(z2 , z1 )
iv) d(z1 , z2 ) ≤ d(z1 , z3 ) + d(z3 , z2 ) for all z1 , z2 , z3 ∈ X First we prove “|u + v| ≤ |u| + |v|
for all u, v ∈ C”.
Consider
|u + v|2 = (u + v) · (u + v)

= (u + v) · (u + v)

= (u · u) + u · v + u · v + (v · v)

= |u|2 + u · v + u.v + |v|2

≤ |u|2 + 2|uv| + |v|2

= |u|2 + 2|u||v| + |v|2

= (|u| + |v|)2
Thus |u + v|2 ≤ (|u| + |v|)2 .Taking non-negative square root of both sides
|u + v| ≤ |u| + |v|
This proves that |u + v| ≤ |u| + |v| for all u, v ∈ C. Consider
d(z1 , z2 ) = |z1 − z2 |

= |z1 − z3 + z3 − z2 |

= |(z1 − z3 ) + (z3 − z2 )|

≤ |z1 − z3 | + |z3 − z2 |

= d(z1 − z3 ) + d(z3 − z2 )

Problem 1.1.20 (Product metric space) Let < X, d1 > and < Y, d2 > be metric
spaces. Prove that the cartesian product X × Y is a metric space under the metric given
by
d((x1 , y1 ), (x2 , y2 )) = max{d1 (x1 , x2 ), d2 (y1 , y2 )}
Solution: Let x = (x1 , y1 ), y = (x2 , y2 ) and z = (x3 , y3 ) be in X × Y.
i)
d(x, y) = max{d1 (x1 , x2 ), d2 (y1 , y2 )}

S25011: REAL ANALYSIS Page 14

 ≥ d1 (x1 , x2 ) (∵ max{a, b} ≥ a, b)

≥0
ii)
d(x, y) = 0

⇐⇒ max{d1 (x1 , x2 ), d2 (y1 , y2 )} = 0

⇐⇒ d1 (x1 , x2 ) = d2 (y1 , y2 ) = 0

⇐⇒ x1 = x2 and y1 = y2

⇐⇒ x = y
iii)
d(x, y) = max{d1 (x1 , x2 ), d2 (y1 , y2 )}

= max{d1 (x2 , x1 ), d2 (y2 , y1 )}

= d(y, x)
iv)d(x, y) = max{d1 (x1 , x2 ), d2 (y1 , y2 )}. So d(x, y) = d1 (x1 , x2 ) or d(x, y) = d2 (y1 , y2 ).
Without loss of generality, let d(x, y) = d1 (x1 , x2 ). Then
d(x, y) = d1 (x1 , x2 )

≤ d1 (x1 , x3 ) + d1 (x3 , x2 )

≤ max{d1 (x1 , x3 ), d2 (y1 , y3 )} + max{d1 (x3 , x2 ), d2 (y3 , y2 )}

= d(x, z) + d(z, y)

Self-Test 01

Problem 1.1.21 Which of the following functions d : R × R → R is not a metric on R?

A. d(x, y) = |x − y|
|x−y|
B. d(x, y) = 1+|x−y|

 1 if x ̸= y
C. d(x, y) =
 0 if x = y

D. d(x, y) = x2 − y 2

Problem 1.1.22 Which of the following functions is a metric?

A. d(x, y) = min{|x|, |y|}, ∀x, y ∈ R

B. d(x, y) = |x2 − y 2 |, ∀x, y ∈ R

C. d(x, y) = |x3 − y 3 |, ∀x, y ∈ C

S25011: REAL ANALYSIS Page 15

 D. d(x, y) = | x1 − y1 |, ∀x, y ∈ R − {0}

Short Answer Questions 01

Problem 1.1.23 In a metric space, < X, d >, show that |d(x, y) − d(z, y)| ≤ d(x, y) for
all x, y, z in X.
Solution: Let x, y, z in X. Consider
d(z, y) ≤ d(z, x) + d(x, y)

=⇒ d(z, y) ≤ d(x, z) + d(x, y) (∵ d(x, z) = d(z, x))

=⇒ −d(x, y) ≤ d(x, z) − d(z, y) . . . (I)

Also
d(x, z) ≤ d(x, y) + d(y, z)

=⇒ d(x, z) ≤ d(x, y) + d(z, y) (∵ d(z, y) = d(y, z))

=⇒ d(x, z) − d(z, y) ≤ d(x, y) . . . (II)

From (I) and (II)
−d(x, y) ≤ d(x, z) − d(z, y) ≤ d(x, y)
which implies that |d(x, y) − d(z, y)| ≤ d(x, y).

Problem 1.1.24 Let, < X, d > be a metric space and the function δ : X × X → R be
defined as δ(x, y) = min{1, d(x, y)}. Prove that δ is a metric on X.
Solution: Let x, y, z ∈ X.
i) If d(x, y) < 1 then δ(x, y) = d(x, y) ≥ 0. If d(x, y) ≥ 1 then δ(x, y) = 1 ≥ 0.
ii)
δ(x, y) = 0

⇐⇒ min{1, d(x, y)} = 0

⇐⇒ d(x, y) = 0

⇐⇒ x = y
iii)
δ(x, y) = min{1, d(x, y)}

= min{1, d(y, x)} (∵ d(x, y) = d(y, x))

= δ(y, x)
iv) We consider the following cases.
I Case: d(x, z) < 1 and d(z, y) < 1.
δ(x, y) = min{1, d(x, y)}

≤ d(x, y)

≤ d(x, z) + d(z, y) (by triangle inequality)

S25011: REAL ANALYSIS Page 16

 = min{1, d(x, z)} + min{1, d(z, y)} (∵ −d(x, y), d(z, y) < 1)

= δ(x, z) + δ(z, y)
II Case: d(x, z) ≥ 1 or d(z, y) ≥ 1
Without of loss of generality, let d(x, z) ≥ 1. Consider
δ(x, y) = min{1, d(x, y)} ≤ 1

= δ(x, z) (δ(x, z) = 1 as d(x, z) ≥ 1)

≤ δ(x, z) + δ(z, y) (∵ δ(z, y) ≥ 0)

Thus δ(x, y) ≤ δ(x, z) + δ(z, y) in all cases.

Summary:

• A metric is generalisation of the distance on real line R or on the Euclidean

plane R2 . The distance on real line is the absolute value of difference between
any two real numbers (points) on it. The distance between any two points on the
Euclidean plane R2 is given by the distance formula. These distance functions
satisfy the following four properties.
i) The distance between any two points is non-negative
ii) The distance between two points is 0 if and only if they coincide.
iii) The distance is symmetric.
iv) The distance between any two points is always less equal to the sum of
distances of a path joining these two points when one or more points is introduced
between these two points.
d is metric on set X if it satisfies the above four properties.

• Every subset of metric space X is again a metric space under the metric induced
from X. This is why we call every subset of metric space as subspace.

• Every normed linear space is a metric space. If < N, +, ·, || · || > is a normed

space then < N, d > is a metric space where d : N × N → R is given by
d(x, y) = ||x − y||.

• There are three standard norms on Rn which give three equivalent metrics on Rn :
Euclidean metric, Sum metric and Sup metric. They generate same topologies
on Rn . We generally study Rn with the Euclidean metric defined on it.

• A discrete metric d can be defined on any set X. It is defined by the property that
the distance between any two distinct points is 1 and the distance between two
points is 0 when they coincide. The discrete metric has many peculiar properties.

S25011: REAL ANALYSIS Page 17

 • If < X, d1 > and < Y, d2 > are metric spaces then the Cartesian product X ×Y is
also metric space under the metric given by d(x, y) = max{d1 (x1 , x2 ), d2 (y1 , y2 )}
where x = (x1 , y1 ) and y = (x2 , y2 ). There are other metrics which can be defined
on X × Y . Also the concept of product metric space can be generalised to more
than two sets.

END OF UNIT EXERCISES

Problem 1.1.25 Let < X, d > be a metric space and the function δ : X × X → R be
d(x,y)
defined as δ(x, y) = 1+d(x,y)
. Prove that δ is a metric on X.

Problem 1.1.26 Let the function d : (0, ∞) × (0, ∞) → R be defined as d(x, y) =

1 1
x
− y
.Prove that d is a metric on (0, ∞).
√
Problem 1.1.27 Let the function d : R × R → R be defined as d(x, y) = x − y.Prove
that d is a metric on R.

Keywords: Metric Space, Discrete Metric Space, Normed Linear Space, Product Metric
Space.

REFERENCES

(1) Section 4.2, Methods of Real Analysis by Richard R. Goldberg, 2nd edition,
Oxford & IBH Publishing Co. Pvt Ltd

(2) Section 1.1, Topology of Metric Spaces by S. Kumaresan, 2nd edition, Narosa
Publishing House

(3) Chapter 2, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,

Mc Graw HillInc.,USA.

(4) Chapter 3, Real Analysis, by N.L. Carothers, 1st edition, Cambridge University
Press.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 18

 Unit 01-02: Open and Closed Sets

Learning Objectives:
After successful completion of this unit, you will be able to

i. Get the relationship between open balls and open sets.

ii. State the properties of open sets in metric spaces.

iii. State the properties of closed sets in metric spaces.

iv. Establish equivalence of metrics defined on the same set.

v. Know the structure of open sets in R.

Introduction: Using the definition of metric, we can introduce the concept of open balls
and subsequently that of open sets. The open sets in R has specific structure. Any open
set in R is countable union of disjoint open intervals. Arbitrary union of open sets is open
and finite intersection of open sets is open in any metric space. Arbitrary intersection of
closed sets is closed and finite union of closed sets is closed in any metric space.

Definition 1.2.1 Open sets: Let X be a metric space. A subset U of X is said to be

open if for every x ∈ U there is r > 0 such that
B(x, r) ⊆ U
A subset U of metric space X is open if for each point x of U there is an open ball
containing x which is entirely contained in U .

Theorem 1.2.2 Every open ball in a metric space is an open set.

Proof: Let X be a metric space, x ∈ X and r > 0. Consider the open ball B(x, r).
Let y ∈ B(x, r). Select a real number s such that 0 < s < r − d(x, y). We claim that
B(y, s) ⊆ B(x, r). Let z ∈ B(y, s). This implies that d(z, y) < s. Consider
d(z, x) ≤ d(z, y) + d(y, x)

< s + d(y, x)

< r − d(x, y) + d(y, x)

=r
Therefore z ∈ B(x, r). Thus z ∈ B(y, s) implies that z ∈ B(x, r). This proves the
claim that B(y, s) ⊆ B(x, r). Thus for every y ∈ B(x, r) there is s > 0 such that
B(y, s) ⊆ B(x, r). Hence B(x, r) is open in X.

Problem 1.2.3 Is the set Qc of irrationals open in R?

Solution: Let x ∈ Qc and r > 0. Then the open ball B(x, r) = (x − r, x + r) contains
rationals. Therefore B(x, r) ̸⊆ Qc . Since this is true for any r > 0, Qc is not open in R.

S25011: REAL ANALYSIS Page 19

Theorem 1.2.4 Let X be a metric space. Then the following statements hold
i) ϕ and X are open in X.
ii) An arbitrary union of open sets in X is open in X.
iii) A finite intersection of open sets in X is open in X.
Proof:
i) Consider the statement
“ For every x ∈ ϕ there exists r > 0 such that B(x, r) ⊆ ϕ” . . . (I)
The statement (I) is conditional statement where the proposition x ∈ ϕ is false. Hence it
is vacuously true. This means that the empty set ϕ is open in X.
To prove that X is open in X, take x ∈ X and r > 0. Then B(x, r) = {y ∈ X : d(y, x) <
r} is clearly a subset of X. So for every x ∈ X and r > 0, B(x, r) ⊆ X. This proves that
X is open in X.
ii) Let {Ui : i ∈ I} be an arbitrary collection of open sets in X. We want to prove that
S
the set U = Ui is open in X. Let x ∈ U . Then x ∈ Ui for some i ∈ I. Since Ui is open
i∈I
in X, there is r > 0 such that
B(x, r) ⊆ Ui ⊆ U
So for every x ∈ U , there is r > 0 such that B(x, r) ⊆ U . This proves that U is open in
X.
iii) Let U1 , U2 , . . . , Un be open sets in X. Take U = U1 ∩ U2 ∩ . . . ∩ Un . We want to prove
that U is open in X. Let x ∈ U . Then x ∈ Ui for i = 1, 2, 3 . . . , n. Since each Ui is open
in X, there is ri > 0 such that B(x, ri ) ⊆ Ui for i = 1, 2, 3 . . . , n. Take r > 0 such that
0 < r < min{r1 , r2 , . . . , rn }. We claim that B(x, r) ⊆ U . Let z ∈ B(x, r). Then
d(x, z) < r

=⇒ d(x, z) < ri for i = 1, 2, . . . , n (∵ r < ri as 0 < r < min{r1 , r2 , . . . , rn } ≤ ri )

=⇒ z ∈ B(x, ri ) for i = 1, 2, . . . , n

=⇒ z ∈ Ui for i = 1, 2, . . . , n (∵ B(x, r) ⊆ Ui for i = 1, 2, . . . , n)

=⇒ z ∈ U
Thus z ∈ B(x, r) implies that z ∈ U .This proves our claim that B(x, r) ⊆ U . We proved
that for every x ∈ U there is r > 0 such that B(x, r) ⊆ U . So U is open in X.

Theorem 1.2.5 An open set in usual metric space R is union of countable number of
pairwise disjoint open intervals.
Proof: Let U open set in R. We want to write U as a union of countable number of
pairwise disjoint open intervals. Take x ∈ U . Since U is open in R, there is r > 0 such
that B(x, r) = (x − r, x + r) ⊆ U . Take Jx as union of all such open intervals containing
S
x. Precisely Jx = Ix where Ix is an open interval containing x and contained in U . It

S25011: REAL ANALYSIS Page 20

is clear that Jx is the largest open interval containing x and contained in U .
Let Jx = (a, b). For x and y in U , we claim that Jx = Jy or Jx ∩ Jy = ϕ.
If Jx = Jy then we are done. Let Jx ̸= Jy . Suppose on contrary Jx ∩ Jy ̸= ϕ. Take
z ∈ Jx ∩ Jy . Then Jx and Jy are open intervals containing common point z. Hence Jx
and Jy are open interval containing x. Also Jx ∪ Jy is contained in U as Jx , Jy ⊆ U .
So Jx ∪ Jy is open interval containing x and contained in U . Since Jx is largest interval
Jx ∪ Jy ⊆ Jx . Also Jx ⊆ Jx ∪ Jy . Therefore Jx ∪ Jy = Jx . Similarly Jx ∪ Jy = Jy . This
gives Jx = Jy which is a contradiction. Hence Jx ∩ Jy = ϕ. This proves our claim that
Jx = Jy or Jx ∩ Jy = ϕ.
Consider a class τ of all disjoint intervals Jx whose union is U . From each of these disjoint
interval choose exactly one rational say rx . Let τ ′ be collection of these rationals. Then
[
U= Jx . . . (I)
rx ∈τ ′
Since Jx = Jy or Jx ∩ Jy = ϕ, rx = ry only when x = y. As rationals are countably many,
the union in (I) is countable. Thus we have written U as countable union of pairwise
disjoint open intervals.

Definition 1.2.6 Closed sets: A subset F of metric space X is called closed set in X
if the complement U = X − F of F in X is open in X.

Theorem 1.2.7 Let X be a metric space. Then the following statements hold
i) ϕ and X are closed in X.
ii) Arbitrary intersection of closed sets in X is closed in X.
iii) Finite union of closed sets in X is closed in X.
Proof:
i) Consider ϕc = X − ϕ = X and X c = X − X = ϕ. Since X and ϕ are open in X, ϕ and
X are closed in X.
ii) Let τ = {Fi : i ∈ I} be an arbitrary collection of closed sets in X. We want to prove
T
that the set F = Fi is closed in X. Let U = X − F . Note that U is complement of F
i∈I
in X.

\ [
U =X −F =X − Fi = (X − Fi )
i∈I
Each X − Fi is open in X because each Fi is closed in X. As the arbitrary union of open
sets is open, U is open in X. Hence F is closed in X.
iii) Let F1 , F2 , . . . , Fn be closed sets in X. We want to prove that F = F1 ∪ F2 ∪ · · · ∪ Fn
is closed in X. To prove this, we prove that the complement U = X − F of F in X is
open in X. Consider
U =X −F

S25011: REAL ANALYSIS Page 21

 = X − (F1 ∪ F2 ∪ · · · ∪ Fn )

= (X − F1 ) ∩ (X − F2 ) ∩ · · · ∩ (X − Fn )
The sets (X − F1 ), (X − F2 ), . . . , (X − Fn ) are open in X as F1 , F2 , . . . , Fn are closed in
X. Also the finite intersection of open set is open. It follows that U is open in X. Hence
F is closed in X.

Definition 1.2.8 Equivalent metrics: Let d1 , d2 be metrics on X. If there are constants

c and c′ greater than 0 such that
cd1 (x, y) ≤ d2 (x, y) ≤ c′ d1 (x, y)
for all x, y in X then metrics d1 and d2 are called equivalent metrics on X.
If d1 and d2 are equivalent metrics on X then a subset U of X is open in the metric
space < X, d1 > if and only if it is open in the metric space < X, d2 >. The same is true
for closed sets. As such collection of open (closed) sets in the metric space < X, d1 >
equals to the collection of open (closed) sets in the metric space < X, d2 >. This means
that < X, d1 > and < X, d2 > are topologically equivalent when d1 and d2 are equivalent
metrics on X. All the topological properties of these two structures are same. We ask
the reader to prove the following assertions.
i) If d1 and d2 are equivalent metrics on X then a subset U of X is open in the metric
space < X, d1 > if and only if it is open in the metric space < X, d2 >.
ii) The relation of equivalent metrics on the set X is an equivalence relation.
iii) The Euclidean metric, the sum metric and the sup metric on Rn are equivalent metrics.

The last assertion implies that Rn is topologically same with respect to the Euclidean,
the sum and the sup metric on it. It is thus sufficient to study Rn with any one of these
metrics. Generally we study Rn with the Euclidean metric on it.

Theorem 1.2.9 Let Y be a subspace of a metric space X. A subset A of Y is open in

Y if and only if there is an open set U in X such that A = U ∩ Y .
Proof:
I Part: Let A = U ∩ Y where U is an open subset of X . We want to prove that A is
open in Y . Take x ∈ A. Then x ∈ U as A = U ∩ Y . Since U is open in X, there is an
open ball BX (x, r) ⊆ U . The subscripts X and Y are used to distinguish open balls in
X and Y respectively.
Now,
BY (x, r) = BX (x, r) ∩ Y ⊆ U ∩ Y ⊆ A
Thus for every x ∈ A, there is r > 0 such that BY (x, r) ⊆ A. This proves that A is open
in Y .
II Part: Let A be open in Y . We want to find some open subset U of X such that

S25011: REAL ANALYSIS Page 22

A = U ∩ Y . Take x ∈ A. Since A is open in Y , there is r > 0 such that BY (x, rx ) ⊆ A.
Take
[
U= BX (x, rx )
x∈A
where BY (x, rx ) ⊆ A. Then U is open in X as arbitrary union of open sets is open.
Finally, we prove that A = U ∩ Y
.Consider 
[
U ∩Y = BX (x, rx ) ∩ Y
x∈A
[
= [BX (x, rx ) ∩ Y ]
x∈A
[
= BY (x, rx )
x∈A

=A (∵ BY (x, rx ) ⊆ A)

Theorem 1.2.10 Let Y be a subspace of a metric space X. A subset B of Y is closed

in Y if and only if there is a closed set F in X such that B = F ∩ Y .
Proof: We shall use the following theorem:
“ Let Y be a subspace of a metric space X. A subset A of Y is open Y if and only if
there is an open set U in X such that A = U ∩ Y .” . . . (I)
I Part: Let F be a closed subset of X such that B = F ∩ Y . We want to prove that B
is closed in Y . Take U = X − F and A = Y − B. As F is closed in X, U is open in X.
Consider
A=Y −B

= Y − (F ∩ Y )

= (X − F ) ∩ Y (∵ F, Y ⊆ X)

=U ∩Y
Since U is open in X, by (I) A is open in Y . Therefore the complement B of A in Y is
closed.
II Part: Let B be a closed subset of Y . We want to find some closed subset F of X
such that B = F ∩ Y . Since B is closed in Y , the complement A = Y − B is open in Y .
Then by (I) there is an open set U in X such that A = U ∩ Y . Take F = X − U . Since
U is open in X, F is closed in X. Finally we prove that B = F ∩ Y . Consider
B =Y −A

= Y − (U ∩ Y )

= (X − U ) ∩ Y (∵ U, Y ⊆ X)

=F ∩Y

S25011: REAL ANALYSIS Page 23

Definition 1.2.11 Interior point: Let A be a subset of metric space X. A point x ∈ X
is called interior point of A if there is r > 0 such that B(x, r) ⊆ A. The set of all interior
points of A is denoted by A◦ or Int(A).

Definition 1.2.12 Exterior point: Let A be a subset of metric space X. A point x ∈ X

is called exterior point of A if there is r > 0 such that B(x, r) ⊆ (X − A). The set of all
exterior points of A is denoted by Ext(A).

Definition 1.2.13 Boundary point: Let A be a subset of metric space X. A point

x ∈ X is called boundary point of A if B(x, r) ∩ A ̸= ϕ and B(x, r) ∩ (X − A) ̸= ϕ for
every r > 0. The set of all boundary points of A is denoted by ∂(A) or bdr(A).

Definition 1.2.14 Limit point: Let A be a subset of metric space X. A point x ∈ X

is called limit point of A if B(x, r) ∩ A ̸= ϕ for every r > 0. The set of all limit points of
A is called closure of A and it is denoted by A or lim(A).

Definition 1.2.15 Cluster point: Let A be a subset of metric space X. A point x ∈ X

is called cluster point of A if B(x, r) − {x} ∩ A ̸= ϕ for every r > 0.

Theorem 1.2.16 Every closed ball in metric space is closed.

Proof: Let X be a metric space, a ∈ X and r > 0. Consider a closed ball

B[a, r] = {x ∈ X : d(x, a) ≤ r}
To prove that B[x, r] is closed, we prove that its complement is open in X. Let U be the
complement of B[x, r] in X. Then
U = X − B[a, r]

= {x ∈ X|d(x, a) > r}
and so U = {x ∈ X|d(x, a) > r}. We want to prove that U is open in X. For that let
x ∈ U and select a real number s such that
0 < s < d(x, a) − r . . . (I)
We claim that B(x, s) ⊆ U . Let z ∈ B(x, s). Then
d(z, x) < s . . . (II)
By triangle inequality
d(x, a) ≤ d(x, z) + d(z, a) = d(z, x) + d(z, a)
Therefore
d(x, a) − d(z, x) ≤ d(z, a) . . . (III)
For z ∈ B(x, s), consider the following argument.
d(x, z) < s

=⇒ −d(x, z) > −s

S25011: REAL ANALYSIS Page 24

 =⇒ d(x, a) − d(x, z) > d(x, a) − s

=⇒ d(x, a) − d(x, z) > d(x, a) + r − d(x, a)

=⇒ d(x, a) − d(x, z) > r . . . (IV )

From (III) and (IV), we have d(z, a) > r and so z ∈ U . Thus z ∈ B(x, s) implies that
z ∈ U . This proves our claim that B(x, s) ⊆ U . We proved that for every x ∈ U , there
is s > 0 such that B(x, s) ⊆ U . So U is open in X implying that its complement B[a, r]
is closed in X.

Theorem 1.2.17 Let X be a metric space and A be subset of X. Then the following
statements are true.
i) A◦ is the largest open set contained in A.
ii) A is open if and only if A = A◦ .
Proof:
i) I Part: A◦ is open set.
A◦ = {x ∈ A| there is r > 0 such that B(x, r) ⊆ A}
Let x ∈ A◦ . Then by the definition of A◦ , there is r > 0 such that B(x, r) ⊆ A. This is
true for every x ∈ A◦ . So by the definition of open set, A◦ is open.
II Part: A◦ is the largest open contained in A.
By I part, A◦ is open. Let U be any open subset contained in A i.e. U is open and
U ⊆ A. Let x ∈ U . Since U is open, there is r > 0 such that B(x, r) ⊆ U ⊆ A. By
definition of A◦ , x ∈ A◦ . This proves that U ⊆ A◦ . Thus U is open is X and U ⊆ A
implies U is contained in A◦ . This means that any open set contained in A is contained
in A◦ . In other words, A◦ is the largest open set contained in A.
ii) I Part: Let A be open in X. We want to prove that A = A◦ .
A◦ = {x ∈ A : there is r > 0 such that B(x, r) ⊆ A}
So clearly
A◦ ⊆ A . . . (I)
Conversely, let x ∈ A. Since A is open, there is r > 0 such that B(x, r) ⊆ A. Again by
definition of A◦ , x ∈ A◦ . Thus x ∈ A implies x ∈ A◦ and hence
A ⊆ A◦ . . . (II)
From (I) and (II) we have A = A◦
II Part: Let A = A◦ . We want to prove that A is open. Let x ∈ A. Then x ∈ A◦ . By
definition of A◦ , there is r > 0 such that B(x, r) ⊆ A. Thus for every x ∈ A, there is
r > 0 such that B(x, r) ⊆ A. Hence A is open in X.

Solved Problems 01

S25011: REAL ANALYSIS Page 25

Problem 1.2.18 Show that every set in discrete metric space is open.
Solution: Let < X, d > be discrete metric space and U be subset of X. Let x ∈ U .
Take a real number r such that 0 < r ≤ 1. Then B(x, r) = {x} ⊆ U . Thus for every
x ∈ U there is r > 0 such that B(x, r) ⊆ U . This proves that U is an open subset of X.
We proved that every subset of discrete metric space is open.

Every set in discrete metric space is open.

Problem 1.2.19 Let F be a finite subset of metric space X. Show that F is closed in X.
Solution: Let F = {x1 , x2 , . . . , xn } and U = X − F . Take x ∈ U . Select a real number
r such that
0 < r < min{d(x, x1 ), d(x, x2 ), d(x, x3 ), . . . , d(x, xn )}
We claim that B(x, r) ⊆ U . Suppose on contrary that B(x, r) contains one of the points
of F = {x1 , x2 , . . . , xn } say xi . Then d(xi , x) < r and
d(xi , x) < r < min{d(x, x1 ), d(x, x2 ), d(x, x3 ), . . . , d(x, xn )} ≤ d(xi , x)
This implies that d(xi , x) < d(xi , x) which is a contradiction. Hence B(x, r) contains no
point of F implying that B(x, r) ⊆ U . Thus for every x ∈ U there is r > 0 such that
B(x, r) ⊆ U . This proves that U is an open subset of X. As U is complement of F in
X, F is closed in X.
Self-Test 01
1

T
Problem 1.2.20 If Gn = 0, 1 + n
for n ∈ N, then Gn is

A. closed

B. open

C. both open and closed

D. neither open nor closed

∞
Problem 1.2.21 In R, let Fn = − n1 , n1 , ∀n ∈ N. Then

T
Fn is :
n=1

A. {0}

B. ϕ

C. both open and closed

D. neither open nor closed

Problem 1.2.22 Let A be a closed subset of R, A ̸= ϕ, A ̸= R. Then A is

A. the closure of interior of A

B. a countable set

S25011: REAL ANALYSIS Page 26

 C. a compact set

D. not open

Problem 1.2.23 The segment (a, b) of X axis in R2 is

A. neither open nor closed

B. open

C. closed

D. both open and closed

Problem 1.2.24 The segment [a, b] of X axis in R2 is

A. neither open nor closed

B. open

C. closed

D. both open and closed

Problem 1.2.25 An open set in R is

A. an interval

B. finite union of open intervals

C. countable union of disjoint open intervals in R

D. intersection of open intervals

Short Answer Questions 01

Problem 1.2.26 Show that every open interval in usual metric space R is open.
Solution: Let (a, b) be an open interval in R. Let x ∈ (a, b). Take r such that
0 < r < min{x − a, b − x}. We claim that
B(x, r) = (x − r, x + r) ⊆ (a, b) . . . (I)
Let y ∈ (x − r, x + r). Then x − r < y < x + r. Also r < (x − a) and r < (b − x). These
imply that
x + (a − x) < x − r < y < x + r < x + (b − x)

=⇒ a < x − r < y < x + r < x + (b − x)

=⇒ a < x − r < y < x + r < b

=⇒ a < y < b

S25011: REAL ANALYSIS Page 27

 =⇒ y ∈ (a, b)
Thus y ∈ (x − r, x + r) =⇒ y ∈ (a, b). This proves that the claim (I).

Problem 1.2.27 Is the set Q of rationals open in R?

Solution: Let x ∈ Q and r > 0. Then the open ball B(x, r) = (x − r, x + r) contains
irrationals. Therefore B(x, r) ̸⊆ Q. Since this is true for any r > 0, Q is not open in R.

Summary:

• If X is a metric space then the open ball B(x, r) with the center at point x of X
and radius r > 0 is the set of all those points of X which are at a distance less
than r from the point x.

• Open balls in the Euclidean metric space R are open intervals in R. The converse
is also true.

• Let < X, d > be a discrete metric

 space. Then
 {x} if r ≤ 1
B(x, r) =
 X if r > 1

• If X is a metric space then the closed ball B[x, r] with the center at point x of
X and radius r > 0 is the set of all those points of X which are at a distance
less than or equal to r from the point x.

• Closed balls in the Euclidean metric space R are closed intervals in R. The
converse is also true.

• Let < X, d > be a discrete metric

 space. Then
 {x} if r < 1
B[x, r] =
 X if r ≥ 1

• A subset U of metric space X is open if for each point x of U there is an open

ball containing x which is entirely contained in U .

• Every open ball in a metric space is an open set. In particular, every open
interval in the Euclidean metric space R is open in R.

• An open set in the Euclidean metric space R is union of countable number of

pairwise disjoint open intervals in R.

• Every set in a discrete metric space is open.

• In any metric space X, ϕ and X are open in X. An arbitrary union of open sets
in X is open in X. A finite intersection of open sets in X is open in X.

S25011: REAL ANALYSIS Page 28

 • A subset F of metric space X is closed if its complement in X is open in X.

• Every closed ball in a metric space is a closed set. In particular, every closed
interval in the Euclidean metric space R is closed in R.

• A finite set in any metric space is closed.

• Every set in a discrete metric space is closed.

• In any metric space X, ϕ and X are closed in X. An arbitrary intersection of

closed sets in X is closed in X. A finite union of closed sets in X is closed in X.

• Let A be a subset of metric space X. A point x in X is an interior point of A

if there is some open ball containing x which is contained in A. The set of all
interior points of A is denoted by A◦ . It is the largest open set contained in A.
The set A is open if and only if A = A◦ .

• Let A be a subset of metric space X. A point x in X is a boundary point of A if

every open ball containing x contains a point of A as well as a point of X − A. In
other words, x is boundary point of A if every open ball containing x intersects
A as well as its complement in X.

• Let A be a subset of metric space X. A point x in X is a limit point of A if

either x is interior point of A or x is boundary point of A. Thus a limit point of
A is either in interior of A or at the boundary of A. The set of all limit points
of A is denoted by A. It is the smallest closed set containing A. The set A is
closed if and only if A = A.

• A subset D of metric space X is dense in X if every open ball in X contains a

point of D. In other words, D is dense in X if every point of X is a limit point
(interior or boundary point) of D.

• A subset A of metric space X is bounded if A can be enclosed in some open ball

in X.

End of Unit Exercises

Problem 1.2.28 Let X be a metric space and f : X → X be continuous. Prove that the
set E = {x ∈ X : f (x) = x} is closed in X.

Problem 1.2.29 Let A be a finite subset of metric space X. Show that X − A is open
in X.

S25011: REAL ANALYSIS Page 29

Problem 1.2.30 Let Y be a subspace of a metric space X. Let y ∈ Y and r > 0. Prove
that
BY (y, r) = BX (y, r) ∩ Y
Here BX (y, r) and BY (y, r) denote open balls in metric spaces X and Y respectively, with
center y and radius r.

Keywords: Open Balls, Closed Balls, Interior Points, Exterior Points, Boundary Points,
Limit Points, Open Sets, Closed Sets.

REFERENCES

(1) Sections 5.4 and 5.5, Methods of Real Analysis by Richard R. Goldberg, 2nd
edition, Oxford & IBH Publishing Co. Pvt Ltd

(2) Sections 1.2, 2.1 to 2.7, Topology of Metric Spaces by S. Kumaresan, 2nd edition,
Narosa Publishing House

(3) Chapter 2, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,

Mc Graw HillInc.,USA.

(4) Chapter 4, Real Analysis, by N.L. Carothers, 1st edition, Cambridge University
Press.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 30

 Unit 01-03: Sequences in Metric Spaces

Learning Objectives:
After successful completion of this unit, you will be able to

i. Understand the convergence of sequences in metric spaces.

ii. Define closed sets in terms of limit points.

iii. Identify dense sets in metric spaces.

Introduction: The concept of convergence of sequence of real numbers can be generalised

to convergence of sequence in metric space. Further closed sets can be defined in terms
of limit points. A set in metric space is dense if every point of metric space is limit point
of the set.

Definition 1.3.1 Sequence: A sequence in metric space X is a function from the set
N of natural numbers to set X. The sequence is generally exhibited as (xn ) where xn is
image of natural number n.

N: 1 2 3 ··· n ···
↓ ↓ ↓ ↓ ··· ↓ ···
X : x1 x2 x3 ··· xn ···

Example: (xn ) = ( n1 ) is sequence in R

Definition 1.3.2 Subsequence: Let (xn ) be a sequence in metric space X and S = {nk :
n1 < n2 < n3 < · · · } be an infinite subset of N. The sequence xnk is called subsequence of
sequence (xn ). The subsequence (xnk ) is obtained by deleting some terms of the sequence
(xn ) without disturbing the order of its elements.
Example: (xnk ) = (x2 , x4 , x6 , . . .) is a subsequence of the sequence (xn ) = (x1 , x2 , x3 , . . .).
Here nk = 2k.

Definition 1.3.3 Convergent sequence: A sequence (xn ) in metric space X is said to

converge to x in X if for every ϵ > 0 there is natural number N such that
n ≥ N =⇒ d(xn , x) < ϵ
i.e.
n ≥ N =⇒ xn ∈ B(x, ϵ)
In this case we write xn → x in X or that lim xn = x.
n→∞

Theorem 1.3.4 If the limit of a sequence in a metric space exists then it is unique.
Proof: Let (xn ) be a convergent sequence in metric space X converging to x and y.
We want to prove that x = y. Suppose on contrary that x ̸= y. Then d(x, y) > 0. Take

S25011: REAL ANALYSIS Page 31

ϵ = d(x, y) > 0. Since xn → x in X, there is a natural number N1 such that
ϵ
n ≥ N1 =⇒ d(xn , x) < . . . (I)
2
Similarly as xn → y in X, there is a natural number N2 such that
ϵ
n ≥ N2 =⇒ d(xn , y) < . . . (II)
2
Take N = max{N1 , N2 }. Then from (I) and (II)
ϵ
n ≥ N =⇒ d(xn , x), d(xn , y) <
2
In particular for n = N
ϵ
d(xN , x), d(xN , y) <
2
Consider
d(x, y) ≤ d(x, xN ) + d(xN , y)
ϵ ϵ
< +
2 2
=ϵ

= d(x, y)
So d(x, y) < d(x, y). This is a contradiction. Hence x = y.

Definition 1.3.5 Let X be a metric space and A ⊆ X. A point x ∈ X is called limit

point of A if
B(x, r) ∩ A ̸= ϕ
for every r > 0.

Theorem 1.3.6 Let A be the subset of metric space X. Then a point x in X is limit
point of A if and only if there is a sequence (xn ) in A that converges to x in X.
Proof:
I Part: Let x be a limit point of A. We want to construct a sequence (xn ) in A that
converges to x in X. For each natural number n, consider the open ball B x, n1 . Since

x is limit point of A,  
1
B x, ∩ A ̸= ϕ
n
for every n. Take xn ∈ B x, n1 for each n ∈ N. Then


1
d(xn , x) <
n
Since n1 → 0, xn → x in X.
II Part: Let (xn ) be a sequence in A that converges to x in X. We want to prove that
x is limit point of A. Let r > 0. Since xn → x, there is a natural number N such that
n ≥ N =⇒ d(xn , x) < r
i.e.
n ≥ N =⇒ xn ∈ B(x, r)
In particular for n = N , xN ∈ B(x, r). Also xN ∈ A. So xN ∈ B(x, r) ∩ A implying that
B(x, r) ∩ A ̸= ϕ

S25011: REAL ANALYSIS Page 32

This is true for every r > 0. Hence x is limit point of A.

Definition 1.3.7 Let A be a subset of metric space X.The following statements give
equivalent definitions of limit point.
i) x in X is limit point of A ⇐⇒ B(x, r) ∩ A ̸= ϕ for every r > 0.
ii) x in X is limit point of A ⇐⇒ U ∩ A ̸= ϕ for every open set U containing x.
iii) x in X is limit point of A ⇐⇒ there is a sequence (xn ) in A such that xn → x in X.

Theorem 1.3.8 A subset E of metric space X is closed if and only if E contains all its
limits (i.e. E = E).
Proof:
I Part: Let E be a closed subset of metric space and x ∈ X be a limit point of E. We
want to prove that x ∈ E. Suppose on contrary that x ̸∈ E. As E is closed, U = X − E
is open. Since x ̸∈ E, x ∈ U . The set U is open. Therefore there is r > 0 such that
B(x, r) ⊆ U . This implies that B(x, r) ∩ E = ϕ. This contradicts the fact that x is limit
point of E. Hence x ∈ E.
II Part: Let E contains all its limit points. We want to prove that E is closed. To prove
this, we prove that the complement U = X − E of E is open in X. Suppose on contrary
U is not open in X. This implies that the negation of the following statement
“For every x ∈ U there is r > 0 such that B(x, r) ⊆ U .”
is true. i.e. the statement
“ There exists x ∈ U such that B(x, r) ̸⊆ U for every r > 0.”
is true. This implies that
B(x, r) ∩ U ̸= ϕ
for every r > 0. Therefore x is limit point of E. By hypothesis E contains all its limit
points. Hence x ∈ E. This contradicts the fact that x ∈ U . So U is open in X.

Theorem 1.3.9 A subset E of metric space X is closed if and only if the following
statement holds.
(xn ) is sequence of points in E that converges to x implies that x is in E.
Proof: I Part: Let E be a closed subset of X. Let (xn ) be a sequence of points in E
that converges to x in X. We want to prove that x ∈ E. As xn → x in X and xn ∈ E
for all n ∈ N, x is limit point of E. Since E is closed, E contains all its limit points. So
x must be in E.
II Part: Conversely assume that every convergent sequence of points in E converges
to a point in E. We want to prove that E is closed in X. Let x be a limit point of E.
Then there is a sequence (xn ) of points in E that converges to x in X. By hypothesis,
x ∈ E. So E contains all its limit points. Therefore E is closed in X.

S25011: REAL ANALYSIS Page 33

Definition 1.3.10 The following statements give equivalent definitions of closed sets.
i) A subset E of metric space X is closed if and only if X − E is open in X.
ii) A subset E of metric space X is closed if and only if E contains all its limit points.
iii) A subset E of metric space X is closed if and only if every convergent sequence of
points in E converges to a point in E.

Theorem 1.3.11 Let E be the subset of metric space X. Then the following statements
are true.
i)E is the smallest closed set containing E.
ii) E is closed if and only if E = E.
Proof:
i)
I Part: E is closed in X.
E = {x ∈ X : B(x, r) ∩ E ̸= ϕ for all r > 0}.
To prove that E is closed set, we prove that the complement U = X − E is open.
U =X −E

= {x ∈ X : B(x, r) ∩ E = ϕ for some r > 0}

= {x ∈ X : B(x, r) ⊆ U for some r > 0}

By the definition of open set, U is open in X. Hence E is closed in X.
II Part: E is the smallest closed set containing E.
By (I) part, E is a closed set. Let F be a closed set containing E. We want to prove that
E ⊆ F . Let x ∈ E. Then
B(x, r) ∩ E ̸= ϕ for every r > 0

=⇒ B(x, r) ∩ F ̸= ϕ for every r > 0 (∵ E ⊆ F )

=⇒ x is limit point of F

=⇒ x ∈ F (∵ F is closed )
This proves that E ⊆ F . We proved that every closed set containing E contains E. So
E is the smallest closed set containing E.
ii)
Note that E is the set of all limit points of E.
I Part: Let E be a closed subset of metric space X and x ∈ X be a limit point of E.
We want to prove that x ∈ E. Suppose on contrary x ̸∈ E. As E is closed, U = X − E
is open. Since x ̸∈ E, x ∈ U . The set U is open. Therefore there is r > 0 such that
B(x, r) ⊆ U. This implies B(x, r) ∩ E = ϕ. This contradicts the fact that x is limit point
of E. Hence x ∈ E.
II Part: Let E contains all its limit points. We want to prove that E is closed. To prove

S25011: REAL ANALYSIS Page 34

this, we prove that the complement U = X − E of E is open in X. Suppose on contrary
U is not open in X. This implies that the negation of the following statement is true.
“For every x ∈ U , there exists r > 0 such that B(x, r) ⊆ U ”
i.e. the statement “There exists x ∈ U such that B(x, r) ̸⊆ U for every r > 0”
is true. Thus there exists some x in U , for which B(x, r) ̸⊆ U for every r > 0. This
implies that B(x, r) ∩ U ̸= ϕ for every r > 0. This implies that x is limit point of E.
As E contains all its limit points, x ∈ E. This contradicts the fact that x ∈ U . So our
assumption is wrong. Hence U must be open in X.

Definition 1.3.12 A subset D of metric space X is called dense in X if B(x, r) ∩ D ̸= ϕ

for every x ∈ X and r > 0.

Theorem 1.3.13 A subset D of metric space X is dense in X if and only if D = X.

Proof:

D = {x ∈ X : B(x, r) ∩ D ̸= ϕ for every r > 0}

I Part: Let D be a dense set in X. We want to prove that D = X. As D is dense in X,
B(x, r) ∩ D ̸= ϕ for every x ∈ X and r > 0. It easily follows that D = X.
II Part: Let D = X. We want to prove that D is dense in X. D = X. By the definition
of D, B(x, r) ∩ D ̸= ϕ for every x ∈ X and r > 0. This proves that D is dense in X.
The following statements give equivalent definitions of dense sets.
i) D is dense in X if and only if B(x, r) ∩ D ̸= ϕ for every x ∈ X and r > 0.
ii) D is dense in X if and only if D = X
iii) D is dense in X if and only if every point x of X is limit point of D.

Definition 1.3.14 A subset A of metric space X is called bounded if A ⊆ B(x0 , r0 ) for

some x0 ∈ X and r0 > 0.

Definition 1.3.15 Let A be subset of metric space X. The diameter of A is defined as

diam(A) = sup{d(x, y) : x, y ∈ A}.

Definition 1.3.16 A sequence (xn ) in metric space X is called Cauchy if for every ϵ > 0,
there is a natural number N such that
m, n ≥ N =⇒ d(xm , xn ) < ϵ

Theorem 1.3.17 Every convergent sequence in metric space is Cauchy sequence.

Proof: Let X be a metric space and let (xn ) be a sequence in X converging to the
point x in X. We want to prove that (xn ) is Cauchy. Let ϵ > 0. Since xn → x in X,
there is a natural number N such that
ϵ
n ≥ N =⇒ d(xn , x) <
2
S25011: REAL ANALYSIS Page 35
For m, n ≥ N ,
ϵ ϵ
d(xm , xn ) ≤ d(xm , x) + d(x, xn ) <
+ =ϵ
2 2
Thus for every ϵ > 0, there is a natural number N such that
m, n ≥ N =⇒ d(xm , xn ) < ϵ
This proves that (xn ) is a Cauchy sequence.

Definition 1.3.18 A sequence xn in metric space X is bounded if there is x ∈ X and

r > 0 such that
xn ∈ B(x, r)
for all n ∈ N.
A sequence (xn ) in X is bounded if all terms of (xn ) can be enclosed in some open ball
in X.

Theorem 1.3.19 Every Cauchy sequence in metric space is bounded.

Proof: Let (xn ) be a Cauchy sequence in metric space X. Since (xn ) is Cauchy for
ϵ = 1 > 0, there is a natural number N such that
m, n ≥ N =⇒ d(xm , xn ) < 1
In particular for m = N
n ≥ N =⇒ d(xN , xn ) < 1
This implies
n ≥ N =⇒ xn ∈ B(xN , 1)
i.e.
xN , xN +1 , . . . ∈ B(xN , 1) . . . (I)
Take R = max{1, d(x1 , xN ), d(x2 , xN ), . . . , d(xN − 1, xN )}. Then
d(x1 , xN ), d(x2 , xN ), . . . , d(xN − 1, xN ) ≤ R
This implies that
x1 , x2 , . . . , xN −1 ∈ B(xN , R) . . . (II)
As R ≥ 1, B(xN , 1) ⊆ B(xN , R). So from (I) and (II),
xn ∈ B(xN , R) for all n ∈ N
This proves that (xn ) is bounded.
From the last two theorems, we have the following theorem.

Theorem 1.3.20 Every convergent sequence in metric space is bounded.

Solved Problems 01

Problem 1.3.21 Show that the subset A of metric space X is bounded if and only if A
is empty or diam(A) is finite.

S25011: REAL ANALYSIS Page 36

Solution:
I Part: Let A be a bounded subset of X. We define diam(ϕ) = 0 < ∞. Let A ̸= ϕ.
We want to prove that diam(A) < ∞. Since A is bounded, there is x0 in X and r0 > 0
such that A ⊆ B(x0 , r0 ). Let x, y are in A. Then x, y are in B(x0 , r0 ) implying that
d(x, x0 ) < r0 and d(x0 , y) < r0 . Consider
d(y, x) ≤ d(y, x0 ) + d(x0 , x) < r0 + r0 = 2r0
Thus d(y, x) < 2r0 for x, y ∈ A. This means that 2r0 is upper bound for the set
{d(y, x)|x, y ∈ A}. Therefore
sup{d(y, x) : x, y ∈ A} ≤ 2r0

=⇒ diam(A) ≤ 2r0

=⇒ diam(A) < ∞
II Part: We want to prove that if A = ϕ or diam(A) < ∞ then A is bounded.
If A = ϕ then A is bounded. Let A ̸= ϕ and D = diam(A) < ∞. Then
D = sup{d(y, x) : x, y ∈ A}

=⇒ d(y, x) ≤ D for all x, y ∈ A

=⇒ d(y, x) < D + 1 for all x, y ∈ A

=⇒ d(x0 , x) < D + 1 for every x ∈ A where x0 is some fixed point in X

=⇒ x ∈B(x0 , D + 1) for every x ∈ A

=⇒ A ⊆ B(x0 , D + 1)

=⇒ A is bounded

Problem 1.3.22 Let X be a metric space, x ∈ X and r > 0. Show that diam(B(x, r)) ≤
2r.
Solution: Let y, z ∈ B(x, r). Then d(y, x) < r and d(z, x) < r. Consider
d(y, z) ≤ d(y, x) + d(x, z) < r + r = 2r
Thus d(y, z) ≤ 2r for y, z ∈ B(x, r). This means that 2r is upper bound for the set
{d(y, z) : y, z ∈ B(x, r)}. Therefore
sup{d(y, z) : y, z ∈ B(x, r)} ≤ 2r

=⇒ diam(B(x, r)) ≤ 2r

Self-Test 01

Problem 1.3.23 Which of the following statements is true?

A. Every Cauchy sequence in metric space is convergent

S25011: REAL ANALYSIS Page 37

 B. A sequence {an } in Cauchy sequence in R if and only if |an+1 − an | → 0 as
n→∞

C. If |an+1 − an | → 0 as n → ∞ then {an } in Cauchy sequence in R

D. If {an } in Cauchy sequence in R then|an+1 − an | → 0 as n → ∞

Problem 1.3.24 Let {xn } and {yn } be two Cauchy sequences in a metric space < X, d >.
Then the sequence {d(xn , yn )}

A. always converges

B. is Cauchy but need not converge

C. need not be Cauchy

D. converges to 0

Problem 1.3.25 Consider the interval (−1, 1) and a sequence {αn } of elements in it.
Then

A. every limit point of {αn } is in (−1, 1)

B. every limit point of {αn } is in [−1, 1]

C. the limit point of {αn } can only be in {−1, 0, 1}

D. the limit point of {αn } cannot be in {−1, 0, 1}

Problem 1.3.26 The number of limit points of the set { 21n + 1

2m
: m, n ∈ N} is

A. infinitely many

B. 2

C. finitely many (more than 2)

D. 1

Short Answer Questions 01

Problem 1.3.27 Show that the set Q of rationals is dense in R.

Solution: Let x ∈ R and r > 0. Then B(x, r) = (x − r, x + r). There is a rational
number q between real numbers x−r and x+r. So q ∈ (x−r, x+r)∩Q i.e. q ∈ B(x, r)∩Q.
Hence B(x, r) ∩ Q ̸= ϕ. This is true for every x ∈ R and for every r > 0. Therefore Q is
dense in R i.e. Q = R.

Problem 1.3.28 Let A be a subset of metric space X. Prove that ∂(A) = ∂(X − A).

S25011: REAL ANALYSIS Page 38

Solution: x ∈ ∂(A)
⇐⇒ B(x, r) ∩ A ̸= ϕ and B(x, r) ∩ (X − A) ̸= ϕ for every r > 0
⇐⇒ B(x, r) ∩ (X − A) ̸= ϕ and B(x, r) ∩ A ̸= ϕ for every r > 0
⇐⇒ B(x, r) ∩ (X − A) ̸= ϕ and B(x, r) ∩ (X − (X − A)) ̸= ϕ for every r > 0
⇐⇒ x ∈ ∂(X − A)
Therefore ∂(A) ⊆ ∂(X − A) and ∂(X − A) ⊆ ∂(A). This proves that ∂(A) = ∂(X − A).
Summary:

• A sequence (xn ) in metric space X is a function from the set N of natural numbers
to the set X. Here xn denote the image of n.

• A subsequence of a sequence is obtained by removing some terms of the sequence

without disturbing the order of elements in the sequence.

• A sequence (xn ) in metric space X is bounded if all terms of the sequence (xn )
can be enclosed in some open ball in X.

• A sequence (xn ) in metric space X is said to converge to point x in X if all terms

except first few finite terms of xn can be made arbitrarily closed to x.

• A sequence (xn ) in metric space X is Cauchy sequence in X if all terms except

first few finite terms of xn can be made arbitrarily close to each other.

• Every convergent sequence in metric space is Cauchy sequence but not conversely.

• Every convergent sequence in metric space is bounded.

• Every Cauchy sequence in metric space is bounded.

End of Unit Exercises

Problem 1.3.29 Show that the set Qc of irrationals is dense in R.

Problem 1.3.30 Is Z dense in the Euclidean metric space R? Justify your answer.

Problem 1.3.31 Can a finite set be dense in the Euclidean metric space R?Justify your
answer.

Problem 1.3.32 Give an example of sequence in metric space which is Cauchy but not
convergent.

Keywords: Sequence in Metric Space, Subsequence, Cauchy Sequence, Bounded Se-

quence, Convergent Sequence.

S25011: REAL ANALYSIS Page 39

REFERENCES

(1) Chapter 4, Methods of Real Analysis by Richard R. Goldberg, 2nd edition, Oxford
& IBH Publishing Co. Pvt Ltd

(2) Chapter 2, Topology of Metric Spaces by S. Kumaresan, 2nd edition, Narosa

Publishing House

(3) Chapter 2, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,

Mc Graw HillInc.,USA.

(4) Chapter 3, Real Analysis, by N.L. Carothers, 1st edition, Cambridge University
Press.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 40

 Unit 01-04: Continuity

Learning Objectives:
After successful completion of this unit, you will be able to

i. Establish the concept of sequential continuity in metric spaces.

ii. State equivalent definitions of continuity in metric spaces.

iii. Understand the algebra of continuous functions defined on metric spaces.

Introduction: The continuity of functions on metric spaces can be characterised by using

metric, sequence, open sets and closed sets. The algebra and properties of continuous
functions on metric spaces are studied with these characterisations of continuity.

Definition 1.4.1 Let X and Y be metric spaces. A function f : X → Y is said to be

continuous if
xn → x in X =⇒ f (xn ) → f (x) in Y
.

Problem 1.4.2 Show that the function f : R → R defined by f (x) = x2 is continuous on

R.
Solution: For any x in R
xn → x in R

=⇒ lim xn = x
n→∞

=⇒ lim x2n = x2
n→∞

=⇒ x2n → x2 in R
Thus xn → x in R implies f (xn ) → f (x) in R. Hence f is continuous on R.

Theorem 1.4.3 Let X, Y and Z be metric spaces. If f : X → Y and g : Y → Z are

continuous functions then g ◦ f : X → Z is continuous.
Proof : For x ∈ X,
xn → x in X

=⇒ f (xn ) → f (x) in Y (∵ f : X → Y is continuous )

=⇒ g(f (xn )) → g(f (x)) in Z (∵ g : Y → Z is continuous )

=⇒ (g ◦ f )(xn ) → (g ◦ f (x)) in Z
Thus xn → x in X implies (g ◦ f )(xn ) → (g ◦ f )(x)) in Z. Hence g ◦ f is continuous on
X.

The composition of continuous functions is continuous.

S25011: REAL ANALYSIS Page 41

Theorem 1.4.4 Let X and Y be metric spaces. For a function f : X → Y , the following
statements are equivalent.
i) f is continuous at x ∈ X.
ii) For every ϵ > 0 there is δ > 0 such that
d(x′ , x) < δ =⇒ d(f (x′ ), f (x)) < ϵ.
iii) For every open set V in Y containing f (x), there is an open set U in X containing x
such that f (U ) ⊆ V .

Proof:
i) =⇒ ii): We assume that f is continuous at x ∈ X. Suppose on contrary that ii) is
not true. Then there is ϵ > 0 such that
∀δ > 0∃x′ d(x′ , x) < δ and d(f (x′ ), f (x)) ≥ ϵ

is true. This implies that  

1
∀n ∈ N∃xn d(xn , x) < and d(f (xn ), f (x)) ≥ ϵ
n
=⇒ xn → x in X and f (xn ) ̸→ f (x) in Y
This contradicts the fact that f is continuous at x ∈ X. Hence ii) must be true.
ii) =⇒ iii): We assume that ii) is true. The statement ii) can be written as


∀ϵ > 0∃δ > 0 f (B(x, δ)) ⊆ B(f (x), ϵ) . . . (I)
Let V be an open set in Y containing f (x). Then there is ϵ > 0 such that
B(f (x), ϵ) ⊆ V
By (I) there is δ > 0 such that
f (B(x, δ)) ⊆ B(f (x), ϵ)
This implies that
f (U ) ⊆ V
where U = B(x, δ). We thus obtained an open set U in X containing x such that
f (U ) ⊆ V .
iii) =⇒ i): We assume that iii) is true. Let xn → x in X. Take ϵ > 0. Let
V = B(f (x), ϵ). Then V is an open set in Y containing f (x). By hypothesis there is an
open set U in X containing x such that
f (U ) ⊆ V = B(f (x), ϵ) . . . (II)
As U is open and x ∈ U , there is r > 0 such that
B(x, r) ⊆ U . . . (III)
Since xn → x in X, there is N ∈ N such that
n ≥ N =⇒ xn ∈ B(x, r)
This implies that
n ≥ N =⇒ xn ∈ U (from (III))


=⇒ n ≥ N =⇒ f (xn ) ∈ f (U )

S25011: REAL ANALYSIS Page 42



=⇒ n ≥ N =⇒ f (xn ) ∈ B(f (x), ϵ) (by (II))


=⇒ n ≥ N =⇒ d(f (xn ), f (x)) < ϵ)
So f (xn ) → f (x) in Y . Thus
xn → x in X =⇒ f (xn ) → f (x) in Y
This proves that f is continuous at x.
Remark: The above theorem gives equivalent definition of continuity at a point of
metric space.

Theorem 1.4.5 Let X and Y be metric spaces. A function f : X → Y is continuous if

and only if f −1 (U ) is open in X whenever U is open in Y .
Proof:
I Part: Let f be continuous and U be open in Y . We want to prove that f −1 (U ) is
open in X. Take x ∈ f −1 (U ). Then f (x) ∈ U . So U is an open set in Y containing f (x).
The continuity of f at x implies the existence of an open set W in X containing x such
that
f (W ) ⊆ U . . . (I)
As W is open and x ∈ W , there is r > 0 such that
B(x, r) ⊆ W . . . (II)
Now
f (W ) ⊆ U (from (I))

=⇒ f −1 f (W ) ⊆ f −1 (U )

=⇒ B(x, r) ⊆ f −1 (U ) (from (II))

Thus for every x ∈ f −1 (U ) there is r > 0 such that B(x, r) ⊆ f −1 (U ). This proves that
f −1 (U ) is open in X.
II Part: Let f −1 (U ) be open in X whenever U is open in Y . We want to prove that
f is continuous on X. Take a point x ∈ X and let ϵ > 0. Take U = B(f (x), ϵ). Then
U is an open set in Y . By hypothesis f −1 (U ) is open in X. Note that x ∈ f −1 (U ) as
f (x) ∈ B(f (x), ϵ) = U . Since f −1 (U ) is open in X, there is δ > 0 such that
B(x, δ) ⊆ f −1 (U )
This implies that
B(x, δ) ⊆ f −1 (B(f (x), ϵ))

=⇒ f B(x, δ) ⊆ f f −1 B(f (x), ϵ)

∵ f (f −1 (C)) ⊆ C



=⇒ f B(x, δ) ⊆ B(f (x), ϵ)

x′ ∈ B(x, δ) =⇒ f (x′ ) ∈ B(f (x), ϵ)

=⇒

d(x′ , x) < δ =⇒ d(f (x′ ), f (x)) < ϵ

=⇒

S25011: REAL ANALYSIS Page 43

So f is continuous at x. As x was arbitrary point of X, f is continuous on X.

A function from one metric space to another metric space is continuous if and only if
inverse image of every open set (in codomain) is open (in domain).

Theorem 1.4.6 Let X and Y be metric spaces. A function f : X → Y is continuous if

and only if f −1 (V ) is closed in X whenever V is closed in Y .
Proof:
I Part: Let f be continuous and V be closed in Y . We want to prove that f −1 (V ) is
closed in X. Consider
V closed in Y
=⇒ Y − V is open in Y
=⇒ f −1 (Y − V ) is open in X (by previous theorem)
=⇒ f −1 (Y ) − f −1 (V ) is open in X
=⇒ X − f −1 (V ) is open in X
=⇒ X − X − f −1 (V ) is closed in X

=⇒ f −1 (V ) is closed in X
II Part: Let f −1 (V ) be closed in X whenever V is closed in Y . We want to prove that
f is continuous on X. Consider
U is open in Y
=⇒ Y − U is closed in Y
=⇒ f −1 (Y − U ) is closed in X (by hypothesis)
=⇒ f −1 (Y ) − f −1 (U ) is closed in X
=⇒ X − f −1 (U ) is closed in X
=⇒ X − X − f −1 (U ) is open in X

=⇒ f −1 (U ) is open in X
Thus f −1 (U ) is open in X whenever U is open in Y . Hence by previous theorem, f is
continuous on X.

A function from one metric space to another metric space is continuous if and only if
inverse image of every closed set (in codomain) is closed (in domain).

Definition 1.4.7 Let X and Y be metric spaces. A function f : X → Y is called

homeomorphism of X onto Y if
i) f is one-one
ii) f is onto
iii) f is continuous on X

S25011: REAL ANALYSIS Page 44

iv) f −1 is continuous on Y
If there is some homeomorphism from X to Y , we say that X and Y are homeomorphic.
The term homeomorphism in analysis (or topology) is used to denote the fact that two
structures are identical. We leave it to the reader to prove that the relation of homeo-
morphism is an equivalence relation on collection of metric spaces.

Theorem 1.4.8 Let X be a metric space and, x ∈ X and c ∈ R. The following state-
ments are true.
i) If f, g : X → R are continuous at x ∈ X then so are f +g, f g and cf . So the set C(X, R)
of all continuous functions from X to R is a vector space under pointwise addition and
scalar multiplication.
ii) If f : X → R is continuous at x and f (x) ̸= 0 then there is r > 0 such that the
1
function h : B(x, r) → R defined by h(x) = f (x)
is continuous on B(x, r).
Proof:
i) Consider
xn → x in X

=⇒ f (xn ) → f (x) in R and g(xn ) → g(x) in R (∵ f, g : X → R are continuous)

=⇒ lim f (xn ) = f (x) and lim g(xn ) = g(x)

n→∞ n→∞

=⇒ lim (f (xn ) + g(xn )) = f (x) + g(x), lim (f (xn )g(xn )) = f (x)g(x) and lim cf (xn ) = cf (x)
n→∞ n→∞ n→∞

=⇒ (f + g)(xn ) → (f + g)(x), (f g)(xn ) → (f g)(x), and (cf )(xn ) → (cf )(x) in R

Thus
xn → x in X =⇒ (f +g)(xn ) → (f +g)(x), (f g)(xn ) → (f g)(x) and (cf )(xn ) → (cf )(x) in R
This proves that f + g, f g and cf are continuous on X. So f + g, cf ∈ C(X, R) whenever
f, g ∈ C(X, R). The set C(X, R) is thus closed under addition and scalar multiplication.
The rest of the properties of vector space can easily be verified.
ii) Let f (x) > 0. Take ϵ = f (x)
2
. Since f is continuous at x, there is r > 0 such that
d(x , x) < r =⇒ |f (x′ ) − f (x)| < ϵ
′

This implies that

f (x)
x′ ∈ B(x, r) =⇒ |f (x′ ) − f (x)| <
2
 
′ f (x) ′ 3
=⇒ x ∈ B(x, r) =⇒ 0 < < f (x ) < f (x)
2 2
So f (x′ ) ̸= 0 whenever x′ ∈ B(x, r). The proof is similar when f (x) < 0. The function
1
h : B(x, r) → R is defined as h(x) = f (x)
. To prove that h is continuous on B(x, r),
consider
xn → x in B(x, r)

=⇒ xn → x in X

S25011: REAL ANALYSIS Page 45

 =⇒ f (xn ) → f (x) in R (∵ f : X → R is continuous)

=⇒ lim f (xn ) = f (x)

n→∞
 
1 1
=⇒ lim =
n→∞ f (xn ) f (x)
=⇒ lim h(xn ) = h(x)
n→∞

=⇒ h(xn ) → h(x) in R
Thus
xn → x in B(x, r) =⇒ h(xn ) → h(x) in R
This proves that h is continuous on B(x, r).

Definition 1.4.9 Let X and Y be metric spaces. A function f : X → Y is said to be

uniformly continuous if for every ϵ > 0 there is δ > 0 such that
x, y ∈ X and d(x, y) < δ =⇒ d(f (x), f (y)) < ϵ
In continuity, δ depends both on point x and ϵ. In uniform continuity, δ depends only on
ϵ. It is clear that every uniformly continuous function is continuous.

Definition 1.4.10 Let X and Y be metric spaces. A map f : X → Y is called open map
if
U is open in X =⇒ f (U ) is open in Y.

Definition 1.4.11 Let X and Y be metric spaces. A map f : X → Y is called closed

map if
U is closed in X =⇒ f (U ) is closed in Y

Solved Problems 01

Problem 1.4.12 Show that the function f : R2 → R defined by f (x, y) = x + y is

continuous on R2 .
Solution: For any (x, y) in R2 ,
(xn , yn ) → (x, y) in R2

=⇒ xn → x in R and yn → y in R

=⇒ xn + yn → x + y in R

=⇒ f (xn , yn ) → f (x, y) in R
Thus (xn , yn ) → (x, y) in R2 implies f (xn , yn ) → f (x, y) in R. Hence f is continuous on
R2 .

S25011: REAL ANALYSIS Page 46

Problem 1.4.13 Let X be a metric space and f, g : X → R be continuous functions on
X. Prove that the function h : X → R2 defined by h(x) = (f (x), g(x)) is continuous on
X.
Solution: For any x ∈ X,
xn → x in X

=⇒ f (xn ) → f (x) in R and g(xn ) → g(x) in R (∵ f, g : X → R are continuous )

=⇒ (f (xn ), g(xn )) → (f (x), g(x)) in R2

=⇒ h(xn ) → h(x) in R2
Thus xn → x in X implies h(xn ) → h(x) in R2 . Hence h is continuous on X.
Self-Test 01

Problem 1.4.14 Pick out function from R → R that is discontinuous only at countably
many point(s).

A. f (x) = sin(x)(x ∈ Q), f (x) = cos x(x ̸∈ Q)

B. f (x) = 1(x ∈ Q), f (x) = −1(x ̸∈ Q)

C. f (x) = x3 − 8x + 6(x ∈ Q), f (x) = x4 − 8x + 5(x ̸∈ Q)

x8 −sin x+x7
D. f (x) = x2 +7

Problem 1.4.15 Let f bea real

√
valued function defined on R as follows
 p+ √2 − p if t = p , p, q ∈ Z, (p, q) = 1 ≥ 0
q+ 2 q q
f (x) =
 0 if t is irrational
A. f is not continuous at t = 1

B. f is continuous at all rationals

C. f is not continuous at all irrationals

D. f is continuous at all rationals

Problem 1.4.16 The number of homeomorphism(s) between (0, 1) and (2, 3) is

A. 0

B. countably infinite

C. uncountable

D. 1
1 1
Problem 1.4.17 Consider the function f (x) = x
on [1, ∞) and g(x) = x
on x > 0.

A. both f and g are uniformly continuous

S25011: REAL ANALYSIS Page 47

 B. g is uniformly continuous but f is not

C. neither f nor g is uniformly continuous

D. f is uniformly continuous but g is not

Problem 1.4.18 Let u and d be usual and discrete metric on R respectively. Define
I :< R, u >→< R, d > as I(x) = x for all x ∈ R. Then

A. I is bounded.

B. I is not continuous.

C. I is continuous but not uniformly continuous

D. I is uniformly continuous

Problem 1.4.19 Let f : R → R and g : [a, b] → R be defined as f (x) = x2 , ∀x ∈ R and

g(x) = x2 , ∀x ∈ [a, b].

A. f and g are uniformly continuous

B. f is uniformly continuous and g is continuous

C. f is continuous and g is uniformly continuous

D. f and g are continuous but f and g are not uniformly continuous

Problem 1.4.20 Which of the following is false?

A. if f : R → R is Lipschitz map then f is uniformly continuous

B. any map f : N → R is uniformly continuous

C. if f : R → R is continuous and f (x) → 0 as x → ±∞ then, f is uniformly

continuous

D. if f : (0, 1) → R is continuous then f is uniformly continuous

Problem 1.4.21 Which of the following subspaces of R are not homeomorphic?

A. [0, 1) and [0, ∞)

B. (0, 1) and [0, 1]

C. the set of all integers and the set of all even integers

D. [1, 2] and [2, 4]

S25011: REAL ANALYSIS Page 48

Short Answer Questions 01

Problem 1.4.22 Prove that the constant function from one metric space to another
metric space is continuous.
Solution: Let X and Y be metric spaces and y0 ∈ y. Define the function f : X → Y
as f (x) = y0 for all x ∈ X. For any x ∈ X,
xn → x in X

=⇒ f (xn ) = y0 for all n in N

=⇒ f (xn ) → y0 in Y

=⇒ f (xn ) → f (x) in Y
Thus xn → x in X implies f (xn ) → f (x) in Y . Hence f is continuous.

Problem 1.4.23 Show that the function f : R2 → R defined by f (x, y) = x · y is

continuous on R2 .
Solution: For any (x, y) in R2 ,
(xn , yn ) → (x, y) in R2

=⇒ xn → x in R and yn → y in R

=⇒ xn · yn → x · y in R

=⇒ f (xn , yn ) → f (x, y) in R
Thus (xn , yn ) → (x, y) in R2 implies f (xn , yn ) → f (x, y) in R. Hence f is continuous on
R2 .

Summary:

• Let X and Y be metric spaces. A function f : X → Y is said to be continuous

at point x if for every ϵ > 0, there is δ > 0 such that
d(x, x′ ) < δ =⇒ d(f (x), f (x′ )) < ϵ
The function f is continuous at x if images of f can be made as close as we want
to the point f (x) in Y by taking pre-images sufficiently close to the point x in
X.

• Let X and Y be metric spaces and f : X → Y . If f is continuous at every

point of X then f is said to be continuous on X. The simple interpretation of
continuity is that images are close to each other whenever pre-images are close
to each other.

S25011: REAL ANALYSIS Page 49

 • Let X and Y be metric spaces. A function f : X → Y is continuous if inverse
image of every open (closed) set in Y under f is open (closed) in X.

• The composition of continuous functions is continuous.

• Let X be a metric space. The set C(X, R) of all continuous functions from X to
R is a vector space under point-wise addition and scalar multiplication.

• Let X and Y be metric spaces. A function f : X → Y is said to be uniformly

continuous if for every ϵ > 0 there is δ > 0 such that
x, y ∈ X and d(x, y) < δ =⇒ d(f (x), f (y)) < ϵ
In continuity, δ depends both on point x and ϵ. In uniform continuity, δ depends
only on ϵ. It is clear that every uniformly continuous function is continuous.

• Let X and Y be metric spaces. A map f : X → Y is open(closed) map if image

of every open(closed) set in X is open(closed).

• Let X and Y be metric spaces. If there is a continuous bijective open map

from X to Y then X and Y are said to be homeomorphic. The term homeomor-
phism in topology is used to denote the fact that two structures are topologically
equivalent.

End of Unit Exercises

Problem 1.4.24 Prove that the identity function one metric space X is continuous.

Problem 1.4.25 Let f : R → R be continuous. Prove that the function h : R2 → R2

defined by h(x, y) = (f (x + y), f (x − y)) is continuous on R2 .

Problem 1.4.26 Let X, Y be metric spaces and D be dense in X. If f, g : X → Y are

continuous functions such that f (x) = g(x) for all x ∈ D, prove that f = g.

Problem 1.4.27 The following table gives some homeomorphisms between subsets of R.

S25011: REAL ANALYSIS Page 50

 X Y homeomorphism f : X → Y
   
d−c d−c
(a, b) (c, d) f (x) = c+ b−a (x − a) or f (x) = d− b−a (x − a)
   
d−c d−c
[a, b] [c, d] f (x) = c+ b−a (x − a) or f (x) = d− b−a (x − a)
   
d−c d−c
(a, b] (c, d] f (x) = c+ b−a (x − a) or f (x) = d− b−a (x − a)
   
d−c d−c
[a, b) [c, d) f (x) = c+ b−a (x − a) or f (x) = d− b−a (x − a)
 
d−c
(a, b] [c, d) f (x) = d− b−a (x − a)
 
d−c
[a,b) (c,d] f (x) = d− b−a (x − a)
−π π


,
2 2
R f (x) = tan x
−π π
f (x) = tan−1 x


R ,
2 2
x
(0, 1) (1, ∞) f (x) = 1−x
x
(1, ∞) (0, 1) f (x) = 1+x
1
(0, 1] [1, ∞) f (x) = x
1
[1, ∞) (0, 1] f (x) = x

Verify that the above functions are homeomorphisms between respective sets.

Keywords: Continuity of Functions on Metric Spaces, Sequential Continuity, Algebra

of Continuous Functions, Homemorphism, Open and Closed Maps.

REFERENCES

(1) Chapter 5, Methods of Real Analysis by Richard R. Goldberg, 2nd edition, Oxford
& IBH Publishing Co. Pvt Ltd

(2) Chapter 3, Topology of Metric Spaces by S. Kumaresan, 2nd edition, Narosa

Publishing House

(3) Chapter 2, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,

Mc Graw HillInc.,USA.

(4) Chapter 5, Real Analysis, by N.L. Carothers, 1st edition, Cambridge University
Press.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 51

CREDIT 02

53
 Unit 02-01: Connected Metric Spaces

Learning Objectives:
After successful completion of this unit, you will be able to

i. Define connected metric spaces.

ii. State properties of connected metric spaces.

iii. Identify connected subsets of R.

Introduction: A connected metric space can be thought as single piece. It can be

characterised using disconnection in terms of open sets and closed sets and also using
characteristic functions. The only connected subsets of R are intervals. The removal of
a single point from R makes it disconnected whereas removal of single (even countably
many points) keep R2 connected. This fact can be used to prove that R and R2 are not
homeomorphic as connected sets are preserved under continuous maps.

Definition 2.1.1 A metric space X is said to be connected if the only subsets of X that
are both open and closed in X are ϕ and X.

Definition 2.1.2 Let A be a subset of metric space X. Then A is also metric space. We
say that A is connected subset of X if A as a metric space is connected.

Definition 2.1.3 Let X be metric space and A, B ⊆ X. Sets A and B form a disconnec-
tion of X if

i. A, B are proper subsets of X whose union is X. (i.e. A, B ̸= ϕ, X and X = A∪B)

ii. A and B are disjoint. (i.e. A ∩ B = ϕ)

iii. A and B are both open and closed in X.

Theorem 2.1.4 X is disconnected if and only if there is a disconnection of X.

Proof:
I Part: Let X be disconnected. Then there is A ⊆ X with A ̸= ϕ and A ̸= X which is
both open and closed in X. Take B = X − A. Then B ̸= ϕ, X. As A is open and closed,
B is also open and closed. Also

A ∪ B = A ∪ (X − A) = X and A ∩ B = A ∩ (X − A) = ϕ
{A, B} thus forms disconnection of X.
II Part: Let {A, B} be a disconnection of X. This implies that A ̸= ϕ and A ̸= X
which is both open and closed in X. Thus X has a non-trivial subset which is both open
and closed in X. Hence X is disconnected.

S25011: REAL ANALYSIS Page 54

Theorem 2.1.5 A metric space X is connected if and only if every continuous function
f : X → {1, −1} is a constant function.

Proof:
I Part: Let X be a connected space and f : X → {1, −1} be continuous function.We
want to prove that f is a constant function. Suppose on contrary that the function f is
not constant. Take A = f −1 ({1}) and B = f −1 ({−1}). Since {1} is closed subset of R
and f is continuous; A = f −1 ({1}) is closed subset of X. Similarly B = f −1 ({−1}) is
closed subset of X. Further A is open as A = X − B. Thus A is both open and closed
in X. So X is not connected. This is a contradiction. Hence f must be constant.
II Part: We give a proof by contraposition. Assuming that X is not connected, we
construct a non-constant continuous function from X to {1, −1}.
Let X be not connected. Then there exist two disjoint proper non-empty subsets A and
B in X such that A and B are both open and closed in X,X = A ∪ B and A ∩ B = ϕ.
Define: f : X → {1, −1}

 1 if x ∈ A
f (x) =
 −1 if x ∈ B
Now,
f −1 ({−1}) = B, f −1 ({1}) = A, and f −1 ({−1, 1}) = A ∪ B = X
Thus the inverse image of every closed subset of {1, −1} is closed subset of X. Hence f
is continuous.

Theorem 2.1.6 Let X be a metric space. If A and B are two connected subsets of X
such that A ∩ B ̸= ϕ then A ∪ B is connected.

Proof: Let c ∈ A ∩ B and f : A ∪ B → {1, −1} be continuous. Since A is connected, f

is constant on A. Therefore
f (a) = f (c) for all a ∈ A . . . (I)

Similarly,
f (b) = f (c) for all b ∈ B . . . (II)

From (I) and (II),

f (x) = f (c) for all x ∈ A ∪ B

Thus f is constant on A ∪ B. We have shown that any continuous function f : A ∪ B →

{1, −1} is a constant function. This proves that A ∪ B is connected.

S25011: REAL ANALYSIS Page 55

Theorem 2.1.7 Let {Ai |i ∈ I} be a collection of connected subsets of a metric space X
S
with property that Ai ∩ Aj ̸= ϕ for all i, j ∈ I.Then the set A = i Ai is connected.
Proof: Let i0 ∈ I. Let f : A → {−1, 1} be a continuous function. Since Ai0 is connected,
f is constant on Ai0 .Without loss of generality, f (x) = 1 for every x ∈ Ai0 . Define the
set U as
[
U= Ai
i∈I,i̸=i0

Let y ∈ U . Then y ∈ Ai1 for all i1 ∈ I and i1 ̸= i0 . By hypothesis Ai0 ∩ Ai1 ̸= ϕ. Take
z ∈ Ai0 ∩ Ai1 . As Ai1 is connected, f is constant on Ai1 . Hence
f (y) = f (z) = 1
S
Thus f (x) = 1 for all x ∈ i∈I Ai . So every continuous function f : A → {−1, 1} is
S
constant. This proves that Ai is connected.
i∈I

The union of pairwise non-disjoint collection of connected sets is connected.

Theorem 2.1.8 Let {Ai |i ∈ I} be a collection of connected subsets of a metric space X.

T S
If i Ai ̸= ϕ then the set A = i Ai is connected.
T S
Proof: Take x0 ∈ i Ai . Let f : i Ai → {−1, 1} be continuous. We want to prove that
S S
f is constant on i Ai . Let, x ∈ i Ai . Then x ∈ Ai for some i ∈ I. As Ai is connected,
f is constant on Ai . Hence
f (x) = f (x0 )
S S
Since x is arbitrary point of i Ai , f is constant on i Ai . Thus every continuous function
S
f : i Ai → {−1, 1} is constant. This proves that A is connected.

The union of non-disjoint collection of connected sets is connected.

Theorem 2.1.9 If X is connected metric space and g : X → Y is continuous then g(X)

is connected.
Proof:
f : g(x) → {1, −1} is continuous function.
=⇒ f ◦ g → {−1, 1} is continuous.
=⇒ f ◦ g is constant on X. (∵ X is connected)
=⇒ f : g(X) → {−1, 1} is constant on g(X).
Thus any continuous function from g(x) to {1, −1} is constant and hence g(x) is con-
nected.

The image of connected set under continuous map is connected.

Theorem 2.1.10 The interval [a, b] in Euclidean metric space R is connected.

S25011: REAL ANALYSIS Page 56

Proof: Suppose on contrary that [a, b] is not connected. Then [a, b] = A ∪ B where
A and B are non empty open and closed proper subsets of metric space [a, b] such that
[a, b] = A ∪ B and A ∩ B = ϕ. Without loss of generality, let a ∈ A. Define the set S as
S = {x ∈ [a, b] : [a, x] ⊆ A}
As A is open and a ∈ A, there is r > 0 such that [a, r) ⊆ A. So [a, y] ⊆ A where y
is such that 0 < y < r − a. Hence y ∈ S. Also S is bounded as S ⊆ [a, b]. So S is
non-empty upper bounded subset of R. By lub axiom, the supremum of S exists finitely.
Let c = sup S. By the definition of supremum, for each n ∈ N there is xn ∈ S such that
1
c− n
≤ xn ≤ c. So the sequence (xn ) in A converges to c in A implying that c is limit
point of A. As A is closed subset of metric space
 [a, b], c ∈ A. Further
∞ ∞
[ 1 [
[a, c) ⊆ a, c − ⊆ [a, xn ] ⊆ A
n=1
n n=1
Thus [a, c) ⊆ A and c ∈ A. Therefore [a, c] ⊆ A implying that c ∈ S by definition of S.
Clearly c ≤ b. Suppose c < b. As A is open in [a, b], c ∈ A and c < b, there is s > 0 such
that (c − s, c + s) ⊆ A. Also c ∈ S. All these imply that [a, c + t] ⊆ A where 0 < t < s.
So c + t ∈ S by definition of S. This contradicts the fact c = sup S. Hence c = b. Again
by definition of S, this implies that [a, c] = [a, b] ⊆ A and so [a, b] = A. This results into
the fact that B = ϕ. This is a contradiction. Hence [a, b] must be connected.

Theorem 2.1.11 (Connected subsets of R) A set J is connected subset of R if and

only if J is an interval.
Proof:
I Part: Let J be a connected subset of R. Suppose on contrary that J is not an interval
in R. Then there are points x and y ∈ J and point z such that x < z < y and z ̸∈ J.
Define f : R − {z} → {1, −1} as 
 −1 if s < z
f (s) =
 1 if s > z
Then the restriction of f to J is a continuous function from J to {1, −1} which is not
constant. Therefore J is not connected. This is a contradiction. So our assumption is
wrong. Hence J must be an interval.
II Part: Let J be an interval and x, y ∈ J. Without loss of generality, assume that
x < y. Let f : J → {1, −1} be continuous. Since [x, y] is connected subset of J, f is
constant on [x, y]. So f (x) = f (y). As x, y are arbitrary points of J, f is constant on J.
Thus any continuous function from J to {1, −1} is constant. Hence J is connected.

Theorem 2.1.12 (Intermediate value theorem): Let f : [a, b] → R be continuous

function. Assume that y is a point between f (a)and f (b) (i.e. either f (a) ≤ y ≤ f (b) or
f (b) ≤ y ≤ f (a)). Then there exists x ∈ [a, b] such that f (x) = y.

S25011: REAL ANALYSIS Page 57

Proof:
f : [a, b] → R is continuous


=⇒ f [a, b] is connected subset of R


=⇒ f [a, b] = interval
Now
f (a), f (b) ∈ f [a, b] and f (a) ≤ y ≤ f (b) or f (b) ≤ y ≤ f (a)


=⇒ y ∈ f [a, b]

=⇒ y = f (x) for some x ∈ f [a, b]

Theorem 2.1.13 (Existence of nth root) Let α ∈ [0, ∞) for n ∈ N. Then there exists
unique x ∈ [0, ∞) such that xn = α.
Proof:
Existence Part: The proof is obvious when α = 0. Let α > 0. Consider the continuous
function f : [0, ∞) → [0, ∞) defined as f (t) = tn . By Archimedean property of real
numbers, there is n0 ∈ N such that n0 > α.
Now f (0) = 0 < α and fn (0) = nn0 ≥ n0 ≥ α. So applying intermediate value theorem to
the restriction of f to the connected interval [0, n0 ], we see that there exists x ∈ [0, n0 ]
such that f (x) = xn = α.
Uniqueness Part: For x ∈ [0, ∞),
f (x) = xn

=⇒ f ′ (x) = nxn−1 > 0 (∵ n ∈ N)

=⇒ f is increasing on (0, ∞)

=⇒ f is one-one


=⇒ f (x1 ) = f (x2 ) = α =⇒ x1 = x2

Theorem 2.1.14 If X and Y are connected metric spaces then the product metric space
X × Y is also connected.
Proof: Let f : X × Y → {1, −1} be a continuous map. Fix the point (x0 , y0 ) ∈ X × Y .
Take (x, y) ∈ X × Y . Define ix : Y → X × Y and iy : X → X × Y as ix (y) = (x, y)
and iy (x) = (x, y) for all (x, y) ∈ X × Y . The maps ix and iy are continuous. Since the
image of connected set under a continuous map is connected and X, Y are connected,
the subsets ix (y) = {x} × Y and iy (x) = X × {y} of X × Y are connected. In particular
X × {y0 } and {x0 } × Y are connected. The point (x0 , y0 ) lies in both sets X × {y0 }
and {x0 } × Y . The restriction of f to either of these sets are continuous and hence f is
constant on them. This implies that

S25011: REAL ANALYSIS Page 58

 f (x0 , y0 ) = f (x, y0 ) for all x ∈ X
and
f (x, y0 ) = f (x, y) for all y ∈ Y
Therefore
f (x, y) = f (x0 , y0 ) for all (x, y) ∈ X × Y
f is thus constant on X × Y . We have shown that any continuous function from X × Y
to {1, −1} is constant. This proves that X × Y is connected.
Solved Problems 01

Problem 2.1.15 Let f, g : [0, 1] → R be continuous functions. Assume that f (x) ∈ [0, 1]
for all x, g(0) = 0 and g(1) = 1. Show that f (x) = g(x) for some x ∈ [0, 1].
Solution: If f (0) = 0 then f (x) = g(x) for x = 0. If f (1) = 1 then f (x) = g(x) for
x = 1. So let f (0) ̸= 0 and f (1) ̸= 1. As f (x) ∈ [0, 1] we see that 0 < f (0) ≤ 1 and
0 ≤ f (1) < 1. As f and g are continuous on [0, 1], g − f is also continuous on [0, 1]. Now
(g − f )(0) = g(0) − f (0)

= 0 − f (0)

= −f (0) < 0 (∵ f (0) > 0)

(g − f )(1) = g(1) − f (1)

= 1 − f (1)

= 1 − f (1) > 0 (∵ f (1) < 1)

Therefore by intermediate value theorem, there exists x ∈ [0, 1] such that
(g − f )(x) = 0

=⇒ g(0) − f (0) = 0

=⇒ g(x) = f (x)

Problem 2.1.16 Let f : X → R be a non-constant continuous function on a connected

metric space. Show that f (X) is uncountable and hence X is uncountable.
Solution: We shall use the following results.
(i) The image of connected set under continuous map is connected.
(ii) The image of countable set under any map is countable.
(iii) A subset J of R is connected if and only if it is an interval.
f : X → R is continuous and X is connected. By result (i), f (X) is connected subset
of R. By result (iii), f (X) is interval. Since f is non-constant, f (X) contains at least
two points. Also any interval (except the singleton closed interval) contains uncountably
many real numbers. It follows that f (X) is uncountable. By result (ii), X can’t be

S25011: REAL ANALYSIS Page 59

countable for if it is then f (X) is countable. We have obtained the desired conclusion
now.
Self-Test 01

Problem 2.1.17 Consider the set A = {x sin( x1 ) : x ∈ (0, 1)} as a subset of R. Then

A. A is closed

B. A is compact

C. A is connected

D. A is not bounded

Problem 2.1.18 The set R of reals with usual metric is

A. connected and complete

B. compact

C. compact and connected

D. complete but not connected

Problem 2.1.19 Let f : R3 → R be a continuous function.If D = {(x, y, z) ∈ R3 :

x2 + y 2 + z 2 ≤ 4}. Then f (D)

A. need not be an interval in R

B. is always an interval of the form (a, b)

C. is union of two disjoint closed sets in R

D. is always an interval of the form [a, b]

Problem 2.1.20 Let G be an open set in Rn . Two points x, y ∈ G are said to be

equivalent if they can be joined by a continuous path completely lying inside G. Number
of equivalence classes is

A. only one

B. at most finite

C. at most countable

D. can be finite, countable or uncountable

Short Answer Questions 01

Problem 2.1.21 Show that R∗ = R − {0} is disconnected.

S25011: REAL ANALYSIS Page 60

Solution:
(0, ∞) = (0, ∞) ∩ R∗ =⇒ (0, ∞) is open in R∗
(0, ∞) = [0, ∞) ∩ R∗ =⇒ (0, ∞) is closed in R∗
Thus (0, ∞) is both open and closed R∗ .
Therefore R∗ is disconnected.

Problem 2.1.22 X is a metric space such that given any two points x, y ∈ X there exists
connected set A such that x, y ∈ A. Prove that X is connected.
Solution: Let x0 ∈ X be fixed and x ∈ X be arbitrary. Let f : X → {−1, 1} be
continuous. By assumption there is connected subset A of X such that x0 , x ∈ A. Since
f is continuous on A and A is connected , f is constant on A. Therefore f (x) = f (x0 ).
Since x is any point of X, it follows that f is constant on X. Thus we have shown
that any continuous function f : X → {−1, 1} is a constant. Hence X is connected.
Summary:

• A connected metric space is “ single piece”. A metric space X is connected

if there is no disconnection of X. By a disconnection we mean a pair of non-
empty proper open and closed subsets of X which form a partition of X. This is
equivalent to saying that X is connected if there is no non-empty proper subset
of X which is both open and closed in X. So intervals in R, line segments and
circular discs in R2 are connected. However if we remove a single point from
interval or line segment, it gets disconnected. The same is not true in case of
circular discs. This fact can be used to prove that R is not homeomorphic to R2 .

• A metric space X is connected if and only if every continuous function from X

to a finite set is a constant function.

• The union of pairwise non-disjoint collection of connected sets is connected.The

union of non-disjoint collection of connected sets is connected.

• A finite subset of metric space is connected if and only if it is a singleton set.

• The only connected subsets of R are intervals.

• The image of connected set under continuous map is connected.

• A finite Cartesian product of connected spaces is connected.

End of Unit Exercises

Problem 2.1.23 If X is a discrete metric space and |X| ≥ 2 then X is disconnected.

S25011: REAL ANALYSIS Page 61

Problem 2.1.24 When is a finite subset of a metric space is connected?

Problem 2.1.25 Prove that C = {(x, y) ∈ R2 |x2 + y 2 = 1} is connected.

Problem 2.1.26 Let A be non-empty subset of R. Assume that every point of A is

rational. What can you conclude?

Problem 2.1.27 Let f : R → R be continuous and f (x) ∈ Q for all x ∈ R. Prove that f
must be constant.

Problem 2.1.28 Let A be connected subset of a metric space X. If A ⊆ B ⊆ Ā then

prove that B is connected. In particular, Ā is connected.

Problem 2.1.29 Prove that any polynomial with real coefficient of odd degree has at least
one real root. i.e. if p(x) = an xn + an−1 xn−1 + · · · + a1 x + a0 , aj ∈ R for 0 ≤ j ≤ n, an ̸= 0
and n is odd then there exists α ∈ R such that p(α) = 0.

Keywords: Connected Metric Space, Product of Connected Metric Spaces, Connected

Subsets of R.

REFERENCES

(1) Sections 6.1 and 6.2, Methods of Real Analysis by Richard R. Goldberg, 2nd
edition, Oxford & IBH Publishing Co. Pvt Ltd

(2) Chapter 5, Topology of Metric Spaces by S. Kumaresan, 2nd edition, Narosa

Publishing House

(3) Chapter 2, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,

Mc Graw HillInc.,USA.

(4) Chapter 6, Real Analysis, by N.L. Carothers, 1st edition, Cambridge University
Press.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 62

 Unit 02-02: Complete Metric Spaces

Learning Objectives:
After successful completion of this unit, you will be able to

i. Define complete metric spaces.

ii. State properties of complete metric spaces.

iii. Establish the completeness of discrete metric spaces.

Introduction: A metric space is complete if every Cauchy sequence in it is convergent.

A closed subset of complete metric space is complete and a complete subset of any metric
space is closed. Any discrete metric space is complete.

Definition 2.2.1 A metric space X is complete if every Cauchy sequence in X is conver-

gent.

Theorem 2.2.2 If Y is a closed subset of complete metric space X then Y is complete.

Proof: Consider the following argument. (xn ) is a Cauchy sequence in metric space Y
=⇒ (xn ) a Cauchy sequence in metric space X
=⇒ there is x in X such that xn → x in X (∵ X is complete)
=⇒ x is limit point of Y
=⇒ x ∈ Y (∵ Y is closed)
Thus(xn ) is convergent in Y if (xn ) is a Cauchy sequence in Y . Hence Y is complete.

A closed subset of complete metric space is complete.

Theorem 2.2.3 If Y is a complete subset of metric space X then Y is closed in X.

Proof: Let y be a limit point of Y . Then there is a sequence (yn ) in Y which converges
to y in X. Every convergent sequence is Cauchy. So (yn ) is Cauchy sequence in Y . As
Y is complete, it must be convergent (and convergent to y) in Y . It now follows that
y ∈ Y . Thus Y contains all its limit points and so is closed.

A complete subset of any metric space is closed.

Theorem 2.2.4 Cantor’s Intersection Theorem: Let X be a complete metric space

and (Fn ) be a sequence of closed and bounded subsets of X such that F1 ⊆ F2 ⊆ F3 · · ·.
∞
T
If lim diam(Fn ) = 0 then Fn contains precisely one point.
n→∞ n=1
Proof: Take a sequence (xn ) in X such that xn ∈ Fn for all n ∈ N. Let ϵ > 0. Since
lim diam(Fn ) = 0, there is N ∈ N such that
n→∞
diam(FN ) < ϵ . . . (I)

S25011: REAL ANALYSIS Page 63

Now,
m, n ≥ N

=⇒ xm , xn ∈ FN (∵ F1 ⊆ F2 ⊆ F3 · · · )

=⇒ d(xm , xn ) < diam(FN )

=⇒ d(xm , xn ) < ϵ ( from (I))

Thus for ϵ > 0 there is N ∈ N such that
m, n ≥ N =⇒ d(xm , xn ) < ϵ
This proves that (xn ) is a Cauchy sequence in X. As X is complete, (xn ) is convergent.
Let xn → x. Note that xn , xn+1 , xn+2 , . . . ∈ Fn as F1 ⊆ F2 ⊆ F3 · · ·. So x is limit point of
∞
T
Fn . As Fn is closed, x ∈ Fn . This holds for any n ∈ N. Hence x ∈ Fn . The first part
n=1
of the proof is now over.
∞
T
To prove the uniqueness of x, let y ∈ Fn . Again from (I),
n=1
d(x, y) ≤ diam(Fn ) < ϵ
Since ϵ is arbitrary positive real number, we must have d(x, y) = 0 implying that x = y.
The proof is now complete.
Solved Problems 01

Problem 2.2.5 If X is a discrete metric space, prove that X is complete.

Solution: Let (xn ) be a Cauchy sequence in X. Then there is N ∈ N such that
1
m, n ≥ N =⇒ d(xm , xn ) <
2
As d is discrete metric on X,
m, n ≥ N =⇒ d(xm , xn ) = 0

=⇒ (m, n ≥ N =⇒ xm = xn )

=⇒ xN = xN +1 = xN +2 = · · ·

=⇒ lim xn = xN
n→∞

=⇒ xn → xN in X
We proved that every Cauchy sequence in X is convergent. So X is complete.

A discrete metric space is complete.

Problem 2.2.6 Prove that R is complete metric space with Euclidean metric on it.
Solution: Let (xn ) be a Cauchy sequence in R. Then (xn ) is bounded. For each n ∈ N,
define
Mn = sup{xn , xn+1 , xn+2 , xn+3 , . . .}
and
mn = inf{xn , xn+1 , xn+2 , xn+3 , . . .}

S25011: REAL ANALYSIS Page 64

Then lim sup xn = lim Mn and lim inf xn = lim mn . As mn ≤ Mn for all n ∈ N,
n→∞ n→∞ n→∞ n→∞
lim inf xn ≤ lim sup xn . . . (I)
n→∞ n→∞
Take ϵ > 0. As (xn ) is Cauchy, there is n ∈ N such that
m, n ≥ N =⇒ |xm − xn | < ϵ
This implies that
ϵ
n ≥ N =⇒ |xN − xn | <
2
 
ϵ ϵ
=⇒ n ≥ N =⇒ xN − < xn < xN +
2 2
ϵ ϵ
=⇒ xN − is lower bound and xN + is upper bound of the set
2 2
{xn , xn+1 , xn+2 , xn+3 , . . .} for n ≥ N
 
ϵ ϵ
=⇒ n ≥ N =⇒ xN − ≤ mn ≤ Mn ≤ xN +
2 2
=⇒ (n ≥ N =⇒ Mn − mn ≤ ϵ)

=⇒ lim Mn ≤ lim mn + ϵ
n→∞ n→∞

=⇒ lim sup xn < lim inf xn + ϵ

n→∞ n→∞
As ϵ is arbitrary positive real number,
lim sup xn ≤ lim inf xn . . . (II)
n→∞ n→∞
From (I) and (II),
lim sup xn = lim inf xn
n→∞ n→∞
Hence the sequence (xn ) is convergent. We proved that every Cauchy sequence in R is
convergent. So R is complete.
Self-Test 01

Problem 2.2.7 Which of the following spaces is not complete?

A. the open interval (0, 1)

B. Rn

C. {0} ∪ { n1 : n ∈ N}

D. C(R)

Problem 2.2.8 If {In } is a decreasing sequence of closed and bounded intervals in R

such that inf(bn − an ) = 0 where In = [an , bn ] then which of the following is true?
T
A. In is empty
T
B. In has exactly one point
T
C. In has exactly two points

S25011: REAL ANALYSIS Page 65

 T
D. In has infinitely many points

Short Answer Questions 01

Problem 2.2.9 If xn → x in R and yn → y in R, prove that (xn , yn ) → (x, y) in R2 .

Solution: Let ϵ > 0. Then there are N1 , N2 ∈ N such that
ϵ
n ≥ N1 =⇒ |xn − x| <
2
and
ϵ
n ≥ N2 =⇒ |yn − y| <
2
Take N = max{N1 , N2 }. Then
ϵ
n ≥ N =⇒ |xn − x|, |yn − y| <
2
This implies that
n ≥ N =⇒ |xn − x| + |yn − y| < ϵ
 p  √
2
=⇒ n ≥ N =⇒ (xn − yn ) + (x − y) < ϵ 2 (∵ a2 + b2 ≤ |a + b| for any a, b ∈ R+ ∪ {0})


=⇒ n ≥ N =⇒ d((xn , yn ), (x, y)) < ϵ
Thus for ϵ > 0 there is N ∈ N such that
n ≥ N =⇒ d((xn , yn ), (x, y)) < ϵ
Hence (xn , yn ) → (x, y) in R2 .

Problem 2.2.10 Let (xn , yn ) → (x, y) in R2 . Prove that xn → x in R and yn → y in R.

Solution: For ϵ > 0, there is N ∈ N such that
n ≥ N =⇒ d((xn , yn ), (x, y)) < ϵ
This implies that
p
n ≥ N =⇒ (xn − yn )2 + (x − y)2 < ϵ

√ √
=⇒ n ≥ N =⇒ |xn − x|, |yn − y| < ϵ (∵ |a| = a2 ≤ a2 + b2 for any a, b ∈ R)
Thus for ϵ > 0 there is N ∈ N such that
n ≥ N =⇒ |xn − x|, |yn − y| < ϵ
This proves that xn → x in R and yn → y in R.

Summary:

• A metric space X is complete if and only if every Cauchy sequence in X is

convergent. This is equivalent to saying that X is complete if every Cauchy
sequence in X has a convergent subsequence.

• A closed subset of complete metric space is complete.

• A discrete metric space is complete.

• Rn is complete with Euclidean metric on it.

S25011: REAL ANALYSIS Page 66

End of Unit Exercises

Problem 2.2.11 Prove that R2 is complete metric space with Euclidean metric on it.

Problem 2.2.12 Prove that R equipped with the metric d(x, y) = |ex − ey | is not com-
plete metric space.

Problem 2.2.13 If (xn ) and (yn ) are Cauchy sequences in metric space X, prove that
d(xn , yn ) is Cauchy sequence in the Euclidean metric space R.

Keywords: Complete Metric Spaces, Cauchy Sequence, Discrete Metric Space.

REFERENCES

(1) Sections 6.3 and 6.4, Methods of Real Analysis by Richard R. Goldberg, 2nd
edition, Oxford & IBH Publishing Co. Pvt Ltd

(2) Sections 2.1 to 2.5, Topology of Metric Spaces by S. Kumaresan, 2nd edition,
Narosa Publishing House

(3) Chapters 2 and 3, Principles of Mathematical Analysis, by Walter Rudin, 3nd

edition, Mc Graw HillInc.,USA.

(4) Chapter 7, Real Analysis, by N.L. Carothers, 1st edition, Cambridge University
Press.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 67

 Unit 02-03: Totally bounded subsets of metric spaces

Learning Objectives:
After successful completion of this unit, you will be able to

i. Define the concept of totally bounded sets in metric spaces.

ii. State the relation between bounded and totally bounded sets in metric spaces.

iii. Define totally bounded sets using Cauchy sequences.

Introduction: A subset of metric space is bounded if it can be enclosed in some open

ball. A subset of metric space is totally bounded if it can be covered by finitely many
open balls of any given radius, however small. Every totally bounded set is bounded but
not conversely. A subset in Euclidean metric space Rn is bounded if and only if it is
totally bounded.

Definition 2.3.1 A subset Y of metric space X is called totally bounded if for every
n
S
ϵ > 0, there are finitely many points x1 , x2 , . . . , xn in X such that Y ⊆ B(xi , ϵ). In
i=1
this case we say that the set {x1 , x2 , . . . , xn } is ϵ dense in Y or that the set {x1 , x2 , . . . , xn }
is ϵ net for Y .

Theorem 2.3.2 A subset Y of metric space X is totally bounded if and only if for every
n
S
ϵ > 0 there are finitely many sets Y1 , Y2 , . . . , Yn such that Y ⊆ Yi and diam(Yi ) < ϵ
i=1
for i = 1, 2, . . . , n.
Proof: I Part: Let Y be totally bounded subset of X. Then there are finitely many
n
B(xi , 2ϵ ).Taking Yi = B(xi , 2ϵ ), we easily see
S
points x1 , x2 , . . . , xn in X such that Y ⊆
i=1
n
S
that Y ⊆ Yi and diam(Yi ) < ϵ for i = 1, 2, . . . , n.
i=1
II Part: Conversely suppose for every ϵ > 0 there are finitely many sets Y1 , Y2 , . . . , Yn
n
S
such that Y ⊆ Yi and diam(Yi ) < ϵ for i = 1, 2, . . . , n.. Take yi ∈ Yi . As diam(Yi ) < ϵ,
i=1
n
S n
S
Yi ⊆ B(xi , 2ϵ). This holds for all i = 1, 2, . . . , n. Since Y ⊆ Yi , Y ⊆ B(xi , 2ϵ). ϵ
i=1 i=1
was arbitrary. So it now follows that Y is totally bounded.

Theorem 2.3.3 If a subset Y of metric space X is totally bounded then Y is bounded.

Proof: As Y is totally bounded, there are subsets Y1 , Y2 , . . . , Yn of X such that Y ⊆
n
Yi and diam(Yi ) < 12 for i = 1, 2, . . . , n. Select points y1 , y2 , . . . , yn such that yi ∈ Yi
S
i=1
n
for i = 1, 2, . . . , n. Take any two points y and y ′ in Y . As Y ⊆ Yi , y ∈ Yi and y ′ ∈ Yj
S
i=1
for some i, j ∈ {1, 2, . . . , n}. Without loss of generality, let i ≤ j. Consider
d(y, y ′ ) ≤ d(y, yi ) + d(yi+1 , yi+2 ) + · · · + d(yj−1 , yj ) + d(yj , y ′ )

S25011: REAL ANALYSIS Page 68

 ≤ d(y, yi ) + d(y1 , y2 ) + · · · + d(yi+1 , yi+2 ) + · · · + d(yj−1 , yj ) + · · · d(yn−1 , yn ) + d(yj , y ′ )


 
1 1 1 1 1
< + + ··· + + (∵ diam(Yi ) < for i = 1, 2, . . . , n)
2 2 2 2 2
1 n−1 1
< + +
2 2 2
n+1
<
2
Thus d(y, y ′ ) < n+1
2
for all y, y ′ ∈ Y . This proves that Y is totally bounded.

Every totally bounded subset of metric space is bounded.

Theorem 2.3.4 Let Y be a subset of metric space X. If Y is totally bounded then every
sequence in Y has a Cauchy subsequence.
Proof: Let (yn ) be a sequence of points in Y . As Y is totally bounded, it can be covered
by finitely many subsets of Y of diameter less than 1. There is at least one of these sets
say Y1 which contains yn for infinitely many n’s. Choose n1 such that yn1 ∈ Y1 .
Again Y1 is totally bounded and can be covered by finitely many subsets of Y1 of diameter
less than 21 . There is at least one of these sets say Y2 which contains yn for infinitely many
n’s. Choose n2 > n1 such that yn2 ∈ Y2 .
Continuing in this way we have a sequence (Yn ) of subsets of Y such that Y1 ⊇ Y2 ⊇
Y3 ⊇ . . . and diam(Yn ) < n1 . Further (ynk ) is a sequence such that ynk , ynk+1 , . . . ∈ Yn for
any n. As diam(Yn ) < n1 , (ynk ) is a Cauchy sequence ( a subsequence of (yn )). We thus
proved that every sequence in Y has a Cauchy subsequence.

Theorem 2.3.5 (Bolzano-Weierstrass Theorem) Every bounded infinite subset of

R has a limit point in R.
Proof: Let S be a bounded infinite subset of R. As S is infinite, there is a sequence
(xn ) of points in S such that xm ̸= xn whenever m ̸= n. Since S is totally bounded,
by previous theorem there is a Cauchy subsequence of (xnk ) of the sequence (xn ). R is
complete and so (xnk ) is convergent in R. Let xnk → x in R. Then x is limit point of S
in R.

Theorem 2.3.6 Let Y be a subset of metric space X. If every sequence in Y has a

Cauchy subsequence then Y is totally bounded.
Proof: Suppose on contrary that Y is not totally bounded.Then there is some ϵ > 0
such that Y has no finite ϵ net for X. Take y1 ∈ Y . Then the set {y1 } does not form
ϵ net for Y . So there is y2 ∈ Y such that d(y1 , y2 ) ≥ ϵ. Again the set {y1 , y2 } does not
form ϵ net for Y . So there is y3 ∈ Y such that d(y1 , y3 ), d(y2 , y3 ) ≥ ϵ. Continuing in this
way, we can form a sequence (yn ) in Y such that d(ym , yn ) ≥ ϵ for all m, n ∈ N. Then

S25011: REAL ANALYSIS Page 69

(yn ) is a sequence in Y which doesn’t have any Cauchy subsequence. This contradicts
the hypothesis that every sequence in Y has a Cauchy subsequence. Hence Y must be
totally bounded.

A metric space is totally bounded if and only if every sequence in it has a Cauchy subsequence.

Solved Problems 01

Problem 2.3.7 Prove that a subset of Euclidean metric space R is bounded if and only
if it is totally bounded.
Solution: Let S ⊆ R. If S is totally bounded then S is bounded by the last theorem.
Conversely suppose that S is bounded. Then S ⊆ [−r, r] for some r > 0.Take ϵ > 0. By
Archimedian property of real numbers, there is n ∈ N such that n > 8rϵ so that n 4ϵ > 2r.
h i
For i = 1, 2, . . . , n,define the set Si as Si = −r + (i−1)ϵ
4
, −r + iϵ
4
. Then it is clear that
n
Si and diam(Si ) = 2ϵ < ϵ for i = 1, 2, . . . , n. As ϵ was arbitrary, we
S
S ⊆ [−r, r] ⊆
i=1
conclude that S is totally bounded.

A subset of Euclidean metric space R is bounded if and only if it is totally bounded.

Problem 2.3.8 Prove that discrete metric space X is totally bounded if and only if X
is finite set.
Solution:
I Part: Let X be a finite set with discrete metric on it. Take X = {x1 , x2 , . . . , xn } and
select ϵ > 0. Then the set X forms ϵ net for
n itself as
[
X⊆ B(xi , ϵ)
i=1
Thus X has ϵ net for any given ϵ > 0. Hence X is totally bounded.
II Part: Let < X, d > be a discrete metric space which is totally bounded. Suppose
on contrary that X is infinite set. Take ϵ = 21 . Then for any x1 , x2 , . . . , xn in X,
[ [ [
X ̸⊆ B(x1 , ϵ) B(x2 , ϵ) B(x3 , ϵ) . . . B(xn , ϵ) = {x1 , x2 , . . . , xn }
1
So there is no ϵ net for X when ϵ = 2
implying that X is not totally bounded. This is a
contradiction. Hence X must be a finite set.
Self-Test 01

Problem 2.3.9 The interval [0, ∞) in the discrete metrics space < R, d > is

A. both bounded and totally bounded.

B. bounded but not totally bounded.

C. neither bounded nor totally bounded.

S25011: REAL ANALYSIS Page 70

 D. none of these.

Problem 2.3.10 Pick out the subspace which is not totally bounded.

A. interval (0, 1) in the Euclidean space R.

B. rectangle [0, 1] × [0, 1] in the Euclidean space R2 .

C. closed unit ball in the set l2 of infinite sequences.

D. Cantor set in the Euclidean space R.

Short Answer Questions 01

Problem 2.3.11 Give an example of a metric space which is bounded but not totally
bounded.
Solution: Let < X, d > be a metric space where X = {2, 4, 6, 8, . . .} and d is a discrete
metric on X defined by 
 0 if x = y
d(x, y) =
 1 if x ̸= y
So < X, d > is a discrete metric space. Since X = B(1, 2), X is bounded. (X is enclosed
in an open ball with center 1 and radius 2).
A discrete metric space X is totally bounded if and only if X is finite (by previous
example). As X is infinite set, it is not totally bounded.

Problem 2.3.12 Prove that every finite subset of any metric space is totally bounded.
Solution: Let F = {x1 , x2 , . . . , xn } be a finite subset of metric space X. Take ϵ > 0.
Then the set F forms ϵ net for itself as n
[
F ⊆ B(xi , ϵ)
i=1
Thus F has ϵ net for any given ϵ > 0. Hence F is totally bounded.

Every finite subset of any metric space is totally bounded.

Summary:

• A finite set F in metric space X is said to form ϵ net for X if for any point x in
X, there is y ∈ F such that d(x, y) < ϵ. Stated differently, F is ϵ net for X if for
every point in X there is a point in F which is at a distance less than ϵ from it.

• A metric space X is totally bounded if there is ϵ net of X for any given ϵ > 0.

• Every totally bounded subset(subspace) of metric space is bounded.

• A subset of the Euclidean metric space R is totally bounded if and only if it is

bounded.

S25011: REAL ANALYSIS Page 71

 • A metric space is totally bounded if and only if every sequence in it has a Cauchy
subsequence.

End of Unit Exercises

Problem 2.3.13 Give an example to prove that totally boundedness is not preserved
under homeomorphisms of metric spaces.

Problem 2.3.14 Prove that a metric space X is totally bounded if and only if every
1
sequence (xn ) in X has a subsequence (xnk ) such that d(xnk , xnk+1 ) < 3k
for every k.

Keywords: Bounded Sets, Totally Bounded Sets, Cauchy Sequence, Subsequence, Bounded
subsets of R.

REFERENCES

(1) Section 6.3, Methods of Real Analysis by Richard R. Goldberg, 2nd edition,
Oxford & IBH Publishing Co. Pvt Ltd

(2) Sections 2.4 and 4.3, Topology of Metric Spaces by S. Kumaresan, 2nd edition,
Narosa Publishing House

(3) Chapter 2, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,

Mc Graw HillInc.,USA.

(4) Chapter 7, Real Analysis, by N.L. Carothers, 1st edition, Cambridge University
Press.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 72

 Unit 02-04: Compact Metric Spaces

Learning Objectives:
After successful completion of this unit, you will be able to

i. Define compact metric spaces.

ii. State equivalent definitions of compact metric spaces.

iii. Understand relations amongst closed, complete and compact metric spaces.

Introduction: A metric space is compact if it is complete and totally bounded or if

every sequence in it has a convergent subsequence. The compact metric space can also be
characterised using open covers. A metric space X is compact if and only if every open
cover of X has a finite subcover. A closed subset of compact metric space is compact
and a compact subset of any metric space is closed. Compact sets are preserved under
continuous maps. A subset in Euclidean metric space Rn is compact if and only if it is
closed and bounded.

Definition 2.4.1 A metric space X is called compact if X is complete and totally

bounded.

Theorem 2.4.2 A metric space X is compact if and only if every sequence in X has a
convergent subsequence.
Proof:
I Part: Let X be a compact metric space. We want to prove that every sequence in
X has a convergent subsequence. Let (xn ) be a sequence in X. As X is compact, it is
totally bounded. Hence the sequence (xn ) has a Cauchy subsequence say (xnk ). Also X
is complete as it is compact. The sequence (xnk ) is thus a Cauchy sequence in complete
metric space X. Hence (xnk ) must be convergent in X. We thus have a subsequence
(xnk ) of the sequence (xn ) which is convergent in X.
II Part: Suppose every sequence in X has a convergent subsequence. We want to prove
that X is compact i.e. X is complete and totally bounded.
Every convergent sequence is Cauchy. So every sequence in X has a Cauchy subsequence.
Hence X is totally bounded.
To prove that X is complete, take a a Cauchy sequence (xn ) in X. By hypothesis it
has a convergent subsequence say (xnk ). Let xnk → x in X. If a Cauchy sequence has
a convergent subsequence then the sequence is also convergent. It now follows that the
sequence (xn ) converges to x in X. We proved that every Cauchy sequence in X is
convergent. Hence X is complete.

S25011: REAL ANALYSIS Page 73

A metric space is compact if and only if every sequence in it has a convergent subsequence.

Theorem 2.4.3 If Y is a closed subset of compact metric space X then Y is compact.

Proof: Consider the following argument.
(yn ) is a sequence of points in Y
=⇒ (yn ) is a sequence of points in X (∵ Y ⊆ X )
=⇒ (yn ) has a convergent subsequence say (ynk ) in Y converging to a point say y in X
The above step follows from the previous theorem as X is compact. This further implies
that
y is a limit point of Y
=⇒ y ∈ Y (∵ Y is closed)
Thus ynk → y in Y . We proved that every sequence in Y has a subsequence which
converges in Y . It now follows by previous theorem that Y is compact.

A closed subset of compact metric space is compact.

Theorem 2.4.4 If Y is a compact subset of metric space X then Y is closed in X.

Proof: Let y be a limit point of Y . Then there is a sequence (yn ) in Y which converges
to y in X. Every convergent sequence is Cauchy. So (yn ) is Cauchy sequence in Y . Y is
compact and hence complete. So the sequence (yn ) must be convergent (and convergent
to y) in Y . It now follows that y ∈ Y . Thus Y contains all its limit points and so is
closed.

A compact subset of any metric space is closed.

Definition 2.4.5 Let X be a metric space and τ = {Ai : i ∈ I} be a collection of some

Ai then τ is called open cover of X. If τ ′ = {A1 , A2 , A3 , . . . An }
S
open sets in X. If X ⊆
i∈I
n
Ai then τ ′ is called finite subcover of X.
S
is finite subset of τ such that X ⊆
i=1

Theorem 2.4.6 If X is compact metric space then every open cover of X has a finite
subcover.
Proof: Suppose on contrary that τ = {Ai : i ∈ I} is an open cover of X which does not
have any finite subcover. X is totally bounded as X is compact. So there are finitely
many closed and bounded subsets of X which cover X and whose diameters are less than
1. (Note that if sets are not closed then we would replace them by their closures and
diam(S̄) = diam(S).) Then at least one of these sets say S1 which can’t be covered by
finitely many sets in τ .
Again S1 is totally bounded and can be covered by finitely many closed and bounded

S25011: REAL ANALYSIS Page 74

subsets of S1 of diameter less than 12 . There is at least one of these sets say S2 which
can’t be covered by finitely many sets in τ .
Continuing in this way we have a sequence (Sn ) of closed subsets of X such that S1 ⊇ S2 ⊇
∞
S3 ⊇ . . . and diam(Sn ) < n1 . As X is compact and hence complete, the set
T
̸ ϕ
Sn =
n=1
∞
T S
by Cantor’s intersection theorem. Take x ∈ Sn . As x ∈ X ⊆ Ai , x ∈ Ai for some
n=1 i∈I
i ∈ I. Since Ai is open, there is r > 0 such that B(x, r) ⊆ Ai . By Archimedian property
1 1 1
of real numbers, there is N ∈ N such that N > r
i.e. r < N
. Since diam(SN ) < r < N
,
SN ⊆ B(x, r) ⊆ Ai . So Ai is a single set in τ which covers SN . This contradicts the fact
that SN can’t be covered by finitely many sets in τ .Hence our assumption is wrong. So
every open cover of X must have a finite subcover.

Theorem 2.4.7 If every open cover of metric space X has a finite subcover then X is
compact.
Proof: Let (xn ) be a sequence in X which has no convergent subsequence. Then for
each x ∈ X there exists ϵx > 0 such that d(xn , x) < ϵx only for finitely many n’s. Without
loss of generality we could assume that 0 < d(xn , x) < ϵx for no n but is still possible
that xn = x for finitely many n’s. The collection {B(x, ϵx ) : x ∈ X} is an open cover of
m
S
X. By hypothesis, there exist finitely many xn ’s such that X = B(xk , ϵxk ). But each
k=1
m
S
B(xk , ϵxk ) contains only finitely many xn ’s. Therefore B(xk , ϵxk ) = X contains only
k=1
finitely many xn . This is a contradiction. Hence every sequence (xn ) in X must have
convergent subsequence proving that X is compact.
By virtue of last two theorems,

A metric space X is compact if and only if every open cover of X has a finite subcover.

Theorem 2.4.8 Let X be a compact metric space and f : X → Y be a continuous

function from metric space X to metric space Y . Then f (X) is compact.
Proof: Let τ = {Ai : i ∈ I} be an open cover of f (X). Since each Ai is open
in Y and f is continuous, f −1 (Ai ) is open in X for all i ∈ I. So τ ′ = {f −1 (Ai ) :
i ∈ I} is an open cover of X. As X is compact, it must have a finite subcover say
{f −1 (A1 ), f −1 (A2 ), . . . , f −1 (An )}. Then {A1 , A2 , . . . , An }. is a finite subcover of f (X).
We proved that every open cover of f (X) has a finite subcover. Hence f (X) is compact.

A continuous image of compact space is compact.

Theorem 2.4.9 Let X be a compact metric space and R be equipped with Euclidean
metric. If f : X → R is continuous then f attains its maximum and minimum values at
points of X.

S25011: REAL ANALYSIS Page 75

Proof: X is compact and f : X → R is continuous. The continuous image of compact
space is compact. So f (X) is compact and hence bounded subset of R. Let
M = sup{f (x) : x ∈ X}
Suppose on contrary that f (x) ̸= M for any x ∈ X. Then f (x) < M for all x ∈ X.
Define the function h : X → R as
1
h(x) =
M − f (x)
The continuity of f implies the continuity of h. So h is continuous function on compact
metric space X and hence h(X) must be bounded. So there is K > 0 such that
h(x) ≤ K
for all x ∈ X. This implies that
1
≤ K for all x ∈ X
M − f (x)
1
=⇒ f (x) ≤ M − for all x ∈ X
K
1
=⇒ M − upper bound of {f (x) : x ∈ X}
K
1
=⇒ M ≤ M − (∵ M is supremum of {f (x) : x ∈ X})
K
1
=⇒ 0 ≤ −
K
=⇒ 0 ≤ −1
This is a contradiction. Hence f (x) = M for some x ∈ X. This proves that f attains its
maximum value at a point of X. The proof for minimum value is similar and is left to
the reader.

A continuous real valued function on compact metric space attains its bounds.

Solved Problems 01

Problem 2.4.10 Prove that every finite subset of any metric space is compact.
Solution: Let F = {x1 , x2 , . . . , xn } be a finite subset of metric space X. Take τ = {Ui :
S
i ∈ I} as an open cover of F . Then each Ui is open in X and F ⊆ Ui . So for each xk
i∈I
in F , there is ik ∈ I such that xk ∈ Uik . This implies that
[ [ [
F = {x1 , x2 , . . . , xn } ⊆ Ui1 Ui2 . . . Uin
. So If τ ′ = {U1 , U2 , U3 , . . . Un } is a finite subcover of τ which cover F . We thus proved
that every open cover of F has a finite subcover. Hence F is compact in X.

Every finite subset of any metric space is compact.

S25011: REAL ANALYSIS Page 76

Problem 2.4.11 Give an open cover of (0, 1) in the Euclidean metric space R which does
not have any finite subcover. Hence conclude that (0, 1) is not compact in the Euclidean
metric space R.
Solution: For each x ∈ (0, 1), let Ux = ( x2 , x+1
2
). Then each Ux is an open interval
(open set) in R and τ = {Ux : x ∈ (0, 1)} is an open cover of (0, 1).
Suppose on contrary that there is a finite subcover say τ ′ = {U1 , U2 , . . . , Un } of τ which
cover (0, 1). Note that Ui = x2i , xi2+1 for i = 1, 2, . . . , n. Define xmin as xmin =

xmin
min{x1 , x2 , . . . , xn } and take y as 0 < y < 2
. Then y ̸∈ Ui for i = 1, 2, . . . , n. This
contradicts the fact that τ ′ = {U1 , U2 , . . . , Un } is a finite subcover of τ which cover (0, 1).
We conclude that τ is an open cover of (0, 1) which does not have any finite subcover. A
metric space is compact only if every open cover of it has a finite subcover. We may now
conclude that (0, 1) is not compact in the Euclidean metric space R.
Self-Test 01

Problem 2.4.12 Let f be a continuous real valued function on [0, 1] and {(x, f (x)) : x ∈
[0, 1]} be its graph. Then

A. f is continuous implies its graph is an open set in R2

B. f is continuous implies its graph is an compact set in R2

C. f has at least one fixed point

D. f is continuous implies f is an open map

Problem 2.4.13 Which of the following subspaces of the Euclidean metric space R is
compact?

A. R

B. {1, 2, 3, 4, 5, 6}

C. (0, 1)

D. [2, 4] − {3}

Problem 2.4.14 Which of the following is a compact subset of the Euclidean metric
space R2 ?

A. the X−axis

B. the set of all points of R2 having integer coordinates

C. the set {(x, y) : |x| + |y| ≤ 1}

D. the set {(x, y) : y 2 = x}

S25011: REAL ANALYSIS Page 77

Problem 2.4.15 Consider a discrete metric space Rd . For a < b,

A. all closed intervals [a, b] are compact

B. all closed intervals [a, b] are connected

C. no non-trivial closed interval [a, b] is compact

D. every closed subset of closed interval [a, b] is compact

Problem 2.4.16 Let S = {(x, y) ∈ R2 : −1 ≤ x ≤ 1 and − 1 ≤ y ≤ 1} and T =

S − {(0, 0)} , the set obtained by removing the origin from S. Let f be a continuous
function from T to R. Select the false option.

A. Image of f must be connected

B. Image of f must be compact

C. Any such a continuous function f can be extended to a continuous function

from S to R

D. If f can be extended to a continuous function from S to R then image of f is

bounded

Short Answer Questions 01

Problem 2.4.17 Consider R with Euclidean metric on it.

a) Give an example of a subspace of R which is connected but not compact.
b) Give an example of a subspace of R which is compact but not connected.
Solution:
a) The set S = (0, 1) is connected as it is interval. It is not compact as it is not closed
(or complete).
b) The set T = {0, 1} is connected as it is finite set. It is not connected as it is not
interval in R. Recall that the only connected subsets of R are intervals.

Problem 2.4.18 Give an example of a function f : [10, 20) → R which attains a minimum
value but not the maximum value.
Solution: Define f : [10, 20) → R as
f (x) = x
f is increasing function on [10, 20) and f clearly attains its minimum value at x = 10
and the minimum value is f (10) = 10.
20+x
For any x ∈ [10, 20), take y = 2
. Then y ∈ [10, 20) and y > x. Hence f (y) > f (x).
Thus for every x ∈ [10, 20), there is y ∈ [10, 20) such that f (y) > f (x).Hence f does not
attain its maximum value at any point of [10, 20).

S25011: REAL ANALYSIS Page 78

Summary:

• A metric space X is compact if it is complete and totally bounded.

• A metric space X is compact if and only if every sequence in X has a convergent

subsequence.

• A closed subset of compact metric space is compact.

• An open cover of a metric space X is a collection τ of open sets in X such that

S
X ⊆ U . If there are finitely many sets in τ that cover X then τ is said to
U ∈τ
have a finite subcover for X.

• A metric space X is compact if and only if every open cover of X has a finite
subcover.

• A continuous image of compact space is compact.

• A continuous real valued function on compact metric space attains its bounds.

End of Unit Exercises

Problem 2.4.19 Prove that Cantor set in the Euclidean metric space R is compact.

Problem 2.4.20 Let K be a compact subset of the Euclidean metric space R. Prove that
sup K and inf K exist and are in K.

Problem 2.4.21 Give an example to prove that infinite union of compact sets may not
me compact.

Problem 2.4.22 If K1 and K2 are compact subsets of metric space X, prove that K1 ∪K2
is also compact subset of X.

Keywords: Compact Metric Space, Closed Subset of Compact Metric Space, Open
Cover,Subcover, Continuous Functions on Compact Spaces.

REFERENCES

(1) Sections 6.5 to 6.8, Methods of Real Analysis by Richard R. Goldberg, 2nd edi-
tion, Oxford & IBH Publishing Co. Pvt Ltd

(2) Chapter 4, Topology of Metric Spaces by S. Kumaresan, 2nd edition, Narosa

Publishing House

S25011: REAL ANALYSIS Page 79

 (3) Chapter 2, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,
Mc Graw HillInc.,USA.

(4) Chapter 8, Real Analysis, by N.L. Carothers, 1st edition, Cambridge University
Press.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 80

CREDIT 03

81
 Unit 03-01: Riemann Integral

Learning Objectives:
After successful completion of this unit, you will be able to

i. Understand the concept of sets of measure 0.

ii. Define lower integrals, upper integrals and Riemann integrals using lower and
upper sums of bounded functions.

Introduction: A measure of set is its size. It is different from the cardinality of sets.
Two sets may have same cardinality but different measures or vice-versa. For intervals
in R, the length of interval is the measure. For subsets of R2 , area is the measure. For
subsets of R3 , volume is the measure. The concept of integrating function f over set S
can be thought as follows.
A set S is divided into smaller parts. On each part, the function f is considered to be
P
constant.The sum of products of value of f and measure part is evaluated. (value of f×
measure of part) The integration of f is obtained by taking limit of this sum as number
of parts increases to infinity (i.e. parts become infinitesimal and the function can really
be assumed to constant on it)
R P
f = limit of (value of f × measure of part)
S

In case of Riemann integration, S = [a, b] is partitioned into subintervals and in limiting

position there are infinitely many infinitesimal subintervals on which f becomes constant.
The sum of products of values of f and lengths of subintervals in this limiting position is
nothing but the value of Riemann integral of f over [a, b]. With little efforts of visualisa-
tion or doing it on paper(by drawing graph of f (x) and splitting [a, b] into subintervals),
the reader may find out that the value of Riemann integration of f over [a, b] is nothing
but the area under a curve of f (x) enclosed between two vertical lines x = a and x = b.

Definition 3.1.1 A subset E of R is of measure zero if for every ϵ > 0 there are finite or
countably many open intervals I1 , I2 , . . . such that
[ X
E⊆ In and |In | ≤ ϵ
n n

[ X
m(E) = 0 ⇐⇒ ∀ϵ > 0 ∃In such that E ⊆ In and |In | < ϵ
n n

If there are finitely many intervals Im

1 , I2 , . . . , Im then
m
[ X
E⊆ In and |In | < ϵ
n=1 n=1

S25011: REAL ANALYSIS Page 82

If there are countably many interval∞I1 , I2 , . . . then
∞
[ X
E⊆ In and |In | < ϵ
n=1 n=1

Theorem 3.1.2 A finite subset of R is of measure 0.

Proof: Let E={x1 , x2 , . . .} be a finite subset of R. Let ϵ > 0. For k = 1, 2, . . . , n; take

n
ϵ ϵ

S
Ik = xk − 4n , xk + 4n . Then E = {x1 , x2 , . . . , xn } ⊆ Ik . Consider
k=1
n n
X X ϵ
|Ik | =
k=1 k=1
2n
n
ϵ X
= 1
2n k=1
ϵ
= ·n
2n
ϵ
=
2
<ϵ
Thus for every ϵ > 0 there are finitely many open interval which cover E and their length
add up to less than ϵ > 0. Hence the set E is of measure zero.

A finite subset of R is of measure 0.

Theorem 3.1.3 Every countable subset of R has measure zero.

Proof: Let E = {x1 , x2 , . . .} be a countable subset of R.

ϵ ϵ ϵ
Take ϵ > 0 for n = 1, 2, 3, . . . and consider I1 = 22
, I2 = 23
, . . . , In = 2n+1
.
Therefore the corresponding interval
 are 
ϵ ϵ
I1 = x1 − 2 , x1 +
2 22
 
ϵ ϵ
I2 = x2 − 3 , x2 +
2 23
...
 
ϵ ϵ
In = xn − , xn +
2n+1 2n+1
...
∞
S
Then E = {x1 , x2 , . . .} ⊆ |In |. Consider
n=1
∞ ∞
X X ϵ
|Ik | =
n=1 n=1
2n+1
∞
X ϵ
=
n=1
2n+1

S25011: REAL ANALYSIS Page 83

 ∞
X 1
=ϵ
n=1
2n+1
1
22
=ϵ 1
1− 2
ϵ
=
2
<ϵ
Thus for every ϵ > 0 there are countably many open interval which cover E and their
length add up to less than ϵ. Hence the set E is of measure zero.

Every countable subset of R has measure zero.

n
S
Theorem 3.1.4 If E1 , E2 , . . . , En are sets of measure 0 then Ek is of measure 0.
k=1
Sn
Proof: Let, E = Ek = E1 ∪ E2 ∪ . . . ∪ En and take ϵ > 0. Since E1 is of measure
k=1
0 then are finitely/countably many open interval which cover Ek and whose length add
ϵ
up to less than n
. This holds for k = 1, 2, . . . , n.Take union of all such open intervals.
A finite union of finite/countable sets is finite/countable. So the union of all these open
intervals is countable/finite.This union covers E and further their length add up to less
ϵ ϵ ϵ
than n
+ n
+ ··· + n
=ϵ
Thus for every ϵ > 0 there are finitely/countably many open intervals which cover E and
their length add up to less than ϵ > 0. Hence the set E is of measure zero.

A finite union of sets of measure 0 is of measure 0.

∞
S
Theorem 3.1.5 If E1 , E2 , . . . , En are sets of measure 0 then En is of measure 0.
n=1
∞
S
Proof: Let, E = Ek = E1 ∪ E2 ∪ · · · ∪ En · · · and take ϵ > 0. Since En is of measure
n=1
0 then are finitely/countably many open interval which cover En and whose length add
ϵ
up to less than 2n
. This holds for n = 1, 2, . . . ,. Take union of all such open intervals.
A countable union of finite/countable sets is finite/countable. So, the union of all these
open interval is countable. This union covers E and further their length add up to less
ϵ ϵ ϵ
than 2
+ 22
+ ··· + 2n
+ . . . = ϵ.
Thus for every ϵ > 0 there are countably many open intervals which cover E and their
length add up to less than ϵ. Hence the set E is of measure zero.

A countable union of sets of measure 0 is of measure 0.

Results/Remarks:
i) A finite set is of measure 0.
ii) A countable sets is of measure 0.

S25011: REAL ANALYSIS Page 84

iii) A finite union of sets of measure 0 is of measure 0.
iv) A countable union of sets of measure 0 is of measure 0.
v) A subset of set of measure 0 is of measure 0.
vi) An uncountable set may be of measure 0 (for example: Cantor set).

Definition 3.1.6 A statement or property is said to hold almost everywhere (a.e.) in

[a, b] if the set of points of [a, b] where the statement or property does not hold is of
measure 0.

Problem 3.1.7 Let f : [a, b] → R and D = {x ∈ [a, b] : f is continuous at x}. If f is

continuous almost everywhere on [a, b] then m(D) = 0. Conversely, if m(D) = 0 then f
is continuous a.e. on [a, b].

Definition 3.1.8 Let I be a bounded interval and f : I → R be a bounded function on

I. We define
m[f ; I] = inf f (x)
x∈I
M [f ; I] = sup f (x)
x∈I
ω[f ; I] = M [f ; I] − m[f ; I]

Problem 3.1.9 Consider the function f : [0, 2π] → R defined by f (x) = sin x. Here for
the function f on I = [ −π , π ]; m[f ; I] = −1, M [f ; I] = 1 and ω[0; I] = 1 − (−1) = 2.
2 2

Definition 3.1.10 Let [a, b] be a closed bounded interval. By a partition of [a, b], we
mean a finite set {a = x0 , x1 , x2 , x3 , . . . , xn = b} of points in [a, b] such that a = x0 <
x1 < x2 < x3 < . . . < xn = b. The intervals Ik = [xk−1 , xk ] are called sub-intervals of the
partition.

Definition 3.1.11 Let σ = {a = x0 , x1 , x2 , . . . , xn = b} be a partition of [a, b]. Let f :

[a, b] → R be a bounded function. Ik denote the sub-interval [xk−1 , xk ] for k = 1, 2, . . . , n
The lower and upper sum of f with respect to partition σ are denoted by L[f ; σ]and
U [f ; σ] respectively and are defined as
n
X
L[f ; σ] = m[f : Ik ]|Ik |
k=1
n
X
U [f ; σ] = M [f : Ik ]|Ik |
k=1

Definition 3.1.12 Let f : [a, b] → R be a bounded function. The lower integral of f over
[a, b] is defined as
Z b
f = sup{L[f ; σ] : σ is partition of [a, b]}
a

S25011: REAL ANALYSIS Page 85

or simply as Z b
f = sup L[f ; σ]
σ a
Lower integral is supremum of all lower sums.

Definition 3.1.13 Let f : [a, b] → R be a bounded function. The upper integral of f

over [a, b] is defined as Z
b
f = inf{U [f ; σ] : σ is partition of [a, b]}
a
or simply as Z b
f = inf U [f ; σ]
a σ
Upper integral is infimum of all upper sums.
Rb Rb
Definition 3.1.14 Let f : [a, b] → R be a bounded function. If a
f = a
f then the
function f is said to be Riemann integrable and in this case we define
Zb Z b Z b
f= f= f
a a
a
The set of all Riemann integrable function over [a, b] is denoted by R[a, b].
( Z b Z b )
R[a, b] = f : [a, b] → R : f= f
a a

Definition 3.1.15 A partition σ ∗ is called refinement of partition σ of [a, b] if σ ⊆ σ ∗ .

Theorem 3.1.16 Let f : [a, b] → R be a bounded function. If σ and τ are any two
partitions of [a,b] then U [f ; σ] ≥ L[f ; τ ].
Proof: The proof is based on the following idea:
“ A refinement of partition is obtained by adding points to the partition. If σ ∗ is refine-
ment of σ then every subinterval of σ will also be subinterval of σ ∗ but not conversely.
Lower sums get increased and upper sums get decreased on taking refinement of parti-
tions”.
Let σ ∗ be the refinement of σ obtained by adding a single point of σ. Take σ =
{x0 , x1 , . . . , xk−1 , xk , xk+1 . . . , xn } and σ ∗ = {x0 , x1 , . . . , xk−1 , xk ∗, xk , xk+1 . . . , xn } Let Ij
denote the subinterval [xj−1 , xj ] for j = 1, 2, . . . , n. Let Ik∗ = [xk−1 , x∗k ] and Ik∗∗ = [x∗k , xk ]
It is clear that Ik = [xk−1 , xk ] = Ik∗ ∪ Ik∗∗ and |Ik | = |Ik∗ | + |Ik∗∗ |.Also
M [f : Ik∗ ] ≤ M [f : Ik ] and M [f : Ik∗∗ ] ≤ M [f : Ik ] . . . (I)

Consider n
X
U [f ; σ ∗ ] = M [f : Ij ]|Ij | + M [f : Ik∗ ]|Ik∗ | + M [f : Ik∗∗ ]|Ik∗∗ |
j=1
n
X
≤ M [f : Ij ]|Ij | + M [f : Ik ]|Ik∗ | + M [f : Ik ]|Ik∗∗ |
j=1 j̸=k

S25011: REAL ANALYSIS Page 86

 n
X
M [f : Ij ]|Ij | + M [f : Ik ] |Ik∗ | + |Ik∗∗ |


=
j=1 j̸=k
n
X

= M [f : Ij ]|Ij | + M [f : Ik ] Ik |
j=1 j̸=k
n
X
= M [f : Ij ]|Ij |
j=1

= U [f ; σ]
U [f ; σ ∗ ] ≤ U [f ; σ]
By induction we can generalise the result for any refinement of σ. Thus
σ ∗ ⊇ σ =⇒ U [f ; σ ∗ ] ≤ U [f ; σ] . . . (II)
Similarly
τ ∗ ⊇ τ =⇒ L[f ; τ ] ≤ L[f ; τ ∗ ] . . . (III)
Taking σ ∗ = σ ∪ τ in (II)
U [f ; σ ∪ τ ] ≤ U [f ; σ] . . . (IV )
Taking τ ∗ = σ ∪ τ in (III)
L[f ; τ ] ≤ L[f ; σ ∪ τ ] . . . (V )
It is clear that
L[f ; σ ∪ τ ] ≤ U [f ; σ ∪ τ ] . . . (V I)
From (IV),(V) and (VI)
L[f ; τ ] ≤ L[f ; σ ∪ τ ] ≤ U [f ; σ ∪ τ ] ≤ U [f ; σ]
This proves that
L[f ; τ ] ≤ U [f ; σ]

Solved Problems 01

Problem 3.1.17 Let f (x) = x on [0, 1] and σ = {0, 31 , 23 , 1} be the partition of [0, 1].
Compute U [f, σ] and L[f, σ].

Solution: I1 = [0, 31 ], I2 = [ 13 , 23 ], I3 = [ 32 , 33 ].
U [f, σ] = M [f, I1 ]|I1 | + M [f, I2 ]|I2 | + M [f, I3 ]|I3 |
     
1 1 2 1 1
= + +1
3 3 3 3 3
1 2 1
= + +
9 9 3
2
=
3
and
L[f, σ] = m[f, I1 ]|I1 | + m[f, I2 ]|I2 | + m[f, I3 ]|I3 |

S25011: REAL ANALYSIS Page 87

      
1 1 1 2 1
=0 + +
3 3 3 3 3
1 2
=0+ +
9 9
1
=
3
Problem 3.1.18 Let f (x) = x2 on [0, 1] and σ = {0, 13 , 23 , 1} be the partition of [0, 1].
Find U [f, σ] and L[f, σ].
Solution: I1 = [0, 31 ], I2 = [ 13 , 23 ], I3 = [ 32 , 33 ].
U [f, σ] = M [f, I1 ]|I1 | + M [f, I2 ]|I2 | + M [f, I3 ]|I3 |
     
1 1 4 1 1
= + +1
9 3 9 3 3
1 4 1
= + +
27 27 3
14
=
27
and
L[f, σ] = m[f, I1 ]|I1 | + m[f, I2 ]|I2 | + m[f, I3 ]|I3 |
     
1 1 1 4 1
=0 + +
3 9 3 9 3
1 4
=0+ +
27 27
5
=
27

Self-Test 01

Problem 3.1.19 If f : [a, b] → R is bounded function then

R1 R1
A. 0 f < 0 f
R1 R1
B. 0 f > 0 f
R1 R1
C. 0 f ≤ 0 f
R1 R1
D. 0 f ≥ 0 f

Problem 3.1.20 Define f on [0, 1] as 

 x2 if x ∈ Q
f (x) =
 x3 if x ̸∈ Q
Then

A. f is not Riemann integrable on [0, 1]

R1 1
B. f is Riemann integrable on [0, 1] and f (x)dx = 4
0

S25011: REAL ANALYSIS Page 88

 R1 1
C. f is Riemann integrable on [0, 1] and f (x)dx = 3
0
R1 R1
D. 0
f> 0
f

Problem 3.1.21 Which of the following subset of R2 has finite positive Lebesgue measure?

A. S1 = {(x, y) : x2 + y 2 > 1}

B. S2 = {(x, y) : x2 + y 2 < 1}

C. S3 = {(x, y) : x = y} +{(x, y) : x = −y}

D. S4 = {(x, y) : x = y} +{(x, y) : x = y}

Short Answer Questions 01

Problem 3.1.22
a) Show that the set of all rational numbers is of measure zero.
b) Show that the set of all irrational numbers is not of measure zero.
Solution:
a) Every singleton set is of measure zero. A countable union of sets of measure zero is of
S
measure zero. Since Q = {q}, The set Q of rationals is of measure zero.
q∈Q
b) Suppose on contrary that the set Qc = R − Q of irrationals is of measure zero. Then
R = Q ∪ Qc is of measure zero as finite union of sets of measure zero is of measure zero.
This is a contradiction as R is not of finite measure. Hence the set Qc of irrationals is
not of measure zero.

Problem 3.1.23 State True or False with justification. “Let f be continuous on [0, 1]
and g(x) = f (x) at almost every x ∈ [0, 1]. Then g is continuous almost everywhere in
[0, 1].”
Solution: FALSE. For example, take f : R → R as f (x) = 1 for all x ∈ R and
g : R → R as 
 0 if x ∈ Q
g(x) =
 1 if x ̸∈ Q
Then f is continuous on [0, 1], and g(x) = f (x) except on Q. Since Q is of measure zero,
g(x) = f (x) at almost every x ∈ [0, 1]. Here g(x) is discontinuous everywhere in [0, 1].

Summary:

• By measure of set E, we mean the size of E. If E is an interval then its measure

is its length (distance between end-points). If E is rectangle including its interior
then its measure is area of E. If E is cube then its measure is volume of E.

S25011: REAL ANALYSIS Page 89

 • Size of all sets can’t be measured. So the non-measurable sets exist.

• m(E) denotes measure of E.

• A set E is of measure zero if it is of negligible size. It is easy to see that

all countable sets are of measure zero. Further the countable union of sets of
measure zero is of measure zero. The precise definition of sets of measure zero
is given in the next point.

• A subset E of R is said to be of measure zero if for every ϵ > 0 there are countable
number of open intervals In such that E ⊆ ∪n In and |In | < ϵ.

• m(E) = 0 ⇐⇒ ∀ϵ > 0∃ countable number of open intervals that cover E and

whose length add up to be less than ϵ.

• A statement is said to hold at almost every (a.e.) point of E if the set of points
of E where the statement does not hold is of measure zero.

• A partition σ of closed and bounded interval [a, b] is a finite subset {x0 , x1 , . . . , xn }

of [a, b] such that a = x0 < x1 < x2 < . . . < xn = b. This partition of [a, b] gives
a division of [a, b] into subintervals [x0 , x1 ], [x1 , x2 ], [x3 , x4 ], . . . , [xn−1 , xn ].

• A refinement τ of partition σ is obtained by adding more points to the partition

σ. Precisely, τ is refinement of σ if σ ⊆ τ .

• Let f be the bounded function on the closed and bounded interval [a, b] and σ
be a partition of [a, b] and I1 , I2 , . . . , In be corresponding subintervals. The lower
sum of f over σ is defined as n
X
L[f ; σ] = mk |Ik |
k=1
where mk = inf f (x) = infimum of f (x) over Ik .
x∈Ik

• Let f be the bounded function on the closed and bounded interval [a, b] and σ be
a partition of [a, b] and I1 , I2 , . . . , In be corresponding subintervals. The upper
sum of f over σ is defined as n
X
U [f ; σ] = Mk |Ik |
k=1
where Mk = sup f (x) = supremum of f (x) over Ik .
x∈Ik
Since mk ≤ Mk ,
L[f ; σ] ≤ U [f ; σ]

• If σ ∗ is refinement of σ then L[f ; σ] ≤ L[f ; σ∗].

σ ⊆ σ ∗ =⇒ L[f ; σ] ≤ L[f ; σ∗]

S25011: REAL ANALYSIS Page 90

 So the lower sum of refinement has more value.

• If σ ∗ is refinement of σ then U [f ; σ] ≥ Uσ∗ f .

σ ⊆ σ ∗ =⇒ U [f ; σ] ≥ Uσ∗ f
So the upper sum of refinement has less value.

• Let σ and τ be any partitions of [a, b], let ξ be a refinement of both σ and τ . For
a bounded function f on [a, b], we have
L[f ; τ ] ≤ L[f ; ξ] ≤ U [f ; ξ] ≤ U [f ; σ]
Observe that any lower sum of any partition less equals to any upper sum of any
partition.

• Let f be a bounded function on closed bounded interval [a, b]. The lower integral
Rb
a
f (x)dx is defined as
Z b
f (x)dx = sup L[f ; σ]
a σ∈Ω
where Ω is the collection of all partitions of [a, b] and sup denotes supremum
(least upper bound). The lower integral is obtained by taking supremum of all
lower sums over all partitions of [a,Zb]. If σ is any partition of [a, b] then
b
L[f ; σ] ≤ f (x)dx
a
The lower integral greater equals to any lower sum.

• Lower Sums goes on increasing on refining partitions and ultimately equals to

lower integral.

• Let f be a bounded function on closed bounded interval [a, b]. The upper integral
Rb
a
f (x)dx is defined as Z
b
f (x)dx = inf U [f ; σ]
a σ∈Ω
where Ω is the collection of all partitions of [a, b] and inf denotes infimum (great-
est lower bound). The upper integral is obtained by taking infimum of all upper
sums over all partitions of [a, b]. If σZ is any partition of [a, b] then
b
U [f ; σ] ≥ f (x)dx
a
The upper integral less equals to any upper sum.

• Upper Sums goes on decreasing on refining partitions and ultimately equals to

Upper integral.

• If σ is any partition of [a,

Z bb] then Z b
L[f ; σ] ≤ f (x)dx ≤ f (x)dx ≤ U [f ; σ]
a a
Lower Sum ≤ Lower Integral ≤ Upper Integral ≤ Upper Sum

S25011: REAL ANALYSIS Page 91

 • A bounded function f is Riemann integrable on [a, b] if its lower and upper
integrals are equal. Z b Z b
f is Riemann integrable ⇐⇒ f (x)dx = f (x)dx
a a

• If f is Riemann integrable then

Lower Sums goes on increasing on refining partitions and ultimately equals to
Riemann integral
and
Upper Sums goes on decreasing on refining partitions and ultimately equals to
Riemann integral

• If f is Riemann integral then the value of lower (or upper integral) is denoted
Rb
by f (x)dx.
a
So if f is Riemann integrable then
Rb Rb Rb
f (x)dx = a f (x)dx = a f (x)dx
a

• The collection of all Riemann integrable functions on [a, b] is denoted by R[a, b].
f is Riemann integrable ⇐⇒ f ∈ R[a, b].

End of Unit Exercises

Problem 3.1.24 Let A be not of measure zero. If B ⊂ A and if B is of measure zero,

prove that A − B is not of measure zero.

Problem 3.1.25 Let f (x) = x on [0, 1] and σn = {0, n1 , n2 , . . . , nn } be the partition of [0, 1].
Find lim U [f, σn ] and lim L[f, σn ].
n→∞ n→∞

Problem 3.1.26 Let f (x) = x2 on [0, 1] and σn = {0, n1 , n2 , . . . , nn } be the partition of

[0, 1]. Evaluate lim U [f, σn ] and lim L[f, σn ].
n→∞ n→∞

Keywords: Sets of Measure Zero, Partition of Interval, Lower Sum, Upper Sum, Lower
Integral, Upper Integral, Riemann Integral.

REFERENCES

(1) Sections 7.1 and 7.2, Methods of Real Analysis by Richard R. Goldberg, 2nd
edition, Oxford & IBH Publishing Co. Pvt Ltd.

(2) Sections 7.1 and 7.2, Introduction to Real Analysis by Bartle Robert G and
Sherbert Donald R, 4nd edition, Wiley India.

S25011: REAL ANALYSIS Page 92

 (3) Chapter 6, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,
Mc Graw HillInc.,USA.

(4) Chapter 14, Real Analysis, by N.L. Carothers, 1st edition, Cambridge University
Press.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 93

Unit 03-02: Necessary and Sufficient Conditions for Riemann Integrability

Learning Objectives:
After successful completion of this unit, you will be able to

i. Prove necessary and sufficient conditions for Riemann Integrability of functions

over closed and bounded intervals.

ii. Distinguish between Riemann and non-Riemann integrable functions

Introduction: Geometrically, a real valued function f is Riemann integrable on [a, b]

if we can determine area under the graph of f (x) and X axis enclosed between two
vertical lines y = a and y = b. Rigorously, f is Riemann integrable on [a, b] if and only
if f is continuous almost everywhere on [a, b]. Thus real valued function with finitely or
countable many discontinuities on [a, b] is Riemann integrable on [a, b] .

Theorem 3.2.1 Let f be a bounded function on [a, b]. Then f ∈ R[a, b] if and only if
for ϵ > 0 there is a partition σ of [a, b] such that
U [f ; σ] < L[f ; σ] + ϵ
Proof:
I Part: Let for every ϵ > 0 there be a partition σ of [a, b] such that
U [f ; σ] < L[f ; σ] + ϵ . . . (I)
We want to prove that f ∈ R[a, b]. We have
Z b Z b
f ≤ U [f ; σ] and L[f ; σ] ≤ f
a a
So from (I)
Z b Z b
f< f +ϵ
a a
As this holds for every ϵ > 0Z
b Z b
f≤ f . . . (II)
a a
Z b Z b
f≤ f . . . (III)
a a
From (II) and (III)
Z b Z b
f= f
a a
Hence f ∈ R[a, b].
II Part: Let f ∈ R[a, b] and take ϵ > 0. By definition of upper integral (infimum)
there is partition σ1 of [a, b] such that Z
b
U [f ; σ1 ] < f +ϵ
a

S25011: REAL ANALYSIS Page 94

 Zb
=⇒ U [f ; σ1 ] < f +ϵ . . . (IV )
a
By definition of lower integral (supremum)
Z b there is partition σ2 of [a, b] such that
L[f ; σ2 ] > f −ϵ
a

Zb
=⇒ f − ϵ < L[f ; σ2 ]
a

Zb
=⇒ f < L[f ; σ2 ] + ϵ . . . (V )
a

Take σ = σ1 ∪ σ2 . Then σ is refinement of σ1 and σ2 . Hence

U [f ; σ] ≤ U [f ; σ1 ] . . . (V I)
From (IV) and (VI)
Zb
U [f ; σ] < f +ϵ . . . (V II)
a
Similarly
L[f ; σ2 ] ≤ L[f ; σ] . . . (V III)
From (V) and (VIII)
Zb
f < L[f ; σ] + ϵ . . . (IX)
a
Finally from (VII) and (IX)
U [f ; σ] < L[f ; σ] + ϵ

Theorem 3.2.2 If f ∈ R[a, b] then f is continuous almost everywhere on [a, b].

Proof: Define the set E as E = {x ∈ [a, b] : f is discontinuous at x}. For each m ∈ I,
1
define the set Em as Em = {x ∈ [a, b] : ω[f ; x] ≥ m
}. Consider
x∈E

⇐⇒ ω[f ; x] > 0
1
⇐⇒ ω[f ; x] ≥ for some m ∈ I
m
⇐⇒ x ∈ Em for some m ∈ I
∞
[
⇐⇒ x ∈ Em
m=1
So we proved
∞
[
E= Em . . . (I)
m=1

S25011: REAL ANALYSIS Page 95

Take ϵ > 0 and fix m ∈ I. Since f ∈ R[a, b], there is a partition σ of [a, b] such that
ϵ
U [f ; σ] − L[f ; σ] < m
. This implies that
k=n
X ϵ
ω[f ; Ik ]|Ik | = U [f ; σ] − L[f ; σ] < . . . (II)
k=1
m
Here Ik′ s denote subintervals of the partition σ. Define sets Em
′
and Em ′′ ′
as Em = E∩σ
′′ ′ ′ ′′ ′
and Em = E − Em . Then Em = Em ∪ Em . As Em is finite set,
′
m(Em ) = 0 . . . (III)
′′
If x ∈ Em then x ∈ Ij for some j. Let Ij1 , Ij2 , . . . , Ijr denote the subintervals of σ that
′′
contains a point of Em . Note that ω[f ; Ij ] ≥ ω[f ; x] ≥ m1 . Consider
r r
X 1 X
|Ijk | ≤ ω[f ; Ijk ]
k=1
m k=1
r
X 1 ϵ
=⇒ |Ijk | < (f rom(II))
k=1
m m
r
X
=⇒ |Ijk | < ϵ
k=1
′′
Thus we proved that Em can be covered by finitely many intervals whose length add up
to less than ϵ. As ϵ was arbitrary,
′′
m(Em )=0 . . . (IV )
′ ′′
From (III) and (IV ), it follows that m(Em ) = m(Em ∪ Em ) = 0. This holds for every
m ∈ I. A countable union of sets of measure 0 is of measure 0. It now follows from
(I) that the set E is of measure 0. This proves that the function f is continuous almost
everywhere on [a, b].

Theorem 3.2.3 If f is continuous almost everywhere on [a, b] then f ∈ R[a, b].

Proof : Let ϵ > 0. By Archimedian property of real numbers, there is m ∈ I such that
2(b−a) b−a
m> ϵ
. This implies that m
< 2ϵ . Define the set Em as Em = {x ∈ [a, b] : ω[f ; x] ≥
1
m
}. As f ∈ R[a, b], m(Em ) = 0, there are open intervals I1 , I2 , . . . In which are subsets
of [a, b],they cover Em and
∞
X ϵ
|In | < . . . (I)
n=1
2ω[f ; [a, b]]
By known result, Em is closed subset of compact set [a, b]. A closed subset of compact set
is compact. Hence Em is compact. This implies that the open cover{In : n ∈ I} of Em
has a finite subcover, say In1 , In2 , . . . , Inr . Then there are finitely many closed intervals
J1 , J2 , . . . , Js such that    
r
[ [ s
[
[a, b] =  Ink   Jk 
k=1 k=1

S25011: REAL ANALYSIS Page 96

 |Jk |
For k = 1, 2, . . . , s, a partition σk can be constructed such that U [f ; σk ] − L[f ; σk ] < m
Ss
(give a proof for this). Define the partition σ as σ = σk . Then
k=1
s
X r

X
U [f ; σ] − L[f ; σ] = U [f ; σk ] − L[f ; σk ] + ω[f ; Ink ]|Ink |
k=1 k=1
r
b−a X
< + ω[f ; [a, b]] |Ink |
m k=1
ϵ ϵ
< + ω[f ; [a, b]] ( from (I))
2 2ω[f ; [a, b]]
ϵ ϵ
= +
2 2
=ϵ
Thus for every ϵ > 0 there is a partition σ of [a, b] such thatU [f ; σ] < L[f ; σ] + ϵ. This
proves that f ∈ R[a, b].
Necessary and sufficient conditions for Riemann Integrability: The following
statements give necessary and sufficient conditions for Riemann integrability.
Rb Rb
i) f ∈ R[a, b] ⇐⇒ a
f= a
f
ii) f ∈ R[a, b] ⇐⇒ for every ϵ > 0 there is a partition σ of [a, b] such that U [f ; σ] −
L[f ; σ] < ϵ
iii) f ∈ R[a, b] ⇐⇒ f is continuous a.e. on [a b]
iv) f ∈ R[a, b] ⇐⇒ m(D(f )) = 0
Here D(f ) = {x ∈ [a, b] : f is discontinuous at x}
v) f ∈ R[a, b] ⇐⇒ set of discontinuities of f is of measure 0.

Solved Problems 01

Problem 3.2.4 Identify functions from the following functions which are in R[0, 1]?
1 2 3
a) The characteristic
 function of the set {0, 10 , 10 , 10 , . . . , 1}
 sin 1 if 0 < x ≤ 1
x
b) f (x) =
 7 if x = 0

 1 if x = p is rational in [0, 1] and g.c.d. of p, q is 1
q q
c) f (x) =
 0 if x is irrational in [0,1]
d) The characterstic function of a set E ⊂ [0, 1] such that E and both [0, 1] − E are dense
in [0, 1].
Solution:
1 2 3
a) f = the characteristic function of the set {0, 10 , 10 , 10 , . . . , 1}
1 2 3
D(f ) = the set of discontinuities of f in [0, 1] = {0, 10 , 10 , 10 , . . . , 1}

S25011: REAL ANALYSIS Page 97

D(f ) is finite set and hence it is of measure zero. Therefore f is continuous almost
everywhere in [0, 1]. So f ∈ R[0, 1]
b) D(f ) = the set of discontinuities of f in [0, 1] = {0}
D(f ) is finite set and hence it is of measure zero. Therefore f is continuous almost
everywhere in [0, 1]. So f ∈ R[0, 1]
c) D(f ) = the set of discontinuities of f in [0, 1] = Q
D(f ) is countable set and hence it is of measure zero. Therefore f is continuous almost
everywhere in [0, 1]. So f ∈ R[0, 1]
d) f = the characterstic function of a set E ⊂ [0, 1] such that E and both [0, 1] − E are
dense in [0, 1]
D(f ) = the set of discontinuities of f in [0, 1] = [0, 1]
Therefore f is discontinuous everywhere in [0, 1]. So f ̸∈ R[0, 1]

Problem 3.2.5 f ∈ R[a, b] and c1 , c2 , . . . ck ∈ [a, b]. The function g : [a, b] → R is defined
as 
 f (x) if x ̸∈ {c , c , . . . , c }
1 2 k
g(x) =
 1 if x ∈ {c1 , c2 , . . . , ck }
Rb Rb
Prove that g ∈ R[a, b] and g= f.
a a

Solution: Define sets D and E as D = {x ∈ [a, b] : f is discontinuous at x} and E =

{x ∈ [a, b] : g is discontinuous at x}. D is the set of discontinuities of f (x) in [a, b] and E
is the set of discontinuities of g(x) in [a, b]. m(D) = 0 as f is integrable. It is clear that
E ⊂ D ∪ {c1 , c2 , . . . ck }. This implies that m(E) ≤ m(D) + m({c1 , c2 , . . . ck }) = 0 + 0 = 0.
So m(E) = 0 implying that g is integrable . Hence g − f is integrable. Consider
Zb Zc1 Zc1 Zc2 Zck Zck Zb
(g − f ) = (g − f ) + (g − f ) + (g − f ) + · · · + (g − f ) + (g − f ) + (g − f )
a a c1 c1 ck−1 ck ck

Zc1 Zc1 Zc2 Zck Zck Zb

= 0+ 1+ 0 + ··· + 0+ 1+ 0
a c1 c1 ck−1 ck ck

= 0 + 0 + 0 + ··· + 0 + 0 + 0

=0
Zb
(g − f ) = 0
a

Zb Zb
=⇒ g− f =0
a a

S25011: REAL ANALYSIS Page 98

 Zb Zb
=⇒ g= f
a a

Self-Test 01

Problem 3.2.6 A function f : [a, b] → R is Riemann integrable if and only if it is bounded

and

A. it is discontinuous only at finite number of points

B. it is either monotone or it has only finite number of discontinuities

C. the set of its discontinuities is of Lebesgue measure zero

D. the set of its discontinuities is countable

Problem 3.2.7 Let f : [0, 1] → R be defined

 as
 1 if x ∈ Q
f (x) =
 0 if x ∈
̸ Q
Then

A. f is Riemann integrable on [0, 1]

B. f is continuous on [0, 1]

C. f is not Riemann integrable on [0, 1]

D. f is continuous almost everywhere on [0, 1].

Problem 3.2.8 A function f : [a, b] → R is Riemann integrable if it is bounded and

A. it is discontinuous only at finite or countable number of points

B. it is either monotone

C. the set of its discontinuities is of Lebesgue measure zero

D. all of these

Short Answer Questions 01

Problem 3.2.9 State True or False with justification. Let f ∈ R[a, b] and f (x) = g(x)
except for a finitely many points in [a, b] then g ∈ R[a, b].
Solution: TRUE. Let
S = {x ∈ [a, b] : f (x) ̸= g(x)}

D(f ) = {x ∈ [a, b] : f (x) is discontinuous}

D(g) = {x ∈ [a, b] : g(x) is discontinuous}

S25011: REAL ANALYSIS Page 99

Since S is finite, its measure is zero. As f ∈ R[a, b], the measure of D(f ) is 0. Therefore
S ∪ D(f ) is of measure zero.
Now, D(g) ⊂ S ∪ D(f ). So D(g) is of measure zero and hence g is continuous almost
everywhere in [a, b]. This implies that g ∈ R[a, b].

Problem 3.2.10 Let f ∈ R[a, b] and c 

∈ [a, b]. The function g : [a, b] → R is defined as
 f (x) if x = ̸ c
g(x) =
 1 if x = c
Rb Rb
Prove that g ∈ R[a, b] and g= f.
a a
Solution: Let D = {x ∈ [a, b] : f is discontinuous at x} and E = {x ∈ [a, b] :
g is discontinuous at x}. D is the set of discontinuities of f (x) in [a, b] and E is the set of
discontinuities of g(x) in [a, b]. m(D) = 0 as f is integrable. It is clear that E ⊂ D ∪ {c}.
This implies that m(E) ≤ m(D) + m({c}) = 0 + 0 = 0. So m(E) = 0 implying that g is
integrable. Hence g − f is integrable. Consider
Zb Zc Zc Zb
(g − f ) = (g − f ) + (g − f ) + (g − f )
a a c c

Zc Zc Zb
= 0+ 1+ 0
a c c

=0+0+0

=0
So
Zb
(g − f ) = 0
a

Zb Zb
=⇒ g− f =0
a a

Zb Zb
=⇒ g= f
a a

Summary:

• Recall the notation

f is Riemann integrable ⇐⇒ f ∈ R[a, b].

Rb Rb
• f ∈ R[a, b] ⇐⇒ a
f (x)dx = a
f (x)dx

S25011: REAL ANALYSIS Page 100

 • f ∈ R[a, b] ⇐⇒ ∀ϵ > 0∃( partition σ of [a, b] such that U [f, σ] < L[f, σ] + ϵ)
In words, f is Riemann integrable if and only if lower and upper sum can be
arbitrary close to each other.

• f ∈ R[a, b] ⇐⇒ f is continuous almost everywhere on [a, b].

• f ∈ R[a, b] ⇐⇒ m(D(f )) = 0
Here D(f ) denotes the set of discontinuities of f in [a, b].

End of Unit Exercises

Problem 3.2.11 State True or False with justification. If f ∈ R[a, b] and if f (x) = g(x)
except for a countably infinite number of points in [a, b], then g ∈ R[a, b].

Problem 3.2.12 Give an example to show that the composition of Riemann Integrable
functions need not be Riemann Integrable.

Keywords: Riemann Integrability, Necessary Conditions for Riemann Integrability, Suf-

ficient Conditions for Riemann Integrability.

REFERENCES

(1) Section 7.3, Methods of Real Analysis by Richard R. Goldberg, 2nd edition,
Oxford & IBH Publishing Co. Pvt Ltd.

(2) Sections 7.1 and 7.2, Introduction to Real Analysis by Bartle Robert G and
Sherbert Donald R, 4nd edition, Wiley India.

(3) Chapter 6, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,

Mc Graw HillInc.,USA.

(4) Chapter 14, Real Analysis, by N.L. Carothers, 1st edition, Cambridge University
Press.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 101

 Unit 03-03: Properties of Riemann Integrals

Learning Objectives:
After successful completion of this unit, you will be able to

i. State properties of Riemann integrable functions.

ii. Understand algebra of Riemann integrable functions.

Introduction: In this unit, we shall study properties of Riemann integrable functions on

on [a, b]. The geometrical interpretation of Riemann integral as area under curve is useful
in understanding properties of Riemann integrable functions. For example, the Riemann
integral of non-negative Riemann integrable function is non-negative as its graph lies on
or above X axis and so the area enclosed is non-negative.

Theorem 3.3.1 If f ∈ R[a, b]) and a < c < b then f ∈ R[a, c], f ∈ R[c, b] and
Zb Zc Zb
f= f+ f
a a c
Proof:
I Part:
E = {x ∈ [a, b] : f is discontinuous at x}
E1 = {x ∈ [a, c] : f is discontinuous at x}
E2 = {x ∈ [c, b] : f is discontinuous at x}
Now
f ∈ R[a, b]

=⇒ m(E) = 0

=⇒ m(E1 ) = m(E2 ) = 0 (∵ E1 , E2 ⊆ E)

=⇒ f ∈ R[a, c] and f ∈ R[c, b]

II Part: Since f ∈ R[a, b],
Z b Z b Zb
f= f= f . . . (I)
a a
a
Let σ be a partition of [a, c] and τ be a partition of [c, b].Then σ ∪ τ is a partition of [a, b]
Z b
L[f ; σ] + L[f ; τ ] = L[f ; σ ∪ τ ] ≤ f ( from (I))
a

This holds for any partition σ on [a, c]. Keeping τ fix and taking supremum over all
partitions of [a, c]
Zb
sup L[f ; σ] + L[f ; τ ] ≤ f
σ
a

S25011: REAL ANALYSIS Page 102

 Z c Zb
=⇒ f + L[f ; τ ] ≤ f
a
a

Zc Zb Z c Zc
=⇒ f + L[f ; τ ] ≤ f (∵ f= f as f ∈ R[a, c])
a
a a a

This holds for any partition τ of [c, b]. Taking supremum over all partitions on [c, b]
Zc Zb
f + sup L[f ; τ ] ≤ f
σ
a a

Zc Z b Zb
=⇒ f+ f≤ f
c
a a
 
Zc Zb Zb Z b Zb
=⇒ f+ f≤ f ∵ f= f as f ∈ R[a, b] . . . (II)
 
c
a c a c
Again Z b
U [f ; σ] + U [f ; τ ] = U [f ; σ ∪ τ ] ≥ f
a
Z b
U [f ; σ] + U [f ; τ ] ≥ f (from(I))
a
This holds for any partition of [a, c]. Keeping τ fix and taking infimum over all partition
on [a, c]
Zb
inf U [f ; σ] + U [f ; τ ] ≥ f
σ
a
 
Z c Zb Z c Zc
=⇒ f + U [f ; τ ] ≥ f . . . (III) ∵ f= f as f ∈ R[a, c]
 
a a
a a
This holds for any partition τ of [c, b]. Taking infimum over all partitions τ of [c, b].
Zc Zb
f + inf U [f ; τ ] ≥ f
σ
a a

Zc Z b Zb
=⇒ f+ f≥ f
c
a a
 
Zc Zb Zb Z b Zb
=⇒ f+ f≥ f ∵ f= f as f ∈ R[a, b]
 
c
a c a c

S25011: REAL ANALYSIS Page 103

From (II) and (III)

Zb Zc Zb
f= f+ f
a a c

Theorem 3.3.2 If f ∈ R[a, b] and λ is any real number then λf ∈ R[a, b] and
Zb Zb
λf = λ f
a a

Proof: If λ = 0 then the result is obvious. Let λ ̸= 0.

I Part:

E = {x ∈ [a, b] : f is discontinuous at x}
E ′ = {x ∈ [a, b] : λf is discontinuous at x}
If λ ̸= 0 then

f is continuous at x ⇐⇒ λf is discontinuous at x
This implies that
E = E′

=⇒ m(E) = m(E ′ ) = 0 (∵ m(E) = 0 as f ∈ R[a, b])

=⇒ λf ∈ R[a, b]
II Part:
I Case: λ > 0
Let σ = {x0 , x1 , x2 , . . . xn } be a partition of [a, b] and Ik = [xk−1 , xk ] for k = 0, 1, 2, . . . , n.
Since λ > 0

M [λf ; Ik ] = λM [f : Ik ] for k = 1, 2, . . . , n (∵ λ < 0)

Hence n n
X X
M [λf : Ik ]|Ik | = λ M [f : Ik ]|Ik |
k=1 k=1

=⇒ U [f ; σ] = λU [f ; σ]
This is true for all partitions of [a, b]. Taking infimum over all partition of [a, b]
inf (U [λf : σ]) = inf (λU [f ; σ])
σ σ

=⇒ inf (U [λf : σ]) = λ inf (U [f ; σ])

σ σ
Z b Z b
=⇒ λf = λ f
a a

S25011: REAL ANALYSIS Page 104

 Zb Zb
=⇒ λf = λ f (∵ f, λf ∈ R[a, b])
a a
II Case: λ < 0
Let σ = {x0 , x1 . . . . xn } be a partition of [a, b] and Ik = [xk−1 , xk ] for k = 1, 2, ..., n.
Consider
M [λf ; Ik ] = λm[f : Ik ] for k = 1, 2, . . . , n (∵ λ < 0)
n
X n
X
=⇒ M [λf : Ik ]|Ik | = λ m[f : Ik ]|Ik |
k=1 k=1

=⇒ U [λf : σ] = λL[f ; σ]
This is true for all partitions of [a, b]. Taking infimum over all partitions of [a, b]



inf U [λf : σ] = inf λL[f ; σ]
σ σ



=⇒ inf U [λf : σ] = λ sup L[f ; σ] (∵ λ < 0)
σ σ
Z b Z b
=⇒ λf = λ f
a a
Z b Z b
=⇒ λf = λ f (∵ f, λf ∈ R[a, b])
a a

Theorem 3.3.3 If f, g ∈ R[a, b] then f + g ∈ R[a, b] and

Zb Zb Zb
f +g = f + g
a a a
Proof:
I Part: Define sets Ef ,Eg and Ef +g as
Ef = {x ∈ [a, b] : f is discontinuous at x}
Eg = {x ∈ [a, b] : g is discontinuous at x}
Ef +g = {x ∈ [a, b] : f + g is discontinuous at x}
As f, g ∈ R[a, b]
m(Ef ) = m(Eg ) = m(Ef ∪ Eg ) = 0 . . . (I)
By algebra of continuous real valued functions
f and gare continuous at x =⇒ f + g is continuous at x
On taking contraposition of this statement we get the statement

f + g is discontinuous at x =⇒ f is discontinuous at x or g is discontinuous at x

This implies that
x ∈ Ef +g =⇒ x ∈ Ef or x ∈ Eg


=⇒ x ∈ Ef +g =⇒ x ∈ Ef ∪ Eg

=⇒ Ef +g ⊂ Ef ∪ Eg

S25011: REAL ANALYSIS Page 105



=⇒ m(Ef +g ) = 0 from (I)

=⇒ f + g ∈ R[a, b]
II Part: Let σ = {x0 , x1 , . . . , xn } be a sub-division of [a, b] and Ik = [xk−1 ; xk ] for
k = 1, 2, . . . , n. For any k in {1, 2, . . . , n}, consider
M [f + g; Ik ] ≤ M [f ; Ik ] + M [g; Ik ]
n
X n
X n
X
=⇒ M [f + g; Ik ]|Ik | ≤ M [f ; Ik ]|Ik | + M [g; Ik ]|Ik |
k=1 k=1 k=1

=⇒ U [f + g; σ] ≤ U [f ; σ] + U [g; σ]
Z b
=⇒ f + g ≤ U [f ; σ] + U [g; σ]
a
Zb
=⇒ f + g ≤ U [f ; σ] + U [g; σ] . . . (II)
a
As f ∈ R[a, b], there is a partition σ1 of [a, b] such that
ϵ
U [f ; σ1 ] < L[f ; σ1 ] +
2
Similarly as g ∈ R[a, b], there is a partition σ2 of [a, b] such that
ϵ
U [g; σ2 ] < L[g; σ2 ] +
2
Adding these inequalities,
U [f ; σ1 ] + U [g; σ2 ] < L[f ; σ1 ] + L[g; σ2 ] + ϵ
Z b Z b
=⇒ U [f ; σ1 ] + U [g; σ2 ] < f+ g+ϵ
a a
Z b Z b
=⇒ U [f ; σ1 ] + U [g; σ2 ] < f+ g+ϵ
a a
Let σ = σ1 ∪ σ2 . Then σ is refinement of σ1 and Z b From σ2 .
Z b the above inequality,
U [f ; σ] + U [g; σ] < f+ g+ϵ
a a
Taking infimum over all partitions of [a, b], Z b Z b
inf U [f ; σ] + inf U [g; σ] ≤ f+ g+ϵ
σ σ a a
Z b Z b Z b
=⇒ (f + g) ≤ f+ g+ϵ
a a a
Z b Z b Z b
=⇒ (f + g) ≤ f+ g+ϵ
a a a

This holds for every ϵZ> 0. Hence Z

b b Z b
(f + g) ≤ f+ g . . . (III)
a a a

S25011: REAL ANALYSIS Page 106

 Rb Rb Rb Rb
Since f, g, f + g ∈ R[a, b], −f, −g, −(f + g) ∈ R[a, b] and −f = − f , −g = − g
a a a a
Rb Rb
and −(f + g) = − (f + g). The inequality (III) holds for any Riemann integrable
a a
functions f and g. ZReplacing f byZ−f and Zg by −g in (III).
b b b
(−f − g) ≤ −f + −g
a a a
Z b Z b Z b
=⇒ −(f + g) ≤ −f + −g
a a a
Z b Z b Z b
=⇒ − (f + g) ≤ − f +− g
a a a
Z b Z b Z b
=⇒ (f + g) ≥ f+ g . . . (IV )
a a a
From (III) and (IV )
Z b Z b Z b
(f + g) = f+ g
a a a

Theorem 3.3.4 If f is non-negative real valued Riemann integrable function defined on

[a, b] then
Zb
f ≥0
. a

Proof: Let σ = {x0 , x1 , . . . , xn } be a sub-division of [a, b] and Ik = [xk−1 ; xk ] for

k = 1, 2, . . . , n. Let S be the set of all possible lower sums of the function f .
S = {L[f ; τ ] : τ is sub-division of [a, b]}
Then Z b
f = sup S . . . (I)
a
Since f (x) ≥ 0 for all x ∈ [a, b], m[f ; Ik ] ≥ 0 for k = 1, 2, . . . , n. This implies that
X n
L[f ; σ] = m[f ; Ik ]|Ik | ≥ 0
k=1
As σ was any sub-division of [a, b], 0 is lower bound of S. Hence
sup S ≥ 0
Z b
=⇒ f ≥0 (from (I) )
a

Zb
=⇒ f ≥0 (∵ f ∈ R[a, b])
a

S25011: REAL ANALYSIS Page 107

Theorem 3.3.5 If f, g ∈ R[a, b] and f (x) ≤ g(x) for all x ∈ [a, b] then
Zb Zb
f≤ g
a a
Proof: We shall use the following result:
Rb
“If h ∈ R[a, b] and h(x) ≥ 0 for all x ∈ [a, b] then h ≥ 0.” . . . (*)
a
Define h : [a, b] → R as h(x) = g(x) − f (x). As f, g ∈ R[a, b], h ∈ R[a, b]. Also
f (x) ≤ g(x) for all x ∈ [a, b] implies that h(x) ≥ 0 for all x ∈ [a, b]. Hence by result (∗),
Zb
h≥0
a

Zb
=⇒ (g − f ) ≥ 0
a

Zb Zb
=⇒ g− f ≥0
a a

Zb Zb
=⇒ g≥ f
a a

Zb Zb
=⇒ f≤ g
a a

Theorem 3.3.6 If f ∈ R[a, b] then |f | ∈ R[a, b] and

Zb Zb
f ≤ |f |
a a
We shall use the following result:
Rb Rb
“ If g, h ∈ R[a, b] and g(x) ≤ h(x) for all x ∈ [a, b] then g≤ h”. . . . (*)
a a
Proof: I Part: Define sets D and E as
D = {x ∈ [a, b] : f is discontinuous at x}
E = {x ∈ [a, b] : |f | is discontinuous at x}
As f ∈ R[a, b],
m(D) = 0 . . . (I)
Now,
f is continuous at x =⇒ |f | is continuous at x
On taking contraposition
|f | is discontinuous at x =⇒ f is discontinuous at x
Hence
x ∈ E =⇒ x ∈ D

S25011: REAL ANALYSIS Page 108

 =⇒ E ⊆ D

=⇒ m(E) = 0 ( from (I))

=⇒ |f | ∈ R[a, b]
II Part:
f (x) ≤ |f (x)| and − f (x) ≤ |f (x)| for all x ∈ [a, b]
Zb Zb Zb Zb
=⇒ f≤ |f | and −f ≤ |f |
a a a a

Zb Zb Zb Zb
=⇒ f≤ |f | and − f≤ |f |
a a a a

Zb Zb Zb Zb
=⇒ f≤ |f | and f ≥− |f |
a a a a

Zb Zb Zb
=⇒ − |f | ≤ f≤ |f |
a a a

Zb Zb
=⇒ f ≤ |f |
a a

Solved Problems 01

Problem 3.3.7 Let f be continuous on [0, 1] and σn = {0, n1 , n2 , . . . , nn }. If x∗k is any point
in the interval [ k−1
n
, nk ], show that
n
1X
L[f, σn ] ≤ f (x∗k ) ≤ U [f, σn ]
n k=1
Further prove that
n Z1
1X
lim f (x∗k ) = f (x)dx
n→∞ n
k=1 0

Solution: For partition σn ,

m[f, Ik ] ≤ f (x∗k )
n n
1X 1X
=⇒ m[f, Ik ] ≤ f (x∗k )
n k=1 n k=1
n n
X 1 1X
=⇒ m[f, Ik ] ≤ f (x∗k )
k=1
n n k=1
n n
X 1X 1
=⇒ m[f, Ik ]|In | ≤ f (x∗k )(∵ |In | = )
k=1
n k=1 n

S25011: REAL ANALYSIS Page 109

 n
1X
=⇒ L[f, σn ] ≤ f (x∗k ) . . . (I)
n k=1
Similarly,
f (x∗k ) ≤ M [f, Ik ]
n n
1X 1X
=⇒ f (x∗k ) ≤ M [f, Ik ]
n k=1 n k=1
n n
1X X 1
=⇒ f (x∗k ) ≤ M [f, Ik ]
n k=1 k=1
n
n n
1X X 1
=⇒ f (x∗k ) ≤ M [f, Ik ]|In |(∵ |In | = )
n k=1 k=1
n
n
1X
=⇒ f (x∗k ) ≤ U [f, σn ] . . . (II)
n k=1
From (I) and (II), we get
n
1X
L[f, σn ] ≤ f (x∗k ) ≤ U [f, σn ] . . . (III)
n k=1
Take arbitrary ϵ > 0.The uniform continuity and Riemann integrability of f implies the
existence of sufficiently large n such that

U [f, σn ] − L[f, σn ] < ϵ . . . (IV )

Now,
Z1
L[f, σn ] < f (x)dx
0
Z1
=⇒ −L[f, σn ] > − f (x)dx
0
Z1
=⇒ U [f, σn ] − L[f, σn ] > U [f, σn ] − f (x)dx
0
Z1
=⇒ ϵ > U [f, σn ] − f (x)dx (from (IV ))
0
Z1
=⇒ U [f, σn ] < f (x)dx + ϵ
0
Z1
=⇒ lim U [f, σn ] ≤ f (x)dx (∵ ϵ is arbitrary)
n→∞
0

S25011: REAL ANALYSIS Page 110

Also
Z1
lim U [f, σn ] ≥ f (x)dx
n→∞
0
Therefore
Z1
lim U [f, σn ] = f (x)dx . . . (V )
n→∞
0
Similarly,

Z1
lim L[f, σn ] = f (x)dx . . . (V I)
n→∞
0
Now from (III),
n
1X
lim L[f, σn ] ≤ lim f (x∗k ) ≤ lim U [f, σn ]
n→∞ n→∞ n n→∞
k=1
Z1 n Z1
1X
=⇒ f (x)dx ≤ lim f (x∗k ) ≤ f (x)dx ( from (V ) and (V I))
n→∞ n
0 k=1 0

n Z1
1X
=⇒ lim f (x∗k ) = f (x)dx
n→∞ n
k=1 0

Problem 3.3.8 If f is non-positive real valued Riemann integrable function defined on

[a, b] then
Zb
f ≤0
a
Solution: Let σ = {x0 , x1 , . . . , xn } be a partition of [a, b] and Ik = [xk−1 , xk ] for
k = 1, 2, . . . , n. Let S be the set of all possible upper sums of the function f .
S = {U [f ; τ ] : τ is partition of [a, b]}
Then Z b
f = inf S . . . (I)
a
Since f (x) ≤ 0 for all x ∈ [a, b], M [f ; Ik ] ≤ 0 for k = 1, 2, . . . , n. This implies that
X n
U [f ; σ] = M [f ; Ik ]|Ik | ≤ 0
k=1
As σ was any partition of [a, b], 0 is upper bound of S. Hence
sup S ≤ 0

=⇒ inf S ≤ 0 (∵ inf S ≤ sup S)

Z b
=⇒ f ≤0 (from (I) )
a
Zb
=⇒ f ≤0 (∵ f ∈ R[a, b])
a

S25011: REAL ANALYSIS Page 111

Self-Test 01
 
1 1 1
Problem 3.3.9 The value of lim n+1
+ n+2
+ ··· + n+n
equals to
n→∞

A. log 2

B. e
R2 1
C. x
dx
1

D. ∞
Rb
Problem 3.3.10 If f is Riemann integrable on [a, b] and f (x)dx > 0 then
a

A. f (x) > 0 for all x ∈ [a, b]

B. f is monotonically increasing on [a, b]

C. f (x) ≥ 0 for all x ∈ [a, b]

D. none of these

Short Answer Questions 01

Problem 3.3.11 Prove that π

2 Z2
2π 2x 4π 2
≤ dx ≤
9 sin x 9
π
6
Solution:
π π
≤x≤
6 2
   
π π
=⇒ sin ≤ sin x ≤ sin
6 2
1
=⇒ ≤ sin x ≤ 1
2
1
=⇒ 1≤ ≤2
sin x
2x
=⇒ 2x ≤ ≤ 4x
sin x
π π π
Z2 Z2 Z2
2x
=⇒ 2xdx ≤ dx ≤ 4xdx
sin x
π π π
6 6 6
π
"
2 2
# Z2 " #
π π 2x π2 π2
=⇒ − ≤ dx ≤ 2 −
4 36 sin x 4 36
π
6

S25011: REAL ANALYSIS Page 112

 π
2 Z2
2π 2x 4π 2
=⇒ ≤ dx ≤
9 sin x 9
π
6

Problem 3.3.12 If f ∈ R[a, b], show that |f | ∈ R[a, b].

Solution: f ∈ R[a, b]
=⇒ f is continuous almost everywhere in [a, b]
=⇒ |f | is continuous almost everywhere in [a, b] (∵ f is continuous at c implies that |f |
is continuous at c)
=⇒ |f | ∈ R[a, b]

Summary: Let f be Riemann integrable over [a, b] and α, β be real numbers.

Rb Rb Rb
• (αf (x) + βg(x))dx = α f (x)dx + β g(x)dx
a a a
The Riemann integration in this sense is linear.
Rb Rc Rb
• f (x)dx = f (x)dx + f (x)dx for any c ∈ [a, b].
a a c

Rb Rb
• |f | ∈ R[a, b] and f (x)dx ≤ |f (x)|dx.
a a

• If f is non-negative real valued Riemann integrable function defined on [a, b] then

Rb
f ≥ 0.
a

• If f is non-positive real valued Riemann integrable function defined on [a, b] then

Rb
f ≤ 0.
a

Rb Rb
• If f, g ∈ R[a, b] and f (x) ≤ g(x) for all x ∈ [a, b] then f≤ g.
a a

End of Unit Exercises

Problem 3.3.13 If f is continuous on [0, 1], prove that

n   Z1
1X k
lim f = f (x)dx
n→∞ n n
k=1 0

Problem 3.3.14 Evaluate the following limits.

a) lim n1 ( n1 )2 + ( n2 )2 + · · · + ( nn )2
 
n→∞
b) lim n1 sin πn + sin 2π nπ


n
+ · · · + sin n
n→∞
1 3 6 3n

c) lim n e n + e n + · · · + e n
n→∞

S25011: REAL ANALYSIS Page 113

Problem 3.3.15
a) If 0 ≤ x ≤ 1, show that
x2 x2
√ ≤√ ≤ x2
2 1+x
b) Prove that
Z1
1 x2 1
√ ≤ √ dx ≤
3 2 1+x 3
0

Problem 3.3.16 Let f be continuous on the closed bounded interval [a, b] and f (x) ≥ 0
for all x ∈ [a, b]. If f (c) > 0 for some c ∈ [a, b], prove that
Zb
f (x)dx > 0
a
Keywords: Algebra of Riemann Integrable Functions, Properties of Riemann Integrable
Functions.

REFERENCES

(1) Section 7.4, Methods of Real Analysis by Richard R. Goldberg, 2nd edition,
Oxford & IBH Publishing Co. Pvt Ltd.

(2) Sections 7.1 and 7.2, Introduction to Real Analysis by Bartle Robert G and
Sherbert Donald R, 4nd edition, Wiley India.

(3) Chapter 6, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,

Mc Graw HillInc.,USA.

(4) Chapter 14, Real Analysis, by N.L. Carothers, 1st edition, Cambridge University
Press.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 114

Unit 03-04: Mean Value Theorems and Fundamental Theorems of Calculus

Learning Objectives:
After successful completion of this unit, you will be able to

i. Understand mean value theorems.

ii. Define derivatives of functions.

iii. Prove fundamental theorems of calculus using mean value theorems.

Introduction: The Mean Value Theorems are central point of calculus. The Mean
Value Theorems are used to prove Fundamental Theorems of Calculus which give relation
between integration and derivative of function.

We recall the definition of derivative from the Calculus.

Definition 3.4.1 Let f be a real valued function on an interval I of R and c ∈ I. If

lim f (x)−f
x−c
(c)
exists then f is said to be differentiable at c. In this case we define the
x→c
derivative f ′ (c) of f at c as
f (x) − f (c)
f ′ (c) = lim
x→c x−c
If f ′ (x) exists for all x ∈ I, we say the f is differentiable in I.

Theorem 3.4.2 Let f : [a, b] → R be continuous function. If f attains its maximum

value at the point c and if f is differentiable at c then f ′ (c) = 0.
Proof: Suppose on contrary that f ′ (c) ̸= 0. Let f ′ (c) > 0. Consider
f ′ (c) > 0
f (x) − f (c)
=⇒ lim >0
x→c x−c
!
′
f (x) − f (c) f (c)
=⇒ there is δ > 0 such that |x − c| < δ =⇒ f ′ (c) − <
x−c 2
f ′ (c)
 
f (x) − f (c)
=⇒ c − δ < x < c + δ =⇒ 0 < <
2 x−c
 
f c + 2δ − f (c)
=⇒ 0 <
c + 2δ − c
 
δ
=⇒ f (c) < f c +
2
This contradicts the fact that f attains its maximum value at the point c. Similarly, we
get a contradiction when f ′ (c) < 0. Hence we must have f ′ (c) = 0.

The proof of the following theorem is similar to the proof of the theorem that we just
proved.

S25011: REAL ANALYSIS Page 115

Theorem 3.4.3 Let f : [a, b] → R be continuous function. If f attains its minimum
value at the point c and if f is differentiable at c then f ′ (c) = 0.

Theorem 3.4.4 (Rolle’s theorem) Let f : [a, b] → R be continuous on [a, b] and

differentiable in (a, b). If f (a) = f (b) then there is c ∈ (a, b) such that f ′ (c) = 0.
Proof: f is continuous real valued function on compact set [a, b]. Hence f attains its
bounds. So there are points c and d in [a, b] where f (c) = m , f (d) = M and m and M
are minimum and maximum values of f respectively. If m = M then f is constant and
we are done in this case as f ′ (x) = 0 for all x ∈ (a, b) when f is constant. Let m < M .
Then the numbers m and M both can not be equal to the equal values f (a) and f (b).
Without loss of generality, let f (c) = m ̸= f (a) for some c ∈ (a, b). The last theorem
then implies that f ′ (c) = 0. We thus obtained the point c ∈ (a, b) with f ′ (c) = 0.

Theorem 3.4.5 (Lagrange’s Mean Value Theorem) Let f : [a, b] → R be continu-

ous on [a, b] and differentiable in (a, b). Then there is c ∈ (a, b) such that f ′ (c) = f (b)−f
b−a
(a)
.
Proof: Define g : [a, b] → R as
f (b) − f (a)
g(x) = f (x) − x
b−a
The continuity of f on [a, b] implies the continuity of g on [a, b]. Also the differentiability
of f in (a, b) implies the differentiability of g in (a, b). Further it is easy to observe that
g(a) = g(b). We can now apply Rolle’s theorem to get a point c ∈ (a, b) such that
f (b)−f (a) f (b)−f (a)
g ′ (c) = 0. This implies that f ′ (c) − b−a
= 0 i.e. f ′ (c) = b−a
.

Theorem 3.4.6 Let f be continuous on interval I and f ′ (x) > 0 for all x ∈ I except
possibly at the end-points of I. Then f is increasing on I.
Proof: Let a, b ∈ I and a < b. Then f is continuous on [a, b] and differentiable in (a, b).
f (b)−f (a)
By Lagrange’s Mean Value Theorem, there is c ∈ (a, b) such that f ′ (c) = b−a
. As
f ′ (c) > 0, f (b) − f (a) > 0 i.e. f (a) < f (b). Thus a < b implies that f (a) < f (b). Hence
f is increasing on I.

The proof of the following theorem is similar to the proof of the theorem that we just
proved.

Theorem 3.4.7 Let f be continuous on interval I and f ′ (x) < 0 for all x ∈ I except
possibly at the end-points of I. Then f is decreasing on I.

Theorem 3.4.8 (Cauchy’s Mean Value Theorem) Let f, g : [a, b] → R be contin-

uous on [a, b] and be differentiable in (a, b). If g ′ (x) ̸= 0 for all x ∈ (a, b) then there is
′
c ∈ (a, b) such that fg′ (c)
(c)
= fg(b)−g(a)
(b)−f (a)
.
Proof: First of all note that g(a) ̸= g(b) by Rolle’s theorem. Define h : [a, b] → R as
f (b) − f (a)
h(x) = f (x) − g(x)
g(b) − g(a)

S25011: REAL ANALYSIS Page 116

The continuity of f and g on [a, b] implies the continuity of h on [a, b]. Also the differen-
tiability of f and g in (a, b) implies the differentiability of h in (a, b). Further it is easy
to observe that h(a) = h(b). We can now apply Rolle’s theorem to get a point c ∈ (a, b)
f (b)−f (a) ′ f ′ (c) f (b)−f (a)
such that h′ (c) = 0. This implies that f ′ (c) − g(b)−g(a)
g (c) = 0 i.e. g ′ (c)
= g(b)−g(a)
.

Theorem 3.4.9 (First Fundamental Theorem of Calculus) Let f be continuous

Rx
on the closed interval [a, b] If F (x) = f (t)dt for all x ∈ [a, b] then F is differentiable on
a
[a, b] and F ′ (x) = f (x) for all x ∈ [a, b].
Proof: Let x ∈ [a, b] and h > 0 be such that x + h ∈ [a, b]. f is continuous on compact
set [x, x + h]. Hence f attains its maximum and minimum values at points of the interval
[x, x + h]. Let m be the minimum value of f and M be the maximum value of f on
[x, x + h]. For all x ∈ [x, x + h]
m ≤ f (x) ≤ M
x+h
Z x+h
Z x+h
Z
=⇒ mdt ≤ f (t)dt ≤ M dt
x x x
x+h
Z
=⇒ m (x + h − x) ≤ f (t)dt ≤ M (x + h − x)
x
x+h
Z
=⇒ mh ≤ f (t)dt ≤ M h
x

=⇒ m ≤ θ(h) ≤ M . . . (I)
x+h
1
R
where θ(h) = h
f (t)dt. By intermediate value theorem, there is c(h) ∈ [x, x + h] such
x
that
f (c(h)) = θ(h) . . . (II)
Consider
x ≤ c(h) ≤ x + h

=⇒ lim x ≤ lim c(h) ≤ lim (x + h)

h→0 h→0 h→0

=⇒ x ≤ lim c(h) ≤ x
h→0

=⇒ lim c(h) = x . . . (III)

h→0
Now, from (II)
lim f (c(h)) = lim θ(h)
h→0 h→0
 
  x+h
Z
1
=⇒ f lim c(h) = lim  f (t)dt (∵ f is continuous)

h→0 h→0 h
x

S25011: REAL ANALYSIS Page 117

  
F (x + h) − F (x)
=⇒ f (x) = lim
h→0 h
=⇒ f (x) = F ′ (x)

Theorem 3.4.10 (Second Fundamental Theorem of Calculus)

Let f be a continuous function on the closed bounded interval [a, b] and ϕ′ (x) = f (x) for
all x ∈ [a, b]. Then
Zb
f (x)dx = ϕ(b) − ϕ(a)
a
Rx
Proof: Let F (x) = f (t)dt for all x ∈ [a, b]. Then by the first fundamental theorem of
a
calculus,
F ′ (x) = f (x) for all x ∈ [a, b] . . . (I)
Also
ϕ′ (x) = f (x) for all x ∈ [a, b] . . . (II)
From (I) and (II),
F ′ (x) − ϕ′ (x) = 0 for all x ∈ [a, b]

=⇒ (F − ϕ)′ (x) = 0 for all x ∈ [a, b]

=⇒ F − ϕ is constant on [a, b]

=⇒ F (b) − F (a) = ϕ(b) − ϕ(a)

Zb Za
=⇒ f (t)dt − f (t)dt = ϕ(b) − ϕ(a)
a a

Zb
=⇒ f (t)dt = ϕ(b) − ϕ(a)
a

Theorem 3.4.11 Let ϕ be a real valued function defined on the closed intervale [a, b]
such that ϕ′ is continuous on [a, b]. Let A = ϕ(a) and B = ϕ(b). If f is continuous on
ϕ([a, b]) then
ZB Zb
f (x)dx = f (ϕ(u))ϕ′ (u)du
A a

Rx
Proof: Define F : ϕ [a, b] → R as F (x) = f (t)dt. Then by first fundamental
a
theorem of calculus,
F ′ (x) = f (x) for all x ∈ [a, b]
Define G : [a, b] → R as G(u) = F (ϕ(u)). Differentiating with respect to u,
G′ (u) = F ′ (ϕ(u))ϕ′ (u) = f (ϕ(u))ϕ′ (u)

S25011: REAL ANALYSIS Page 118

This holds for u ∈ [a, b]. So
Zb
f (ϕ(u))ϕ′ (u)du
a

Zb
= G′ (u)du
a

= G(b) − G(a)



= F ϕ(b) − F ϕ(a)

= F (B) − F (A)
ZB
= F ′ (x)dx
A

ZB
= f (x)dx
A
This completes the proof.
Solved Problems 01
Rx
Problem 3.4.12 If f is continuous on [a, b] and if F (x) = f (t)dt for all x ∈ [a, b], prove
a
that F is continuous on [a, b].
Solution: Let c ∈ [a, b]. Then Z x
F (x) − F (c) = f (t)dt
c
Since f is integrable, we know Z x say |f (x)|
Z xit is bounded, Z x≤ M over [a, b]. Then
− M dt ≤ f (t)dt ≤ M dt
c c c
which means Z x
−M (x − c) ≤ f (t)dt ≤ M (x − c)
c
Thus we get
|F (x) − F (c)| ≤ M |x − c|
This implies that
− M |x − c| ≤ F (x) − F (c) ≤ M |x − c|


=⇒ lim −M |x − c| ≤ lim F (x) − F (c) ≤ lim M |x − c|
x→c x→c x→c


=⇒ 0 ≤ lim F (x) − F (c) ≤ 0
x→c


=⇒ lim F (x) − F (c) = 0
x→c

=⇒ lim F (x) = F (c)

x→c
This proves that f is continuous on c. Since c was arbitrary point of [a, b], F is continuous
on [a, b].

S25011: REAL ANALYSIS Page 119

 Rx
Problem 3.4.13 If f is continuous on [a, b], if f (x) > 0 and F (x) = f (t)dt for all
a
x ∈ [a, b], prove that F is strictly increasing on [a, b].
Solution: For x < y, consider y
Z Zx
F (y) − F (x) = f (t)dt − f (t)dt
a a
Zy Za
= f (t)dt + f (t)dt
a x
Za Zy
= f (t)dt + f (t)dt
x a
Zy
= f (t)dt
x

>0 (∵ f (x) > 0 on [a, b] and x < y)

Thus
x < y =⇒ F (x) < F (y)
This proves that F is strictly increasing on [a, b].
Self-Test 01

Problem 3.4.14 A function f : [−1, 1] → R defined by f (x) = 1 − |x| is

A. continuous on [−1, 1]

B. f ′ (x) does not exist for all x ∈ (−1, 1)

C. f ′ (0) does not exist

D. f ′ (x) does not exist for x = ±1

Problem 3.4.15 A function f : R → R defined by f (x) = 1 − |x + 2| + |x − 2| is

A. discontinuous on [−2, 2]

B. f ′ (x) does not exist for all x ∈ (−2, 2)

C. f ′ (0) does not exist

D. f ′ (x) does not exist for x = ±2

Problem 3.4.16 If g(x) = sin x and f (x) = cos x, then the point at which the conclusion
of Cauchy’s mean value theorem holds in − π4 , π4
 

π
A. 6

S25011: REAL ANALYSIS Page 120

 B. 0

π
C. 4

−π
D. 6

Short Answer Questions 01

Rx
Problem 3.4.17 If f is continuous on (−∞, ∞) and if F (x) = f (t)dt for all x in
0
(−∞, ∞), prove that F ′ (x) = f (x) for all x ∈ (−∞, ∞).
Solution: The result follows from first fundamental theorem of calculus applied on
interval [0, x]. For detailed answer, repeat the proof of first fundamental theorem of
calculus on the interval [0, x].

Problem 3.4.18 Let f be continuous on the closed bounded interval [a, b] and f (x) ≥ 0
Rb
for all x ∈ [a, b].If f (x)dx = 0 , prove that f is identically zero on [a, b].
a
Solution: Suppose on contrary that f is not identically zero on [a, b]. Then there is
c ∈ [a, b] such that f (c) > 0.Since f (x) is continuous at x = c and f (c) > 0, there is δ > 0
such that
1
a < c − δ < x < c + δ < b =⇒ 0 < f (c) < f (x)
2
This implies that
Zc+δ Zc+δ
1 1
f (x)dx > f (c) dx = 2δf (c) = δf (c) . . . (I)
2 2
c−δ c−δ
Since f (x) ≥ 0 on [a, b],
Zc−δ Zb
f (x)dx ≥ 0 and f (x)dx ≥ 0 . . . (II)
a c+δ

Now,
Zb Zc−δ Zc+δ Zb
f (x)dx = f (x)dx + f (x)dx + f (x)dx
a a c−δ c+δ

> 0 + δf (c) + 0( from (I) and (II))

= δf (c)

>0
Rb
So f (x)dx > 0. This is a contradiction. Hence f must be identically 0 on [a, b].
a

Summary:

• Mean Value Theorems are used to prove Fundamental Theorems of Calculus.

S25011: REAL ANALYSIS Page 121

 • Fundamental Theorems of Calculus give relation between Integration and Deriv-
ative.

• Mean Value Theorems are Central Point of Calculus.

End of Unit Exercises

Problem 3.4.19 Give Geometrical Interpretation of Rolle’s Mean Value Theorem and
Lagrange’s Mean Value Theorem.

Problem 3.4.20 Test whether the conditions of Rolle’s Mean Value Theorem hold for
given functions on the interval and if so find the value of c.
x2 +ab


a) f (x) = log (a+b)x on [a, b] b) f (x) = (x − a)m (x − b)n on [a, b]

Problem 3.4.21 Test whether the conditions of Lagrange’s Mean Value Theorem hold
for given functions on the interval and if so find the value of c.

a) f (x) = (x − 1)(x − 2)(x − 3) on [0, 4] b) f (x) = tan−1 x on [0, 1]

Problem 3.4.22 a) Show that the value of c when Cauchy’s Mean Value Theorem is
applied to the functions f (x) = sin x and g(x) = cos x on [a, b] is the arithmetic mean of
a and b.
b) Show that the value of c q
when Cauchy’s Mean Value Theorem is applied to the func-
√
tions f (x) = x and g(x) = x1 on [a, b] is the geometric mean of a and b.

Keywords: Rolle’s Mean Value Theorem, Lagrange’s Mean Value Theorem, Cauchy’s
Mean Value Theorem, First Fundamental Theorem of Calculus, Second Fundamental
Theorem of Calculus.

REFERENCES

(1) Sections 7.5 to 7.8, Methods of Real Analysis by Richard R. Goldberg, 2nd edi-
tion, Oxford & IBH Publishing Co. Pvt Ltd.

(2) Section 7.3, Introduction to Real Analysis by Bartle Robert G and Sherbert
Donald R, 4nd edition, Wiley India.

(3) Chapter 6, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,

Mc Graw HillInc.,USA.

(4) Chapter 14, Real Analysis, by N.L. Carothers, 1st edition, Cambridge University
Press.

S25011: REAL ANALYSIS Page 122

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 123

CREDIT 04

125
 Unit 04-01: Convergence and Uniform Convergence of Sequence of
Functions

Learning Objectives:
After successful completion of this unit, you will be able to

i. Define pointwise and uniform convergence of sequence of functions.

ii. Understand difference between point wise and uniform convergence of sequence
of functions.

Introduction: In this unit, we shall define sequence of functions and study its pointwise
and uniform convergence. The uniform convergence of sequence of functions is stronger
than the pointwise convergence of sequence of functions. So every uniformly convergent
sequence of functions converges pointwise.

Definition 4.1.1 Let {fn }∞

n=1 be a sequence of real valued function on set E.

We say that {fn }∞

n=1 converges to the function f on E if lim fn (x) = f (x) for all x ∈ E
n→∞

OR

Let {fn }∞ ∞
n=1 be a sequence of real value function on set E. We say that {fn }n=1 converges

point wise to the function f on E if for every point x of E and for every ϵ > 0 there is
N ∈ I such that
n ≥ N =⇒ |fn (x) − f (x)| < ϵ.
While evaluating lim fn (x), x is kept fixed (constant).
n→∞


fn → f on E ⇐⇒ ∀x ∈ E ∀ϵ > 0 ∃N ∈ I n ≥ N =⇒ |fn (x) − f (x)| < ϵ
 
fn → f on E ⇐⇒ ∀x ∈ E lim fn (x) = f (x)
n→∞

Definition 4.1.2 : Let {fn }∞

n=1 be a sequence of real valued function on set E. We say

that {fn }∞
n=1 converges uniformly to the function f on E if for every ϵ > 0 there is N ∈ I

such that
n ≥ N ; x ∈ E =⇒ |fn (x) − f (x)| < ϵ


fn → f uniformly on E ⇐⇒ ∀ϵ > 0∃N ∈ I n ≥ N : x ∈ E =⇒ |fn (x) − f (x)| < ϵ

It is clear that the uniform convergence implies point wise convergence.

fn → f uniformly on E =⇒ fn → f pointwise on E
If the type of convergence is not specified then the convergence is always point wise
convergence. So fn → f means fn convergence to f point wise. It may not be uniform
convergence.
S25011: REAL ANALYSIS Page 126
i) In point wise convergence , N depends on ϵ and the point x of E .
ii) In uniform convergence N depends only on ϵ . The same N works for all points of the
domain.

iii) Note that for fixed x ∈ E{fn (x)} is sequence of real numbers . So {fn (x)}∞
n=1 denotes

sequence of real numbers for x ∈ E and {fn }∞

n=1 is sequence of functions.

Result: The sequence {fn }∞

n=1 does not converges uniformly to f on E if and only if

there exist some ϵ > 0 for which there is no N ∈ I satisfying the statement
n ≥ N ; x ∈ E =⇒ |fn (x) − f (x)| < ϵ

∞
Theorem 4.1.3 The sequence {f
! n }n=1 of functions converges uniformly to f on E if and
only if lim sup |fn (x) − f (x)| =0
n→∞ x∈E

Proof:
I Part :
lim (sup |fn (x) − f (x)|) = 0
n→∞ x∈E
!
=⇒ ∀ϵ > 0 ∃N ∈ I n ≥ N =⇒ sup |fn (x) − f (x)| < ϵ
x∈E


=⇒ ∀ϵ > 0 ∃N ∈ I n ≥ N ; x ∈ E =⇒ |fn (x) − f (x)| < ϵ (∵ |fn (x) − f (x)| ≤ sup |fn (x) − f (x)|)
x∈E
Hence {fn }∞
n=1 converges uniformly to f on E.

II Part: Let {fn }∞

n=1 be converge uniformly to f on E .Then for every ϵ > 0, there is

N ∈ I such that
n ≥ N ; x ∈ E =⇒ |fn (x) − f (x)| < ϵ
It means that ϵ is upper bound of the set {|fn (x) − f (x)| : x ∈ E} whenever n ≥ N . So
sup |fn (x) − f (x)| < ϵ whenever n ≥ N .
x∈E
Stated differently
n ≥ N =⇒ sup |fn (x) − f (x)| < ϵ
x∈E
!
=⇒ lim sup |fn (x) − f (x)| =0
n→∞ x∈E
Remark: The result in the above theorem !
fn → f uniformly on E ⇐⇒ lim sup |fn (x) − f (x)| = 0
n→∞ x∈E
!
fn ̸→ f uniformly on E ⇐⇒ lim sup |fn (x) − f (x)| ̸= 0
n→∞ x∈E

Theorem 4.1.4 Cauchy criterion for uniform convergence

Let {fn }∞ ∞
n=1 be a sequence of real valued function on a set E. Then {fn }n=1 converges

S25011: REAL ANALYSIS Page 127

uniformly on E( to some functionf ) if and only if for every ϵ > 0 there is N ∈ I such
that
m, n ≥ N ; x ∈ E =⇒ |fm (x) − fn (x)| < ϵ
Proof: I Part:
Let {fn }∞
n=1 be converge uniformly to f on E. Then for every ϵ > 0 there is N ∈ I such

that
ϵ
n ≥ N ; x ∈ E =⇒ |fn (x) − f (x)| <
2
 
ϵ ϵ
=⇒ m, n ≥ N ; x ∈ E =⇒ |fm (x) − f (x)| < and |fn (x) − f (x)| <
2 2


=⇒ m, n ≥ N ; x ∈ E =⇒ |fm (x) − f (x)| + |fn (x) − f (x)| < ϵ


=⇒ m, n ≥ N ; x ∈ E =⇒ |fm (x) − f (x) − fn (x) + f (x)| < ϵ


=⇒ m, n ≥ N ; x ∈ E =⇒ |fm (x) − fn (x)| < ϵ
II Part: Let for every ϵ > 0 there is N ∈ I such that
m, n ≥ N ; x ∈ E =⇒ |fm (x) − fn (x)| < ϵ . . . (I)
It is clear that {fn (x)}∞
n=1 is a Cauchy sequence of real numbers for all x ∈ E. Since R

is complete, {fn (x)}∞

n=1 is convergent for all x ∈ E. So lim fn (x) exists for all x ∈ E.
n→∞
Define the function f : E → R as f (x) = lim fn (x)
n→∞
Taking limit as m → ∞ in (I)
n ∈ N ; x ∈ E =⇒ lim |fm (x) − fn (x)| < ϵ
m→∞

=⇒ (n ∈ N ; x ∈ E =⇒ | lim fm (x) − fn (x)| < ϵ)

m→∞

=⇒ (n ∈ N ; x ∈ E =⇒ |f (x) − fn (x)| < ϵ)

=⇒ (n ∈ N ; x ∈ E =⇒ |fn (x) − f (x)| < ϵ)

This proves that {fn }∞
n=1 converges uniformly to f on E.

Theorem 4.1.5 Dini’s Theorem

Let {fn }∞
n=1 be a sequence of continuous real valued function on the compact metric space

M such that f1 (x) ≤ f2 (x) ≤ · · · fn (x) ≤ · · · for all x ∈ E. If {fn }∞

n=1 converges point

wise on to the continuous function f on M then {fn }∞ n=1 converges uniformly to f on M .

Proof: For eachN ∈ I define gn on M as gn = f − fn . Consider
f1 (x) ≤ f2 (x) ≤ · · · ≤ fn (x) ≤ . . . ≤ f (x)for all x ∈ M

=⇒ −f1 (x) ≥ −f2 (x) ≥ . . . − fn (x) ≥ for all x ∈ E

=⇒ f (x) − f1 (x) ≥ f (x) − f2 (x) ≥ · · · f (x) − fn (x) ≥ for all x ∈ E

=⇒ g1 (x) ≥ g2 (x) ≥ . . . gn (x) ≥ for all x ∈ E . . . (I)

Also
lim gn (x) = f (x) − lim fn (x) = f (x) − f (x) = 0 . . . (II)
n→∞ n→∞

S25011: REAL ANALYSIS Page 128

Take ϵ > 0. Then from (II) there is N (x) ∈ I such that
ϵ
n ≥ N (x) =⇒ |gn (x) − 0| <
2
ϵ
n ≥ N (x) =⇒ gn (x) <
2
In particular, for n = N (x)
ϵ
gN (x) (x) < . . . (III)
2

The function gN (x) (x) is continuous as functions f and fN (x) are continuous. The con-
tinuity of gN (x) at a point x implies the existence of neighbourhood (open ball) Bx of x
such that
y ∈ Bx =⇒ gN (x) (y) < ϵ . . . (IV )
A metric space X is compact if and only if every open cover of X has some finite sub-cover.
The collection {Bx |x ∈ m} forms an open covering of M . Since M is compact there is a
finite subcover of this open covering. Let {Bx1 , Bx2 , ......., Bxk } be a finite subcover of M .
Take N = max{Nx1 , Nx2 , . . . , Nxk } and y ∈ M . Then y ∈ Bxj for some j ∈ {1, 2, . . . , k}.
Consider
n ≥ N, y ∈ M

=⇒ gn (y) ≤ gN (y) ( from (I))

=⇒ gn (y) ≤ gN (y) ≤ gNi (y) for i = 1, 2, . . . , k ( from (I) as N ≤ Nxi for i = 1, 2 . . . , k)

=⇒ gn (y) < ϵ ( from (III) and (IV ) as y ∈ Bxj )

=⇒ |gn (y) − 0| < ϵ

Thus
n ≥ N ; y ∈ M =⇒ |gn (y) − 0| < ϵ

=⇒ {gn }∞
n=1 converges uniformly to 0 on M

=⇒ {fn }∞
n=1 converges uniformly to f on M

Remark: The Dini’s theorem holds even when {fn }∞

n=1 is non-increasing.

Solved Problems 01

Problem 4.1.6 Prove that the sequence {fn }∞

n=1 of functions converges to f on [0, 1]

where fn (x) = xn and 

 0 if 0 ≤ x < 1
f (x) =
 1if x = 1
Is the convergence uniform? Justify your answer.

Solution: I Part : fn : [0, 1] → R, fn (x) = xn

I Case : 0 ≤ x < 1

S25011: REAL ANALYSIS Page 129

In this case
lim fn (x) = lim xn
n→∞ n→∞

=0

= f (x)
II Case : x = 1
In this case,
lim fn (x) = lim xn
n→∞ n→∞

= lim 1n
n→∞

= lim 1
n→∞

= f (x)
Thus lim fn (x) = f (x) for all x ∈ [0, 1]. Hence{fn }∞
n=1 converges point wise to f on
n→∞
[0, 1].
1
II Part : Let ϵ = 2
> 0. Suppose on contrary there is N ∈ I such that

1
n ≥ N ; x ∈ [0, 1] =⇒ |fn (x) − f (x)| < . . . (I)
2
 
n 1
=⇒ n ≥ N ; x ∈ [0, 1] =⇒ |x − f (x)| <
2
 
1
=⇒ n ≥ N ; x ∈ [0, 1) =⇒ |xn − 0| <
2
 
1
=⇒ n ≥ N ; x ∈ [0, 1) =⇒ xn <
2
1
=⇒ lim xn ≤ lim
x→1− x→1 2
1
=⇒ 1n ≤
2
1
=⇒ 1 ≤
2
This a contradiction . Hence there is no n such that the statement (I) holds. So {fn }∞
n=1

does not convergence uniformly to f on [0,1].

Problem 4.1.7 Prove that the sequence {gn }∞

n=1 of function converges to g on [0, ∞)

where gn (x) = x
1+nx
and g(x) = 0 for all x ∈ [0, ∞).Does {gn }∞
n=1 converge uniformly to

g on [0, ∞)? Justify your answer.

Solution:
x
I Part : gn : [0, ∞) → R, gn (x) = 1+nx
I Case : x > 0

S25011: REAL ANALYSIS Page 130

In this case
x
lim gn (x) = lim
n→∞ n→∞ 1 + nx
x
n
= lim
n→∞ 1 +x
n
lim x
n→∞ n
=
lim 1 + x
n→∞ n
0
=
0+x
=0

= g(x)

II Case: x = 0
In this case clearly lim gn (x) = lim gn (0) = lim 0 = 0 = g(x)
n→∞ n→∞ n→∞
Thus lim gn (x) = g(x) for all x ∈ [0, ∞). Hence{gn }∞
n=1 converges point wise to g on
n→∞
[0, ∞).
II Part :Let ϵ > 0
By Archimedean property of real numbers,there is N ∈ I such that
1
<ϵ . . . (I)
N
Consider
n≥N
1 1
=⇒ ≤
n N
1
=⇒ <ϵ (from (I) )
n
Thus
1
n ≥ N =⇒ <ϵ . . . (II)
n
Consider
0<1

=⇒ nx < 1 + nx
nx
=⇒ <1
1 + nx
x 1
=⇒ <
1 + nx n
This holds for all x ∈ [0, ∞). Thus
x 1
x ∈ [0, ∞) =⇒ < . . . (III)
1 + nx n
Now,
x
n ≥ N, x ∈ [0, ∞) =⇒ <ϵ (from (II) and (III))
1 + nx

S25011: REAL ANALYSIS Page 131



=⇒ n ≥ N, x ∈ [0, ∞) =⇒ gn (x) < ϵ


=⇒ n ≥ N ; x ∈ [0, ∞) =⇒ |gn (x) − g(x)| < ϵ
This proves that {gn }∞
n=1 converges uniformly to g on [0, ∞)

Problem 4.1.8 Show that the sequence {hn }∞

n=1 of functions converges point wise to
nx
the function h on R or (−∞ , ∞) where hn (x) = 1+n2 x2
and h(x) = 0 for all x ∈ R.
Does{hn }∞
n=1 converge uniformly to h on R ? Justify your answer.
Solution: I Case : x ̸= 0
In this case

nx
lim hn (x) = lim
n→∞ n→∞ 1 + n2 x2
x
n
= lim
n→∞ 12 + x2
n
lim x
n→∞ n
=
lim 12 + x2
n→∞ n
0
=
0 + x2
=0

= h(x)
II Case : x = 0
In this case clearly

lim hn (x) = lim 0 = 0 = h(x)

n→∞ n→∞
∞
This holds for all x ∈ R. Hence {hn }n=1 converges point wise to h on R.
hn n1 − h n1 = 1
− 0 = 21 ≥ 14 . This holds for



2
all n ∈ I. Hence
1
sup |hn (x) − h(x)| ≥ for all n ∈ I
x∈E 4
!
=⇒ lim sup |hn (x) − h(x)| ̸= 0
n→∞ x∈E

=⇒ {hn }∞
n=1 does not converges uniformly to h on R

Self-Test 01

Problem 4.1.9 The sequence {fn (x)}∞

n=1 = {x
n+2 ∞
}n=1

A. pointwise convergent to a continuous function on [0, 1].

B. not pointwise convergent on [0, 1].

S25011: REAL ANALYSIS Page 132

 C. pointwise convergent but not uniformly convergent on [0, 1].

D. uniformly convergent on [0, 1].

n
Problem 4.1.10 The sequence {fn (x)}∞ x ∞
n=1 = { 1+xn }n=1

1
A. pointwise convergent on [0, 10 ] but not pointwise convergent on [0, 1].

1
B. uniformly convergent on [0, 10 ] and pointwise convergent on [0, 1].

C. uniformly convergent on [0, 1].

D. none of these.

Short Answer Questions 01

Problem 4.1.11 Show that the sequence {hn }∞

n=1 of functions converges point wise to
nx
the function h on R or (−∞ , ∞) where hn (x) = 1+n2 x2
and h(x) = 0 for all x ∈ R.
Does{hn }∞
n=1 converge uniformly to h on R ? Justify your answer.
Solution: I Case : x ̸= 0
In this case

nx
lim hn (x) = lim
n→∞ n→∞ 1 + n2 x2
n
x
= lim
n→∞ 12 + x2
n
lim x
n→∞ n
=
lim 12 + x2
n→∞ n
0
=
0 + x2
=0

= h(x)
II Case x = 0 :
In this case clearly

lim hn (x) = lim 0 = 0 = h(x)

n→∞ n→∞

This holds for all x ∈ R. Hence {hn }∞

n=1 converges point wise to h on R.
1
Let ϵ = 2
. Suppose on contrary that there is N ∈ I such that
1
n ≥ N ; x ∈ R =⇒ |hn (x) − h(x)| < 2

S25011: REAL ANALYSIS Page 133

This implies that
nx 1
n ≥ N ; x ∈ R =⇒ 2 2
−0 <
1+n x 2
!
nx 1
=⇒ n ≥ N ; x ∈ R =⇒ 2 2
−0 <
1+n x 2
1
In particular for n = N and x = n ∈ R
N N1 1
2 1 <
1 + n N2 2
1 1
=⇒ <
1+1 2
1 1
=⇒ <
2 2
∞
This is a contradiction. Hence {hn }n=1 does not converge uniformly to h on R.

Problem 4.1.12 Show that the sequence {χn }∞

n=1 of functions converges point wise to

\ {0} → R = (−∞, ∞) is given by

the constant function K where χn : R
 1 if x ∈ [−n, n]
χn (x) =
 0 if x ∈
/ [−n, n]

and K(x) = 1 for all x ∈ R Does {χn }∞

n=1 converge uniformly to K on R ? Justify your

answer.
Solution:
χn is a characteristic function of [−n, n]. Let x ∈ R. By Archimedean property of real
numbers, there are positive integers N1 and N2 such that x < N1 and −x < N2 . Take
N = max{N1 , N2 } so that x < N1 ≤ N and − x < N2 ≤ N . This implies that
−N ≤ x ≤ N =⇒ −(N + 2) ≤ −(N + 1) ≤ −N ≤ x ≤ N ≤ N + 1 ≤ N + 2 . . .

=⇒ χN (x) = χN +1 (x) = χN +2 (x) = .... = 1

=⇒ lim χn (x) = K(x)

n→∞
This holds for all x ∈ R. Hence {χn }∞
n=1 converges to K on R.

Suppose on contrary that there is N ∈ I such that

1
n ≥ N, x ∈ R =⇒ |χn (x) − K(x)| <
2
This implies that,
1
n ≥ N ; x ∈ R =⇒ |χn (x) − 1| <
2
In particular for n = N and x = N + 1
1
|χn (N + 1) − 1| <
2
1
=⇒ |0 − 1| <
2
1
=⇒ 1<
2
This is a contradiction . Hence {χn }∞
n=1 does not converges uniformly to 1 on R.

S25011: REAL ANALYSIS Page 134

Summary:

• Let τ denote the collection of all real valued functions defined on E. τ = {f :

E → R}
A sequence of real valued functions on a set E is a function from the set I of
positive real numbers to the set τ .

I: 1 2 3 ··· n ···
↓ ↓ ↓ ↓ ··· ↓ ···
τ : f1 f2 f3 ··· fn ···

The above sequence is denoted as {fn }∞

n=1 . Here fn denote image of n. Note

that f1 : E → R, f2 : E → R, f3 : E → R, . . . , fn : E → R, . . . are real valued

functions defined on E.

E: ... ... ... x ... ... ...

f1 : ... ... ... f1 (x) ... ... ...
f2 : ... ... ... f2 (x) ... ... ...
f3 : ... ... ... f3 (x) ... ... ...
... ... ... ... ... ... ... ...
fn : . . . ... ... fn (x) ... ... ...
... ... ... ... ... ... ... ...

• Let {fn }∞
n=1 be a sequence of real valued functions defined on E. If lim fn (x) =
n→∞
f (x) for all x ∈ E then {fn }∞
n=1 is said to converge pointwise to f on E.

In pointwise convergence of {fn }∞

n=1 , point x of E is kept fixed and then the

convergence of the sequence {fn (x)}∞

n=1 of real numbers is considered.

So x is kept fixed or treated as constant while evaluating lim fn (x) in pointwise

n→∞
pointwise
convergence. We denote the pointwise convergence of {fn }∞
n=1 to f by fn −−−−−→
f.

pointwise
• fn −−−−−→ f ⇐⇒ ∀x( lim fn (x) = f (x))
n→∞

pointwise
• fn −−−−−→ f ⇐⇒ ∀x(sequence of real numbers {fn (x)}∞
n=1 converges to f (x))

pointwise
• fn −−−−−→ f ⇐⇒ ∀x∀ϵ > 0∃N ∈ I(n ≥ N =⇒ |fn (x) − f (x)| < ϵ)

S25011: REAL ANALYSIS Page 135

 E: ... ... ... x ... ... ...
f1 : ... ... ... f1 (x) ... ... ...
... ... ... ... ↓ ... ... ...
f2 : ... ... ... f2 (x) ... ... ...
... ... ... ... ↓ ... ... ...
f3 : ... ... ... f3 (x) ... ... ...
... ... ... ... ↓ ... ... ...
fn : . . . ... ... fn (x) ... ... ...
... ... ... ... ↓ ... ... ...
f ... ... ... f (x) ... ... ...
• A sequence of functions converges to a function (if it is convergent) just as a
sequence of real numbers converges to a real number (if it is convergent).

• Let {fn }∞
n=1 be a sequence of real valued functions defined on E. If for every

ϵ > 0 there is N ∈ I such that

x ∈ E, n ≥ N =⇒ |fn (x) − f (x)| < ϵ
unif orm
We denote the uniform convergence of {fn }∞
n=1 to f by fn −
−−−−→ f .
The positive integer N in the above definition does not depend on the point x.
It only depends on ϵ. The same N works for all x in E.
In pointwise convergence, N depends on ϵ as well as the point x.
It is clear that the uniform convergence is stronger than the pointwise conver-
gence.
pointwise unif orm
fn −−−−−→ f =⇒ fn −−−−−→ f
unif orm
• fn −−−−−→ f ⇐⇒ ∀ϵ > 0∃N ∈ I(x ∈ E, n ≥ N =⇒ |fn (x) − f (x)| < ϵ)
unif orm
• fn −−−−−→ f ⇐⇒ ∀ϵ > 0∃N ∈ I(x ∈ E, m, n ≥ N =⇒ |fm (x) − fn (x)| < ϵ)
unif orm
• fn −−−−−→ f ⇐⇒ lim sup fn (x) − f (x) = 0
n→∞ x∈E

End of Unit Exercises

sin nx
Problem 4.1.13 If fn (x) = n
on [0, 1], show that there exists N ∈ I such that
1
x ∈ [0, 1], n ≥ N =⇒ |fn (x) − 0| <
100
xn
Problem 4.1.14 Let fn (x) = 1+xn
on [0, 1]. Show that {fn }∞
n=1 converges pointwise on

[0, 1]. Further if {fn }∞

n=1 converges pointwise to f on [0, 1], show that there is NO N ∈ I

such that
1
x ∈ [0, 1], n ≥ N =⇒ |fn (x) − f (x)| <
8

S25011: REAL ANALYSIS Page 136

Problem 4.1.15 Let χn be a characteristic function of the open interval (0, n1 ) and let
fn (x) = nχn (x) on [0, 1].
a) Show that {fn }∞
n=1 converges to 0 on [0, 1].

b) Show that there is no N ∈ I such that

1
x ∈ [0, 1], n ≥ N =⇒ |fn (x) − 0| <
4
R1
c) Calculate lim fn .
n→∞ 0
R1 R1
d) Hence or otherwise, prove that lim fn (x)dx ̸= lim fn (x)dx.
0 n→∞ n→∞ 0

Problem 4.1.16 Let fn (x) = nx(1 − x2 )n on [0, 1]

∞
a) Show that {f
( n }n=1 converges to 0 on [0, 1].
)∞
1
converges to 12 on [0, 1].
R
b) Show that fn
0
n=1
R1 R1
c) Hence or otherwise, prove that lim fn (x)dx ̸= lim fn (x)dx.
0 n→∞ n→∞ 0

Keywords: Sequence of Functions, Pointwise Convergence of Sequence of Functions,

Uniform Convergence of Sequence of Functions.

REFERENCES

(1) Sections 9.1 and 9.2, Methods of Real Analysis by Richard R. Goldberg, 2nd
edition, Oxford & IBH Publishing Co. Pvt Ltd.

(2) Chapter 8, Introduction to Real Analysis by Bartle Robert G and Sherbert

Donald R, 4nd edition, Wiley India.

(3) Chapter 7, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,

Mc Graw HillInc.,USA.

(4) Chapter 10, Real Analysis, by N.L. Carothers, 1st edition, Cambridge University
Press.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 137

 Unit 04-02: Properties of functions preserved under uniform convergence

Learning Objectives:
After successful completion of this unit, you will be able to

i. State properties of sequence of functions preserved under uniform convergence.

ii. Understand continuity, integrability and differentiability of functions under uni-

form convergence of sequence of functions.

Introduction: Under the uniform convergence of sequence of functions, continuity and

Riemann Integrability are preserved. The differentiability of functions is preserved with
some additional conditions.

Theorem 4.2.1 Let {fn }∞

n=1 be a sequence of real valued functions on metric space M

which converges uniformly to f on M . If each {fn }∞

n=1 is continuous at point a of M then

f is also continuous at a.
Proof: Let ϵ > 0. Since {fn }∞
n=1 converges uniformly to f on M , there is N ∈ I such

that
ϵ
n ≥ N, x ∈ M =⇒ |fn (x) − f (x)| < . . . (I)
3
Each {fn }∞
n=1 is continuous at point a. In particular, fN is continuous at x = a. Hence
there is δ > 0 such that

ϵ
d(x, a) < δ =⇒ |fN (x) − fN (a)| < . . . (II)
3
For d(x, a) < δ, consider
|f (x) − f (a)|

= |(f (x) − fN (x)) + (fN (x) − fN (a)) + (fN (a) − f (a))|

≤ |f (x) − fN (x)| + |fN (x) − fN (a)| + |fN (a) − f (a)|

ϵ ϵ ϵ
< + + (from (I) and (II))
3 3 3
=ϵ
Thus for every ϵ > 0 there is δ > 0 such that
d(x, a) < δ =⇒ |f (x) − f (a)| < ϵ
This proves that f is continuous at x = a.
Continuity of functions at a point is preserved under uniform convergence of sequence of functions.

Theorem 4.2.2 Let {fn }∞

n=1 be a sequence of real valued functions on metric space M

which converges uniformly to f on M . If each {fn }∞

n=1 is continuous on M then f is also

continuous on M .

S25011: REAL ANALYSIS Page 138

Proof: Let a ∈ M . Take ϵ > 0. Since {fn }∞
n=1 converges uniformly to f on M , there is

N ∈ I such that
ϵ
n ≥ N, x ∈ M =⇒ |fn (x) − f (x)| < . . . (I)
3
Each {fn }∞
n=1 is continuous at point a. In particular, fN is continuous at x = a. Hence

there is δ > 0 such that

ϵ
d(x, a) < δ =⇒ |fN (x) − fN (a)| < . . . (II)
3
For d(x, a) < δ, consider
|f (x) − f (a)|

= |(f (x) − fN (x)) + (fN (x) − fN (a)) + (fN (a) − f (a))|

≤ |f (x) − fN (x)| + |fN (x) − fN (a)| + |fN (a) − f (a)|

ϵ ϵ ϵ
< + + (from (I) and (II))
3 3 3
=ϵ
Thus for every ϵ > 0 there is δ > 0 such that
d(x, a) < δ =⇒ |f (x) − f (a)| < ϵ
This proves that f is continuous at x = a. As a was arbitrary point of M , f is continuous
on M .
Continuity of functions on set is preserved under uniform convergence of sequence of functions.

Theorem 4.2.3 If {fn }∞

n=1 is a sequence of functions in R[a, b] which converges uniformly

to f on [a, b] then f ∈ R[a, b].

Proof: {fn }∞
n=1 converges uniformly to f on [a, b]. So for ϵ = 1, there is N ∈ I such that
n ≥ N ; x ∈ [a, b] =⇒ |fn (x) − f (x)| < 1
In particular for n = N ,
x ∈ [a, b] =⇒ |fN (x) − f (x)| < 1 . . . (I)
Consider
fN ∈ R[a, b]

=⇒ fN is bounded on [a, b]

∃M ′ > 0 such that x ∈ [a, b] =⇒ |fN (x)| < M ′

=⇒ . . . (II)
From (I) and (II)
x ∈ [a, b] =⇒ |f (x)| ≤ |f (x) − fN (x)| + |fN (x)| < M ′ + 1 = M
Thus
x ∈ [a, b] =⇒ |f (x)| < M

=⇒ f is bounded on [a, b] . . . (III)

S25011: REAL ANALYSIS Page 139

Define the set D as D = {x ∈ [a, b] : f is discontinuous}. For each n ∈ I, take the set En
∞
S
as En = {x ∈ [a, b] : fn is discontinuous at x}. Let E = En . Consider
n=1
fn ∈ R[a, b] for all n ∈ I

=⇒ fn is continuous a.e on [a, b] for all n ∈ I

=⇒ m(En ) = 0 for all n ∈ I

∞
!
[
=⇒ m En = 0
n=1

=⇒ m(E) = 0 . . . (IV )
Now,
x ∈ [a, b] − E

=⇒ x ̸∈ E
∞
[
=⇒ x ̸∈ En
n=1

=⇒ x ̸∈ En for any n ∈ I

=⇒ fn is continuous at x for all n ∈ I

=⇒ f is continuous at x( as continuity is preserved uniform convergence. )

=⇒ x ̸∈ D

=⇒ x ∈ [a, b] − D
Thus
x ∈ [a, b] − E =⇒ x ∈ [a, b] − D

=⇒ D ⊆ E

=⇒ m(D) = 0 (from(IV ))

=⇒ f is continuous a.e on [a, b] . . . (V )

From (III) and (V), f is bounded and continuous a.e on [a, b]. Hence, f ∈ R[a, b].

A sequence of Riemann integrable functions converges to a Riemann integrable func-

tions under uniform convergence i.e. Riemann integrability is preserved under uniform
convergence of sequence of functions.

S25011: REAL ANALYSIS Page 140

Theorem 4.2.4 If {fn }∞
n=1 is a sequence of a functions in R[a, b] which converges uni-

formly to the function f on [a, b] then f is in R[a, b] and

Zb Zb Zb
lim fn (x) dx = f (x) dx = lim fn (x) dx
n→∞ n→∞
a a a

Proof:
I Part: f ∈ R[a, b]. Refer to the proof of previous theorem.
II Part: Let ϵ > 0. Since {fn }∞
n=1 converges uniformly to f on [a, b], there is N ∈ I

such that
ϵ
n ≥ N ; x ∈ [a, b] =⇒ |fn (x) − f (x)| <
b−a
Z b !
Z b
ϵ
=⇒ n ≥ N =⇒ |fn (x) − f (x)|dx < dx
a a b−a
!
Z b
ϵ
=⇒ n ≥ N =⇒ |fn (x) − f (x)|dx < (b − a)
a b−a
!
Z b
=⇒ n ≥ N =⇒ |fn (x) − f (x)|dx < ϵ
a
   
Z b Z b Z b
=⇒ n ≥ N =⇒ fn (x) − f (x)dx < ϵ ∵ g ≤ |g|
a a a
 
Z b Z b
=⇒ n ≥ N =⇒ fn (x)dx − f (x)dx < ϵ
a a

This proves that

Zb Zb Zb
lim fn (x) dx = f (x) dx = lim fn (x) dx
n→∞ n→∞
a a a

Theorem 4.2.5 Let fn be differentiable on [a, b] for all n ∈ I and fn′ be continuous on
[a, b]. If {fn }∞ ′ ∞
n=1 converges tof on [a, b] and {fn }n=1 converges uniformly to g on [a, b]

then g(x) = f ′ (x) for all x ∈ [a, b]. In other words

lim fn′ (x) = f ′ (x) for all x ∈ [a, b]


n→∞

Proof: {fn }∞
n=1 converges uniformly to f on [a, b]. So
lim fn (x) = f (x) for all x ∈ [a, b] . . . (I)
n→∞
Also
{fn′ }∞
n=1 converges uniformly to g on [a, b] and so
lim fn′ (x) = g(x) for all x ∈ [a, b]. . . . (II)
n→∞

S25011: REAL ANALYSIS Page 141

Further each fn′ is continuous on [a, b] and hence fn′ ∈ R[a, b] for all n ∈ I. The Riemann
integrability is preserved under uniform convergence implying that g ∈ R[a, b].
Zx Zx
′
lim fn (t)dt = lim fn′ (t)dt for all x ∈ [a, b]
n→∞ n→∞
a a
Zx


=⇒ lim fn (x) − fn (a) = g(t)dt for all x ∈ [a, b] ( from (II))
n→∞
a
Zx
=⇒ lim fn (x) − f (a) = g(t)dt for all x ∈ [a, b]
n→∞
a
Zx
=⇒ f (x) − f (a) = g(t)dt for all x ∈ [a, b]
a
Zx
d
=⇒ f ′ (x) = g(t)dt for all x ∈ [a, b]
dx
a

=⇒ f ′ (x) = g(x) for all x ∈ [a, b] ( by first fundamental theorem of calculus )

′
=⇒ lim fn (x) = g(x) = lim fn′ (x) for all x ∈ [a, b] ( from (I) and (II))
n→∞ n→∞

Theorem 4.2.6 If A is dense in metric space M and {fn }∞

n=1 is a sequence of continuous

functions on M converging uniformly on A then {fn }∞

n=1 converges uniformly on M .

Proof: Take x ∈ M and ϵ > 0. As fm and fn are continuous at x, there are δ1 and δ2
such that
ϵ
|x − x′ | < δ1 =⇒ |fm (x) − fm (x′ )| <
3
′ ′ ϵ
|x − x | < δ2 =⇒ |fn (x) − fn (x )| <
3
Taking δ = min{δ1 , δ2 }, we get
ϵ
|x − x′ | < δ =⇒ |fm (x) − fm (x′ )| < . . . (I)
3
ϵ
|x − x′ | < δ =⇒ |fn (x) − fn (x′ )| < . . . (II)
3
As A is dense in M , there is a ∈ A such that
|x − a| < δ
So from (I) and (II),
ϵ
|fm (x) − fm (a)| < . . . (III)
3
ϵ
|fn (x) − fn (a)| < . . . (IV )
3
The uniform convergence of {fn }∞ n=1 on A implies the existence of N ∈ I such that
ϵ
y, y ∈ A, m, n ≥ N =⇒ |fm (y) − fn (y ′ )| <
′
3
In particular for y = y ′ = a,
ϵ
m, n ≥ N =⇒ |fm (a) − fn (a)| < . . . (V )
3
S25011: REAL ANALYSIS Page 142
 ϵ
From (III), (IV ) and (V ) m, n ≥ N =⇒ |fm (x) − fm (a)| < 3
, |fm (a) − fn (a)| <
ϵ ϵ
3
, |fn (x) − fn (a)| < 3
ϵ
=⇒ (m, n ≥ N =⇒ |fm (x) − fm (a) + fm (a) − fn (a) − fn (x) + fn (a)| < 3
+ 3ϵ + 3ϵ )
ϵ
=⇒ (m, n ≥ N =⇒ |fm (x) − fn (x)| < 3
+ 3ϵ + 3ϵ )
=⇒ (m, n ≥ N =⇒ |fm (x) − fn (x)| < ϵ)
Recall that x was arbitrary point of M . Also observe that N is independent of x. It
follows that {fn }∞
n=1 converges uniformly on M .

Theorem 4.2.7 Let g be a continuous function on the closed bounded interval [a, b].If
{fn }∞
n=1 is a sequence of continuous functions which converge uniformly on [a, b] to f then
Rb Rb
lim fn g = f g.
n→∞ a a
Proof: A continuous function on closed bounded interval is bounded. As g is continuous
function on closed bounded interval [a, b], g is bounded. So there is M > 0 such that
x ∈ [a, b] =⇒ |g(x)| ≤ M . . . (I)
Let ϵ > 0. Since {fn }∞n=1 converges uniformly to f on [a, b], there is N ∈ I such that
ϵ
x ∈ [a, b], n ≥ N =⇒ |fn (x) − f (x)| <
M
This implies that
ϵ
x ∈ [a, b], n ≥ N =⇒ |fn (x) − f (x)| <
M
 
ϵ
=⇒ x ∈ [a, b], n ≥ N =⇒ |g(x)||fn (x) − f (x)| < |g(x)|
M
 
ϵ
=⇒ x ∈ [a, b], n ≥ N =⇒ |g(x)||fn (x) − f (x)| < M ) ( from (I))
M


=⇒ x ∈ [a, b], n ≥ N =⇒ |g(x)||fn (x) − f (x)| < ϵ


=⇒ x ∈ [a, b], n ≥ N =⇒ |fn (x)g(x) − f (x)g(x)| < ϵ

=⇒ {fn g}∞
n=1 converge uniformly to f g on [a, b]

Zb Zb
=⇒ lim fn g = lim fn g
n→∞ n→∞
a a

Zb Zb
=⇒ lim fn g = fg
n→∞
a a

Solved Problems 01

Problem 4.2.8 Show that {fn }∞

n=1 converges to f on [0, 1] where fn : [0, 1] → R is defined

as 
 2n if 1 ≤ x ≤ 2
n n
fn (x) =
 0 otherwise
Does {fn }∞
n=1 converge uniformly to f on [0, 1]? Justify your answer.

S25011: REAL ANALYSIS Page 143

Solution:
I Part: Fix x ∈ (0, 1]. By Archimedean property of real numbers, there is N ∈ I such
that
1
x>
2N
1
=⇒ 2N >
x
1
=⇒ N >
2x
1 1 1
=⇒ x > > > > ···
2N 2(N + 1) 2(N + 2)
=⇒ fN (x) = fN +1 (x) = fN +2 (x) = · · · = 0

=⇒ lim fn (x) = 0 = f (x)

n→∞
This holds for all x ∈ (0, 1]. Also for x = 0
lim fn (x) = lim 0 = 0 = f (x)
n→∞ n→∞
Thus
lim fn (x) = lim 0 = 0 = f (x) for all x ∈ [0, 1]
n→∞ n→∞
Hence {fn }∞
n=1 converges pointwise to f on [0, 1].

II Part: Consider
Z1
lim fn (x)dx
n→∞
0
Z1
= f (x)dx
0
Z1
= 0dx
0

=0
So
Z1
lim fn (x)dx = 0 . . . (I)
n→∞
0
R1
Next, we find the value of the integral lim fn (x)dx.
n→∞ 0
Z1
lim fn (x)dx
n→∞
0
 
1 2
Zn Zn Z1
 
= lim  fn (x)dx + fn (x)dx + fn (x)dx

n→∞ 
0 1 2
n n

S25011: REAL ANALYSIS Page 144

  
2
Zn
 
= lim 0+ fn (x)dx + 0
n→∞  
1
n
"  #
2 1
= lim 2n −
n→∞ n n
2n
= lim
n→∞ n
=2
So
Z1
lim fn (x)dx = 2 . . . (II)
n→∞
0
From (I) and (II)
Z1 Z1
lim fn (x)dx ̸= lim fn (x)dx
n→∞ n→∞
0 0
Hence {fn }∞
n=1 does not converge uniformly to f on [0, 1].

xn
Problem 4.2.9 Let fn (x) = n
on [0, 1].
a) Show that {fn }∞
n=1 converges uniformly to 0 on [0, 1].
b) Prove that {fn′ }∞
n=1 converges to 0 on [0, 1)?

Solution: a) Let ϵ > 0. By Archimedian property of real numbers, there is N ∈ I such

that N > 1ϵ . Then

1
x ∈ [0, 1], n ≥ N =⇒ n >
ϵ
!
1
=⇒ x ∈ [0, 1], n ≥ N =⇒ <ϵ
n
!
xn xn
 
1
=⇒ x ∈ [0, 1], n ≥ N =⇒ <ϵ ∵ ≤ as x ∈ [0, 1]
n n n
!
xn
=⇒ x ∈ [0, 1], n ≥ N =⇒ −0 <ϵ
n
!
=⇒ x ∈ [0, 1], n ≥ N =⇒ |fn (x) − 0| < ϵ

This proves that {fn }∞

n=1 converges uniformly to 0 on [0, 1].
nxn−1
b) lim fn′ (x) = lim n
= lim xn−1 = 0 if x ∈ [0, 1).
n→∞ n→∞ n→∞
If x = 1 then lim fn′ (x) = 1. Thus {fn }∞
n=1 converges to f on [0, 1] where f : [0, 1] → R
n→∞

S25011: REAL ANALYSIS Page 145

is defined as 
 0 if 0 ≤ x < 1
f (x) =
 1 if x = 1
So {fn′ }∞
n=1 converge to 0 on [0, 1).
Self-Test 01

Problem 4.2.10 Let {fn }∞

n=1 be a sequence of real valued differentiable functions on [a, b]

that converges uniformly to the differentiable function f on [a, b]. Then

A. {fn′ }∞ ′
n=1 converges pointwise but not uniformly to f on [a, b].

B. {fn′ }∞ ′
n=1 converges uniformly to f on [a, b].

C. {fn′ }∞ ′
n=1 does not converge pointwise to f on [a, b].

D. none of these.

Problem 4.2.11 Let {fn }∞

n=1 be a sequence of Riemann integrable functions on [a, b] that

converges uniformly to the function f on [a, b]. Then

Rb Rb
A. { fn (x)dx}∞
n=1 converges to f (x)dx.
a a

Rb Rb
B. { fn (x)dx}∞
n=1 does not converge to f (x)dx.
a a

Rb Rb
C. { fn (x)dx}∞
n=1 converges to f (x)dx when each fn is differentiable.
a a

Rb Rb
D. { fn (x)dx}∞
n=1 converges to f (x)dx only when each fn is continuously differ-
a a
entiable.

Problem 4.2.12 Let {fn (x)}∞ n ∞

n=1 = {(−x) }n=1 on [0, 1]. Then which of the following is

true.

A. there exist a pointwise convergent subsequence of {fn }.

B. {fn } has no pointwise convergent subsequence.

C. {fn } converges pointwise everywhere.

D. {fn } has exactly one pointwise convergent subsequence.

Short Answer Questions 01

Problem 4.2.13 If {fn }∞ ∞ ∞

n=1 and {gn }n=1 converge uniformly on E, prove that {fn +gn }n=1

converge uniformly on E.

S25011: REAL ANALYSIS Page 146

Solution: Let ϵ > 0. Since {fn }∞ ∞
n=1 and {gn }n=1 converges uniformly on E, there are

positive integers N1 and N2 such that

ϵ ϵ
x ∈ E, n ≥ N1 =⇒ |fn (x) − f (x)| < and x ∈ E, n ≥ N2 =⇒ |gn (x) − g(x)| <
2 2
Take N = max{N1 , N2 }. Then
ϵ ϵ
x ∈ E, n ≥ N =⇒ |fn (x) − f (x)| < and |gn (x) − g(x)| <
2 2
This implies that
ϵ ϵ
x ∈ E, n ≥ N =⇒ |fn (x) − f (x)| + |gn (x) − g(x)| < +
2 2


=⇒ x ∈ E, n ≥ N =⇒ |fn (x) − f (x)| + |gn (x) − g(x)| < ϵ


=⇒ x ∈ E, n ≥ N =⇒ |fn (x) − f (x) + gn (x) − g(x)| < ϵ


=⇒ x ∈ E, n ≥ N =⇒ |(fn (x) + gn (x)) − (f (x) + g(x))| < ϵ
This proves that {fn + gn }∞
n=1 converge uniformly on E.

Problem 4.2.14 If gn (x) = n1 e−nx on [0, ∞), prove that {gn }∞

n=1 converges uniformly to

0 on [0, ∞).
Solution:
x≥0

=⇒ −x ≤ 0

=⇒ −nx ≤ 0

=⇒ e−nx ≤ e0 (∵ ex is increasing function)

=⇒ e−nx ≤ 1
Thus
x ∈ [0, ∞), n ∈ I =⇒ e−nx ≤ 1 . . . (I)
Let ϵ > 0. By Archimedian property of real numbers there is N ∈ I such that

1
N> . . . (II)
ϵ
Now
n≥N
1 1
=⇒ ≤
n N
1
=⇒ < ϵ ( from (II))
n
So from (I)
e−nx
x ∈ [0, ∞), n ≥ N =⇒ <ϵ
n
!
e−nx
=⇒ x ∈ [0, ∞), n ≥ N =⇒ −0 <ϵ
n


=⇒ x ∈ [0, ∞), n ≥ N =⇒ |gn (x) − 0| < ϵ

S25011: REAL ANALYSIS Page 147

This proves that {gn }∞
n=1 converges uniformly to 0 on [0, ∞).

Summary:

• A sequence of continuous functions converges to a continuous function under

uniform convergence.
unif orm unif orm
fn −−−−−→ f =⇒ fn −−−−−→ f
|{z} |{z}
continuous continuous

• A sequence of Riemann integrable functions converges to a Riemann integrable

function under uniform convergence.
unif orm unif orm
fn −−−−−→ f =⇒ fn −−−−−→ f
|{z} |{z}
Riemann integrable on [a,b] Riemann integrable on [a,b]
Further in this case
Zb Zb
lim fn (x)dx = lim fn (x)dx
n→∞ n→∞
a a

• A monotone sequence {fn }∞

n=1 of continuous real valued functions converging to

a continuous function f on compact space gets uniformly converge to f .

pointwise unif orm
fn −−−−−→ f on compact space M =⇒ fn −−−−−→ f on M
|{z} |{z}
monotone continuous

pointwise unif orm

• fn −−−−−→ f and fn′ −−−−−→ g on [a, b] =⇒
|{z}
continuous
 ′
′
g(x) = f (x) = lim fn′ (x) = lim fn (x) on [a, b]
n→∞ n→∞

End of Unit Exercises

Problem 4.2.15 Let f be a uniformly continuous real-valued function defined on R and

let fn (x) = f (x + n1 ) on R. Prove that {fn }∞
n=1 converges uniformly to f on R.

xn
Problem 4.2.16 Let fn (x) = 1+xn
on [0, 1].
a) Show that {fn }∞ 1
n=1 converges uniformly on [0, 4 ]

b) Does {fn }∞
n=1 converge uniformly on [0, 1]?

Problem 4.2.17 Let {fn }∞

n=1 be a sequence of continuous real-valued functions that

converge uniformly on the closed bounded interval [a, b] and for each n ∈ I, define Fn as
Rx
Fn (x) = fn (t)dt on [a, b] . Show that {Fn }∞
n=1 converges uniformly on [a, b].
a

Problem 4.2.18 If {fn }∞

n=1 be a sequence of continuous functions that converges uni-

formly on [0, 1], prove that there exists M > 0 such that
x ∈ [0, 1], n ∈ I =⇒ |fn (x)| ≤ M

S25011: REAL ANALYSIS Page 148

Problem 4.2.19 Give an example of pointwise converging sequence of functions {fn }∞
n=1

on R for which there is no constant M > 0 such that

x ∈ R, n ∈ I =⇒ |fn (x)| ≤ M

Problem 4.2.20 Give an example of monotone sequence {fn }∞

n=1 of functions converging

pointwise on R+ that does not converge uniformly on R+ .

Problem 4.2.21 If {fn }∞n=1 is a sequence of functions which converges uniformly to the

continuous function f on R, prove that lim fn x + n1 = f (x) on R.

n→∞

Problem 4.2.22 Let fn (x) = nx(1 − x2 )n on [0, 1]

a) Show that {fn }∞
n=1 converges pointwise to 0 on [0, 1].

b) Show that {fn }∞

n=1 does not converge uniformly on [0, 1].

sin nx
Problem 4.2.23 Let fn (x) = n
on [0, 1]. Show that
a) {fn }∞
n=1 converges uniformly to 0 on [0, 1].

b) {fn′ }∞
n=1 does not converge pointwise on [0, 1].

Keywords: Continuity under Uniform Convergence of Sequence of Functions, Riemann

Integrability under Uniform Convergence of Sequence of Functions, Differentiability under
Uniform Convergence of Sequence of Functions.

REFERENCES

(1) Section 9.3, Methods of Real Analysis by Richard R. Goldberg, 2nd edition,
Oxford & IBH Publishing Co. Pvt Ltd.

(2) Chapter 8, Introduction to Real Analysis by Bartle Robert G and Sherbert

Donald R, 4nd edition, Wiley India.

(3) Chapter 7, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,

Mc Graw HillInc.,USA.

(4) Chapter 10, Real Analysis, by N.L. Carothers, 1st edition, Cambridge University
Press.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 149

Unit 04-03: Convergence and Uniform Convergence of Series of Functions

Learning Objectives:
After successful completion of this unit, you will be able to

i. Define pointwise and uniform convergence of series of functions.

ii. State sufficient conditions for convergence of series of functions.

Introduction: In this unit, we shall define series of functions and study its pointwise
and uniform convergence. The results on convergence of series of functions are proved us-
ing the corresponding results of convergence of sequence of functions. Under the uniform
convergence of series of functions, continuity and Riemann Integrability are preserved.
∞
un is defined as an ordered pair < {un }∞ ∞
P
Definition 4.3.1 A series of functions n=1 , {sn }n=1 >
n=1
where {un }∞ ∞
n=1 and {sn }n=1 are sequences of real valued functions on E and sn =

u1 + u2 + u3 + · · · + un for all n ∈ I. ∞
P
The notation u1 + u2 + u3 + · · · is also used to denote the un series.
n=1

i) If sequence {sn }∞
n=1 of functions converges pointwise to f on E then we say that the
∞
P
series un converges pointwise to f on E. In this case we write
n=1
∞
X
un = f on E
n=1
or as ∞
X
un (x) = f (x) for all x ∈ E
n=1

If lim sn (x) = lim u1 (x) + u2 (x) + · · · + un (x) = f (x) for all x ∈ E then the series
n→∞ n→∞
P∞
un converges pointwise to f on E.
n=1
ii) If sequence {sn }∞
n=1 of functions converges uniformly to f on E then we say that the
P∞
series un converges uniformly to f on E.
n=1
∞
P
Theorem 4.3.2 Let u1 , u2 , u3 . . . be real valued functions on M . If un converges
n=1
uniformly to f on M and if each un is continuous at a point a of M then f is also
continuous at the point a.
Proof: We shall use the following theorem.
“Let {gn }∞ ∞
n=1 be a sequence of real valued functions defined on metric space M . If {gn }n=1

converges uniformly to g of M and if each gn is continuous at point a of M then g is

continuous at the point a.” . . . (*)
For each n ∈ I define sn as
sn = u1 + u2 + u3 + · · · + un

S25011: REAL ANALYSIS Page 150

So {sn }∞
n=1 is a sequence of real valued functions on M . Consider
un is continuous at a for all n ∈ I =⇒ sn is continuous at a for all n ∈ I . . . (I)
X∞
un converges uniformly to f on M =⇒ {sn }∞ n=1 converges uniformly to f on M . . . (II)
n=1
By applying result (∗) to (I) and (II), we conclude that f is continuous at the point a
of M .

∞
P
Theorem 4.3.3 Let u1 , u2 , u3 . . . be real valued functions on M . If un converges
n=1
uniformly to f on M and if each un is continuous on M then f is also continuous on M .

Proof: We shall use the following theorem.

“ Let {gn }∞ ∞
n=1 be a sequence of real valued functions defined on metric space M . If {gn }n=1

converges uniformly to g of M and if each gn is continuous on M then g is continuous on

M .” . . . (*)
For each n ∈ I define sn as
sn = u1 + u2 + u3 + · · · + un
So {sn }∞
n=1 is a sequence of real valued functions on M . Consider
un is continuous on M for all n ∈ I

=⇒ sn is continuous on M for all n ∈ I . . . (I)

Also ∞
X
un converges uniformly to f on M
n=1

=⇒ {sn }∞
n=1 converges uniformly to f on M . . . (II)
By applying result (∗) to (I) and (II), we conclude that f is continuous on M .

∞
P
Theorem 4.3.4 (Weierstrass M-Test) Let uk be a series of real valued functions
k=1
∞
P
on a set E. If there exist positive numbers M1 , M2 , . . . with Mk < ∞ such that
k=1
∞
X ∞
X
uk ≪ Mk for all x ∈ E
k=1 k=1
∞
P
then uk converges uniformly on E.
k=1

Proof: There exist N1 ∈ I such that

n ≥ N1 =⇒ |un (x)| ≤ Mn for all x ∈ E . . . (I)

For each n ∈ I define
sn = u1 + u2 + · · · + un and tn = M1 + M2 + · · · + Mn

S25011: REAL ANALYSIS Page 151

 ∞
So {sn }∞ uk and {tn }∞
P
n=1 is a sequence of partial sums of the series n=1 is a sequence of
k=1
∞
P
partial sums of the series Mk . For m > n ≥ N1 and x ∈ E, consider
k=1
sm (x) − sn (x)
m
X n
X
= uk (x) − uk (x)
k=1 k=1

m
X
= uk (x)
k=n+1
m
X
≤ uk (x) (by triangle inequality)
k=n+1
m
X
≤ Mk (from (I))
k=n+1
m
X n
X
= Mk − Mk
k=1 k=1

= tm − tn

≤ |tm − tn |
This holds for all x ∈ E. So
m > n ≥ N1 ; x ∈ E =⇒ sm (x) − sn (x) ≤ |tm − tn | . . . (II)
Let ϵ > 0. Consider the series
∞
X
Mk < ∞
k=1
∞
X
=⇒ Mk is convergent series of positive real numbers
k=1

=⇒ {tn }∞
n=1 is convergent

=⇒ {tn }∞
n=1 is Cauchy


=⇒ ∃N2 ∈ I m > n ≥ N2 =⇒ |tm − tn | < ϵ . . . (III)
Take N = max{N1 , N2 }. Then from (II) and (III).
m > n ≥ N ; x ∈ E =⇒ sm (x) − sn (x) < ϵ

=⇒ {sn }∞
n=1 converges uniformly on E (by Cauchy criterion for uniform convergence)
∞
X
=⇒ uk converges uniformly on E
k=1

S25011: REAL ANALYSIS Page 152

Solved Problems 01
∞
P
Problem 4.3.5 Show that the series un converges pointwise to the function f on[0, 1]
n=0
where un (x) = x(x − 1)n and

0 if x = 0

f (x) =
1 if 0 < x ≤ 1

∞
P ∞
P
(i.e. prove that un = f (x) for all x ∈ [0, 1].) Does un converges uniformly to f on
n=0 n=0
[0, 1]? Justify your answer.
Solution:
I Part: We shall use the following result
∞
P
“Let u0 , u1 , . . . be real valued function on the metric space M. If un converges uniformly
n=1
to f on M and if each un is continuous at point a ∈ M then f is also continuous at the
point a”. . . . (*)
Let sn = u0 + u1 + · · · + un forn = 0, 1, 2, . . ..{sn }∞
n=1 is a sequence of partial sum of the
∞
P
series un .
n=1
If x = 0 then clearly
lim sn (x) = lim (u0 (x) + u1 (x) + . . . + un (x)) = 0 = f (x)
n→∞ n→∞
For x ∈ (0, 1], consider
lim sn (x) = lim (u0 (x) + u1 (x) + . . . + un (x))
n→∞ n→∞

= lim (x + x(1 − x) + x(1 − x)2 + . . . + x(1 − x)n )

n→∞

= x lim (1 + (1 − x) + (1 − x)2 + . . . + (1 − x)n )

n→∞

(1 − (1 − x)n+1 )
= x lim
n→∞ 1 − (1 − x)
(1 − (1 − x)n+1 )
= x lim
n→∞ x
= lim 1 − (1 − x)n+1


n→∞

= 1 − lim (1 − x)n+1
n→∞

=1−0 (∵ 0 < 1 − x < 1 as 0 < x ≤ 1)

=1

= f (x)
Thus
lim sn (x) = f (x) for all x ∈ [0, 1]
n→∞

S25011: REAL ANALYSIS Page 153

 =⇒ {sn }∞
n=1 converges pointwise to f on [0, 1]
∞
X
=⇒ un converges pointwise to f on [0, 1]
n=1
X∞
=⇒ un = f (x) for all x ∈ [0, 1]
n=1
n
II Part : un (x) = x(1 − x) for n = 0, 1, 2, . . . Each un is continuous at x = 0 as it is
polynomial continuous at function. However f is discontinuous at x = 0 . It follows from
∞
P
the result (*) that the series un does not converge uniformly to f on [0, 1].
n=1

∞
e−nx xn is uniformly convergent on [1, 100].
P
Problem 4.3.6 Show that the series
n=1

Solution: First we find the maximum of f (x) = xe−x on [1, 100]. Consider
f (x) = xe−x

=⇒ f ′ (x) = e−x − xe−x

Hence
f ′ (x) = 0

=⇒ e−x − xe−x = 0

=⇒ e−x (1 − x) = 0

=⇒ x=1
Also
f ′′ (x) = −e−x − (e−x − xe−x ) = xe−x − 2e−x
Now
f (100) = 100
e100
, f (1) = 1e , f ′ (1) = 0, f ′′ (1) = 1
e
− 2
e
<0
It follows that the maximum of f (x) on [1, 100] is 1e . For x ∈ [1, 100]
1
xe−x ≤
e
 n
1
=⇒ (xe−x )n ≤
e
 n
1
=⇒ xn e−nx ≤
e
X∞ ∞
X  1 n
n −nx
=⇒ x e ≪
n=1 n=1
e
∞
1 n
converges as it is a geometric series and 1e < 1. So by Weierstrass
P

The series e
n=1
∞
xn e−nx on [1, 100] of follows.
P
M-test, the uniform convergence
n=1
Self-Test 01

S25011: REAL ANALYSIS Page 154

 ∞
P
Problem 4.3.7 If un is a series of differentiable functions that converge pointwise to
n=1
f on [a, b] then

A. f is differentiable on [a, b].

B. f is not continuous on [a, b].

C. f is continuous on [a, b] but may not be differentiable on [a, b].

D. none of these.
∞
sin nx
P
Problem 4.3.8 The series n
n=1

A. converges pointwise but not uniformly on [0.1, π2 ].

B. converges uniformly on [0.1, π2 ].

C. does not converge on [0.1, π2 ].

D. converges pointwise to a discontinuous function on [0.1, π2 ].

Short Answer Questions 01

∞
xn converges pointwise to the function f (x) =
P
Problem 4.3.9 Prove that the series
n=1
x
1−x
on (−1, 1).
Solution: For each n ∈ I, un : (−1, 1) → R is defined as
un (x) = xn
∞
Let {sn }∞
P
n=1 be a sequence of partial sums of the series un . For x ∈ (−1, 1), consider
n=1

lim sn (x) = lim u1 (x) + u2 (x) + · · · + un (x)
n→∞ n→∞

= lim (x + x2 + · · · + xn )
n→∞

x (1 − xn )
= lim
n→∞ 1−x
 
x n
= 1 − lim x
1−x n→∞
 
x n
= ∵ lim x = 0 as |x| < 1
1−x n→∞

= f (x)
∞
xn converges pointwise to
P
Thus lim sn (x) = f (x) for all x ∈ (−1, 1). So the series
n→∞ n=1
x
the function f (x) = 1−x
on (−1, 1).
∞
sin nx
P
Problem 4.3.10 Prove that the series n2
converges uniformly on R.
n=1

S25011: REAL ANALYSIS Page 155

Solution: For each n ∈ I, define un : R → R as un (x) = sinn2nx and Mn = 1
n2
.
sin nx |sin nx| 1
2
≤ 2
≤ 2 for all x ∈ R and n ∈ I
n n n
1
=⇒ un (x) < 2 for all x ∈ R and n ∈ I
∞ ∞
n
1
P P
Also Mn = n2
is p series with p = 2 > 1 and so is convergent.
n=1 n=1
∞
X
Mn < ∞ . . . (II)
n=1
Applying Weierstrass M-test to (I) and (II), we obtain the uniform convergence of the
∞
sin nx
P
series n2
on R.
n=1

Summary:
∞
P
• The infinite series un of real valued functions on set E can be thought as an
n=1
infinite sum of the terms of the sequence {un }∞
n=1 of real valued functions on E.
∞
P
The series un is also denoted by u1 + u2 + · · · + un + · · ·
n=1
∞
P
To “define” the series un of real valued functions on , two sequences need
n=1
to be specified. First the sequence {un }∞
n=1 of real valued functions and the

sequence {sn }∞ ∞
n=1 = {u1 + u2 + · · · + un }n=1 of partial sums.
∞
P
• All the properties of the series un are defined using the properties of the
n=1
sequence {sn }∞
n=1 .
∞
P
• A series un of real valued functions on E converges pointwise to f on E if
n=1
the sequence {sn }∞ ∞
n=1 = {u1 + u2 + · · · + un }n=1 converges pointwise to f on E.
∞
P P∞
In this case we write un = f or un (x) = f (x) for all x ∈ E.
n=1 n=1
∞ ∞
P P pointwise
The pointwise convergence of un to f on E is denoted by un −−−−−→ f .
n=1 n=1

• A series of functions converges if and only if sequence of partial sums of functions

converges.
∞
pointwise pointwise
un −−−−−→ f on E ⇐⇒ {sn }∞
P
n=1 −−−−−→ f on E
n=1
∞
pointwise pointwise
un −−−−−→ f on E ⇐⇒ {u1 + u2 + · · · + un }∞
P
n=1 −−−−−→ f on E
n=1

• A series of functions converges to a function (if it is convergent) just as a series

of real numbers converges to a real number (if it is convergent).
∞
P
• A series un of real valued functions on E converges uniformly to f on E if
n=1
the sequence {sn }∞ ∞
n=1 = {u1 + u2 + · · · + un }n=1 converges uniformly to f on E.
∞ ∞
P P unif orm
We denote the uniform convergence of un to f on E by un −−−−−→ f .
n=1 n=1

S25011: REAL ANALYSIS Page 156

 • A series of functions converges uniformly if and only if sequence of partial sums
of functions
∞
converges uniformly.
unif orm unif orm
X
un −−−−−→ f on E ⇐⇒ {sn }∞
n=1 −
−−−−→ f on E
n=1
∞
unif orm unif orm
X
un −−−−−→ f on E ⇐⇒ {u1 + u2 + · · · + un }∞
n=1 −
−−−−→ f on E
n=1

• It is clear that the uniform convergence is stronger than the pointwise conver-
gence.
∞ ∞
pointwise unif orm
X X
un −−−−−→ f =⇒ un −−−−−→ f
n=1 n=1

• A series of continuous functions converges to a continuous function under uniform

convergence.
∞ ∞
unif orm unif orm
X X
un −−−−−→ f =⇒ un −−−−−→ f
|{z} |{z}
n=1 continuous n=1 continuous

• A sequence of Riemann integrable functions converges to a Riemann integrable

function under uniform convergence.

End of Unit Exercises

∞
x2
P
Problem 4.3.11 Does the series (1+x2 )n
uniformly converge on R?
n=0
∞
1
is uniformly convergent on R+ .
P
Problem 4.3.12 Show that the series n2 +x2
n=1

Keywords: Series of Functions, Pointwise Convergences of Series of Functions, Uniform

Convergence of Series of Functions, Properties preserved under Uniform Convergence of
Series of Functions.

REFERENCES

(1) Sections 9.3 to 9.5, Methods of Real Analysis by Richard R. Goldberg, 2nd edi-
tion, Oxford & IBH Publishing Co. Pvt Ltd.

(2) Chapter 9, Introduction to Real Analysis by Bartle Robert G and Sherbert

Donald R, 4nd edition, Wiley India.

(3) Chapter 7, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,

Mc Graw HillInc.,USA.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA

S25011: REAL ANALYSIS Page 157

OER
BOOKS

S25011: REAL ANALYSIS Page 158

 Unit 04-04: Power Series

Learning Objectives:
After successful completion of this unit, you will be able to

i. Use Weirstrass M-Test for convergence of power series.

ii. State sufficient conditions for convergence of power series.

iii. Differentiate power series term by term.

iv. Integrate power series term by term.

Introduction: In this unit, we shall prove some sufficient conditions for uniform conver-
gence of power series on intervals. With suitable conditions, uniformly convergent power
series can be term by term integrated and differentiated on its interval of convergence.
∞
ak xk converges for x = x0 then it converges
P
Theorem 4.4.1 If the power series
k=0
uniformly on [−x1 , x1 ] where x1 is any number such that 0 < x1 < |x0 |.
∞
ak xk converges absolutely for
P
Proof: By known result for series of real numbers, if
k=0
x = x0 then it converges absolutely for any x with |x| < x0 . This implies that it converges
absolutely for x = x1 where x1 is a real number such that 0 < x1 < |x0 |. So the series
∞
|ak |xk1 converges. Also
P
k=0
∞
X ∞
X
k
0 < |x| < x1 =⇒ ak x ≪ |ak |xk1
k=0 k=0
∞
X ∞
X
=⇒ ak x k ≪ Mk for all x ∈ [−x1 , x1 ] where Mk = |ak |xk1 . . . (I)
k=0 k=0
As seen before,
∞
X ∞
X
Mk = |ak |xk1 is convergent . . . (II)
k=0 k=0
Applying Weierstrass M-test to (I) and (II), we obtain the uniform convergence of the
∞
ak xk on [−x1 , x1 ].
P
series
k=0
∞
P
Theorem 4.4.2 Let un be a series of continuous non-negative real valued functions
n=1
∞
P
on compact metric space M . If the series un converges to a continuous function f on
n=1
∞
P
M then the series un converges uniformly to f on M .
n=1
Proof: We shall use the following theorem.
“Dini’s Theorem: Let {fn }∞
n=1 be a monotone sequence of continuous real valued

functions on compact metric space < M, d >. If {fn }∞

n=1 converges pointwise on M to a

continuous function f then {fn }∞

n=1 converges uniformly to f on M .”

S25011: REAL ANALYSIS Page 159

For each n ∈ I define sn = u1 + u2 + · · · + un . {sn }∞
n=1 is a sequence of partial sums of
∞
P
the series un where each un is continuous non-negative real valued function defined
n=1
on compact metric space M . This implies that
uk (x) ≥ 0 for all x ∈ M and k ∈ I

=⇒ un+1 (x) ≥ 0 for all x ∈ M and n ∈ I

=⇒ sn+1 (x) − sn (x) ≥ 0 for all x ∈ M and n ∈ I

=⇒ sn+1 (x) ≥ sn (x) ≥ 0 for all x ∈ M and n ∈ I

=⇒ {sn }∞
n=1 is a monotone sequence of functions . . . (I)
As each un is continuous on M ,
sn is continuous on M for all n ∈ I . . . (II)
∞
P
Also as the series un converges to a continuous function f on M ,
n=1
{sn }∞
n=1 converges to the continuous function f on M . . . (III)
The statements (I), (II) and (III) assert that the hypothesis of Dini’s theorem is satisfied.
∞
P
Hence by Dini’s theorem, we conclude that the series un converges uniformly to f on
n=1
M.
∞ ∞
an xn on (−1, 1) then f is
P P
Theorem 4.4.3 If the series an converges and f (x) =
n=0 n=0
continuous on (−1, 1).
Proof : ∞
X
an converges
n=0
∞
X
=⇒ an xn converges for x = 1
n=0
X∞
=⇒ an xn converges uniformly on (−1, 1)
n=0
A series of continuous functions converges to a continuous function under uniform con-
vergence. Since un (x) = an xn is continuous on (−1, 1) for each n ∈ I, it follows that
∞ ∞
an xn is continuous on (−1, 1).
P P
f (x) = un (x) =
n=0 n=0

Theorem 4.4.4 Let A be a dense subset of metric space M and u1 , u2 , . . . be continuous

∞
P P∞
functions on M . If un converges uniformly on A then the series un converges
n=1 n=1
uniformly on M .
Proof : We shall use the following result:
“If A is dense in metric space M and {fn }∞
n=1 is a sequence of continuous functions on

M converging uniformly on A then {fn }∞

n=1 converges uniformly on M ” . . . (∗)
Take {sn }∞ ∞
n=1 = {u1 + u2 + · · · + un }n=1 .

S25011: REAL ANALYSIS Page 160

Now, ∞
X
un converges uniformly on A
n=1

=⇒ {sn }∞
n=1 converges uniformly on A

=⇒ {sn }∞
n=1 converges uniformly on M ( by (∗))
∞
X
=⇒ un converges uniformly on M
n=1
∞
P
Theorem 4.4.5 Let uk be a series of functions in R[a, b] which converges uniformly
k=1
to f on [a, b]. Then f ∈ R[a, b] and
Zb X ∞ ∞ Zb
X
uk (x)dx = uk (x)dx
a k=1 k=1 a
Proof: We shall use the following result.
“ Let {fn }∞n=1 be a sequence of functions in R[a, b] which converges uniformly to f on
Rb Rb
[a, b]. Then f ∈ R[a, b] and lim fn (x)dx = lim fn (x)dx” . . . (∗)
n→∞ a a n→∞
For each n ∈ I define sn = u1 + u2 + · · · + un . The sequence{sn }∞
n=1 is then a sequence of
∞
P ∞
P
partial sums of the series uk . As the series uk converges uniformly to f on [a, b],
k=1 k=1
{sn }∞
n=1 converges uniformly to f on [a, b] . . . (I)
Also un ∈ R[a, b] for each n ∈ I. This implies that
un ∈ R[a, b] for each n ∈ I . . . (II)
Applying result (∗) to (I) and (II), we get that f ∈ R[a, b] and
Zb Zb
lim sn (x)dx = lim sn (x)dx
n→∞ n→∞
a a

Zb X
n Zb n
X
=⇒ lim uk (x)dx = lim uk (x)dx
n→∞ n→∞
a k=1 a k=1

n Z b Zb n
X X
=⇒ lim uk (x)dx = lim uk (x)dx
n→∞ n→∞
k=1 a a k=1

∞ Z b Zb X
∞
X
=⇒ uk (x)dx = uk (x)dx
k=1 a a k=1

Theorem 4.4.6 Let u1 , u2 , . . . be differentiable and u′1 , u′2 , . . . be continuous functions on

∞ ∞
u′k converges uniformly to f on
P P
[a, b]. If uk converges to f on [a, b] and the series
k=1 k=1
∞
′
u′k (x)
P
[a, b] then f (x) = for all x ∈ [a, b].
k=1
Proof: We shall use the following result.
“Let each of fn be differentiable and each of fn′ be continuous on [a, b]. If {fn }∞
n=1 converges

S25011: REAL ANALYSIS Page 161

to f on [a, b] and if {fn′ }∞
n=1 converges uniformly to 
g on [a, b] then
′
′ ′
g(x) = lim fn (x) = f (x) = lim fn (x)
n→∞ n→∞
for all x ∈ [a, b]” . . . (*)
For each n ∈ I, define sn = u1 + u2 + · · · + un . The sequence{sn }∞
n=1 is then a sequence
∞
un . Each un is differentiable and each u′n is continu-
P
of partial sums of the series
n=1
ous on [a, b]. The sum of differentiable functions is differentiable and hence each sn is
differentiable on [a, b]. Also the continuity of all u′n implies the continuity of each of s′n .
X∞
uk converges to f on [a, b] =⇒ {sn }∞ n=1 converges to f on [a, b] . . . (I)
k=1
∞
X
u′k converges uniformly to f on [a, b] =⇒ {s′n }∞
n=1 converges uniformly on [a, b] . . . (II)
k=1
Applying result (∗) to (I) and (II),  ′
lim s′n (x) = lim sn (x)
n→∞ n→∞

∞ ∞
!′
X X
=⇒ u′n (x) = un (x)
n=1 n=1
∞
X
=⇒ u′n (x) = f ′ (x)
n=1
This holds for all x ∈ [a, b].
∞
an xn = f (x) on (−S, S) for some S > 0 then f is differentiable on
P
Theorem 4.4.7 If
n=0
∞
′
nan xn−1 for all x ∈ (−S, S).
P
(−S, S) and f (x) =
n=1
p ∞
an x n
P
Proof: Take L = lim sup n
|an |. By root test (Appendix Unit C.6), the series
n→∞ n=0
p
converges on (−S, S) if lim sup n |an xn | < 1. This implies that
n→∞  
p
n
lim |an | |x| < 1
n→∞

=⇒ L|x| < 1
1
=⇒ |x| <
∞
L
n
P
an x = f (x) for all x ∈ (−S, S). This implies the uniform convergence of the series
n=0
∞
an xn on (−S, S). Hence its term by term differentiation is valid on (−S, S). For
P
n=0
x ∈ (−S, S),
∞
X
f (x) = an x n
n=0

=⇒ f (x) = a0 + a1 x + a2 x2 + · · · + an xn + · · ·

=⇒ f ′ (x) = a1 + 2a2 x + · · · + nan xn−1 + · · ·

S25011: REAL ANALYSIS Page 162

 ∞
X
=⇒ f ′ (x) = nan xn−1
n=1

∞
an xn on (−S, S) for some S > 0 then for any k ∈ I then
P
Theorem 4.4.8 If f (x) =
n=1
∞
f (k) (x) exists for all x ∈ (−S, S) and f (k) (x) = n(n − 1) · · · (n − (n − k))an xn−k .
P
n=k
Proof: We shall use the following theorem.
∞
an xn converges to f (x) on (−S, S) then f is differentiable on (−S, S) and f ′ (x) =
P
“ If
n=0
∞
nan xn−1 for all x ∈ (−S, S).”
P
. . . (∗)
n=1
Applying the result (∗) on (−S, S) to f (x) = a0 + a1 x + a2 x2 + · · · + an xn + · · ·,
f ′ (x) = a1 + 2a2 x + · · · + nan xn−1 + · · ·
Again applying the result (∗) on (−S, S) to f ′ (x)
f ′′ (x) = 2a2 + (3 · 2 · 1)a3 x + · · · + n(n − 1)an xn−2 + · · ·
Continuing in this way,
f ′′′ (x) = (3 · 2 · 1)a3 + (4 · 3 · 2 · 1)a4 x + · · · + n(n − 1)(n − 2)an xn−3 + · · ·
..
.

=⇒ f (k) (x) = (k(k − 1)(k − 2) · · · 1)ak + ((k + 1)k · · · 1)ak−1 x + · · · + (n(n − 1) · · · (n − k + 1)an xn−k
∞
X
(k)
=⇒ f (x) = n(n − 1) · · · (n − (n − k))an xn−k
n=k

Solved Problems 01
∞
nx2
P
Problem 4.4.9 Prove that the series n3 +x3
is uniformly convergent on [0, 100]. Hence
n=1
∞
nx2 n

P
or otherwise prove that lim n 3 +x3 = n3 +1
.
x→1 n=1
Solution: For x ∈ [0, 100], n ∈ I
0 ≤ x ≤ 100

=⇒ x2 ≤ 1002 and x3 ≥ 0

=⇒ nx2 ≤ n1002 and n3 + x3 ≥ n3

1 1
=⇒ nx2 ≤ n1002 and ≤ 3
n3 +x 3 n
nx2 1002
=⇒ ≤
n 3 + x3 n2
∞ ∞
X nx2 X 1002
=⇒ ≪
n=1
n 3 + x3 n=1
n2
∞ ∞
1002 nx2
P P
The series n2
converges. The uniform convergence of the series n3 +x3
on [0, 100]
n=1 n=1
follows by the Weierstrass M-Test.

S25011: REAL ANALYSIS Page 163

 ∞
nx2
P
Let f (x) = n3 +x3
on [0, 100]. A series of continuous functions converges to a continu-
n=1
nx2
ous function under uniform convergence. Since un (x) = n3 +x3
is continuous on [0, 100] for
each n ∈ I, it follows that f (x) is continuous on [0, 100]. In particular f (x) is continuous
as x = 1. So ∞
! ∞
X nx2 X n
lim f (x) = f (1) =⇒ lim =
x→1 x→1
n=1
n3 + x3 n=1
n3 +1
∞
P
Problem 4.4.10 If |an | < ∞, prove that
n=0
Z1 X ∞
! ∞
n
X an
an x dx =
n=0 n=0
n+1
0
Solution:
x ∈ [0, 1]

=⇒ 0 ≤ x ≤ 1

=⇒ xn ≤ 1 for all n ∈ I

=⇒ |an |xn ≤ |an | for all n ∈ I

∞
X ∞
X
n
=⇒ an x ≪ |an |
n=1 n=1
∞ ∞
an xn on [0, 1]
P P
Since |an | < ∞, by Weierstrass M-Test the uniform convergence of
n=1 n=1
∞
an xn can be term by term integrated on [0, 1].
P
follows.So the series
n=1
Z1 X
∞ ∞ Z
X
1
n
an x dx = an xn dx
0 n=0 n=0 0

Z1 X
∞ ∞
n
X an xn+1 1
=⇒ an x dx = [ ]
n=0 n=0
n+1 0
0
Z1 X
∞ ∞
X an
=⇒ an xn dx =
n=0 n=0
n+1
0

Self-Test 01
∞
1
P
Problem 4.4.11 The series n3 +x3
is
n=0

A. pointwise convergent but not uniformly convergent on [0, ∞).

B. uniformly convergent on [0, ∞).

C. not pointwise convergent on [0, ∞).

D. none of these.

S25011: REAL ANALYSIS Page 164

 ∞
xn
P
Problem 4.4.12 The series n
is
n=0

A. pointwise convergent but not uniformly convergent on [−0.5, 0.5].

B. uniformly convergent on [−0.5, 0.5].

C. not pointwise convergent on [−0.5, 0.5].

D. none of these.

Short Answer Questions 01

Problem 4.4.13 Deduce the2series3

x x
x− + − · · · = log(1 + x) for |x| < 1
2 3
from the series
1
1 − x + x2 − · · · = for |x| < 1
1+x
Solution: Consider
(−1)n xn
lim
n→∞ (−1)n−1 xn−1

= lim |x|
n→∞

= |x|
∞
(−1)n xn = 1 − x + x2 − · · · is convergent
P
By ratio test (Appendix Unit C.6), the series
n=0
∞
(−1)n xn is uniformly convergent
P
when |x| < 1. This results into the fact that the series
n=0
on (−1, 1). Hence its term by term integration is valid on (−1, 1).
1
1 − t + t2 − · · · = for |t| < 1
1+t
Zx Zx Zx Zx
1
=⇒ 1dt − tdt + t2 dt − · · · = dt for |x| < 1
1+t
0 0 0 0
2 3
x x
=⇒ x − + − · · · = log(1 + x) for |x| < 1
2 3
Problem 4.4.14 Deduce the series
1
1 + 2x + 3x2 + · · · = for |x| < 1
(1 − x)2
from the series
1
1 + x + x2 + · · · = for |x| < 1
1−x
Solution: Consider
(−1)n xn
lim
n→∞ (−1)n xn−1

= lim |x|
n→∞

= |x|

S25011: REAL ANALYSIS Page 165

 ∞
xn = 1 + x + x2 + · · · is convergent when |x| < 1. This results
P
By ratio test the series
n=0
∞
xn is uniformly convergent on (−1, 1). Hence its term by
P
into the fact that the series
n=0
term differentiation is valid on (−1, 1).
1
1 + x + x2 + · · · = for |x| < 1
1−x
 
d d d 2

d 1
=⇒ (1) + (x) + x + ··· = for |x| < 1
dx dx dx dx 1 − x
1
=⇒ 1 + 2x + 3x2 + · · · = for |x| < 1
(1 − x)2
Summary:

• Weierstrass M-Test: A series of real valued functions dominated by fixed con-

vergent series∞of positive real
∞
numbers converges
∞
uniformly.
unif orm
X X X
∀x( uk (x) ≪ Mn ) =⇒ un −−−−−→ f on E
k=1
|n=1{z } n=1
convergent

• ∞ ∞
unif orm
X X
k
ak x converges for x0 =⇒ un −−−−−→ f on (−|x0 |, |x0 |)
k=1 n=1

• A series of continuous non-negative real valued functions on compact space con-

verging
∞
to a continuous function converges uniformly.
∞
pointwise unif orm
X X
un −−−−−→ f on |{z}
M =⇒ un −−−−−→ f
|{z} |{z}
n=1 non−ve continuous compact n=1

• A uniformly convergent series of Riemann integrable functions can be term by

∞
integrated. Z Z Z Z
unif ormly
X
un −−−−−−→ f =⇒ (u1 +u2 +· · ·+un +· · · ) = u1 + u2 +· · ·+ un +· · ·
|{z}
n=1 Riemann integrable

∞ ∞
pointwise unif orm
u′n
P P
• un −−−−−→ f and −−−−−→ on[a, b] =⇒
n=1 n=1 |{z}
continuous
(u1 + u2 + · · · + un + · · · )′ = u′1 + u′2 + · · · + u′n + · · · on [a, b]
∞
∞ ′ ∞
n n
nan xn−1
P P P
• an x converges on (−r, r) for some r > 0 =⇒ u n an x =
n=0 n=0 n=1
More generally, !(k)
X∞ ∞
X
n
un an x = n(n − 1) · · · (n − (k − 1))an xn−1
n=0 n=k

End of Unit Exercises

∞ ∞
an xn converge uniformly on [0, 1].
P P
Problem 4.4.15 If |an | < ∞, show that
n=0 n=0

S25011: REAL ANALYSIS Page 166

Problem 4.4.16 By using term by term differentiation of power series, deduce the series
x2 x4
cos x = 1 − + − · · · on R
2! 4!
from the series
x3 x5
sin x = x − + − · · · on R
3! 5!
x2 x4 x2n
Problem 4.4.17 If 1 + 1!
+ 2!
+ ··· + n!
+ · · · = f (x) on R, show that f ′ (x) = 2xf (x)
on R.

Keywords: Power Series, Term by Term Differentiation of Power Series, Term by Term
Integration of Power Series, Weirstrass M test for Convergence of Power Series.

REFERENCES

(1) Sections 9.3 to 9.5, Methods of Real Analysis by Richard R. Goldberg, 2nd edi-
tion, Oxford & IBH Publishing Co. Pvt Ltd.

(2) Chapter 9, Introduction to Real Analysis by Bartle Robert G and Sherbert

Donald R, 4nd edition, Wiley India.

(3) Chapter 7, Principles of Mathematical Analysis, by Walter Rudin, 3nd edition,

Mc Graw HillInc.,USA.

MOOCS
YOUTUBE VIDEOS
WIKIPEDIA
OER
BOOKS

S25011: REAL ANALYSIS Page 167

 Answers to Self-Tests and End of Unit Exercises

Credit 01: Metric Spaces

Unit 01-01: Metric Spaces

1.1.21: Option D 1.1.22: Option D

1.1.25 Solution: Let x, y, z ∈ X.

d(x,y)
i) As d(x, y) ≥ 0, δ(x, y) = 1+d(x,y)
≥ 0.
ii)
δ(x, y) = 0
d(x, y)
⇐⇒ =0
1 + d(x, y)
⇐⇒ d(x, y) = 0

⇐⇒ x=y
iii) Since d(x, y) = d(y, x),
d(x, y) d(y, x)
δ(x, y) = = = δ(y, x)
1 + d(x, y) 1 + d(y, x)
iv) We prove the triangle inequality by the following argument.
d(x, y) ≤ d(x, z) + d(z, y)

=⇒ 1 + d(x, y) ≤ 1 + d(x, z) + d(z, y)

1 1
=⇒ ≥
1 + d(x, y) 1 + d(x, z) + d(z, y)
−1 −1
=⇒ ≤
1 + d(x, y) 1 + d(x, z) + d(z, y)
−1 −1
=⇒ 1 + ≤1+
1 + d(x, y) 1 + d(x, z) + d(z, y)
1 + d(x, y) − 1 1 + d(x, z) + d(z, y) − 1
=⇒ ≤
1 + d(x, y) 1 + d(x, z) + d(z, y)
d(x, y) d(x, z) + d(z, y)
=⇒ ≤
1 + d(x, y) 1 + d(x, z) + d(z, y)
d(x, y) d(x, z) d(z, y)
=⇒ ≤ +
1 + d(x, y) 1 + d(x, z) + d(z, y) 1 + d(x, z) + d(z, y)
d(x, y) d(x, z) d(z, y)
=⇒ ≤ +
1 + d(x, y) 1 + d(x, z) 1 + d(z, y)
169
 =⇒ δ(x, y) ≤ δ(x, z) + δ(z, y)

1 1
1.1.26 Hint: To prove that x
− y
= 0 if and only if x = y, use the fact that
x, y ∈ (0, ∞).

1.1.27 Hint: To prove triangle inequality,

p
0 ≤ 2 |x − y||y − z|
p
=⇒ |x − z| ≤ |x − z| + 2 |x − y||y − z|
p
=⇒ |x − z| ≤ |x − y| + |y − z| + 2 |x − y||y − z|
p p p
=⇒ |x − z| ≤ |x − y| + |y − z|

=⇒ d(x, z) ≤ d(x, y) + d(y, z)

Unit 01-02: Open and Closed Sets

1.2.20: Option B 1.2.21: Option A 1.2.22: Option A

1.2.23: Option A 1.2.24: Option C 1.2.25: Option C

1.2.28 Solution: Consider the following argument.

A point x in X is limit point of E
=⇒ there is a sequence xn in E such that xn → x in X
=⇒ f (xn ) → f (x) in X ( ∵ f is continuous )
=⇒ lim f (xn ) = f (x)
n→∞
=⇒ lim xn = f (x) ( ∵ f (xn ) = xn as xn ∈ D )
n→∞
=⇒ x = f (x) ( ∵ lim xn = x as xn → x )
n→∞
=⇒ x ∈ E (by definition of E)
Thus x is limit point of E implies that x is in E. In other words, E contains all its limit
points. This proves that E is closed in X.

1.2.29 Solution: Let A = {x1 , x2 , . . . , xn } and U = X − A. Take x ∈ U . Select a real

number r such that
0 < r < min{d(x, x1 ), d(x, x2 ), d(x, x3 ), . . . , d(x, xn )}
We claim that B(x, r) ⊆ U . Suppose on contrary that B(x, r) contains one of the points
of A = {x1 , x2 , . . . , xn } say xi . Then d(xi , x) < r and
d(xi , x) < r < min{d(x, x1 ), d(x, x2 ), d(x, x3 ), . . . , d(x, xn )} ≤ d(xi , x)
This implies that d(xi , x) < d(xi , x); which is a contradiction. Hence B(x, r) contains no
point of A implying that B(x, r) ⊆ U . Thus for every x ∈ U there is r > 0 such that
B(x, r) ⊆ U . This proves that U is an open subset of X.

S25011: REAL ANALYSIS Page 170

1.2.30 Solution:
BY (y, r) = {z ∈ Y : d(z, y) < r}

= {z ∈ X : d(z, y) < r} ∩ Y

= BX (y, r) ∩ Y
The similar result is true for closed balls.

Unit 01-03: Sequences in Metric Spaces

1.3.23: Option D 1.3.24: Option A 1.3.25: Option B

1.3.26: Option A

1.3.29 Solution: Let x ∈ R and r > 0. Then B(x, r) = (x − r, x + r). There is an

irrational number q between real numbers x − r and x + r. So q ∈ (x − r, x + r) ∩ Qc i.e.
q ∈ B(x, r) ∩ Qc . Hence B(x, r) ∩ Qc ̸= ϕ. This is true for every x ∈ R and for every
r > 0. Therefore Qc is dense in R i.e. Qc = R.

1.3.30 Solution: As B(0.5, 0.2) ∩ Z = ϕ, Z is not dense in the Euclidean metric space
R.

1.3.31 Hint: A finite set can not be dense in n the Euclidean metric space R. Consider
maximum (or minimum) value of finite set. Then construct an open ball to the right
(left) of maximum (minimum) point which does not intersect the finite set.

1.3.32 Solution: The sequence n1 in the Euclidean metric space (0, 1) is Cauchy but

not convergent.

Unit 01-04: Continuity

1.4.14: Option D 1.4.15: Option C 1.4.16: Option C

1.4.17: Option D 1.4.18: Option A 1.4.19: Option C

1.4.20: Option D 1.4.21: Option B

1.4.24 Solution: Define the function f : X → x as f (x) = x for all x ∈ X. For any
x ∈ X,
xn → x in X

=⇒ f (xn ) → f (x) in X (∵ f (xn ) = xn and f (x) = x)

Thus xn → x in X implies f (xn ) → f (x) in X. Hence f is continuous.

1.4.25 Solution: For any (x, y) ∈ R2 ,

(xn , yn ) → (x, y) in R2

=⇒ xn → x in R and yn → y in R

S25011: REAL ANALYSIS Page 171

 =⇒ lim xn = x and lim yn = y
n→∞ n→∞

=⇒ lim (xn + yn ) = x + y and lim (xn − yn ) = x − y

n→∞ n→∞

=⇒ xn + yn → x + y in R and xn − yn → x − y in R

=⇒ f (xn + yn ) → f (x + y) in R and f (xn − yn ) → f (x − y) in R (∵ f : R → R is continuous )

=⇒ (f (xn + yn ), f (xn − yn )) → (f (x + y), f (x − y)) in R2

=⇒ h(xn , yn ) → h(x, y) in R2
Thus (xn , yn ) → (x, y) in R2 implies h(xn , yn ) → h(x, y) in R2 . Hence h is continuous on
R2 .

1.4.26 Solution: Let x ∈ X. Since D is dense in X, x is limit point of D. So there is

a sequence (xn ) in D such that xn → x in X. Consider
f (x) = lim f (xn ) (∵ f is continuous at x)
n→∞


= lim g(xn ) ∵ f (xn ) = g(xn ) as xn ∈ D
n→∞

= g(x) (∵ g is continuous at x)
Thus f (x) = g(x) for all x ∈ X. Hence f = g.

Credit 02. Connected, Complete and Compact Metric Spaces

Unit 02-01: Connected Metric Spaces

2.1.17 Option A 2.1.18: Option A 2.1.19 Option D

2.1.20: Option C

2.1.23 Solution: Any subset in a discrete metric space X is both open and closed in
X. Let a ∈ X. Since |X| ≥ 2,
A = {a} =
̸ ϕ, X

=⇒ A ⊆ X which is both open and closed

=⇒ X is disconnected

A non-singular discrete metric space is disconnected.

2.1.24 Solution: We prove: “ A finite subset of metric space is connected if and only if
it is a singleton set.”
I Part: Let A be singleton subset of a metric space. Then ϕ and A are the only subsets
of A which are both open and closed in A. Hence A is connected.
II Part: Let A be a finite and connected subset of metric space X. Suppose on contrary
that |A| ≥ 2. Then the set {a} where a ∈ A is proper subset of A. Since any finite set is

S25011: REAL ANALYSIS Page 172

closed, {a} and A − {a} are closed. It follows that {a} is both open and closed in A. So
A is not connected. This is a contradiction. Hence |A| = 1 i.e. A is singleton.

2.1.25 Solution: Define f : [0, 2π] → R2 as

f (θ) = (cos θ, sin θ)
Since [0, 2π] is connected and continuous. The image of connected set is connected.
Therefore C is connected.

2.1.26 Solution:
A must be singleton subset of R. This follows from:
i) A subset A of R is connected if and only if it is an interval.
ii) Any interval except possibly [a, a] = {a}, (a ∈ R) contains irrationals.

2.1.27 Solution: Let a, b ∈ R and a < b. Consider

f : [a, b] → R is continuous.


=⇒ f [a, b] is connected.


=⇒ f [a, b] is interval.



=⇒ f [a, b] is singleton (∵ f [a, b] ⊂ Q)

=⇒ f (a) = f (b)
Since a and b are any two points of R, f is constant on R.

2.1.28 Solution: Let f : B → {1, −1} is a continuous function. As A is connected, f is

constant on A. Suppose f (x) = 1 for all x ∈ A. Take b ∈ B.
The function f is continuous. So for the open subset {f (b)} of {1,-1}, there is an open
subset U of A containing b such that f (U ) ⊆ {f (b)}. Since b ∈ B ⊆ Ā, b is limit point of
A. Therefore A ∩ U ̸= ϕ. Take a ∈ A ∩ U . Then
f (a) ∈ f (A ∩ U ) ⊆ f (U ) ⊆ {f (b)}

=⇒ f (a) ∈ {f (b)}

=⇒ f (a) = f (b) = 1
Thus we have shown that any continuous function f : B → {1, −1} is constant. Hence
B is connected. In particular, B = Ā is connected.

2.1.29 Solution: Without loss generality assume that an = 1 (for if an ̸= 1 then divide
by a to the equationp(α) = 0). For x ̸= 0 we write,
p(x) = xn q(x)
an−1 an−2 a1 a0
q(x) = 1 + + 2 + · · · + n−1 + n
x x x x

S25011: REAL ANALYSIS Page 173

For |x| > 1, we have
|an−1 | |an−2 | |a1 | |a0 |
|q(x) − 1| ≤ + + ··· + + n
|x| |x|2 |x| n−1 |x|
|an−1 | |an−2 | |a1 | |a0 |
≤ + + ··· + +
|x| |x| |x| |x|
A
=
|x|
n−1
1
P
where A = |aj |. Thus if |x| > max{1, 2A} then |q(x) − 1| < 2
. Hence q(x) > 0
j=0
for such x. Let |β| > max{1, 2A}. Then q(β) and q(−β) are both positive. But then
p(β) > 0 and p(−β) < 0. Hence by intermediate value theorem there exists α ∈ [−β, β]
such that p(α) = 0.

Unit 02-02: Complete Metric Spaces

2.2.7: Option A 2.2.8: Option B

2.2.11 Solution: Let (xn , yn ) be a Cauchy sequence in R2 . Then for ϵ > 0 there is
N ∈ N such that
m, n ≥ N =⇒ d((xm , ym ), (xn , yn )) < ϵ
This implies that
p
m, n ≥ N =⇒ (xm − xn )2 + (ym − yn )2 < ϵ

√ √
=⇒ m, n ≥ N =⇒ |xm − xn |, |ym − yn | < ϵ (∵ |a| = a2 ≤ a2 + b2 for any a, b ∈ R)
Thus for ϵ > 0 there is N ∈ N such that
m, n ≥ N =⇒ |xm − xn |, |ym − yn | < ϵ
This proves that (xn ) and (yn ) are Cauchy sequences in R. As R is complete,(xn ) and (yn )
are convergent in R. Let xn → x and yn → y in R. This then implies that (xn , yn ) → (x, y)
in R2 . We proved that every Cauchy sequence in R2 is convergent. So R2 is complete.
In general it can be proved that Rn is complete with Euclidean metric on it.

Rn is complete with Euclidean metric on it.

2.2.12 Hint: Prove that (−n) is a Cauchy but not convergent sequence in the metric
space R, where R is equipped with the metric d(x, y) = |ex − ey |.

2.2.13 Hint: Use definition of Cauchy sequence for sequences (xn ) and (yn ) and apply
triangle inequality.

Unit 02-03: Totally bounded subsets of metric spaces

2.3.9 Option B 2.3.10: Option C

S25011: REAL ANALYSIS Page 174

 x
2.3.13 Solution: Define f : (0, 1) → (0, ∞) as f (x) = 1−x
. Then f is homeomorphism
in the Euclidean metric space R. Here (0, 1) is totally bounded subset of R but (0, ∞) is
not totally bounded subset of R.

2.3.14 Hint: Apply theorems 2.3.4 and 2.3.6.

Unit 02-04: Compact Metric Spaces

2.4.12 Option B 2.4.13: Option B 2.4.14 Option C

2.4.15 Option C 2.4.16: Option B

∞
Fn , where F0 = [0, 1], F1 = 0, 13
T   S 2 
2.4.19 Solution: The Cantor set K = 3
, 1 , F2 =
 1 S 2 1 S 2 7 S 8  n=1
0, 9 ,
9 3
,
3 9 9
, 1 , . . .. In general Fn is finite union of closed intervals (closed
sets) in the Euclidean metric space R. Hence each Fn is closed. As arbitrary intersection
of closed sets is closed, K is closed. Also K is bounded as K ⊆ [0, 1] in the Euclidean
metric space R. Thus the Cantor set K is closed and bounded in the Euclidean metric
space R. Hence it is compact in the Euclidean metric space R.

2.4.20 Hint: K is bounded since K is compact. So sup K and inf K exists. By the
definition of sup K , for every n ∈ N, there exists xn ∈ K such that sup K − xn < 1/n.
Then lim xn = sup K where xn ∈ K for every n ∈ N. So sup K is limit point of K. As
n→∞
K is closed, sup K is in K. The proof for infimum is similar.
∞
S
2.4.21 Solution: [−n, n] = R. Here each closed interval [−n, n] is compact in
n=1
the Euclidean metric space R but the Euclidean metric space R is not compact as it is
unbounded.

2.4.22 Hint: Take open cover τ of K1 ∪ K2 . This is also an open cover of K1 and
K2 . As K1 and K2 are compact, the open cover τ will have finite subcovers say τ1 and τ2
that cover K1 and K2 respectively. The union τ1 ∪ τ2 of these finite subcovers is a finite
subcover of τ that covers K1 ∪ K2 . Thus every open cover of K1 ∪ K2 has finite subcover
proving that K1 ∪ K2 is compact.

Credit 3. Riemann Integration

Unit 03-01: Riemann Integral

3.1.19 Option C 3.1.20: Option A

3.1.24 Solution: Suppose on contrary that A − B is of measure zero. Then A =

B ∪ (A − B) is of measure zero as fintie union of sets of measure zero is of measure zero.
This is a contradiction. Hence A − B is not of measure zero.

S25011: REAL ANALYSIS Page 175

3.1.25 Solution: Ik = [ k−1
n
, nk ] for k = 1, 2, . . . n.
Xn
U [f, σn ] = M [f, Ik ]|In |
k=1
n
X k1
=
k=1
nn
n
1 X
= 2 k
n k=1
1 n(n + 1)
=
n2 2
2
n +n
=
2n2
Therefore
1
n2 + n 1+ n 1+0 1
lim U [f, σn ] = lim = lim = =
n→∞ n→∞ 2n2 n→∞ 2 2 2
Similarly, n
X
L[f, σn ] = m[f, Ik ]|In |
k=1
n
X k−11
=
k=1
n n
n−1
X k1
=
k=1
nn
n−1
1 X
= 2 k
n k=1
1 n(n − 1)
=
n2 2
2
n −n
=
2n2
Therefore
1
n2 − n 1− n 1−0 1
lim L[f, σn ] = lim 2
= lim = =
n→∞ n→∞ 2n n→∞ 2 2 2
3.1.26 Solution: Ik = [ k−1
n
, k
n
] for k = 1, 2, . . . n.
Xn
U [f, σn ] = M [f, Ik ]|In |
k=1
n
X k2 1
=
k=1
n2 n
n
1 X 2
= 3 k
n k=1
1 n(n + 1)(2n + 1)
=
n3 6

S25011: REAL ANALYSIS Page 176

Therefore
n(n + 1)(2n + 1) 1(1 + n1 )(2 + n1 ) 1(1 + 0)(2 + 0) 1
lim U [f, σn ] = lim = lim = =
n→∞ n→∞ 6n3 n→∞ 6 6 3
Similarly, n
X
L[f, σn ] = m[f, Ik ]|In |
k=1
n
X (k − 1)2 1
=
k=1
n2 n
n
1X
= (k − 1)2
n k=1
n−1
1 X 2
= k
n3 k=1
1 n(n − 1)(2n − 1)
=
n3 6
Therefore
n(n − 1)(2n − 1) 1(1 − n1 )(2 − n1 ) 1(1 − 0)(2 − 0) 1
lim L[f, σn ] = lim = lim = =
n→∞ n→∞ 6n3 n→∞ 6 6 3
Unit 03-02: Necessary and Sufficient Conditions for Riemann Integrability
3.2.6: Option C 3.2.7: Option C 3.2.8: Option D

3.2.11 Solution: FALSE.

 For example, take f : R → R as f (x) = 1 for all x ∈ R and
 0 if x ∈ Q
g : R → R as g(x) =
 1 if x ̸∈ Q
Then f is continuous on [0, 1] and hence f ∈ R[0, 1] . Also g(x) = f (x) except on Q
which is countably infinite. Since Q is of measure zero, g(x) = f (x) at almost every
x ∈ [0, 1]. Here g(x) is discontinuous everywhere in [0, 1]. Hence g ̸∈ R[0, 1]

3.2.12 Hint: Find Riemann integrable functions f, g on [0, 1] such that gof is 0 for
each irrational and gof is 1 for each rational.

Unit 03-03: Properties of Riemann Integrals

3.3.9 Option A 3.3.10: Option D

3.3.13 Solution: The result is obtained by taking x∗k = k

n
in the Example 3.3.7. For
detailed answer, repeat the answer by taking x∗k = k
n
.

3.3.14 Solution: To evaluate the limits, we use the following result.

n   Z1
1X k
lim f = f (x)dx
n→∞ n n
k=1 0
a) Take f (x) = x2 . "
 2  2  2 # n  
1 1 2 n 1X k
lim + + ··· + = lim f
n→∞ n n n n n→∞ n n
k=1

S25011: REAL ANALYSIS Page 177

 Z1
= f (x)dx
0
Z1
= x2 dx
0
" #1
x3
=
3
0
1
=
3
b) Take f (x) = sin πx.
  n  
1 π 2π nπ 1X k
lim sin + sin + · · · + sin = lim f
n→∞ n n n n n→∞ n k=1 n
Z1
= f (x)dx
0
Z1
= sin πxdx
0
−1  1
= cos πx 0
π
−1
= (−1 − 1)
π
2
=
π
c) Take f (x) = e3x .
n  
1 3 6 3n
 1X k
lim e + e + ··· + e
n n n = lim f
n→∞ n n→∞ n n
k=1
Z1
= f (x)dx
0
Z1
= e3x dx
0
1  3x 1
= e 0
3
1
= (e3 − 1)
3
3.3.15 Solution:
a)
0≤x

=⇒ 1≤1+x

S25011: REAL ANALYSIS Page 178

 √
=⇒ 1≤ 1+x
1
=⇒ √ ≤1 . . . (I)
1+x
Also
x≤1

=⇒ 1 + x ≤ 2
√ √
=⇒ 1 + x ≤ 2
1 1
=⇒ √ ≤√ . . . (II)
2 1+x
From (I) and (II),
x2 x2
√ ≤√ ≤ x2
2 1+x
b) From part a) ,

0≤x≤1
x2 x2
=⇒ √ ≤√ ≤ x2
2 1+x
Z1 Z1 Z1
1 2 x2
=⇒ √ x dx ≤ √ dx ≤ x2 dx
2 1+x
0 0 0
Z1
1 x2 1
=⇒ √ ≤ √ dx ≤
3 2 1+x 3
0

3.3.16 Solution: Since f (x) is continuous at x = c and f (c) > 0, there is δ > 0 such
that
1
a < c − δ < x < c + δ < b =⇒ 0 < f (c) < f (x)
2
This implies that
Zc+δ Zc+δ
1 1
f (x)dx > f (c) dx = 2δf (c) = δf (c) . . . (I)
2 2
c−δ c−δ
Since f (x) ≥ 0 on [a, b],
Zc−δ Zb
f (x)dx ≥ 0 and f (x)dx ≥ 0 . . . (II)
a c+δ
Now,
Zb Zc−δ Zc+δ Zb
f (x)dx = f (x)dx + f (x)dx + f (x)dx
a a c−δ c+δ

> 0 + δf (c) + 0 ( from (I) and (II))

= δf (c)

S25011: REAL ANALYSIS Page 179

 >0

Unit 03-04: Mean Value Theorems and Fundamental Theorems of Calculus

3.4.14 Option C 3.4.15: Option D

3.4.19 Hint: For Rolle’s Mean Value Theorem, draw graph of any differentiable func-
tions on some interval [a, b] such that f (a) = f (b). Then identify the point on graph for
which the tangent is parallel to X axis.
For Lagrange’s Mean Value Theorem, draw graph of any differentiable functions on some
interval [a, b] Then identify the point on graph for which the tangent is parallel to the
secant line joining points a, f (a) and b, f (b).

3.4.20 Solution: a) f (x) is continuous on [a, b] and f (x) is differentiable in (a, b). Also

a2 +ab a2 +ab



f (a) = log (a+b)a
= log (a2 +ab)x
= log 1 = 0

and

b2 +ab b2 +ab



f (a) = log (a+b)b
= log (a+b2 )x
= log 1 = 0

The hypothesis of Rolle’s Mean Value theorem is satisfied. To find c, consider

f ′ (c) = 0

1
=⇒ (c2 +ab)
(2c) = 0
(a+b)c

=⇒ c2 + ab = 0

√
=⇒ c = ab

b) f (x) is continuous on [a, b] and f (x) is differentiable in (a, b). Also

f (a) = (a − a)m (a − b)n = 0

and

f (a) = (b − a)m (b − b)n = 0

S25011: REAL ANALYSIS Page 180

The hypothesis of Rolle’s Mean Value theorem is satisfied. To find c, consider

f ′ (c) = 0

=⇒ m(c − a)m−1 (c − b)n + n(c − a)m (c − b)n−1 = 0

=⇒ (c − a)m−1 (c − b)n−1 (mc − mb + nc − na) = 0

=⇒ mc − mb + nc − na = 0

=⇒ c(m + n) − (mb + na) = 0

mb+na
=⇒ c = m+n

3.4.21 Solution: a) f (x) is continuous on [0, 4] and f (x) is differentiable in (0, 4). So
the hypothesis of Lagrange’s Mean Value Theorem is satisfied. To find value of c, consider

f (4)−f (0)
f ′ (c) = 4−0

−6−6
=⇒ (c − 2)(c − 3) + (c − 1)(c − 3) + (c − 1)(c − 2) = 4

=⇒ 3c2 − 12c + 11 = −3

=⇒ 3c2 − 12c + 8 = 0

√ √
2 3 2 3
=⇒ c = 2 + 3
or c = 2 − 3

=⇒ c = 3.1547 or c = 0.8453

b) f (x) is continuous on [0, 1] and f (x) is differentiable in (0, 1). So the hypothesis of
Lagrange’s Mean Value Theorem is satisfied. To find value of c, consider

f (1)−f (0)
f ′ (c) = 1−0

π
1 −0
=⇒ 1+c2
= 4
1

1 π
=⇒ 1+c2
= 4

S25011: REAL ANALYSIS Page 181

 4
=⇒ 1 + c2 = π

q
4
=⇒ c = ± π
−1

q
=⇒ c = π4 − 1 (∵ c ∈ (0, 1))

3.4.22 Solution: a) The functions f (x) = sin x and g(x) = cos x are continuous on
[a, b] and differentiable in (a, b). So the hypothesis of Cauchy’s Mean Value Theorem is
satisfied. To find c, consider

f ′ (c) f (b)−f (a)

g ′ (c)
= g(b)−g(a)

cos c sin b−sin a

=⇒ − sin c
= cos b−cos a

2 sin( b−a ) cos( b+a )

=⇒ − cot c = 2
−2 sin( b−a
2
) sin( b+a )
2 2

=⇒ − cot c = − cot (b+a)

2

b+a
=⇒ c = 2

=⇒ c is arithmetic mean of a and b.

√ q
1
b) The functions f (x) = x and g(x) = x
are continuous on [a, b] and differentiable in
(a, b). So the hypothesis of Cauchy’s Mean Value Theorem is satisfied. To find c, consider

f ′ (c) f (b)−f (a)

g ′ (c)
= g(b)−g(a)

1 −1 √ √
c 2
=⇒ 2
1 −3
= √ 1b− √a1
−2c 2 b
− a

√ √ √ 
b− a
=⇒ −c = − ba √ √
a− b

√
=⇒ c = ab

=⇒ c is geometric mean of a and b

S25011: REAL ANALYSIS Page 182

 Credit 4. Sequence and Series of Functions

Unit 04-01: Convergence and Uniform Convergence of Sequence of Functions

4.1.9 Option C 4.1.10: Option B

4.1.13 Solution: Take N = 101. Then

x ∈ [0, 1], n ≥ N =⇒ n > 100
 
1 1
=⇒ x ∈ [0, 1], n ≥ N =⇒ <
n 100
! !
sin nx 1 sin nx 1
=⇒ x ∈ [0, 1], n ≥ N =⇒ < ∵ ≤
n 100 n n
!
sin nx 1
=⇒ x ∈ [0, 1], n ≥ N =⇒ −0 <
n 100
 
1
=⇒ x ∈ [0, 1], n ≥ N =⇒ |fn (x) − 0| <
100
4.1.14 Solution: If 0 ≤ x < 1 then
xn
lim fn (x) = lim
n→∞ n→∞ 1 + xn
0
=
1+0
=0
If x = 1 then
xn
lim fn (x) = lim
n→∞ n→∞ 1 + xn
1
=
1+1
1
=
2

Thus {fn }∞
n=1 converges pointwise to fon[0, 1] where f : [0, 1] → R is defined as
 0 if 0 ≤ x < 1
f (x) =
 1 if x = 1
2
Suppose on contrary that there is N ∈ I such that
1
x ∈ [0, 1], n ≥ N =⇒ |fn (x) − f (x)| <
8
 
1
=⇒ x ∈ [0, 1), n ≥ N =⇒ |fn (x) − f (x)| <
8
!
xn 1
=⇒ x ∈ [0, 1), n ≥ N =⇒ n
−0 <
1+x 8
−
Letting x → 1
1 1 1 1
−0 < =⇒ <
2 8 2 8

S25011: REAL ANALYSIS Page 183

This is a contradiction. Hence no such N exists.

4.1.15 Solution: a) Fix x in [0, 1]. By Archimedian property of real numbers there is
N ∈ I such that N > x1 .Consider
n≥N
1
=⇒ n >
x
1
=⇒ <x
n
 
1
=⇒ x ̸∈ 0,
n
=⇒ χn (x) = 0
Thus
n≥N

=⇒ χn (x) = 0

=⇒ χN (x) = χN +1 (x) = χN +2 (x) = · · · = 0

=⇒ N χN (x) = (N + 1)χN +1 (x) = (N + 2)χN +2 (x) = · · · = 0

=⇒ lim nχn (x) = 0

n→∞

=⇒ lim fn (x) = 0
n→∞
This holds for all x ∈ [0, 1]. Hence {fn }∞
n=1 converges to 0 on [0, 1].

b) Suppose on contrary that there is N ∈ I such that

1
x ∈ [0, 1], n ≥ N =⇒ |fn (x) − 0| <
4
 1
=⇒ x ∈ [0, 1], n ≥ N =⇒ nχn (x) <
4
 1 
=⇒ x ∈ [0, 1] =⇒ N χN (x) < . . . (I)
4
Choose x in [0, 1] such that x < N1 . Note that such x always exists, however large N may
be. For this choice of x in [0, 1],  
1
x ∈ 0,
N
=⇒ χN (x) = 1

=⇒ N χN (x) = N
1
=⇒ N < ( from (I))
4

S25011: REAL ANALYSIS Page 184

This is a contradiction. Hence no such N exists.
c)
1
Z1 Zn Z1
fn (x)dx = fn (x)dx + fn (x)dx
0 0 1
n

1
Zn Z1
= nχn (x)dx + nχn (x)dx
0 1
n

1
Zn
=n 1dx + 0
0
1
=n
n
=1

R1 R1
Thus fn (x)dx = 1 and therefore lim fn (x)dx = 1.
0 n→∞ 0
R1 R1
d)From a), lim fn (x)dx = 0dx = 0
0 n→∞ 0
R1
From c), lim fn (x)dx = 1.
n→∞ 0
R1 R1
So lim fn (x)dx ̸= lim fn (x)dx.
0 n→∞ n→∞ 0

4.1.16 Solution: a)If x = 0 or x = 1, it is clear that lim fn (x) = 0. For 0 < x < 1,
n→∞
consider
lim fn (x) = lim nx(1 − x2 )n
n→∞ n→∞
 
nx ∞
= lim form
n→∞ (1 − x2 )−n ∞
x
= lim
n→∞ (1 − x2 )−n log(1 − x2 )
x
= lim (1 − x2 )n
log(1 − x2 ) n→∞
 
x 2 n 2
= .0 ∵ lim (1 − x ) = 0 as 0 < (1 − x ) < 1
log(1 − x2 ) n→∞

=0
Thus for all x ∈ [0, 1], lim fn (x) = 0. This proves that {fn }∞
n=1 converges to 0 on [0, 1].
n→∞
b)
Z1 Z1
fn (x)dx = nx(1 − x2 )n dx
0 0

S25011: REAL ANALYSIS Page 185

 Z1
n
= (1 − x2 )n (−2x)dx
−2
0
" #x=1
2 n+1
n (1 − x )
=
−2 n+1
x=0
 
n 1
= 0−
−2 n+1
n
=
2(n + 1)
Therefore
Z1
n 1
lim fn (x)dx = lim =
n→∞ n→∞ 2(n + 1) 2
( )∞ 0
R1 1
This proves that fn converges to 2
on [0, 1].
0
n=1
R1
c) From a), lim fn (x)dx = 0.
0 n→∞
R1
From b), lim fn (x)dx = 21 .
n→∞ 0
R1 R1
So lim fn (x)dx ̸= lim fn (x)dx.
0 n→∞ n→∞ 0

Unit 04-02: Properties of functions preserved under uniform convergence

4.2.10: Option D 4.2.11: Option A 4.2.12: Option A

4.2.15 Solution: Let ϵ > 0. Since f is uniformly continuous on R, there is δ > 0 such
that
|x′ − x′′ | < δ =⇒ |f (x′ ) − f (x′′ )| < ϵ . . . (I)
{ n1 }∞
n=1 is Cauchy sequence. So there is N ∈ I such that
1 1
m, n ≥ N =⇒ − <δ
m n
 
   
1 1
=⇒ x ∈ R, m, n ≥ N =⇒ x+ − x+ < δ
m n
 
   
1 1
=⇒ x ∈ R, m, n ≥ N =⇒ f x + −f x+ < ϵ ( from (I))
m n
 
=⇒ x ∈ R, m, n ≥ N =⇒ fm (x) − fn (x) < ϵ

=⇒ {fn }∞
n=1 converge uniformly on R.

S25011: REAL ANALYSIS Page 186

  
1 1



As f is continuous, lim fn (x) = lim f x + n
=f lim x + n
= f (x + 0) = f (x).
n→∞ n→∞ n→∞
So {fn }∞ ∞
n=1 converge to f on (−∞, ∞). It follows that {fn }n=1 converge uniformly to f

on R.

4.2.16 Solution: a) For each n ∈ I, fn is increasing on [0, 14 ].(Try to graph fn or find

its derivative). Hence
1
4n
sup fn (x) = 1
x∈[0, 14 ] 1+ 4n
This implies that
1
4n
lim sup fn (x) = lim 1
n→∞ n→∞ 1 +
x∈[0, 14 ] 4n

0
=
1+0
=0
Thus

lim sup fn (x) − 0 = 0

n→∞
x∈[0, 41 ]

This proves that {fn }∞

 1
n=1 converges uniformly on 0, 4
b)If 0 ≤ x < 1 then

xn
lim fn (x) = lim
n→∞ n→∞ 1 + xn
0
=
1+0
=0
If x = 1 then
xn
lim fn (x) = lim
n→∞ n→∞ 1 + xn
1
=
1+1
1
=
2
Thus {fn }∞
n=1 converges pointwise to fon[0, 1] where f : [0, 1] → R is defined as
 0 if 0 ≤ x < 1
f (x) =
 1 if x = 1
2
Suppose on contrary that {fn }∞
n=1 converges uniformly to f on [0, 1]. Then there is N ∈ I
such that

1
x ∈ [0, 1], n ≥ N =⇒ |fn (x) − f (x)| <
4
S25011: REAL ANALYSIS Page 187
  
1
=⇒ x ∈ [0, 1), n ≥ N =⇒ |fn (x) − f (x)| <
4
!
xn 1
=⇒ x ∈ [0, 1), n ≥ N =⇒ − 0 <
1 + xn 4
Letting x → 1−
1 1 1 1
−0 < =⇒ <
2 4 2 4
This is a contradiction. Hence {fn }∞
n=1 does not converge uniformly on [0, 1].
(Alternatively, observe that each fn is continuous at x = 1 but f is not continuous at
x = 1. As sequence of continuous functions converge to a continuous function under
uniform convergence, it follows that {fn }∞
n=1 does not converge uniformly to f on [0, 1].)

4.2.17 Solution: Let ϵ > 0. Since {fn }∞

n=1 converge uniformly to f on [a, b], there is

N ∈ I such that
ϵ
t ∈ [a, b], m, n ≥ N =⇒ |fm (t) − fn (t)| <
b−a
This implies that
Zx Zx
ϵ
x ∈ [a, b], m, n ≥ N =⇒ |fm (t) − fn (t)|dt < dt
b−a
a a
Zx !
ϵ
=⇒ x ∈ [a, b], m, n ≥ N =⇒ |fm (t) − fn (t)|dt < (x − a)
b−a
a
Zx !  
x−a
=⇒ x ∈ [a, b], m, n ≥ N =⇒ |fm (t) − fn (t)|dt < ϵ ∵ <1
b−a
a
Zx !  Zx Zx 
=⇒ x ∈ [a, b], m, n ≥ N =⇒ (fm (t) − fn (t))dt < ϵ ∵ g(t)dt ≤ |g(t)|dt
a a a
Zx Zx !
=⇒ x ∈ [a, b], m, n ≥ N =⇒ fm (t)dt − fn (t)dt < ϵ
a a
!
=⇒ x ∈ [a, b], m, n ≥ N =⇒ |Fm (x) − Fn (x)| < ϵ

=⇒ {Fn }∞
n=1 converges uniformly on [a, b].

4.2.18 Solution: Let {fn }∞

n=1 converge uniformly to f on [0, 1]. A sequence of contin-

uous functions converge to a continuous function under uniform convergence. Therefore

f is continuous. A continuous function on closed bounded interval is bounded. So there
is constant M ′ > 0 such that
x ∈ [0, 1] =⇒ |f (x)| ≤ M ′ . . . (I)

S25011: REAL ANALYSIS Page 188

As {fn }∞
n=1 converge uniformly to f , there is N ∈ I such that
x ∈ [0, 1], n ≥ N =⇒ |fn (x) − f (x)| < 1
Consider
 
′
x ∈ [0, 1], n ≥ N =⇒ |fn (x) − f (x)| + |f (x)| < M + 1 ( from (I))
 
′
=⇒ x ∈ [0, 1], n ≥ N =⇒ |fn (x) − f (x) + f (x)| < M + 1
 
′
=⇒ x ∈ [0, 1], n ≥ N =⇒ |fn (x)| < M + 1
 
′
=⇒ x ∈ [0, 1] =⇒ |fN (x)|, |fN +1 (x)|, |fN +2 (x)|, . . . , < M + 1 . . . (II)
A continuous function on closed bounded interval is bounded. So there are constants
M1 , M2 , . . . , MN −1 such that
x ∈ [0, 1] =⇒ |f1 (x)| ≤ M1 , |f2 (x)| ≤ M2 , . . . , |fN −1 (x)| ≤ MN −1 . . . (III)
Take M = max{M1 , M2 , . . . , MN −1 , M ′ + 1}. Then from (II) and (III),
x ∈ [0, 1], n ∈ I =⇒ |fn (x)| ≤ M

ex
4.2.19 Solution: Consider the sequence {fn }∞
n=1 on (−∞, ∞) where fn (x) = n
. Since
x
lim fn (x) = lim e = ex lim n1 = 0 on R, {fn }∞
n=1 converges pointwise to 0 on R.
n→∞ n→∞ n n→∞
Suppose on contrary that there is M > 0 such that
x ∈ R, n ∈ I =⇒ |fn (x)| ≤ M
In particular for n = 1 and x = log 2M ,
|f1 (log 2M )| ≤ M

=⇒ elog 2M ≤ M

=⇒ 2M ≤ M

=⇒ 2≤1
This is a contradiction. Hence no such M exists.

−x
4.2.20 Solution: Consider the sequence {fn }∞
n=1 on R
+
where fn (x) = e n . Since
−x
lim fn (x) = lim e n = e0 = 1 on R+ , {fn }∞ +
n=1 converges pointwise to 1 on R .
n→∞ n→∞
+
Also for x ∈ R

n1 ≤ n2
1 1
=⇒ ≥
n1 n2
x x
=⇒ ≥
n1 n2
−x −x
=⇒ ≤
n1 n2

S25011: REAL ANALYSIS Page 189

 −x −x
=⇒ e n1 ≤ e n2

=⇒ fn1 (x) ≤ fn2 (x)

Thus
x ∈ [0, ∞) =⇒ f1 (x) ≤ f2 (x) ≤ f3 (x) ≤ . . .
In this sense, {fn }∞
n=1 is monotone on [0, ∞).

Suppose on contrary that {fn }∞

n=1 is uniformly convergent on [0, ∞).Then there is N ∈ I

such that
1
x ∈ [0, ∞), n ≥ N =⇒ |fn (x) − 1| < 1 −
e
In particular for n = N, x = N ,

1
|fN (N ) − 1| < 1 −
e
−N 1
=⇒ |e N − 1| < 1 −
e
1 1
=⇒ −1 <1−
e e
1 1
=⇒ 1− <1−
e e
This is a contradiction. Hence {fn }∞
n=1 does not converge uniformly on [0, ∞).

4.2.21 Solution: Let ϵ > 0. Since {fn }∞

n=1 converge uniformly to f on R, there is

N1 ∈ I such that    
1 1 ϵ
x ∈ R, n ≥ N1 =⇒ fn x + −f x+ < . . . (I)
n n 2
The continuity of f at x implies the existence of δ > 0 such that
ϵ
|x′ − x| < δ =⇒ |f (x′ ) − f (x)| < . . . (II)
2
As n1 converges to 0, there is N2 ∈ I such that
1
n ≥ N2 =⇒ < δ . . . (III)
n2
Take N = max{N1 , N2 }. Consider
n≥N
1
=⇒ < δ ( from (III) )
n
 
1
=⇒ x+ −x <δ
n
 
1 ϵ
=⇒ f x + − f (x) < ( from (II) )
n 2

S25011: REAL ANALYSIS Page 190

Thus  
1 ϵ
n ≥ N =⇒ f x+ − f (x) < . . . (IV )
n 2
From (I) and (IV ),
     
1 1 ϵ 1 ϵ
n ≥ N =⇒ fn x + −f x+ < and f x + − f (x) <
n n 2 n 2
 
     
1 1 1 ϵ ϵ
=⇒ n ≥ N =⇒ fn x + −f x+ +f x+ − f (x) < + 
n n n 2 2
 
 
1
=⇒ n ≥ N =⇒ fn x + − f (x) < ϵ
n
 
1
=⇒ lim fn x + = f (x)
n→∞ n

4.2.22 Solution: a)If x = 0 or x = 1, it is clear that lim fn (x) = 0. For 0 < x < 1,
n→∞
consider,
lim fn (x) = lim nx(1 − x2 )n
n→∞ n→∞
 
nx ∞
= lim form
n→∞ (1 − x2 )−n ∞
x
= lim
n→∞ (1 − x2 )−n log(1 − x2 )
x
= lim (1 − x2 )n
log(1 − x2 ) n→∞
 
x 2 n 2
= · 0 ∵ lim (1 − x ) = 0 as 0 < (1 − x ) < 1
log(1 − x2 ) n→∞

=0
Thus for all x ∈ [0, 1], lim fn (x) = 0. This proves that {fn }∞
n=1 converges to 0 on [0, 1].
n→∞
b)
Z1 Z1
fn (x)dx = nx(1 − x2 )n dx
0 0
Z1
n
= (1 − x2 )n (−2x)dx
−2
0
" #x=1
n (1 − x2 )n+1
=
−2 n+1
x=0
n 1
= (0 − )
−2 n+1
n
=
2(n + 1)

S25011: REAL ANALYSIS Page 191

Therefore
Z1
n 1
lim fn (x)dx = lim =
n→∞ n→∞ 2(n + 1) 2
( )∞ 0
R1 1
This proves that fn converges to 2
on [0, 1].
0
n=1
R1 R1
Thus lim fn (x)dx = 0 and lim fn (x)dx = 21 .
0 n→∞ n→∞ 0
R1 R1
So lim fn (x)dx ̸= lim fn (x)dx. Hence {fn }∞
n=1 does not converge uniformly on [0, 1].
0 n→∞ n→∞ 0

4.2.23 Solution: a) Let ϵ > 0. By Archimedian property of real numbers, there is

N ∈ I such that N > 1ϵ . Then

1
x ∈ [0, 1], n ≥ N =⇒ n >
ϵ
 
1
=⇒ x ∈ [0, 1], n ≥ N =⇒ <ϵ
n
! !
sin nx sin nx 1
=⇒ x ∈ [0, 1], n ≥ N =⇒ <ϵ ∵ ≤
n n n
!
sin nx
=⇒ x ∈ [0, 1], n ≥ N =⇒ −0 <ϵ
n


=⇒ x ∈ [0, 1], n ≥ N =⇒ |fn (x) − 0| < ϵ
This proves that {fn }∞
n=1 converges uniformly to 0 on [0, 1].

b) lim fn′ (x) = lim cos nx

n
.n = lim cos nx which does not exist except for x = 0. Hence
n→∞ n→∞ n→∞
{fn′ }∞
n=1 does not converge on [0, 1].

Unit 04-03: Convergence and Uniform Convergence of Series of Functions

4.3.7 Option D 4.3.8: Option B

4.3.11 Solution: Define the sequence {sn }∞

n=1 of functions on R as
sn = u0 + u1 + u2 + · · · + un
x2
where uk (x) = (1+x2 )k
. Consider
sn (x) = u1 (x) + u2 (x) + · · · + un (x)
x2 x2 x2
= + + · · · +
(1 + x2 ) (1 + x2 )2 (1 + x2 )n
x2 1 − (1+x1 2 )n
= × 1
1 + x2 1 − 1+x 2

x2 1 − (1+x1 2 )n
= × x2
1 + x2 1+x2

S25011: REAL ANALYSIS Page 192

 1
=1−
(1 + x2 )n
Therefore    
1 1
lim sn (x) = lim 1 − = 1 − lim
n→∞ n→∞ (1 + x2 )n n→∞ (1 + x2 )n
So 
 = 1 − 0 = 1 if x ̸= 0
lim sn (x) =
n→∞  = 1 − 1 = 0 if x = 0
∞
x2
P
Thus the series un (x) = converges to f (x) on R where
(1+x2 )n
n=0 
 1 if x ̸= 0
f (x) =
 0 if x = 0
A series of continuous functions converges to a continuous function under uniform con-
x2
vergence. un (x) = is continuous on R for each n ∈ I. However f is not continuous
(1+x2 )n
∞
x2
P
at x = 0. It follows that the series un (x) = (1+x 2 )n does not converge uniformly on R.
n=0
4.3.12 Solution:
x2 ≥ 0

=⇒ n2 + x2 ≥ n2
1 1
=⇒ ≤ 2
+x 2n2 n
∞ ∞
X 1 X 1
=⇒ ≪
n=1
n2 + x2 n=1
n2
∞ ∞
1 1
on R+ follows.
P P
n2
< ∞. By Weierstrass M-Test the uniform convergence of n2 +x2
n=1 n=1

Unit 04-04: Power Series

4.4.11 Option B 4.4.12: Option B

4.4.15 Solution:
x ∈ [0, 1]

=⇒ 0≤x≤1

=⇒ xn ≤ 1 for all n ∈ I

=⇒ |an |xn ≤ |an | for all n ∈ I

∞
X ∞
X
n
=⇒ an x ≪ |an |
n=1 n=1
∞ ∞
an xn on [0, 1] follows.
P P
|an | < ∞. By Weierstrass M-Test the uniform convergence of
n=1 n=1

4.4.16 Solution: Consider

x2n+1
(2n+1)! x2n+1 (2n − 1)!
lim x2n−1
= lim ×
n→∞
(2n−1)!
n→∞ (2n + 1)! x2n−1

S25011: REAL ANALYSIS Page 193

 x2
= lim
n→∞ 2n(2n + 1)

=0
x3 x5
So by ratio test the series x − 3!
+ 5!
− · · · converges on R. Hence its term by term
differentiation is justified.
!
d d x 3 x5
(sin x) = x− + − ···
dx dx 3! 5!
3x2 5x4
=1− + − ···
3! 5!
x2 x4
=1− + − ···
2! 4!
= cos x

4.4.17 Solution: Consider

x2n+2
(n+1)! x2n+2 n!
lim x2n
= lim × 2n
n→∞ n→∞ (n + 1)! x
n!

x2
= lim
n→∞ n + 1

=0

<1
x2 x4 2n
So by ratio test the series 1 + 1!
+ 2!
+ · · · + xn! + · · · converges on R. Hence its term by
term differentiation is justified.

2x 4x3 6x5 2nx2n−1

f ′ (x) = 0 + + + + ··· + + ···
1! 2! 3! n!
2x 4x3 6x5 2nx2n−1
= + + + ··· + + ···
1! 2! 3! n!
!
x2 x4 x2(n−1)
= 2x 1 + + + ··· + + ···
1! 2! (n − 1)!

= 2xf (x)

S25011: REAL ANALYSIS Page 194

 APPENDIX 1

Foundations of Mathematics

Section 1.1 What is Mathematics

Mathematics is a study of sets with structures. There are three basic structures.
i) Algebraic Structure
ii) Topological Structure
iii) Ordered Structure
For example if we are studying arithmetic of real numbers then we are studying the al-
gebraic structure < R, +, · >. The study of Group theory is a study of the structure
< G, ∗ >, where ∗ is a binary operation on set G satisfying certain properties. The study
of metric spaces is the study of the structure < X, d > where the function d : X × X → R
satisfies certain properties.
Mathematics may broadly be categorised in the following branches.
1)Pure Mathematics: Algebra(Group Theory,Ring Theory,Field Theory,Vector Spaces,
Matrices and Determinants etc.), Analysis(Real Analysis, Complex Analysis), Topology,
Differential Equations, Number Theory,Geometry...
2) Discrete Mathematics: Combinatorics, Graph Theory, Lattice Theory,...
3) Applied Mathematics: Differential Equations, Numerical Analysis, Operations Re-
search, Cryptography, Computational Geometry, Statistics (as applied real analysis),...

The above classification is not the precise one. The mode of study and approach of certain
theory may categorise it in different branches.

Section 1.2 Logic as foundation of Mathematics

The Logic forms foundation for Mathematics. The Logic is the language of Mathematics.
The rules of logic distinguishes between valid and invalid statements in Mathematics.
Any proof in Mathematics can be translated into a sequence of logical statements and
the validity of proofs can be established by using the rules of inferences. If there seems
a problem with some proof (or argument), its validity may be checked by translating it
into a sequence of logical statements and then applying rules of inferences.

195
To ideal course on Foundations of Mathematics will have the following outline.

Logic −→ Set theory −→ Number System

Problem 1.2.1 Is the statement 2 ≤ 3 true? Justify your answer.

Solution: The statement 2 ≤ 3 is logically equivalent to the statement 2 < 3 or 2 = 3.
Here the first statement is true and the second is false. The disjunction p ∨ q is true when
p is true and q is false. Refer to the truth table of p ∨ q. It now follows that the statement
2 ≤ 3 is true.

This example illustrates the use of rules of logic in understanding the clear meaning of
statements in Mathematics. The or (disjunction ∨) in Mathematics is always inclusive
or. The statement p ∨ q is true when at least one of p and q is true. The statement p ∨ q
is true includes three possibilities.

• p is true and q is false

• p is false and q is true

• p is true and q is true

S25011: REAL ANALYSIS Page 196

It includes the third possibility. So is inclusive or. (There is an exclusive or in logic
which is true when exactly one of p and q is true.)

Problem 1.2.2 Prove that the empty set ϕ is subset of every set.
Solution: Let S be a set. Consider the statement
x ∈ ϕ =⇒ x ∈ S . . . (I)
It is a conditional statement p =⇒ q, where the statement p is false and the statement
q may be true or false. If p is false then the conditional statement p =⇒ q is true
(vacuously true). Refer to the truth table of p =⇒ q. It follows that the statement (I)
is true, which means that ϕ is a subset of S.

Problem 1.2.3 Prove that the empty set ϕ is a semi-group.

Solution: Semi-group: G is a set and ∗ : G × G. The structure < G, ∗ > is a semi-
group if
∗ is associative i.e. (a ∗ b) ∗ c =a ∗ (b ∗ c) for all a, b, c ∈ G
The above definition can be written as
Semi-Group: G is a set and ∗ : G × G. The structure < G, ∗ > is a semi-group if

• a, b, c ∈ G =⇒ (a ∗ b) ∗ c = a ∗ (b ∗ c)

The property a, b, c ∈ G =⇒ (a∗b)∗c =a∗(b∗c) in a definition is a conditional statement

p =⇒ q. For G = ϕ, the statement is
a, b, c ∈ ϕ =⇒ (a ∗ b) ∗ c = a ∗ (b ∗ c)
The hypothesis p : a, b, c ∈ ϕ is false. If p is false then the conditional statement p =⇒ q
is true (vacuously true). Refer to the truth table of p =⇒ q. It follows that the
statement
a, b, c ∈ ϕ =⇒ (a ∗ b) ∗ c = a ∗ (b ∗ c)
is true, which means that ϕ is a semi-group.

Problem 1.2.4 Prove that the empty set ϕ is not a group.

Solution: In algebra texts, the reader may find the following two (equivalent) definitions
of groups.
Group: G is a non-empty set and ∗ : G × G. The structure < G, ∗ > is a group if

• ∗ is associative i.e. (a ∗ b) ∗ c =a ∗ (b ∗ c) for all a, b, c ∈ G

• There is an identity element for ∗ in G i.e. there is e ∈ G such that a∗e =e∗a = a
for all a ∈ G
• Every element in G has inverse with respect to ∗ i.e. for every a ∈ G there is
b ∈ G such that a ∗ b =b ∗ a = e

OR

S25011: REAL ANALYSIS Page 197

Group: G is a set and ∗ : G × G. The structure < G, ∗ > is a group if

• ∗ is associative i.e. (a ∗ b) ∗ c =a ∗ (b ∗ c) for all a, b, c ∈ G

• There is an identity element for ∗ in G i.e. there is e ∈ G such that a∗e =e∗a = a
for all a ∈ G
• Every element in G has inverse with respect to ∗ i.e. for every a ∈ G there is
b ∈ G such that a ∗ b = b ∗ a = e

According to the first definition of group, the empty set is not a group. The second
definition can be written as
Group: G is a set and ∗ : G × G. The structure < G, ∗ > is a group if

• a, b, c ∈ G =⇒ (a ∗ b) ∗ c = a ∗ (b ∗ c)

• ∃e ∈ G∀a ∈ G(a ∗ e = e ∗ a = a)

• ∀a ∈ G∃b ∈ G(a ∗ b = b ∗ a = e)

The second property is not true for the empty set as it doesn’t contain any element.
Hence the empty set is not a group.

Section 1.3 Role of definitions

There are two important properties of definitions.

i) The definition is always fundamental in Mathematics.
Let’s see an example.
i2 = i.i
√ √
= −1 −1
p
= (−1)(−1)
√
= 1

=1
√ √ √
We proved that i2 = 1! Where is the mistake? The mistake is in the step a b = ab.
This rule is valid only when a and b are non-negative. But more important question is
What is the proper way to calculate the value of i2 . The answer is: use DEFINITION
of product of two complex numbers. The complex numbers are defined as a + ib where
√
a, b ∈ R and i denotes the symbol −1. Then addition, subtraction, multiplication
and division of complex numbers are defined. The multiplication of complex numbers is
defined as
(a + ib)(c + id) = (ac − bd) + i(ad + bc)

S25011: REAL ANALYSIS Page 198

So the proper way to calculate the value of i2 is
i2 = i.i

= (0 + 1i)(0 + 1i)

= (0 − 1) + (0 + 0)i

= −1

Definitions are always fundamental in Mathematics. Everything in Mathematics begins

with definitions. We can’t argue with definitions. We have to accept definitions. If you
think some argument is not correct,go back to the definitions and then start from it.
ii) Every definition in Mathematics is if and only if or biconditional ( p ⇐⇒
q) type of statement.
Examples:

Definition 1.3.1 A real number x is positive if and only if x > 0.

So if real number x is positive then x > 0. Conversely if x > 0 then x is positive real
number

Definition 1.3.2 △ABC is equilateral if and only if all sides of △ABC have equal
lengths.
So if △ABC is equilateral then all sides of △ABC have equal lengths. Conversely if all
sides of △ABC have equal lengths then △ABC is equilateral.
It is important to note that Every definition in Mathematics is of this type. The standard
convention is to write if instead of if and only if. So the above definitions could be stated
as

S25011: REAL ANALYSIS Page 199

Definition 1.3.3 A real number x is positive if x > 0.

Definition 1.3.4 △ABC is equilateral if all side of △ABC have equal lengths.
In above definitions, if means if and only if. The both way implications are used. For
example, if △ABC is equilateral then we shall use the fact that all side of △ABC have
equal lengths. On the other hand to prove that △ABC is equilateral, we shall prove that
all sides of △ABC have equal lengths.

Section 1.4 Axiomatic Theories in Modern Mathematics

The theories in Mathematics are of axiomatic types. There are defined/undefined con-
cepts and axioms for these concepts. Based on these axioms, theories are developed. For
example in Euclidean geometry, point is undefined concept and based on certain axioms;
the Euclidean geometry is developed. In set theory, set is undefined concept and based on
certain axioms; the set theory is developed. In many introductory texts on Real Analysis
or Calculus, it is assumed that there is a set R of real numbers (undefined concept) and
algebraic,ordered and completeness properties (axioms) are stated based on which the
properties of real numbers are proved.

S25011: REAL ANALYSIS Page 200

 Section 1.5 Sets

The set is the most fundamental concept of Mathematics. The meaning is that anything
and everything that we deal with in Mathematics is ultimately a set. For example, a
relation (binary relation) on set A is a subset (set) of A × A. A function f : A → B is
a subset (set) of A × B. A sequence is a function and so it is also a set. The cardinal
numbers are equivalence classes (of similarity relation) and so they are sets. The ordered
pair (x, y) is defined as (x, y) = {{x}, {x, y}} and so it is a set. The series of real
numbers (defined as ordered pair of two sequences) is also a set. You may be surprised
to know that numbers are also sets. For example,the natural numbers are defined as
0 = ϕ, 1 = {0}= {ϕ}, 2 = {1, 0}= {{ϕ}, ϕ} . . ..
The reader may find that the rigorous treatment of set theory is not so easy. It has taken
centuries to formalize the logic and set theory in rigorous way. For example, what is the
concept of the empty set? It is an axiom that the empty set exists. The empty set is a set
containing no elements. Intuitively set can be thought as a collection of objects. So how
a set can be empty? (The analogy may be helpful in this case. The sets can be thought
as boxes containing objects and then the empty set is a box containing no object).
About the uniqueness of empty set: A − A = ϕ and B − B = ϕ′ for sets A and B. So
whether both ϕ and ϕ′ are equal. It is by axiom that the empty set is unique.

Section 1.6 Cardinality of Sets

• Intuitively the cardinality of sets can be thought as a number of elements in

a set. Two sets have same cardinality if there is one-to-one correspondence
between them. If there is one to one correspondence between two sets then they
are similar. The similarity is an equivalence relation on the collection of sets.
The cardinality of a set may rigorously be defined as an equivalence class of this
relation.

• A set is infinite if it is similar to a proper subset of itself.

• A set is finite if it is empty or can be put in one-to-one correspondence with the

set {1, 2, 3, . . . , n} for some positive integer n.

• A set is denumerable if it can be put in one-to-one correspondence with the set

{1, 2, 3, . . . , } for some positive integer n. The cardinality of denumerable sets is
denoted by χ0 .

• Empty set is the only set with cardinality 0.

S25011: REAL ANALYSIS Page 201

 • Intuitively, a set is countable if its element can be counted as 1, 2, 3, . . . i.e. as
first element, second element, third element,....A set is countable if it is finite or
denumerable (Note: in some texts, denumerable sets are defined as countable).

• A set is uncountable if it is not countable.

The following are few examples of countable sets.

I: 1 2 3 ···
↓ ↓ ↓ ↓ ···
N: 1 2 3 ···

This proves that the set I of positive integers is denumerable and hence counatable.

I: 1 2 3 ···
↓ ↓ ↓ ↓ ···
2N : 2 4 6 ···

This proves that the set 2N of positive even integers is denumerable and hence countable.

I: 1 2 3 ···
↓ ↓ ↓ ↓ ···
Z: 0 −1 1 ···

This proves that the set Z of integers is denumerable and hence countable.
The rationals can also be counted. So the set Q of rationals is denumerable and hence
countable.

• A subset of countable set is countable.

• A countable union of countable sets is countable. So a finite union of finite sets is
finite(countable), a finite union of denumerable sets is denumerable(countable),
a denumerable union of finite sets is countable and a denumerable union of
denumerable sets is denumerable (countable).
• A finite cartesian product of finite sets is finite (countable).
• A finite cartesian product of denumerable sets is denumerable (countable).
• A denumerable cartesian product of denumerable sets is uncountable.
• Thus a countable cartesian product of countable sets may not be countable.
• There is no way of counting the elements of the set Qc of irrationals. Hence it is
uncountable. If a ̸= b then intervals (a, b).[a, b], [a, b), (a, b] are uncountable.
• The Cantors theorem says that the cardinality of a set A is always less than the
cardinality of its power set P(A). So there is no largest set. For if U is the
largest set then P(U )contains more elements than U .

S25011: REAL ANALYSIS Page 202

 • χ0 denotes the cardinality of denumerable set. It is the smallest possible cardinal
number that an infinite set can have.
• |N| = |Z|= |Q| = χ0
• |(a, b)| =|[a, b]| =|(a, b]| = |[a, b)| =|Q| =|R| =2χ0 = c
χ
• n < χ0 < 2χ0 < 22 0 < ....
There are infinitely many infinites in cardinal numbers. The smallest (unity) is
χ0 .

Section 1.7 Remarks on Definitions and Proofs

The definitions in Mathematics are of fundamental importance. It is advised to write

definitions in few different (and equivalent) ways. Try to write every definition as a
biconditional statement p ⇐⇒ q. The clear understanding of definitions will ease out
the learning curve of any theory. Once the definition is understood, you may try to write
it in your own way. This may be used as a test of understanding.

The proofs are to be avoided in paragraphs. If proofs are written as a sequence of logical
statements (step by step), they are better understood. This may not be possible for the
lengthy and complex proofs. However proofs may be divided in parts and sub-proofs may
be written as a sequence of logical statements. The proof is not completely understood
unless one can view it as a single idea.

Most of the statements in Mathematics may be written as a conditional statement

p =⇒ q. The direct proofs may be preferred whenever possible in proving condi-
tional statements. In direct proofs, one begin with the hypothesis p and after a sequence
of logical statements it ends with the conclusion q.

statement p
=⇒ statement 1
=⇒ statement 2
=⇒ · · · · · · · · · · · ·
=⇒ statement q

S25011: REAL ANALYSIS Page 203

 APPENDIX 2

Sequences of Real Numbers

Section 2.1 Sequence and subsequence

• A sequence of real numbers is a function from the set I of positive real numbers
to the set R of real numbers.
I: 1 2 3 ··· n ···
↓ ↓ ↓ ↓ ··· ↓ ···
R : s1 s2 s3 ··· sn ···

The above sequence is denoted as {sn }∞

n=1 . Here sn denote image of n.

• A subsequence {sni }∞ ∞
i=1 of {sn }n=1 is obtained by removing some of the terms of

the sequence {sn }∞ ∞

n=1 without disturbing the order of elements of {sn }n=1 .

◦ Every sequence is subsequence of itself.

{sn }∞
n=1 : s1 s2 s3 · · · sn · · ·
{sni }∞
i=1 : s1 s2 s3 ··· sn i ···

{sni }∞ ∞
i=1 is a subsequence of {sn }n=1 . Here ni = i.

◦
{sn }∞
n=1 : s1 s2 s3 ··· sn ···
{sni }∞
i=1 : s1 s3 s5 ··· sn i ···

{sni }∞ ∞
i=1 is a subsequence of {sn }n=1 . Here ni = 2i − 1.

Section 2.2 Convergence of sequence

• A sequence {sn }∞
n=1 is said to converge to L if for every ϵ > 0, there is N ∈ I

such that
n ≥ N =⇒ |sn − L| < ϵ
In this case we write lim sn = L.
n→∞

• lim sn = L ⇐⇒ ∀ϵ > 0∃N ∈ I(n ≥ N =⇒ |sn − L| < ϵ)

n→∞
lim sn = L ⇐⇒ ∀ϵ > 0∃N ∈ I(n ≥ N =⇒ Sn ∈ (L − ϵ, L + ϵ))
n→∞
lim sn = L ⇐⇒ ∀ϵ > 0∃N ∈ I(SN , SN +1 , SN +2 . . . ∈ (L − ϵ, L + ϵ))
n→∞
205
 lim sn = L ⇐⇒ ∀ϵ > 0∃N ∈ I (L − ϵ, L + ϵ)
n→∞ | {z }
↑
sN ,sN +1 ,sN +2 ,...
lim sn = L ⇐⇒ sn is close to L for sufficiently large n.
n→∞

• If {sn }∞ ∞
n=1 and {tn }n=1 are convergent sequences of real numbers then
lim (sn + tn ) = lim sn + lim tn
n→∞ n→∞ n→∞
lim (sn − tn ) = lim sn − lim tn
n→∞ n→∞ n→∞
lim (sn .tn ) = lim sn . lim tn implying that lim sm m
n = ( lim sn ) for any positive
n→∞ n→∞ n→∞ n→∞ n→∞
integer m. 
sn
 lim sn
n→∞
lim = provided lim tn ̸= 0
n→∞ tn lim tn n→∞
n→∞

Section 2.3 Monotone sequence

• A sequence {sn }∞
n=1 of real numbers is increasing if
s1 ≤ s2 ≤ s3 ≤ · · · ≤ sn ≤ sn+1 ≤ · · ·
{sn }∞
n=1 is increasing ⇐⇒ (n ∈ I =⇒ sn ≤ sn+1 )

An increasing sequence is also called non-decreasing.

The sequence {sn }∞ ∞
n=1 = {n}n=1 is increasing as 1 ≤ 2 ≤ · · · ≤ n ≤ n + 1 ≤ · · ·

• A sequence {sn }∞
n=1 of real numbers is decreasing if
s1 ≥ s2 ≥ s3 ≥ · · · ≥ sn ≥ sn+1 ≥ · · ·
{sn }∞
n=1 is decreasing ⇐⇒ (n ∈ I =⇒ sn ≥ sn+1 )

A decreasing sequence is also called non-increasing.

The sequence {sn }∞ 1 ∞
n=1 = { n }n=1 is decreasing as
1
1
≥ 1
2
≥ ··· ≥ 1
n
≥ 1
n+1
≥ ···

• {sn }∞ ∞ 1 ∞
n=1 is monotone if it is increasing or decreasing sequence. {n}n=1 and { n }n=1

are monotone sequences. The sequence {(−1)n }∞

n=1 is not monotone as it is

neither increasing nor decreasing sequence.

Section 2.4 Bounded sequence

• A sequence {sn }∞
n=1 of real numbers is bounded if there is M > 0 such that

|sn | ≤ M for all n ∈ I.

• {sn }∞
n=1 is bounded ⇐⇒ ∃M > 0(n ∈ I =⇒ |sn | ≤ M )

{sn }∞
n=1 is bounded ⇐⇒ ∃M > 0(n ∈ I =⇒ −M ≤ sn ≤ M )

{sn }∞
n=1 is bounded ⇐⇒ ∃M > 0(n ∈ I =⇒ sn ∈ [−M, M ])

{sn }∞
n=1 is bounded ⇐⇒ ∃M > 0([−M, M ])
| {z }
↑
s1 ,s2 ,s3 ,...
{sn }∞ ∞
n=1 is bounded ⇐⇒ all terms of {sn }n=1 can be enclosed in some closed

and bounded interval.

S25011: REAL ANALYSIS Page 206

 {sn }∞ ∞
n=1 is bounded ⇐⇒ all terms of {sn }n=1 can be enclosed in some open and

bounded interval.

• The terms of { n1 }∞ n ∞
n=1 can be enclosed in interval (0, 2). The terms of {(−1) }n=1

can be enclosed in (−2, 2).So the sequences { n1 }∞ n ∞

n=1 and {(−1) }n=1 are bounded

sequences.

• The terms of {n}∞ ∞

n=1 and {−n}n=1 can’t be enclosed in some bounded intervals.

So the sequences {n}∞ ∞

n=1 and {−n}n=1 are unbounded.

Section 2.5 Cauchy sequence

• A sequence {sn }∞
n=1 is Cauchy if for every ϵ > 0, there is N ∈ I such that
m, n ≥ N =⇒ |sm − sn | < ϵ

• {sn }∞
n=1 is Cauchy ⇐⇒ ∀ϵ > 0∃N ∈ I(m, n ≥ N =⇒ |sm − sn | < ϵ)

{sn }∞
n=1 is Cauchy ⇐⇒ ∀ϵ > 0∃N ∈ I(m > n ≥ N =⇒ |sm − sn | < ϵ)

{sn }∞
n=1 is Cauchy ⇐⇒ sm and sn are close to each other for sufficiently large

m and n.

• lim sn = L ⇐⇒ sn is close to L for sufficiently large n.

n→∞
The following argument gives a non-rigorous proof of the fact that every Cauchy
sequence is convergent.
{sn }∞
n=1 converges to L

=⇒ sm and sn are close to L for sufficiently large m, n

=⇒ sm and sn are close to each other for sufficiently large m, n
(Analogy: If two persons are close to third person then two persons are also close
to each other).
=⇒ {sn }∞
n=1 is Cauchy

Section 2.6 Limit superior and inferior

• Let {sn }∞
n=1 be an upper bounded sequence of real numbers. For each n in I,

define
Mn = sup{sn , sn+1 , sn+2 , sn+3 , . . .}
Since {sn+1 , sn+2 , sn+3 , . . .} ⊆ {sn , sn+1 , sn+2 , sn+3 , . . .},
Mn+1 ≤ Mn for all n ∈ I
Thus
M1 ≥ M2 ≥ M3 ≥ . . . ≥ Mn+1 ≥ Mn ≥ . . . =⇒ {Mn }∞
n=1 is decreasing

S25011: REAL ANALYSIS Page 207

 So if {Mn }∞ ∞
n=1 is lower bounded, then {Mn }n=1 converges and in this case we

define
lim sup sn = lim Mn
n→∞ n→∞
If {Mn }∞ ∞
n=1 is not lower bounded, then {Mn }n=1 diverges to −∞ and in this case

we define
lim sup sn = −∞
n→∞

• If {sn }∞
n=1 is not upper bounded then we define
lim sup sn = ∞
n→∞

• Let {sn }∞
n=1 be a lower bounded sequence of real numbers. For each n in I, define
mn = inf{sn , sn+1 , sn+2 , sn+3 , . . .}
Since {sn+1 , sn+2 , sn+3 , . . .} ⊆ {sn , sn+1 , sn+2 , sn+3 , . . .}
mn+1 ≥ mn for all n ∈ I
Thus
m1 ≤ m2 ≤ m3 ≤ . . . ≤ mn+1 ≤ mn ≤ . . . =⇒ {mn }∞
n=1 is increasing

So if {mn }∞ ∞
n=1 is upper bounded, then {mn }n=1 converges and in this case we

define
lim inf sn = lim mn
n→∞ n→∞
If {mn }∞ ∞
n=1 is not upper bounded, then {mn }n=1 diverges to ∞ and in this case

we define
lim inf sn = ∞
n→∞

• If {sn }∞
n=1 is not lower bounded then we define
lim sup sn = −∞
n→∞

{sn }∞
n=1 is upper bounded {Mn }∞
n=1 is lower bounded =⇒ lim sup sn = lim Mn
n→∞ n→∞

{sn }∞
n=1 is upper bounded {Mn }∞
n=1 is not lower bounded =⇒ lim sup sn = −∞
n→∞

{sn }∞
n=1 is not upper bounded =⇒ lim sup sn = ∞
n→∞

{sn }∞
n=1 is lower bounded {mn }∞
n=1 is upper bounded =⇒ lim inf sn = lim mn
n→∞ n→∞

{sn }∞
n=1 is lower bounded {mn }∞
n=1 is not upper bounded =⇒ lim sup sn = ∞
n→∞

{sn }∞
n=1 is not lower bounded =⇒ lim sup sn = −∞
n→∞

S25011: REAL ANALYSIS Page 208

 • The number lim sup sn gives the idea about how large the terms of {sn }∞
n=1 can
n→∞
be when n becomes large and large.

• The number lim inf sn gives the idea about how small the terms of {sn }∞
n=1 can
n→∞
be when n becomes large and large.

• It follows that
{sn }∞
n=1 converges ⇐⇒ lim sup sn = lim inf sn
n→∞ n→∞

Section 2.7 Results

• Archimedian property of real numbers: For every real number x, there is a

positive integer n such that n > x. For example if x = 0 then n can be taken as
1 or any positive integer. If x = 10.9 then n can be taken as 11 or any positive
integer greater than 10. This simple property is useful in proving many of the
1
results such as lim = 0 and some inclusions.
n→∞ n

• An increasing and upper-bounded sequence of real numbers converges to its

supremum.
s1 ≤ s2 ≤ . . . and {sn }∞
n=1 is upper bounded =⇒ lim sn = sup{sn , sn+1 . . .}
n→∞

• An increasing and non upper-bounded sequence of real numbers diverges to in-

finity.
s1 ≤ s2 ≤ . . . and {sn }∞
n=1 is not upper bounded =⇒ lim sn = ∞
n→∞

• A decreasing and lower-bounded sequence of real numbers diverges to its infi-

mum.
s1 ≥ s2 ≥ . . . and {sn }∞
n=1 is lower bounded =⇒ lim sn = inf{sn , sn+1 . . .}
n→∞

• A decreasing and non lower-bounded sequence of real numbers diverges to minus

infinity.
s1 ≥ s2 ≥ . . . and {sn }∞
n=1 is not lower bounded =⇒ lim sn = −∞
n→∞

• Every convergent sequence of real numbers is bounded.

{sn }∞ ∞
n=1 is Convergent =⇒ {sn }n=1 is bounded.

• Every Cauchy sequence of real numbers is bounded.

{sn }∞ ∞
n=1 is Cauchy =⇒ {sn }n=1 is bounded.

• A sequence of real numbers is Convergent if and only if it is Cauchy.

{sn }∞ ∞
n=1 is Convergent ⇐⇒ {sn }n=1 is Cauchy.

• {sn }∞
n=1 is convergent ⇐⇒ lim sup sn = lim inf sn
n→∞ n→∞

S25011: REAL ANALYSIS Page 209

 APPENDIX 3

Series of Real Numbers

Section 3.1 Series of real numbers

∞
P
• An infinite series an of real numbers can be thought as an infinite sum of the
n=1
∞
terms of the sequence {an }∞
P
n=1 of real numbers. The series an is also denotes
n=1
as a1 + a2 + · · · + an + · · ·
P∞
To “define” the series an , two sequences need to be specified. First the
n=1
sequence {an }∞ ∞
n=1 of real numbers and the sequence {sn }n=1 = {a1 + a2 + · · · +

an } ∞
n=1 of partial sums.
∞
P
• All the properties of the series an are defined using the properties of the
n=1
sequence {sn }∞
n=1 .

Section 3.2 Convergence and divergence

∞
be a series of real numbers and {sn }∞ ∞
P
• Let n=1 = {a1 + a2 + · · · + an }n=1 be its
n=1
sequence of partial sums.
∞
an converges if the sequence {sn }∞
P
◦ the series n=1 converges
n=1
∞
an diverges if the sequence {sn }∞
P
◦ the series n=1 diverges
n=1
∞
P
• the series an converges ⇐⇒ lim sn exists.
n=1 n→∞
Infinite sum exists if limit of finite sum exists.
∞
P
• the series an diverges ⇐⇒ lim sn does not exist.
n=1 n→∞

∞
P
• If the series an is convergent then its value is defined as
n=1
∞
X
an = lim sn = lim (a1 + a2 + · · · + an )
n→∞ n→∞
n=1
Infinite sum = limit of finite sum

Section 3.3 Series of non-negative real numbers

∞
P
• an is a series of non-negative real numbers if an ≥ 0 for all n ∈ I.
n=1
211
 ∞
P ∞
P
• The notation an < ∞ denotes that the series an of non-negative real
n=1 n=1
numbers is convergent.
∞
P
• The series an of non-negative real numbers is convergent ⇐⇒ the sequence
n=1
of its partial sums {sn }∞ ∞
n=1 = {a1 + a2 + a3 + · · · + an }n=1 is bounded.
∞
an < ∞ ⇐⇒ the sequence {sn }∞ ∞
P
n=1 = {a1 +a2 +a3 +· · ·+an }n=1 is bounded.
n=1
∞
P ∞
P
• The notation an = ∞ denotes that the series an of non-negative real
n=1 n=1
numbers is divergent.
∞
P
• The series an of non-negative real numbers is divergent ⇐⇒ the sequence
n=1
of its partial sums {sn }∞ ∞
n=1 = {a1 + a2 + a3 + · · · + an }n=1 is unbounded.
∞
an = ∞ ⇐⇒ the sequence {sn }∞ ∞
P
n=1 = {a1 + a2 + a3 + · · · + an }n=1 is
n=1
unbounded.
∞
P n
P ∞
P
• If the series an of non-negative real numbers is convergent then ak ≤ an
n=1 k=1 n=1
for all n ∈ I.

Section 3.4 Absolute and conditional convergence

∞
P ∞
P
• an converges absolutely if |an | converges.
n=1 n=1
∞
X ∞
X
an converges absolutely ⇐⇒ |an | converges.
n=1 n=1

• Every absolutely∞ convergent series of real∞ numbers is convergent.

X X
|an | converges =⇒ an converges
n=1 n=1
∞
X ∞
X
|an | < ∞ =⇒ an converges
n=1 n=1

• By virtue of the above result we have

∞
P ∞
P ∞
P
an converges absolutely ⇐⇒ an converges and |an | converges.
n=1 n=1 n=1
P∞ P∞ P∞
an converges absolutely ⇐⇒ an converges and |an | < ∞
n=1 n=1 n=1
∞
P ∞
P ∞
P
• an converges conditionally if an converges and |an | diverges.
n=1 n=1 n=1
P∞ ∞
P ∞
P
an converges conditionally ⇐⇒ an converges and |an | diverges.
n=1 n=1 n=1
P∞ P∞ P∞
an converges conditionally ⇐⇒ an converges and |an | = ∞
n=1 n=1 n=1

• It is clear that the absolute convergence and convergence happen to be same

P∞
for the series an of non-negative real numbers. In other words, the series
n=1

S25011: REAL ANALYSIS Page 212

 of non-negative real numbers can’t converge conditionally. It either converges
absolutely or diverges.

Section 3.5 Rearrangement of Series:

∞
P ∞
P
• A rearrangement ani of the series an is obtained by rearranging the terms
i=1 n=1
∞
P
of an . Precisely, there is a bijective function (one-to-one correspondence)
n=1
f : I → I and f (i) = ni so that ani = af (i)
∞
P ∞
P
• Any rearrangement ani of absolutely convergent series an is convergent
i=1 n=1
∞
P ∞
P
and ani = an
i=1 n=1
The sum of absolutely convergent series remains unchanged on rearrangement.
∞
P
• Any rearrangement ani of convergent series of non-negative real numbers
i=1
∞
P ∞
P ∞
P
an is convergent and an i = an
n=1 i=1 n=1
The sum of convergent series of non-negative real numbers remains unchanged
on rearrangement.

• The conditionally convergent series of real numbers can be made to converge to

P∞
any real number by rearranging its terms. Precisely, if an is conditionally
n=1
∞
P
convergent and x is any real number then there is a rearrangement ani of
i=1
∞
P ∞
P
an such that ani = x.
n=1 i=1

Section 3.6 Basic tests for convergence and divergence

∞
P ∞
P
• The series bn is said to dominate the series an of real numbers if there is
n=1 n=1
N ∈ I such that

n ≥ N =⇒ |an | ≤ |bn |

In this case we write ∞ ∞

X X
an ≪ bn
n=1 n=1
∞
P ∞
P ∞
P
If bn dominates the series an , we also say that an is dominated by
n=1 n=1 n=1
∞
P
bn .
n=1
P∞ ∞
P
an is dominated by bn if and only if there is N ∈ I such that
n=1 n=1
|aN | ≤ |bN |, |aN +1 | ≤ |bN +1 , |aN +2 | ≤ |bN +2 |, . . .

S25011: REAL ANALYSIS Page 213

 ∞
X ∞
X
an ≪ bn ⇐⇒ ∃N ∈ I(|aN | ≤ |bN |, |aN +1 | ≤ |bN +1 , |aN +2 | ≤ |bN +2 |, . . .)
n=1 n=1
∞
P
• Necessary Condition for convergence of series If the series an converges then
n=1
lim an = 0.
n→∞

• Standard Series:
∞
xn converges ⇐⇒ −1 < x < 1.
P
◦ The geometric series
n=1
∞
1
P
◦ The p-series np
converges ⇐⇒ p > 1.
n=1
∞
P (−1)n
◦ The series np
converges conditionally if 0 < p < 1.
n=1
∞
P ∞
P ∞
P
• If an and bn are convergent and α, β are real numbers then (αan + βbn )
n=1 n=1 n=1
∞
P ∞
P ∞
P
is convergent and (αan + βbn ) = α an + β bn .
n=1 n=1 n=1
∞
P ∞
P ∞
P
• Comparison Test for Convergence: If an is dominated by bn where bn
n=1 n=1 n=1
∞
P
converges absolutely, then an also converges absolutely.
n=1
∞
X ∞
X ∞
X ∞
X
an ≪ bn and |bn | < ∞ =⇒ |an | < ∞
n=1 n=1 n=1 n=1
Dominating series converges =⇒ Dominated series converges.
Big series converges =⇒ Small series converges.
∞
P ∞
P ∞
P
• Comparison Test for Divergence: If an is dominated by bn and |an | =
n=1 n=1 n=1
∞
P
∞ then |bn | = ∞.
n=1
X∞ ∞
X ∞
X ∞
X
an ≪ bn and |an | = ∞ =⇒ |bn | = ∞
n=1 n=1 n=1 n=1
Dominated series diverges =⇒ Dominating series diverges
Small series diverges =⇒ Big series diverges
∞
P
• Comparison Test in limit form for convergence: If bn converges absolutely
n=1
∞
|an | P
and if lim exists then an converges absolutely.
n→∞ |bn | n=1

∞
P |an |
• Comparison Test in limit form for divergence: If |an | = ∞ and if lim
n=1 n→∞ |bn |
∞
P
exists then |bn | = ∞
n=1

S25011: REAL ANALYSIS Page 214

 ∞
P
• Ratio Test for Convergence and Divergence: Let an be a series of non-zero
n=1
|an+1 |
real numbers. a = lim inf |an |
, A= lim sup |a|an+1
n|
|
n→∞ n→∞
Then
∞
an+1 P
◦ if lim < 1 then |an | < ∞.
n→∞ an n=1
∞
an+1 P
◦ if lim > 1 then an diverges.
n→∞ an n=1

an+1
◦ if lim an
= 1 then the test fails.
n→∞
√
• Root Test for Convergence and Divergence If lim sup n a
n = A then the series
n→∞
∞
P
of real numbers an
n=1

◦ converges absolutely if A < 1

◦ diverges if A > 1

◦ the test fails when A = 1

• Application of Root Test to Power Series: Let {an }∞

n=1 be a sequence of real

numbers. Then
√ ∞
an xn converges absolutely for all x
P
◦ If lim n an = 0, the series
n→∞ n=1

√ ∞
1
an xn converges absolutely for |x| <
P
◦ If lim n an = L > 0, the series L
n→∞ n=1
1
and diverges for |x| > L

• Cauchy Condensation Test : Let {an }∞n=1 be a non-increasing sequence of posi-

∞ ∞
2n a2n converges if and only if
P P
tive numbers. The series an converges.
n=1 n=1

Section 3.7 Further tests for convergence and divergence

∞
P
• (an+1 − an ) converges ⇐⇒ lim an exists.
n=1 n→∞
In this case,
∞
X
(an+1 − an ) = lim an − a1
n→∞
n=1
∞
P
• an is a convergent series of decreasing positive terms =⇒ lim nan = 0
n=1 n→∞

• {an }∞
n=1 is a decreasing sequence of positive numbers and lim an = 0 =⇒ the
n→∞
∞
(−1)n+1 an is convergent.
P
alternating series
n=1

• Rabbes Test:

S25011: REAL ANALYSIS Page 215

  an  ∞
P
◦ lim n( an+1 − 1) > 1 =⇒ an converges.
n→∞ n=1

 an  ∞
P
◦ lim n( an+1 − 1) < 1 =⇒ an diverges.
n→∞ n=1
∞
an a bn
P
• Gauss test: Let an be a series of positive terms and an+1
= 1+ n
+ np
. If
n=1
{bn }∞
n=1 is a bounded sequence of real numbers then
∞
P
◦ a > 1 =⇒ an converges
n=1
∞
P
◦ a ≤ 1 =⇒ an diverges
n=1
∞
P
• Logarithmic test: Let an be a series of positive terms.
n=1
∞
an
P
◦ lim n log an+1 > 1 =⇒ an < ∞
n→∞ n=1
∞
an
P
◦ lim n log an+1 < 1 =⇒ an = ∞
n→∞ n=1

• DeMorgan’s and Bertrand’s Test

∞
an
  P
◦ lim [n( an+1 − 1) − 1] log n > 1 =⇒ an converges.
n→∞ n=1
∞
an
  P
◦ lim [n( an+1 − 1) − 1] log n < 1 =⇒ an diverges.
n→∞ n=1
∞
n
Cr xr converges absolutely
P
• ◦ |x| < 1 =⇒ the binomial series
r=1
∞
n
Cr xr converges conditionally
P
◦ x = 1 the binomial series
r=1

an nk +Ank−1 +Bnk−2 +Cnk−3 +···

• If an+1
= nk +ank−1 +bnk−2 +cnk−3 +···
, where k is positive integer then
∞
P
◦ A − a − 1 > 0 =⇒ an converges
n=1
∞
P
◦ A − a − 1 < 0 =⇒ an diverges
n=1

• Cauchy’s Integral Test: If f is a non-negative decreasing function on [1, ∞),

X∞ Z∞
f (n) converges ⇐⇒ f (x)dx converges
n=1 1
∞
P ∞
P
• Abel’s Test: an converges and bn is monotone and convergent =⇒
n=1 n=1
∞
P
an bn converges.
n=1

S25011: REAL ANALYSIS Page 216

 • The product of two absolutely convergent series is absolutely convergent. Pre-
∞
P ∞
P ∞ P
P n
cisely, if an and bn are absolutely convergent then the series ( ak bn−k )
n=0 n=0 n=0 k=0
is also absolutely convergent.
∞
P ∞
P
• The product of two convergent series may diverge. Precisely, if an and bn
n=0 n=0
∞ P
P n
are convergent then the series ( ak bn−k ) may diverge.
n=0 k=0

S25011: REAL ANALYSIS Page 217

 APPENDIX 4

Geometric series and p series

Section 4.1 Geometric series

∞
xn converges if and only if −1 < x < 1.
P
Theorem 4.1.1 The geometric series
n=0
∞ ∞
xn and let sn = a0 + a1 + a2 + · · · + an
P P
Proof : Take an =
n=1 n=1
I Case: x < −1
In this case, {an }∞ ∞
n=1 is not-bounded. So the sequence {an }n=1 is divergent. Therefore
P∞
an is divergent.
n=1
II Case: x = −1
In this case {s2n }∞ ∞
n=1 converges to 0 and {s2n+1 }n=1 converges to 1. So the sequence
∞
{sn }∞
P
n=1 is divergent. This implies the divergence of the series an .
n=1
III Case: −1 < x < 1
In this case
sn

= 1 + x + x2 + · · · + xn
1 − xn
=
1−x
So
1 − xn
 
1−0 1 n
lim sn = lim = = ∵ lim x = 0 as − 1 < x < 1
n→∞ n→∞ 1 − x 1−x 1−x n→∞
∞
x
P
Thus the series an is convergent and converges to 1−x
n=1
IV Case: x ≥ 1
In this case, {an }∞
n=1 is non-decreasing and not-bounded. Hence it diverges to ∞ and so
∞
P
lim an ̸= 0. Therefore an is divergent.
n→∞ n=1

∞
xn converges ⇐⇒ −1 < x < 1
P
The geometric series
n=1

Section 4.2 p series

∞
1
P
Theorem 4.2.1 The p-Series np
converges if and only if p > 1
n=1
219
Proof : I Case: p > 1
Consider
1 1 1 1 j−1 1
+ p + p + ··· + p ≤ 2
(2j−1 )p (2j−1+ 1) j−1
(2 + 2) j
(2 − 1) (2 )p
j−1
| {z }
2j−1 terms
Note that each term in the above sum is less than or equal to the first term. With this
observation
∞
1
P
np
n=1
= 11p + 21p + 31p + · · · + n1p + · · ·
= 1 + ( 21p + 31p ) + ( 41p + 51p + 61p + 71p ) + · · · + ( (2j−1
1
)p
+ (2j−11+1)p + (2j−11+2)p + · · · + (2j −1)
1
p )+· · ·

1 1 1
≤ 1 + 2. 2p + 22 . (22 )p + · · · + 2j−1 (2j −1)p + · · ·
Thus ∞
X 1 1 1 1
≪ 1 + 2. p + 22 . 2 p + · · · + 2j−1 j + ···
n=1
n p 2 (2 ) (2 − 1)p
1
The series on right is a geometric series with common ratio 2p−1 < 1 (∵ p > 1). By
∞
1
P
comparison test, it follows that np
is convergent.
n=1
II Case: p = 1
∞ ∞
1
P P
Let an = n
and take sn = a1 + a2 + a3 + · · · + an . Consider
n=1 n=1
s1 = 1
1 3
s2 = 1 +
=
2 2
1 1 3 1 1
s4 = s2 + + > + + = 2
3 4 2 4 4
1 1 1 1 1 1 1 1 5
s8 = s4 + + + + > 2 + + + + =
5 6 7 8 8 8 8 8 2
∞
Continuing in this way, s2n > n+2
2
. So the sequence {s n } n=1 contains a divergent subse-
∞
1
P
quence and hence it is divergent. This implies the divergence of the series n
.
n=1
III Case: p < 1
In this case,
1 1
< p
n n
So ∞ ∞
X 1 X 1
≪
n=1
n n=1
np
∞ ∞
1 1
P P
From II Case, n
is divergent. So by comparison test, the series np
is divergent.
n=1 n=1

∞
1
P
The p− series np
converges ⇐⇒ p > 1.
n=1

S25011: REAL ANALYSIS Page 220

 Feedback Form
School of Sciences
(Formerly School of Architecture, Science & Technology)
Yashwantrao Chavan Maharashtra Open University,
Nashik – 422 222

FEEDBACK SHEET FOR THE STUDENT

Dear Student,
You have gone through this book, it is time for you to do some thinking for us.
Please answer the following questions sincerely. Your response will help us to
analyse our performance and make the future editions of this book more useful.
Your response will be completely confidential and will in no way affect your
examination results. Your suggestions will receive prompt attention from us.

Please submit your feedback online at this QR Code or at following link

https://forms.gle/rpDib9sy5b8JEisQ9

or email at: [email protected]

or send this filled “Feedback Sheet” by post to above address.

(Please tick the appropriate box)

Pl. write your Program Code Course Code & Name

Style
01. Do you feel that this book enables you to learn the subject independently without
any help from others?

Yes No Not Sure

02. Do you feel the following sections in this book serve their purpose? Write the
appropriate code in the boxes.
Code 1 for “Serve the purpose fully”
Code 2 for “Serve the purpose partially”
Code 3 for “Do not serve any purpose”
Code 4 for “Purpose is not clear”
Warming up Check Point Answer to Check Points

To Begin With Summary References

 Objectives Key Words

03. Do you feel the following sections or features, if included, will enhance self - learning
and reduce help from others?
Yes No Not Sure

Index
Glossary
List of “Important Terms Introduced”
Two Colour Printing

Content

04. How will you rate your understanding of the contents of this Book?

Very Bad Bad Average Good Excellent

05. How will you rate the language used in this Book?
Very Simple Simple Average Complicated Extremely Complicated

06. Whether the Syllabus and content of book complement to each other?
Yes No Not Sure

07. Which Topics you find most easy to understand in this book?
Sr.No. Topic Name Page No.
 08. Which Topics you find most difficult to understand in this book?
Sr.No. Topic Name Page No.

09. List the difficult topics you encountered in this Book. Also try to suggest how
they can be improved.
Use the following codes:
Code 1 for “Simplify Text”
Code 2 for “Add Illustrative Figures”
Code 3 for “Provide Audio-Vision (Audio Cassettes with companion Book)”
Code 4 for “Special emphasis on this topic in counseling”

Sr.No. Topic Name Page No. Required Action Code

10. List the errors which you might have encountered in this book.

Sr.No. Page Line Errors Possible Corrections

No. No.
11. Based on your experience, how would you place the components of distance
learning for their effectiveness?
Use the following code.
Code 1 for “Most Effective” Code 3 for “Average” Code 5 fo r “Least Effective”

Code 2 for “Effective” Code 4 for “less Effective”

Printed Book Counseling Lab Journal

Audio Lectures Home Assignment YouTube Videos

Video Lectures Lab-Experiment Online Counseling

12. Give your overall rating to this book?

1. 2. 3. 4. 5.

13. Any additional suggestions:

Thank you for your co-operation!