Elementary Analysis through Examples and Exercises

Kluwer Texts in the Mathematical Sciences

VOLUME 10

A Graduate-Level Book Series

The titles published in this series are listed at the end of this volume.
Elementary Analysis
through Examples
and Exercises

by

John Schmeelk
Department of Mathematical Sciences,
Virginia Commonwealth University,
Richmond, Virginia, U.S.A.

Djurdjica Takaci
and
Arpad Takaci
Institute of Mathematics,
University ofNo vi Sad,
Novi Sad, Yugoslavia

Springer- Science+Business Media, B. V.

A C.I.P. Catalogue record for this book is available from the Library of Congress

ISBN 978-90-481-4590-4 ISBN 978-94-015-8589-7 (eBook)

DOl 10.1007/978-94-015-8589-7

Printed on acid-free paper

All Rights Reserved

© 1995 Springer Science+Business Media Dordrecht
Originally published by Kluwer Academic Publishers in 1995.
Softcover reprint of the hardcover 1st edition 1995

No part of the material protected by this copyright notice may be reproduced or

utilized in any fonn or by any means, electronic or mechanical,
including photocopying, recording or by any infonnation storage and
retrieval system, without written pennission from the copyright owner.
Table of contents

Preface VII

1 Real numbers 1
1.1 (R, +, .,:::;) as a complete totally ordered field 1
1.1.1 Basic notions . . . . . . 1
1.1.2 Examples and exercises. 5
1.2 Cuts in Q . . . . . . . . . . . . 29
1.2.1 Basic notions . . . . . . 29
1.2.2 Examples and exercises. 31
1.3 The set R as a topological space . 37
1.3.1 Basic notions . . . . . . 37
1.3.2 Examples and exercises .. 39

2 Functions 45
2.1 Real functions of one real variable 45
2.1.1 Basic notions . . . . . . . 45
2.1.2 Examples and exercises .. . 49
2.2 Polynomials, rational and irrational functions 72
2.2.1 Basic notions . . . . . . 72
2.2.2 Examples and exercises. 75

3 Sequences 83
3.1 Introduction 83
3.1.1 Basic notions . . . . . . 83
3.1.2 Examples and exercises. 85
3.2 Monotone sequences . . . . . . 102
3.2.1 Basic notions . . . . . . · 102
3.2.2 Examples and exercises. · 102
3.3 Accumulation points and subsequences · 116
3.3.1 Basic notions . . . . . . 116
3.3.2 Examples and exercises. · 117
3.4 Asymptotic relations . . . . . . · 135
3.4.1 Basic notions . . . . . . · 135
3.4.2 Examples and exercises. · 136

v
VI

4 Limits of functions 141

4.1 Limits . . . . . . . . . . . . . . · 141
4.1.1 Basic notions . . . . . . · 141
4.1.2 Examples and exercises. · 143
4.2 Asymptotes for the graphs of functions · 170
4.2.1 Basic notions . . . . . . · 170
4.2.2 Examples and exercises. · 171
4.3 Asymptotic relations . . . . . . .174
4.3.1 Basic notions . . . . .. · 174
4.3.2 Examples and exercises. · 174

5 Continuity 181
5.1 Continuity at a point . . . . .. · 181
5.1.1 Basic notions ..... . · 181
5.1.2 Examples and Exercises · 183
5.2 Uniform continuity . . . . . . . .204
5.2.1 Basic notions ..... . .204
5.2.2 Examples and exercises. .205

6 Derivatives 213
6.1 Introduction . . . . . . . . . . . · 213
6.1.1 Basic notions . . . . . . · 213
6.1.2 Examples and exercises. · 217
6.2 Mean value theorems . . . . . . .244
6.2.1 Basic notions . . . . . . .244
6.2.2 Examples and exercises. .245
6.3 Taylor's formula. . . . . . . . . .252
6.3.1 Basic notions . . . . . . .252
6.3.2 Examples and exercises. · 253
6.4 L'Hospital's Rule . . . . . . . . .262
6.4.1 Basic notions . . . . .. .262
6.4.2 Examples and Exercises .263
6.5 Local extrema and monotonicity of functions .270
6.5.1 Basic notions . . . . . . .270
6.5.2 Examples and exercises. · 271
6.6 Concavity . . . . . . . . . . . . .279
6.6.1 Basic notions . . . . . . .279
6.6.2 Examples and exercises. .280

7 Graphs of functions 285

Bibliography 313

Index 315
 Preface

It is hard to imagine that another elementary analysis book would contain ma-
terial that in some vision could qualify as being new and needed for a discipline
already abundantly endowed with literature. However, to understand analysis, be-
ginning with the undergraduate calculus student through the sophisticated math-
ematically maturing graduate student, the need for examples and exercises seems
to be a constant ingredient to foster deeper mathematical understanding. To a
talented mathematical student, many elementary concepts seem clear on their first
encounter.
However, it is the belief of the authors, this understanding can be deepened with a
guided set of exercises leading from the so called "elementary" to the somewhat more
"advanced" form. Insight is instilled into the material which can be drawn upon and
implemented in later development. The first year graduate student attempting to
enter into a research environment begins to search for some original unsolved area
within the mathematical literature.
It is hard for the student to imagine that in many circumstances the advanced
mathematical formulations of sophisticated problems require attacks that draw upon,
what might be termed elementary techniques. However, if a student has been guided
through a serious repertoire of examples and exercises, he/she should certainly see
connections whenever they are encountered.
The seven chapters in this book contain, in the authors' opinion, a wide variety of
problems, exercises and examples implemented by many instructors. The book can
be used both by complete beginners in analysis, as well as by students that already
have gone successfully through some calculus courses. The presented material is
self-contained and the exposition is mostly deductive. Occasionally some notions
are used before their formal definition, though, presumably, they will usually be
known to the reader.
The content is in fact elementary but the strategy employed is to navigate
through a wide assortment of problems connected to the nature of the chapter. A
minimal amount of expository discussion is included at the start of each new section
with a maximum emphasis placed on well selected examples and exercises capturing
the essence of the material. It is the intention of the authors to have students as-
semble a very valuable collection of well-thought-out problems. We have separated
the problems into examples and exercises. The examples contain a complete solu-
tion. In the exercises students are left to solve the problem. Often, answers only are
provided. One should note that the included exercises in some cases are generally

VII
viii PREFACE

found in very advanced texts when they are needed to prove a more profound result.
We feel that the students should study some of the pertinent advanced exercises in
the early developmental stages, so that they can enhance their mathematical skills.
Chapter 1 introduces the field of real numbers from the traditional axiomatic
presentation, as well as the R. Dedekind construction presented in 1872. Comparing
the two methods is, in our opinion, vital to understanding new concepts. Contrast
oftentimes invites the student to see why and where things differ and also where
things go wrong. The induction principle is also included with a very complete set
of examples and exercises implementing the process varying from the traditional
problems to what we term the somewhat unusual problems.
Chapter 2 introduces the concepts of relations and functions. It includes a com-
prehensive review of the basic ideas and terminology implemented to express the
different concepts. Various examples presented are oftentimes not included in a
very elementary text, like, for instance, the Dirichlet function. Also included is the
connection between rational functions and their representation using partial fraction
decomposition.
Chapter 3 introduces the concept of a sequence of real numbers. The traditional
theory surrounding sequences is presented such as limits, monotonicity and Cauchy
sequences. However, it goes on to identify asymptotic behavior for sequences as
well as proving some important estimates on the number e, which is the basis of
what is probably the most important function in analysis, namely the exponential
function. The discrete case is often drawn into the continuous case by examining
such functions as the logarithmic, f(x) = In x, x > 0, and showing its connection to
the discrete version, fn = In n, n E N. Furthermore, iterative schemes are examined
and how they can be implemented to generate converging sequences. Again the
Landau notation commonly called "big oh" and "small oh" are included, showing
how their asymptotic behavior definition can be reflected into sequences.
Chapter 4 introduces classical notion of limit to include the left and right limits.
Special emphasis is placed on their connections to points of accumulation. Several
examples illustrate the change of variable technique and how this process fosters
good arithmetic behavior, so that a precise limit can be computed. The line asymp-
totes for the graphs of the functions are given as the geometrical application of
limits. The Landau notation is again revisited and several examples and exercises
show the students how these relationships control asymptotic behaviors.
Chapter 5 presents the classical notion of continuity and its relationship to points
and points of accumulation. Left-side and right-side continuity, uniform continuity
and theorems relating these properties are presented in a traditional manner. The
delta-epsilon techniques together with their quantifiers are discussed in detail to-
gether with accompanying examples and exercises. Several somewhat unusual exam-
ples are presented to illustrate connections to the notion of the order of discontinuity
at a point.
Chapter 6 introduces the derivative using a detailed discussion involving the left
and right-hand derivatives. The traditional results regarding derivatives and their
geometrical interpretations are included together with a rich assortment of examples
and exercises. It continues to develop differentials and their connections to deriva-
PREFACE ix

tives and the derivative remainders. Again examples illustrating these notions and
calculated for some rather sophisticated functions are given. This chapter devel-
ops many applications for derivatives. The traditional results are included with
several applications not necessarily found in many elementary texts. It identifies
and proves the Laguerre, Hermite and Chebychev polynomials normally termed the
special functions in mathematical physics have real roots. Several inequalities are
proven using the mean value theorem for the differential calculus. The Taylor and
Maclaurin formulas are included with a generous amount of examples and exercises.
Of course, all of the standard information regarding preliminary curve sketching
implementing first and second order derivative tests is included, together with a
presentation of L'Hospital's rule.
Chapter 7 contains a detailed account for graphing functions. For this chapter
the asymptotes to include slanted asymptotes, concavity, (local) maximums and
minimums and points of inflection are presented again with some rather interesting
functions. Let us note here each example is endowed with the appropriate graph of
the given function. The figures were carefully produced, though they primarily serve
as an illustration of the analytically achieved results. For such methodical reasons,
the unit lengths on the x- and y-axis are in some figures nonequal.
The extensive bibliography found at the end of the book is to provide the student
with a wide range of resource material. It is also there to help indicate that this set
of examples and exercises could benefit students studying from among a very broad
spectrum of mathematical disciplines.
The origin of this book lies in several analysis and calculus courses that the
authors gave to students on various levels and with, occasionally, quite different
mathematical backgrounds. The experience we got through these lectures, is en-
dowed, we hope successfully, in the given examples and exercises. We would like to
thank many colleagues and students for their remarks, corrections, new problems
(at least for us), their original, better or more precise solutions, and some other
contribution(s) to this book. In particular, it is our pleasure to thank academician
Dr. Olga Hadzic, from the Institute of Mathematics at the University of Novi Sad,
for the careful reading of some early versions of the manuscript and her numerous
improvements in the text. The hard job of preparing the figures was done by Mr.
Milan Manojlovic.
At last but not least, we wish to express our gratitude to Kluwer Academic
Publishers for their kind and generous support throughout the production of our
manuscript. We are especially thankful to Dr. Paul Roos and Ms. Anneke Pot for
their time and effort in coordinating our work.

J. SCHMEELK
DJ. TAKACI
Novi Sad & Richmond, January 1995. A. TAKACI
Chapter 1

Real numbers

1.1 (R, +,', <) as a complete totally ordered field

1.1.1 Basic notions
Definition 1.1. The field of real numbers R = {x, y, z, ooo} is a set with two
operations addition (+) and multiplication (j, and a binary relation termed less
than or equal (S;) defined with the following axiomso

(Rl) (\lx,y,zER) (x+y)+z=x+(y+z)

(associative law for addition);

(R2) (30 E R) (\Ix E R) x +0 = 0 +x = x

(existence of the additive identity element);

(R3) (\Ix E R) (3(-x) E R) (-x) +x = x+ (-x) =0

(existence of the additive inverse);

(R4) (\lx,yER) x+y=y+x

(commutative law for addition);

(R5) (\lx,y,ZER) (xoy)oz=xo(yoz)

(associative law for multiplication);

(R6) (31 E R \ {O}) (\Ix E R) x ° 1 = 1 ° x = x

(existence of the multiplicative identity element);

(R7) (\Ix E R \ {O}) (3x- 1 E R \ {O}) X-loX = X ° X-I = 1

(existence of the multiplicative inverse);

(R8) (\lx,y,zER) xo(y+z)=xoy+xoz, (x+y)oz=xoz+yoz

(distributive law of multiplication over addition);

(R9) (\Ix, y E R) x y = yo x
0

(commutative law for multiplication);

1
2 CHAPTER 1.

(RIO) (\:Ix,y,z E R) (x::; Y I\. y::; z) =? x::; z

(transitivity of the binary relation ::;);

(Rll) (\:Ix, y E R) (x ~ Y I\. Y ~ x) =? x = y

(antisymmetry of the binary relation ~);

(RI2) (\:Ix, y E R) x ~ y VY ~ x
(total ordering of the binary relation ~);

(RI3) (\:Ix, y, z E R) x ~ Y =? (x + z ~ y + z)
(compatibility of the binary relation ~ with addition);

(R14) (\:Ix,yER) (O~xI\.O~y)=?O~x·y

(compatibility of the binary relation ~ with multiplication);

(RI5) Let X and Y be two nonempty subsets of the set R with the property

(\:Ix E X) (\:Iy E Y) x ~ y.

Then there exists an element c E R, such that

(\:Ix E X) (\:Iy E Y) x ~ c~ y

(completeness of the set R).

From the axioms (Rl), (R2) and (R3) it follows that (R, +) is a group, while by
(R4) it becomes a commutative (or: Abelian) group. In the same manner, from the
axioms (RS), (R6), (R7) and (R9) it follows that (R \ {O},·) is also a commutative
group. The axiom (R8) gives the distributivity of multiplication over addition. The
axioms (RIO), (Rll) and (RI2) show that R is a totally ordered set with the binary
relation ~, while the axioms (RI3) and (RI4) give the connection of ::; with the
operations of addition and multiplication. The axioms (RI) - (RI4) define R as a
totally ordered field, and, finally, with (RIS) the set R becomes a complete totally
ordered field.
Fram now on we shall define b - a := b + (-a) and call the operation " - "
subtraction, define alb := a· b- 1 , b:l 0, and call the operation "I" division. Instead
of x . y we shall often write xy for multiplication.
Let x and y be two real numbers such that x ~ y. We define another binary
relation " :2: " by
(Vx,y E R) y:2: x ¢=:? x ~ y.
If both x ~ y and x :I y, we shall simply write x < y. Also, we shall define
x >y ¢=:? Y < x.

Definition 1.2. The absolute value of a real number a is defined by

a, if a> 0;
{ if a = 0;
lal = ~~,
if a < O.
REAL NUMBERS 3

A real number x is positive iff x > 0, resp. x E R is negative iff x < O.

We shall define the distance between two elements x and y from the set R as
Ix - yl. Clearly, the absolute value lal of a real number a gives its distance from the
origin, i.e., from O.
Definition 1.3. An element mER is a lower bound (resp. upper bound) of
a nonempty set Xc R if for every x E X it holds m :::; x (resp. m ~ x).
A set Xc R is bounded from below (resp. bounded from above) if it has
at least one lower (resp. upper) bound.
A set X C R is bounded if it is both bounded from below and from above. A
set X C R is unbounded if it is not bounded.

Definition 1.4. An element i E R is the infimum of a set X C R if the following

two conditions hold.
(i1) i :::; X for every x E X {i. e., i is a lower bound for X)j
(i2) for every i1 > i there exists an element Xl E X such that Xl < i 1 .

Definition 1.5. An element s E R is the supremum of a set X C R if the

following two conditions hold.
(sl) s ~ x for every x E X {i.e., s is an upper bound for X)j
(s2) for every Sl < s there exists an element Xl E X such that Xl > sl'

From Definition 1.4 (resp. Definition 1.5) it follows that the infimum is the greatest
lower bound (resp. the supremum is the smallest upper bound) of a set. In Example
1.50, using axiom (R15), we shall prove that every subset of R bounded from below
has an infimum, while every subset of R bounded from above has a supremum.

Definition 1.6. The infimum (resp. supremum) of a set X C R is called mini-

mum of X (resp. maximum of X) if it belongs to the set X.

We shall denote by inf X, sup X, min X and max X the infimum, supremum, mini-
mum and maximum of the set X C R respectively.

Bounded intervals. Let a and b be two real numbers such that a < b. Then the
following subsets of R (a,b), [a,b], (a, b] and [a,b) are bounded intervals with
endpoints a and b :

open interval: (a, b) := {x E RI a < x < b}j

closed interval: [a, b] := {x E RI a :::; x :::; b}j
other bounded intervals: (a, b] := {x E RI a < x :::; b} and
[a, b) := {x E RI a:::; x < b}.
Unbounded intervals. Let a and b be two real numbers. Then the sets (a, +00),
[a, +00), (-00, b), (-00, b] and (-00, +00) are unbounded intervals, defined as
follows:

unbounded to the right: (a, +00) := {x E RI x> a}j

[a, +00) := {x E RI x ~ a}j
4 CHAPTER 1.

unbounded to the left: (-00, b) := {x E Rj x < b};

(-oo,b]:= {x E Rj x::; b}.
We also put (-00, +00) := {x E R} = R, which is an interval both unbounded
to the right and to the left.

Note that bounded (resp. unbounded) intervals are also bounded (resp. unbounded)
sets in R in the sense of Definition 1.3.

Definition 1.7. A nonempty set X C R is inductive if for every x E X it holds

that the real number x + 1 is also in X.

Definition 1.8. The set of natural numbers, denoted by N, is the smallest in-
ductive set containing the real number 1.
Hence N = {I, 2, 3, ... }.

Quite often we shall use the set No := N U{O}.

The fundamental property of the set N gives the following statement, called the
mathematical induction principle.

Theorem 1.9. Assume a subset X of the set of natural numbers N has the following
two properties.

(mipl) 1 EX;

(mip2) if n E X, then n +1 is also in X.

Then necessarily X = N.

Definition 1.10. The set of integers, denoted by Z, includes the natural numbers,
their additive inverses and the additive identity element, zero.
Hence, Z = { ... ,-2,-1,O,1,2,3, ... }.

The pair (Z, +) is an Abelian group, but (Z \ {O},·) is not.

The number m E Z is a divisor of a number n E Z if there exists an element
k E Z such that n = km.
An integer is even (resp. odd) if it is divisible (resp. not divisible) by 2.
A natural number p, p > 1, is prime if in the set N it has no divisors other than
1 and p itself.
The integers p and q are relatively prime if their only common divisors are the
numbers 1 and -1.
If the largest common divisor of the integers p and q is the number 1, then we
r.q
call the fraction irreducible.

Definition 1.11. The set of rational numbers (or: fractions), denoted by Q,

make the real numbers of the form p . q-l, where p, q E Z, q =f. 0.
The irrational numbers are those real numbers that are not rational.
REAL NUMBERS 5

The set of rational numbers Q satisfies the first fourteen axioms from Definition
1.1, hence endowed with addition, multiplication and the binary relation:::: it also
becomes a totally ordered field. However, the set Q does not satisfy the completeness
axiom (R15).

Theorem 1.12. The Archimedes theorem.

For every two real numbers x > 0 and y there exists a natural number n such that
nx > y.

Theorem 1.13. The Cantor theorem.

Let for every n E N be given a closed interval [an, bnl and assume that for m > n it
holds [am, bml C [an, bn ), i.e., an :::: am :::: bm :::: bn- Then it holds

n [an, bnl of- 0.

nEN

It is important to note that the Cantor theorem is not true in the set of rational
numbers Q.
In Example 1.51 we shall prove that axiom (RI5) is equivalent with the con-
junction of the Archimedes and the Cantor theorem. More precisely, the system of
fifteen axioms (R1) - (RI5) is equivalent to the system of fourteen axioms (Rl) -
(RI4) plus the Archimedes and the Cantor theorem.

1.1.2 Examples and exercises

Example 1.14. Using axioms (Rl), (R2), (R3) and (R4) show that

a) in the set R there exists a unique additive identity;

b) in the set R every element has a unique additive inverse;

c) for given real numbers a and b the equation a + x = b has a unique solution in
R.

Solutions.

a) From the axiom (R2) it follows only the existence of an additive identity. Suppose
there exist two such additive identities, say 01 and O2 . Then from (R2) and
(R4) it follows
01 = 01 + O2 = O2 + 01 = O2 .

b) By axiom (R3) for a given element x E R there exists an additive inverse.

Suppose there exist two such inverse elements to x, say Xl and X2. Then from
(R2), (Rl) and (R4) it follows

Xl = Xl +0 = Xl + (X + xz) = (Xl + X) + X2 = X2 + (Xl + X) = X2 +0= X2.

6 CHAPTER 1.

c) Using axioms (R4), (Rl) , (R3) and (R2), let us check that the number x :=
b + (-a) is a solution of the given equation:

a +x = a + (b + (-a)) = a + (( -a) + b) = (a + (-a)) + b = 0 + b = b + 0 = b.

Let us suppose that Xl is some other solution of the given equation, i.e., it
holds b = a + Xl. Then

( -a) + b = (-a) + (a + Xl) = (( -a) + a) + Xl = 0 + Xl = Xl.

Hence Xl = (-a) + b = b + (-a), which implies the uniqueness of the solution.

Example 1.15. Using the axioms (R1), (R2), ... , (R14) and Example 1.14 show
that for every x, y, x', y' E R it holds

a) X· 0 = 0; b) -x = (-1) . X;
c) -(-x) = X; d) x( -y) = -(xy) = -(x)y;
e) (-x)(-y) = xy; f) X s:; 0 -¢::::::} - X 2 0;
g) X s:; y -¢::::::} -x 2 -y; h) (x < Ol\.y < 0) =} xy > 0;
i) (x s:; Y I\. x' s:; y') =} X +x' s:; y + y'; j) (x < 0 I\. Y > 0) =} xy < 0;
k) 0<1; 1) x > 0 -¢::::::} x-I> O.

Solutions.

a) Using (R6), (RS) and (R2) we have

x + x ·0 = x . 1 + x . 0 = x . (1 + 0) = x . 1 = x,
hence x + x . 0 = x. Now from Example 1.14 c) it follows that x . 0 is also an
additive identity; from Example 1.14. a) then it follows x . 0 = o.

b) From axiom (RS) and a) it follows

x + (-1) . x = (1 + (-1)) . x = 0 . x = x . 0 = O.
Thus (-1) . x is also an opposite number to x; from Example 1.14.b) then it
follows (-1)· x = -x.

c) By the definition of the element -( -x) we can write (-x) + (-( -x)) = 0, while
by the definition of the opposite element -x it holds (-x) + x = O. Using the
uniqueness of the solution of the equation a + x = b (see Example 1.14 c)) we
obtain the statement.

d) From the relations

O=x·O=x(y+(-y))=xy+x(-y), hence xy+x(-y)=O

and xy + (-(xy)) = 0 (see c)), it follows x( -y) = -(xy).

e) The statement follows from d) and c).
REAL NUMBERS 7

f) By axiom (R13) we have

x :s: 0 ::::} (-x) + x :s: (-x) + 0 ::::} 0 :s: -x.
g) x:S: y::::} ((-x) + (-y)) + x:S: ((-x) + (-y)) +y::::} -y:S: -x.

h) Using f), (R14) and e) we have

(x < OAy < O)::::} (-x> OA -y > O)::::} (-y)(-x) > 0 <===} xy> O.

i) Since x :s: y ::::} x + x' :s: x' + y and x' :s: y' ::::} x' + y :s: y' + y, from (RIO) and
(R4) the statement follows.

j) (x < 0 A 0 < y) ::::} (0 < -x A 0 < y) ::::} (0 < (-x)y) ::::} (0 < (( -l)x)y)
::::} (0 < (-l)(xy))::::} (0 < -(xy))::::} (xy < 0).

k) The axiom (R6) implies 1 of. 0, hence by (R12) it is either 1 > 0 or 1 < O. Suppose
1 < 0; then from h) and 1 of. 0 it holds (1 < 0 Al < 0) ::::} (0 < 1·1) ::::} 0 < l.
This is a contradiction, hence necessarily 1 > O.

1) First of all, X-I of. O. Suppose X-I < O. Then from j) it follows
(x- I < 0 A 0 < x) ::::} (x· X-I < 0) ::::} (1 < 0),
which is in contradiction with k).

Exercise 1.16. Prove that

a) (x· y = 0) ::::} (x = 0 V Y = 0);

b) (\:Ix E R)(-l). (-x) = x;

c) (\:Ix E R)(-x)· (-x) = X· x =: x 2 ;

d) (\:Ix,y,z E R)(x < yAy:S: z)::::} x < z;

e) (\:Ix,y,z,wER) (x:S:yAz<w)::::}(x+z<y+w).
Exercise 1.17. Prove that
a) in the set R there exists a unique multiplicative identity;
b) for every x of. 0 there exists a unique multiplicative inverse, denoted by x-\

c) the equation a . x = b has a unique solution in R, provided that a E R, a of. 0,

and b E R.

Example 1.18. Show that for every x E R the following properties of the absolute
value hold.
a) Ixl = 0 <===} x = 0; b) 1- xl = Ixl;
c) -lxl:S: x :s: Ixl; d) Ixl:S: E <===} -E:S: x :s: E;
e) Ixl < E <===} -E < x < E.
8 CHAPTER 1.

Solutions.

a) Follows immediately from Definition 1.2.

b) For x = 0 this is trivial. For x < 0 it holds -x > 0 (see Example 1.15 f», hence
by Definition 1.2 Ixl = -x and I - xl = -x. Thus Ixl = 1- xl- The third case
(x > 0) is handled similarly.

The equalities c) and d) follow from Examples 1.15 d), e), h) and j).

Example 1.19. Prove that for every x, y E R the following properties of the abso-
lute value hold.

a) Ix + yl S; Ixl + Iyl; b) Ix - yI21Ixl-lyll;

c) Ix' yl = Ixl' Iyl; d) y = M'

Ixl Ixl
y =J O.

Solutions.

a) Using (RI3), from the inequalities

-Ixl S; x S; Ixl and -Iyl S; y S; Iyl,

(see Example 1.18 c» it follows

-(Ixl + Iy\) S; x +y S; Ixl + IYI, or Ix + yl S; Ixl + Iyl·

b) Using a), we obtain the following inequalities:

Ixl = Ix - y + yl S; Ix - yl + Iyl and Iyl = Iy - x + xl S; 1- (x - y)1 + Ixl·

Hence
-Ix - yl S; Ixl-Iyl S; Ix - yl '* Ilxl- Iyll S; Ix - yl·

Relations c) and d) follow immediately from Definition 1.2.

Exercise 1.20. Show that for every Xl, X2, ••• , Xn E R the following holds.

Ik=l
txkl:= IXI + X2 + ... + xnl S; t
k=l
IXkl·

Example 1.21. Solve the following equation and inequation.

a) Ix - 11 - Ix + 21 = 1; b) 13x + 21> 41x - 11·

Solutions.
REAL NUMBERS 9

a) From the definition of the absolute value, for x E (-=, -2) (~ x + 2 < 0)
the given equation can be written as -(x - 1) - (-(x + 2)) = 1, which is
equivalent to 3 = 1. Hence there is no solution in the interval (-=, -2) . For
-2 < x :S: 1 the given equation can be written as -(x - 1) - (x + 2) = 1, so
in the interval (-2, 1] the only solution is x = -1. For x > 1 the equation has
the form x-I - (x + 2) = 1, or -3 = 1, which means that no solutions exist
in the interval (1,+=).

b) Every point from the interval (~, 6) is a solution of the given inequation.

Exercise 1.22. Solve the following equations.

a) Ix + 31 = 5; b) Ix - ll-12x - 71 = 2;
c) Ixl- Ix - 11 + 31x - 21- 21x - 31 = x + 1.

Answers. a) x = -8 V x = 2. _-1-
b) x = 4 Vx
3·
° c) x = -1 V x:::: 3.

Exercise 1.23. Solve the following inequations.

a) 12x + 11 :S: 14x - 71; b) IX+ll < 1 x:f. -5;

x+5 '

5X + 21> 2 3
c) 12x - 3 - , x :f. 2; d)
IX:ll>x:l'
x:f. -1;

e) Ix 2 - xl-Ixl < 1; f) Ix 3 - x 2 1< Ix 2 + xl;

g) 11 -Ix - 111 < 1; h) Ilx + 11 - Ix - 111 < 1.

Answers.

a) x E (-=,1] U[4, +=). b) x E (-3,+=).

c) x E (-=,-8] U (~, D U G,+=)· d) xE(-I,O).

e) x E (-1, 1 + J2). f) x E (1 - J2, 0) U(O, 1 + J2).

g) XE(-I,I)U(I,3). h) x E (-~, D.
Example 1.24. Prove that the sum of the first n, n E N, terms of a geometric
sequence with quotient q :f. 1 is given by

l: l-l =
n
qn -1 q :f. 1. (1.1)
k=l q- 1'
10 CHAPTER 1.

Solution. For n = 1 the formula reduces to qO = q - 1, which is true. Assume that

q-l
(1.1) is true for some n = m; then we must show that it is true also for n = m + 1.
To that end, we have
qm _ 1 qm _ 1 + qm+! _ qm
L
m+!
qk-1 = _ _ + qm = -=--------=---_---=_
k=l q- 1 q- 1

Hence
qm+1 - 1
Lq
m+! k
= .
k=l q- 1
Thus we obtained formula (1.1) for n = m + 1. By the mathematical induction
principle it follows that formula (1.1) holds for all natural numbers n.
Example 1.25. Prove the following formula for n EN:
lOn+1 - 9n - 10
3 + 33 + 333 + ... + 33 ... 3 = 7 (1.2)
'----... ' 2
n addends
IOn+! - 9n - 10
Solution. Let us put f(n) := 27 ' n E N. For n = 1 formula (1.2)
becomes
10 2 - 9·1 - 10
3= 27 '
which is true. Assume that (1.2) holds for some n = m. Then we have using Example
(1.24 )
10m +1 - 9m - 10
3 + 33 + 333 •+ ... + 33 ... 3' = + 33 ... 3
, 27 '--v--'
m + 1 addends m + 1 ciphers

= 2\ . (10 m +! - 9m - 10 + 3·27· (lorn + ... + 101 + 10°))

= ~. (10 m +! _ 9m - 10 + 81 . _10...,..m..,..+_1_--,--I)
27 10-1
= _1_ . (9. 10m +! _ 81m - 90 + 81· lO m +1 - 81)
27·9

= 2/ 9 . 9 . (lO m +2 - 9( m + 1) - 10) .
Simplifying the last expression on the right-hand side gives

p+ 33 + 333 f ... + 33 ... 3, = f(m + 1),

m + 1 addends
i.e., formula (1.2) for n = m + 1. The mathematical induction principle gives the
validity of formula (1.2) for every n E N.
REAL NUMBERS 11

Example 1.26. Using the mathematical induction principle, prove the following
formulas for n E N.
n n
n(n+1).
a) Lk=
2
, b) L(2k - 1) = n 2;
k=l k=l

n
n(n + 1)(2n + 1)., n 2
= n(4n -1).,
c) Lk 2 = d) L(2k _1)2
k=l 6 k=l 3
n
n(n+1)(n+2). n 1 n
e) Lk(k+1)= , f)
k=l 3 E(2k-1)(2k+1) = 2n+1

Solutions.
1·2
a) For n = 1 the given formula reduces to 1 = -2-' which is true. If the formula is
true for n = m, mEN, then

E =E +
m+l
k
m
k (m + 1) = m(~ + 1) + (m + 1) = (m + 1)2(m + 2).

Hence the formula is true for n = m + 1. By the mathematical induction

principle, the set of all m for which the given formula is true coincides with
the set of natural numbers N.
. 1 1
f ) For n = 1 thestatement becomes t h e equahty - 1 3 = 2 . Let us assume
. ·1 + 1
that the given formula holds for n = m. Then we have

m+1 1
E . "~. m . + 1
k=l

m(2m + 3) + 1 (m + 1)(2m + 1) m+1

(2m + 1)(2m + 3) (2m+ 1)(2m + 3) 2m+3

Example 1.27. Prove the following formulas for n E N.

a) En (n)2
E k3 = k , hence by Example 1.26 a) it follows

t e = (n(n+1))2.
k=l 2'
n
b) L k(k + l)(k + 2) = n(n + l)(n + 2)(n + 3)
k=l . .
12 CHAPTER 1.

Solution. a) The formula is true for n = 1. Supposing that the formula is true for
n = m, (for some mEN), let us prove it for n = m + 1. Namely, from the inductive
assumption and Example 1.26.a) it follows

m+l
{; k3
m
= {; k3 + (m + 1)3 = {; k
(m)2 + (2 (m(m+1»)
2 (m + 1) + (m + 1)2)

m)2 +2 (m)
= ( {;k {;k (m+1)+(m+1)2= ( m ) 2 = (m+l)2
(;k+(m+l) (;k

Example 1.28. The binomial coefficient (~), n E N, kENo, 0 ~k~ n, is

defined by
( n)
k :=
n!
k!·(n-k)!·
(By definition, we put O! = 1.)
a) Prove the following formula for the binomial coefficients.

k + (k +
( n) +1
n 1) -_ (nk + 1) ' n EN, kENo, 0 ~ k ~ n- 1.

b) Prove the binomial formula

(a + bt = t
k=O
(~)an-kbk, a,b E R, n E N. (1.3)

Solution. Part a) is straightforward, so we go to

b) For n = 1 formula (1.3) reduces to the true formula

(a + W= G)a1-Ob1-1+ G)a1-1b1-O.
Assume (1.3) is true for some n = m. Then we have using a)

(a+b)m+1 (E (7)a m
-
kbk) (a+b)

f (m)a m- k+1bk + (m)a m- kbk+1

k=O k k=O k
f
(;)a m +1 + E((7) + 1)) am- k+1bk + (:)bm+1
(k:

am+1 + f (m + 1) am-k+1bk + bm+1 'El (m + 1) am-k+1bk.

=
k=lk k=O k
REAL NUMBERS 13

Example 1.29. Prove the following formulas .

. na . (n+1)a
n sIn-· sm
a) L sin(ka) = 2. a 2 , a ~ {27r£1£ E Z};
k=l sm"2

n cos -
na (n + l)a
. cos -'-------'.-
b) L cos(ka) = 2. a 2 , a ~ {27r£1£ E Z}.
k=l sm"2
. a . 2a
SIn-sm-
Solution. a) For n = 1 the given formula becomes sin a = ~ a 2 ,which is
sm-
2
true. Supposing that the formula is true for n = m, let us prove it for n = m + l.
· ma . (m+1)a
m+I SIn -2- . sm -'---2----'--
L sin(ka) . a + sin((m + l)a)
k=I sm 2

. ma )
· (m+1)a sm
sm -'-----'-- ( 2a (m+1)a
+ 2 cos --'------'--
2 sin _ 2
2

· (m+1)a
(. (m + l)a
sm 2
--~a'--- sm cos -
a
+ cos (m + l)a sm
. a)
-
sin - 2 2 2 2
2

· (m+1)a . (m+2)a
sm . sm 22
. a
sm-
2
Example 1.30. Prove the following formulas.

a)
k=I
n cos 2k = 2 SInsin-
n X x
x'
n
0 < x < 7r;
2n

b) /2 + ,/2 -+; ... + V2, = 2 cos 2:+1 .

n square roots

Solutions.

a) For n = 1 the given formu Ia becomes cos -x = sm x x ,wh·ICh IS

. equivalent wit h
2 21 sin-
21
sin(2y) = 2 sin y cos y for 0 < y < 27r. Next, we shall assume that the given
14 CHAPTER 1.

formula holds for some n = m. Then we have

m+l x smx x
n cos 2k
k=l
x . cos .
2m sin- 2 +1
m
~-

2m
x
sin x . cos 2m +1 slnx
x . x x
2m. 2 cos - _ . sm 2m + 1 2m+1 . sin 2m +1
2m +1

b) For n = 1 we have the correct equality .J2 = 2 cos ~. Assume the formula is true
for n = m. Then

,r;~ J2 ~ ... + ~ = /2 + 2 cos 2:+ 1

m+l square roots

+ cos
= 2\1 1 - -- 2:+1 = 2 cos (~(_Jr_))
2 2m+1 = 2 cos 2:+ 2 '

which means that the formula is also true for n = m + 1.

Remark. We used the equalIty cos

.
"20: = VI + 2coso: for 0 ::; 0: ::; Jr.
Example 1.31. Prove the following inequalities for n E N.

a) (1 + x)n ::::: 1 + nx, x> -1, (the Bernoulli inequality);

b) (1 + xI)(l + X2)··· (1 + x n ) ::::: 1 + Xl + X2 + ... + x n ,

(the generalized Bernoulli inequality), where Xk > -1, k = 1, ... ,n,
and all Xk are of the same sign;

n+1)n
c) n! < ( -2- , n> 1.

Solutions.

a) For n = 1 we have (1 + x)l ::::: 1 + 1· x, which is always true. Let us suppose now
that the inequality is true for n = m, i.e., (1 + x)m ::::: 1 + mx. We shall prove
next that it is then true for n = m + 1 and x > -1.

(1 + x)m+1 (1 + x)(l + x)m ::::: (1 + x)(l + mx)

1+(m+1)x+mx2:::::1+(m+l)x, since mx2:::::0.
REAL NUMBERS 15

c) For n = 2 the given inequality 2! < (2; 1 )2 is clearly true. Assume the inequal-
ity is true for n = m, then it holds
m + l)m
(m + I)! (m + 1) . m! < (m + 1) ( -2-

2. (m;2)m+1 1
(1+--
1 )m+l.
m+l

From the Bernoulli inequality (see a)) it follows

(1 + -1)
-
m+l
m+l
1
2: 1 + (m + 1)· - - = 2,
m+l

hence
m + 2)m+1 1 (m + 2)m+1
(m + I)! < 2· ( - - .- = --
2 2 2
Example 1.32. Prove the following inequalities for n E N.
a) Xl +X2 + ... +X n 2: n, where Xk > 0, k = 1, ... ,n,
and XIX2··· Xn = 1;

1 1 1 1
b) - - + - - + ... +->_.
n +1 n +2 2n 2'
1 1 1
c) --+--+···-->1·
n+1 n+2 3n + 1 '
1 1
d) fo < 1 + v'2 + ... + fo < 2..jii, n > 1.

Solutions.
a) We shall use the mathematical induction. For n = 1 the statement is trivial.
Assume it is true for n = m, i.e., that for any m positive numbers whose
product is 1, the given inequality holds.
So let Xl, X2, ... , X m , Xm+l be m + 1 positive numbers whose product is equal
to 1. Then either all Xk-S are equal to 1, in which case the left-hand side of
the inequality becomes just m + 1, or at least one of them is less than 1, say
Xm < 1 and, of course, at least one is greater than 1, say Xm+l > 1. Then the
m numbers Xl, X2, ... , Xm-l, X m · Xm+l satisfy the inductive assumption, hence

Xl +X2+·· ·+Xm-l +xm ·Xm+l 2: m {::::::} Xl +X2+·· ·+Xm-l 2: m-Xm ·Xm+l.

Using the last inequality we obtain

Xl + X2 + ... + Xm + Xm+l 2: m - Xm . Xm+l + Xm + Xm+l

= m+ 1+ xm(1 - Xm+l) + Xm+l - 1 = m + 1 + (1 - Xm)(Xm+l - 1).

16 CHAPTER 1.

Since by assumption 1 - Xm > 0 and Xm+l - 1 > 0, it follows that

Xl + X2 + ... + Xm + Xm+l 2: m + 1.

d) For n = 2 the given formula becomes

1
V2 < 1 + V2 < 2V2,
which is true, since 1.41 < V2 < 1.42. Assume that the formula is true for
some n = m. Then we have
111 1
1+-+ .. · + - + >vm+-.
V2 Vm vm + 1 vm + 1
Once we prove that the expression
1
vm+ ~-vm+1 (1.4)
m+1
is positive, the left-hand side of the given formula follows by the mathematical
induction principle. To that end, the expression in (1.4) is equal to

1 ~ / m( m + 1) + 1 - (m + 1) / m( m + 1) - m
vm+ -ym+1= = ,
vm + 1 vm + 1 vm + 1
which is positive.
For the right-hand side inequality in the given formula, from the inductive
assumption
1 1
1+ - + ... + - < 2vm
V2 Vm '
we have
1 1 1 1
1+-+ .. · + - + <2vm+-.
V2 Vm vm + 1 vm + 1
The right-hand side of the given formula will follow once we prove the following
inequality.
2/m(m+1)+1
vm + 1 - 2Vm + 1 < O.
In fact, we have

2/m(m+l)+1 2/m(m + 1) + 1 - 2m - 2
vm+ 1 -2vm+ 1 vm+l

v4m 2 + 4m - v4m 2 + 4m + 1
vm+l
which is clearly less than zero.
REAL NUMBERS 17

Example 1.33. The expressions

Xl + X2 + ... + Xn
A:= A(Xl,X2,'" ,x n ) := , XI,X2,··· ,Xn E R,
n

9 := 9(XI, X2,"" xn) := y'XI . X2··· X n , Xb X2,"" Xn ;:::: 0,

n
H:=H(XI,X2,""X n ):= 1 1 l' XI,X2,···,Xn >0,
-+-+ ... +-
Xl X2 Xn

are respectively called the arithmetic, geometric and harmonic mean of the numbers
Prove the following inequalities.
Xl, X2, ... , X n .

a) Xl + X2 + ... + Xn ;:::: y'XI . X2 ... X n ,

XI,X2, .•. ,Xn ;:::: 0;
n
n
b) y'XI . X2 ... xn;:::: 1 1 1 ' XI,X2"",Xn > O.
-+-+ ... +-
Xl X2 Xn

Remarks. The inequalities a) and b) mean that it holds

A;::::9;::::H

for XI,X2"",Xn > O.

Note that the equalities in a) and b) occur if and only if Xl = X2 = ... = Xn .
We suggest to the reader to check that for n = 2 it holds
A·H = 92 •

Solutions.
a) For n = 1 the inequality reduces to a trivial equality. For n = 2 it becomes
Xl + X2 ~ (VX1 - v'XZ)2
2 - V XIX2 = 2 ;:::: O.
Suppose now that the inequality is true for n = m > 2. Then
Xl + X2 + ... + Xm
Xl + X2 + ... + Xm + Xm+1 m
m
+ Xm+l
m+l m+l

>
m y'XI . X'2-;~ + Xm+1
m+l
From the last inequality we have
+ X2 + ... + Xm + Xm+l
m +1
Xl
- m+ijXl . X2 •.. Xm+l

> m y'XI' X2" 'X m + xm+1 m+ijXI ,x2",xm+l'

m+ 1
18 CHAPTER 1.

Putting pm(m+1) := Xl . X2 ... Xm and qm+l := Xm+b we obtain

m y'XI-:X;-·--;-;.1;';; + Xm+l
m+~XI . X2 ... Xm . Xm+l
m+1

mpm+l + qm+l _ pmq p _ qm))

= _1_ (mpm(p _ q) _ q(m
m+1 m+1

= p --
- q mp (m - qp m-l - q2 pm-2 - ... - q m)
m+1

p - q ((pm _ qpm-l) + (pm _ q2 p m-2) + ... + (pm _ qm))

m+1

p - q (pm-l(p _ q) + pm-2(p2 _ q2) + ... + (pm _ qm))

m+ 1
(p - q)2 (pm-l + pm-2(p + q) + ... + (pm-l + pm-2 q + ... + qm-l)) ~ o.

b) Applying the inequality a) to the numbers ~, ~, ... ,~, we obtain

Xl X2 Xn
1 1 1
-+-+
Xl X2
... +-
Xn >
/1
n _ • _
1
.•• _
1
=
1
,
n Xl X2 Xn y'XI . X2 ... Xn

which is equivalent with the statement.

Exercise 1.34. Prove that for n E N, n > 2, it holds

a) n! > 2n-1., b) (n!)2 > nn; c) 3 nn! > nn > 2nn!;

d) (2n - I)!! := (2n - 1)(2n - 3)···3·1 < 2n . nL

Example 1.35. Show that the sum of two natural numbers is again a natural num-
ber.

Solution. Let m, n E Nj using the mathematical induction principle, we shall

prove that m + n is also a natural number. Let us denote by N 1 the following set.

N 1 := {n E NI ('tim E N) m+n EN}.

The set N (of natural numbers) is an inductive one (see Definitions 1.7 and 1.8),
which means that for every mEN it follows m +1 EN, hence 1 E N 1. Assuming
that n E N I , i.e., m+n E N for every mEN, it follows from axiom (R1) that the
number m + (n + 1) = (m + n) + 1 is also naturaL Hence, by the mathematical
induction principle Nl = N.
REAL NUMBERS 19

Example 1.36. Using the mathematical induction principle, prove the following
properties for the set of natural numbers.

a) If m, n E N such that m > n, then m - n is also a natural number;

b) the minimum of the set N is the number 1;

c) if m, n E N such that m > n, then m ~ n + 1;

d) every nonempty subset A ofN has a minimum.
Solutions.

a) Let n = 1. Then from the assumption it follows m > 1, hence there exists an
element pEN such that m = p + 1. This is equivalent to p = m - 1, which
means that the statement is true for n = 1. Suppose now that it is true for an
element kEN; i.e., that it holds m > k =? m - kEN. Assume m > k + 1.
Then m - k > 1, so from the (already proved) statement for n=l, it follows

m - k > 1 =? ((m - k) - 1) EN=? (m - (k + 1)) E N.

Hence the statement is true for k + 1 EN, provided it is true for kEN.

b) Since by (R6) 1 EN, we have to prove that for every n E N it holds 1 ~ n. For
n = 1 this is trivial. Let us suppose that 1 ~ k for some kEN. Then from
Examples 1.15 k) and 1.15 i) it follows

(1 ~ k 1\ 0 ~ 1) =? 1 ~ k + 1.

c) Suppose m, n E Nand m > n. Then from a) it follows (m - n) E N and from

b) it follows m - n ~ 1. Hence m ~ n + 1.

d) Let us suppose the contrary, i.e., that there exists a nonempty set A C N which
has no minimum. From b) it follows 1 rf- A. Let us define a set B by

B = {n E N I (Va E A) n < a}.

We shall prove that B = N.
By b), for every a E A it holds 1 ~ a; since 1 < a, it follows that 1 E B.
Now assume that nEB and we shall prove (n + 1) E B. Firstly, since A has
no minimum, for every a E A there exists a' E A such that a' < a. This implies

nEB =:} n < a' =? (n + 1 < a' + 1 ~ a) =? n + 1 < a,

which means that (n + 1) E B. By the mathematical induction principle, it
follows that B = N. By the definition of the set B, it follows An B = 0,
hence A = AnN = 0. This is in contradiction with the assumption that A is
nonempty.
20 CHAPTER 1.

Exercise 1.37. Let m, n E N. Prove that there exist unique numbers q E No and
rENo, 0:::; r < n, such that m = nq + r.

Exercise 1.38. Prove that the product of natural numbers (resp. integers) lS a
natural number (resp. an integer).

Exercise 1.39. Prove that

a) a natural number can not be both even and odd;

b) every natural number is either even or odd;

c) if n is an even natural number, then so is n 2.

Exercise 1.40. Prove that (Q, +,',:::;) is a totally ordered field, i.e., that Q satisfies
the axioms (Rl) - (R14).

Example 1.41. If a and b are two rational numbers, a < b, prove then that there
exists an element cEQ, such that a < c < b.

Solution. Let us put c := a ; b. It holds

a a a b b b
a=-+-<-+-=c<-+-=~
2 2 2 2 2 2
(This procedure shows that actually there exist infinitely many rational numbers
between any two given rational ones.)

Example 1.42. Prove that the equation

x2 = 2 (1.5)

has no solution in the set of rational numbers Q.

Solution. Let us suppose that there exists a rational solution x = r of equation

p
(1.5); we may assume that r > O. Then r = -, where p and q are natural numbers
q
2
with largest common divisor 1, i.e., the fraction E is irreducible. Since r2 = P2 = 2,
q q
it follows p2 = 2q2. This implies that p2 is an even number, hence p is even itself.
We can write p = 2k for some kEN, which gives us q2 = 2P. Now q2 is also an
even number, and finally so is q. But then the fraction E is reducible, since both
q
its numerator and denominator are even. This is a contradiction, hence there is no
solution of equation (1.5) in the set Q.

Example 1.43. Prove that the equation (1.5) has a solution in the set R.
REAL NUMBERS 21

Solution. Let us define two sets X and Y as follows.

X:= {x E R+I x 2 < 2} and Y:= {y E R+I y2 > 2},

where R+ = {x E RI x > O} is the set of positive real numbers. Since 12 = 1 < 2

and 22 = 4 > 2, it follows that 1 E X and 2 E Y, hence both X and Yare nonempty.
Clearly, it holds X n Y = 0. If x and yare positive numbers, in view of axiom (R14)
it holds
x < y <¢:::=} x 2 < y2.
This means that any element of X is smaller than any element from Y, hence the
conditions of axiom (R15) are satisfied. By (R15), there exists a real number r such
that
(Vx E X) (Vy E Y) x::; r ::; y. (1.6)
Let us prove that r2 = 2. By (R12) there are three possibilities, out of which exactly
one is true, namely r2 < 2 or r2 > 2 or r2 = 2. Of course, our goal is to prove that
the last case is true; we shall show that the first two cases are impossible. Suppose
first r2 < 2. Then the real number rl := 2 r + 1 is in the set X, since
r+2
2 r2 + 2r + 1 r2 - 2
r1 - 2 = 4 . ( )2 - 2 = 2 . ( )2 < O.
r+2 r+2

However, it also holds

r +1 r2 - 2
r - rl = r - 2 - - = - - <0
r+2 r+2 '
which means that rl E Y, hence rl E X n Y - a contradiction, since the last
intersection is empty. The assumption r2 > 2 gives a contradiction in an analogous
way, so the third possibility remains, namely r2 = 2.
Remark. By Example 1.42, any solution x of the equation (1.5) is not a rational
number. In fact, there are two real numbers satisfying x 2 = 2, namely V2 and
-V2.
Example 1.44. Show that

a) the root vn for m, n E N is either an integer or an irrational number;

b) Vn + y"nTI is an irrational number for every n E N;

c) In + Vn is an irrational number for every n E N.

Solutions.

a) Let us assume the contrary, i.e., vn = E,q where p and q are integers with largest
common divisor 1 and Iql i- 1. Then n =
pm, which implies that p and q have
qm
a common divisor different from 1, which is a contradiction.
22 CHAPTER 1.

b) For every natural number n it holds

n2<n(n+l)«n+l?,

hence n (n + 1) is not a complete square. Then by a), the root j n (n + 1) is

an irrational number. Since

(Vn + vn+1)2 = 2n + 1 + 2jn(n + 1),

it follows that the number (fo + Vn+1)2 is irrational, hence so is the number
fo+Vn+1·
c) If n = b2 for some bEN, then from the equality

jn + Vn = ..jb2 + b = jb(b + 1)
it follows that jn + fo is not rational.

If n is not a square of some integer, then fo is not rational, hence jn + fo

is neither a rational number.

Exercise 1.45. Prove that

a) the number ..j3 is irrational;

b) the numbers 5...;2 and 3 + ...;2 are irrational, using that ...;2 is irrational;

c) In: 1 is an irrational number for every n E N.

Example 1.46. Prove that the Archimedes theorem (Theorem 1.12) holds in the
set of rational numbers, i.e.,

(Va E Q, a > 0) (Vb E Q) (:In E N) na > b.

Solution. If a > 0 and b < 0, then we can take n = 1. So we can suppose that
ml nl
a = - , b = - , for some natural numbers ml, m2, nl and n2. Put n := m2nI + 1;
m2 n2
then
ml nl
na = (m2nl + 1) - > ml nl ~ - = b,
m2 n2
since mIn2 ~ 1.

Example 1.47. Using the Archimedes theorem in R (see Theorem 1.12), show that
. 1
a) for every x E R, x 1: 0, there exzsts an n E N such that 0 < - < Ixl;
n

b) if for a nonnegative number x it holds that for every n E N x < ~, then x = 0;

n
REAL NUMBERS 23

c) for every x E R there exist numbers nI, n2 E N such that -nI < x < n2;

d) for every x E R there exists a unique ko E Z such that ko ~ x < ko + 1;

e) for every x, y E R, x < y, there exists a rational number r, such that x < r < y.
(Compare with Example 1.41')

Solutions.

a) Assume x i' 0, then Ixl > 0. Thus from Theorem 1.12 it follows that for the real
numbers 1 and Ixl there exists an n E N such that nlxl > 1, or ~n < Ixl.

b) If we suppose that x > 0, then from a) it follows that there exists an n E N such
that x > ~. (Put b = 1 and a = Ixl in the Archimedes theorem.)
n
°
c) Since 1 > (see Example 1.15 k», there exists an n E N such that n· 1 > x.
Analogously, there exists an n' E N such that n' . 1 > -x or -n' < x. Putting
nI = n' and n2 = n, we get the statement.

d) Applying the Archimedes theorem, there exists a kEN such that k·l > Ixl ~ x,
hence k > x. Let ko be the largest among the numbers 0, ±1, ... , ±k which
are not greater than x. Then ko ~ x < ko + 1. Clearly, by the construction,
the integer ko is unique.

Remark. The greatest integer part of a real number x denoted by [xl is defined
to be the largest integer less or equal to x.

e) If x < y, then y - x > 0, hence by a) there exists an n E N such that ~ < y - x.

n
By d), there exists a unique p E Z such that p = [nx], i.e., p ~ nx < p + 1.
Then it holds p + 1 < ny, since otherwise p + 1 ~ ny would imply
1
n(y - x) ~ p + 1 - P =? Y - x ~ -,
n

··
a contrad IctIon. P +-
S·mce - 1 E Q an d x p +-1 < y, £or th e ratlOna
<- . 1 numb er
n n
p+l
r we can take - - .
n
Remark. From Example 1.47 e) it immediately follows that between any two real
numbers a and b, a < b, there are infinitely many rational numbers and infinitely
many irrational numbers.

Example 1.48. 1 Let [an, bnl, n E N, be a sequence of closed intervals with the
following properties

(i) m <n =? am ~ an ~ bn ~ bm (every interval is contained the previous one);

1 In this example, we use the notion of the limit of a sequence - see Definition 3.1.
24 CHAPTER 1.

(ii) n~oo
lim (bn - an) = 0 (the lengths of intervals tend to zero).

Prove that then there exists exactly one real number a belonging to every interval
[an, bnl·
Solution. By the Cantor theorem, the intersection n [an, bnl is not empty. Hence,
nEN
there exists a real number a which belongs to all intervals [an, bn ], i.e., an :=:; a :=:; bn
for every n E N. We have to prove that a is the only such real number. Assume
there exists another real number fJ i' a such that an :=:; fJ :=:; bn for every n EN.
Assume a < fJ, then it holds an :=:; a < fJ :=:; bn for every n E N. This implies
o < fJ - a < bn - an, a contradiction with the condition (ii). The assumption a > fJ
is handled in an analogous way.
Example 1.49. Is the Cantor theorem (Theorem 1.13) true when applied to afam-
ily of arbitrary intervals (In)nEN such that

n >m =:} In C 1m? (1.7)

Solution. In general, without the assumption that the intervals are closed, the
Cantor theorem is not true. Namely, put In = (0, ~),
n E N, then the relation (1.7)
n
holds. However, in view of Example 1.47 b), the intersection n In is empty.
nEN

Example 1.50. Let us denote by (R15) , the following statement.

(R15)' Every nonempty set X C R bounded from below has an infimum.
a) Prove that every non empty set X C R bounded from above has a supremum
(resp. bounded from below has an infimum).

b) Prove that if in the system of axioms (Rl) - (R15) one replaces the last one,
namely (R15), with (R15) , then the system of axioms (Rl) - (R14) and (R15) ,
becomes equivalent with the system (Rl) - (R15).
(Then we say that the axiom (R15) is equivalent with (R15)'.)
Solution.
a) We shall prove only the statement that if a nonempty set X C R is bounded
from above, then it has a supremum. Let the set Y be defined by
Y = {y E RI (Vx E X) y 2 x}.
By assumption, the set Y is non empty and it holds
(Vx E X) (Vy E Y) x:=:; y,

hence, we can apply axiom (R15). It gives us the existence of a number c E R,

such that
(Vx EX) (Vy E Y) x:=:; c:=:; y. (1.8)
By construction of the set Y, it holds c = min Y = supX. Namely, if c ~ Y,
then there exists an x E X such that x > c, which is in contradiction with
(1.8).
REAL NUMBERS 25

b) In a) we proved that the system (R1) - (R15) implies (R15)'. We have yet to
prove that if the axioms (R1) - (R14) and (R15)' hold, then the statement in
(R15) (from Definition 1.1) follows.
Assume that X and Yare two nonempty sets from R such that

('<Ix E X)('<Iy E Y) x::; y. (1.9)

The set Y is then bounded from below, hence by (R15)' the set Y has an
infimum c. By (i1) from Definition 1.4, for every y E Y it holds c ::; y. Finally,
let us show that ('<Ix E X) x::; c. Otherwise, there exists an element Xl E X
such that Xl > c. But then the number Xl is a lower bound for Y greater than
c, which is in contradiction with (i2) from Definition 1.4.

Example 1.51. Prove that the axiom (R15) is equivalent with the conjunction of
the Archimedes and the Cantor theorem (see Theorems 1.12 and 1.13).

Solution. We shall omit the proof of the statement that the system of axioms
(Rl) - (R14) plus (R15)', being equivalent to the system (R1) - (R15), implies the
Archimedes and the Cantor theorem. We shall only prove that assuming axioms
(R1) - (R14) and both the Archimedes and the Cantor theorem, every nonempty set
bounded from below has an infimum. Then by Example 1.50 b), the axiom (R15)
follows.
Assume that the nonempty set B C R is bounded from below with a number aI,
and let bl E B. Then we claim that for every mEN there exists the largest number
nm E No such that
nm
am := al +- m 2
is still a lower bound for B. In fact, if al = inf B, then nm = o. If al oj:. inf B, then
there exists nm E N such that

('<Ix E B) am = al + ~: ::; x.
If this were not true, then for some mEN there would be no such n = n m , but
n
rather it would hold al + 2m ::; X for every n E N and every X E B. But then
2: ::; X - al for every n E N, which contradicts the Archimedes theorem.
1 nm + 1
Put now bm := am + -2 m = al + ---; 2m then bm E B for every mEN. By
construction,
al ::; a. ::; ... ::; am ::; ... ::; bm ::; ... ::; b. ::; bl ,
hence
[aI, bl ] ~ [a., b.] ~ ... [am, bm] ~ ....
By the Cantor theorem, there exists a real number b, such that

bEn [am' bm].

mEN
26 CHAPTER 1.

Then it holds am :::; b:::; bm for every mEN. Let us show next that for every 6 > 0
there exists an mEN such that b - 6 :::; am. In the contrary, there would exist an
6> 0 such that for every mEN it holds am < b - 6. This would imply

= al + -nm nm
al + --- - +1 = a m + -21m -
am 2m < b - 6 <
- 2m 6 6.

Thus it follows
1 m 1
6 < -m
<¢:::=} 2 <-.
2 6
Using the Bernoulli inequality (see Example 1.31 a)) it would then follow
1
- > 2m = (1
6
+ l)m 2: 1 + 1 . m
for every mEN, which contradicts the Archimedes theorem.
So we obtained that for given 6 > 0 there exists an mEN such that b- 6 :::; am.
In an analogous way, one can prove that for given 6 > 0 there exists an mEN such
that b:::; bm :::; b + 6 (do that!).
Putting these conclusions together, we obtain that for given 6 > 0 there exists
an mEN such that
b - 6 :::; am :::; b :::; bm :::; b + 6. (1.10)
Finally, let us show that b = inf B. By the construction, for every x E B there exists
an element bm 2: b such that bm :::; x, hence b is a lower bound for B. Further on,
from relation (1.10) it follows that b is also the greatest lower bound for the set B.
Remark. This example shows that the Cantor theorem does not hold for the set of
rational numbers Q. Namely, as we shall see in Examples 3.33 and 3.36, the subset
X of Q defined by
X:= {(I +~) n I n EN}
is bounded from above, but has no supremum in Q. In fact, it does have a supremum
in R and it is the irrational number e.

Example 1.52. Let the set A be given by A = {~ ± 3n: 11 n EN}. Prove that
2
inf A = 0 and sup A = 3'

Solution. We have for n E N

1 n 1 n 1
- ± -- > - - -- = > O.
3 3n + 1 - 3 3n + 1 3(3n + 1)
This means that the set A is bounded from below with O. By Example 1.50 it has
an infimum. We prove next that just 0 is that infimum. If, however, a positive
number 6 would be the infimum of A, then clearly 6 :::; 112 (put n = 1). Putting

no := [~(1 - 36)] + 1 it would then hold

96
1 n
n > no :::} 0 < - - - - <
+1
6
3 3n '
REAL NUMBERS 27

a contradiction.
Let us prove now that the supremum of A is ~. Firstly, ~ is an upper bound of
A, since it holds for every n E N

2 (1- ± - - >
--
3 3
n) -3(3n+l)
3n+l
1 >0
.
Next, for every e > 0 we choose no as above; then for every n > no it holds
2 1 n 2
- - e < -
3 3
+- - < -.
3n +1 - 3
2
The last inequality implies that 3" is the supremum of the set A, since it means
that for every e > 0 there exists an element x from A which belongs to the interval
( ~3 - e ~).
'3
Example 1.53. For a nonempty set X C R define

-X:= {-xl x EX}.

Prove that

a) inf( -X) = - supX; b) sup( - X) = - inf X,

provided that in a) (resp. in b)) X is bounded from above (resp. from below).

Solutions. We shall prove only part a), since b) is quite analogous.

By Example 1.50 the set X has a supremum; let us denote it by M. Then since
M is an upper bound for X, it holds

(Vx E X) x:::; M -¢=? (Vx E X) - x2 -M

Thus it holds
(Vy E (-X)) y 2 -M. (1.11)
Since M is the smallest upper bound for X, it holds

(Ve > 0) (3XI E X) Xl > M - e -¢=? (Ve > 0) (3XI EX) - Xl < -M + e.
Thus it follows
(Ve> 0) (3X2 E (-X)) X2 < -M +e. (1.12)
The relations (1.11) and (1.12) mean that -M is the infimum of the set -x.
Example 1.54. Let X and Y be two nonempty subsets of R bounded from below
(resp. from above). Put

S := {s = x + yl x E X, Y E Y}.
Prove that S has an infimum (resp. a supremum) and it holds

inf S = inf X + inf Y (resp. sup S = sup X + sup Y).

28 CHAPTER 1.

Solution. In view of Example 1.50, if the sets X and Yare bounded from below,
then they have an infimum; let us denote by ml = inf X and m2 = inf Y. Hence it
holds
(VX E X) x 2 ml and (Vy E Y) y 2 m2.
Let s be an arbitrary element from S; then there exist x E X and y E Y such that
s = x + y. Then we have
s = x + y 2 ml + m2,
hence the set S is bounded from below by the sum ml + m2. Let us show that the
last number is also the greatest lower bound for S. For given $, there exist Xl E X
and Yl E Y such that
$ $
Xl < ml +2 and Yl < m2 + 2·
This implies
(V$ > 0) (:lSI := Xl + Yl E S) Sl = Xl + Yl < ml + m2 + $,
and thus we obtain that ml + m2 is the infimum of the set S.
Exercise 1.55. Find, if any, the infimums and supremums of the following sets and
check whether they are also their minimums or maximums.
5n --
a) X= { - 11 nEN }; b) X={~lnEN};
7n +2
c) X {I + 2.(~I)n In EN};
= d) X = {E 2\ I n EN} ;

e) X={:lm,nEN,m<n}.
Answers.

a) infX = minX =~, supX =~. b) inf X = 0, sup X = max X = 2.

c) infX = minX = -1, sup X = 2. d) infX = minX =~, sup X = 1.

e) inf X = 0, sup X = l.
Exercise 1.56. Let X and Y be two nonempty bounded subsets of R, and put
X - Y := {z E RI (:Ix E X) (:ly E Y) z = X - y}.
Prove that sup(X - Y) = sup X - inf Y.
Exercise 1.57. Assume X and Yare two nonempty subsets of R+ (0, +=)
bounded from below, and put
Z := {z E RI (:Ix E X) (:ly E Y) z = X· y}.
Prove that
a) inf Z = inf X . inf Y if X and Yare bounded from below;

a) sup Z = sup X . sup Y if X and Yare bounded from above.

REAL NUMBERS 29

1.2 Cuts in Q
1.2.1 Basic notions
In Subsection 1.1 we gave the axioms (R1) - (R15) that defined the set of real
numbers R as a complete totally ordered field, (R, +,', ~). One can prove that,
up to an isomorphism, there exists a unique totally ordered field (Q, +"'~) which
satisfies the axioms (R1) - (R14). However, as one can see from Example 1.51, in
Subsection 1.1.2, Q does not satisfy the last axiom (R15).
The goal of this section is to expose an effective construction of the set of real
numbers R starting from the set of rational numbers Q. This method, developed
firstly by the German mathematician R. Dedekind in 1872, uses the so-called cuts
in Q. 2

Definition 1.58. A cut 0: in the set of rational numbers Q (shortly; cut) ~s a

subset of Q with the following three properties.

(c1) 0:=10 1\ o:=lQ;

(c2) (Va,b E Q) (a E 0:) b < a =} b E 0:;

(c3) 0: has no maximum.

The set of all cuts will be denoted by R.

Every rational number defines a cut as follows.

Theorem 1.59. Let r be a rational number and put

r* := {p E QI p < r}. (1.13)

Then r* is a cut, called rational cut, defined by the rational number r.

The set of rational cuts will be denoted by Q. In particular, 0* (the zero-cut)

and 1* (the one-cut) are rational cuts. We shall see in Example 1.69 that Q is a
proper subset of the set of all cuts R; those cuts that are not rational, will be called
irrational cuts.
In R we define the relation " -< " by

(Vo:,~ER) o:-<~ {::::::} (3PEQ) pE(~\o:). (1.14)

We also put 0: :5 ~ if either 0: = ~ or 0: -< ~. The notations" »- " and " ~ " are
interpreted analogously. In Example 1.62 we shall prove that :5 is a total ordering
in R.
2 Another important construction of the set R starting from Q uses the so-called Cauchy se-
quences (see Definition 3.8), and was elaborated by another German mathematician, namely G.
Cantor. It is interesting to note that Cantor and Dedekind almost simultaneously announced their
constructions to the mathematical community.
30 CHAPTER 1.

If a cut a satisfies a >- 0* (resp. a t 0*) then it is called a positive cut (resp.
nonnegative cut). The set of positive cuts will be denoted by R+. The negative
and nonpositive cuts are defined analogously.
The addition in R, denoted by EB, is defined by

a EB 13 := {x + YI x E a, y E f3} (1.15 )

for arbitrary cuts a,f3 E R. We shall show that (R, EB) is an Abelian group, with
the additive identity 0*.
The absolute value lal of a cut a is defined by

lal := { a, ~f a to:; (1.16)

8a, If a -< 0 ,
where 8a is the additive inverse of a for EB (see Example 1.65 b)). Clearly, for every
a E R it holds lal t 0*, and lal = 0* iff a = 0*.
In order to define the multiplication in R, denoted by 0, we shall start with
the multiplication in R+

a0f3:= {xyl x E a, y E f3}U{x E QI x::::; O} (1.17)

for arbitrary positive cuts a,f3 E R+. It is an easy task to show that a 0 13 is a
positive cut when a,f3 E R+. Next, we put for every cut a

a 0 0* = 0*.

Finally, if a and 13 are arbitrary elements in R, then we put

a0f3 := { lal 0 1131, if eithe~ a :::S 0* and 13 :::S 0*, or a t 0* and 13 t 0*; (1.18)
81a101f3l, otherWIse.

We shall show in Example 1.66 that (R \ 0*,0) is an Abelian group, with the
multiplicative identity 1*.
Moreover, the distributive law holds in R.

(Va,f3"ER) (aEBf3)0,=(a0,)EB(f30,). (1.19)

So we obtain that (R, EB, 0,:::s) is a totally ordered field. In Example 1.70 we
shall show that it is isomorphic to the field (R, +, " ::::;). This means that there is a
bijection3 <ft : R -+ R such that for every a,f3 E R it holds

a :::S 13 {=::> <ft( a) ::::; <ft(f3); <ft( a EB (3) = <ft( a) + <ft(f3); <ft( a 0 (3) = <ft( a) . <ft(f3).
Moreover, <ft can be chosen with the property that its restriction to the set of rational
cuts Q satisfies the condition

(Vr E Q) <ft(r*) = r.
3See Subsection 2.1.1.
REAL NUMBERS 31

1.2.2 Examples and exercises

Example 1.60. Show that if a rational number x does not belong to a cut a, then
for every yEa it holds x > y.

Solution. Assume that there exists ayE a such that y 2 x. Then by property
(c2) from Definition 1.58 it follows that x E a, contradicting the assumption x rf- a.

Example 1.61.

a) Prove Theorem 1.59, i.e., that the set r* = {p E QI p < r} is a cut for every
rational number r.

b) Let a be a cut. Prove that the set Q \ a has a minimum iff a is a rational cut.
Solutions.

a) Take a rational number r and define the set r* by (1.13). We have to check the
three conditions (cl) - (c3) from Definition 1.58.

• The set r* is nonempty, since it contains, for instance, the rational number
r - 1, and since r + 1 rf- r*, it is a proper subset of Q.
• The second condition is satisfied by definition.
• Clearly, the supremum of r* is r, which by definition is not in r*.

Thus it follows that r* is a cut.

b) By property (cl) from Definition 1.58, the set Q \ a is nonempty. We shall

analyze two cases. The first is when a is a rational cut and the second, when
it is not.
Clearly, if a = r* for some r E Q, then r rf- a, which implies that r E (Q \ a).
Thus r = inf(Q \ a) = min(Q \ a). Let a be a cut from R \ Q. Assume that
the set Q \ a has a minimum r. Then by Example 1.60 for every x E a it holds
x < r, hence by Theorem 1.59 a = r*, which is a contradiction.

Example 1.62. Prove that the binary relation ~ is a total ordering in the set of
cuts R.

Solution. Let a, (3 be two arbitrary cuts. We have to prove that either a ~ (3 or

(3 ~ a.
If a and (3 are equal as sets, then by definition a ~ (3 and (3 ~ a. Let a =I- (3. We
have to prove that exactly one of the following two inclusions is true

a C (3 or (3 C a. (1.20)

Assume the contrary to 1.20. Then there exist two rational numbers a and b such
that a E (a \ (3) and b E ((3 \ a).
We have

a E (a \ (3) {::=} (a E a 1\ a rf- (3) => (Vx E (3) x < a (1.21 )

32 CHAPTER 1.

(see Example 1.60). Analogously

b E ((3 \ 0:) ~ (b E (3 1\ b 1. 0:) =? (Vy EO:) Y < b. (1.22)

From (1.21) we obtain that b E (3 implies b < a, while a E 0: implies a < b, hence
we have a contradiction.

Example 1.63.

a) If 0: and (3 are arbitrary cuts, then the set, := 0: Ell (3 defined by relation (1.15),
is also a cut.

b) Prove that the product of two positive cuts given by relation (1.17) is again a
positive cut.
Hence by relation (1.18), the product of two arbitrary cuts is also a cut.

Solution. We shall prove only part

a) We have to prove that the set, satisfies the properties (cl) - (c3) from Definition
1.58.

• Since and (3 are cuts, then (cl) implies that there exist two rational numbers
0:
a E and bE (3, hence the rational number c := a + b is in ,. On the other
0:
hand, again (cl) implies that there exist rational numbers a' 1. 0: and b' 1. (3,
hence the rational number c' := a' + b' is not in ,. Thus, is neither empty
nor equals to the whole of Q, which means that it satisfies (cl).

• Assume c E , and let x E Q, x < c. We have to prove that x E ,. From

c E , it follows that there exist a E 0: and b E (3, such that c = a + b. Put
d := c - x > O. Then it holds
d d d d
x = c - d = (a + b) - (- + -) = (a - -) + (b - -).
2 2 2 2

Now a - ~ is in 0: and b - ~ is in (3 by (c2), hence their sum x is in ,. Thus

, satisfies (c2).

• Finally, we have to prove that, has no maximum. Assume the contrary, i.e.,
that there exists a maximum Co in ,. Then by the definition of Ell given in (1.15)
there exist two rational numbers ao E 0: and bo E (3 such that Co = ao + boo
By (c3), the cuts 0: and (3 have no maximums, hence there exist al E 0: and
bI E (3 such that al > ao and bI > boo But then al + bI is an element from,
such that
al + bI > ao + bo = Co,

which means that Co is not the maximum of,. Hence, has no maximum, and
thus it satisfies (c3).
REAL NUMBERS 33

Exercise 1.64. Prove that the operations EEl and 129 satisfy the commutative law,

(Va, (3 E R) a EEl (3 = (3 EEl a, a 129 (3 = (3 129 a,

the associative law,

(Va, (3, , E R) (a EEl (3) EEl, = a EEl ((3 EEl ,), (a 129 (3) 129, = a 129 ((3 129 ,),

and also the distributive law,

(Va, (3" E R) a 129 ((3 EEl ,) = (a 129 (3) EEl (a 129 ,).

Example 1.65. Prove that

a) the zero cut O· is the additive identity for EEl;

b) the additive in verse for a E R for the operation EEl is the set 8a, where

8a:={-xEQlxE(Q\a), xf;a},

where a is the minimum of the set Q \ a, provided this minimum exists.

Solution. We shall prove only part

a) Let us put
al:= a EEl 0* = {a+xl x E a, x < O}.
By Example 1.63, al is a cut. We have to prove the set-equality

al = a, or equivalently al C a A a Cal'

• If y E aI, then by (1.15) there exist rational numbers a E a and x < 0 such
that y = a + x. Hence y < a, which by (c2) means that yEa. Thus al Ca .

• If a E a, then by (c3) there exists a' E a, such that a' > a. Hence

a = a + (-a' + a') = (a - a') + a' = a' + (a - a').

Since a' E a and (a - a') E 0', it follows that a E al' Thus a Cal-
We proved that a EEl O· = a. By Example 1.64, it also holds

0* EEl a = a EEl O· = a.

Remark. From Examples 1.63 and 1.65 and Exercise 1.64 it follows that (R, EEl) is
an Abelian group.

Example 1.66. Prove that

a) the one-cut 1* is the multiplicative identity for 129;

34 CHAPTER 1.

b) for given 0' E R+, the set 0'8 1* is the multiplicative inverse element of 0' for 0,
where
0'8 1*:= {x E QI x ~ O}U{~I x E (Q\a)' x =la,} (1.23)

if a is the minimum of Q \ 0', provided this minimum exists.

For 0' -< 0*, prove that its inverse is the cut 8(10'18 1*).

Solution. We shall prove only part b) for the case 0' >- 0*. First, we shall prove
that the set 0'8 1* is a cut.

• Since 0' is a cut, it follows from (c1) applied to 0' that 0'8 1* is nonempty and
is a proper subset of Q.

• Let x E 0'81* and assume that the rational number y satisfies y < x; we have
to prove that y E 0'8 1*. If y ~ 0, then there is nothing left to prove. So we can
assume that 0 < y < x, which implies 0 < ~x < ~.
y
Using the last inequality, it
holds
x E 0'8 1* => ~ E (Q \ 0') => ~ E (Q \ 0') => y E 0'81*.
X Y

• Finally, we have to prove that 0'81* has no maximal element. The set Q \ 0'
either has a minimum a > 0, or it has not.
In the first case, it holds by definition of 0'8 1* (see relation (1.23)) that the
positive rational number 1/ a is not in 0'81*. Thus 0'81* has then no maximum.
In the case when Q \ 0' has no minimum, the following holds

(Vx E (Q \ 0') (:lx' E (Q \ 0')) x > x' > 0) => ~x < ~.

x'

But -1IS
. .III 0'
81*
.
x'

Hence it follows that 0'81* is a cut.

U sing Example 1.63 it follows that for 0' >- 0* the set 0' 0 0'8 1* is also a cut.
We shall prove now that 0' 0 0'8 1 * = 1*. This is equivalent to the conjunction of
inclusions
(0' 0 0'8 1* C 1*) 1\ (1 * C 0' 00'8 1*).

• Let z E (0' 0 0'81*). If z ~ 0, then z E 1*. Hence we can suppose z > 0; by

definition there exist positive numbers x E 0' and y E 0'8 1* such that z = xy.
Hence

=> -1 => -1 > x =>

81*
yEa E (Q \ 0') z = xy < 1 => z E 1* .
y y
REAL NUMBERS 35

• Let now z E 1*. If z :::; 0, then z E a®a 81 ' by the definition of the last product.
So we can assume that 0 < z < 1. Since a is a cut, there exists a positive
rational number x E a such that ~ = y f!. a; otherwise the complement of a
z
. x
m Q would be empty. Put y:= - E Q \ {a}, then
z

~ E a(-l)' =} z = X· ~ =} z E a 129 a 81 '

y y

Exercise 1.67. Prove that (Q, 81, ®,:::S) is a totally ordered field.

Exercise 1.68. Prove the following properties of rational cuts:

a) if a is a cut, then r E Q is in a iff r* -< a;

b) if a and (3 are cuts, then there exists a rational cut r* such that a -< r* -< (3;

c) (Vr,p E Q) r < p -¢=::} r* -< pO;

d) (Vr,p, q E Q) r +p = q -¢=::} r* 81 p* = q*;

e) (Vr,p, q E Q) r· p = q -¢=::} r* 129 p' = q*.

Example 1.69. Prove that there exists a positive irrational cut (3, i. e., a cut with
the property that Q \ (3 has no minimum.

Solution. Let us put

(3 := {x E Q+I X2 < 2} U{x E QI x :::; O}.

(Of course, Q+ = {x E Q I x > O}.) That the set ;3 satisfies properties (cl) and
(c2) is trivial. For (c3), assume that (3 has a maximum Yo E (3. Then the rational
number YI := 2 Yo + 1 is such that YI > Yo, but still yi
< 2 (compare to Example
Yo + 2
1.42). Hence ;3 is a cut.
Finally, let us prove that ;3 is an irrational cut. Assume the contrary, i.e., that
there exists the smallest number Xo in Q \ ;3. But then the number Xl := 2 Xo +1
Xo +2
satisfies xi > 2, Xl > 0 and Xl < Xo. Hence Xl is also in Q \;3, but is smaller than
Xo, a contradiction.

Example 1.70. The Dedekind theorem.

Let A and B be two classes of cuts with the following properties.

(D!) a cut is in one and only one of these classes;

(D2) neither of these classes is empty;

(D3) for every cut a E A and every cut ;3 E B it holds a :::S ;3.
36 CHAPTER 1.

Then there exists a unique cut, such that

(Va E A) (V{3 E B) a ~, ~ {3. (1.24 )

Solution. Let us show the existence of the cut ,. The class A either has a maximum,
or it has not. In the first case, put c := max A. Then the sought cut is , := c*. Next
we assume that A has no maximum. Let us put

G = {x E QI x* E A}.

We shall prove that G is a cut (i.e., we shall check the properties (cl) - (c3) from
Definition 1.58) and, moreover, show that G satisfies the inequalities in (1.24). This
will then imply G = , is a cut.

• Since A is not empty, there exists a cut a E A. By assumption, A has no

maximum, hence there exists another cut al E A such that al >- a. By
Example 1.68 b), there exists a rational cut in R r* E A such that a ~ r* ~ al'
By the definition of G, r is in G.
Since B is also nonempty, there exists a cut {3 E B. Since {3 is a cut, there
exists a rational number p such that p rt {3. Hence

{3 -< p* =} p* E B =} p* rt A=} P rt G.

Thus the set G is neither empty nor does it equal the whole of Q.

• Let rEG and take some rational number rl < r. Then it holds

r* E A=} r; E A=} rl E G.

Thus (c2) holds for G.

• In order to show property (c3) for the set G, we have to prove that G has
no maximum. Let us assume the contrary, i.e., that there exists a maximum
m E Q of G. We claim that then m* is the maximum in A, of course for the
binary relation ~ . Once we prove this, we shall get a contradiction with our
assumption that the class A has no maximum. This will then mean that G
can not have a maximum, i.e., (c3) holds for G. In order to prove that m* is
the maximum of A, under the assumption that G has a maximum m, we shall
suppose that m* is not the maximum of A. Then there exists a cut a E A
such that a >- m*. By Example 1.68 b), there exists a rational cut m~ such
that m* -< m;' -< a. But then m can not be the maximum of G, since ml > m
and ml E G.

Thus it follows that G is a cut. Let us take any a E A. Since A has no maximum,
there exists another cut al E A such that al >- a. From Example 1.68 b) we find
a rational cut r* such that a -< r* -< al' By definition, the rational number r is in
G, hence G >- a. Let now {3 E B. Then either G ~ {3 (as we claim), or G >- {3. In
the latter case there exists a rational cut p* E B, such that {3 -< p* -< G. But then
REAL NUMBERS 37

pEG, hence p* E A, which contradicts to the assumption that the classes A and E
are disjoint.
So we obtained that the cut G has the property

('10: E A) (Vf3 E E) 0: ::S G ::S f3.

Hence we can put / := G.

Let us prove that this / is unique. Assume that there exist two cuts /1 and
/2 which satisfy relation (1.24) and let /1 --< /2. There exists a rational number r
such that /1 --< r* --< /2' But then /1 --< r* implies r* E E, while r* --< /2 implies
r* E A. Thus the classes A and E are not disjoint, contrary to the assumption.
The contradiction came from the assumption that there were two cuts which satisfy
(1.24), hence the cut / = G is the unique cut satisfying (1.24).
Remark. Dedekind's theorem is, in fact, axiom (R15) from Definition 1.1 given
at the beginning of Chapter 1. In the previous examples and exercises we either
proved or just left to the reader to prove the properties of R analogous to those
given by the first fourteen axioms (R1) - (R14) from Definition 1.1. With the last
example, we thus proved that the set of rational cuts R is isomorphic to the set of
real numbers R introduced in Definition 1.1.

Exercise 1.71. Prove th(J,t every nonempty set X C R, bounded from above, has
a supremum in R, using' Dedekind 's theorem, Example 1.70 (compare to Example
1.50).

Exercise 1.72. Starting from the set of rational numbers Q given by axioms (Rl)
- (R14), the construction of cuts in Q gave the set of cuts. The last turned out to
be isomorphic to the set R given by the axioms (Rl) - (R15).
Now, if one starts from the set of real numbers R given by axioms (Rl) - (R15),
and defines the cuts in R in the manner of Definition 1.58, does one get a set
nonisomorphic to R?

Answer. No. In fact, Dedekind's theorem shows that the cuts in R give again real
numbers, hence no new elements arise from the above construction in R.

1.3 The set R as a topological space

1.3.1 Basic notions
Definition 1.73. A neighborhood of a point Xo E R is any set U(xo) C R
which contains an open interval (xo - c, Xo + c) for some c > O.

Clearly, every neighborhood of Xo E R contains Xo, while every interval ( a, b) that

contains Xo is also its neighborhood.

Definition 1.74. A nonempty set A C R is an open set, if it is the neighborhood

of each of its points.
38 CHAPTER 1.

By definition, the empty set is also open.

Definition 1.75. A set A C R is a closed set, if its complement in R, the set

R \ A, is open.
Any "open interval" (a, b) is an open set, while any "closed interval" [a, b] is a closed
set in the sense of the last two definitions (prove that!)

Definition 1.76. Let A be a subset of the set of real numbers R.

a) A point Xo E R is an interior point of the set A if A is a neighborhood of Xo.

The interior of the set A, denoted by A 0, is the set of interior points of A.

b) A point Xo E R is a point of closure of the set A if in every neighborhood of

Xo there exists at least one point from A.

The closure of a set A, denoted by if, is the set of points of closure of A.

c) The point Xo E R is an accumulation point of the set A if in every neighbor-

hood of Xo there exists at least one point from A different from Xo. The set of
accumulation points of A will be denoted by A'.

d) The point Xo E A is an isolated point of the set A if there exists a neighborhood

of Xo that contains no other points from A.

e) The point Xo E R is a boundary point of the set A if in every neighborhood

of Xo there exists at least one point from A and at least one point from its
complement R \ A. The set of boundary points of a set A is denoted by GA.

A set is bounded if it is both bounded from below and from above (compare to
Definition 1.3).

Definition 1.77. A set K C R is compact if it is both bounded and closed.

A set is infinite if there exists a bijection 4 that maps it onto one of its proper
subsets. The number sets N, Z, Q and R are all infinite sets. The sets that are not
infinite are finite. An infinite set X is countable if there exists a bijection between
X and the set of natural numbers N. The sets N, Z and Q are countable, however
the set R is not.

Theorem 1.78. The Bolzano-Weierstrass theorem.

Every infinite and bounded set A c R has at least one accumulation point in R
(which is not necessarily in A).

Definition 1.79. A collection of sets {Bil i E I} is a covering of a set A c R

if for every x E A there exists an index i from the index set I such that x E B i . If,
additionally, all the sets Bi are open, then the collection {Bi liE I} is called open
covering.
4See Subsection 2.1.1.
REAL NUMBERS 39

Definition 1.80. A set A C R has the Heine-Borel property if every open

covering of A has a finite subcovering.

In other words, if {Oi liE I} is a collection of open sets whose union covers A, then
the Heine-Borel property of A implies the existence of a finite set 11 C I such that

AC U Oi.
iEh

Remark. In the Hausdorff topological spaces, (see Example 1.89) the Heine-Borel
property is often used for the definition of compactness. In Example 1.96 we shall
prove that a set in R is compact (= bounded and closed) iff it has the Heine-Borel
property.5

1.3.2 Examples and exercises

Example 1.81. Find the interior, the closure, all accumulation points, isolated
points and boundary points for the following sets.

Ill}.
a) A=[O,l]; b) B = { 0, 1, "2' 4"' ... , 2n ' • .. ,

c) C= {;n In EN} ; d) D = N = {1,2, ... }.

Answers.
a) AO = (0,1), A = [0,1], A' = [0,1]' no isolated points of A, aA = {O, I}.
b) W = 0, B = B, B' = {O}, all points from B \ {O} are isolated, aB = B.
c) Co = 0, C= CU{O}, C' = {O}, all points from C are isolated, ac = C.
d) DO = 0, D = D, D' = 0, all points from D are isolated, aD = D.

Example 1.82. Show that if 0: E R is an accumulation point of a set A C R, then

in every neighborhood of 0: there exist infinitely many points from A.

Solution. Let U( 0:) be a neighborhood of 0:, then by Definition 1.73 there exists
an interval (0: - C1, 0: + C1) c U(o:). This interval is also a neighborhood of 0:. Now
since 0: is an accumulation point of A, by Definition 1.76 c), there exists an element
0:1 i- 0: such that
0:1 E (0:-Cl,O:+C1)nA.
Put C2 := (10: - 0:11)/2. Then in the interval (0: - C2,0: + C2) there exists an
element 0:2 E A which is different both from 0: and 0:1. Continuing this procedure
ad infinitum, we get infinitely many points from A in the interval (0: - C1, 0: + cd,
hence also in the given neighborhood U(o:) of the point 0:.
5However, there exist topological spaces in which there exist sets that are bounded and closed,
but still do not satisfy the Heine-Borel property!
40 CHAPTER 1.

Example 1.83. Prove that every accumulation point of a set A c R is also in the
closure of A. Is the opposite statement true'?

Solution. Let 0' be an accumulation point of the set A. Then in every neighborhood
U(O') of 0' there exists a point f3 E A, f3f. 0'. By Definition 1.76 b), 0' is then in the
closure of A.
The opposite statement is not true, as will be seen from the following example.
Let A := (0-,1) U{2}. Then the closure of A is A = [0,1] U{2}, while the set of
accumulation points of A is A' = [0,1]. Namely, the point 2 is not an accumulation
point of A, but rather its isolated point, since in the interval (2 - ~,2 +~) there are
no points from A different from 2 itself.

Example 1.84. Prove that for every set A c R it holds A = AUA', i.e., the
closure of a set is equal to the union of that set and its accumulation points.

Solution. By Definition 1. 76 b) it holds A c A, while by Example 1.83 A' c A,

hence A U A' c A. For the opposite inclusion, assume x E A. Then either x E A (in
which case there is nothing left to prove) or x ~ A. In the last case, we claim that
x is an accumulation point of A. Otherwise, there would exist a neighborhood of x
disjoint from A; however, then x could not be in the closure of A.

Example 1.85. Prove that the closure of a set is closed.

Solution. Let A be the closure of the set A c R. We shall prove that the set
B := R \ A is open. Let f3 E B. By Example 1.84,

f3 ~ A -¢=} (f3 ~ A 1\ f3 ~ A').

In view of Definition 1.76 c), it follows that there exists a neighborhood U(f3) of f3
disjoint from A. By Definition 1.73, U(f3) contains an open interval (f3 - c, f3 + c),
which thus contains no accumulation points of A. Hence, the last interval is contained
in B, which means that B is the neighborhood of each of its points. By Definition
1. 74, the set B is then open, hence A = R \ B is closed.

Exercise 1.86. A set is closed iff it contains all of its accumulation points.

Example 1.87. The closure A of a set A c R is the smallest closed set that con-
tains A.

Solution. In Example 1.85 we proved that A is closed. Assume B c R is a closed

set that contains A. We have to prove that A c B. In the contrary, there would
exist an element 0' E A \ B. From Example 1.84 it would then follow that 0' E A \ A
(since B contains A). Hence 0' is an accumulation point of A, which contradicts to
Exercise 1.86.
Remark. This example justifies the term "closure".

Exercise 1.88. The open sets in R satisfy the following three properties.

(tsl) The sets Rand 0 are open.

REAL NUMBERS 41

(ts2) An arbitrary union of open sets is open.

(ts3) A finite intersection of open sets is open.

Remark. A family T of sets contained in a set X, T : = {Oil i E I}, where I is

a set of indices, is called topology on X, if it satisfies the properties (ts1) - (ts3)
on X, and the set X is called topological space. (Of course, in (ts 1) one has to
replace "R" with "X".) The elements of T are then called open sets.
Thus from Exercise 1.88 it follows that if one defines open sets in R as in Defi-
nition 1.74, the set R becomes a topological space.

Example 1.89. Prove that for every two real points x and y, x -=J y, there exist
two disjoint neighborhoods of these points.
. . r r r r
SolutIon. Let r := Ix - yl > 0, then the mtervals (x - 3' x + 3) and (y - 3' y + 3)
are disjoint neighborhoods of x and y respectively.
Remark. A topological space with this property is called Hausdorff space. Thus
we proved that the set R with the topology given with the open sets from Definition
1. 74, is a Hausdorff space.

Exercise 1.90. Check whether the sets from Example 1.Bl are compact.

Answers.

a) The set A is compact, since it is bounded and closed.

b) Firstly, the set B is bounded, since B C [0,1]. The only accamulation point of
B is 0, which belongs to B, hence it is also closed. Thus B is compact.

c) The set C is bounded, but does not contain its accumulation point 0. Hence C
is not closed, which implies that it is not compact.

d) The set of natural numbers N has no accumulation points, hence N N.

However, it is not bounded, hence N is not compact.

Example 1.91. Show that every nonempty subset S of a compact set K C R has
an infimum and a supremum, which both belong to K.

Solution. By Definition 1.77 the set K is bounded, hence so is its subset S. By

Example 1.50, there exist real numbers a and f3 such that inf S = a and sup S = f3.
We have to prove that a, f3 E K.
We shall only prove that a E K. So assume a ¢: K. Since K is closed, the
°
set R \ K is open. This means that there exists an c > such that the interval
(a - c,a + c) is disjoint with K. Since S C K, it follows (a - c,a + c)nS = 0.
But then there is no element Sl in S such that Sl < a + c, which contradicts the
assumption a = inf S.

Exercise 1.92. Prove that

42 CHAPTER 1.

a) a finite set has no accumulation points;

b) every finite set is compact.

Exercise 1.93. Prove that a closed subset of a compact set is itself compact.

Example 1.94. A set K c R is compact if and only if its every infinite subset has
an accumulation point which belongs to K.

Solution. In view of Exercise 1.92, it is enough to observe the case when K is

infinite .

• The condition is necessary. Assume that K is compact, i.e., that (by

Definition 1.77) it is bounded and closed. Let S be an infinite subset of K.
The set S is bounded, being a subset of a bounded set. By the Bolzano-
Weierstrass theorem (Theorem 1.78), the set S has at least one accumulation
point 0: E R. The assumption S C K, implies that 0: is also an accumulation
point for K. Now K is closed, hence by Example 1.86 it contains all of its
accumulation points. This means that 0: E K .

• The condition is sufficient. Assume that every infinite subset of K has

an accumulation point which belongs to K. We have to prove that K is both
closed and bounded.
Let us prove first that K contains all its accumulation points, which is, by
Example 1.86, equivalent with the statement that K is closed. Let;3 be an
accumulation point of K; ~e have to prove that ;3 is in K. By Definition 1. 76 c),
for every c > 0 there exists at least one point al ~ ;3 in the set (;3-c, ;3+c) n K.
. d1 d1 n .
Put d l := 1;3 - all· Next, m the set (;3 - 2,;3 + 2) K there eXIsts a pomt
.
a2 ~ ;3. Continuing this procedure ad infinitum, we construct an infinite set
S = {aI, a2, ... } contained in K. By construction, ;3 is an accumulation point
of S, hence ;3 E K.
Finally, we have to prove that K is bounded. Assume that K is unbounded.
We shall construct an infinite subset S of K which has no accumulation points
in K. Let al be an arbitrary element from K. By the Archimedes theorem,
there exists a natural number nl such that lall < nl (see Theorem 1.12).
Since K is unbounded, there exists an element a2 E K such that nl < la21.
Choose n2 E N such that n2 > la21; next, choose a3 > n2, etc. In this way
we constructed an infinite set S = {aI, a2, ... } contained in K, which has no
accumulation points in K. This contradicts the assumption that K contains
all the accumulation points of its infinite subsets.

Exercise 1.95. Prove that every infinite sequence of elements from a compact set
has a convergent subsequence (see Definition 3.44).

Example 1.96. Prove that a necessary and sufficient condition for the compactness
of a set K C R is that it has the Heine-Borel property.
REAL NUMBERS 43

Solution .

• The condition is necessary. Assume that K is a compact set in R, i.e.,

that K is bounded and closed, but it does not satisfy the Heine-Borel property.
This means that there exists an open covering {Oil i E I} of K which has no
finite sub covering. A compact set is bounded, thus there exists an interval
lab bl ] containing K. Let us divide this interval on two equal parts and denote
by [a2' b2] that half of lab bl ] which has the property that there is no finite
sub covering of the set [a2' b2]nK. Continuing this procedure, we come to
an infinite sequence of closed intervals [an, bnL n E N, contained always in
the previous one and having the property that their lengths tend to zero as
n ----+ 00. By Example 1.48, there exists a unique real number 0: contained in
each interval [an' bn].
We shall prove now that 0: is an accumulation point of K, i.e., that in every
neighborhood of 0: there exist infinitely many elements from K. In fact, if U( 0:)
is a neighborhood of 0:, then by Definition 1. 73 there exists an 6 > 0 such that
(0: - 6,0: + 6) C U(o:). Since the lengths of intervals [an, bn ], n E N, tend to
zero, there exists an integer no such that for n > no it holds

[an,bn] C (0:-6,0:+6) C U(o:).

By construction, for n > no every interval [an, bn] has infinitely many elements
from K, thus it follows that 0: is an accumulation point of K. A closed set
contains its accumulation points, hence 0: E K.
The collection of sets {Oil i E I} was assumed to be an open covering of K,
hence there exists an index i l E I such that 0: E Oil' Moreover, there exists an
integer nl with the property that for every n > nl it holds [an, bn] C Oil' So
we covered all these intervals with only one open set Oil' though we assumed
that no finite sub covering could cover any of them .

• The condition is sufficient. Assume that K has the Heine-Borel property;

we have to prove that K is compact. Let us assume the contrary, namely that
K is not compact. By Example 1.94, there exists an infinite set S C K with
no accumulation point in K. Hence for every point x E (K \ S) there exists
an open set Ox such that S n Ox = 0. For every yES there is an open set
n
Oy from the open covering such that Oy S = {y}; otherwise, y would be an
accumulation point of S belonging to K.
Clearly, it holds
K C ( U Ox)U(U Oy),
xEK\S yES

but since the set S is infinite, it is impossible to find a finite sub covering of
K, which is a contradiction with the Heine-Borel property.

Example 1.97. Does the set A = {;'I n EN} have the Heine-Borel property?
44 CHAPTER 1.

Solution. We shall show in two ways that A does not have the Heine-Borel prop-
erty.

• First method. The number 0 is the only accumulation (cluster) point of the
set A. Thus A is not a closed set implying that that it is neither compact. By
Example 1.96 this is equivalent with the statement that A has no Heine-Borel
property.

• Second method. Let us observe the open intervals

On = (~n - ~,
n n
~ + ~),
n
n E N.

Clearly A c U On, but the family {On I n E N}, has no finite sub covering of
nEN
the set A. This follows from the following relations:

1 n2 - 1 n2 - 1 n2 - 1 1 1
--= = <--=---
n+l (n+l)(n 2 -1) n 3 +n 2 - n - l n3 n n3 '

for n = 2,3, ....

Exercise 1.98. Let uS) define a function d : R X R -+ [0, +=) called distance by
the formula
d(x,y) = Ix - yl, X,y E.R.
Prove the following properties of the distance d.

(Ml) (Vx,y E R) d(x,y) > 0 if x i- y and d(x,y) = 0 if x = y;

(M2) (Vx,y E R) d(x,y) = d(y,x);

(M3) (Vx,y,z E R) d(x,z)::; d(x,y) + d(y,z).
Remark. A set X with a function d : X X X -+ [0, +=) satisfying the upper
conditions (Ml) - (M3) (with R replaced by X) is called a metric space.
Chapter 2

Functions

2.1 Real functions of one real variable

2.1.1 Basic notions
Let A and B be two nonempty sets. By definition, a relation f from A into B is a
subset of the direct product A X B.
Definition 2.1. A relation f is a function which maps the set A into the set B if
the following two conditions hold:
(ft) for every x E A there exists an element y E B such that the pair (x,y) is in f;

(£2) if the pairs (x, Yd and (x, Y2) are in f, then necessarily Yl = Y2·
Then we write f : A -+ B and say that "f maps A into B". If the pair (x, y) E f,
we shall write Y = f (x). The set A is then called the domain, while the set B is
called the codomain of the function f. The element x from the domain A is called
the independent variable, while the element y from the codomain B is called the
dependent variable.
The set
f(A) = {y E BI (3x E A) f(x) = y}
is called the range of the function f. By Definition 2.1, f(A) C B and, in general,
the range is a proper subset of the codomain.
Clearly, a function f is determined with the triple (A, B, f). This also means
that two functions II : Al -+ Bl and h : A2 -+ B2 are equal if and only if their
domains are equal, i.e., Al = A 2, their codomains are equal, i.e., Bl = B 2, and, of
course, it holds fl (x) = f2 (x) for all x E A.
A function f : A -+ B is a one-to-one function or injective if for every pair
Xl and X2 from the domain A it holds

Xl f- X2 =} f(Xl) f- f(X2).

A function f : A -+ B is surjective if for every y E B there exists X E A such

that f(x) = y. Clearly, for a surjective function its codomain and its range are the

45
46 CHAPTER 2.

same, i.e., f(A) = B. It is usual to say for a surjective function f : A -+ B that it

maps A onto B.
A function f : A -+ B is a bijection if it is both a one-to-one function and a
surjection.
Suppose the function f : A -+ B is a bijection. Then, for every y E B there
exists a unique element x E A such that y = f(x). Now, the relation from B to A
given by
r l := ((y,x) E B x AI y = f(x)}
is a function on B which will be called the inverse function for f. The inverse
function is also a bijection and it holds

rl(y) =x {::=} f(x) =y for every pair (x,y) E A x B.

Let two functions f A -+ Band 9 B -+ C be given. Then the function

9 0 f :A -+ C given by

go f(x) := g(f(x)) for every x E A

is called the composite function of the functions 9 and f.

Let us remark that if a function f : A -+ B is a bijection and f- l : B -+ A its
inverse function, then

rIO f(x) = x for every x E A and f 0 rl(y) = y for every y E B.

In this book, we shall observe only those functions whose domains and codomains
are some subsets of the set of real numbers R. Such functions are called real func-
tions of the real variable; shortly, we shall call them simply functions. Most
often, a function is given with the analytical expression (formula) y = f(x), where
the sets A and B, i.e., the domain and the codomain of the function f, are not ex-
plicitly given. If not stated otherwise, in that case we shall always assume that for
such a function the codomain is the whole set of real numbers R, while the domain
is the largest set for which the given formula has sense. This domain will be called
the natural domain of the function f.
The graph G j of a function F: A -+ B is the subset of the set R2 = R x R
given by
G j = {(x, f(x))1 x E A}.
Quite often, our main task will be to draw the graph of a function given with some
formula. Some of the geometric properties of the graph are defined next.
A point Xo from the domain of a function f is a zero (or: root) of a function f if
it holds f(xo) = O. Geometrically, this means that the point (xo,O) is the common
point of the graph of f and the x-axis.
A function f : A -+ B is monotonically increasing (resp. monotonically
decreasing) on the set X c A if for every pair of elements Xl and X2 from the set
X it holds

Xl < X2 ::::} f(xd < f(X2) (resp. Xl < X2 ::::} f(xd > f(X2))'
FUNCTIONS 47

A function f : A - t B is monotonically nondecreasing (resp. monotonically

nonincreasing) on the set X C A if for every pair of elements Xl and X2 from the
set X it holds

Xl < X2 =:} f(xd ~ f(X2) (resp. Xl < X2 =:} f(XI) 2: f(X2)).

The importance of the monotonicity of a function is demonstrated in the follow-
ing theorem.

Theorem 2.2. If the function f : A - t B is a surjection and is either monotonically

increasing or monotonically decreasing on the whole domain A, then f is a bijection.

A function f: A - t B has a local maximum (resp. local minimum) in the

point Xo E A if there exists a number c > 0 such that

(VxE(xo-c,xo+c)nA) f(x)~f(xo) (resp. f(x)2:f(xo)).

A function f : A -t B has a global maximum (resp. global minimum) In

the point Xo E A if

(Vx E A) f(x) ~ f(xo) (resp. f(x) 2: f(xo)).

In the following, we usually omit the terms local and global and speak just about
. .
mInImUm or maXImum.
A function f : A - t B is bounded on the set X c A if there exists a number
C > 0 such that
(Vx EX) If(x)1 ~ C.
Geometrically, this means that the graph of the function f over the set X is settled
between two horizontal lines, namely y = C and y = -C.
The set X c R is symmetric (to the origin) if for every point X E X it holds
that the point -x is also in X. (Notice that for X -I- 0 the points X and -x are
symmetric to the origin.)
Suppose that the domain A of a function f : A - t B is symmetric. Then f is
an
even function if for every X E A it holds f( -x) = f(x);

odd function iffor every x E A it holds f(-x) = -f(x).

Geometrically, the graph of an even function is symmetric to the y-axis, while the
graph of an odd function is symmetric to the origin.
A number T -I- 0 is called the period of the function f : A - t B if for all x E A
the points x + T and x - T are also in A and it holds

(Vx E A) f(x + T) = f(x).

The smallest positive period, if it exists, is called the basic period of the func-
tion f. Clearly, if we know the basic period T of a function, then it is enough to
draw its graph on any set X C A of the length T.
48 CHAPTER 2.

A function f : A -+ B is called concave upward on the interval (a, b) C A if

for every pair Xl, X2 E ( a, b) and for every a E (0, 1) it holds

f(O'xI + (1 - O')x2) ::; O'f(xd + (1 - O')f(x2)'

Geometrically, if a function f is concave upward on the interval ( a, b), then the

segment connecting any two points on its graph is above the graph.
A function f : A -+ B is called concave downward on the interval (a, b) c A
if for every pair Xl, X2 E (a, b) and for every a E (0,1) it holds

f(O'xI + (1 - O')x2) 2: O'f(xd + (1 - O')f(x2)'

Geometrically, if a function f is concave downward on the interval (a, b), then the
segment connecting any two points on its graph is under the graph. I
The basic elementary functions are the following ones:

• power function: y = x S , X E R, for a fixed sEN

(in particular, for s = 0 this function is a constant one);

• exponential function: y = aX, X E R, for a > 0 and a =f. 1;

• logarithmic function: y = loga x, X E (0, +00), for a > 0 and a =f. 1;

• trigonometric functions:

y = smx, x E R; y = cos x, x E R;

y = tan x, x ER \ {(2k; 1)?l" IkE Z} ; y = cot x, x E R \ {hi k E Z}.

• the inverse functions (with the possible restriction of the domains) of the
up to now mentioned basic elementary functions.
In particular, such are the so-called inverse trigonometric functions:

f( x) = arcsin X', x E [-11"/2,11"/2]; f(x) = arccos x, x E [0,11"];

f(x) = arctan x, x E R; f(x) = arccotx, x E R.

In particular, if a = e (see Example 3.33), for the exponential and logarithmic

function we shall write

y = eX = exp(x), x E R, and y =lnx, x> O.

An elementary function is obtained by finitely many applications of the alge-

braic operations: addition, subtraction, multiplication and division, as well as the
operations of composition of the basic elementary functions.
I Some authors use the words "convex" instead of "concave upward" and "concave" instead of
"concave downward" .
FUNCTIONS 49

2.1.2 Examples and exercises

Example 2.3. Let the sets A and B be given by A = {I, 2, ... , n} and B = {I, 2, ... , m}
for some natural numbers nand m.

a) Determine the number of all functions f : A -+ B, i.e., that map the set A into
the set B.

b) What is the relation between nand m if the function f : A -+ B is a bijection?

Determine then the number of all bijections from A onto B.

Solutions.

a) The number of all functions f : A -+ B is equal to the number of all variations

with repetition of m elements of the class n, i.e., it is mn.

b) Clearly, the function f : A -+ B can be a bijection only if n = m. The number of

all such bijections is equal to the number of all permutations without repetition
of n elements, i.e., it is n! := n· (n - 1)···2·1.

Example 2.4. Assume f : A -+ B is a function. Prove that for arbitrary sets

X, YeA the following holds.

a) X C Y =? f(X) C f(Y) (is the opposite implication always true?);

b) f(X U Y) = f(X) U f(Y);

c) f(X n Y)C f(X) n f(Y) (compare with b)).

Solutions.

a) If the element y is in f(X), then there exists an element x E X such that

f(x) = y. Since Xc Y, it follows that x E Y which implies f(x) E f(Y), i.e.,
y E f(Y).
That the opposite implication is in general false, shows the following coun-
terexample.
Let the function f : Z -+ Z be given by f(x) = x\ and let X = {I, -I}
and Y = {1,2}. Then f(X) = {I} C {1,16} = f(Y), but it is not true that
XcY.

b) Let y E f(X U Y). Then there exists an element x E XU Y, such that y = f(x).
This implies x E X or x E Y, hence y E f(X) or y E J(Y). SO we obtain that
y E J(X) U J(Y). Thus we proved the inclusion J(X U Y) c J(X) U J(Y).
The opposite inclusion J(X) U J(Y) c J(X U Y) is proved in an analogous
way (do it!).
50 CHAPTER 2.

c) If y E J(Xny), then there exists an element x E Xny such that f(x) = y.

From x E X n Y it follows that both x E X and x E Y, hence y E f(X) and
y E J(Y). This means that y E f(X) n f(Y).
The following counterexample shows that in c) the equality does not hold
always.

Let A = B = R, and let J(x) = x 2 , X E R, and take X = {-1,2}, Y = {1,2}.

Then J(X) = J(Y) = {1,4}. But then f(X n Y) = f( {2}) = {4}, while
J(X) n J(Y) = {1,4}, which shows that the sets f(X n Y) and J(X) n f(Y)
are different.

Example 2.5. Let f : A --t Rand 9 : B --t R be two functions. Determine whether
the functions f and 9 given below are equal.

a) f(x) = #, A = R, g(x) = x3 , B = R;
b) f(x) = #, A = [0,+00), g( x) = x3 , B = [0, +00 );

2x
c) f(x) = -,
x
A = R \ {O}, g(x) = 2, B = R \ {o};

2x
d) f(x) = -,
x
A =R \ {O}, g( x) = 2, B = R;

e) f(x) = sin 2 x + cos 2 x, A = R, g(x) = 1, B = R;

f) f(x)=lnx 2 , A=(O,+oo), g(x) = 2In(x), B = (0, +00);
g) f(x)=lnx 2 , A=R\{O}, g(x) = 2In(x), B = (0,+00).

Solutions.

b), c), e) and f) The functions f and 9 are equal.

a) The functions f and 9 are not equal, since f( -1) = 1, while g( -1) = -1.
(However, the restriction of their domains from the whole R to the interval
[0, +00) makes them equal (see b).)

d) and g) The functions f and g are not equal, since their domains are different.

Example 2.6. Determine the largest set A C R such that the following analytical
expressions have sense. (We call then A the natural domain of the function given
with that formula.)
FUNCTIONS 51

a) f(x) = (x - 3) - ;
2-x ~ b) f(x) = x· vcos Vi;

c) f(x) = VSin(x 2); d) f(x) = In 3 (sin(;));

e) f(x) = arccos ( ~) ;
l+x
f) f (x) = In (arcsin G~ ~) );
g) f(x) = In(x + 4); h) f(x)=ln(sinv'-x2).
~
Solutions.

a) First, the given formula has no sense for x = 2, since for that value of x the
denominator becomes zero. Further on, since the natural domain of the square
root function g( t) = v't is the interval [0, +00 ), (or, equivalently, if t 2: 0), the
given formula has sense if and only if

x+22:0 ~ (x+2)(2-x) 2:0, (2.1 )

2-x (2 - X)2
provided that x i- 2.
Th~ quadratic function h(s) = (s + 2)(2 - s), s E R, has nonnegative values
for s E [-2,2] (draw the graph of the function h - it is a parabola!). Applying
this to equation (2.1) gives that A C [-2,2]; since we have to exclude the
point x = 2, we finally have A = [-2,2).
The other method of finding the natural domain A of the function f is to use
the equivalence

2x _+ x2 2: 0 ~ ( (x + 22: 0 1\ 2 - x> 0) V (x + 2 :S 0 1\ 2 - x < 0) ) ,

which, of course, gives the same result (check that!).

b) The natural domain of the function g( t) = v't is the set {t E RI t 2: O}, which
means that A C [0, +00). Further on,

coss2:0 ~ sE U [(4k-1)~,(4k+1)~],
kEZ 2 2
or, equivalently,
Jr Jr
(4k-1)"2:Ss:S(4k+1)"2' kEZ.
Using the condition A C [0, +00) and putting in the last relation s = Vi, x 2:
0, we obtain that if x 2: 0, then cos( Vi) 2: 0 holds if and only if

Jr2 Jr2 Jr2

0<
- X
-<-,
4 or (4k - 1)24 :S x :S (4k + 1)24
52 CHAPTER 2.

for some natural number k. This means that the natural domain of f is the
set

A = [0, :2] U C~N [(4k _1?:2, (4k + 1)2:2]) .

c) The function g(t) = sint is nonnegative iff 2k7r::; t::; (2k + 1)11", k E Z, hence

sin(x 2 ) 2: 0 ~ 2k7r::; x 2 ::; (2k + 1)11", for kENo.

Since for every x E R it holds # = lxi, we obtain finally

A = {x E RI (:lk E No) V2k7r ::; Ixl ::; V(2k + 1)11" }

= (u
kENo
[(V2k7r,V(2k+1)1I"J) u(U kENo
[-V(2k+1)1I",-V2k7r]).

d) The natural domain of the logarithmic function g(t) = In t is the open interval
(0, +00), hence the natural domain A of the function f will be the set of all
x E R such that sin( ~) > o. The last inequality is true iff there exists an
x
integer k such that
11"
2k7r < - < (2k + 1)11".
x
Solving these inequalities by x gives three cases.

1
• If k = 0, then 0 < - < 1, hence x E (1,+00);
x

• if k E Z and k > 0 (i.e., kEN), then x E Ck ~ l' 21k) ;

• if k E Z and k < 0, then x E (2k ~ l' 21k) .

So we obtained that

-
A _ ( U (2k+1'2k
_1 ~)) U ( U (2k+1'2k
_ 1 1 )) U(l,+oo).
kEZ,k<O kEN

e) The formula g(t) = arccost has sense iff t E [-1,1]'

hence the natural domain A
of the given function f is determined by the inequality

2x 2 2
-1 -< -
1 +-
x 2 -< 1 ~ -1 - x -< 2x -< 1 + x ,

which is true for every x E R. So we obtained that A = R.

FUNCTIONS 53

f) The formula g(u) = In u has sense for u > 0, which means that we must have

arcsin-- >
x+2
5-x
° ~ --
x+2
5-x
>° ~ x E (-2,5).

As in e), we must also have

Ix+21~1 x+2
I.e., -1<--<l.
5-x -5-x-
Thus we have
x+2
1--1
5-x
~ 1 ~ x E (-00,3/2].

Hence, the natural domain of f is the interval A = ( -2,~] .

g) The formula g( u) = In( u + 4) has sense for u E (-4, +00), while the inequality
°
Ixl - x > is possible if and only if x < 0. Hence A = (-4,0).
h) The natural domain of this function is the empty set.

Example 2.7. The function f is defined on the closed interval [0,1]. Determine the
set A on which the following composite functions can certainly be defined.

a) h(x) = f(x 4 ); b) h(x) = f(sinx);

c) h(x) = f(x + 3); d) h(x) = f(lnx).

Solutions.

a) If the range ofthe function g( x) =

x4 is the interval [0,1], then the interval [-1,1]
is the (largest possible) domain of g. Hence, the domain A of the composite
function h = fog is the interval [-1, 1].

b) The largest set for which sin( x) E [0,1] is the union U [2h, (2k + 1)11-].
kEZ

c) The range of the function g(x) = 3 + x is [0,1]. Hence

x E [0,1] ~ x E [-3,-2] =: A.

d) lnx E [0,1] iff x E [1,e].

Example 2.8. Let the function f : (0, +00) -+ R be given with the formula

f(D=x+~.
Find the formula for f.
54 CHAPTER 2.

Solution. Since we have

f (~) = x + Vf+X2 = x (1 + J:2 + 1) = l (1 + J(~r + 1) ,

x
it follows that
f(X)=~(l+Vf+X2), x>O.

Example 2.9. Let the function f : R \ {-I} --+ R be given with the formula
x
f(x) = - - . Find the function fn, n E N, where
l+x
fl = f, f2 =f 0 fl and fn =f 0 fn-l for n = 2,3, ....
Determine also the natural domains of these composite functions.

Solution. The range of the function f is the set R \ { -I}. Then it holds
x
x
h(x) =f h(x) = f(f(x)) = f (_X_) = ~
1 + 2x
0
l+x 1+--
l+x
Clearly, the natural domain of the last formula is the set R \ {-1/2}, though the
definition of f2 reduces its domain to the set R \ {-I, -1/2}.
Let us prove by mathematical induction that for n = 2,3, .. , it holds

fn(x) = - -x , x ER \{ II
-1'--2' }.
... ,--
1 + nx n
We proved already this formula for n = 2. Suppose it holds for n k, for some
natural number k > 1. Then

fk+l(X) (f 0 fk)(X) = f(fk(X)) = f C: kJ

x
1 + kx x
1 + _x_ 1 + (k + l)x
1 + kx
Ill}
for x E R \ { -1, - 2' ... , - k' - k + 1 .

Example 2.10. The function f : R --+ R is given with the formula

x
f(x) = J1 + x2'
Find the formula that gives the functions fn defined on R by

fn = f 0 f 0 ' · · 0 f, n = 2,3, ....

'-..-'
n
FUNCTIONS 55

Solution. Let us calculate fz first:

x

f,(x) ~ f(J(x)) ~ (II: x,) ~ 11+ (b)'

Vf+X2 x
VI + 2x 2

Using the mathematical induction, the reader should check that for every n = 2,3, .. ,
it holds
x
fn(x) = f(Jn-l(X)) = I X E R.
1 + nx
2'

Example 2.11. Find the function f given by the following formula.

1
a) f(x-2)= x+3' x?,=-3; b) f (~) = x4 + 1;

c) f (x + ~) = x2+ :2' x ?'= 0; d) f(x 2) = 1- x 3 ,x 2': O.

Solutions.

a) Putting t := x - 2 we get x = t + 2. Clearly if x E R \ {-3}, then t E R \ {-5}.

1
Hence f(t) = - - for t ?'= -5.
t+5
1 + X4
b) f(x) = -4-' X
x
?'= o.
c) Since X2+~ = (x+.!. )2_2, we shall use the change of variable t:= x+.!., x?'= o.
x x x
So we obtain
f(t)=t 2 -2, t?,=O.
(Of course, the last formula is defined for t = 0; however, we have to omit this
value because of the definition of the function f.)

d) f(x) = 1 - X 3/ 2 , X 2': o.
Example 2.12. Show that the function f : R -+ R given by f(x) = g(lxl), x E R,
is an even function for any function g : R -+ R.

Solution. For every x E R it holds

f( -x) = g(l- xl) = g(lxl) = f(x),

which proves that f is an even function.

Example 2.13. Prove that every function f whose domain A is a symmetric set
can be written as a sum of an even and an odd function.
56 CHAPTER 2.

Solution. The function f can be written as the sum

1 1
f(x) = 2(f(x) + f( -x)) + 2(f(x) - f( -x)), x E A.

1
If we put fl(X) := 2(f(x) + f( -x)), x E A, then

1
fl(-X) = 2(f(-x) + f(-(-x))) = fl(X) for every x E A,

i.e., h is an even function.

Analogously we show that the function h(x) := ~(f(x) - f( -x)), x E A, is odd.

Example 2.14. Prove that

a) the sum of two even (resp. of two odd) functions is an even (resp. an odd)
function,-

b) both the product of two even or of two odd functions is an even function,-
c) the product of an even and an odd function is an odd function.
Solutions.
a) Suppose f and 9 are odd functions, both defined on a symmetric set A c R.
Then for their sum f + 9 it holds

(f+g)(-x) f( -x) + g( -x) = - f(x) + (-g(x))

-(f(x) + g(x)) = -((f + g) (x)),
for every x E A.

c) Suppose f is an even and 9 an odd function, both defined on a symmetric set

A c R. Then for their product f 9 it holds

(fg)( -x) = f( -x) . g( -x) = f(x) . (-g(x)) = - f(x)g(x) = -(fg)(x),

for every x E A.

Remark. One must distinguish the product fg from the composition fog of two
functions f and g.

Example 2.15. If the number T > 0 is a period of a periodic function f : A --+ R,

then the number kT, k E Z \ {O}, is also a period of f.

Solution. We shall prove this statement first for positive integers k, using the
principle of the mathematical induction. By assumption, the statement is true for
k = 1. If it is true for kEN, then

f(x + (k + l)T) = f((x + kT) + T) = f(x + kT) = f(x), x E A.

FUNCTIONS 57

Hence by the principle of the mathematical induction, f is a periodic function with

periods kT, kEN.
Using this, we show that -kT, kEN, is also a period of f :

f(x + (-kT)) = f(x - kT) = f((x - kT) +kT) = f(x), x E A.

Thus every number of the form kT, k E Z \ {O}, is a period of f.

Example 2.16. Show that if the function f : R -t R is periodic with the period T,
then the function 9 : R -tR given by g( x) = f( ax + (3), x E R, is periodic with the
period T/a, provided that a -=I- o.
Solution. For every x E R it holds

g(x+~) =f(a(x+~) +(3) =f((ax+(3)+T) = f(ax + (3) =g(x).

Example 2.17. Check whether the following functions are periodic; if yes, find their
basic periods T, if any.

a) f(x) = sin 2 (x), x E R; b) f(x) = sin(x 2 ), x E R;

c) f(x) = sin lxi, x E R; d) f(x) = cos lxi, x E R.

Solutions.

a) It holds that the sin 2 x = ~(1 - cos(2x)), x E R. From Example 2.16 it follows
that the function g(x) = cos(2x), x E R, is periodic with the basic period
27r /2 = 7r. Hence, the function f is also periodic with the basic period 7r.

b) The zeros of the function f are of the form ±>/h, kEN. Let us show that the
distance between the zeros of f tends to zero as k -t 00, see Subsection 4.1.1.

lim
k-+oo /j(k + 1)7r - Jk;/ = }~~ /(k + 1~7r + Jk; = O.

This implies that the function f is not periodic.

c) Only the numbers 27r and 7r are candidates for the basic period of f. However,

f (- i) = 1, f (- i+ 27r ) = f c;) = -1,

f (~)
4
= v'2
2'
f (7r + ~)
4
= f (57r) = _ v'2
4 2·
Hence, the function f is not periodic.

d) Periodic with basic period T = 27r (check!).

58 CHAPTER 2.

Example 2.18. Determine the basic periods of the following functions, if they exist.
x l I
a) f(x) = cos 4" + cos x + 2" cos(3x) + :3 cos(5x), x E R;

b) f(x) = cos(x) + cos(V3x), x E R;

c) f(x) = vtanx, x E U [br, (2k + l)~) ;

kEZ 2

2 7r 2 7r 2)
d) f(x)=tanyfx, 7r )
xE [0,- U ( (2k-l)2-,(2k+l)2- .
4 kEN 4 4

Solutions.
a) Firstly we have

X4
f(x + T) - f(x) = ( cos - + T- - cos 4"X) + (cos(x + T) - cosx)

1 1
+ 2" (cos(3(x + T)) - cos(3x)) + :3 (ms(5(x + T)) - cos(5x))

. T . 2x + T 2' T . 2x + T
- 2 Sln-Sln--- - sm-sm---
8 8 2 2
. 3T . 6x
- sm-sm
+ 3T - 2: 5T . lOx + 5T
-sm-sm----
2 2 3 2 2
Now f(x + T) - f(x) == 0 if and only if for some m, n,p, q from N it holds
2p7r 2q7r
T = 8n7r = 2m7r , T = -3 = -5
From the last equalities we have that m = 4n, p = l2n, q = 20n, and therefore
the given function is periodic with basic period T = 87r.

b) The function g(x) = cosx, x E R, is periodic with the basic period 27r, while
the function h(x) = cos(V3x), x E R, is periodic with the basic period ~.
However, the function f = 9 +h is not periodic, because there exist no nonzero
27r
integers k and I such that 27rk = V31.

c) Since for T > 0 it holds

tan(x + T) - tan x
Jtan(x + T) - Jtan(x) = Jtan(x + T) + Jtan(x)

sinT
- cos(x + T) cos x (jtan(x + T) + jtan(x)) ,
FUNCTIONS 59

°
the last expression is identically equal to iff T = k1r for some k E Z. Therefore
the given function is periodic with basic period T = 7r.
d) Not periodic.
Example 2.19. Prove that the function f given with the formula
f (x) = x - [x], x E R,
is periodic and find its basic period (if it exists), where [x] is the greatest integer part
of x, see Example 1.47 d).
Solution. For x = z + r, where z E Z and °: ;
r < 1, it holds [x] = z. (Note that
[x] = x iff x E Z, otherwise x-I < [x] < x.) Then
f(x) = f(z + r) = z + r - z = r.
If x E Z, then f(x) = 0. Thus
(Vx E Z) f(x) = f(x + 1).
If x E R \ Z, then from the decomposition x = z + r, where z E Z and °< r < 1, it
follows
f(x + 1) = (x + 1) - [x + 1] = x + 1 - (z + 1) = x - z = r = f(x).
Hence the basic period of f is 1.
Example 2.20. Check whether the Dirichlet function given by the formula
D( )
x
= {I, 0,
if x is a rational number;
if x is an irrational number,
is periodic.
Solution. Let us prove first that every rational number r is a period of the function
D. Namely, since the sum of two rational numbers is again rational, it follows that
D(x + r) = 1 = D(x) for every rational number x, while the sum of a rational and
°
an irrational number is an irrational number. Hence D( x + r) = = D( x) for every
irrational number x. Thus we proved that every rational number r is the period
of D. However, since there does not exist a smallest positive rational number, it
follows that the Dirichlet function has no basic period.
(Let us add that no irrational number is the period of D - check that!)
Example 2.21. Show that the following functions f : A ---+ B are bijections and
find their inverse functions.
a) f(x) = 3x + 4, A = B = R;

x2
b) f(x) = 3' A = (-00,0]' B = [0,+00);

c) f(x) = I-x, A=B={XERll x l>l};

I+x

d) f(x) = sinh x := ~(exp(x) - exp( -x)), A = B = R.

60 CHAPTER 2.

Hint. Use Theorem 2.2.

Solutions.
a) Let us show first that f is a surjection. Namely, let y E R given. Solving the
equation y = 3x + 4 by x gives x = ~(y - 4) E R. Then for x E R it holds

f (x) = 3 G (y - 4)) + 4 = y,

which implies the surjectivity of f. Since for every pair xI, X2 E R it holds

Xl < X2 ~ 3XI < 3X2 ~ 3XI + 4 < 3X2 + 4 ~ f(XI) < f(X2),
it follows that f is monotonically increasing on R. Applying Theorem 2.2 we
obtain that f is a bijection. Then we know that there exists an inverse function
g. In order to find its analytical expression, we formally replace x and y in the
given formula y = 3x + 1 and then attempt to solve the equation for y. This
gIves us
1
x = 3y + 4 ~ y = 3(x - 4).
Hence the inverse function 9 : R -t R is given by the formula
1
g(x) = 3(x - 4).

b) Let y 2 ° be given. From the equality y = ~

2
, x ~ 0, it follows x = -y'3Y.
Then f(x) = y and hence f is a surjection.
Next, for every pair XI,X2 E (-00,0] it holds

Xl < X2 ~ xi/3 > x;/3 ~ f(xt} > f(X2)'

Hence f is monotonically decreasing on (-00,0]. From Theorem 2.2 it follows

that f is a bijection from (-00,0] onto [0, +00). Thus the inverse function
g: [0, +00) - t (-00,0] exists and is given by the formula, g(x) = -ffx.

c) Let y such that Iyl > 1 be given. Then solving the equation
I-x
y=l+x

by x it follows that a unique x such that Ixl > 1 exists, namely x = 1 - y. So

l+y
we showed that f is a surjection.
Let us prove now that f is monotonically decreasing on (-00, -1). To that
end, first we assume that XI, X2 E (-00, -1). Then we have
1 - Xl 1 - X2
Xl < X2 ~ 1- Xl> 1- X2 ~ - - > - - ~ f(XI) > f(X2),
1 + Xl 1 + X2
FUNCTIONS 61

hence f is monotonically decreasing on (-00, -1). Analogously we prove that

f is also monotonically decreasing on (1, +00). In view of Theorem 2.2, it
follows that f is a bijection, thus it has an inverse function g. Exchanging x
I-x l-y
and y in the formula y = - - , we get the equation x = - - . Solving it
l+x l+y
by y we get
I-x
g(x) = -1-' Ixl > 1,
+x
which means that the function f is equal to its inverse function g.

d) The function f = sinh x is monotonically increasing on R, because it is a sum of

two such functions. Solving the equation y = ~(exp( x) - exp( -x» by x gives

= y ± fj+l
2x
e -1
y =- --
2e
~ 2ye X - e2x +1 = 0 ~ eX y2 + 1.

Hence
x = In(y + fj+l).
Clearly, for given y E R this x E R has the property y = f(x). Thus f is a
surjection and from the monotonicity of f it follows that it is a bijection. The
formula that gives the inverse function 9 : R ----* R is

g(x) = In(x + R+1).

Example 2.22. Assume the function f : A ----* B is a bijection, and denote by f- 1 :

B ----* A its inverse function. Prove that for arbitrary sets X c A and Yl, Y2 C B
the following holds.
a) f(X) -:f: 0 =? X -:f: 0;

b) Yi c Y 2 ~ f- 1 (y1 ) c r 1 (y2 );

c) f- 1 (Yi n Y2) = f- 1 (Yi) n f- 1 (12);

d) f- 1 (Y1 u 12) = f- 1 (Yi) u f- 1 (12).

Solutions.
a) Since the function f is a bijection, it holds that its codomain B is equal to the
range of f, the set f(A). If the set f(X) is nonempty, i.e., if there exists an
element y E f(X), then it follows that there exists a unique element x from
X such that f(x) = y.
b) Assume Yi C Y2 , and let x be an arbitrary element from the set f- 1 (Yi). Denote
y = f( x); since f is a bijection, x is the unique element from the domain X
whose image is y. By hypothesis y is also in yz, which implies x E f- 1 (y2 ).
Assume now j-l(Yi) C j-l(12), and take y E Yi. Then the element x
f-l(y) is in f- 1 (Yi), hence x E j-l(y2 ). But then it holds y = j(x) E Y2.
62 CHAPTER 2.

c) Using that f is a bijection, we have

xEf- l (yl ny2 ) {:=::} (3 l yE(Yiny2 )) f(x)=y

{:=::} (y E Yi) /\ (y E Y2 ) {:=::} (x E f-l(ll)) /\ (x E f- l (y2 ))

{:=::} x E (J-l(Yi)nf- l (12)).

d) Analogous to c).

Exercise 2.23. Find the largest sets A and B such that the function f : A ---+ B is
a bijection, and then find its inverse function g : B ---+ A, if
1 x
a) f(x) = x 5 ; b) f(x) = x + 1·
Answers ..
a) A = B = R \ {O}, g(x) = X- 1/5 .

x
b)A=R\{-I}, B=R\{I}, g(x)=-.
I-x
Example 2.24. Using the graph of the function g(x) = x 2 , X E R, sketch the graph
of the following functions.

1 .
a) f(x) = 1 +x 2 ' b) f(x) = ex2 ;

c) f(x) = sinx 2 ; d) f(x) = Inx 2 •

Solutions.

y
I
\ I
y y.K \ I
\ I
\ I
\ I \ I
\ I
,,
1
\ I

, .... ., I

% , .- %
o 1 0
I
1

1
Fig. 2.1. f(x) = 1 + x 2 Fig. 2.2. f(x) = ex2

a) The function is defined on the interval (-00, +00), is even and has no zeros. It
has a maximum at x = 0 (both local and global), is increasing on (-00,0) and
decreasing on (0,+00), concave upward. (Figure 2.1.)
FUNCTIONS 63

b) The function f(x) = ex2 is defined on the interval (-00,+00), is even and
has no zeros. It has a minimum at x = 0, is decreasing on (-00,0) and is
increasing on (0, +00), concave upward. (Figure 2.2.)

c) The function f( x) = sin x 2 is defined on the interval ( -00, +00 ), is even, but not
periodic (see Example 2.17 b)). The zeros of f are at x = ±.Jh, k = 0,1, ....
This function has local maximums at x = ±J% + 2k7r, k = 0,1, ... , and local
minimums at x = ±J3; + 2br, k= 0,1, .... (Figure 2.3.)

y
y

Fig. 2.3. f(x) = sinx 2 Fig. 2.4. f(x) = Inx 2

d) The function f(x) = In x 2 is defined on (-00,0) U (0, +00), is even and has zeros
at Xl = -1, X2 = 1. It has no extrema (i.e., no minimums or maximums) , is
decreasing on (-00,0) and increasing on (0, +00). (Figure 2.4.)

Example 2.25. Using the graph of the function g(x) = cos x, x E R, (see Figure
2.5), sketch the graph of the following functions.
1
a) f(x) = - ; b) f(x) = cos n x, n E N;
cos x

1
c) f(x) = Icos xl; d) f(x) = -Icos
-I;x

1
e) f (x) = In I cos x I; f) f(x) = In I cosx I·

Solutions.

a) The function f( x) = _1_ is defined on the union of the intervals

cos x

((2k -1)%, (2k + 1)%), k = 0, ±1, ±2, ... ,

where it holds cos x 01 o. The function f is even, periodic with basic period
271" and has no zeros.
64 CHAPTER 2.

If x E ( -~,~) , then it holds

1
0< cos x < 1, i.e., - - >- 1 ,
cos x
meaning that on this interval the function has a mInImUm at x = O. On
the interval (~, 3;,)
the function has a maximum at x = 7f. The vertical
7f 7f
asymptotes of the graph are x = -2 and x = 2· (Figure 2.6.)

J
y
U.II
1\

y I
1t "

~r
1
Fig. 2.5. f( x) = cos x Fig. 2.6. f(x) = cos x

b) The function f(x) = cos n x is defined on the interval (-00, +(0) is even and its
zeros are at the points ~ + k7f, k = 0, ±1, ±2, ....
If n = 2k, kEN, the function is periodic with basic period 7f. On the interval
[- ~, ~] , the function has minimums at - ~, ~, and a maximum at O. (Figure
2.7 for n = 2.)
If n = 2k + 1, kEN, the function is periodic with basic period 27f. On the
interval [-~, 3;] , the function has a minimum at 7f and a maximum at O.
(Figure 2.8 for n = 3.)

y ~
1

o 1

Fig. 2.7. f(x) = cos 2 X Fig. 2.8. f( x) = cos3 X

FUNCTIONS 65

c) The function f( x) = 'cos x, is defined on the interval (-00, +00), is even and its
zeros are at the points ~ + k1r, k = 0, ±1, ±2, .... This is a periodic function
with basic period 7r.

On the interval [-~, %] , the function has minimums at the points -~, ~,
and a maximum at O. (Figure 2.9.)

d) The function f(x) = -,_1_, is defined on the intervals ((2k -

xcoo
1)~,
2
(2k + 1)~) ,
2
k = 0, ±1, ±2, ... , where cos x =f o. The function is even, periodic with basic
period 7r and has no zeros.

On the interval ( -%,~) , the function has a minimum at o. It has vertical

asymptotes x = -% and x = %. (Figure 2.10.)

r.Y

o 1 Jr/2 1t x

1
Fig. 2.9. f(x) = Icosxl Fig. 2.10. f(x) = I cosxl

e) The function f(x) = In Icos x, is defined on the intervals ((2k - 1)%, (2k + 1)%) ,
k = 0, ±1, ±2, .... The function is even, periodic with basic period 7r and has
zeros at the points k7r, k = O. ± 1 ± 2, ....

On the interval (- ~, ~) , the function has a maximum at x 0. It has

vertical asymptotes x = -% and x = %. (Figure 2.11.)

f) The function f (x) = I ,1 , is defined on the intervals
ncosx
~,
2
(k.
(k + 1) . ~) ,
2
k = 0, ±l, ±2, .... The function is even, periodic with basic period 7r and has
no zeros. The function f has no extrema. The vertical asymptotes are x =
k7r, k = 0, ±l, ±2, .... However, the lines x = (2k + l)i, k = 0, ±1, ±2, ... ,
are not vertical asymptotes even though the function is not defined at these
points. (Figure 2.12.)
66 CHAPTER 2.

1
Fig. 2.11. f(x) =lnlcosxl Fig. 2.12. f(x) = 'nlcosx

Example 2.26. Sketch the graphs of the following functions.

a) f(x)=x+cosx; b) f(x)=xsinx;
c) f(x) = sin.!..; d) f (x) = x sin .!...
x x
Solutions.

:It x

, ,,
,, ,

Fig. 2.13. f(x) = x + cos x Fig. 2.14. f(x) = xsinx

a) The function f (x) = x + cos x can be considered as the sum of the functions
II = x, x E R, and 12 = cos x, x E R.
This function is neither odd nor even, is not periodic and has a zero approxi-
mately at the point -0.739. It has no extrema. (Figure 2.13.)

b) The function f (x) = x sin x is defined for every x E R. This is an even not
periodic function, which has zeros at the points br, k = 0, ±1, ±2, .... At
the points %+ 2k1r, k = 0, ±1, ±2, ... , it holds f(x) = x and at the points

3; + 2k1r, k = 0, ±1, ±2, ... , it holds f(x) = -x. (Figure 2.14.)

FUNCTIONS 67

1 :y

," '"", ,
,, ,,
y=-1 " "

Fig. 2.15. f(x) = sin ~ Fig. 2.16. f(x) = x sin ~

x x

c) The function f(x) = sin ~x is defined for every x E R \ {o}. This is an odd, not
1
periodic function with zeros at the points br' k = ± 1, ±2, .... It is a bounded
1
function, namely -1 ::::: sin - ::::: 1.
x
1
The maximums of f are at the points ( /)- k ' k = 0, ±1, ±2, ... , and
7r2+27r
1
minimums at the points (37r /2) + 2h' k = 0, ±1, ±2, .... (Figure 2.15.)

d) The function f (x) = x . sin ~ is defined for every x E R \ {o}. This is an even,
x
1
not periodic function with zeros k7r' k = ±1, ±2, .... (Figure 2.16.)

Example 2.27. Sketch the graphs of the following functions.

a) f(x) = arcsin x; b) f(x) = arccos x;

c) f(x) = arctan x; d) f(x) = arccotx.

Solutions.

a) The function f: [-1, IJ-t [-~,~] is monotonically increasing. (Figure 2.17.)

68 CHAPTER 2.

y y
~2

lit
-1

-~2 lit

Fig. 2.17. f(x) = arcsin x Fig. 2.18. f(x) = arccos x

b) The function f : [-1,1] -+ [0,7r] is monotonically decreasing.

c) The function f : (-00, +00) -+ [-i, i] is monotonically increasing. The graph

of f has two horizontal asymptotes y = -% and y = %. (Figure 2.19.)

, __ ~M I" __ !~~__
--~----

:t

_____ ________
~ ~:M

o 1
x

Fig. 2.19. f(x) = arctan x Fig. 2.20. f(x) = arccotx

d) The function f : (-00, +00) -+ [0,7r] is monotonically decreasing. The graph of

f has two horizontal asymptotes y = 0 and y = 7r. (Figure 2.20.)
FUNCTIONS 69

Example 2.28. Show the following identities.

a) arcsin x + arccos x = %' Ixl:S:; 1;

b) arcsin x = arccos(v'f=X"2), O:S:; x:S:; 1;

c) arcsin x ± arcsin y = arcsin (x~ ± Yv'f=X"2) ,

larcsinx±arcsinyl:S:; %' Ixl:S:; 1, lyl:S:; 1;

d) arccos x ± arccos y = arccos(xy ± v'f=X"2~),

O:S:; arccos x ±arccosy:S:; 7r, Ixl:S:; 1, lyl:S:; 1;

+ arccot x = 2'
7r
e) arctan x x E R;

f) arctan x ± arctan y = arctan ( -x±y)

-- ,
1 =f xy
7r
I arctan x ± arctan yl < 2·

Solutions.

a) Let us denote a = arcsin x, hence for x E [-1,1] it holds a E [-%, %J . Then

we have

X = sin a = cos ( 27r - a

)
~ 2
7r
- a = arccos x ~ arcsin x + arccos x = 2·
7r

b) Let us put a = arcsin x, x E [0,1], hence a E [0,%], then

x = sin a = cos (% - a) ~ x 2 - 1= - cos 2 a

~ cos a = ~ ~ a = arccos ( ~).

c) If we put a = arcsin x and 13 = arcsiny, a,j3,a + 13 E [-i, iJ ' then x = sina

and y = sin 13, hence cos a = v'f=X"2 and cos 13 = ~. Applying the
function arcsin to the formula

sin( a + 13) = sin a . cos 13 + sin 13 . cos a,

under the assumptions on x and y, we obtain the announced equality.
70 CHAPTER 2.

Example 2.29. Sketch the graphs of the following functions.

eX _ e- x
a) f(x) = sinh x = ; b) f(x) = cosh x = eX + e- x .
2
eX _ e-X
c) f(x) = tanh x = eX +
e- x ; d) f(x) = coth x = eX+ e-
X
eX - e- X

Solutions.

a) The domain and codomain of this function is R. The function is monotonically

increasing and odd. (Figure 2.21.)

"
"
%

%
o 1

Fig. 2.21. f(x) = sinh x Fig. 2.22. f(x) = cosh x

b) The function f : R --t [1, +00) is even and has a minimum at the point x = o.
(Figure 2.22.)

c) The function f : (-00,+00) --t (-1,1) is monotonically increasing and odd.

(Figure 2.23.)

d) The function f : (-00,0) U (0, +00) --t R is monotonically decreasing and odd.
(Figure 2.24.)
FUNCTIONS 71

"

L
-----,
%
.!..=..1 o 1 ,,=-1
---------
%

1
~------~--------.!..=..-!--

Fig. 2.23. f(x) = tanh x Fig. 2.24. f(x) = cothx

Example 2.30. Show for the functions

. 1 1 sinh x
smhx:= -2 (eX - e- X), coshx:= -2 (eX + e- X) and tanh x := --h-'
cos x
whose natural domains is the set of real numbers R, that the following identities
hold for arbitrary real numbers x and y.

a) cosh(x ± y) = cosh x . coshy ± sinh x . sinhyj

b) cosh(2x) = cosh 2 x + sinh2 Xj

c) sinh( x ± y) = sinh x cosh y ± cosh x sinh Yj
d) sinh(2x) = 2sinh(x)· cosh(x)j

tanh x ± tanh y
e) tanh( x ±)
y - .
- 1 ± tanh x . tanh y ,

2 tanh(x)
f) tanh(2x) = 1 + tanh 2 x .

Solutions. Before we prove the given relations, let us note that by the definitions
of the functions sinh and cosh, it follows

eX = cosh x + sinh x, e- X = cosh x - sinh x (2.2)

for every x E R. Using the main property of the exponential function, namely
eX+Y = eX . eY , x E R, and the equalities in (2.2), we obtain for every x, y E R

eX +Y = cosh(x + y) + sinh(x + y) = (cosh x + sinh x)( cosh y + sinh y) (2.3)

72 CHAPTER 2.

and

e-(X+Y) = cosh(x + y) - sinh(x + y) = (cosh x - sinhx)(coshy - sinhy). (2.4)

a) Summing (resp. subtracting) the equality (2.4) with (resp. from) the equality
(2.3), we obtain the given equality.
b) Putx=yina).

e) Using the equalities a) and c), we obtain

sinh(x ± y)
tanh(x ± y) = cosh(x ± y)

sinh x . cosh y ± cosh x . sinh y tanh x ± tanh y

cosh x . cosh y ± sinh x . sinh y 1 ± tanh x . tanh y ,
for every x, y E R.

Exercise 2.31. Prove the following identities for every x, y E R.

x+y x-y
a) cosh x + cosh y = 2 cosh -2- cosh -2-;

. x+y. x-y
b) cosh x - cosh y = 2 smh -2- smh -2-;

c) 'h x
sm + SIn
'h y = 2 SIn x2
'h - + -y cos h -2-;
X - y

d) sm
. h x - SIn x + y sIn
. h y = 2 cos h -2- . h -2-;
x - y

coth x . coth y ± 1
e) coth(x ± y) = cothx ± cothy .

2.2 Polynomials, rational and irrational func-

tions
2.2.1 Basic notions
The function

Pn(x) anx n + an_lX n- 1 + ... + alx + ao, x E R, (or x E C), (2.5)

where the coefficients aj, j = 0,1, ... , n, are real numbers, is called polynomial
function (shortly: polynomial) of degree n E N if the coefficient an :I 0. 2
2In (2.5), the subscript n in Pn(x) stands only to denote the degree of that polynomial.
FUNCTIONS 73

By definition, the constant function is a polynomial of degree zero. Clearly, the

natural domain of any polynomial is the whole set of real numbers R. Occasionally,
however, we shall even allow complex values for x in (2.5).
A number Xo is a zero (or sometimes termed root) of the polynomial Pn(x)
from (2.5) if
Pn(XO) = o.
If a zero Xo of a polynomial is a rational, real or respectively a complex number,
then it will be called rational, real or respectively complex zero.

Theorem 2.32. Fundamental theorem of algebra.

Each polynomial of degree n E N has exactly n (real and/or complex) zeros, among
which some might be equal.

If Xo is a real zero of the polynomial Pn(x), then there exists a unique polynomial
Qn-1 with real coefficients,

Qn-1(X) = bnxn + bn_1x n- 1 + ... + b1x + bo, x E R, (2.6)

of degree n - 1, such that

Pn(X) = (x - XO)Qn-1(X), (2.7)

The method of quick determination of the polynomial Qn-1 (x), the so-called Horner
scheme, is explained below.
In view of Theorem 2.32 and the factorization (2.7), the number Xo is called zero
of order m, mEN, 1 :S m :S n, of the polynomial Pn(x) from (2.5) if there exists
a polynomial Rn-m(x) of degree n - m, such that Rn-m(xo) i- 0 and it holds

Pn(X) = (x - xo)m Rn-m(x).

Any polynomial Pn(x) with real coefficients of the form (2.5), can be written in
a unique way as the product

Pn(X) = an(x - X1)m , ... (x - x r )m'(x 2 + b1x + C1)11 ... (X2 + bsx + cs)l" (2.8)

where for the natural numbers mj, j = 1, ... ,r, and lk' k = 1, ... ,s, it holds

m1 + ... + mr + 2(11 + ... + ls) = n.

In (2.8), the numbers Xj are mutually different zeros of order mj, j = 1, ... ,r, of the
polynomial Pn (x), while the zeros of the polynomials X2 + bkx + Ck, k = 1, ... , s, are
conjugate complex numbers.
In general, it might be a tricky task to find the zeros of a polynomial. However,
in the special (but quite often) case when the polynomial Pn(x) from (2.5) has
only integer coefficients, then the candidates for rational zeros can be found quickly.
Namely, ifthe reduced fraction E, where p E Z and q EN, is the zero of Pn(x), then
q
necessarily the numerator p is a divisor of ao (we write that shortly plao), while the
74 CHAPTER 2.

denominator q is a divisor of an (qla n ) (prove that!). In particular, if an = 1, then

q = 1, hence a rational zero of Pn(x) with integer coefficients is, in fact, an integer.
The Horner scheme allows a quick check whether a real number Xo is or is not
a zero of the polynomial Pn(x). Namely, performing the division of the polynomial
Pn(x) with the polynomial x - Xo (of degree one) gives

Pn(X) = (x - XO)Qn-I(X) + r. (2.9)

where the polynomial Qn-I(X) of degree (n - 1) is given by relation (2.6). Its

coefficients bn - l , ... , bo can be found by the formula

bn- l = an, bn- 2 = xObn- 1 + an-b ... , bo = xObl + aI, (2.10)

or
bk=xobk+1+ak+b k=0,1, ... ,n-1,
and the remainder r is given by

r = xobo + ao.

Clearly, if r = 0, then the number Xo is the zero of the polynomial Pn (x) and
relation (2.9) coincides with (2.7). However, if r 01 0, then the number Xo is not a
zero of the polynomial Pn(x).
The Horner scheme is most easily performed as follows.
First one writes in a row all the coefficients of the given polynomial function Pn(x),
including those that are equal to zero. Using the upper formulas for bk, k =
0,1, .. , n - 1, and the one for r, we make the following two-row scheme:

an an-l an-2 ak ao I Xo
bn - l bn - 2 bk bo Ir.
All we have to do now is to apply the upper analysis, i.e., to check whether the
remainder r is zero or not.
The rational function is the quotient of functions

Pn(x)
R(x) = Qm(x)' Qm(x) 01 0, (2.11)

where Pn(x) and Qm(x) are polynomials of degree nand m.

The most important examples of the rational functions are the following two
types of partial fractions

A Bx+C
(x_a)i,jEN, and (x 2 +bx+c)k' kEN, (2.12)

for some natural numbers j and k. In relation (2.12), A, B, C, a, band c are constants
and the zeros of the polynomial x 2 + bx + c are conjugate complex numbers.
FUNCTIONS 75

Theorem 2.33. A rational function of the form (2.11) can be written in a unique
way as the sum of partial fractions of the form (2.12) and, if n 2': m, a polynomial
of degree n - m. (If n = m, then the last polynomial reduces to a constant.)

This theorem is essential when calculating the integrals of rational functions.

The irrational function is an elementary function which is not rational and
can be defined as a composition of finitely many rational functions, power functions
with rational exponents and arithmetic operations.

2.2.2 Examples and exercises

Example 2.34. Find the rational zeros of the following polynomial functions.

a) P(x) = 2x 2 +4x - 6; b) P(x)=x 3 +3x 2 -4x-12;

c) P(x) = 2X4 - 13x 3 + 28x 2 - 23x + 6; d) P(x)=x 5 -x3 +x 2 -1;

e) P(x) = x7 + 6x 5 - 5x 4 + 9x 3 - 10x 2 + 4x - 5.

Solutions.

a) The given polynomial P( x) is of degree 2, so the roots can be found as the

solutions of the quadratic equation and have the forms Xl = -3, X2 = 1. In
view of Theorem 2.32 these are also the only zeros of P(x); observe that they
are both rational. This means that we have the factorization

2X2 + 4x - 6 = 2(x - l)(x + 3).

Remark. The zeros of the polynomial P( x) could have been also found using
Horner scheme.

b) For the polynomial P( x) we shall find the divisors of 12, since ao 12. The
divisors are
±1, ±2, ±3, ±6, ±12.
Further on, since an = a3 = 1, the only possible rational zeros of the polyno-
mial P(x) are integers from the upper set of divisors.
Let us check whether x = 1 is a zero of P(x)

1 3 -4 -12 I x=l
1 4 0 1-12.

Since the last number in the second row is -12 (i.e., r = -12), the number
x = 1 is not a zero of P(x).
Next, we check whether x = 2 is a zero of P(x)

1 3 -4 -12 I x=2
1 5 6 10.
76 CHAPTER 2.

Now the last number in the second row is 0, (i.e., r = 0), hence x = 2 is a zero
of the given polynomial. So we obtained the factorization

x3 + 3x 2 - 4x - 12 = (x - 2)(x 2 + 5x + 6).

We yet have to find the zeros of the polynomial Q(x) := x 2 + 5x + 6. We can

find them either by solving the quadratic equation

x 2 + 5x + 6 = 0,
or as we shall do, using the Horner scheme.
Let us try the number x = -2

1 5 6 I x =-2
1 3 10.
Hence, the second zero of the polynomial P3(X) is the number x = -2. So we
obtained the factorization

x3 + 3x 2 - 4x - 12 = (x - 2)(x + 2)(x + 3).

(The third zero is -3.)

c) Let us determine first the divisors of 6. They are

±1, ±2, ±3, ±6.

The positive integers that are divisors of 2 are 1 and 2, hence the possible
rational zeros of the given polynomial are

1 3
±1, ±2, ±3, ±6, ±2"' ±2".

U sing the Horner scheme, we obtain that x = 1, x = 2 and x = 3 are the zeros
of the given polynomial:

2 -13 28 -23 6 I x=1

2 -11 17 -6 10 x=2
2 -7 3 10 x=3
2 -1 10.
Now, the last row of this scheme shows that the fourth zero of the given
polynomial is x = ~. So we obtain the factorization

2X4 - 13x3 + 28x 2 - 23x + 6 = 2(x - 1)(x - 2)(x - 3)(x - 1/2).

FUNCTIONS 77

d) The only possible rational zeros of the given polynomial are 1 and -1, hence the
Horner scheme gives
1 0 -1 1 0 -1 I x=1
1 1 0 1 1 10 x =-1
1 0 0 1 10 x =-1
1 -1 1 10.
This shows that

P(x) = x 5 - x 3 + x 2 - 1 = (x + 1)2(x - 1)(x 2 - X + 1). (2.13)

The last factor x 2 - x + 1 has no real, therefore no rational zeros, but rather
its zeros are conjugate complex numbers. This means that the formula (2.13)
is the factorization of the form (2.8).

e) The possible rational zeros of P(x) are ±1,±5. The Horner scheme gives
1 0 6 -5 9 -10 4 -5 I x = 1
1 1 7 2 11 1 5 10 x=1
1 2 9 11 22 23 128.
This shows that x = 1 is a zero of first order, and not of second order, since
the last number in the third row is 28. We leave to the reader to check that the
numbers x = -1, x = 5 and x = -5 are not the zeros of the given polynomial,
hence it has no other rational zeros. Moreover, the polynomial P(x) has no
other real zeros, and the factorization of the form (2.8) is
P(x) = x7+6x5-5x4+9x3-10x2+4x-5

(x - 1)(x 2 + X + 5)(x 2 + 1)2.

Example 2.35. Decompose to partial fractions the following rational functions.

2 x 3 - 2x - 35
a) b)
x 2 - l' x 2 - 2x -15'

x+l x+3
c) d)
x 3 - 2x 2 + X - 2 ' x4 - 5x 2 + 4'

x2 + 1 2X2 - 4x + 3
e) f)
(x-l)3' x4 - 6x 3 + 13x 2 - 12x + 4
Solutions.
a) The zeros of the polynomial x 2 - 1 (the denominator of the given rational func-
tion) are x = 1 and x = -1. Using Theorem 2.33, it follows that it can be
decomposed as the sum of following partial fractions
2 A B
- - = -x -
x 2 -1 -1
+- -.
x+1
78 CHAPTER 2.

From this equality we get the constants A and B in the following way. First,
we add the last two fractions

A B A(x + 1) + B(x - 1)
--+--=
x -1 x +1 x 2 -1
.
Since the denominator of the obtained fraction is the same as that of the given
rational function, their numerators should be also the same, hence

2 = x(A + B) + (A - B).

So we obtain the linear system of equations with unknown A and B

A+B=O, A-B=2, whichgives A=I, B=-l.

This means that finally we have

2 1 -1
-2-- = - - + --.
x -1 x-I x+l

b) Let us notice first that the degree of the numerator is greater than that of the
denominator, so let us do the following transformations:

x 3 - 2x - 35 x3 - 2x 2 - 15x + 2x 2 + 13x - 35
x 2 - 2x -15 x 2 - 2x -15

2X2 + 13x - 35 17 x - 5
x+ 2 =x+2+ 2 •
x - 2x - 15 x - 2x - 15

Remark. Instead of these transformations, one can divide the polynomial

x 3 - 2x - 35 from the numerator with the polynomial x 2 - 2x - 15 from the
denominator, which gives the polynomial x + 2 and the fraction

17x - 5
x 2 - 2x - 15

Further on we have
17x-5 A B A(x+3)+B(x-5)
x 2 - 2x - 15 = -x---5 + -x-+-3 = -----'---x--=-2--'-2-x------'---1-5-----'- ,
which gives us the system of linear equations

A + B = 17, 3A - 5B = -5.
So we obtain A = 10 and B = 7 and finally
x 3 - 2x - 35 2 10 7
=
----::--------::-----------:'CC' X + + -- + --.
x 2 - 2x - 15 x - 5 x + 3
FUNCTIONS 79

c) The polynomial in the denominator can be written as

x3 - 2x2 +X - 2 = (x - 2)(x 2 + 1).

Hence, there exist constants A, Band C, such that

x+1
-A- + ---:---
Bx+C
x3 - 2x 2 + X - 2 X - 22 x +1
Ax 2 + A + Bx 2 - 2Bx + Cx - 2C
(x - 2)(X2 + 1)
or
x +1 (A + B)x 2 + (C - 2B)x + A - 2C
x 2x + X - 2
3 - 2 x 3 - 2x 2 + X - 2
This gives us a system of linear equations

A+B=O, C-2B=1, A-2C=1,

whose solution is A = 3/5, B = -3/5, C = -1/5.

Finally we have
x+1 3 3x + 1
x3 - 2x 2 + X - 2 5(x - 2) 5(x2+1)·

d) The numbers x = 1, x = 2, x = -1 and x = -2 are the zeros of the polynomial

in the denominator, hence

x+3 ABC D
--,-----:-- = - -
X4 - 5x 2 + 4 x-I
+- - +--
x +1 x - 2
+ --.
x +2
Adding together the last fractions gives us

x+3 Ax3 + Ax 2 - 4Ax - 4A + Bx 3 - Bx 2 - 4Bx + 4B

X4 - 5x 2 + 4 X4 - 5x 2 + 4

Cx 3 + 2Cx 2 - Cx - 2C + Dx 3 - 2Dx2 - Dx + 2D
+ X4 - 5x 2 + 4

(A + B + C + D)x3 + (A - B + 2C - 2D)x 2
X4 - 5x 2 + 4

-(4A + 4B + C + D)x - 4A + 4B - 2C + 2D
+ ~--------~~-------------­
X4 - 5x 2 +4
Hence

A + B + C + D = 0, A- B + 2C - 2D = 0,
-(4A + 4B + C + D) = 1, -4A + 4B - 2C + 2D = 3,
80 CHAPTER 2.

which gives
A = -2/3, B = 1/3, C = 5/12, D = -1/12.
So we obtained the decomposition

x +3 -2 1 5 1
-x4:---5-x-=-2-+-4 = 3(x -1) + 3(x + 1) + 12(x - 2) - -12-(x-+ 2)"

e) The number 1 is the zero of order 3 of the denominator. Thus we have

x2+1 ABC
-- + + -,-----
(x - 1)3 x-I (x-l)2 (x-l)3

Ax 2 - 2Ax + A + Bx - B + C
(x - 1)3
which gives the system of linear equations

A=I, -2A+B=0, A-B+C=l.

Its solution is A = 1, B =2 and C = 2, hence we obtain the decomposition

x2 + 1 1 2 2
- - - = -x-I
(x-l)3
- + (x-l)2 +(x-l)3·
---

f) The values x = 1 and x 2 are zeros of order 2 of the polynomial in the

denominator, hence

2X2 - 4x + 3 ABC D
---,--:--::--.,-,:--=-----,-.,-----:- = - - + + -- + ---
x 4 -6x 3 +13x 2 -12x+4 x-I (x-l)2 x-2 (x-2)2

Ax 3 - 5Ax 2 + 8Ax - 4A + Bx 2 - 4Bx + 4B

X4 - 6x 3 + 13x 2 - 12x + 4

Cx 3 - 4Cx 2 + 5Cx - 2C + Dx 2 - 2Dx + D

+ X4 - 6x 3 + 13x 2 - 12x + 4
This gives the system

A + C = 0, -5A + B - 4C + D = 2,
8A - 4B + 5C - 2D = -4, -4A + 4B - 2C + D = 3,

whose solution is A = 2, B = 1, C = -2, D = 3.

Finally we have

2X2 - 4x + 3 2 1 - 2 3
--------- = -- + + - - + -,------
4
x -6x 3 +13x 2 -I2x+4 x-I (x-I)2 x-2 (x-2)2
FUNCTIONS 81

Exercise 2.36. Decompose to partial fractions the following rational functions.

a 3. b 4x 2 + 4x + 26
) x 3 +1' ) (2x-4)(x 2 +1)2'

1 2
c) d)
X4 + l' X4 + x 2 + l'
Solutions.
3 1 x-2
a) ---
x3 + 1 x+l x 2 - X + l'

4x 2 + 4x + 26 1 x+2 3x +4
b) ------
(2x - 4)(X2 + 1)2 x-2 x2+1 (x 2 +1)2'

1 V2x+2 V2x-2
c)
X4 +1 4(x 2 + xV2 + 1) 2
4(x -xV2+1)'

d) 2 x-I. x +1
x4 + x2 + 1 x2 +X + 1
x2 - X + 1
Example 2.37. Prove that the function f( x) = X1 / 2, x 2: 0, is an irrational
function. 3
Solution. Assume that there exist polynomials P and Q such that
1/2 _ P(x)
x -Q(x),x2:0.

Then
1 P(x)
y'x xQ(x)' x> 0,
hence
lim _1_ = lim P(x) = O.
x--++oo y'x x--++oo xQ(x)
So the degree of P(x) is less than the degree of x . Q(x) and the degree of P is ~
than the degree of Q. Thus
·
11m r::: l' P(x) K
X~+OO
yX = 1m
X----++OCl
Q( X ) = ,

for some constant K, giving a contradiction to

lim y'x =
x--++oo
+00.
Exercise 2.38. Prove that the function f( x) = x p / q , x > 0, is an irrational
function, provided that q E N, q> 1, and the fraction E is irreducible.
q
Exercise 2.39. Determine the irrational functions among the ones given or ana-
lyzed in this chapter.

3In the proof, we use the notion of the limit of a function at +00, given in Definition 4.4.
Chapter 3

Sequences

3.1 Introduction
3.1.1 Basic notions
A sequence is a function a : N -+ R. It is usual to write

an := a(n), n E N and a = (an)nEN.

Then an is called the general term of the sequence a.

Definition 3.1. A real numbed! is the limit of a sequence (an)nEN if for every
E > 0 there exists a natural number no = no( E), such that for every n > no it holds
I an -£ 1< E, i.e.,

(VE > 0) (3no EN) (Vn E N) n > no :::} I an - £ 1< E. (3.1)

Then we write lim an

n->(X)
= c.
If a sequence (an)nEN has the limit C, we say that it converges to C. We say that a
sequence diverges if it does not converge. Let us consider two classes of divergent
sequences.

Definition 3.2. A sequence (an)nEN

• tends to plus infinity, if for every positive number M > 0 there exists a
natural number no = no(M), such that for every n > no it holds an > M (then
lim an = +=);
we write n->()()

• tends to minus infinity if for every real number M > 0 there exists a natural
number no = no(M), such that for every n > no it holds an < -M (then we
lim an = -=).
write n->()()

Definition 3.3. A sequence a = (an)nEN is bounded if there exists a positive real

number M such that for every n E N it holds lanl :::; M.

83
84 CHAPTER 3.

Definition 3.4. A point £ E R is an accumulation point of a sequence (an)nEN

if for every E > 0 and every n E N, there exists a number mEN, m > n (m
depending on nand E > 0), such that it holds I am - £ 1< E.

Theorem 3.5. A sequence converges iff it is bounded and has exactly one accumu-
lation point.

Theorem 3.6. If (an)nEN and (bn)nEN are convergent sequences, then it holds

lim (an ± bn) = n---+oo

• n---+oo lim an ± n--+oo
lim bn;

• n---+oo
lim (an· bn) = n---+oo
lim an· n--+oo
lim bn;

lim an
· a n n-+oo
• 11m - = - - -
n-+oo bn lim bn '
n-+oo
where bn f 0 for every n E N and lim bn f O.
n-+oo

Theorem 3.7.

• Assume (an)nEN and (bn)nEN are convergent sequences, with the property that
there exists a natural number no such that for every n > no it holds an ~ bn .
Then
lim an ~ n--+oo
n--+oo
lim bn-

• If for the sequences (an)nEN, (bn)nEN and (Cn)nEN there exists a number no
such that for every natural number n > no it holds

an ~ bn ~ Cn,

then the following implication holds

(n--+oo
lim an = n--+oo
lim Cn = £) :::} (lim
n--+oo
bn = f).
Definition 3.8. A sequence (fn)nEN is a Cauchy sequence if it satisfies the fol-
lowing condition

(VE > 0) (3no E N) (Vm, n E N) m, n > no :::} Ifm - fnl < E. (3.2)

This condition is often replaced with the condition

(Vf > 0) (3no E N) (Vn,p EN) n > no :::} Ifn+p - fnl < f. (3.3)

Of course, both in (3.2) and (3.3) the natural number no depends on f.

Theorem 3.9. A sequence of real numbers converges iff it is a Cauchy sequence.

This property of the set R is called completeness. 1

lOne can prove that the axiom (R15) from Subsection 1.1 is equivalent with the last statement,
which explains the term "completeness of R" in (R15).
SEQUENCES 85

3.1.2 Examples and exerCIses

Example 3.10. Show by Definition 3.1 that the following sequences have a limit
which is equal to zero.

1 n-l
a) a n -_. b) b =-_.
n2 + 2'
- ,
n n

1 1
c) c - . d) en -- _ ".
n - ~n + 3' n.

1
e) dn = - a> O.
n Ol '

Solutions.

a) We have to show that for every given positive number e there exists a natural
number no = no (e) such that the following implication holds

(\:In E N) (n > no =? Ian - 01 < e), i.e., (\:In EN) (n > no =? ~n < e).

From the last relation it follows that for every n which has the property n > ~,
. 11 I 1.
It holds ;:; - 0 = ;:; < e. ThIs means that we can take no = [1] + 1, wheree
~
[xl is the greatest integer part of x defined in Example 1.47.

Remark. It is sufficient to consider only the case when e > 0 is "small" (say when
o < e < 1). Namely, such a restriction does not loose on generality, because if there
exists no = no(c) such that for every n E N, n > no, it holds Ian - £1 < e (see
relation (3.1)), then

(\:leI 2: e) (\:In E N) n> no =? Ian -£1 < el.

Further on, we shall consider only the case 0 < e < 1.

b) For every given e > 0 there exists a natural number no = no( e) such that

n-l
(\:In E N)(n > no =? Ian - 01 < e), i.e., (\:In E N)(n > no =? -2-- < e).
n +2

Using the following relations

n- 1 n -1 n2 1
--<---n --<-<e
n2 + 2 n2 + 2 - n2 +2 n '

and the case a), we obtain that we can choose no such that no = [~] + 1.
86 CHAPTER 3.

c) For given c > 0 we shall find a natural number no = no(c) such that for every
n > no it holds

1~I<c.
From the implications

(1-:-<l=n1=+=3 - 01 = -«,nl+ 3 < c) =;. C: 3 < c3) =;. (n + 3 > c13) ,

it follows that we can take no = no( c) = [:3] + 1.
d) (1~_01=~<_1
n! n! 2n - 1
<c) =;. (In~«n-1)ln2.)
c

Thu, we can take n. ~ n.(e) ~ [lnl~t) 1+ l.

e) If a is a positive rational number, then we have

I ~-ol
nOi =~
nOi <c, hence
1
- < nOi
c
or
1
\IE < n.

So we can take no = [ ~] + 1.
If a is a positive irrational number, then there exists a rational number f3 such
that it holds 0 < f3 < a and n f3 < n Oi . Hence

1:01 - 01 = :01 < :f3' and we can take no = [~] + 1.

Remark. The number no = no(c) from Definition 3.1 is not uniquely determined.
Namely, if for given c > 0 we can find no = no(c) such that the implication in
(3.1) holds, then any natural number nl > no could also give the implication in
(3.1). This also means that in finding no(c) we can occasionally use rather rough
approximations. In fact, it is important to understand that the existence of no =
no (c) for given c > 0 is essential.

Example 3.11. Show by Definition 3.1 that the following sequences have a limit
which is equal to one.

a) -+ 1 .
a n -_ n b) bn -- nl(i
yu, a > O·, c) en = y'ii.
n+2'
Solutions.
SEQUENCES 87

a) From the relations

n+l_ l l=_I_<t:,
In+2 n+2
it follows that for each t: > 0, it holds
1 1
n+2 > ~ or n> - - 2.
t: t:
Therefore we can take for no any arbitrary natural number greater than the
number lit: - 2. For example, put no := no(t:) = [~] + 1. Then for every
n > no it holds
n +-1 -1 I < t:.
I-
n+2
b) If we suppose that a = 1, then it is trivial that lim bn = 1.
n--+oo

If we suppose that a > 1, namely a = 1 +b, b > 0, then for every n EN, n > 1,
there exists a real number c( n) > 0 such that it holds

\11 + b = 1 + c(n), i.e., 1 + b = (1 + c(n)t.

Using the Bernoulli inequality

(1 + kt 2 1 + nk, k> -1, n E N,

we obtain 1 + b 21 + n· c(n), wherefrom we have c(n):::; !!.-. This means that

n
for every t: > 0 it holds
b
I\11 + b - 11 = c(n) :::; - < t:,
n
provided that no := [blc] + 1.
. 1
If we suppose that 0 < a < 1, l.e., a = h' h> 1, then from Theorems 3.6 and
3.7 it follows that
1 = 1.
lim
n--+oo yra = lim Vh
n--+oo

Thus lim
n--+oo
yra = 1 for any a > O.

c) Using the binomial formula

vn = 1 + d(n), d(n) > 0, l.e.,

n(n - 1)
n = (1 + d(n)t = 1 + nd(n) + 2 d2 (n) + ... + dn(n),
we obtain the estimation

n>
n(n - 1) d2 (n),
2 wherefrom we have d(n) < J~
n l'
88 CHAPTER 3.

For every e > 0, it holds

Iyrn - 11 = d( n) < J 2
n -1
< e, provided that n > [~e + 1] .
For given e choose no = [:2 + 1] + 1.
Example 3.12. Prove the following limits:

0, Iql < 1;
r qn = { 1, q = 1; b) lim nbqn = 0, Iql < 1, bE R;
a) n~~ +=, q > 1;
n .... =

an na
c) lim -, = 0, a E R; d) lim -, = 0, a E R.
n---+oo n. n-+oo n.

Solutions.

a) First of all, let us consider the case when Iql < 1. We can write then
Iql = 1 h'
1+ h > 0.

Then, for given e > 0, using the Bernoulli inequality, the following estimation
holds
1 1 1 1
Iqn - 01 -- (1 + h)n <- - - < - < e'
1 + nh - nh for every n > he'

So we can take no(e) = [hI]e + 1. Thus it follows for Iql < 1 that lim qn
~=
= 0.
If q = 1, then it is obvious that lim qn = l.
n .... =
If q > 1, then we have

qn =~,
rn
°< .,. < 1, hence lim qn
n-+oo
= +=,
because lim
n .... =.,.n = 0.
Remark. In the case q :::; -1, the sequence qn does not converge. If in particular
q = -1, then it has two accumulation points 1 and -1.

°it reduces to a).

b) In the case when b < 0, the statement is trivial, while for b =
If b > 0, then we can choose k, kEN, such that k> b. For h > °and n > 2k
then it holds

(1 + ht > ( n ) hk = n(n - 1)··. (n - k + 1) hk > (~)k hk~

k k! 2 k!'
SEQUENCES 89

because
n
n > 2k ==?- n > 2k - 2 ==?- 2n > n + 2k - 2 ==?- n - k + 1 > 2'

If we take h > 0, such that Iql = ~h < 1, then from the previous estimations
1+
we have

Inbqn _ 01 __ n bl q In __ (1 +
nb
h)n
n b2_
<_ k k! -
nkhk -
2kk! b-k
--y;;;-n <
2 k k'. b-k
--y;;;-n 1 < c:,

for every
n> [k_b1Wl
V2kkf +1,
where b1 E Q such that b < b1 < k.

c) If it holds that lal < 1, then from a) and Example 3.10 d) it follows that
an
lim -, = O.
n~oo n.
If lal > 1, then it holds

I
an
n! -
1_ -;J
lanl _ lal lal lal lal lal lal
- T . 2 . 3' .. k . k + 1 .,. -:;;'

where kEN is chosen sufficiently big in order to satisfy the following estima-
tions
~ >1
k -
and ~
k +1
< l.
Therefore we can write

o < I~I < ~. ~. ~ ... ~. (~)n-k = M. (~)n-k

- n! 1 2 3 k k +1 k +1 '

where M is a constant. From a) it follows that

lim M. I I )n-k =
( _a_ O.
n->oo k+1

From the last two relations and Theorem 3.7 it holds that lim an, exists and
n ...... oo n.
is equal to zero.

d) Let us suppose that a > 0 and let the natural numbers k and n satisfy the
following inequalities
n n
k > a and n > 2k, hence n - k + 1 > 2' ... , n > 2'
90 CHAPTER 3.

Then from the estimations

na nk nk
n! < n! (n - k)! (n - k + l)(n - k + 2) ... n

1 nk 2k
<
(n - k)! (%r - (n - k)!'
.----

it follows that for every E > 0, there exists a natural number no = no(E) such
that it holds
Inal
2k 2k
n! < (n - k)! < n _ k < E,

for every n > no. For example, no = k + 1 + [2 k /EJ.

Exercise 3.13. Show by definition that the following sequences converge to zero.

a) fn = (_l)n 1995
n 1995 ; b) fn = y'n;

c) fn = 1995 n . d) fn = n 1995
f •
n! ' n.
Exercise 3.14. Prove the following.

a) n-->oo
lim n 5 = +00; b) li.~ (~r = +00;

c)
nf
lim - ' = +00;
n-->oo 12 n
d) lim (20 -
n-+oc> n
n) = -00;
e) n-->oo
lim n 1 / k = +00, kEN; f) lim
n-->oo ifrJ = +00.

Example 3.15. Determine the following limits.

2n 5 - 3n 2 + 1 b) lim 3n4 + 2n 2 + 1
a) lim
n-->oo n 5 + 3n + 2 ; n-->oo n3 + 1

c) lim 8n 2 + 3n + 1 d) lim (2n + 1)3 - 8n 3

n-->oo n3 +1 n-->oo n2 + 1

e . (n2 + 2n + 3 -
11m n3 + 5 ) .
f) lim(y'n+2-y'n);
) n-->oo n+ 1 n 2 + 2n + 1 ' n-->oo

g) lim(v'n 2 +2n-n);
Vn+T
n-->oo h) li.~ Vn+2 + y'n+3'
Solutions. We shall repeatedly use Theorem 3.6.
SEQUENCES 91

2n 5 - 3n 2 + 1
5 _ _ ( 3 1)
. n 2 - n3 + n5 _ n-+oo
lim 2 - - + -
n3 n5
(3 1)
a) lim = hm ( 3 2) - ( 3 2 ) = 2.
n-+oo n 5 + 3n + 2 n-+oo n5 1 + - + - lim 1 + 4" + 5"
n4 n 5 n-+oo n n

. 3n4 + 2n 2 + 1
11m . n 4 (3 + 22 + -;)
b) = hm n n
n3 + 1 1) = +00.
n 3 1 +-3
n-+oo n-+oo (

2 n 2(8+-+-
3 1) 3
8+-+- 1
c) lim 8n + 3n + 1 lim n n 2 = lim..!:... n n2 = o. 8 = O.
n3 + 1 1 ) n-+oo n 1
n3 1 + - 3 1 +-3
n-+oo n-+oo (

n n

· (2n+1)3-8n3
11m 1·8n3 +12n 2 +6n+I-8n 3
d) = 1m = 12.
n-+oo 2 n +1 n-+oo n2 + 1

. (n2 + 2n + 3 -
11m n3 + 5 ) 1. n 3 + 3n 2 + 5n + 3 - n 3 - 5
e) = 1m
n-+oo n+ 1 n 2 + 2n + 1 n-+oo (n + 1)2
2 n2 ( 3+ 5 - - 2- )
lim 3n + 5n - 2 = lim n n 2 = 3.
(n + 1)2 1) 2
n 2 1 +-
n-+oo n-+oo (

f) E.~ (Vn+2 - vn) = lim (v'n+2 - yin) (v'n+2 + yin)

2 n-+oo vn+2 + fo)
(

lim
n-+oo (v'n+2 + yin) -- 0.

g) li..~(vln2+2n-n)= lim (vn~n-n)(vn2+2n+n)

n-+oo vin 2 + 2n+n
')n
- = l.
li..~ n ~(V---=~=+~-,---+ 1)

yin. VI + ~
n_~,;n . (~r.3)
1
h) -
v'ii+l
,!i.'1t, vn+2 + v'n+3 = lim
V1+ ;; + V1+ ;; 2
92 CHAPTER 3.

Example 3.16. Determine the limit of each of the following sequences.

1+2+3+"'+n 12 + 22 + 3 2 + ... + n 2
a) b)
n2 n3
1 . 2 +2 . 3 + " . + n( n + 1) . 1 + 3 + 5 + ... + 2n - 1 2n +1
c) d) ---
n3 n+1 2

1 3 5 2n - 1 1 +a + ... + an
e) -
2
+ -22 + -23 + " . + -2-
n
'' f)
l+b+,,·+bn '
o < lal, Ibl < 1.

Solutions.

a) Using mathematical induction we have (see Example 1.26 a»

lim 1 + 2 + 3 +". +n = lim n(n + 1) = ~.

n->oo n2 n->oo 2n 2 2

b) Using mathematical induction we have (see Example 1.26 c))

lim 12 +2 2 +3 2 +"'+n 2 = lim n(n+1)(2n+1) =~

n->oo n3 n->oo 6n 3 3

c) In this case we can write

. 1·2+2·3+3·4+,,·+n(n+1)
hm - - - - - - - - - - ' - - - - - - ' -
n~oo n3

= lim 1(1+1)+2(2+1)+3(3+1)+"'+n(n+1)
n--+oo n3

. 12 + 22 + 3 2 + ... + n 2 + 1 + 2 + 3 + ... +n
hm - - - - - - - - -
n--+oo n

n(n + 1)(2n + 1) n(n + 1)

= lim 6n 3 + 2 1
n->oo n3 3

d) In this case we have (see Example 1.26 b))

. ( 1 + 3 + 5 + ... + 2n - 1 2n + 1)
11m ---
n->oo n +1 2

lim
n->oo
(~_ 2n +
n +1 2
1) = lim -3n -
n->oo 2(n + 1)
1= _ ~. 2
SEQUENCES 93

1 3 5 2n - 1
e) Putting fn = "2 + 22 + 23 + ... + ~' (using Example 1.24) we have

fn f _ fn = ~ + (~ _ ~) + ... + (2n - 1 _ 2n - 3) _ 2n - 1
2 n 2 2 22 22 2n 2n 2n+1

1
1 (1 1 1 ) 2n - 1 1 1 1- 2n"=1 2n - 1
"2 + "2 + 22 + ... + 2n- 1 - 2n+1 ="2 + "2 . 1 - 2n+1 .
1- -
2

Thus we obtain

lim fn
n->oo
. (
hm 1+2 1 - - -2n
n->oo 2n- 1 2
(
--n- 1) 1)
· -1- - 2 1·1m -n
3 - 11m 1·1m -1 = 3.
+ n->oo
n->oo 2n- 2 n->oo 2n 2n

f) In the numerator and the denominator we have geometric sums and therefore we
can write

1 __ an +1
. 1 + a + ... + an . 1_ a 1 - b . 1 - an+1 1- b
hm = hm hm ----,----,-
n->oo 1 + b + ... + bn n->oo 1 - bn+1 1 - a n--+oo 1 - bn+1 1- a
1- b

Exercise 3.17. Determine the limits of the sequences given below.

a) Xn = V16 + ~2; b) Xn = ( 16 + 1
2n r 1 3
/
;

2n+3 . d) Xn = JnT+T + Vn+I.

c) x -
n-J4n 2 +3' fln 3 + 2n + 12 + n'

e) Xn = Jn 2 + n - Jn 2 - n; f) Xn = fin + 1 - fin - 3.

Answers.

1
a) 4. b) 2y!2. c) 1. d) 1. e) 1. f) O.
94 CHAPTER 3.

Exercise 3.18. Determine the limits of the sequences given below.

V'3-3 . yI64 - 1
a) Xn = 1+ \11995' b) Xn = V'8 _ 1 '
1 2 y1n5 + V'5 .
c) Xn = 1 _ V'3 - 1 - \1'9; d) Xn = 5 yr:;;'i + ytSn'

e) Xn= n4+4n
n + 4n +!' f) Xn = C f;
9: 5

16n +n! . 3n2 - 2n + n 3

g) Xn= 4n +(n+1)!' h) Xn = (n 3 )!

Answers.
1 1
a) -1. b) 2. c) d) -.
2 3
1
e) 4' f) O. g) O. h) O.
Exercise 3.19. Determine the limits of the following sequences.
1 - 2 + 3 - ... + (2n - 1) - 2n
a) Xn = v'n2+1

b) Yn = 12 + 32 + ... + (2n - 1 )2
n3 ;

1 (1 1 1)
c) Zn = Vii J2 + V4 + V4 + v'6 + ... + ffn + V2n + 2 .

Answers.
a) -1. b) ±
3'
c) J2
2 .
Example 3.20. Determine the limit of the following sequences.

a)
1 1 1.
Xn = ~ + 3.5 + ... + (2n -1). (2n + 1)'

b) Yn = (1 - ;2) . (1 - ;2) ... (1 - ~2) ;

c) Zn ~ (1 - D ( D {- 1- n( n :+ 1) ) ;
23 - 1 33 -1 n3 - 1
d) w - - - . - - ... - -
n - 23 + 1 33 +1 n3 + 1.
SEQUENCES 95

Solutions.
a) In this case we have

Xn = ~ . (1- ~ + ~ - ~ + ~ - ... + 2n ~ 1- 2n ~ 1) = ~ . ( 1 - 2n ~ J'

and lim
n--+oo
Xn = lim
n--+oo
~
2
(1 __ 1_) = ~.2
2n + 1

b) Using the following transformation 1-~=

1 (j -1)(j
r + 1) ' j = 2,3, ... ,n,
J
we obtain

lim Yn
n--+oo
= lim
n--+oo
(1 - ~)
22
(1 - ~)
32
... (1 - n 2 2..)
= lim (~. 2·4 . 3·5 ... (n - 1)(n + 1)) = lim ~. n + 1 = ~.
n--+oo 22 32 42 n2 n--+oo 2 n 2

c) Since it holds

1 _ .. 1 = (j - 1)(j + 2) .
J(J + 1) j(j + 1) , J = 2,3, ... ,n,
2
we have

lim
n--+oo
Zn lim (1 -
n--+oo
~) (1 - ~) ... (1 _ n(n 1+
3 6 1)
)
2

lim (~. 2·5 ... (n - l)(n + 2)) = lim ~ . n + 2 = ~.

n--+oo2· 3 3· 4 n( n + 1) n--+oo 3 n 3

d) Taking for j = 2,3, ... , n

j3 -1 (j - 1) (j2 + j + 1) (j - 1) (j2 + j + 1)
j3 + 1 (j + 1) (P - j + 1) (j + 1)((j - 1)2 + (j - 1) + 1)'
we obtain
23 1 33
- -1 n3 - 1
lim Wn = lim - - . - - ... - -
n--+oo n--+oo 23 + 1 33 +1 n3 +1
. (1 . 7 2· 13 3· 21 4· 34 (n - 1) ( n 2 + n + 1) )
= l~~ 3·3· ~ . 5·13 . 6·21 ... (n + 1)((n - 1)2 + (n - 1) + 1)
. 2 n2 + n + 1 2
= hm -. =-
n--+oo 3 n(n + 1) 3
96 CHAPTER 3.

Exercise 3.21. Determine the limits of the following sequences.

al a2 an
a) fn=-+-+"'+
aOSI SlS2 Sn-ISn
,

b) gn = In (1 - ;2) + In (1 - ;2) + ... + In (1 - ~2) = ~ In (1 - j\) ,

where in a) (an)nEN is an increasing sequence of positive numbers and the sum
n
Sn := L ak diverges to +00.
k=O
Answers.
a) nlim
..... oo
fn = 1lao. Hint. Use the equalities
al 1 1 a2 1 1 an 1 1
aOSI ao - Sl' SlS2 Sl - S2' ... ,
Sn-ISn Sn-l - Sn'

b) Hint. Use the following transformation

tIn
J=2
(1 - ~2) = ir (1 - ~2) .
J
In
J=2 J
The logarithmic function is a continuous one and therefore we have

lim
n ..... oo
L
j=2
n
In (1) 1 -""72
J
= In lim
n ..... oo
II
n
j=2
(
1 -""72
J
1) 1= In -.
2

Example 3.22. Evaluate

vn 2 + 1 + vn 2 + 2 + ... + vn 2 + n '.
a) . (1
11m 1 1 )
n ..... oo

an
b) lim
n-+oo (1 + a)(l + a 2) ... (1 + an) , a> 0;

c) J~~ (3 _~) (-Win

Solutions.
a) Using the inequalities
n i l 1 n
Vn2 + n-< vn2+ 1 + vn2+ 2 + ... + vn + n-< ---;===
---;===
~ 2

and
n l'
~=1= 1m ~,
·n
11m
v n2 + n n ..... oo v n2 + 1
n ..... oo

from Theorem 3.7 we obtain

+
. (1
11m
n ..... oo vn 2 +1 vn 1+ 2 + ... + vn21+ n)
2
= 1.
SEQUENCES 97

b) If 0 < a < 1, we have

an
0< < an,
(1 + a)(1 + a 2 ) •.• (1 + an)
and using Example 3.12 a) and Theorem 3.7 we obtain
an
lim = O.
n-+oo (1 + a)(1 + a2 ) ..• (1 + an)
If a> 1 we have
an an an
0< < --:----
(1 + a)(1 + a2 ) ••• (1 + an) a· a
2 ••• an a n (n+lJ/2 '

thus
an
lim = O.
n-+oo (1 + a)(1 + a 2 ) ••• (1 + an)
If a = 1, then
·
11m 1 . l'1m - 1 = O.
n-+oo 2 . 2 ... 2 n-+oo 2n
'-v---'
n-terms
Therefore we can write for all a >0
an
lim = O.
n-+oo (1 + a)(1 + a 2 ) .• • (1 + an)
c) Denoting the general term of the given sequence by h n , we have

I)-l/n
( 3 -:;; , n = 1,3,5, ... ;
{
hn =
(3 - :;;1) l/n
, n = 2,4,6, ....

Thus we obtain the inequality

3- l / n < h n < 3 l / n .

Since by Example 3.11 b) it holds

lim 3- l / n
n~oo
= nlim
...... oo
3l / n = 1
'

we have lim h
n-+oo n
= l.

Example 3.23. Prove the implication

lim an
n--+oo
lim lanl = lal.
= a:::} n--+oo
Is the opposite statement true'?
98 CHAPTER 3.

Solution. From Definition 3.1,

('ic > 0) (3no E N) ('in E N) n > no => Ian - al < c,

it follows

('ic > 0) (3no E N) ('in E N) n > no => Ilanl-lall:s Ian - al < c,

and this means that lim lanl = lal.

n~oo

The opposite is not always true. For instance, the sequence given by

_(n+l)(-I)n, nEN,
an - n

does not converge, though the sequence (lanl)nEN does converge to 1.

Exercise 3.24. Examine whether the following implications are true.

a) lim j2n = j2
n---+(X)
=> n---+oo
lim fn = f;

b) lim
n--+oo
P =F
n
lim in = f.
=> n--+(X)
Answers.
a) No. For example, the sequence fn = (_I)n, n E N, does not converge, but
the stationary sequence f? = ((-I)n)2 = 1, n EN, converges to 1.

b) Yes.

Exercise 3.25. Show that there exist sequences (an)nEN and (bn)nEN such that
lim an = +00 and n--+oo
n--+oo
lim bn = +00, but

a) (an + bn)nEN converges; lim (an

b) n--+(X} + bn) = -00; lim (an
c) n--+oo + bn ) = +00;
d) the sequence (an + bn)nEN neither converges nor diverges to +00 or to -00.

Answers. For example, the following sequences can be used.

1 1
a) an = -
n
+ n, bn = - - n;
n
lim (an
n--+oo
+ bn ) = lim
n--+oo
~n = O.

b) a n = -+n,
1
n
1
bn = - - 2n;
n
lim (an
n-+oo
+ bn ) = lim
n--+oo
(~-
n n) = -00.

c) an = -+2n,
1
n
1
bn = -n - n; lim (an
n--+oo
lim (~+
+ bn ) = n--+CXl n n) = +00.
d) an = (_I)n + n, bn = -n; the sequence (an + bn)nEN neither converges nor
diverges to +00 or to -00.

Exercise 3.26. Show that there exist sequences (an)nEN and (bn)nEN such that
lim an = +00 and lim bn = +00, but
n--+oo n--+oo
SEQUENCES 99

. an . an
11m
a ) n~oo -b --
n
o·, b) lim abn = A, A> 0;
n .... oo
c ) 11m -
n-+oo bn
= +00;
n

d) the sequence (~n) neither converges nor diverges to +00 or to -00.

n nEN

Answers. For example, the following sequences can be used.

a) an = n, bn = n 2; lim -an = l'1m -1 = O.

n-+oo bn n--+oo n

b) an = 2n, bn = n; lim an = 2.
bn
n .... oo

c) an = n 2 , bn = n; 11m an = l'1m n = +00.

· -b
n--+oo n--+oo
n

d) an = (3 + (_l)n) n, bn = n; the sequence (~n) neither converges nor

n nEN
diverges to +00 or to -00.

Exercise 3.27. Show that there exist sequences (an)nEN and (bn)nEN such that
lim an = 0 and lim bn = +00, but
n--+
00 n--+oo

a) nlim
.... oo
an . bn = 0; b) nlim
.... oo
an . bn = A, A ~ 0; c) lim an . bn = +00'
n--+oo '

d) the sequence (an' bn)nEN neither converges nor diverges to 00 or to -00.

Answers. For example, the following sequences can be used.

1
a) an = 2'
n
bn = n; lim (an' bn ) = n--+oo
n--+oo
lim!..n
= O.

1
b) an = -, bn = 3n; lim (an' bn ) = 3.
n n .... oo

1
c) an = -, bn = n2; lim (an' bn) = +00.
n n .... oo

( _l)n
d) an = - - , bn = n 2 ; clearly, the sequence (an' bn)nEN neither converges nor
n2
diverges to +00 or to -00.

Exercise 3.28 .

• Assume that the sequences (an)nEN and (bn)nEN satisfy either lim an = +00
n .... oo
and lim bn
n--+oo
= +00, or lim an
n-+oo
= -00 and lim bn
n-+oo
= -00. Prove then

lim anb n = +00.

n .... oo
100 CHAPTER 3.

• Assume that the sequences (an)nEN and (bn)nEN satisfy either lim an =
n--->oo
-00
and lim bn = +00, or lim an = +00 and lim bn = -00. Prove then
n---+oo n--+oo n--+oo

lim anbn =
n--->oo
-00.
Exercise 3.29. Show that there exist sequences (an)nEN and (bn)nEN such that
lim an = 0 and lim bn = 0, but
n--+ 00 n--+oo

. an . an
11m . an
a ) 11m - = 0; b) -b -1· c ) 11m - = +00·'
n--+oo bn n--+oo -
n , n-+oo bn

d) the sequence (~n) neither converges nor diverges to +00 or to -00.

n nEN

Example 3.30. Examine whether the following are Cauchy sequences.

a) a = 1+1
- + -1
+ ...1
+ _.
n 4 2 9 n '

b) b
n
=sin-
2
1 sin 2
+-22
sin n
+ ... +_.
2n '

l' 2' ,
c) en = c;~ 2· + c;~ 3· + ... + n .c~::·I);

d) dn = 1 + ~ + ~ + ... + ~ (the harmonic sequence) ;

1 1 1
e) en =I+-+-+···+-.
ln2In3 Inn

Solutions.
a) Using Definition 3.8, we shall show that for every c > 0, we can find no = no(c)
such that for every n 2 no and for every pEN it holds la n+p - ani < c. First
we have

Ian
+p
- an I = 11 + 4~ + ... + ~ + 1 + ...
n 2 (n+lF
+ 1 - 1 - ~4 -
(n+pF
... - ~
n2
I
1 1 1
=
(n+lF + (n+2F + ... +---
(n+p)2
1 1 1
<
n(n+l)
+ (n+l)(n+2) + ... +(n+p-l)(n+p)
---
11 1 1 1 1 1 1
= - - - - + - - - - - + - - - - - + ... + ---
n n+l n+l n+2 n+2 n+3 n+p-l n+p
1 1 1
= ---- <-.
n n +p n
SEQUENCES 101

Then from the inequalities

1
lan+p - ani < - < e, for every n> [I/e] + 1,
n
it follows that we can take no = no(e) = [I/e] + 1. The sequence (an)nEN is
Cauchy's, hence from Theorem 3.9 it follows that it is a convergent sequence.

b) For given e > 0 and arbitrary pEN, we have

Ibn+p - bn I ISin(n2n+1+ 1) + ... + sin(n2n+p+ p) I

Isin(n+I)1 Isin(n+p)1
< 2n+1 + ... + 2n+p

1 1 1 1 1
< 2n+1 + ... + 2n+p S 2n+1--I = 2n <e
1- -
2

for every n > [~~~e] + 1. This means that the sequence (bn)nEN is Cauchy's.
c) The sequence (Cn)nEN is Cauchy's, because for given e > 0 it holds

cos(n + I)! cos(n + 2)! cos(n + p)! I

Icn +p - Cn I = I(n + 1) (n + 2) + (n + 2) (n + 3) + ... + (n + p) (n + p + 1)

1 1 1
< (n + I)(n + 2) + (n + 2)(n + 3) + ... + -'--(n-+-p-)(-n + p + 1)
1 1 1
~----,- - < <e
(n+I) (n+p+I) (n+I) ,

provided that n> [I/e] + 1.

d) In this case we shall show that the sequence (dn)nEN is not Cauchy's. This means
that there exists e > 0, such that for every no, there exist n > no and pEN
satisfying Idn+p - dn I > e.
Let us take e = 1/4. From

1 1 . 1 P
Id n+ -dnl = - - + - - + ... + - - > - - ,
p n+I n+2 n+p n+p
it follows that for n = p we have
1 1
Idn+p - dnl = Id2n - dnl 2: 2" > 4' for every n E N.
So the sequence (dn )nEN is not Cauchy's and does not converge.
102 CHAPTER 3.

e) The sequence (en)nEN diverges, because from the inequality In x < x, which holds
for x > 1, it follows that for c = 1/3 it holds

1 1 1 p P
Ie n+p -el-
n -In(n+1)
+
In(n+2)
+ ... +
In(n+p)
>
In(n+p)
>--.
n+p

For p = n we obtain
1 1
Ien+p - en I > "2 > 3"'
Therefore the sequence (en)nEN is not Cauchy's, hence it diverges.

3.2 Monotone sequences

3.2.1 Basic notions
Definition 3.31. A sequence (an)nEN is

• monotonically increasing (resp. monotonically nondecreasing) if for

every n E N it holds an < an+l (resp. an :S an+l);

• monotonically decreasing (resp. monotonically nonincreasing) if for

every n E N it holds an> an+l (resp. an 2: an+l)'

Theorem 3.32. A monotonically nondecreasing (resp. nonincreasing) sequence

bounded from above (resp. from below) converges.

3.2.2 Examples and exercises

Example 3.33. Show that the sequence

an = (1 + ~r, (3.4)

a) is monotonically increasing;
b) is bounded from above with the number 3.

Solution.

a) Let us show that the sequence given by relation (3.4) is monotonically increasing
in two ways .
• First method. We shall consider the quotient ~ and show that it is less
an+l
than l.

an ( 1 + ;:;
1) n ( n + 1)
-n- n
1 1 n +1
an+l - (1 _l_)n+l = n+2 . n+2 - (n(n+2))n' n+2'
+n+1 n+1 n+1 (n+1)2
SEQUENCES 103

Using the Bernoulli inequality

(1 + ht ~ 1 + nh, h> -1,

it holds
( n(n+2))n _
(n+l)2 -
(1- 1 (n+l)2
)n > I _ _
-
n
(n+l)2
and
1 n+l n 3 + 3n 2 + 3n + 1
~< n -------::----<l.
an+1 - 1- (n+l)2 n+2 n 3 + 3n 2 + 3n + 2

• Second method. Next we consider the difference an+! - an and show that
it is positive. We have

1 )n+1 ( l)n
an+! - an = ( 1 + n + 1 - 1 + :;;:

t ((
j=O
n -: 1 )
J
1 . _ ( n ) ~)
(n+l)J j nj
+ _1
(n+l)n+1

= t((n+l)n ... (n- j +2)_n(n-l) ... (n- j +l)) _1

j=l j!(n + l)i j! (n)j +.

n 1{(
L"7j" 1)
1· 1 - - - . (1 - -2)
- ... (1 -j-
-l-)
j=l J . n+l n+l n+l

- (1 - ~)
n
. (1 - ~)
n
... (1 - ~)
n
}+ 1
(n+l)n+1
> 0,

b ecause 1 - -j - > 1- )- an d ( 1 ) +1 > 0, .

J, n E
N.
n+l n n+ln
b) Using the binomial formula, we have
(1 + ~)n = 1 + 1 + n(n - 1) + ... + n(n - l)(n - 2)··· (n - (n - 1))
n 2!n 2 n!nn'
From the following inequalities

n(n - 1) = ~ (1 _ ~) < ~,
2n 2 2 n 2

n(n - l)(n -
3! . n 3
2) = ~3! (1 _~)n (1 _~)n < ~3! '

n( n - 1) ... (n - (n - 1)) = ~ (1 _ ~) (1 _ ~) ... (1 _ n -

n! . nn n! n n n
1) < ~,
n!
104 CHAPTER 3.

we obtain
1 1 1 1 1 1
(1+~f < 1+1+-+-+···+-<1+1+-+-+···+-
2! 3! n! 2 22 2n - 1

1+ 1 - (~r < 1+ 2 ~ 3.
1--
2

r ~ (1 ~r 2,
Note that since (an)nEN is monotonically increasing, it also holds

an = (1 + ~ + = n EN.
Remark. From Theorem 3.32 it follows that the given sequence converges. Its limit
is the irrational number e (see Example 3.36).
Example 3.34. Determine the following limits.

a) lim
n--+oo
(1 + -n1r n ; b) If ;
lim(I--
n--+oon

. en + 3) 3n+2 .

r
c) hm - 2 - , d) lim (n+lf
-- ;
n
n---+oo n- 1
n-->oo

1 n2
--1
2
e) lim (n2 - f) lim (n---)
n-->oo n2 + 1 n-->oo n +1

In Vn+T - In yin.
g) lim , h) lim n . (In Vn+T -In yin).
n-->oo n n-->oo

Solutions.

a) lim (1+!..)3n = (lim

n---+oo n n--+oo
(1 + !..)n)3
n
=e3.

b) lim (1 - -
1) = li..~ ( n1 )
n
n =. (
1
1) n 1
. lim - - 1
n-->oo 1 + __
1
n-->oo n -- hm 1 + --1 n _ 1
n- 1 n-->oo n-
1 1
-·1 = -
e e
¥.~
2n + 3 3n+2
c) n---+oo
lim ( -2n- ) = n---+oo
lim (1 + -2)
n . lim (1 + -)
3 3n
2n
n-+oo
3 2
= n--+oo
lim
(
1+ -2n
1 )
3
2n/3) 9/2
( lim
n-->oo
(1 + _1
2n
) .1 = e 9/ 2

3
SEQUENCES 105

( 1 +-l)n
d) lim (~)n _. (n (1 + ~))n
n-->oo n - 1 - J!...~ ( 7)
n 1--
lim n
n-->oo ( 1- - 1) n
n n

J~~ (1 + ~r
2
lim (1_1)(-n)(-l) =e •
n---+oo -
n
The last example can be done also as follows.

n+l)n
lim ( - -
n-->oo n - 1
= lim ( 1 + - - )n-l . lim ( 1 + - -
n-->oo n- 1 n-->oo n- 1
2 2)
. (
= hm 1 + - -
2) 2(n-l}/2
=
(. (
hm 1 + - -
2) (n-l}/2) 2
= e2 •
n-->oo n - 1 n-->oo n - 1

n2)
e lim
) "_00
n2-1
(n, +
-
1)
n2
= lim (
"_00
(1- 21 -)
(1+;!,
n
r 2 = e- 2 •

·
f) n--+oo
n- 1
hm (n -+)1
n2
.
= n--+oo
hm (
l)n)n
(( 1--
n
1 +-
n
1 r . -2 n
hm (e ) =
= n--+oo o.

· InJnTI-Iny'n . 1 (n+1)1/n 1 . ( l)l/n

g) hm = hm -In - - = -In hm 1 + -
n-->oo n n-->oo 2 n 2 n-->oo n
1
= -In 1 = o.
2

h) · n· (lnJnTI -lny'n) = hm
hm . 1
-In (n+l)n
- - = 1/2.
n--+oo n--+oo 2 n
Example 3.35. Prove the following estimations.

Eo JT < k. k!'
k 1 1
0< e - kEN. (3.5)

Solution. First we can write

( 1 + -l)n = Lk (n). 51 + Ln (n). 5'1 0 < k < n. (3.6)

n j=O J n j=k+l J n
The second addend can be estimated as follows.
t n(n-1)(n-2!, ... (n-j+1).~
Ln
j=k+1
(
J)
~ ~
nj
--
j=k+l J. nJ

J=k+l
~
L..J ~(
., 1 -
J.
1)(1 - -n2) ... (1 -
-
n
j-l)
- - .
n
106 CHAPTER 3.

The last expression is less or equal than

~! (1 _~) (1 _~) ... (1 _k: 1) .

. {_I
k+1
(1 _~)
n
+ 1 (1 _~)
(k+1)(k+2) n
(1 _~)
n
+ ...

+ (k+1)(k+2) k ) ( 1 -k+1)
1 ... (k+n)· (1-;;: -n- .. · ( 1 -n-1)}
-n- .

Thus we obtain
t
j=k+l
(~)~=
) nJ
t
j=k+l
n(n-1)(n-2) ... (n-j+1).~
j! nj

1( k+1
< k!
1 + (k+1)2
1 + ... + (k+1)n-k
1) < k!'
1 k+1'1_~
1 1
k+1
So it holds

Ln (n)
. 1
j<
1
k.k'· (3.7)
j=k+l ) n .
From relations (3.6) and (3.7) we obtain for 0 < k < n :

o< ( 1 + -l)n Lk (n)

- j=O )' 1
nj = . Ln (n)
j
1 1
n j < k. k! . (3.8)
n J=k+l

The last estimations imply

0::; 1. (1 + -n1) - l'

1m
n_oo
n ~
1m L..J
n_ooj~)
( n) 1 -1kk"
. ~ ::;
~ '.

In order to obtain relation (3.5), let us show

1 = k -.
1
lim
n_oo L )~
k

J=O
( )
~
nJ L).!
J=O
(3.9)

From the relations

+ 1)
(7 )~j n(n - l)(n - 2) ... (n - j
j! n j

~ (1 - ~) (1 _~) ... (1 _j : 1) ,
it follows that for fixed j E N it holds

lim (
n_oo
~)~
) nJ
~, '
= ).
SEQUENCES 107

wherefrom we obtain relation (3.9).

Remark. From the estimations given by relation (3.5), it follows that

· ~
11m 1 = 1·1m (1 + -1 + -1 + ... + -1 ) = e.
L.J -:-
n--+(X).J'.
3=0 n--+(X) l'. .2' • n'
Example 3.36. Prove that the real number e is not rational.

Solution. Let us suppose the opposite, i.e. that e is a rational number. Then it
can be written as e =~, where p, q E N, 2 < ~ < 3 and lcd(p, q) = 1. As
q q
usual, lcd(p, q) stands for the largest common divisor of the numbers p and q. If in
relation (3.5) we put k = q, then we obtain

1 1 1) <_.1
O<e- ( 1+1+-+-+···+-
2! 3! q! q. q!'
or

o < q [e q! - q' + 3i
( q! + q! + 2i q' + ... + q')]
qi < 1.

Taking e = ~ , we have
q

q' q' ... +-':'

O<q· [p(q-1)!- ( q!+q!+-':'+-':'+ q,)] <1 (3.10)
2! 3! q! .

It is obvious that the numbers

q! q! q!
p(q - I)!, q! and q'. + q'. + -2! + -3! + ... + -q!
are integers and therefore the expression

p(q - 1)'. - (q'. + q'. + -.:.2!q' + -.:.3!q' + ... + -.:.q!q,)

represents an integer. The supposition q > 1 is in contradiction with relation (3.10).

Remark. e ~ 2.718281828459045.
. 1
Example 3.37. Prove that hm
n~oo
nCi =
vn!
o.
Solution. Using mathematical induction, let us show first

n! > (~r, n E N. (3.11)

108 CHAPTER 3.

For n = 1 the inequality is correct. Let us suppose that it is correct for n = k, i.e.,

k! > (~) k • For n = k + 1 we have

(k + I)! = k!(k + 1) > ( "3k ) k (k + 1) = (

+
k 1) k+1 3

(1 + ~f'
(3.12)

In Example 3.33 it is shown that (1 + ~f < 3, so from relation (3.12) it follows

that
(k+1)!> (k;lf+l

So the relation (3.11) is correct. Therefore for given s > 0 one can find no such that
it holds
1
1 I

\Yn! = \Yn! <

1
iCir =;
1 3
< s,

for every n > no(s) = [3/s] +1.

Example 3.38. Let us consider the sequence given by
1 )n+l
an = ( 1 + ; , n EN.

a) Show that the sequence is monotonically decreasing.

b) Determine n---+oo
lim an'

c) Prove the inequality + -) <-.

-1- < In (11 1
n+ 1 n n
d) Prove the equality

lim (~+ _1_ + ... + ~) = In 2.

n---+oo n n +1 2n

e) Prove the equality hm. (1-+--+1 ... +-+

1 ... +-1) =lnk, kEN.
,,---+00 n n +1 2n kn
Solutions.
a) Using the Bernoulli inequality we have

( 1 )n+l
an 1 +; ( 1 )n+l n+1
a n +l 1 + _1_) n+2 = 1 + -n-:-(n-+-2-) n+2
(
n+1

> ( 1+ n +1 ) n +1
.--=
n 3 + 4n 2 + 4n + 1
>1.
n(n+2) n+2 n 3 +4n 2 +4n
SEQUENCES 109

This means that an > an+1, so the sequence (an)nEN is monotonically de-
creasing.

b) The sequence (an)nEN is monotonically decreasing and bounded from below (for
example with zero), and therefore from Theorem 3.32 it follows that this is a
convergent sequence and its limit can be ordered as

lim
n-+oo
(1+-n1)n+l = n-+oo
l ) n . ( 1+- =e.
lim (1+-
n
1)
n

c) From a), b) and Example 3.33 it follows

l)n
( 1+; (l)n+l
<e< 1+;

Since a logarithmic function is monotonically increasing, it holds

nln (1 +~) < 1 < (n + l)ln (1 +~),

wherefrom we have

In (1 +~)
n
< ~
n
and In (1 +~) >
n
_1_.
n+1
So we can write
-1- <In (1+-
1 ) <-.
1
n+ 1 n n
d) Using the inequality proved in c), we obtain the following inequalities for n > 1.

In (1 +~) <
1
n
< In (1 + _1_)
n-1
,
In(1+ n :1) < n+1 < In
1
(1 +~) ;

In C2: 1)
n <
1
2n
< In(~).
2n -1
Summing the previous inequalities, we get

L2n In ( 1 + k1 ) < L2n k1 < L2n In ( 1 + 1)

k_ 1 .
k=n k=n k=n
Using the transformations

L2n In (1
1 + k) = In
2n
II (1 + ~) = In (n + 1 . n + 2 ... 2n + 1)
k=n k=n k n n +1 2n'
110 CHAPTER 3.

we obtain
In (2n n+ 1) < f k=n
~ < In (~) .
k n - 1
The logarithmic function is continuous on the interval (0, +00) and therefore
we have
lim In
n--+oo
(2n n
+ 1) = lim In (~) = In2.
n--+oo n - 1
Using Theorem 3.7 we obtain finally
2n 1
lim
n--+oo
E -k = In2.
k=n

e) The proof is similar to the one in d), so we omit it.

Example 3.39. Prove that the sequence

n 1
an = E - -Inn
k=l k

is monotonically decreasing and bounded from below.

Solution. From Example 3.38 c) it follows

an - a n +l = In (1 + ~) - _1_ > 0,
n n+1

implying that an > a n +l, i.e., (an)nEN is a monotonically decreasing sequence.

The sequence is bounded from below, because (see Example 3.38 b))

an
n 1
E--lnn>Eln 1+- -Inn
n ( 1)
k=l k k=l k

In (2 . ~ . ~ ... n + 1 . ~) = In n + 1 > 0 .
2 3 n n n

From Theorem 3.32 it follows that the sequence (an)nEN converges and has a limit
lim an =: T' The limit T is called Euler's constant.
n--+oo

Example 3.40. Prove the following two inequalities.

a) (~r <n!<e(~r, nEN;

r
b) --<In(l+r)<r, rEQ, r>-l.
l+r

Solutions.
SEQUENCES 111

a) The proof of the left-hand side inequality is essentially the same to the proof of
relation (3,11) in Example 3,37, So let us prove the right-hand side inequality,
To that end, using the mathematical induction and the fact that (1 + ~) > 2 n

(n;lf
for n > 1, it follows
n!<
Further on we have

(n+1)n (l)n
( ~)n = e (!.?:)n, - 2 = e (!.?:)n, 1 + ~ < e (!.?:)n
2 2 e (%f 2 e 2

b) Let us take r = p/q, p,q E N, Then we obtain using Example 3,38 c)

In(1 + r) = In ( 1 + -p) = In (q-+ q +-

-1, - q +p )
2 , , , ---=----''---
q q q+1 q+p-1

In (1 + ~)q + In (1 + _1_)
q+1
+ ' " + In (1 + q+p-1
1 )
1 I I p
< -+--+'''+
q q+1 q+p-1
<-,
q
Thus
In(l +r) < r, (3,13)
Also, from Example 3,38 c) it follows
P
In(l + r) > -
1
- +-
IIp
+'" + -q+p > -q+p = - - = --,
q r
(3,14)
q+1 q+2 E+1 r+1
q
If we take -1 < r < 0 and r E Q, then we can put r = -rl, where 0 < rl < L
Denoting by r2 = _r_l-, it follows that r2 E Q and r2 > 0, From relations
1 - rl
(3,13) and (3,14) we obtain
r2
- - < In(1 + r2) < r2'
1 + r2

Putting r2 = _r_l- in the last relation, we get

1 - rl

rl < In ( 1 + -r-
l ) < -rl
-,
1 - rl 1 - rl
Then it follows

1+ 1-
-r- = -
-rl
1+
- < -In (
r
- - = In(l -
rl )
rl 1- rl
rl) = In(l + r) < -rl = r,
and these relations immediately give the inequalities in b),
112 CHAPTER 3.

Example 3.41. Show that the lollowing sequences converge.

a) In = F~Je + Je + ... ..fi =../e + In-I, II =..fi, e> 0;

1 (2) a
b) In+! = 2 a + In , ft = 2' 0 < a ::::; 1;

c) In+! = ~ (In + ;J, 11 =I 0, b ~ 0;

d) In+! = ~ ((m -1)ln + l;,nb- 1)' II =I 0, b> 0, mEN, n > 2;

e) In+!=ln-sinln, 0::::;ft<1r.
Solutions.
a) Using mathematical induction, we shall prove that this sequence is monotonically
increasing and bounded from above.
It is obvious that it holds h = Je+..fi > ft· Assume In> In-I. Then we
have
Je + In> Je + In-l =? In+! > In·
Next we show that for all n E N it holds In < ..fi + 1. Firstly we have
ft = ..fi < ..fi + 1. If we suppose In = ..fi < ..fi + 1 for some natural number
n, then it follows

In+l = J e+ In < /e + yC + 1 < J( yC + 1)2 = yC + 1.

Thus we showed that (fn)nEN monotonically increases and is a bounded
se-
quence. From Theorem 3.32 it follows that there exists a real number £ such
that it holds
lim In = £ and also
n~oo
lim In-l = £. n~oo

The limit £ can be ordered from

I; = e + In-l lim
=? n---+oo I; = c + lim In-l
n---+oo
=? £2 = e + £,
and it has the form
£ _ 1± v'I+Tc
1~ - 2 .
All terms of our sequence are positive and therefore the number
R
q =
1 - ../1 + 4c <0
2
can not be the searched limit. Hence the limit of the sequence (fn)nEN is

£ = £2 = 1 + v'I+Tc
2 .
SEQUENCES 113

b) We shall show that the sequence (fn)nEN is monotonically increasing and bounded
from above. It holds that in > 0 for every n E Nand

2)) a 1 (a)2 a
i2 = 21 ( a + (a4 = 2 + 2 2 > 2 = It.

If we suppose that in > in-lOr in - in-I> 0, then we have

2(fn+l - in) = i~ - iLl = (fn + in-l)(in - in-I) > O.

Further on, we have !t =~ :s ~ < 1. Assuming that in < 1, we obtain

1(a + i~ ) = 2a + j2; < 1.

i n+1 = 2
Thus the sequence is convergent and there exists an £ E R such that

lim
n--+-oo
in = £ = lim
n--+CX)
in-I'
The limit £ can be ordered from
1
£= 2 (a + £2) =} £1,2 = 2 ± J4=4a
So the limit of (fn )nEN IS
. 2 .

£=£2 = 1-~.

In this case we could not accept the number £1 = 1 +vr=-a > 1, (for 0 < a <
1) as the limit of (fn)nEN, because we have shown in < 1, for every n E N.

c) We have to analyze tree cases, depending on band !t'

• Case 1. If b = 0, then it holds

in = 2n!tl
-
and lim
n->oo
in = O.

• Case 2. If b > 0 and il > 0, then in > 0 for some n E N implies

in+1 = ~ (in + ;J > O.

So we have in > 0 for every n. From the transformations

b b
2in+l = in + in =} 2in+1 - 2Vb = in - 2Vb + in

=} 2(fn+1 - Vb) = (in _ {[) 2 > 0,

114 CHAPTER 3.

we can conclude that fn > 0 for n > 1. Using the last inequality we get

fn+1 - fn = 'l(b
2 fn - fn
) =
b-F:' < 0,
'12 . ~

wherefrom we obtain that the given sequence monotonically decreases (at

least for n :::: 2). Therefore this sequence is convergent and there exists
an £ E R such that
lim fn+1 = £ = n--+oo
n--+oo
lim fn.
The number £ can be ordered from the given recurrence relation

fn+! = ~ (fn + YJ .
Thus £= ~ (£ + D> O. Since fn > 0 for all n E N, we get finally
£= 0.
• Case 3. If we have b > 0 and f1 < 0, analogously it can be shown that
the sequence is monotonically increasing and it holds that fn < 0 for
every n E N. This means that the considered sequence converges with
the limit x = -0.

Remark. In the Cases 2 and 3, the sequence is not decreasing (resp. increasing),
but it is rather eventually decreasing (resp. eventually increasing); i.e., the sequence
becomes monotone starting from an index no. This situation happens, for instance,
when in Case 2 it is assumed 0 < II < 0.

d) In Example 1.33 we have shown that the arithmetic mean is not smaller than
the geometric mean of any finitely many positive numbers. Thus

b
(m - l)fn + jm-1
> m/ jm-1 . _b_ = Vb
fn+! = m n
Vn f~-l

for every n E N. The sequence (fn)nEN is monotonically decreasing because

fn+! - fn =m 1( b
f:::- 1 - fn
) b-f:::
= mf:::- 1 s:; o.

Similarly as in c), we can show that lim fn =

n->oo
'Vb.
e) Let us show that the given sequence is bounded. To that end, let us show
o s:; fn < 7r for every n E N. This is true for fl. Now if we suppose that
o s:; fn < 7r, then 0 s:; sin(fn) < fn, which implies
OS:; fn+1 = fn - sin(fn) s:; fn < 7r.
SEQUENCES 115

From the last relation it also follows that the sequence is monotonically de-
creasing and therefore it converges. The function sin x is continuous and the
limit i can be ordered from the equation

i=i-sini, (3.15)

which follows from

lim fn+! = i = nlim

n--+ (Xl --+
sin(fn) = sin i.
00

Equation (3.15) has two solutions on the interval [0,7r], namely i1 = 0 and
i2 = 7r. From the inequalities fn :::; it < 7r it follows that the limit is i = i1 = O.

Exercise 3.42. Prove that the sequences given by the following recurrence formulas
converge and determine their limits.

a) a n +1 = ';2 + an) a1 > 0; b) bn +! ="31 ( 2bn + 125)

b~ , b1 #- 0;

1
c) cn +! 1 - c~, C1 = 2' d) dn+! = dn (1 - dn), 0< d1 < 1.

Answers.

a) 2. b) 5. c)VS- I . d) O.

Example 3.43. Prove that the sequences (fn)nEN and (gn)nEN given by

fn+! = fn + gn f n gn > 0, > 0,

2 gn+! -- fn2 + gn , f1 gl

converge to the same limit.

Solution .

• First method. It is obvious that fn > 0 and gn > 0 for every n E N. Also it
holds that fn+! ;::: gn+! for every n E N (comparison between the arithmetic
and the harmonic mean - see Example 1.33). Therefore we have

f n+1 - f n = fn +
2
gn _ f
n
= gn - fn
2
<0
-, for every n> 2,n E N.

Thus the sequence (fn)nEN is monotonically decreasing, while from

gn+! 2fn > 1, for n = 2,3, ... ,

gn fn + gn
it follows that the sequence (gn)nEN is monotonically increasing.
116 CHAPTER 3.

The sequences (fn)nEN and (9n)nEN are bounded from below, resp. above,
because
fn+! 2': 9n+l 2': ... 2': 92 and 9n+!:::; fn+! :::; ... :::; h,
for every n E N. So these two sequences converge. Let us denote

f = lim fn
n~oo
and 9 = n--+oo
lim 9n·

Then we have

lim fn+l
n-+oo
= ~2 (lim
n-+oo
fn + lim 9n )
n-+oo
=;. f =f+
29 =;. f = 9.
From the equality
2fn9n
9n+l - fn + 9n
- ,
it follows
fn+19n+! = fn9n = fn-19n-l = ... = 1t91'
and we obtain

fn9n = f191 , n = 1,2, ... , hence f· 9 = f191 .

Finally we have
f =9 = jlt91'

• Second ll1ethod. From the equalities

fn+!9n+l = fn9n, fn9n = f191,

it follows
1 n+l
= fn
2
+ f191
2fn
=~
2
(I +
n
f191)
fn .

So we obtain a sequence of the same form as in Example 3.41 c).

3.3 Accumulation points and subsequences

3.3.1 Basic notions
Definition 3.44. Assume that a sequence (an)nEN is defined with the function a,
i.e., a(n) = an, n E N. A subsequence (anJkEN of the sequence (an)nEN is a
restriction of the function a to an infinite subset {nk IkE N} of the set N, satisfying
the condition nl < n2 < ... < nk < ....

Theorell1 3.45. Any subsequence of a convergent sequence converges and they both
have the same limit.
SEQUENCES 117

A subsequence (fnJEN of a sequence (fn)nEN which has the property fnk = £ for
some number £ and for all kEN is called stationary subsequence. Clearly,
then £ is an accumulation point of the sequence (fn)nEN.
Definition 3.46.

• The limes inferior of a sequence (an )nEN is the limit

I = nlim
..... oo
(inf{ am 1m;?: n}).

• The limes superior of a sequence of real numbers (an)nEN is the limit

s= nlim
..... oo
(sup{fml m;?: n}).

Each of the last two limits either exist in the (usual) sense of Definition 3.1, or is
+00 or -00 (see Definition 3.2). In any of these cases we shall write

I = liminf an and S = lim sup an.

n~oo n~oo

Let us note that the limes inferior and the limes superior of a sequence are resp.
the smallest and the largest accumulation point of that sequence.

3.3.2 Examples and exercises

Example 3.47. If at least one of the following conditions is fulfilled:
(apl) the set {nl fn = £} is infinite;
(ap2) for every e > 0, the set (£ - e,£ + e) n Unl n E N} is infinite,
then the point £ is an accumulation point for the sequence (fn)nEN' Prove.

Solution. If the set MI = {nl fn = £} is infinite, then its elements can be ordered
into a monotonically increasing sequence which diverges to +00. So we can put
fnk = £ for kEN and nl < nz < ... < nk < ... -+ +00. This means that for every
mEN there exist nk E M I , such that it holds nk > m. Then, for arbitrary e > 0,
we have
Ifnk - £1 = 1£ - £1 < e.
We obtain that from condition (ap1) it follows that £ is an accumulation point for
the sequence (fn)nEN.
If for every e > 0, the set (£ - e,£ + e) n Unl n E N} is infinite, then the set
Mz = {nl Ifn - £1 < e} can be ordered into a monotonically increasing sequence
nl < nz < ... which diverges to +00. This means that, for every mEN, there exists
kEN, such that nk E M z and nk > m. So for arbitrary e > we have °
Ifnk - £1 < e.
We obtain that from condition (ap2) it follows that £ is an accumulation point for
the sequence (fn)nEN.
118 CHAPTER 3.

Example 3.48. If the point £ is an accumulation point of the sequence (fn)nEN,

then at least one of the conditions (ap1) or (ap2) from Example 3.41 are fulfilled.
Prove.

Solution. Let us suppose that £ is an accumulation point for the sequence (fn)nEN
and that the condition (apl) is not fulfilled (this means that the sequence (fn)nEN
has no stationary subsequence (fnk )nEN with the property (fnk hEN = £, for every
kEN). We have to show that then condition (ap2) is fulfilled.
Let us suppose that this is not true, namely that there exists an 6 > 0 such
that the set (£ - 6, £ + 6) n Un I n E N} is finite. If n1 is the smallest natural
number such that n > n1 => fn f. £, then the supposition means that the finite set
(£ - 6, £ + 6) n Unl n E N} is either empty, or there exists a natural number n2 > n1
such that for every n > n2 => fn (j. (£ - 6, £ + 6). Hence this means that £ is not an
accumulation point for the sequence (fn)nEN, a contradiction.
Remark. The statements in Examples 3.47 and 3.48 can be together expressed as
follows.
A real number £ is an accumulation point for the sequence (fn)nEN if and only if
either the sequence (fn)nEN has a stationary subsequence whose every element is
equal to £, or every interval containing £ has an infinite number of terms of the
sequence (fn)nEN (or both).

Example 3.49. Find a sequence with

a) two accumulation points;

b) three accumulation points;

c) infinitely many accumulation points;

d) only one accumulation point, but the sequence diverges.

Solutions.

a) The sequence (fn)nEN , where fn = (_I)n, has two accumulation points £1 = 1

and £2 = -1, because Jzk = 1 and Jzk+1 = -1 for k = 0,1,2, ... (see condition
( ap 1) from Example 3.47).
n 27rn
b) The sequence (gn)nEN ,where gn = - - sin - - , n E N, has three accumu-
n+ 1 3
lation points. Namely, its stationary subsequence

3k
g3k = 3k + 1 sin(27rk) = 0

converges to 0 and this means that 0 is an accumulation point for the sequence
(gn)nEN. Further on, we have

·
11m +
3k 1 . (k
= l'1m -k--sm 2 7r + -27r) V3
=-.
k-->oo 3 + 2
93k+1
k-->oo 3 2
SEQUENCES 119

So V; is another accumulation point for the sequence (gn)nEN. Finally from

3k+2 . (
g3k+2 = 3k + 3 sm 2k1r
41r)
+ :3

it follows that - V; is the third accumulation point for the sequence (gn)nEN·
As another example of a sequence with three accumulation points, one can
n1r
take the sequence with the general term h n = cos 2' n EN.

c) We shall use the sequence whose accumulation point is every positive rational
number given by the following scheme

1/1 -+ 1/2 1/3 -+ 1/4 ... l/n ...

,/ /' ,/
2/1 2/2 2/3 2/4 ... 2/n ...
1 /' ,/
3/1 3/2 3/3 3/4 ... 3/n ...

k/1 k/2 k/3 k/4 ... k/n ... ,

or
h = 1, h = 1/2, h = 2/1, f4 = 3/1, f5 = 2/2, ....
d) The sequence (an)nEN, where an = n + (-l)nn, n E N, has only one accu-
mulation point, because it holds that a2k+1 = 0, for k = 1, ... (see Example
3.47 (ap1)).
But the sequence (an)nEN does not converge, because

a2k = 4k, hence lim a2k+l =

n-+oo
+00.

Exercise 3.50. Show that the sequence (fn)nEN, given by

2n
hn = 1 and f2n+l = -2--'
n+1
satisfies both conditions of Example 3.47.

Example 3.51. Assume that I = liminf an is a real number. Prove that then I
n-+oo
is the smallest accumulation point of the sequence (an)nEN.

Solution. The sequence

fn = inf{aml m > n}
is a monotonically nondecreasing one, and from Definition 3.46 it follows lim fn = I.
n-+oo
Then, for given E > 0, there exists a natural number no = no (E), such that it holds

n > no (E) =} I - E < fn ::::; I + E.

120 CHAPTER 3.

From the definition of the sequence (fn)nEN, it follows that for every n E N there
exists a number mn = mn(c) 2: no satisfying In ::; a mn ::; In + c. The numbers
mn can be chosen so that the sequence (mn)nEN becomes monotonically increasing.
Then for every mn it holds

I - c ::; a mn ::; I + c.
So I is an accumulation point of the sequence (an)nEN; we have yet to show that it
is also its smallest accumulation point. From the inequalities

n > no 1\ m > n :::} I - c ::; In ::; am

and since c > 0 is arbitrary, it follows that I is the smallest accumulation point for
the sequence (an)nEN.
Remarks.

1. If the sequence (an )nEN is not bounded from below, it is obvious that

lim (inf{aml m::; n}) =

n--+oo
-00.

In this case we usually write

liminf an = -00.
n--+oo

But it is easy to find an example of the sequence (an)nEN such that

liminf
n--+oo
an = +00.

For an example, take simply an = n, n E N.

2. In the previous example it was shown that the point I satisfying I = liminf an
n--+oo
is the smallest accumulation point of the sequence (an)nEN. So we can prove
that in this case for every c > 0 it holds that

• there are infinitely many terms an such that an < I + c;

• there are at most finitely many terms an such that an < I-c.

3. Similarly as in Example 3.51, it can be shown that the point S satisfying that
S = lim sup an is the biggest accumulation point of the sequence (an)nEN. So
n--+oo
we can show that in this case for every c > 0 it holds that

• there are infinitely many terms an such that an > S - c;

• there are at most finitely many terms an such that an > S + c.
SEQUENCES 121

Example 3.52. Determine lim sup fn and liminf fn for the following sequences.
n-+oo n-+oo

1
a) fn = [( _1)n -I)n + -, n E N;
n

b) fn = a + bn(-l)n
1 + n(-l)n, a < b, n E N·,

n2 • 21m
c) fn = -1--2 sm-, n E N;
+n 3
d) fn = 1 + (_I)n+! + 3(_I)n(n-l)/2, nEN.

Solutions.

a) From 12k = [(_1)2k -1)· 2k + 2lk = 2lk , it follows lim 12k

k--+oo
= 0; further on,

from f2k+! = [( _1)2k+! - 1)· (2k + 1) + - 21+k1 ,it follows lim f2k =
k--+oo
-00.

Thus it holds
lim sup fn = 0 and liminf fn = -00.
n--+oo n--+oo

b)
a + b· 2k
hk= 1. nl , k=I,2, ... , lim f2k = b=limsupfn;
k-+oo n-+oo
b
a+--
f2k-l = 2k 11, k = 1,2, ... , lim f2k-l = a = liminf fn.
1 + __ k--+oo n--+oo

2k -1
y'3(3k - 2)2 . y'3.
c) f3k-2 = 2(1 + (3k _ 2)2)' k = 1,2, ... , t:'I~,f3k-2 = 2 = h~~p fn;
_ -y'3(3k - 1)2 . -y'3. .
hk-l - 2(1 + (3k _ 1)2)' k = 1,2, ... , 1:.~ hk-l = -2- = h~~f fn;
(3k)2 .
hk = 1 + (3k)2 ·0, k = 1,2, ... , kI:.~ f3k = O.
d) f4k = 1 + (_1)4k+! + 3(_1)4k(4k-l)/2 = 3,

14k+! = 5 = limsupfn,
n--+oo
f4k+2 = -3 = liminf
n--+oo
fn and !4k+3 = -l.

Exercise 3.53. Prove that a necessary and sufficient condition for a sequence (fn)nEN
to converge is
lim sup fn = liminf in = lim in = i. (3.16)
n-+oo n-+oo n-+oo

Answer. The statement immediately follows from Theorem 3.5, which claims that
a sequence converges iff it is both bounded and has exactly one accumulation point.
122 CHAPTER 3.

Exercise 3.54. If (fnkhEN is a subsequence of a convergent sequence (fn)nEN,

prove then
lim inf fn ::; lim inf fnk ::; lim sup fnk ::; lim sup fn.
n~oo k~oo k-too n-too

Answer. Clearly, every accumulation point of a subsequence is the accumulation

point of the sequence, but not every accumulation point of the sequence is necessarily
an accumulation point of its subsequence.
Exercise 3.55. Prove Theorem 3.45, i.e., the following.
Any subsequence of a convergent sequence has the same limit as the sequence.

Example 3.56. Let us suppose that (an)nEN and (bn)nEN are bounded sequences
and satisfy the following condition:

an ::; bn for n 2: no·

Prove that then it holds

lim inf an ::; lim inf bn and lim sup an ::; lim sup bn .
n-too n-too n-+oo n--+oo

Solution. We shall prove only the first inequality.

Let us denote by

a = liminf an and b = liminf bn ,

n--+oo n--+oo

and suppose the opposite, i.e., a > b. This means that there exists a real number
d > 0 such that d = a-b.
Since a is the smallest accumulation point of the sequence (an)nEN, then there
exists nl E N such that it holds
d
an > a - 3 for all n > nl'

Since b is the smallest accumulation point of the sequence (bn)nEN, then for infinitely
many indices n E N it holds
d
bn < b + 3'
Hence for infinitely many indices n E N it holds
d d
bn < b + 3 < a - 3 < an, hence bn < an-

This is the opposite of the assumption an ::; bn , n E N, n > no.

Exercise 3.57. The sequence (Xn)nEN is given as follows.

n n7f
a) Xn=(-1)n+l(3+~); b) x = 1 + --cos-.
n n+2 2
Determine inf{xn\ n EN}, sup{xn\ n EN}, liminf Xn and lim sup Xn , and then
n-+oo n-+oo
compare them.
SEQUENCES 123

Answers.

a) inf{xnl n E N} = -4, liminfxn = -3, limsupxn = 3, sup{xnl n E N} = 5.

n---+oo n-+oo

b) inf{xnl n E N} = liminf
n-+oo Xn = 0, Sup{xnl n E N} = limsupxn = 2.
n-+oo
Example 3.58. If (fn)nEN is a sequence of positive numbers, prove then the fol-
lowing relations.
1 1
a) limsupfn = -liminf(-fn); b) lim inf -1 = lim sup fn
n-+oo n---+oo
n---+oo n n---+oo

Solutions.

a) Let us denote
limsupfn = L, L E R. (3.17)
n-+oo
Then for every c > 0, there are

• infinitely many terms fn such that fn > L - c;

(3.18)
• at most finitely many terms fn such that fn > L + c.
So from relations (3.18) it follows that for every c > 0 there are

• infinitely many terms - fn such that - fn < -L + c;

• at most finitely many terms - fn such that - fn < - L - E;.

The terms - fn belong to the sequence (- fn)nEN. Thus

liminf(-fn)
n-+oo = -L.

Remark. If lim sup fn = +00, then lim inf( - fn)

n-+oo
= -00.
n-+oo
b) Let us take L from relation (3.17) and assume L > O. Then from relations (3.18),
for every c > 0, c < L, there are
1 1 1 1
• infinitely many terms -1
n
such that -1 < - L
n
_
c
=: -
L
+ C1;
. 1 1 1 1
• at most fimtely many terms -1 such that -1 < - - =: - - C2,
n n L+c L
where C1,C2 > O. The terms l/fn belong to the sequence (l/fn)nEN.
This means that
1 1
lim inf fn
n---+oo
=L= lim sup fn
n---+oo

If, however, lim sup fn = 0, then for every c > 0, there are
n-+oo
124 CHAPTER 3.

• infinitely many terms fn such that fn < c

• at most finitely many terms fn such that fn > c.

This means that for every c > 0 there are

•
. fi mte
III
. Iy many terms -I
1 such t h at -1 > -,
1
n fn c
• at most finitely many terms fI such that II < ~.
n n c
So we have
1 __ 1
liminf -I =
n---+oo n
+00 - limsupfn
n---+oo

Example 3.59. If (In)nEN and (gn)nEN are two sequences of real numbers, then
the following inequalities hold.

lim inf fn
n-+oo
+ lim sup gn
n-+oo
a) liminf fn + liminf gn :::; liminf(Jn + gn) < {
+ lim inf gn
n---+oo n-+oo n---+oo
lim sup fn
n-+oo n---+oo

< limsup(Jn + gn) :::; lim sup fn + lim sup gn;

n---+oo n--+oo n---+oo

liminf fn . lim sup gn

n-+oo n-+oo
b) liminf fn ·liminf gn :::; liminf(Jn . gn) < {
n-+oo n-+oo n-+oo
lim sup fn . lim inf gn
n-+oo n---+oo

< limsup(Jn' gn):::; limsupfn ·limsupgn-

n-><XJ n-><XJ n---><XJ

In b) we additionally assume that fn >0 and gn > 0, for all n E N.

Solutions.

a) Let us denote by

f = liminf fn,
n-+oo
9 = liminf gn
n-+oo
and £ = liminf(Jn
n-+oo
+ gn);
F=limsupfn, G=limsupgn and L=limsup(Jn+gn)'
n---+oo n-+oo n-+oo

1. First we shall prove that f +g :::; £. If we suppose the opposite, i.e., £ < f +g,
then it holds d := (J + g) - £ > O. For every c > 0, there are

• at most finitely many terms fn < f - c;

• at most finitely many terms gn < 9 - c.
SEQUENCES 125

So it follows that, for every c > 0, there are

• at most finitely many terms in + gn < i + 9 - 2c. (3.19)

This holds also for c := ~. Since £ is an accumulation point o~ the sequence

Un + gn)nEN, there are infinitely many terms in + gn < £ + "3' But this is in
contradiction with relation (3.19). So we have proved the first inequality.
2. Now we shall prove that £ ~ i + G. For every c > 0 there are
• infinitely many terms in < i + C;
• infinitely many terms gn < G+c, but only finitely many terms gn > G+c.

So it follows that for every c > 0 there are

• infinitely many terms in + gn < i + G + 2c . (3.20)

Since £ is the smallest accumulation point of the sequence Un + gn)nEN, it

follows that £ ~ i + G.
3. Let us prove that i + G ~ L. Assume the opposite, i.e., L < i + G. Then
it holds U + G) - L = d2 > O. For every c > 0 there are

• infinitely many terms in > i - c;

• infinitely many terms gn > G-c, but only finitely many terms gn > G+c,

So it follows that for every c > 0, there are

• infinitely many terms in + gn > i + G - 2c . (3.21 )

Since L is the greatest accumulation point of the sequence (in + gn)nEN, there
are

• at most finitely many terms in + gn > L + ~2

and this relation is in contradiction with relation (3.21) for c = ~2.

4. Finally, we shall prove L ~ F + G. Let us suppose the opposite, i.e.,
L> F + G, then it holds L - (F + G) = d3 > O. For every c > 0 there are
• at most finitely many terms in > F + C;
• at most finitely many terms gn > G + c.
So it follows that for every c > 0 there are

• at most finitely many terms in + gn > F + G + 2c. (3.22)

Since L is an accumulation point of the sequence Un + gn)nEN, there are

126 CHAPTER 3.

• infinitely many terms fn + gn > L _ ~3.

But this relation is in contradiction with relation (3.22), for c = d3 /3.

b) We shall only prove that f . 9 :S C, where we used the notations

f = liminf fn
n~oo
and 9 = liminf gn,
n~oo
C = liminf(fn . gn).
n~oo

If at least one of the real numbers f or 9 is zero, then the inequality f 9 :S C

is trivial. Assume f > 0 and 9 > 0 and suppose that it holds C < f . g. Then,
for every c > 0, c < min{f,g}/2, there are

• at most finitely many terms fn < f - c;

• at most finitely many terms gn < 9 - c.

So it follows that, for every c > 0, there are

• at most finitely many terms fn . gn < f· 9 - c(f +9 - c).

This is in contradiction with the fact that C is an accumulation point of the

sequence (fngn)nEN. In fact, there exist infinitely many terms in every interval
surrounding C.
The other relations can be proved similarly as in a).

Exercise 3.60. The following sequences are given

a) fn = (-1 )n(n+1)/2 sin2 mr. gn = (_1)n(n+1)/2 cos 2 -

n7r'
2 ' 2 '
1
b) fn = 2 + (-1 )n; gn = 2 - (_l)n + 2"( _It(n+l)/2.

Determine the smallest and the largest accumulation point for every sequence and
determine

limsup(fn
n->oo
+ gn), liminf(fn
n->oo
+ 9n), limsup(fn . 9n),
n->oo
liminf(fn
n->oo
. 9n).

Answers.

a) liminf fn = -1,
n->oo
liminf
n->oo
9n = -1, liminf(fn
n->oo
+ 9n) = -1,

limsupfn = 1,
n->oo
limsupgn = 1,
n->oo
limsup(fn
n->oo
+ gn) = l.

1 3
b) liminf fn = 1,
n->oo
liminf
n->oo
gn = -2' liminf(fn
n->oo
. 9n) = 2"
7 9
lim sup fn = 3,
n->oo
limsuP9n
n-+oo
= 2' limsup(fn . 9n) =
n->oo
2'
SEQUENCES 127

Example 3.61. Two sequences (Xn)nEN and (Yn)nEN are given and the first one
converges, i.e., there exists an x, such that x = lim Xn . Then it holds
n->oo

a) liminf(xn
n-+CX)
+ Yn) = n-+oo
lim Xn + liminf
n-+oo
Yn;

b) if additionally Xn > 0 and Yn > 0 for all n EN, then

liminf(xn
n-+oo
. Yn) = n-+(X)
lim Xn ·liminf
n--+oo
Yn.

Solutions.

a) From Example 3.59 we have

lim inf Xn
n--+oo
+ n-+oo
lim inf Yn:S: lim inf(xn
n-+oo
+ Yn):S: lim sup Xn
n-+oo
+ n--+oo
lim inf Yn

and from the convergence of the sequence (Xn)nEN, i.e.,

lim Xn = lim
n-+oo
inf Xn = lim sup Xn
n--+oo n->oo

(see Example 3.53), we obtain the given equality.

b) Left to the reader.

Example 3.62. Prove the following relations:

inf{xnl n E N}:S: liminfxn:S: limsupxn:S: sup{xnl n EN}.

n-+oo n-+oo

Solution. We shall prove only the first relation. Since i := inf {xn I n E N} is the
largest lower bound of the last set, it holds that i :s: Xn for every n E N. Therefore
we have
lim (inf{xml m ::::: n}) = liminf
i:S: n--+oo n--+oo
Xn .

Remark. See Exercise 3.57 a).

Example 3.63. Let us suppose that the sequence (fn)nEN has two accumulation
points a and b and assume lim inf fn = -00 and lim sup fn = +00. Prove that then
n--+(X) n--+oo
there exist subsequences (nl)IEN, (nkhEN, (np)PEN and (nq)qEN of the set N such
that
lim fn!
1-+00
-00, lim fnk = a, k--+oo

lim fnp
p--+ 00
=b and lim fnq
g--+(X)
= +00.
Solution. We shall construct only two subsequences of (fn)nEN converging to a
and +00 respectively. First put

M = {n E NI fn = a}.
128 CHAPTER 3.

If the set M is infinite, then it can be written in a unique way as an increasing se-
quence (nk)kEN which diverges to +00. Clearly, (fnk)kEN is a stationary subsequence
of (fn)nEN which converges to a.
If, however, the set M is finite or empty, then there exists an no E N such that
('tin E N) n > no ::::? fn i- a.
Put n1 := no + 1 and C1 := IfnI - al/2. Since a is an accumulation point of (fn)nEN,
there exists a natural number n2 > n1 such that
fn2 E (a-Cl,a+cl).
Putting C2 := Ifn2 - al/2, there exists a term f n3 in the interval (a - C2, a + C2) ,
etc. Continuing this procedure ad infinitum, we obtain a subsequence (fnk) kEN of
the sequence (fn)nEN which converges to the point a.
A subsequence (fnq)qEN which diverges to +00 can be obtained from the fact
that the sequence (fn)nEN is not bounded from above. Namely, denote by nl the
smallest natural number such that fnl > 1. Next, put n2 for the smallest integer
greater than fnl + nl' Continuing in this manner we get a monotonically increasing
sequence (nq )qEN which diverges to infinity. This sequence is the set of indices of a
monotonically increasing subsequence (fnq)qEN which diverges to +00.
Exercise 3.64. If a sequence (Xn)nEN is bounded from below, then there exists
lim (inf{xml m ~ n}).
n ..... oo

If a sequence (Xn)nEN is bounded from above, then there exists

lim (sup{xml m ~ n}).
n ..... oo

Prove.
Example 3.65. Let (fn)nEN be a sequence and denote by
f1 + h + ... + fn
F n := , n E N, (3.23)
n
its sequence of arithmetic means. Prove then the following implication.
lim fn
n~oo
=f ::::? lim Fn
n--+-oo
= f.
Solution. From the condition nlim
..... oo
fn = f, it follows that

('tiC> 0) (3no E N) ('tin E N) n > no ::::? Ifn - fl < c/2.

Then we can write
IFn - fl = III + 12 + ... : fno + ... + fn - fl

1
< -(If1 - fl + 112 - fl + ... + Ifno-1 - fl + Ifno - fl + ... + Ifn - fl)
n

A n - no A
< - + --- . - < - + -,
C C

n n 2 n 2
SEQUENCES 129

where A = Ih - fl + Ih - fl + ... + Ifno - fl· There exists a natural number

nl = nl(s) such that
A s
n > nl => - < -.
n 2
Put n2 = n2(s) := max{no, nd. Then we obtain

(\is> 0) (3n2 E N) (\in EN) n > n2 => IFn - fl < s.

Remark. The opposite statement does not necessarily hold. For example, the
sequence fn = (_I)n, n E N does not converge, but lim Fn = 0, if Fm n E N are
n-+oo
defined as in relation (3.23).
Example 3.66. Let (gn)nEN be a sequence of positive numbers and denote by

Gn := n-+oo
lim ylglg2'" gn (3.24)

its sequence of geometric means. Prove then the following implication

lim gn = 9
n---+oo
=> n--+oo
lim G n = g,

provided that the limit 9 > O.

Solution. The logarithmic function is continuous, hence from

lim gn = 9 it follows that lim In gn = In lim gn = In g.

n--+oo n--+oo n---+oo

Since it holds
1
InGn = -(lngl + Ing2 + ... + lngn),
n
we can apply the previous example and obtain

lim In G n
n--+oo
= In g, hence finally lim.Jgl ... gn
n---+oo
= g.
Remark. The opposite does not necessarily hold. Find an example!
Example 3.67. Prove that if a sequence (gn)nEN has the property

lim (gn - gn-d = g, then lim gn = g.

n--+oo n-+oo n
Solution. Let us denote by fn = gn - gn-l, go:= O. Then

lim fn = n--+oo
n-+oo
lim (gn - gn-l) = 9

and let us calculate the sequence of arithmetic means of the sequence (fn)nEN. We
have
Fn = h + h + ... + fn = gl - go + g2 - gl + g3 - g2 + ... + gn - gn-l = gn.
n n n
From Example 3.65, it follows that
· -gn = l'1m 1'n
11m L'
= g.
n-+oon n--+oo
130 CHAPTER 3.

Example 3.68. Prove that if a sequence (hn)nEN of positive numbers has the prop-
erty
lim hhn
n---+oo n-l
= h, then lim
n---+oo
~ = h.

Solution. Let us denote gn = hhn,

n-l ho:= 1. Then from

G n = \/gl··· gn = rj _
hl
ho
... --
hn
hn-
=
1
nhh
V nn

and from Example 3.66, it follows that

nfh;.
lim V Il n
n---+oo
= h.

Example 3.69. Determine the limits of the following sequences, given with their
general terms.

a) fn = W. b) gn =
\j(kn)!
nL.
kEN;
n '

c) hn = ~y'(n+l)(n+2) ... 2n.

n
Solutions.
n!
a) If we denote by Xn := nn' then

Xn
n!
nn
(n - I)!
n(n - l)n-l = (n: 1 1 r- 1
1 )n-l.
Xn-l
(n - l)n-l ( 1+--
n-1

So we have lim ~ = ~e
n---+oo Xn-l
and from the previous example it follows

W 1
lim - - =-.
n-+oo n e

. (kn)!.
b ) In thIS case we put Yn = -k- and obtam
n n

(kn)!
Yn nkn (n - 1)kn kn· (kn - 1) ... (k( n - 1) + 1)
Yn-l (k(n - I))! n kn · (n-1)k
(n - 1 )k(n-l)

1 kn . (kn - 1) ... (k( n - 1) + 1)

=
((l+n~lr-lr
nk
SEQUENCES 131

Thus we have · -Yn-

11m kk an d fi nal1y
= -k
n...... oo Yn-1 e

:;j(kn)! = kk, kEN.

nlim
...... oo nk ek

4
c) lim h n = -.
n ..... oo e

Example 3.70. Let the sequence (Fn)nEN be the sequence of arithmetic means of a
sequence (fn )nEN. Prove that then it holds

liminf fn :::; liminf Fn :::; lim sup Fn :::; lim sup fn.
n~oo n~oo n---+oo n-+oo

Solution. We shall prove only that

lim sup Fn :::; lim sup fn.

n---+oo n---+oo

Let us denote by f = lim sup fn and assume that f ~ O. Then for every c > 0,
n ...... oo
there exists no = no (c) such that for every n > no it holds fn < f +c. So for n ~ no
we have
fl + h + ... + fno + ... + fn
Fn
n

f1+f2+"'+fno + (n-no)(f+c):::; B +f+c,

< n n n
where we put B := f1 + h + ... + fno' For given C1 > 0 there exists n1 E N such
that ~
n
< C1 for every n > n1. Let C2 > 0 be given. Then choose c := c2/2 and
also C1 := cd2. Let us put n2 := max{ no, nd. Then it holds

(\In E N) n > n2 ::::} Fn < f + C2,

and this means that lim sup Fn :::; f.
n ...... oo
The case f < 0 is similar and left to the reader.

Exercise 3.71. Let us suppose that (Pn)nEN is a sequence of positive numbers such
n
that nlim
...... oo
L Pk = +00, and let (an)nEN be a sequence of real numbers. Put
k=l

F n'- PIal + P2 a 2 + ... + Pnan N

n E .
PI + PI + ... + Pn
'- ,

(Fn is called the sequence of generalized arithmetic means of the sequence (an)nEN')
Prove then the following.

lim inf an :::; lim inf Fn :::; lim sup Fn :::; lim sup an.
n---+oo n---+oo n---+oo n--+oo
132 CHAPTER 3,

Thus if the sequence (an)nEN converges, then its sequence of generalized arithmetic
means converges to the same limit,

Example 3.72. The Stolz theorem.

Let us suppose that the sequence of positive numbers (Pn)nEN satisfies the following
two conditions
lim Pn = +00 and Pn+l > Pn-
n-+oo

If there exists lim ~n - ~n-l, prove then the following equality,

n--+CX) n - n-l

' -
l 1m Qn l' Qn - Qn-l ,
= 1m
n-+oo P n n-+oo P n - P n- l

Solution. Using the conditions on the sequence Pn , we can construct the following
two sequences

Pn = PI + P2 + '" + Pn and Qn = PIal + P2 a2 + ," + Pnan'

So we have
Qn - Qn-l _ Pnan _
....:..:.:---,:-.=-----=- - - - - an'
Pn - P n - l Pn
This means that
' Qn - Qn-l
l 1m I'
p p = 1m an'
n--+oo n - n-l n-+oo

If there exists lim an = £, then from the previous exercise it follows that the se-
n-+oo
quence of generalized arithmetic means converges to the same limit, i,e"

' Qn
I1m 0
n-+oo -p =.c
n

Example 3.73. Prove that the statement opposite to the one from the Stolz theorem
is not necessarily true, In other words, even if the limit

lim Qn = £ exists, then the limit

n-+oo Pn

' Qn - Qn-l
l 1m
n-+oo-=-'------=:-----
P n - P n- l
might not exist,

Solution. If we take, for example,

n br
Pn = nand Qn = Esin-, n E N,
k=l 3
then there exists
,7r ,27r , n7r
Q sm - + sm - + . , . + sm -
lim ~ = lim 3 3 3 = 0,
Pn
n--+oo n--+oo n
SEQUENCES 133

, , I'1m Qn - Qn-l
However, t h e Ilmlt p d oes not eXIst,
' b ecause
n---;(XJ Pn - n-l

, ' f Qn
I lmln p - Qn-l
p = I'lmln ' fsln- J3
' mf = --,
n---;(XJ n - n-l n---;oo 3 2

,
IImsup Qn - Qn-l
p p = I'lmsupsm J3
- = -,
,mf

n---;(XJ n- n-l n---;(XJ3 2

Example 3,74. Determine the following limits by using the Stolz theorem,

1 1
k 1+-+ .. , + -
1k + 2k + ' , , + n kEN i b) 9n = V2 Vn,
a) fn = nk+'
n '

lP + 3P + ' , , + (2n + l)P Inn

c) hn = n P+1 ' p E Qi d) Xn = nl / k ' k E Ni

1k + 2k + ' , , + n k _n kEN,
e) Yn = nk - k + 1'

Solutions.

a) Let us denote by Qn = 1k + 2k + ' , , + n k and Pn = nk+l, n E N, Then from

the equalities

, Qn - Qn-l
I1m nk
= lim ---:---,-------,--
Pn - Pn - l
n---;(XJ n---;(XJ nk+ l - (n - l)k+1

nk 1
= lim --,-----0-----,--,----:---0-------,----0--0--
n---;oo n k+l - n k+l + (k + l)nk -'" - (_l)k+l k+1

we obtain that , Qn
I1m I' f 1
n---;oo 1m n = -+
-P = n---;oo k I'
n

1 1 1
b) If we put Qn = 1 + V2 + J3 + '" + Vn and Pn = n, n E N, then we have

1
Vn- -- 0,
, 9n -- I'1m -Qn -- I'1m Qn - Qn-l -- -
I1m
n---;(XJ n---;(XJ Pn n---+(XJ Pn - Pn - l 1

2P
lim h n = p + 1
c) n---;oo
134 CHAPTER 3.

d) Putting Qn = In nand Pn = nl/k, n E N, we have

. Qn - Qn-l
11m = lim In n - In( n - 1) = lim In ( 1 + ~)
n-->oo Pn - Pn - l n-->oo n l / k - (n -l)1/k n-->oo ( ( l)l/k)
nl / k 1- 1- -
n

< E,n;, n -1
1
n'" (1- (1-;;) 1 l/k)

r 1 1 k-l ( 1 j/k
n~~ n _ 1 . n(l/k)-l L
)=0
1 - -)
n
= O.

Thus from Theorem 3.7 it follows that lim Qn - Qn-l exists and
n-->oo Pn - Pn - l

· Qn - Qn-l = 0 :::} l'1m -Qn = 0.

11m
Pn - P n - l
n-->oo n-->oo Pn

e) Let us denote Qn = (k+l)(lk+2k+ .. ·+nk)_n k+l and Pn = (k+1)n\ n E N.

Then we have
Qn lim -'...(k_+_1:....:)(,--1_k_+-,-2,-k_+_._.-,'+~n_k)~__n_k+_l
lim -
Pn
n-+oo n-->oo (k + 1 )nk

lim Qn - Qn-l = lim (k + l)nk - (nk+l - (n - l)k+l) =~

Pn - Pn- l
n-->oo n-->oo (k + l)(nk - (n - l)k) 2

Example 3.75. Let two sequences (an)nEN and (bn)nEN be given and let us define
the sequence (Cn)nEN by

Cn =
a1b n + aZbn - l + ... + anbl , n = 1,2, ....
n
lim an = 0 and Ibnl:::; B for every n E N, then it holds
a) If n-+oo lim Cn = O.
n-+oo

b) If n-+oo
lim an = a and lim bn = b,
n-+oo
then it holds lim Cn = a . b.
n-+oo

Prove.

Solutions.
a) From the implication
lim an
n-+oo
= o:::} n-+oo
lim lanl = 0,
and the Stolz theorem, it follows

lim
n-+oo
t lakl =
~n k=l lim lanl
n-+oo
= o.
SEQUENCES 135

Since the sequence (bn)nEN is bounded, i.e., Ibnl ~ B for every n E N, we have

lall + la21 + ... + lanl Band lim lenl ~ n-+oo

lim B ~ lakl = o.
lenl ~ n n~oo n L...J
k=l

This means that lim en = O.

n--+<XJ
b) Let us put
lim an = a and Xn = an - a for every n EN.
n--+<XJ
Then it holds lim Xn = 0 and
n--+<XJ
(Xl + a)bn + ... + (x n + a)b l
en
n

xlbn+···+xnbl bn+···+bl
~-~~+a· =fn+9n.
n n
From a) it follows that

lim f n = ll·m xlbn + ... + Xn b1

n~oo n--+<XJ n = 0,
while
b +···+bl
lim 9n
n---+oo
= n-+oo
lim a· n
n
= a . b.
Therefore lim en
n--+<XJ
= a . b.

3.4 Asymptotic relations

3.4.1 Basic notions
Definition 3.76. We say that a sequence (fn)nEN is asymptotically equivalent
to a sequence (9n)nEN as n tends to 00 if 9n -I- 0 for every n E N and it holds

lim fn = l.
n--+<XJ 9n
Then we write
fn rv 9n as n -+ 00.

Definition 3.77. Assume that (9n)nEN is a sequence with positive terms. We say
that

• a sequence (fn)nEN is big oh of (9n)nEN as n -+ 00 if there exist an no E N

and a constant K > 0 such that

Ifn I ~ K ·9n, for every n > no,

and we write fn = O(9n) as n -+ 00;

136 CHAPTER 3.

• a sequence (fn)nEN is small oh of (gn)nEN as n -+ (X) if

r in 0
n~~ gn = ,

and we write fn = o(gn) as n -+ (X).

In particular, if the sequences (fn)nEN and (gn)nEN diverge to +(X) and satisfy
the condition
fn = o(gn) as n -+ (X) -¢:=} lim fn = 0
n-+oo 9n '

then we say that the sequence (gn)nEN diverges faster to infinity then the sequence
(fn)nEN and we write
fn -< gn, as n -+ (X).

3.4.2 Examples and exerCIses

Example 3.78. Let us give the following two sequences (fn)nEN and (gn)nEN by

a) fn = n 4 + 3n 2 + 2 and gn = n4 ;
n
b) fn = ~ and 9n =-,
e

c) fn=lnn! and gn=nlnn.

Prove that in is asymptotically equivalent to gn, denoted by fn '" gn as n -+ (X).

Solutions. By Definition 3.76 we can write

fn '" gn as n -+ (X) -¢:=} lim fn = 1

n-+(X) 9n
and we have to check the last limit in all three cases.
r n 4 + 3n 2 + 2 1
) n~~
a n4 =.
b) We shall prove the given asymptotic equivalence in two ways .
• First method. Use Example 3.69 a) .
• Second method. We shall use the Stirling formula
n! = nn e -nJ27rn' e O/(12n), 0 < () < 1,

thus
n! '" nne- n J27rn as n -+ (X). (3.25)
This implies

. V'nI . \jnne -nJ27rneO/(12n)

hm
n-+oo
~ = hm
n-+oo
~ = 1.
e e
SEQUENCES 137

c) Using the Stirling formula we have

1 ()
In n! nlnn - n + -In(27rn) + -
lim - - = lim 2 12n = l.
n_oo n In n n--+oo nlnn

Example 3.79. Prove the following asymptotic relations.

a) n a --< n b for 0 < a < b; b) pn --< qn for 0 < p < q;

c) n a --< qn for a> 0, q > 1; d) qn --< n! for q> 1;

e) lnn--<na for a>O; f) n! --< nn.

Solutions.

e) There exists q E N, such that a> l/q. Then it holds

Inn Inn
-
n a <n-
1/ q •

From the relations

In(n+l)-lnn
-,----'---:--;-----:-;- =
In(I+~)n
n <
q(n+1)(q-1)/q ( l)n
In 1 + - ,
(n + 1)1/q - n 1/ q n((n + 1)1/q - n 1/ q) - n n

it follows
. In(n+l)-lnn
11m = O.
n-+oo (n + 1)1/q - n 1/ q
The last inequality we obtained from the following identity

1 = (yin + 1 - y'n) ({j(n + l)q-1 + -:j(n + l)q-2n + ... + V"'n q- 1 ) ,

which implies
1
yin +1 - y'n:::: q1(n + l)q-1

From the Stolz theorem we have

· -In1n/ = 0, . In n
11m an d t h·IS imp1·1es 11m - - = O.
n q
n---+oo na
n--+oo

Remark. Using this example, we can form a (partial) scale of growths for sequences
diverging to infinity as follows.

Inn --< n a --< n b --< qn --< n! --< nn, 0 < a < b, q> 1.

Example 3.80. If (fn)nEN is a given sequence such that fn > 1 and lim fn =
n-+oo
+00, then it holds qin >- fn' for q > 1.
138 CHAPTER 3.

Solution. Since Un] :s; in < Un] + 1 and qfn ~ qUnl, then

in
-qfn < "'---;-:-:-
[in] +1 nk +1
qUnl qnk

.
T h e sequence gIven by - +-
nk- 1 IS
. a su b sequence 0 f t h e sequence gIven
. by n + 1
qnk qn
From the limit
lim n+1 =0
n->= qn
it follows that
.
11m nk 1
--- =
+ 0 I.e.,
.
hm - j =0.
qfn
n--+CX) qnk ' n->= n

From this example we can form another scale of the growth of sequences which
diverge to infinity,
nG
na -< qn -< qq -< ... , a> 0, q> 1.
G

Example 3.81. Ii it holds q > 1 and an ?- bn , n E N, prove then

qa n ?-In.

Solution. From the following equalities

lim qa n = lim qan-bn = qan(l-bn/a n ) = +00,

qb n
n--+oo n--+oo

b
because lim ~ = 0, lim an = +00, we obtain
n-+oo an n--+oo

. qb n
hm - =0.
n-+oo qan

From this example we can again form the following scale of growth for the sequences
which diverge to infinity
qfo -< qn -< qn2 -< ... , q > 1.
Example 3.82. Ii (fn)nEN is a given sequence such that in >1 and lim in
n->=
= +00,
prove then
In in -< in.

Solution.
In in In(Un]+l) In(nk+1)
--< = .
in [in] nk
. In(nk + 1) . . In(n + 1)
The sequence gIven by IS a subsequence of the sequence gIVen by .
nk n
From Example 3. 79 e) it follows that

lim In(n + 1) = 0, I.e., 1·

1m
In in =
fn
o.
n--+oo n n-+oo
SEQUENCES 139

Example 3.83. If the sequences (fn)nEN and (fn)nEN satisfy fn > 0, 9n > 0,
prove then

a) 0(1) + 0(1) = 0(1); b) 0(1) + 0(1) = 0(1);

c) 0(1) = 0(1); d) O(fn) + 0(9n) = O(fn + 9n);

e) o(fn) + 0(9n) = o(fn + 9n),
where fn = 0(1) means that the sequence (fn)nEN is bounded and fn = 0(1) means
that the sequence (fn)nEN tends to 0, and (fn + 9n)nEN = (fn)nEN + (9n)nEN.
Remark. The asymptotic relations in all cases a) - e) are supposed to be read only
from left to the right.
Solutions.

a) The sum of two bounded sequences is a bounded sequence.

b) The sum of two sequences converging to zero is a sequence converging to zero

too.

c) A convergent sequence is a bounded one.

d) By Definition 3.77 we have

fn = 0(9n) ~ ((3M> 0) ( 3no E N) (n > no) Ifni:::; M9n).

Therefore, the expressions

Wn = O(fn), Un = 0(9n) mean Iwnl:::; Kfn, IUnl:::; M9n,

for some K > 0, M > 0 and n > no, for some no E N. Thus it holds for n > no

IW n + Unl :::; Kfn + M9n :::; L(fn + 9n),

where L = max{K,M}. This means that Wn + Un = O(fn + 9n).
e) The expressions

an = o(fn), bn = 0(9n) mean . -an

hm =0 an d l'1m -bn = O.
in
n-+oo n-+oo 9n

It holds If an I:::; Ifan I and 1f bn I:::; 1bn I· Then we have

n + 9n n n + 9n 9n

lim an + bn = lim ( an + bn ) = O.
n---+oo fn + 9n n---+oo fn + 9n fn + 9n
So we obtain o(fn) + 0(9n) = o(fn +9n).
140 CHAPTER 3.

Exercise 3.84. If the sequences (fn)nEN and (9n)nEN satisfy fn > 0 and 9n > 0,
n E N, prove then
a) G(fn) + O(9n) = G(fn + 9n); b) G(fn)' G(9n) = G(fn . 9n);

c) G(fn)' o(9n) = o(fn . 9n); d) o(fn)' O(9n) = o(fn . 9n).

Example 3.85. If fn rv 9n' n --t 00, then it holds fn - 9n = O(9n). Prove.

Solution. The limit lim fn = 1, means that for every c > 0, there exists no = no(c)
n-+oo 9n

-
such that it holds
I~: 11 < c, for every n > no·
Thus we have

-c < fn - 9n < c, .
meamng l'1m fn - 9n 0
=, hence fn - 9n = 0 (9n ).
9n n-+oo 9n
Exercise 3.86. Prove that fn rv 9n as n --t 00, if the sequences (fn)nEN and (9n)nEN
are 9iven as follows.

) _ 2n 3 + 3n 2 + n +1 _ ~ .
a fn - 3n 2 + 1 ,9n - 3 n ,

n2 •
b) fn=I+2+3+"'+n, 9n= 2'
c) fn = ifr0 + n + 2, 9n = nifii;

2Pn P+1
d) fn=lP+3 P + ... +(2n+l)P, 9n= p+l' pEQ.

Exercise 3.87.
a) Does a "fastest" sequence (9n)nEN exist, i.e., such that
lim fn = 0, for every (fn)nEN?
n-+oo 9n

b) Does a "slowest" sequence exist?

c) If for two sequences in and 9n it holds that fn -< 9n as n --t 00, prove that then
there exists a sequence (Cn)nEN such that
fn -< Cn -< 9n as n --t 00.

Answers.
a) No, because it always holds fn -< f::.

b) No, because it always holds In In -< In.

c) If the given sequences have positive terms, then, for an example, one can take
Cn = Jln9n.
Chapter 4

Limits of functions

4.1 Limits
4.1.1 Basic notions
For the definition of the limit of a function at a point, we need the notion of the
accumulation point of a set, see Definition 1.76 c).

Definition 4.1. Let Xo be an accumulation point of the domain A c R of a function

f: A ----+ R. Then the number L is the limit of the function f as x approaches
Xo iff for every c > 0, there exists a 8 > 0, 8 = 8( c), such that for every x E A with
the properly
0< Ix - xol < 8 it holds If(x) - LI < c.

Then we write f(x) ----+ L when x ----+ XO, x E A, or rather

lim f(x) = L.
x--+xo, xEA

Note that we do not need the function f to be defined in the point Xo. Even if
Xo E A, the value of f at Xo is irrelevant.
Using logical symbols, Definition 4.1 can be expressed as follows.

lim f(x) = L
x--+xo,xEA

~ (Vc> 0) (:38> 0) (Vx E A) 0 <I x - Xo 1< 8 =} If(x) - LI < c.

Definition 4.1 is equivalent to the following.

Definition 4.2. Let Xo be an accumulation point of the domain A of a function

f : A ----+ R. Then the number L is the limit of the function f as x approaches
Xo if for every sequence (Xn)nEN from A \ {xo} converging to Xo it holds

lim f(x n ) = L.
n-+oo

141
142 CHAPTER 4.

If in Definition 4.1 we take only the values x E A greater (resp. smaller) than Xo,
we get the definition of the right-hand limit (resp. left-hand limit) of a function
f: A ~ R at the point Xo. The right-hand limit of f at Xo is denoted by

lim f(x),
x--+xo+, xEA

while the left-hand limit of f at Xo is denoted by

lim f(x).
X--+xo - , xEA

Theorem 4.3. Let Xo be an accumulation point of the domain of a function

f : A ~ R. If both the left-hand and the right-hand limit of f at the point Xo exist,
then a necessary and sufficient condition for the existence of the limit of f at Xo is
the equality
lim f(x) = lim f(x) =: L. (4.1 )
X--+X'o-,xEA x--+xo+,xEA

Clearly, (4.1) is equivalent to lim f( x) = L.

x--+xo,xEA
Most often, the domain A of the function f will be its the natural domain. In
that case, we shall simply write

lim f(x), lim f(x) and lim f(x)

X--+XO X--+Xo+ X--+Xo-

for the limit, right-hand limit and the left-hand limit of the function f at the point
Xo·

Definition 4.4 .

• Assume the domain A of the function f : A ~ R contains the interval (a, +(0)
for some a E R. The number L is the limit of f at +00 if for every c > 0
there exists a number T > a, T = T(c), such that for every x > T it holds
1f(x) - L 1< c. Then we write

lim f(x)
x--++oo
= L. (4.2)

• Assume the domain A of the function f : A ~ R contains the interval (-00, b)

for some b E R. The number L is the limit of f at -00 if for every c > 0
there exists a number T < b, T = T(c), such that for every x < T it holds
1f(x) - L 1< c. Then we write
lim f(x)
;&--+-00
= L. (4.3)

In Definition 4.4 x went in the first case to plus infinity through increasing
positive values, while in the second case it went to minus infinity through decreasing
negative values.
LIMITS OF FUNCTIONS 143

Definition 4.5. Assume the domain A of the function f contains the interval (xo, b)
and assume that for every T > 0, there exists a 8 > 0, 8 = 8(T), such that for every
x E A and x E (xo, Xo + 8) it holds f( x) > T. Then we say that the function f
tends to plus infinity when x --t xo+, and write

lim f(x) = +00. (4.4)

x~xo+

Analogous meanings have the following notations:

lim f(x) = -00, lim f(x) = +00 and lim f(x) = -00.
+
x---+xo X---+XO - x--+xo-

Relation of the limit of a function with the basic operations and inequalities gives
the following statement.

Theorem 4.6. Assume the functions f and 9 are defined on a set A c R and let
Xo be an accumulation point of A. Moreover, assume that the following two limits
exist:
lim f(x) = L and lim g(x) = K.
X---+XO, rEA X---+XO, rEA

Then it holds

• lim
x--+xo, xEA
(J(x) ± g(x)) = L ± K;

• lim
x---+xo,xEA
(J(x)· g(x)) = L· K;

lim f(x) L
• x--->xo,XEAg(x) K'
where we additionally assume that there exists a 8 > 0 such that g( x) =1= 0, for
all x in the set (xo - 8, Xo + 8) n A and, moreover, K =1= 0;

• (38) 0) ("Ix E A) (0 < Ix - xol < 8 =} f(x) ::; g(x)) =} L::; K.

The last equalities and inequalities remain true if the point Xo is replaced with one
of the symbols +00 or -00.

4.1.2 Examples and exercises

Example 4.7. Using Definitions 4.1 and 4.2, show that

x2 - 4
lim = 2.
x-->2 x 2 - 2x

Solutions.

• First, let us use Definition 4.1. For arbitrary c > 0 and x E (1,3), x =1= 2, we
can write
4 _ 21 = IX + 2 _ 21 = Ix - 21.
1xX22 -- 2x x x
144 CHAPTER 4.

So we can take 0 := c, and for x E (1,3) we have

(0 < Ix - 21 < o)::::::?

- 4
IxX22 -2x I
- 2 =
Ix - 21
--
0
< - = c.
x 1

Therefore, we have

(Vc > 0) (30 = c) (Vx E (1,3)) 0 < Ix - 21 < o::::::? IxX22 _ 2x4 - 2I < c .
• In order to apply Definition 4.2, let us consider a sequence (Xn)nEN such that
for every n E N, 1 < Xn < 3, Xn #- 2 and

lim Xn
n--+oo
= 2.
Then we have
X2 - 4 . _
11m Xn _+ 2 - n--+oo
lim Xn + 2
lim f(xn) = E.~ X2n n_ 2x n n--+oo X - --1'-'----- = 2.
n~oo
n n~~ Xn

7r
Example 4.8. Show that the function f(x) = sin -, x E R \ {O}, has no limit at
x
the point Xo = O.

.
So IutlOn. Let us cons]'der two sequences gIven. b
y Xn I= -d an Yn = 2 , n =
n 4n + 1
lim Xn = n-+oo
1,2, ... , which have the same limit 0, i.e., n---too lim Yn = O. But it holds

lim f(xn) = sin trn = 0,

n~oo
lim f(Yn) = sin 7r 4n
n-+-oo
+1 =
2
l.

7r
Therefore from Definition 4.2 it follows that there is no lim sin -.
x---+o x
Remark. Neither the limit lim f(x) nor lim f(x) exists.
x--+o+ x--+o-

Exercise 4.9. Prove that lim f( x) does not exist for the following functions.
x--+o

a) f(x) = cossgn (~); b) f(x) = sgn cos (~) .

Exercise 4.10. Assume the functions f and g have no limits at the point x = Xo.
Does this imply that the following limits also do not exist:

lim (J(x)
X---+XO
+ g(x)) and lim (J(x) . g(x)) ?
x--+xo

Answer. Not necessarily. The limits lim

x--+o
~x and x--+o
lim (-~)
X
do not exist, but it holds

lim
X---+O
(~+
X
(-~))
X
= limO = O.
x--+o
LIMITS OF FUNCTIONS 145

. 1
Example 4.11. Show that hm (
" .... 2 X - 2
)2 = +00.
Solution. Let T > 0 be given. Then from the relations
1
(x-2)2>T, (x-2?<!..
T

it follows that we can find 8 = Jr such that it holds

1 _ 8 1
o < Ix - 21 < v'T - :::}
(x-2)2>T.

From Definition 4.5 it follows that lim ( 1 'n = +00.

" .... 2 X - 2
Example 4.12. Determine the following limits.

2x 3 - 3x 2 - X - 2 + 12x6 + 3x3 .
x8
a) lim , b) lim
" .... 2 x-2 " .... 0 x 7 + 4x 6 + x 5 + x 3 '

3x 2 - 13x - 10
c) lim . d) lim ex +3 _ 3x +1) .
x .... 52x 2 -7x-15' " .... 0 x2- X x3 - X '

rI m
x3 - 8 x 7 + 12x 5 - 13x 4 + 5x 2 + 4x - 9
e) --' f) lim ;
" .... 2 x 2 - 4' x .... l x 5 - 4X4 + 3x 3 - 2x 2 + X + 1

. x 5 - 3x 4 + 5x 3 - 7x 2 + 4
g) hm - - - - - - - - - -
" .... 2 x 5 - 5x 4 + 8x 3 + x 2 - 12x + 4

Solutions.
l' (2
· 2X3_3x2_X-2 1· (x-2)(2x2+x+l)
a ) 11m
" .... 2 x - 2
= 1m
x .... 2 X - 2
= 1m 2x
" .... 2
+ x + 1) = II.

. x8 + 12x6 + 3x 3 . X3(X 5 + 12x 3 + 3)

b) hm = hm = 3.
" ....0 x 7 + 4x 6 + x 5 + x 3 " ....0 x (x 4 + 4x 3 + x 2 + 1)
3

c) lim 3x 2 - 13x - 10 = lim (x - 5)(3x + 2) = lim 3x +2 = 17

" .... 5 2x 2 - 7x - 15 " .... 5 (x - 5)(2x + 3) ".... 5 2x +3 13

d) lim (2X + 3 _ 3~) = lim 2X2 + 2x = -2.

" .... 0 x 2 -x x 3 -x " ....ox(x-l)(x+l)

e) lim x 3 - 8 = lim (x - 2)(x 2 + 2x + 4) = 12 = 3.

" .... 2x 2 -4 " .... 2 (x-2)(x+2) 4

f) If the value x = Xo is a zero of the polynomial Pn (x), then it holds

Pn(x) = (x - XO)Qn-I(X),
146 CHAPTER 4.

where the coefficients of the polynomial Qn-l (x) can be ordered by using the
Horner scheme (see Subsection 2.2.1). For the polynomial in the numerator it
holds
1 0 12 -13 0 5 4 -91 x = 1
1 1 13 0 0 5 91 o.
So we have

x7 + 12x 5 - 13x 4 + 5x 2 + 4x - 9 = (x - 1)(x 6 + x 5 + 13x 4 + 5x + 9).

For the polynomial in the denominator we have

1 -4 3 -2 1 11 x = 1
1 -3 0 -2 -11 0,

and it follows

x5 - 4X4 + 3x 3 - 2X2 +X+1 = (x - 1)(x4 - 3x 3 - 2x - 1).

So we obtain

. x 7 + 12x 5 - 13x 4 + 5x 2 + 4x - 9
hm - - - - - - - - - - - - -
x--+l x 5 - 4X4 + 3x 3 - 2x 2 + X + 1

lim (x - 1)(x 6 + x 5 + 13x 4 + 5x + 9) = _ 29.

x--+l (x - 1)(x4 - 3x 3 - 2x - 1) 5

g) In this case we have the following Horner scheme

1 -3 5 -7 0 41 x = 2
1 -1 3 -1 -21 O.

Therefore, it holds

x5 - 3x 4 + 5x 3 - 7x 2 + 4 = (x - 2)(X4 - x 3 + 3x 2 - X - 2).

Also, we have
1 -5 8 1 -12 41 x =2
1 -3 2 5 -21 o.
So, we can write

x5 - 5x 4 + 8x 3 + x2 - 12x + 4 = (x - 2)(X4 - 3x 3 + 2X2 + 5x - 2).

Therefore it holds

. x 5 -3x 4 +5x 3 -7x 2 +4 . (x-2)(X 4 -x 3 +3x 2 -x-2)

hm
x--+2 x 5 - 5x 4 + 8x 3 + x 2 - 12x + 4
= x-->2
hm
(x -2)(X4 - 3x 3 + 2x 2 + 5x - 2)
= 2.
LIMITS OF FUNCTIONS 147

Exercise 4.13. Determine the following limits,

a) lim x - 3x + 2 3x + 2
2
b) lim x 2 -
x ...... 2 x 2 -4 X""" +00 x2 - 4 ;

c) lim x - 3x + 2
2 X7 + 5x6 - 7x 2 + 5
4X4 -
d) lim '
x ...... -2 (x + 2)2 ; x ...... l x6 - 3x + x + 2x - 1
5 3

Answers.

1 7
a) 4 b) 1. c) + 00, d)
4

Example 4.14. Determine the following limits,

X lOl - 101x + 100

a) xlim
...... l X2
- 2X + 1 ;

'
b) l1m
x ...... l
(m ------
1 - xm
n) '
1 - xn
m,nEN;

x
m+l - xn+l + xn - mx + m - 1
"
m n E N'
,
c) xlim
...... l (x 1)2

' (l+mx)n-(1+nx)m
d) l1m N
' m,n E ,
x ......O x2

Solutions.

, X lOl -101x + 100 l' (x - 1)2(x 99 + 2x 98 +'" + 99x + 100)

a) hm = 1m = 5050,
x ...... 1 x2-2x+l x ...... l (x-l)2

b) Assume m > n > 3. If we introduce t = x-I (t ---+ 0 when x ---+ 1), then we
obtain

'
l 1m (m
- - - - - = lim
1 - xm
X""" 1 1 - xn t ...... o 1 - (1
n) (m+ t)m - -----:---:--
1 - (1
n)
+ t)n

Ji:'6 ( _,m - (7),m-'~ ... - (m': 1)t - _,n - (~),n-': .. - (n ~ 1)')

148 CHAPTER 4.

. + n m( m2 - 1) ) t + t 2 ( -m (n : + n (m
( -m n( n 2- 1) + ...
hm -'----------;---,--------=------'-----::-~___:_'-----'':__--'--_:_--'--.!..--
3) r: 3) )
t-+O t(t m - l + mtm- 2 + ... + m)(tn- l + nt n- 2 + ... + n)

. t n- l (-m+n(mr: n )) +···+nt m- l
+hm-,-----'--------'---,-:'--'--------
t-+O t(t m - l + mt m- 2 + ... + m)(t n- l + nt n- 2 + ... + n)

m·n ( - m-1 --- n-1)- +o(t)

.
11m 22 m- n
t-+O (t m - l + mt m- 2 + ... + m)(t n- l + nt n- 2 + ... + n) -
-
--
2 .

The other cases are left to the reader.

c) Taking t = x-I (t ----t 0 when x ----t 1), we obtain

x m+ l _ x n+ l + xn - mx + m - 1
lim
x-+l (1)2
X -

. (t + 1 )m+1 - (t + 1 )n+1 + (t + l)n - m( t + 1) + m - 1

hm -'------"----'---'---------'-------''------'--------"----
t-+O t2

t 2 ((m+1)m _ n(n+1) + n(n 2-1)) +o(e) m 2 +m-2n

lim 2t 2 2
2
t-+O

Here 0(t 2) stands for a function ¢>(t) with the property lim ¢>(2t )
t-+O t
o (see
Definition 4.54).

d) Similarly as in the previous case we have

. (1 + mx)n - (1 + nx)m n(n - 1) 2 m(m - 1) 2 mn(n - m)
hm = m - n = .
x--->O x2 2 2 2

Exercise 4.15. Determine the following limits.

a) x--->l
lim (-1
3 + ~1)
-x-
; X
3
xm -1
lim~, m,nEN;
b) x-+l x

c) lim (xn - l)(x n- l - 1)··· (x n- m- l -1)

x--->l (x-1)(x 2 -1) ... (xm -1) ,m,nEN, m<n.

Answers.

b) m n(n-1)···(n-m-1)
a) 1. c)
n 1·2···m
Example 4.16. Prove that lim y'X = y'a, a> 0, and lim y'X = 0, n E N.
x-+a x-+O+
LIMITS OF FUNCTIONS 149

Solution. Let us first suppose that a > O. Then for arbitrary e > 0, from

Ivx- ytal = Ix - al Ix - al
< - n-1 < e,
V'x n- 1 + V'xn- 2a + ... + V'a n- 1 V'a -

it follows that we can order 8 such that it holds

Ix - al < 8:= eV'an- 1 =? Ivx - ytal < e.

If a = 0, then, for x > 0 and arbitrary e > 0, it follows that there exists 8 := en
such that it holds
o < x < 8 =? Ivxl < e,
and this means that lim
x--->O+
vx = O.
Remark. It holds that lim
x----++oo
\IX = +00, because for every T > 0 it holds \IX> T,
provided that x > Tn.

Example 4.17. Determine the following limits.

a) lim y'X - 3 b) lim v'x+6 - 3

x--->9 X - 9 ; x--->3 x-3

x-2 . d) lim v'x 2 + X + 1 - x-I

c) ~~ JX+2 - 2' x--->O x

e) lim
x--+8
v'8+X -
3[;;
4 f) lim y'X -
x--->a+
+~va
'/x2-a~ ;a>O.
yX - 2 '.

Solutions.

a) lim y'X-3 = lim (y'X-3)(y'X+3) = lim x-9 = 1/6.

x--->9 x-9 x--->9 (x-9)(y'X+3) x--->9(x-9)(y'X+3)

b) lim v'x+6 - 3 = lim (v'x+6 - 3)( v'x+6 + 3) = 1/6.

x--->3 x- 3 x--->3 (x - 3)( v'x+6 + 3)

· x-2 l' (x-2)(Vx+2+2)

c ) 11m = 1m ---;:===-'"-'-~==-'-
x--->2 Vx+2 - 2 x---+2 (JX+2 - 2)(Vx+2 + 2)
lim (JX+2 + 2)(x - 2) = 4.
x--->2 X - 2

· v'x2+x+l-x-l
d) 11m l' (v'x2+x+I-x-l)(v'x2+x+l+x+1)
= 1m -'-------:~=;;======o-----___:-----'--
x--->o X x--->o x( v'x 2 + X + 1 + x + 1)
-x -1
= lim = - - = -1/2.
x--->ox(v'x 2 +x+1+x+1) 1+1
150 CHAPTER 4,

e lim.J8+X - 4 = lim (.J8+X - 4)(.J8+X + 4)( V9 + 2~ + 4)

) {IX - 2
x---+B x---+B({IX - 2)( V9 + 2{IX + 4)(.J8+X + 4)
= lim (x - 8) ( V9 + 2 {IX + 4) = 12 = ~
x---+B(x - 8)(.J8+X + 4) 8 2

f) lim
x---+a+
VX - va + ~ =
..jx 2 _ a2
lim
x---+a+
(vx - va + ----:vr=;x;=-=a=;;:)
..jx 2 - a2 ..jx 2 - a2
lim ( x-a + 1 )
x---+a+ (VX + va)..jx 2 - a2 vx:ta"
E~+ (VX ~ va' J: ~: + ..jx1+ a) ..;k, =

Example 4.18. Determine

a)
, ..jx
I1m + 2 - ijx + 20 '
x---+7 ~x+9-2 '

b) lim \1f+X
x - 1 n E Z \ {O};
x---+o

C)
, 11
I1m + ax' xV'1 + bx - 1 , m,n E Z \ {O}, a, bE R;
x---+o

d) lim
'11 + P(x)-1 , where P(x) = alX + a2x2 +, .. + anx n ~ 0, mE Z \ {C
x->o X

, (1 - VX)(1 - {IX) .. , (1 - y'x)

e) hm
x---+1
(
1 -)
x n -1 n E N,

Solutions.

a) lim..;x+2 _ ~ , ..;x+2 - 3 + 3 - Vx+2O

x---+ 7 ~ +9_
X 2 hm7
= x-+ x - 7 4r::::-17\ x 7
yx+9-2

, ..;x+2-3 x-7
+ I'1m -----.:...:..
3 - ijx + 20
3
lim 1 1
X _ 7
x->7
x->7 X 7 - 27 112
, ~x+9-2
I1m-----= - 6 1- 27
x---+7 X - 7 32
b) If we change the variables t = 11 + x-I, then from Example 4,16 it follows
that lim 11 + x = 1, and we obtain that t - t 0 when x - t 0, So we have
x---+o

t 1
x->O
n~ 1 = lim
lim y 1 +X x - t---+O (1 +t t)n _ 1 = lim
t---+O
nt + ( n) t + ' , , + tn
2 2 n
LIMITS OF FUNCTIONS 151

c) We assume a, b i- 0, the other cases are left to the reader. Using b), we obtain
. \11
11m + ax· 'V'1 + bx - 1 1.
= 1m
yIf+Tx( ytr:taX - 1) + yIf+Tx - 1
x~o X x~o X

lim 'V'1
x--->o + bx . a . lim ytr:taX - 1 + b . 1·1m ----:----
'V'1 + bx - 1 a + -.
b
x--->O ax x--->O bx = -n m
d) We assume that mEN, the case -m E N is left to the reader.

hm
. r/1 + P(x) - 1
=hm
. r/1 + P(x) - 1 P(x)
._-
x--->O X x--->O P(x) X

\!l+P(x)-l 1
lim
x---.o
P( )
x . lim (al
x~o
+ a2 x + ... + anx n- )

. 1 a1
= al· hm =-.
x--->O r/p + p(x))m-l + r/(l + p(x))m-2 + ... + 1 m

e) After changing of variables t = 1 - x, where t ~ 0 when x ~ 1, we obtain

1,---_Vx...:....x..,...:.)_.:-
lim . .:. .l_-_Vx...:....x--,-)-,-(
( 3
.. . :. .l_-_VX...:....n
.( x--'-)
x--->1 (l-x)n-l

lim ( 1 -
t--->o
v'1=t . 1 - ~ ...
t t
1- vT=t)
t
= ~. ~ ... ~
2 3 n
1
,
n.
.

Exercise 4.19. Determine the limits of the functions given below.

a ) 11· m x · b) 1·1m V9 + 2x - x 2 - v'3 + x + x 2.

x--->o Vl8 + x - 2' x--->2 2x - x 2 '

c) lim
Vl6 + x + x + 4 ; . 1- Vx
d) hm 1 {/X;
x--+-5 yt9 + 2x + 1 x-+1 - X

. y'X-1 \1c+x- \1c-x

e) hm VX
x-+l n x-I ' m,nEN;
f) lim , n E N, c> 0;
x-+O x

'V'1 + ax - \11 + bx , m, n E Nab E R.

g ) x-+O
lim X
"
Answers.
7
a) 12. b) 12. c) 2. d) ~3·
n .
e) -
m
(change the vanable x to (1 + t)mn).

f) 2yIC a b
nc . g) - --.
m n
152 CHAPTER 4.

Example 4.20. Prove that

an) ·00, n >m;

(sgn bm

.
11m anx n + an_Ix n-I+ . . . + ao an
x-++oo bmx m + bm_Ix m- I + ... + bo n=m;
bm '

0, n < m,

whenever an =I 0 and bm =I O.
Solution. The given expression can be written as

anx n + an_Ix n- I + ... + ao

bmx m + bm_Ix m- I + ... + bo

1+ -
an-l
-X
-1 + · · ·+-aoX-n (4.5)
an n-m an an
-·x
bm 1+ bm-I
- - x -1 + · · ·+-bX
o -m
bm bm
Let us suppose that n > m. Since,

·
11m (an-l
--x- 1 + ... + -aoX - n) = 0
x-++oo an an

and
·
11m (b rn -
- - X -1
1
+ ... + -bboX -m) 0
=,
x-++oo brn m
there exist positive numbers Xl and X 2 such that the following inequalities hold.

aox -nl < -1

l-an-1
- x -1+ ... + -
an an - 2'
for X> Xl,

and

I-bb- x - + ... + -bbox -ml < -21

m- 1
m
1
m -
for x > X 2.

So for x ~ max{X1 ,X2 }, we obtain

1+ -
an-l
- x -1 + · · ·+-aoX-n
xn- m an an
1+-
bm-l
-X
-1 + · · ·+-boX -m
bm bm

1- l-an-1
-x -1
+ ... + -aox -nl 1 - -1 n-m

1-
> xn-m. an an > xn-m _ _
2 = _x_

I
1 + bm-l x-I
bm
+ ... + ~x-m
b
bm
1 + -1
2
3
LIMITS OF FUNCTIONS 153

Since lim x m -
X---++CXJ
n = +00, we have

.
11m anx n + an_IX n-l + . . . + ao = (s an )
n - . 00
x->+oo bmx m + bm_ 1 x m- 1 + ... + bo g bm .

If m = n, then from relation (4.5) it follows

.
11m anx n + an-IX n-1 + . . . + ao an
--.
x->+oo bmx + bm_ 1 x - + ... + bo
m m 1 bm

At last, if n < m, then from lim x m -

x---++oo
n = 0 and relation (4.5) we have

.
11m anx n + an-IX n-l + . . . + ao
x->+oo bnx n + bn_1x n- 1 + ... + bo

1 + -an-1 -1 + + aO -n
-X "'-x
an lim xn-m. lim an an = an . 0 . 1 = O.
bm x->+oo x->+oo bm- 1 -1 bo -m bm
l+--x
bm
+"'+-x
bm

Example 4.21. Determine the following limits.

a)
.
11m 5X3 + 3x 2 + 2x + 5 5x 3 + 3x 2 + 2x + 5
. b) lim
x->+oo 5x 3 + x 2 + X + 3 ' x->+oo 5x 4 + x 2 + X + 3

) .
11m 5X4 + 3x 2 + 2x + 5
. d) lim v'x+T
Jx + Jx +,;x;
C
x->+oo 5x 3 + x 2 + X + 3 ' x->+oo

e) lim (~-
x--++oo
v'?"TI) ; f) lim (J16x 2 + x-I - 4x) ;
x-++oo

g) lim (\lx3
x---++oo
+ 3x 2 - JX2 - 2x) ; h) lim X (Jx2
x---++oo
+ 2x - 2JX2 + X + x) ;

i) lim (;;!(x
x-++oo
+ al)(x + a2)" . (x + an) - x) .

Solutions.

3 -
x 3 ( 5+ -+ 2 +-
5)
5x 3 + 3x 2 + 2x + 5 l' X x2 x3 =1
lim 5x 3 + x 2 + X + 3 =
a) x->+oo 1m ( 1 1 3)
x->+oo x3 5 + _ + _ + _
.
X x2 x3

3
( 5 + -; 2 5)
b) lim 5x 3 + 3x 2 + 2x +5 x3 + ;2 + ;J
= O.
x-++oo lim
x->+oo 4 ( 5 + _1 + _1 + _3 )
X x2 x3 X4
154 CHAPTER 4.

3
( 5 + ;; 2 5)
.
11m + 3x z + 2x + 5 X4 + ;2 +~
C) +00.
5X4
=
x-++oo 5x + x Z + x + 3
3 lim
X-++OO 3( 5 + _1 + _1 + _3)
x X XZ x3

Vx·R
Vx+ Vi ~ ~ ~
=l.

d) ,~'foo Jx +v'xTI x-++oo

lim
Vi. 1+
1
;; + x'
e) lim (~- V'x2TI)= lim (~-v9+1)(~+v9+1)
~ + v9+1
x-++oo x- "I l.
x-++oo
-2
lim
x-++oo ~ + v9+1 = 0.
f) x!i~oo (v'16x Z + x-I _ 4x) lim (v'16x Z + x-I - 4x) (v'16x Z + x-I
= + 4x)
v'16x z + x-I + 4x
x-++oo
. x-I 1 -I
= 11m
x-++oo
- 1·1m
v'16x Z + x-I + 4x - x-++oo
4x· 1 + -
x
1
- -
1
- + 4x
=8 J
16x 16x z

g) x!i~oo (V/ X 3 + 3x z - v'x z - 2x) = x!i~ (V/ X 3 + 3x z - x)+ x!i~oo (x - v'x z - 2x)
3x Z . 2x
lim + hm = 2.
x-++oo \/(x3 + 3x2)2 + xV/x 3 + 3xz + xZ x-++oo x + v'x z - 2x

2x (v'xz + 2x - x-I)
h) lim x (v'xz
x-++oo
+ 2x - 2v'x z + x + x) = lim x
x-++oo v'X Z + 2x + x + 2.Jx2+x
X Z+ X
_2x2 1
- lim
- x-++oo (v'X Z + 2x + x + 2v'X Z + x) (v'x Z + 2x + x + 1) 4
i) Taking x =~, where t --+ 0+ when x --+ +00, we obtain
t
\l1+P(t)-1
+ al)(x + az)··· (x + an) -
n
y/(x x= t
where
P(t) = (al +az+·· ·+an)t+(alaZ+ala3+·· ·+an_lan )t2+·· ·+alaZ·· ·antn.
From Example 4.18 d) it follows
lim (n/r-(x-+-a-d-(x-+-a-z-).-.-.(-x-+-an-) - x)
x-++oo V

lim
\II + P(t) - 1
=
al + a2 + ... + an .
t-+O+ t n
LIMITS OF FUNCTIONS 155

Exercise 4.22. Determine the following limits of the functions.

a) lim (x + 2)(3 - 5x). b) lim (3x 3 + 2x - 1)6 .

x--++oo (2x + 1)2 , x--++oo (3x 6 + 8x 2 + 1)3'

c
)
lim
x--++oo
( X41 +
+x
2x
3
3
- x) .
'
d) ·
11m
x--++oo
-
(x2- +
2x + 1
x 3 + 4x 2 -
1 - 2x 2
2) .'

3x 6 -1
(Vx4 + 2x 2 - 1- Vx4 - 1) ;

+
e) lim + 2; f) lim 2x 2 -
x--++oo v'x 12 + 2 x4 x--++oo

g) xg'J'oo (Vx' + Jx' + Vx' + n -

h) lim x 2(JX4
x--++oo
+ X2VX4 + 1 - V2;;t) .
Answers.
5 9
a) b) 27. c) 2. d)
4 4

1
e) 3. f) 2. g) 2· h) J2
8 .

smx
Example 4.23. Knowing that lim - - = 1, determine the following limits.
x--+O x

a)
r1m--;
tan x
b)
r1m-.--,
smax
001 a, bE R;
x--+O X x--+O sm bx

cos 2x 3 - 1
c) lim ( . / . - ~) ; d) lim ;
x--+O sm x SIn x sin x x--+O sin6 2x

e) lim
x2
. f)
r1m tan(a + x) tan(a - x) - tan 2 a
.
x--+o VI + x sin x - vcos x' x--+O x2 '

. cos(a+2x)-2cos(a+x)+cosa
g) 11m 2 .
x--+o X

In f) and g), we assume a 01 O.

Solution.

a) The function f( x) = cos x is continuous at the point x = 0 and therefore we

have
· -
11m tan x 1· sin x 1. 1
- = 1m - - . 1m - - = l.
x--+o x x--+o X x--+o cos x
156 CHAPTER 4,

Slnax
, a'---
' SIn ax l'1m a
b) lIm-,-- = , ax
x-+O sm bx x-+O b, sm bx b
bx

) r ( 2 __ 1_) _r _2_S1_'n_x.,..--.,..-2_S1:-'n-;;x_co_s_x_
c x~ sin 2x sin x sin 2 x - x~ sin 2x sin 2 x
1 - cos x 2 sin 2 ( x /2)
' 2 1 - cos x
l1m, ,
,
= lIm,
x2
,
r 4(x/2)2 1
-
x-+o SIn 2x SIn x x-+o SIn 2x sm x = x~ sin2x sinx 2
2X2 2x x

( sin X3) 2 )
'
d) l1m cos 2x 3 - 1 l' -2 sin 2 x3 l' ~ 1
= 1m = -21m (
x-+o sin6 2x x-+o sin6 2x x-+o 26 (si~:x) 6 32

' X2 l' x 2( VI + x sin x + Vcos x)

e ) l 1m = 1m -~---:-,----''------'-
x--->o vI + xsinx - VCOSX x--->o 1 + xsmx - cos x

lim VI + x sin x +
x-+o 1 - cos x sm x
ycosx = _2_ = ~
1 3'
- -2- + - - -+1
x x 2

' tan( a + x) tan( a - x) - tan 2 a

f) l1m l' 1 (tan 2 a - tan 2 x 2 )
= 1m - - tan a
x--->o x2 x--->o x 2 1 - tan 2 a tan 2 x
, tan 2 x 4 COS 2a
hm --(tan a-I) = - - - ,
x--->o x2 cos 4 a

' cos(a+2x)-2cos(a+x)+cosa
g ) l 1m
x--->o x2

lim
x--->o X
~((cos(a + 2x) - cos(a + x)) - (cos(a + x) - cos a))

lim -1 ( -2 sm
,x ' (a + -
- 'sm 3X) + 2 sm, x , ( a + -X))
- 'sm
x2
x--->o 2 2 2 2
, x
-2sm- 3
ETa x 2 2 (sin(a+ ; ) -sin(a+~))
,x
-2 sm 'x
-, 2 sm -, cos ( a + x )
lim 2 2 = - cos a,
x--->o x2
LIMITS OF FUNCTIONS 157

Exercise 4.24. Determine the following limits.

. sin 5x
11m-.-;
a) b) lim 3x cot 3x;
x-+O SIn x x-+o

. smx d) r1m 1 - cos x .

c) hm . 7 . 9 ;
x-+O SIn x - SIn x x-+O x2

. 1 + sin x - cos x f) lim cot( a + 2x) - 2 cot( a + x) + cot a.

e ) 11 m . ;
x-+o 1 + sm ax - cos ax x-+O x2
In e) and f), we assume a #- O.
Answers.
1
a) 5. b) 1. c) - -
2

1 1 ) 2 cos a
d) 2" e) - f --.
a sin 3 a
Example 4.25. Determine
. sin 47rX b) lim smx - sina
a ) 11m. ;
x-+1 SIn 57rX x---+a x- a

· cot x - cot a
11m k d) 1· tan 3 x - 3 tan x.
c)
....L
, aT 7r;
COS X + 6"
x---+a X - a 1m
x-+tr /3
(tr)'

. . 27r
e) hm xsm-;
X---++CXl X
f) lim 3x 2 (cos
x-++oo
~X - cos ~)
X
.

Solutions.
a) After changing of variables t = x-I, t --+ 0 when x --+ 1, we obtain

lim sin47rx = lim sin(47r(t + 1» = lim sin(47rt) =_~

x-+lsin57rX t-+O sin(57r(t + 1)) t-+o - sin(57rt) 5

. x-a x+a
sin x - sin a 2 sm - - - cos - - -
b) lim = lim 2 2 = cosa.
x---+a X - a x---+a X - a

. cot x - cot a_I. sin(a - x)

11m 1 __I_
e) - 1m. . . 2 .
x-+a X - a x-+a sIn x sIn a x - a SIn a

. t an 3 x - 3 t anx tan x (tan 2x - tan 27r)

. -3
d) hm 7r = hm 7r
x-+tr/3 cos(x+-) x-+tr/3 cos(x+-)
6 6
sin(x - ~)
~ - lim tan x (tan x +tan ":) x (3 X) ~-24.
x-+tr/3 3 cosxcos 3 sin 7r/2-(x+ 6)
158 CHAPTER 4.

e) After changing of variables t = 2Jr, t --+ 0+ when x --+ +00, we obtain

x

·
11m 2Jr l' sin t
x sin - = 1m 2Jr-- = 2Jr.
"'-->+00 x t-->O+ t

f) lim 3x 2 (cos
"'-->+00
~x - cos~)
x
= lim 3x 2 (2 sin ~ sin ~)
"'-->+00 x x
. 1 . 2
sm- sm-
lim 3·2· 2 _ _x . __ x = 12.
"'-->+00 1 2
x x

Exercise 4.26. Determine the following limits.

sin x b) lim sin(x - 1)

a) X-->7l"
lim Jr 2 -
X
2;
x-->l 1 - x3 ;

.
11m 2 cos 2 X + cos x-I d) lim cos(x + Jr) - cos(3x + Jr)
c) .
X-->7l" /3 2 cos 2 X - 3 cos x + 1 ' X-->7l" (x - Jr) 2

Answers.
1 1
a) 2Jr b) c) - 3; d) 4.
3
Example 4.27. Determine the following limits.

a) !~ 2 sin(Vx+T - 1). b) lim

x2 •
c;::;;-;;;; ,
x x-->O VI + x sin x - V cos x

c) lim ytCOSaX - ~ d) lim (sin Vx+T - sin y'X),

x-->o x2 x--++oo

where in c) we assume m, n E N and a, b, E R.

Solutions.
x

2 sin( Vx+T - 1) = 2 lim x

sm~+1
(Vx+l + 1) =l.
a) x-->O
lim x x-->o Vx+T + 1

x2 x 2 (VI + x sin x + vcos x)

b) lim = lim - - - ' - - - - - - - - - - " -
",-->0 VI + x sin x - Vcos X x-->o 1 + x sin x - cos x

. y'lTXsin x + Vcos x 4
hm . =-.
1 - cos x sm x 3
+--
x-->O
----=--
x2 X
LIMITS OF FUNCTIONS 159

c
r y'COS(iX -
) "'~ x 2
y'cOSbX
= "'~
r (y'COS(iX -
x2
1
-
y'cOSbX -
x2
1) = ~(b
2 n
2
_ a2 )
m
•

d) ",.!!foo (sin Vx+T - sin ft) = 2 "'.....

lim +00 (sin Vx+T2 - ft . cos Vx+T2 + ft)
lim 2 sin 1 VxTI + ft
"' ..... +00 2( VxTI + ft) cos 2

·
SInce 1cos Vx+T2 +ft -< 1 for all x E R ,we h ave
1

1",.!!f)sin
. .
Vx+T - sin ft)l::; "' .....lim
+00
12 sin 2( VxTI
1
+ ft) 1. 1 0 T
. hus
hm (sm Vx+T - sin y
X~+OO IX)w= 0.

Exercise 4.28. Determine the following limits.

tan 2 x . b) 1·1m vcos x - ,ycos x .
a) l~ V3 - V2 + cos x ' "' ..... 0 sin 2 x

.
11m {lsin x - ,ysin x
c) . d) lim (cos VxTI - cos ft) .
cos 2 X
"' ..... 7r/2 ' x-++oo

Answers.
1 1
a) 4V3. b)
12 c) 24 d) O.

Example 4.29. Determine

1
7r - 4 arctan - -
a) lim arcsin(x + 2) b) lim 1 +x
"' ..... -2 x2 + 2x ; "' ..... 0 x
Solutions.
a) After changing a variable t = arcsin( x + 2), i.e., sin t = x + 2, where t --+ 0 when
x --+ -2, we obtain
lim arcsin(x + 2) = lim ~ . arcsin(x + 2) = _~ . lim _t_ = _~
"' ..... -2 x2 + 2x "' ..... -2 X X +2 2 t ..... o sin t 2
b) After changing a variable t = 7r - 4 arctan _1_, i.e., x = cot 7r - t - 1, where
x +1 4
t --+ 0 when x --+ 0, we obtain

1 t
7r - 4 arctan 1 + x = lim ~ t 1
1m x
"'1·..... 0 t .....o cot -4--

. 7r-t . 7r t. 1 1
tsm--sm-
J2. v'2
lim
t ..... o . (7r
sm
4
- -7r-- - t)4 = limo t ..... . t
sm-
= 2.
4 4 4
160 CHAPTER 4.

Exercise 4.30. Determine the following limits.

a) lim
arcsIn x
, b)
r
Im--
arctan 2x
x-+o 3x x-+o x
Answers.
1
a) - b) 2.
3
Exercise 4.31. Let us suppose that the following inequalities

gl(X) :::; f(x) :::; g2(X)

are satisfied on the interval (a,{3) \ {xo}, Xo E (a,{3). Additionally, if there exists a
number L such that
lim gl(X) = X-+Xo
X-+XO
lim g2(X) = L,

then, using Theorem 4.6, prove

lim f(x) = L.
X-+Xo

Example 4.32. Show that

a) lim
x-++oo
(1 + ~)X
X
= e; b) lim(l+x)l/x=e.
x-+o

Solutions.

a) Let x take the values of the sequence (xkhEN with the property lim Xk = +00.
k-++oo
Then for every monotonically increasing sequence (nkhEN such that nk E N
and nk -+ 00 when k -+ +00, it holds nk :::; Xk :::; nk + 1 and
111 111
--<-<- hence 1 + - - < 1 + - < 1 +-.
nk + 1 - Xk - nk' nk + 1 - Xk - nk
Wherefrom we have

1 )nk
(1+- - < ( 1+-l)Xk
< ( 1+-l)nk+l
nk + 1 - Xk - nk '
or

(1+- 1 - )nk+l(1 + -1 - : : : ; 1+- )-1 (

l)Xk :::; ( 1+-l)nk(1 + - . 1)
nk + 1 nk + 1 Xk nk nk
Since
1 )nk ( 1 )n k+l
lim ( 1 + - = lim 1 + - - - = e,
k-++oo nk k-++oo nk + 1
we obtain
1 )Xk
lim ( 1 + - = e, l.e., lim (1 + ~) x = e.
X-++OO Xk X-++OO X
LIMITS OF FUNCTIONS 161

b) If x --t 0+, then for t = ~ we have t --t +00 and from a) we obtain
x

lim (1 lim (1 + ~)t = e.

+ x)1fx = t-++oo
x-+o+ t
1 .
If x --t 0-, then for s = - - It holds s --t +00 and we have
x

lim (1
x-+o-
+ x)1fx = (1 _~)-.
s
lim (_S_)'
lim
.-++00 s - 1
.-++00
=

lim (1 + -1 lim (1 + -1)

-).-1 . .-++00 - e . 1 e. = =
• -++00 s- 1 s- 1
Using Theorem 4.3, from the equalities

lim (1
x----+o-
+ X)l/x = lim (1
X~O+
+ X)l/x = e,
it follows that
lim(l
x-+o
+ X)l/X = e.

Example 4.33. Show that

a) lim aX = a Xo , a > 0 , 0
x E R, b) lim Inx = Inxo, Xo > 0;
X~Xo x ....... xQ

c) lim (u(x))v(x)
x ......... xQ
= ab , Xo E R.

In c)! we suppose that u( x) > 0 and that there exist a > 0 and b such that

lim u(x) = a and lim v(x) = b.

X-+XO X-+XO

Solution.

a) Assume first a > 1. In Example 3.11 b) it was shown that

lim a l / n lim a-lin - 1

n ...... oo n-+oo -,

and therefore, for given £ > 0, there exists no E N such that for a > 1 it holds
1 - ~ < a-l/no < aI/n0 < 1 + ~.
aXo aXo
Note that we used the fact that the function aX is monotonically increasing on
1 1 1
the whole R. If we suppose that Ix - xol < -, then from - - < x - Xo < -,
no no no
it follows
1 - ~ < a-l/no < a X- XO < aI/no < 1 + ~,
aXO aXO
wherefrom we have
£ £
< a X- XO _ 1 <
aXO aXO
162 CHAPTER 4.

1
So for 0:= - and Ix - xol < 0, we have
no
laX - axol = axolax-xo -11 < e.
Thus for a > 1, it follows lim aX = aXO for any Xo E R. In the case when

i,
x-+xo

0< a < 1, i.e., a = b> 1, it holds

· a = 1·1m
11m X
-1= -1 = a X
o.
x-xo x-xo bx lfo
b) We proved the following inequalities (see Example 3.38 c))

1 (1) < -n1 and

- - < In 1 + -
n+ 1 n n -1
1 (1)
- - - < In 1 - -
n
1 n > l.
< --,
n
Since the In function is monotonically increasing, we can write

__1_ < In (1-~)

n-1 n
< In (1 +~) <~.
n n

For given e > 0 and e ::;~, there exists no such that

- e < __
~-1
1_ < In (1 - ~) < In (1 +~) < ~ < e.
~ ~ ~
(4.6)

Xo
If we suppose that Ix - Xo I < - , then we have
no
1 x - Xo 1
--<--<-. (4.7)
no Xo no
From relations (4.6) and (4.7) we obtain

-e < In (1 - ~J < In (1 + x :oxo) < In (1 + ~J < e.

This means that

IIn(l+X:oXo)l<e, hence IInx-Inxol<e,

under the condition Ix - xol < 0 with 0:= xo. So we get for Xo > 0
no
lim Inx = Inxo.
X---IoXo

c) From a) and b) we obtain

}~o (u(x))v(x) = }i.~o exp (v(x) In u(x)) = exp Ui.~o v(x) . }i.~o In u(x))
e blna = abo
LIMITS OF FUNCTIONS 163

Example 4.34. If the functions u( x) and v( x) satisfy

lim u(x) = 1 and lim v(x) = 00,

X---+Xo X--+Xo

then we have
lim u(xt(x) = exp (lim (u(x) - 1) v(x)) , (4.8)
X-+Xo X-+Xo

provided the last limit exists.

Solution. We have
lim u( x )v(x) = lim (1 + (u(x) _ 1))l/(U(X)-l)) (u(x)-l)v(x)
X--+Xo X-Io

Ji..q;o exp ((u(x) - l)v(x) In(I + (u(x) - 1))l/(U(X)-l))

exp (Ji..q;o (u( x) - 1) v( x)) .

c
Example 4.35. Determine the following limits.

a) lim - -
x-->+oo c+
1
x-I r .
'
b) lim - X +-3 r+ 2 .
x-->+oo 2x + 2 '

c) lim( JI+X
x-->a
- x)2fx; d) lim( cos x
x-->a
t cot 2 x;

e)
. C1++
hm
x-->a
tanx) 1/ sin 3 x
.
sm x
f)
. (1 + tan X) 1/ sin x
hm
x-->a1+ sin x
Solutions. We shall repeatedly use Example 4.34.
x+I
a) In this case we put u(x) = - - and v(x) = x, hence
x-I

(u(x) - 1) v(x) = ( ; ~~ - 1) x = x 2~ 1.

Thus we obtain

.
11m -+-
(X 1) x = exp ( 1.1m -2X)
- = e 2.
x-->+oo x-I x-->+oo x-I

b) Using the previous method we have

2X+3)X+2
lim ( - -
x-->+oo 2x + 2
= exp ( x-->+oo
lim - - = e1 / 2 = .;e.
X+2)
2x + 2
Also, this example can be done as follows.

2X + 3)X+2
lim ( - - = x-->+oo
lim
(( 1 )2X+2) 1/2 . lim ( 1 + -1)
1 +- - - =.;e.
x-->+oo 2x + 2 2x + 2 x-->+oo 2x + 2
164 CHAPTER 4.

2
c) If u(x) = vr+x - x and v(x) = -, then
x

lim(v'1+x - x) = 1 and lim v(x) = lim ~ = +00.

x-a x_o+ x_o+ x

Thus we have
(u(x) - 1) v(x) = (v'1+x - (x + 1)) ~.
x
From the formula (4.8) we get

lim (v'1+x - x?/x = exp (lim ~(v'l+x - (x + 1))) = e- 1.

x_o+ x_o+ x

We leave to the reader to show that lim (Jf+X - X)2/x = ~, thus

x_o- e

lim(v'l+x - X)2/X --
x---+-o
~•
e

. t2. ( . )-Cos 2 x/(2sin 2 x)

d) hm(cosxt CO x=hm 1-sm2 x =ve.
x---+-o x---+-o

1 + tan x 1
e) If we denote by u(x) = . and v(x) = -'-3-' then we have
1 + smx sm x

(ux ( ) -_ tanx-sinx . _
() -1 ) vx 1 _ tanx(l-cosx)
-- ._-
1
1 + sinx sin3 x 1 + sin x sin3 x

2 . 2 X
sm "2 1
cosx(l+sinx)' sin 2 x'
Therefore we have

1 + tan x 1/ sm
. 3x ( 2 sm
. 2 -X )
lim = exp lim . 2 = ve·
x-a (1 + sinx) x-o cosx(l + smx) sin x
2

f) !~ (1 + t~n X) 1/ sin x
° 1 + smx = 1.
Exercise 4.36. Determine

a lim
X
--
)X ; b) lim (X2 + 2)X2_1
x +1
(
) X_+OO X_+OO x2 - 2 ;

c) lim (1 +cotx)tanx; d) !~ (1 + X 4 )COt 4 x .

x_1f/2
LIMITS OF FUNCTIONS 165

Answers.
1
a) - b) e 4 • c) e. d) e.
e
Exercise 4.37. Prove that

a) lim log a x = log a Xo, a > 0, Xo > 0; lim arccos x = arccosxo, Ixol < 1;
b) X---+Xo
X---+Xo

1 7r
c) lim arctan - = -2;
x ..... o+ x
d) lim arctan x =
x ..... -oo
_!!:..2
Example 4.38. Determine the following limits.

a) lim log5 -1 x+5

X"'" 5 x - 5 ; b) lim xlog3
x ..... +oo
-+
X
3;

. In(1
c ) 11m
+ 4x + x 2) + In(1 - 4x + x 2) . d) lim V4+X - 2 - sinx.
X"'" 0 2x 2 ' o
x .....In(1 + x) ,

. In cos 4x
e) 11m .
x ..... o In cos 3x '
f) lim x 2 ln (sin
x ..... oo
(!!:.2 - !!:.))
x
;
In(1 + 3 X ) • . In(1 +3 X )

g) x ..... -oo In ( 1 + 2x)'

lim h) xl!~oo In(1 + 2x)·
Solutions.
. log5 X - 1 _ 1· log5 X -log5 5 _ 1· 1 (X)1/(X-5)
a ) 11m - 1m - 1m og5 -
x ..... 5 X - 5 X"'" 5 X - 5 X"'" 5 5
X - 5) (5/(X-5)).1/5) 1 1
= xlim logs exp ( In ( 1 + - - = -logs e = -1-.
..... s 5 5 5 n5

b)
x ..... +oo
x +-
· x 1og3 -
11m 5 = -11- . 1·1m 1n ( 1 + - -
+ 3 n 3 x ..... +oo
x + 3 = -1n-3. X
2) x 2
c) lim In(1 + 4x + x 2)2x
x..... o
+2In(1 - 4x + x 2) = lim In(1 - 14x2 + X4)1/(2x2)
x ..... o

= ~~ (In(1- 14x2 + x 4)1/(X2(X 2 -14))) (x 2 - 14)/2 = -7.

d) lim V4+X-2-sinx = hm . (V4+X-2-sinx . (1 n (1+ x )l/X)-l)

x ..... o In(1 + x) x ..... o X

=
. .;4+X-2 -hm--
( x---+o
hm . Sinx) ·hm
. ( In(1+x )I/X)-l_ ~
--4.
x x---+o X x-+o

Incos4x sin 2 4x In(1- sin24xtl/sin24x 16

e) lim = lim - . - - . .2 = -.
x..... o In cos 3x x..... o sm 2 3x In(1- sin 2 3x)-I/sm 3x 9
166 CHAPTER 4.

2
1r2
f) lim x 2 ln (sin (~2 - ~))
x
= ~2 X--++OO
lim In (1 _ sin2 ~)X =
X--++OO X 2

. In(1 + 3X ) _ l' In (1 + 3- t ) _ l' 2t In (1 + 3- t )3 t _

g) 11m - 1m - 1m - . • - 0.
x--+-oo In(1 + 2x) t--++oo In (1 + 2- t ) t--++oo 3t In (1 + 2_t)2

In(1 + 3 X ) x In 3 + In (1 + Ix) ln3

h) lim = lim 3
X--++OO In(1 + 2x) X--++OO X In 2 + In (1 + ;x) ln2'

Example 4.39. Determine

eX -1
a) lim--; b) lim(l+x)b- 1 , bER;
x--+o x x--+o x

. 1 - cosb X d) lim e
X2
- (cos x)2v'3
c) hm 2 ,b E R;
x--+o x x-+o X -

Solutions.

a) Changing the variables t = eX -1, i.e., x = In(t + 1), where t -t °when x -t 0,

we obtain
eX - 1
lim--=lim l (
t ( ")-1 =l.
) = lim In(t+l))t
x--+ox t--+o n t + 1 t--+O

b) For b = 0 it is trivial that the given limit is equal to 0 = b. For b =1= 0, we start
with the following transformation

(1 + x)b - 1 ebln(I+x) - 1 bln(1 + x)

x = bln(l+x)' x

e bln(l+x) _ 1 bln(1 + x)
Since lim b ( ) = 1 and lim = b, we obtain finally
x--+o In 1 + x x--+o x

lim (l+x)b-l =b.

x--+o x

c) For b = °the limit is zero. For b -I- 0, similarly as in b), we can write
l-cosbx l_ebln (cosx) bln(cosx) 1 b
lim = lim . lim = 1· b· - = -.
x--+O x2 x--+o b In( cos x) x--+o x2 2 2

d) From a) and c) we obtain

. e x2 (cos X )2v'3 . e x2 - 1 . 1 - (cos X )2v'3

v'3.
_
hm 2 = hm --2- + hm = 1+
x
n
x--+O x--+o X x--+o X
LIMITS OF FUNCTIONS 167

Example 4.40. Assume a > 0 and determine

aX _ x a XX _ aa aa X
_ a xa
a) Iim--- b) Iim--- c) l i m - - -
x--+a x- a x_a X - a x--+a aX - x
a

Solutions.

a) Using Example 4.39 b), we have

X - a)a
aX _ x a x-a 1 ( 1 + - - -1
Iim--- aa lim a - _ lim a . aa-1
x-+a x - a x_a X - a x_a X- a
a

aa In a - a· aa-1 = aa In(aje).

b) From the transformation

XX _ aa XX _ x a x a _aa
x-a
-x-a
- + x-a
--

and from x_a

Iim( x - a) In x = 0, it follows that

X_ a
l'1m
ealnx (e(x-a)lnx - 1)
· x
11m X
x_a X - a = x-a ( X - a ) 1n X . 1n x a 1
= ana.

Using Example 4.39 b), we have

Iim xa - aa _. aa-1((1+ x - a )a - )
x_a X _
a
- hm
x-+a x-aa 1 = aa ,
a

and therefore we get

lim XX - aa = aa(lna + 1) = aa In(ae).

x_a X - a

aa X _ a xa . aaX-x" - 1 . x. . at - 1 . a"
c) Iim--- a
x--+a aX - x
= x_a
hm . lIma
aX - x a x_a
= hm---a
t_D t
a =a Ina,

because we put t = aX - x a and t --7 0 when X --7 a.

168 CHAPTER 4.

Exercise 4.41. Determine

a) lim (2
x-+o+
+ v'x); b) lim (2
x---+o-
+ v=I);

1
l'1m -; . 1
c) d) hm -;
x-+o- X x-+o+ X

e) lim x + Ixl. f) lim x+lxl.

x-+o- 2x ' x-+o+ 2x '

g) lim arcsin(x
x---.+o-
+ 1); h) lim arcsin( x
x-+o+
+ 1);

i) lim ecotx ; j) lim ecotx ;

x---+o- x-+o+

1 1
k) lim I) lim
+ e2- x + e 2 -1 x
1 ,
x-+2- x2 x-+2+ x2

Answers.

a) 2. b) 2. c) -00. d) +00.
e) O. f) 1. g) 7r /2. h) Senseless.
i) O. j) +00. k) O. I) 1/4.

Example 4.42. Determine

l' sinh x cosh x - 1 c) lim tanhx

a) x2.W--x- ; b) lim ;
x---+o- x2 x-+o- ~

Solutions. For the convenience of the reader, we rewrite the definition of the
hyperbolic functions.

sinh x = eX - e- X
eX + e- X sinh x eX - e- X
cosh x = 2 and tanh x = --- = - - -
2 cosh x eX + e- X

. h -x 2x 1
· -
a ) 11m sm- -
x l'1m e
= x-+o
X
- e l'1m e -x . -
= x-+o e -- - = 1.
x-+o x 2x 2x

. h2 -X 1
2sm ( sm-
. h X) 2
1
b) lim cosh x - 1 = x-+o
lim 2 = - lim ~ -
x-+o x2 2 x-+o _ 2
2

c) lim tanh x = lim ~ . sinh x . _1_ = ~.

x-+o 2x x-+o 2 x cosh x 2
LIMITS OF FUNCTIONS 169

Exercise 4.43. Determine

. 1 1
a) hm arctan - - , lim arctan - - ;
x-+1- 1- x x-+1+ 1- x

1 .
11m 1
.
b) EW-1 +e1/x' x-+o+ 1 + e1 / x '

c) lim
x~-oo
(1 + -111)X x lim
x-++oo (1 + -111)X;
x

d) x.!!~(Jx2+5x-Jx2+2x+1), lim
X~+OO
(J x2+ 5x - J x2+ 2x + 1) .
Answers.
7r 7r 3 3
a) 2' 2 b) 1, O. C) e -1 , e. d) - 2' 2
Example 4.44. Show for a> 1, k > 0 :
xk
a) lim - = 0; b) lim loga x = O.
x-++oo aX x-++oo xk
Solutions.
a) Using Example 3.12 b), we have

lim -
nk
= 0 hence lim (n + l)k
n~oo an n-+oo an = O.
This means that there exists no E N such that, for given c > 0, it holds
(n + l)k
-'-----'-- < c, for every n > no.
an
If we denote n = [x], then we have n::; x < n +1 and
xk (n + l)k
o < -aX < an
< c, for n > no.

b) Taking t = xk, t ----t +00 when x ----t +00, we get

. loga x l . loga t
hm - - = - . hm
x-++oo xk k t-++oo t
From Example 3.79 e) it follows
. loga n
11m - - =0, hence lim log a (n + 1)
n-+oo n n-+oo n = 0,
and this means that for given c > 0, there exists no E N, such that for n > no
it holds
loga(n + 1)
0<- n -< c.
Taking t > no + 1 and n = [t] (thus n > no ), we can write
loga t
o ::; - - ::; loga (n + 1) ::; c, .
I.e., 1m
l' log a
--k-
x
= O.
t n x-++oo x
170 CHAPTER 4.

Example 4.45. Prove lim (x. sin.!..)

x--+o x
= O.
Solution. For arbitrary E > 0, it follows from relations

Ix, sin ~ - 01 = Ix, sin ~I < lxi,

that there exists 8 = E, such that the following implication holds

o < Ixl < 8 =*Ix . sin ~I < E.

Remark. We leave to the reader to show for a > 0

lim
x--+o+
(x"" sin.!..) = O.
X

Example 4.46. Determine the following limits.

a) lim ((x
x-++oo
+ 3) In(x + 3) - 2(x + 2) In(x + 2) + (x + 1) In(x + 1));

b) In bX) , b > 1.
lim In( x In b) In ( -x-
X--+O+ In-
b
Solutions.
a) lim ((x
x-++oo
+ 3) In(x + 3) - 2(x + 2) In(x + 2) + (x + 1) In(x + 1))

. (x + 3)x+3( X + 1)x+l .
X+3)X+2 (x + 3) )
(( :.;;---z
=
"-+~
hm In
( (x + 2)'«") )
=
"-+~hm In +
(X-+-2)'+' (x+2)
-
~ 0.
x+1

(-x-
~b ~b-~-
X)b
In _ ) (
b) lim In(xlnb)ln = lim In(xlnb) ·In 1 + x
X--+O+ X--+O+ In _
b b
In(x/b)/lnb2
. In ( x In b) In b2 1 In b2) 1 b2
1 + ---x
11m (
x . n = n .
x--->O+ In - In-
b b

4.2 Asymptotes for the graphs of functions

4.2.1 Basic notions
Definition 4.47. A line x = Xo is called vertical asymptote for the graph of a
function f :A -+ R if at least one of the following limits
lim f(x) or lim f(x)
X-+Xo+ x-+xo-

is equal either to +CXJ or to -CXJ.

LIMITS OF FUNCTIONS 171

Definition 4.48. A line y = n is called horizontal asymptote when x -+ +00

(resp. when x -+ -(0) for the graph of a function f : A -+ R if it holds
lim f ( x) = n ( resp. lim f ( x) = n).
X~+OO x~-oo

Definition 4.49. A line y = kx + n, k i- 0, is called slanted asymptote when

x -+ +00 (resp. when x -+ -(0) of the graph of a function f : A -+ R if it holds

lim (f(x) - (kx

X~+OO
+ n» = 0 (resp. lim (f(x) - (kx
x-+-oo
+ n» = 0).

The numbers k and n can be ordered from the following.

k = lim f (x ) ( resp. k = lim f (x ) )

x-++oo X x-+-oo x
(4.9)
n = lim (f(x) - kx) (resp n = lim (f(x) - kx)),
x-++oo x-+-oo

provided that both limits in (4.9) when x -+ +00 (resp. when x -+ -(0) exist.

4.2.2 Examples and exercises

Example 4.50 Determine the asymptotes for the graphs of the following functions.
2x I-x
a) f(x) = y'4 _ x2 ; b) f(x) = I n - ;
l+x
x
e) f(x)=I+x 2 ; d) f(x) = vX+T - v'x=l;

e) f(x) = Vi + vi4=X; f) f(x) = arcsin eX.

Solutions.
a) The function f is defined on the interval (-2,2). From
2x . 2x
lim
x-+-2+ y'4 - x 2
= -00, hm
x-+2- y'4 - x 2
= +00,
it follows that the graph of f has vertical asymptotes x = 2 and x = -2.
b) The function f is defined on the interval (-1, 1). From
I-x I-x
lim In - - = +00, lim In - - = -00,
x-+-H 1+x x-+l- 1+x
it follows that the graph of f has vertical asymptotes x = 1 and x = -1.
e) The graph of f has a horizontal asymptote y = 0 when both x -+ +00 and
x -+ -00, because
. x
hm - - = 0
x-++oo 1 +x2
and
x
lim - - =0.
1 +x2
x-+-oo
172 CHAPTER 4.

d) The function f is defined on the interval [1, +00). The graph of f has no vertical
asymptotes, because

lim
x-+1+
(Vx+1 - Vx=1) = Vi.
It has a horizontal asymptote y = 0 when x -t +00, because
lim
x-++oo
(Vx+1 - Vx=1) = O.

e) The function f is defined on the interval [0,4J. Its graph has no asymptotes.
f) The function f is defined on the interval (-00, OJ. It has no vertical asymptotes,
because
.
11m . x 7r
arcsIne =-.
x-+o- 2
However, from
lim arcsin eX = 0,
x---+--oo

it follows that the graph of f has a horizontal asymptote y = 0 when x -t -00.

Example 4.51 Determine the asymptotes for graphs the following functions.
1 1
a) f(x) = -1-2; b) f(x) = x - -;
-x x

c) f(x) = arctan x; d) f(x) = v'X2 + 3;

e) f(x) = In{l + eX); f) f(x) = x 2 + 1.

x-2

Solutions.
a) The natural domain of the function f is the set R \ {I, -I}. Its graph has
vertical asymptotes x = 1 and x = -1, because
.
11m 1
- - - =-00 and lim _1_2 = +00.
x-+1+ 1 - x2 x-+-1+ 1- x

(Note that lim _1_2 = +00, lim _1_2 = -00.) Its graph has a
x-+1- 1 - X x-+-1-1- x
horizontal asymptote y = 0 when x -t +00 and when x - t -00, because

lim _1_ =0 and lim _1_ = O.

1 - x2
x-++oo x-+-oo 1 - x2

b) The graph of the function f has a vertical asymptote x = 0 and a slanted

asymptote y = x when x - t +00 and when x - t -00, because
1
x--
k= x = 1 and
lim _ _
x-+±oo x
n= lim
x-+±oo
(x - ~x - x) = O.
LIMITS OF FUNCTIONS 173

c) The graph of f has two horizontal asymptotes, namely y = -7r/2 when x ----t -CXJ
and y = 7r /2 when x ----t +CXJ.

d) The graph of f has two slanted asymptotes, namely y = -x when x ----t -CXJ and
y = x when x ----t +CXJ.

e) The graph of f has a horizontal asymptote y = 0 when x ----t -CXJ, because

lim In(l
x---+-oo
+ eX) = O.

Note that
lim In(l + eX) = +CXJ,
x---++oo

hence it has no horizontal asymptote when x ----t +CXJ. However, it has a slanted
asymptote y = x when x ----t +CXJ, because

lim In(l + eX) = 1 and lim (In(l + eX) - x) = o.

X_+= x x-+=

f) The graph of f has a vertical asymptote x = 2 and a slanted asymptote y = x +2

when x ----t +CXJ and when x ----t -CXJ.
Exercise 4.52 Determine the asymptotes for the graphs of the following functions.

a) f(x) = x 2 e x ; b) f(x) = e- x2 ;

c) f(x) = e1 / x ; d) f(x)=e- 1 / x2 ;

e) f(x) = e1jx2; f) f(x) = xe- 1 / x .

Answers. The graph of f has a
a) horizontal asymptote y = 0 when x --t -CXJ;
b) horizontal asymptote y = 0 when x ----t +CXJ, and also when x ----t -CXJ;
c) vertical asymptote x = 0 and a horizontal asymptote y = 1 both when x ----t -CXJ
and when x ----t +CXJ;

d) horizontal asymptote y = 1 both when x ----t -CXJ and x ----t +CXJ, because
lim e- 1/ x2
x---+-oo
=1 and lim e- 1 / x2
x---++oo
-
-,
1

but has no vertical asymptotes, because

lim e- 1/ x2
x---+o-
=0 and lim e- 1/ x2
x---+o+'
= o·
e) vertical asymptote x = 0 and a horizontal asymptote y = 1 when x ----t -CXJ and
also when x ----t +CXJ;

f) vertical asymptote x = 0 and a slanted asymptote y = x-I when x ----t -CXJ and
also when x --t +CXJ.
174 CHAPTER 4.

4.3 Asymptotic relations

4.3.1 Basic notions
In this subsection we shall suppose that the domains of the functions f and 9 contain
a set (a,xo)U(xo,b).
Definition 4.53. We say that a function f is asymptotically equivalent to the
function 9 as x approaches Xo if there exists a function 1> such that

f(x) = 1>(x), g(x), x E (a, b), x::J Xo,

and it holds lim
X-+Xo
1>( x) = 1.
Then we write f( x) rv g( x) as x - Xo. A sufficient condition for the asymptotic
equivalence f(x) rv g(x) as x _ Xo is the equality lim f((X)) = l.
X--+Xo 9 x

Definition 4.54. We say that a function f is small oh of the function 9 as x

approaches Xo if there exists a function 1> such that
f(x) = 1>(x) . g(x), x E (a, b), x ::J Xo,
and it holds lim
x-+xo
1>( x) = o.
Then we write f = o(g) as x - Xo. If g(x) ::J 0 for every x::J Xo, then a necessary
and sufficient condition for the asymptotic relation f(x) = o(g(x)) as x - Xo is the
1· l'1m -
equa1ty
f(x)
() = .
0
x--+xo 9 x

In particular, if g(x) == 1, then f(x) = 0(1) as x - Xo. This means that f tends
to zero as x - Xo.
Definition 4.55. We say that a function f is big oh of the function 9 as x
approaches Xo if there exists a constant K > 0 such that

If(x)1 ~ K 'lg(x)l, x E (a, b), x ::J Xo.

Then we write f(x) = O(g(x)) as x - Xo. In particular, if g(x) == 1, x E (a, b), then
f(x) = 0(1) as x - Xo. This means that, for some b > 0, the function f is bounded
on the set (xo - b,xo + b) \ {xo}.
These tree notions, namely the asymptotic equivalence, "big oh" and "small oh",
can also be defined in the cases x - Xo + 0, x - Xo - 0, x _ +00 and x _ -00.

4.3.2 Examples and exercises

Example 4.56. Show that
1 .
a) e - l
X rv
2 sm2x, X-Oj b) sinh x rv x, x _ 0;

In3
c) ( X+l)X rve2, x-+oo; d) In(1 + 3X ) rv -1-In(1 + 2X ), x - +00.
x-I n2
LIMITS OF FUNCTIONS 175

Solutions.
eX -1
a) From the equality lim -.-- = 1/2, by Definition 4.53 it follows
x-+o SIn 2x

Xl.
e -1", "2sm2x, x --t 0.

b) Follows from Example 4.42 a).

c) Follows from Example 4.35 a).

d) Follows from Example 4.38 h).

Example 4.57. Prove that f(x) '" g(x), X --t 0, where

sinx, x E Q; X, x E Q;
f(x) = { and g(x) = {

0, x E R \ Q, 0, x E R \ Q,

with Q denoting the set of rational numbers.

. sm x
SolutIOn. Let us put 1jJ(x) = - - , x i= 0. Then it holds
x

f(x) = 1jJ(x)g(x) and x-+o

lim 1jJ(x) = 1,

and by Definition 4.53 this means that f(x) '" g(x), when x --t 0.

Remark. Note that in this case the limit lim f((X)) does not exist.
x-+o g x

Example 4.58. Determine the following limits.

. sin 3x + 2 arctan 2x + 3x 2
11m In cos x
a) . b) lim 2;
x-+o In(l + 2x + sin 2 x) + xeX ' x-+o tan x

c) lim x (In (1
X-++OO
+ :.)
2
- In :.) ;
2
d) ~~(1 - x 3 )cot X.

Solutions.

a) From the following asymptotic behaviors

sin3x",3x, arctan2x",2x, xe X ", x, In(1+2x+sin 2 x) ",2x+sin 2 x",2x,

when x --t 0, we have

lim sin 3x + 2 arc~an 2x + 3x 2 = lim 7 x = ~.

x-+o In(l + 2x + sm 2 x) + xe x-+o 3x 3
176 CHAPTER 4.

b) Since

2ln cos x = In(l - sin 2 x) '" - sin 2 x '" _x 2 , tan x 2 '" x2 when x -+ 0,

we have
lim In cos x = -1 lim x 2 = _ ~.
x-+o tan x 2 2 x-+o x 2 2

c) lim x (In (1
x-++oo
+ :.)
2 - In:') = lim x In (1
2 x-++oo
+ 3.)
X

= lim x
X~+OO
(3.x + (3.))
x
=
0 lim (2
x~+oo
+0 (1~)) = 2.
X

d) From the expressions

In(l - x 3 ) _x 3
lim(l - x 3 )cot x = l, where £ = lim = lim - - = 0,
x-+o x-+o tan x x-+O x

it follows
lim(l - x 3 )cot x -- I .
x--+o

Exercise 4.59. Prove the following asymptotic relations.

xm xm
a) '" xm 0· b) - - - '" x m - n , -+ +00;
1 + xn
X -+
1 + xn
X
' ,

c) 2-2cosbx",bx 2 , x-+O, b#O; d) arctan 2x '" 2x, x -+ 0;

7r 1 1 7r
e) arctan x '" 2' x -+ +00; f)
-cos m X '" (7r
2 -x
)m' X -+ 2.

In this case, m and n are natural numbers.

Exercise 4.60. Determine the numerical constants a and b such that f(x) '" ax b,
when x -+ Xo, if

a) f(x) = V3x + Jx + ..;x, Xo = 0;

b) f(x) = J3x + Jx +..;x, Xo = +00;

c) f(x) = 5ex ' + (cos x -1)2 + x 6 - 5, Xo = 0;

d) f(x) = sin 2 3x + arcsin 2 x + 2 arctan x 2 , xo=O.

LIMITS OF FUNCTIONS 177

Answers.
1 1
a) a=l , b=-
8 '3 b = -2'
b) a = Vv,

21
c) a -- -4' b -- 4 . d) a = 12, b = 2.

Example 4.61. Show the following asymptotic equalities.

a) x 2 sin ¥i = {J7;'i + o( {J7;'i), x -+ 0; b) eX - 1 = x + o(x), x -+ 0;

c) loga x = o(XC), a> 1, c> 0, x -+ 0+.

Solutions.

a) From the following limits

. x 2sin ¥i = 1 . (X2 sin ¥i - {J7;'i)

hm
x_a {J7;'i
x7
and hm
x_a {J7;'i x7
= 0,

we obtain the given asymptotic equality.

b) Follows from Example 4.39 a).

c) Follows from Example 4.44 b).

Example 4.62. Determine the following limits.

a) lim (2e x /(x+1) _ 1) (x 2+1)/x , · In cos ax

b) 11m ,a, b TO,
-I. .
x--+o •
x_a In cos bx

c) hm
. (1 + sin xcos ax) cot
3
X

x_a 1 + sin x cos bx

Solutions.
x
a) From the following expression e x /(x+1) -1 = --+o(x), x -+ 0, and Example
x+1
4.34 we can write

lim ( 2e x /(x+1)
x--+o
- 1) (x2+1)/x = exp (
lim(
2 e x /(x+1)
x--+o
- 1) .x-+-
2

X
1)
exp ( hm2
x_a
--+o(x) ) .x-+-
. (X
x +1 x
2 1)
exp
' 2(X2-+-1+ (x 2+ l)O(X))) = e2.
(1x_a
1m
x +1 x
178 CHAPTER 4.

b) Using the asymptotic equality In( 1 + x) = x + o( x), x --., 0, we have

a2x 2
a2x2
In ( 1- - + o(x 2)) _ _ + o(x 2)
. In cos ax
hm
x-+o In cos bx
l'
= 1m
x-+o In
(
1_ b2
2
2
+ o(x 2)
)
+ = lim -~}"-;2;----
x-+o b x + o(x 2)
--2-
a2
b2 •

c) Let us first consider the following transformations when x --., O.

(
1 + sin x cos ax ) 3 cos ax - cos bx cos 3 x
-----~-1 ·cot x= .--
1 + sin x cos bx 1 + sin x cos bx sin 2 x

b22
(- -a 2
- x 2 +o(x) 2) (1-2"+o(x
X2 2) )3
1+ (x + o(x)) . (1 _b2;2 + O(X2))' (X2 + o(x 2))

b2 - a2 x 2 + o(x 2)
2
x 2 + o(x 2)
Then we have

. (1 + sin. x cos ax) cot

hm
3
x
x-+o 1 + sm x cos bx

(.
exp hmcot 3 x·
x-+o
(1 + sin. x cos axb -1 ) ) =exp((b
1 + sm x cos x
2 -a 2 )/2).

Example 4.63. Are the following asymptotic equalities correct?

a) 1995x + xcosx = O(x), x --., +00; b) x = O(1995x + xcosx), x --., +00;

c) x = O(x + x cos x), x --., +00; d) v'x 2 + 3 - x = 0 (~) , x --., +00.

Solutions.
a) Since it holds 11995x + xcosxl ::; Clxl, x E R, (for example for x 2: lone can
take C = 1996), it follows that the given statement is true.
b) The statement is true, because for x 2: 1 it holds

Ixl ::; Ix + (1994 + cos x)xl = 1 . 11995x + x cos xl.

c) The statement is not true, because for every Xo > 0 and for every C > 0, there
exists Xl > Xo such that it holds
1
l+cosxl<C or XI>C(X1+XICOSXI).
LIMITS OF FUNCTIONS 179

d) The statement is true, because it holds

1
Iy' X2 + 3 - x I < 3 . -,
x
x:2: 1.

Exercise 4.64. Show

x2
a) l-cosx=2+0(x),x~0; b) aX -1 = xlna + o(xlna), x ~ 0;

c) x + xcosx = O(x), x ~ 0; d) .!.x = 0 (v9+1-lxl), x ~ +00.

Example 4.65. Let f be a positive function on some neighbourhood of the point Xo.
Prove the following asymptotic relations when x ~ Xo.

a) O( O(J)) = O(J); b) 0(0(J)) = 0(J);

c) 0(0(J)) = 0(J); d) 0(J) + O(J) = O(J).

Solutions. We shall prove only parts a) and d), the others are left to the reader.

a) Put 9 = O(J) and h = O(g). Then by Definition 4.55 there exist constants
KI > 0 and K2 > 0 such that in some neighbourhood U of Xo it holds for
x # Xo
Ig(x)1 ::; Kd(x) and Ih(x)1 ::; K 2g(x)
Then on the set U \ {xo} it holds

Ih(x)1 ::; K2Kd(x),

which gives the statement.

d) Put 9 = 0(J) and h = O(J). Then by Definitions 4.54 and 4.55 there exist a
function 1>, a constant K > 0 and a neighbourhood U of Xo such that

g(x) = 1>(x)f(x), lim 1>(x) = 0 and Ih(x)l::; Kf(x), x E U \ {xo}.

X~Xo

Then the sum of the functions 9 and h on the set U \ {xo} can be written as

g(x) + h(x) = 1>(x)f(x) + h(x).

Since 1>( x) tends to zero as x ~ xo, there exists a neighbourhood UI C U of
the point xo, such that on the set UI \ {xo} it holds 11>(x)1 ::; 1. Thus

(Vx E UI \ {xo}) Ig(x) + h(x)1 ::; 1· f(x) + K· f(x) ::; Kd(x)

for KI := 1 + K.
180 CHAPTER 4.

Exercise 4.66. Show that if the functions f and 9 have the properties f( x) =J. 0
and g( x) =J. 0 for x =I- Xo, then as x --t Xo it holds

f(x) rv g(x) -{=} g(x) - f(x) = o(g).

Hint. From the equality lim f((X))

X-Xo 9 x
=1 it follows lim (1 - f((X)))
X-Xo 9 x
= 0, wherefrom
we obtain
. g(x) - f(x) .
hm
X-Xo
()
9
= 0,
X
meanmg g(x) - f(x) = o(g), X --t Xo.

Example 4.67. Let us suppose that f(x) rv h(x) and g(x) rv gl(X), when x --t Xo.
If there exists lim h ((x ))' then there exists lim f(( x)) satisfying
X-Xo gl X x-xo9 X

lim f(x) = lim Il(x).

X-Xog( x) X-Xo gl (x)

Prove.

Solution. From Example 4.66 it follows f( x) = 11 (x) + O(JI (x)) and g( x)

gl (x) + O(gl (x)) when x --t Xo. Therefore we can write

1 + O(Jl(X))
lim f(x) = lim f1(X) + 0(J1(X)) = lim f1(X) . lim f1(X) = lim f1(X)
x-xo g( x) x-xo gl (x) + O(gl (x)) x-xo gl (x) x-xo 1 + ---,O(g=--l,:. .x:..c..))
.,( x-xogl (x)"
gl(X)
Chapter 5

Continuity

5.1 Continuity at a point

5.1.1 Basic notions
Definition 5.1. A function f : A c R ----+ R is continuous at a point Xo E A if
for every e > 0 there exists a number 8 > 0, depending on e and on the point Xo,
such that for every x E A with the property Ix - xol < 8 it holds If(x) - f(xo)1 < c.

Using logical symbols, the last definition can be expressed as follows.

A function f : A C R ----+ R is continuous at a point Xo E A iff

(\Ie> 0) (38) 0) (\Ix E A) Ix - xol < 8 =} If(x) - f(xo)1 < c. (5.1)

If the point Xo E A is an accumulation point of the set A, then the following two,
mutually equivalent, definitions can also be used.

Definition 5.2. A function f : A C R ----+ R is continuous at a point Xo E A,

where Xo is an accumulation point of the domain A, if it holds

lim f(x) = f(xo).

X----l-XO, xEA

Definition 5.3. A function f : A C R ----+ R is continuous at a point Xo E A,

where Xo is an accumulation point of the domain A, if for every sequence (Xn)nEN
of elements from A it holds that

lim f(x n ) = f(xo).

n--+oo

Definition 5.4 .

• Assume that the domain of a function f contains an interval (a, xo]. Then the
function f : A ----+ R is continuous from the left side at the point Xo if

lim f(x) = f(xo).

X---+Xo-

181
182 CHAPTER 5.

• Assume that the domain of a function f contains an interval [xo, b). Then the
function f : A --+ R is continuous from the right side at the point Xo if

lim f(x) = f(xo).

x--+xo+

Theorem 5.5. A function f : (a, b) --+ R is continuous at a point Xo E (a, b) if and

only if both one sided limits of f at Xo exist, the limit of the function f at Xo also
exists, and all these three limits are equal to f(xo), i.e.,

lim f(x) = lim f(x) = lim f(x) = f(xo).

X--+Xo - X-+Xo+ X--+XO

In view of this theorem, one can classify the discontinuities of a function f defined
on an open interval. Namely, if a function f : (a, b) c R --+ R is discontinuous at
a point Xo E ( a, b), then there are three possibilities.
• If the limit lim f(x) exists and equals to some number L
X--+Xo
:f- f(xo), then f has
a removable discontinuity at the point Xo;

• if both one sided limits

lim f(x) =: Ll and lim f(x) =: L2 (5.2)

x--+xo- x--+xo+

exist, but are unequal, Ll :f- L 2, then f has a first order discontinuity at
the point Xo;

• if at least one of the two one sided limits in (5.2) does not exist, then f has a
second order discontinuity at the point Xo.
A function f : A C R --+ R is continuous on a set B C A if it is continuous at
every point from B. It is important to note that all basic elementary functions are
continuous on their natural domains.
If the functions f : Al C R --+ Rand 9 : A2 C R --+ R are continuous at a point
Xo E Al n A2 (resp. on a set B C Al n A 2), then the following functions are also
continuous at the point Xo E Al n A2 (resp. on the set B C Al n A 2) :
• f + 9 (sum of f and g),
• f· 9 (product of f and g),

•£ (quotient of f and g),

9

Of course, for the quotient £ this holds only when g(xo) :f- 0 (resp. when g(x) :f- 0
9
for every x E B).
Assume that a function 9 : A C R --+ R is continuous at a point Xo E A and that
a function f: B C g(A) --+ R is continuous at the point g(xo). Then the composite
function h = fog is also continuous at the point Xo. Shortly we say that the
• composition of continuous functions is continuous.
CONTINUITY 183

5.1.2 Examples and Exercises

Example 5.6. Using Definition 5.1, check the continuity of the following functions
at an arbitrary point Xo from their domains.

a) f(x) = 2x + 1, x E Rj b) f(x) = x 2, X E Rj

c) f(x) =..;x, x E [O,+oo)j d) f(x) = Vx, x E Rj

e) f(x) = arctan x, x E R.

Solutions.
a) Fix Xo E R. Let c > 0 be given. From the inequality

If(x) - f(xo)1 = 21x - xol < 28, x E R,

it follows that choosing 8 := c/2 the implication (5.1) holds, i.e.,

(\Ix E R) Ix - xol < 8 = c/2 =? If(x) - f(xo)1 < c.

b) Fix Xo E R. Let c > 0 be given. Then it holds

If(x)-f(xo)1 = IX2_X~1 = I(X-XO)2+2xO(X-xo)1 ~ Ix- xoI 2+2I xollx-xol·

In order to determine 8 > 0, we assume Ix - xol < 8 and 8 < 1, hence

If(x) - f(xo)1 < 82 + 21xol8 < 8(1 + 2lxol),

and assume that the last expression is less than c. Then choosing

8:= ~min {1, 1 + ~Ixol}'

it follows that

(\Ix E R) Ix - xol < 8 =? If(x) - f(xo)1 < c.

Remark. The number 8 from Definition 5.1 (hence also in the last example) is
not uniquely determined, Namely, if for given c, we find a 8 such that for every
x E (xo - 8, Xo + 8), it holds If(x) - f(xo)1 < c, then for every other 81, 0 < 81 < 8,
it holds
Ix - xol < 81 =? If(x) - f(xo)1 < c.
Again, it is important that for given c > 0 a 8 > 0 exists.
c) Fix first Xo > O. Let c > 0 be given. From the relations

Ix - xol Ix - xol
1v'X - VxOI = ..;x + JXo <JXo,
184 CHAPTER 5.

assuming that Ix - xol < b, it follows

b
If(x) - f (Xo )1 < -..jXo < c.

Hence we can determine b:= c..jXO, which then gives the implication (5.1).
The case Xo = 0 is left to the reader.

d) Fix Xo > 0; the case Xo = 0, being somewhat easier, is omitted.

Let c > 0 be given. If Ix - xol < b, then
Ix -xol
V9 + flxxo + fii
1~-1Xo1 =

Ix-xol b
2 < -3 .
(
~ + 1Xo
2
)
+ ~4 Vay6 X
-(i6
4

Hence b can be determined as b:= c' ~(i6.

e) Let c > 0 be given. Let first Xo = O. Then for all x E R it holds
If(x) - f(O)1 = I arctan x - arctan 01 = Iarctan xl ~ lxi,
see Example 6.89 b). Clearly, choosing b := i, it follows

Ix - 01 < b =} If(x) - f(O)1 < c.

Assume now Xo =1= 0 and let x > Xo; i.e., <51 := x - Xo > O. Using Example 2.28
f), it follows that
x - Xo
Iarctan x - arctan Xo I arctan 1 + xXo

x - Xo bi bi
< < < c,
1 + xXo 2 i:
1 + Xo + UIXO - 1 + Xo - blXo
2

. 1 + x5
prOVIded that bi < C.
1 + cXo
Similarly, if x < Xo, i.e., b2 := -(x - xo) > 0, then
b2
Iarctan x - arctan xol ~ 8 < c, (5.3)
1 + Xo - 2XO
2

1 +x 2
provided that again 82 < 0 c.
1 +cxo
Hence, putting
8 := ~ . min { 1 + x5 c, Ixol }
2 1 + cXo 2
(the latter in order to keep the intervals (xo - 8j, Xo + 8j ), j = 1,2, away from
the point 0) we get the implication in (5.1).
CONTINUITY 185

Exercise 5.7. Check the continuity of the following functions at an arbitrary point
Xo from their domains.

a) f(x) = sinx, x E R; b) f(x) = cosx, x E R;

c) f(x) = arcsin x, x E [-1,11; d) f(x) = arccos x, x E [-1,11;

e) f(x) = tanx, x E R \ {(2k + 1)71-j2\ k E Z};

f) f(x)=cotx, XER\{hlkEZ}.

Exercise 5.B. Check the continuity of the following functions at an arbitrary point
Xo from their domains.

a) f (x) = eX, x E R; b) f(x) = aX, x E R, a> 0;

c) f(x) = lnx, x E (0,+00); d) f(x)=logax, xE(O,+oo),a>O;

e) f(x) = sinh x, x E R; f) f(x) = cosh x, x E R;

g) f(x) = tanh x, x E R; h) f(x) = cothx, x E R \ {O}.

Exercise 5.9. Check the continuity of the following functions at an arbitrary point
Xo from their domains.

a) f(x) = cos(ax + b), x E R; b) f(x)=x 2 +2x+2, xER;

c) f(x) = x3 + 2cosx, x E R; d) f(x) = y'xsin3x, x E R;

2 - 2 cosx
e) f(x) = x cot x, x E R \ {hi k E Z}; f) f(x) = ? ,x E R \ {O}.
x

Example 5.10. Using Definition 5.1, check the continuity of the following functions
at an arbitrary point Xo from their domains.

x # 0; 0, x E 1:= R \ Q;
a) f(x) = { 1:1' b) D(x) = { 1,
x E Q.
-1, x = 0;

Here Q and I stand for the (mutually disjoint) sets of rational and irrational num-
bers, which together make the set of real numbers R.

Remark. The function in b), called Dirichlet's function, was already defined in
Example 2.20.
Solutions.
186 CHAPTER 5.

a) From the definition of the absolute value, it follows that f(x) = 1 if x > 0, and
f( x) = -1 if x :S O. Let us show first that f is continuous at every point
Xo i= O. Firstly, if Ix - xol < ~Ixol, then x and Xo are of the same sign. Hence

If(x) - f(xo)1 = 11:1 - 1::11 = o.

Let c > 0 be given; putting b = I~ol, the following implication holds.
(\Ix E R) Ix - xol < b =? If(x) - f(xo)1 < c.
Next we show that f is discontinuous at Xo = O. Since

lim f(x) = 1 and lim f(x) = -1 = f(-I),

x-+o+ x-+o-

the function f is only continuous from the left at 0, but it is not continuous
at the point O. In fact, f has a first order discontinuity at o.

b) We claim that Dirichlet's function is discontinuous at every real point Xo, rational
or irrational. We shall prove this for Xo E Q. (The other case, namely Xo E I,
is handled in a completely analogous way.)
In order to prove the discontinuity of the function D at xo, we must show
that there exists an c > 0 such that for every b > 0 there exists an x E R (x
depending on b) with the property

Ix - xol < 8 =? ID(x) - D(xo)1 ~ c.

Let us put c := 1/2. Then for every b > 0 there exists an irrational number X5
such that IX5 - xol < b. However, then it holds

ID(x5) - D(xo)1 = 10 - 11 = 1 > 1/2.

Here we used the fact that the set I of irrational numbers is dense in the set
Q of rational numbers; i.e., in every neighborhood of any Xo E Q there exists
at least one irrational number. In fact, there are infinitely many irrational
numbers in any neighborhood of an arbitrary number Xo E Q. (See Example
1.47 e), from which the last statement easily follows.)

Exercise 5.11. Prove that the function f is discontinuous at every x E R, if

-I, x E Q;
f(x) = { 1, x E R \ Q.

Exercise 5.12. If the function f is discontinuous at a point Xo E R, is then the

function p also discontinuous at that point?

Answer. Not necessarily. (See the previous exercise.)

CONTINUITY 187

Example 5.13. Assume that the function f: (a, b) -+ R has the following property
at a point Xo E (a, b).

a) (Vc > 0) (38) 0) ("Ix E (a, b» Ix - xol < 8;

If(x) - f(xo)1 < c =>
b) ("18> 0) (3c > 0) ("Ix E (a, b» If(x) - f(xo)1 < c => Ix - xol < 8;

c) ("18> 0) (3c > 0) ("Ix E (a, b» Ix - xol < 8 => If(x) - f(xo)1 < c.
What can one say about the continuity of f at xo?

Solutions. We shall find three functions that will satisfy the stated implications in
a), b) and c) respectively, but neither of them will turn out to be continuous at the
point Xo. This means that the order of the quantifiers as well as the direction of the
implication in Definition 5.1 are essential.
a) Let us define the function f by (see Figure 5.1).

X, x::; 1;
f(x)= {
x+1, x>1.

lY

J
lY

2 / x
0

'(
Figure 5.1. Figure 5.2.

Clearly, this function is continuous on the set R \ {I}, while it has a first order
discontinuity at the point Xo = 1. Let us show, however, that f does have the
stated property at the point Xo = 1. To that end, for c > 1 we put 8 := c - 1.
Then for every x E R it holds

If(x) - f(1)1 < c => Ix - 11 < 8. (5.4)

For c ::; 1, we put simply 8 := c. Then the set of points x > 1 that satisfy the
inequality If(x) - f(l)1 < c is empty, hence the implication (5.4) is true. For
x < 1 it holds
If(x) - f(1)1 < c -¢:=} Ix - 11 < c.
188 CHAPTER 5.

b) The function f given by

X2, x ¥= 0;
f(x) = { 1, x = 0,

has a removable discontinuity at Xo = 0, while it is continuous on the set

R \ {O}. Let us show that it satisfies the stated condition at the point Xo = O.
b2
For given b > 0 let us choose c such that 1 < c < 1 + P (say c := 1 + 2).
Then for every x E R \ {O} it holds

(If(x)- f(O)1 = Ix 2 -11 < x 2 +1 < c) =?- (Ix-Ol = Vlxl2 + 1 - 1 < ~ < b).

c) Let the function f given by

x2 - 4x + 5, x> 2;
f(x) = { 0, x = 2;
_x 2 + 4x - 5, x<2
(see Figure 5.2). This function has a first order discontinuity at the point
Xo = 2, and is continuous on the set R \ {2}. Let us show that it satisfies the
stated condition at the point Xo = 2. Namely, for given b > 0 let us define
c := 1 + b 2 • Then for every x E R it holds

(Ix - 21 < b) =?- (If(x) - f(2)1 = 1 + (x - 2)2 < 1 +b2 = c).

In fact, the condition stated in c) gives only the boundedness of the function
f in some neighborhood of the point Xo.
Example 5.14. Prove that if the function f : (a, b) -+ R is continuous at a point
Xo E (a, b) and it holds f(xo) ¥= 0, then there exists a b > 0 with the following
property.
f(xo) .
a) (Vx E (a, b) Ix - xol < b =?- f(x) > -2-' promded that f(xo) > 0;

f(xo) .
b) (Vx E (a, b)) Ix - xol < b =?- f(x) < -2-' promded that f(xo) < O.

Solution. We shall prove only the case a), the other one being analogous. By the
definition of the continuity of f at Xo, for c := f(;o) , there exists a b > 0 such that
the following implication holds.
f(xo)
(Vx E (a, b)) Ix - xol < b =?- If(x) - f(xo)1 < -2-.

Hence for all x E (a, b), such that Xo - b < x < Xo + b, it holds
(_f(;o) < f(x) _ f(xo) < f(;o)) =?- (f(;o) < f(x) < 3f~xo)).
CONTINUITY 189

Remark. This example gives an important property of a continuous function f at

some point Xo where f(xo) i- o. Namely, then there exists an interval (xo - 5, Xo + 5)
on which f has the same sign as the number f(xo).
Example 5.15. Check the continuity at 0 of the following functions.

a) f(x) = { l'i:"I, xi- 0;

b) f(x) =
{ "n x
H'
xi- 0;

1, x = 0; 1, x = 0;

c) f(x) =
{ ,in (~) , x
xi- 0;
d) f(x) =
{ "in (~), x
xi- 0;

C, x = 0; 0, x = o.
In c), C stands for some real number. We suggest to the reader to prove the conti-
nuity of these four functions on the set R \ {O}.
Solutions.
a) Let us show that f is continuous at the point O. Using the limit lim smx = 1,
! x-a X
see Example 4.23, we have

11m x ! = 1·1m I-
· f() x I = 1·1m -
sin- sin-
x = 1 and
x_o+ ! x_o+ X x_o+ X

I I
I . •
· f()
I1m x = 1·1m - x = 1·1m -
sm- - sm
-- x = l.
x_o- x_o- x x-o--x
Since also f(O) = 1, in view of Theorem 5.5, f is continuous at o.
b) Since . .
. smx. smx
hm
x_o+
f( x) = lim -,-, = hm -
x_o+ x x_o+ X
= 1 and

· f()
11m x = 1·1m - x = 1·1m -
sin- sin-
x =-1
x--+o- x--+o- Ixl x--+o- -x '
the function f has a first order discontinuity at the point O.
c) Let us prove first that f has no limit at O. If Xn = _1_, n E N, then lim Xn = 0,
n7r n--+oo
2
and if x~ = n E N, then also lim x~ = O. However, it holds
4n + 1 1f
f )'
n-oo

lim f(x n ) = n-+oo

n--+oo
lim sin(n1f) = 0 and lim
n---+oo
f(x~) lim sin ((4n
= n-+oo + 1)~)
2 = l.

In view of Definition 4.2, it follows that f has no limit at o. Hence, from

Theorem 5.5 we obtain that regardless of the choice of the number C, the
function f has a second order discontinuity at O. (See Figure 2.15.)
190 CHAPTER 5.

d) Let us prove that f is continuous at O. For given c > 0 let us choose 8 := c.

Then for every x E R such that 0 < Ix - 01 < 8 it holds

If(x) - f(O)1 = Ix sin.!.x - 01 ~ Ixl = Ix - 01 < 8 = c.

Thus limf(x) = 0 = f(O), and from Theorem 5.5 follows the continuity of f
x-+o
at the point O. (See Figure 2.16.)

Example 5.16. Find the points of discontinuities and their kind for the following
functions.

I, x> 0;
a) sgn x:= { 0, x = 0; b) f(x) = sgn 2 x, x E R.
-1, x < 0;

Solutions.

a) Since
lim sgnx = 1 and lim sgnx = -1,
X-+O+ x--+o-

f has a first order discontinuity at O. (Figure 5.3.)

b) Since sgn 2 x = 1 for x '# 0, it follows that f has a removable discontinuity at O.

'y

2 -

lr
'Y

I
:It
o
-
2 3
:It

0 1

-1
I
Figure 5.3. Figure 5.4.

Example 5.17. Find the points of discontinuities and their order for the following
functions.

a) f(x) = [x], x E R; b) f(x)=ax+b[x], xER, b,#O;

c) f(x) = x[x], x E R; d) f(x) = [x]· sin(7rX), x E R.

CONTINUITY 191

Solutions. Clearly, each of these four functions is continuous on the set R \ Z. We

shall next examine what happens at the integer points.
a) From the definition of the function f(x) = [x], i.e., the one that assigns to a
real number x its greatest integer part [x], it follows that f has a first order
discontinuity at every integer k. (Figure 5.4.)
b) First we find the right- and left-hand limits of f at an arbitrary integer k.
lim (ax
x ..... k+
+ b[x]) = ak + bk = (a + b)k and

lim (ax
x-+k-
+ b[x]) = ak + b(k - 1) = (a + b)k - b.

Since b i- 0, it follows that f has first order discontinuities at every integer k.

c) At the point Xo = 0 the function f is continuous, since f(O) = 0, and it holds
lim x[x]
x ..... o+
= x lim
..... o+
X· lim [x]
X"'" 0+
= 0.0 =0 and

lim x[x]
x-+o-
= x-+o-
lim X· lim [x] = O· (-1) = O.
x-+o-

At an arbitrary integer k i- 0 we have

lim x[x]
x ..... k+
= x lim
..... k+
X· lim [x] = k· k = P
x ..... k+
and

lim x[x]
x--+k-
= x-+k-
lim X· lim [x] = k· (k -1) = P
x-+k-
- k.

Hence f has a first order discontinuity at every integer k i- o.

d) The right and left side limits of f at k are both equal to zero, since
lim [x]· sin(7rX) = k lim sin(7rX) = 0 and
x ..... k+ x ..... k+

lim [x]· sin(7rX)

x-+k-
= (k - 1) lim sin(7rX)
x-+k-
= o.
Since also f(k) = 0, from Theorem 5.5 it follows that f is continuous at every
integer point.
Example 5.18. Find the points of discontinuity and their order for the following
functions.

a) f(x) = { [~], xi- 0;

b) f(x) = { X· [~] ,
xi- 0;

0, x = 0; 0, x = 0;

c) f(X)~{ [:,l,~+n(:,)), xi- 0;

0, x = O.
192 CHAPTER 5.

Solutions.

a) Putting x =~ and t = [t] + r(t), where 0::; r(t) < 1, we obtain

lim
x--+O+
[~]
X
= t--++oo
lim (t - r(t)) = +00.

Analogously it holds
lim
x--+O-
[~]
X
= -00.

This means that f has a second order discontinuity at O.

1
For x #- 0, putting again x = t' we obtain that f has first order discontinuities

at the points i, k E Z.

b) Since it holds lim x . [1/ x]

x--+O
= 1, the function f has a removable discontinuity at
O. At the point x = l/k, k E Z \ {O}, it holds

k-l
·
I1m
x--+1/k+
X· [ -1 ]
X
= l'1m ( -1
~--+o+ k
+€) . [ 1 ]
1/ k + €
= _.
1 I'1m [-k- ]
k ~--+o+ 1 + k€ -k-'

Similarly we can get

lim
x--+l/k-
X· [~]
X
= ~ . k = 1.
k

Hence f has first order discontinuities at the points i, k E Z \ {O}.

c) Let us check first the continuity of f at O. We start from the following two
sequences that converge to zero.

xn = J f. 2 " n E N, and x' _

n -
/
V7r{3+4n)'
2
n E N.

It holds
lim f(x n ) = +00
n~oo
and lim f{x~) = -00.
n~oo

Thus f has no limit or one sided limits at 0, hence f has a second order
discontinuity at O.

We have yet to examine the points ± ~, k E Z \ {O}. We claim that f has

v1kl
first order discontinuities at these points. To that end, we shall find only the
1
right- and left-hand side limits of f at the points 11 kEN, since the other
vk,
CONTINUITY 193

cases are handled similarly. We have

lim
x~l/v'k+
[~].
x
sgn (sin( ~))
X
= lim
e~O+

(~+£r
1
.~+n (~I+£r)
lim [ (1
e--->O+
k ] .sgn (sm. (1 + £VkF
+ £VkF k) = (k - 1) . sgn( sm k) .
and

lim
x~l/v'k-
[~].
x
sgn (sin( ~))
X
= lim
e~O+

(~-£r
1
. sgn (,in I I
(Vk-£
r)
lim [ kVk ]. sgn (sin k ) = k· sgn(sin k).
e~O+ (1-£ kF (1-£ Vkk)2
1
Since the left- and the right-hand side limits of f at Vk are different, the
1
function f has first order discontinuities at the points Vk' kEN.

Example 5.19. Determine the constant C, if possible, in order to obtain (at least
one sided) continuity of the functions given below at the given point xo.

{ Ixl-x x # 0;
a) f(x) = x2 ' xo = 0;
C, x = 0,

{ Ix + II x # -1;
b) f(x) = ~' Xo = -1;
C, x = -1,

c) f(X)~{ CXP(x+D· x # 0;
xo = 0;
C, x = 0,

d) f(x) ~ { C~, m'

c,
x

x
# 0;
= 0,
Xo = 0;

e) f(x) = { xlnx
2 x # 0;
,
Xo = O.
C, x = 0;
194 CHAPTER 5.

Solutions.

a) The function f can be written as

0, x> 0;
2
f(x) = { , x < 0;
x
C, x = o.
Hence
lim f(x) =
x--+o-
+00 and lim f(x) = 0,
x--+o+

which means that f has a second order discontinuity at O. If C = 0, then f

turns out to be continuous from the right at 0, but clearly for no values of C
can f be continuous.

b) Since f is equal to
-I x< -1;
f(x) = { 1,' x> -1;
C, x = -1,
it has a first order discontinuity at x = -1. For C = -1, f is continuous from
the left side, while for C = 1 it is continuous from the right side.

c) Since the following relations hold,

lim f(x) lim exp (x + .!.) = 0,

= x-+o- lim f(x) = lim exp (x + .!.) = +00,
x---+o- X x--+o+ x--+o+ X

the function f has a second order discontinuity at x o. For C 0 the

function f becomes continuous from the left side at o.

d) Since the limit of the function

f(x) = cos 2(:2)' xi- 0,

does not exist at 0, f has a second order discontinuity at x = o. Obviously,

for no values of C can f be continuous at least from one side.

e) Since from Example 4.44 b) we have

lim x In x 2 = lim 2x In Ixl = lim 21n t = 0,

x--+O x--+O t--++oo t

it follows that the function f is continuous at 0 if C = 0, resp. has a removable

discontinuity at 0 if C i- o.
CONTINUITY 195

Exercise 5.20. Determine the constant C, if possible, in order to make the follow-
ing functions continuous at the given point Xo.

{ x-2 x i= 2;
a) f(x)= x 2 -4' Xo = 2;
C, x = 2,

f(x) ={ Cx 3 + 1, x::::: 0;
b) Xo = 0;
x, x < 0,

{ ,;n2x, x <~.
- 2' 7r
c) f(x) = Xo = 2'
+ 1),
7r
C(x x>-
2'

d) f(x) ~{ 1
In(x _1)'
x> 1;
Xo = 1.
C+1, x :S 1,

Answers.

a) The function is continuous at the point Xo = 2 if C = ~.

b) The function has a first order discontinuity at the point Xo = 0, and for no value
of the constant C can f be made continuous.

c) The function is continuous at the point Xo = 7r /2 if C = o.

d) The function is continuous at the point Xo = 1 if C = -1.

Exercise 5.21. Determine the constants a and b, if possible, in order to make the

r-'
following functions f continuous on R.

x<
- 2·,

a) f(x) = ax+~, 2 < x < 3;

y'x=2, x::::: 3;

x i= 2, x i= -2;
{ 24'
1
b) f(x) = x:,
x = 2;
b, x = -2.

Answers.

a) a = 1, b = -2.
196 CHAPTER 5.

b) The function has second order discontinuities at the points Xl = 2 and X2 = -2,
and therefore we can not order the constants a and b to make f continuous on
R.
Example 5.22. Determine the constants a, band c, if possible, in order to obtain
continuity of the functions given below on their domains.

{ 2x, Ixl ~ 1;
a) f(x)= x 2 +ax+b, Ixl > 1;

{ (x-2)' X r/: {-2, 2};

x2 - 4 '
b) f(x) = a, x = -2;
b, X = 2;

1 1
x x +1 x r/: {-1,0,1};
1 1'
c) f(x) = x -1 x
a, x = -1;
b, x = 0;

I
c, x = 1;

(~)
~
x cos
-7r 37r]
sm x X E [2'2 \ {0,7r};
d) f(x)
a, x = 0;
b, x = 7r.
Solutions.
a) Since it holds
lim f(x) = lim 2x = 2, lim f(x) = lim(x 2 +ax+b)=1+a+b,
x-+l- x-+l- X"'" 1+ X"'" I+
lim f(x) = lim (x 2 +ax+b) = 1-a+b, lim f(x) = lim 2x = -2,
x-+-l- x-+-l- x ..... -1+ x ..... -1+
the function f is continuous on R only if
= 1 + a + band - 2 = 1 - a + b.
2
Hence we must choose a = 2, b = -1.

b) Since lim f(x) = -00, f has a second order discontinuity at x = -2,

x ..... -2+
regardless of the choice of a. However, since

x = l'1m (x 2- 2)2 = l'1m -

· f()
11m x --2 = 0,
X"'" 2 x ..... 2 X - 4 x ..... 2 X + 2

f is continuous at the point 2 if b = 0.

CONTINUITY 197

c) It holds
lim f(x) = +00, limf(x) = -1, limf(x) = O.
x~-l- X--+O x--+l

This means that f has a second order discontinuity at x = -1, regardless of

a; for b = -1 it becomes continuous at 0, and finally for c = 0 the function
f becomes continuous at the point 1.

d) We leave to the reader to show that

lim f(x)
x--+o
=1 and lim f(x)
X--+1r
= ~,
2

hence for a = 1 and b = %the function f becomes continuous on its domain.

Example 5.23. Examine the continuity of the compositions of the functions fog
and 9 0 f on R if
a) f(x) = sgnx, g(x)=1+x 2 ;

b) f(x) = sgnx, g(x) = x(l + x);

a) f(x)=sgnx, g(x) = 1 +x - [xl.
Solutions.
a) The function
f(g(x)) = sgn(l + x 2 ) = 1
is continuous on R. However, the function

2, x < 0;
g(J(x)) = 1 + (sgnx)2 = { 1, x = 0;
2, x> 0,
has a removable discontinuity at Xo = O.

b) In this case we have

-1, -1 < x < 0;

f(g(x)) = sgn(x(l + x)) = { 0, x = -1, x = 0;
1, x < -1, x> 0,

x :S 0;
g(J(x)) = sgn(x(l + sgnx)) = { ~: x> o.

Both functions have first order discontinuities at the point x = 0, while the
function fog has the same kind discontinuity at the point x = -1.

c) Both functions are continuous because

f(g(x)) = sgn(l +x - [xl) = 1, g(J(x)) = 1 + sgnx - [sgnx] = 1.

198 CHAPTER 5.

Exercise 5.24 Prove that a monotone function can have only first order disconti-
nuities.
Example 5.25. If the function f : A C R --t R is continuous on the compact set
K C A, then its range,
f(K) := {y E RI (3x E K) y = f(x)},
is also a compact set.
Solution. The set f(K) can be either finite or infinite. If f(K) is finite, then it is
both bounded and closed, i.e., by Definition 1.77 f(K) is then compact.
If f(K) has infinitely many elements, then for the compactness of f(K) it is
enough to prove that every infinite subset S C f(K) has an accumulation point
that belongs to f(K) (see Example 1.94). To that end, let us put
T:= {x E KI (3y E S) y = f(x)}.
Since f is a function, the last set is also infinite; being a subset of the compact set
K, T has an accumulation point Xo that belongs to K. We shall show that f(xo)
is an accumulation point of the set S. To that end, let (Xn)nEN be any sequence
from T that converges to Xo, such that for every n E N it holds Xn i- Xo. Since f is
continuous at Xo, using Definition 5.3, we have
lim f(x n ) = f(xo),
n ..... oo

which implies that f(xo) is an accumulation point of the set S.

Example 5.26. Discuss the continuity of the following functions, and sketch their
graphs.
nX _ n- X
a) f(x) = lim _1_,
n..... oo 2 + xn
x 20; b) f (x) = lim
n ..... oo nX + n- X , x E R;

c) f(x) = nlim
..... oo
y'3 + x 4n , X E R; d) f(x) = nlim
..... oo
(1 - sin 2n x), x E R;

e) f (x) = lim
n ..... oo
(x.
2 + 2 sm x
)4'
n
x E R.

Solutions.
a) It holds
0, xE[O,I);
lim xn = {
n ..... oo
1, x = 1;
+00, x> 1.
Hence the function f is equal to
1/2, o :s x < 1;
f(x) = { 1~3, x = 1;
x> 1,
and the function f has a discontinuity of first order at the point x = 1.
CONTINUITY 199

Remark. This example shows that a function obtained as a limit of a sequence of

continuous functions is not necessarily continuous.

b) We have f(x) = sgnx, hence by Example 5.16 a) the function f has a discon-
tinuity of first order at the point x = 0.

c) We shall prove that

I, Ixl ~ 1;
f(x) = { x\
Ixl > 1.
For Ixl ~ 1 it holds

V'3 ~ Y/3 + x4n ~ Y/3 + 1 = \14,

hence by Example 3.11 b) and Theorem 3.7 we obtain

f(x) = lim
n--->oo
\1'3 + x4n = 1.
For Ixl > 1 it holds

f(x) = n-+oo
lim Y/3+ x4n = n-+oo
limx 4 'Ql+ x4n
3 =X4.

Observe that the limit function f is continuous on R.

d) The function f has removable discontinuities at the points (2k + 1)7r /2, k E Z,
since it equals to

I, x =I- (2k + 1)7r/2, k E Z;

f(x ) = {
0, x = (2k + 1)7r/2, k E Z.

e) We start from the following relations.

12sinxl < 1 ~ (3k E Z) Ix - hi < 7r/6,

7r 57r
12 sinxl > 1 ~ (3k E Z) "6 < Ix - hi < 6'
Hence the limit function f is

x/2, I sinxl < 1/2, l.e., Ix - hi < 7r/6, k E Z;

f(x) = { x/3, I sinxl = 1/2, i.e., x = h ± 7r/6, k E Z;
0, Isinxl > 1/2, i.e., 7r/6 < Ix - hi < 57r/6, k E Z.
At the points Xk = k7r ± 7r /6, k E Z, f has discontinuities of the first order.

Example 5.27. Discuss the continuity of the following functions.

a) Dirichlet's function: D(x) = m----+oo

lim (lim cosn(7rm! x)), x E R;
n--+oo
200 CHAPTER 5.

b) f(x) = x· D(x), where the function D was defined in a};

c) Riemann's function:

m
R(x) = { ~, X=-,
n
mEZ, nEN, lcd(m,n)=l;
0, x E R \ Q.

Compare a) to Example 5.10 b). In c), lcd( m, n) stands for the largest common
m
divisor of the numbers m and n. When lcd( m, n) = 1, the fraction - is irreducible.
n

Solutions.

a) • First method. Assume first that x is a rational number, i.e., x = E for

q
some p E Z and q E N. Then for m > q it holds

m! x = m! E=
q
1 ·2 .. · (q - l)(q + 1)··· m· p,
hence m! x is an even number. This implies

cos(11'm! x) = 1, hence D(x) = 1 for every x E Q.

If x is an irrational number, then for no m can the number m! x be an

integer. But then

Icos (11'm! x)1 < 1, which implies lim cosn(11'm! x) = O.

n--->oo

Passing to the limit in m, the last equality gives

(Vx E R \ Q) D(x) = O.

Thus we obtained the same function as in Example 5.10 b). We proved

there that D is discontinuous at every real point x.
• Second method. Let us prove the same fact in another way, namely
using Definition 5.3.
If Xo is an arbitrary real number, then there exist two sequences, one of
rational, and the other of irrational numbers, both converging to Xo (see
the Remark after Example 1.47). Denoting the first by (1'n )nEN and the
second by (in )nEN, we have

lim rn = xo::::? n-+oo

n-+oo
lim D(rn) = 1 and

lim in = xo::::? n-+oo

n-+oo
lim D(i n ) = O.

Hence for no x E R does the limit lim D(x) exist, which means that D
X--+Xo

has a second order discontinuity at every real point Xo.

CONTINUITY 201

b) If Xo = 0, then it holds for every E >0

If(x) - f(O)1 = Ix· D(x) - 01::; Ixl < E,
provided that Ix - 01 < D := E. Thus f is continuous at o.
We shall prove next that 0 is the only real number at which f is continuous.
To that end, let Xo E R \ {O} and let (rn)nEN and (in)nEN be two sequences
of rational, resp. irrational numbers, such that

lim rn
n~oo
= Xo and lim in
n~oo
= Xo.
Then it holds

lim f(rn)
n~oo
= xo·1 = Xo and lim f(i n) = Xo· 0 = O.
n~oo

This means that f has a second order discontinuity at every real point different
of zero.

c) We shall prove that Riemann's function is continuous at every irrational, but is

discontinuous at every rational number.
Assume first Xo E R \ Q. Let us observe a sequence of rational numbers (rk)kEN
converging to xo, where rk = m k , and assume that lcd(mk' nk) =1 for every
nk
kEN. But then lim nk = +00, hence
k->oo

lim R(rk) = lim 0 = lim R (mk) = lim ~ = 0 = R(xo). (5.5)

k->oo k->oo k->oo nk k->oo nk
It is trivial that for every sequence of irrational numbers (ikhEN converging
to Xo it holds
lim R(ik) = 0 = R(xo). (5.6)
k->oo

Putting the results from (5.5) and (5.6) together, it follows that for every
sequence of real numbers the same remains true. But this means that R is
continuous at every irrational number.
Assume now that Xo is a rational number of the form Xo = m, where
n
lcd(m, n) = 1. By the definition of the function R, it holds R(xo) = .!.. Let us
n
put now
mk+l
rk kEN.
:= ~~

nk
This sequence of rational numbers converges to Xo as k -+ 00; however, it holds

lim R(rk) lim ~k = 0 -:I

= k->oo .!.n = R(xo).
k->oo n

Hence the function R has a discontinuity at every rational point.

202 CHAPTER 5.

Example 5.28. The function f : R -7 R is given by

0, x = OJ
f(x) = { 1/x, x E Q \ {O}j
x, x E 1:= R \ Q.

a) Prove that f is a bijection.

b) Discuss the continuity of the function f at the points x' = 1, x" = -1 and
XIII = O.

c) Discuss the continuity of the function f at the point Xo E R \ {-1, 0, 1}.

Solutions.
a) We have to prove that f is both an injection,
(VXl' X2 E R) Xl f:. X2 ::::} f(Xl) f:. f(X2),
and a surjection,
(Vy E R) (3x E R) y = f(x).
To that end, let us put

9 : Q \ {O} -7 Q \ {O}, g(x) = 1/x. (5.7)

and
h:I -7 I, h(x) = x. (5.8)
Note that f = 9 on the set Q \ {O} and f = h on the set I. It is an easy
exercise to check that both 9 and h are injective and surjective functions. The
ranges of 9 and h are disjoint, while the union of their ranges is the set

(Q\{O})UI=R\{O}.
Thus f is also a bijection on the last set.
Finally, since
(Vx E R) f(x) =0 ~ x =0
it follows that f is a bijection on the whole R.
b) We shall prove the continuity of f at the point x' = 1 by using Definition 5.3.
Let (xkhEN be an arbitrary sequence of positive real numbers converging to
x' = 1 j it is no restriction to assume Xk f:. 0 for every kEN. Our task is to
show
lim f(Xk) = 1 (= f(1)). (5.9)
k--+oo

Let € > be given. Let us put

Al = {k E NI Xk E Q \ {O}} and A2 = {k E NI Xk E R \ Q}.
Then exactly one of the following two possibilities is true.
CONTINUITY 203

(A) Both sets Al and A2 are infinite;

(B) one of the sets Al or A2 is finite, while the other one is infinite.
In the case (A), for j = 1,2, it holds

lim Xk = l. (5.10)
k--+oo, kEAj
Let us define the number i as the following infimum:

i := inf{lxkll k E Ad.
Clearly, either i = 0, or, as we claim, i is positive. Namely, if i = 0, then there
exists a subsequence (Xk,)IEN of the sequence (xkhEAl which converges to O.
(Observe that each Xk, is nonzero.) But the existence of the last subsequence
is in contradiction with (5.10). Hence i > O.

By the definition of f, it holds f(Xk) =~, k E AI, and using the inequality
Xk
i > 0, it follows that
IXk -11
If(xk) - 11
IXkl
(5.11)
< (inf{lxkll k E Adtl ·Ixk -11 = i-I ·Ixk -11.
For given £ > 0, we choose kl E N such that
(Vk E At) k> kl =} (Ixk - 11 < i· E).
This implies

(Vk E At) k> kl =} If(xk) - f(1)1 < i-I. (i· £) = c. (5.12)

Further on, since f(Xk) = Xk for every k E A 2 , for given £ > 0 there exists
k2 E N such that
(Vk E A2) k> k2 =} (If(xk) - f(I)1 = IXk - 11 < E). (5.13)

Thus from (5.12) and (5.13) it follows that the following implication holds.

(Vk E N) k > max{kl , kd =} If(xk) - f(I)1 < £, (5.14)

which means that f is continuous under the assumption (A).

The case (B) is somewhat simpler and is left to the reader as an exercise.
Also, the proof of the continuity of f at the point x" = -1 is completely
analogous to the previous proof, hence omitted.
Finally, we shall prove that f has a discontinuity at XIII = O. Clearly
lim f(x) = lim x = 0 = f(O).
x--+O, xEI x--+O, xEI
204 CHAPTER 5.

For a rational number x such that Ixl < 1 it holds

1
If(x) - f(O)1 = If(x)1 = ~ > 1.

Choose c := ~. Then for every b > 0 there exists a nonzero element Xs E Q

such that
IX51 < b =} If(xs) - f(O)1 > 1 > c.

Hence the limit lim f(x) is certainly not 0 (in fact, it does not exist), thus
x-+O, xEQ
in view of Theorem 5.5, f can not be continuous at O.

c) We shall prove that f is discontinuous in each point of the set R \ {-1, 0,1}.
Let first Xo E Q \ {-1, 0,1}. Then for every kEN there exists an irrational
number Xk, Xk i: Xo, in the interval (xo - 1 ,Xo
-k + -k1 ), thus lim Xk = Xo. But
k-+oo
then
lim f(Xk) = lim Xk = Xo
k-+oo k-+oo
i: ~
Xo
= f(xo),

meaning that f is discontinuous at Xo.

The other case, namely when Xo E I, is analogous and left to the reader.

5.2 Uniform continuity

5.2.1 Basic notions
Definition 5.29. A function f : A C R --+ R is uniformly continuous on the set
X C A if for every c > 0 there exists a number b > 0, depending only on c, such that
for every pair Xl> X2 E X with the property IXI - x21 < 8 it holds If( xd - f( X2) I < c.

Using logical symbols, the Definition 5.29 can be expressed as follows.

A function f :A C R --+ R is uniformly continuous on a set X C A iff

(Vc > 0) (3b > 0) (VXI,X2 E X) IXI - X2! < 8 =} If(xd - f(x2)1 < c. (5.15)

Clearly, a uniformly continuous function on a set X is also continuous at every point

of that set. As will be seen from Example 5.34 below, the opposite is not true in
general. However, the following statement holds.

Theorem 5.30. If a function f : A C R --+ R is continuous on a compact set

K C A, then it is also uniformly continuous on K.

The following two theorems give two important properties of a continuous func-
tion on a compact set, resp. on a closed bounded interval (which is thus also a
compact set).
CONTINUITY 205

Theorem 5.31. If a function f : A c R -t R is continuous on a compact set

K C A, then it attains its minimum and maximum on the set K.

In other words, there exist two numbers Xm and XM in K such that for every x E K
it holds
f(xm) ::::; f(x) ::::; f(XM).
Theorem 5.32. Assume that a function f : A c R - t R is continuous on an
interval [a, bJ c A, and assume} for instance} that f(a) < f(b). Then for every y
such that f(a) < y < f(b), there exists an x E (a, b) such that f(x) = y.

In other words, a continuous function maps an interval onto another interval.

5.2.2 Examples and exercises

Example 5.33. Prove the uniform continuity of the following functions.

a) f(x) = x 2 - 3x -1, x E [3,6J; b) f(x)=2x-l, xER.

Solutions.
a) Of course, the uniform continuity of f follows at once from Theorem 5.30, since f
is continuous on the compact set [3,6J. We shall, however, show that directly.
For every pair Xl, X2 E [3,6J it follows

If(XI)-f(X2)1 = l(xi-3xl-l)-(x~-3x2-1)1 = IXI-X21·lxI+X2-31::::; 9·l xI-X21·

So for given E > 0 we can choose b := E/9, in order to obtain the implication

(V'x!, X2 E [3,6]) IXI - x21 < b =} If(xd - f(X2)1 < E.

b) Since for every pair Xl, X2 E R it holds

If(XI) - f(X2)1 = 2· IXI - x21,

for given E > 0 we can choose b := E/2. Then the following implication holds

(V'Xl?X2 E R) IXI - x21 < b =} If(xd - f(X2)1 < E.

Notice that this function is uniformly continuous on the noncompact set R

(since R is not bounded).

1
Example 5.34. Prove that the function f(x) = -, x> 0, is
X

a) uniformly continuous on the set (c, 1), where 0 < c < 1;

b) not uniformly continuous} though continuous} on the set (0,1 J.

Solutions.
206 CHAPTER ,,).

a) Let 5 > 0 be given. For every pair XI, X2 from the interval (c, 1) it holds

If(Xl) - f(X2)1 = IXI - x21 < IXI - x21

XIX2 - c-

This allows us to choose I) := C2 5, since then it holds

(VXl' X2 E (c,l)) (IXI - x21 < I) =} If(XI) - f(X2)1 < 5).

Question: Why could we not simply apply Theorem 5.30 ?

b) In order to prove that f is not uniformly continuous on the open interval (0, 1),
1 1
we shall use the sequences Xn = -, n E N, and x~ = - - , n E N. Obviously
n n+1
both sequences converge to 0, i.e.,

lim Xn
n---+(X)
= n--+oo
lim x~ = O.

However, it holds

If(xn) - f(x~)1 = If (~) - f (n ~ 1) I = In - (n + 1)1 = 1.

If we put 5 := ~, then for every 0 < I) < 1 there exists an n E N (depending
on I)) such that

IXn - Xn'1_
- n( n
1+ 1,< 8.

But then we have

If(Xn - f(x~)1 = 1> ~= 5.

Finally, let us prove the continuity of the function f at every point Xo E (0, 1].
For given 5 > 0 let us choose

I) := min { 2' 2x5

xo 5} . (5.16)

Then the following implication holds:

If(x) - f(xo)1 = Ix - xol < 21x - xol < 5

X· Xo x5
provided that Ix-xol < I), proving the continuity of f at every point Xo E (0,1].

Remark. It is important to understand that the number I) in (5.16) was depending

not only on 5, but also on the observed point Xo. In fact, in this case it was impossible
to choose I) uniformly in Xo.
CONTINUITY 207

Example 5.35. Discuss the uniform continuity of the following functions on their
domains.
a) f(x)=lnx, xE(O,l); b) f(x)=x·sinx, xE[O,+oo);

c) f (x) ={ . ;;-,
x . sm 1 x E (0, 1r) ;
d) f(x)
1
= eX . cos -, x E (0,1).
0, x = 0; x

Solutions.
a) We shall prove that the function f(x) = Inx, x E (0,1), is continuous at every
point Xo E (0,1), but is not uniformly continuous on the open interval (0,1).
Fix Xo E (0,1) and let c E (0,1) be given. The inequality

If(x) - f(xo)1 = lIn x -lnxol < c

is equivalent to
-E:
-c < I n - < c
X
~ xoe < x < xoeE: .
Xo

Hence we can choose 0 := min{xo - xoe-E:,xoeE: - xo}, so that the following

implication holds:

(Vx E (0,1» Ix - xol < o:::} If(x) - f(xo)1 < c.

In the same manner one can prove the continuity of the logarithmic function
on the whole interval (0, +00).
In order to prove that f is not uniformly continuous on (0,1), we shall start
from two sequences converging to zero, (Xn)nEN and (X~)nEN, where
1 x 'n ..- 1
Xn := en and - e n +1 ' for n E N.

Put c := 1/2. Then for every 0 > °there exists an n = n5 E N such that
e-1
IX n - x~1 = en+! < o.
For such n it holds, however,

If(x n ) - f(x~)1 = I( -n) - (-(n + 1»1 = 1 > c,

contradicting the uniform continuity of f on (0,1).

b) We shall prove that the given function f is not uniformly continuous on the
interval [0, +00).

Let Xn = n1r and x~ = n1r + ~.

n
Then for given 0> 0, there exists n = n5 such
that
IX n - x~1 = ~n < o.
208 CHAPTER 5.

Put c = 1; then it holds for n > 2 :

If(xn) - f(x~)1 = ln7r sin(n7r) - (n7r + ~) sin (n7r + ~) I

. 1 ( n7r
1(-l)nsm-1
n
+ -n1) (2
> -. -1) (n7r + -)
7r n
1 > 2 > c.
n
We used the inequality
. 2
sIn X > -x,
7r °< X < "2.
7r

Hence f is not uniformly continuous on [0, +00). Notice, however, that f is a

product of two uniformly continuous functions on [0, +00) (prove that).

c) Let us observe first that If(x)1 ::; Ixi for all x E [0,7r).
°
Let c > be given; we shall assume that °< 6 < 7r (this is no restriction). If
xI, X2 E [0, ~], then it holds for 151 := 6/2 :
2
IXI - x21 < 151 '* If(xd - f(X2)1 < Xl + X2 < 6, (5.17)
6
If Xl, X2 E [2,7r], then

If(xd - f(X2)1 = IXI sin ~

Xl
- X2 sin ~I
X2

IXl sIn. -Xl1 - X2 sm -

X2
. 1 . 1
+ X2 sm -
Xl
.
- X2 sm -
Xl
1 I

< IXI - x211sin ~I + X212 sin Xl - X2 cos _X_I_+_X_21

Xl 2XIX2 2XIX2

- x21
< IXI - x21 + 2X2 IXI2XIX2 ::; IXI - x21
1
+ -IXI
Xl
- x21·

Thus If(xd - f(X2)1 < IXI - x21 (1 + D.This shows that choosing 152 :=
2
_6_ the following implication holds
2+6
(VXI' X2 E [c,7r)) IXI - x21 < 152 '* If(xd - f(X2)1 < 6. (5.18)
Putting relations (5.17) and (5.18) together, we obtain that for given 6 > °
we can choose 15 as
2
6 6 }
15 := min {
2' 2 + c '

in order to get the implication

(VXI' X2 E [0,7r)) IXI - x21 < 15 '* If(xd - f(X2)1 < 6.

Thus we proved the uniform continuity of f on (0, 7r).

CONTINUITY 209

°
d) Since 1- (0,1), the given function is continuous in each point of its domain. We
shall prove next that it is not uniformly continuous on (0,1).
1 1 .
Let Xn = - - and x~ = (
2mf 2n + 1) 71" for n E N, hence It holds
lim
n~oo
Xn = n--+oo
lim x~ = 0.

Now, for given 8 > °there exists n = n8 such that

IX n - x~1 = 1 < 8.

However, for every n it holds

If(x n ) - f(x~) I /ex p C~71") cos 2n71" - exp C2n ! 1 )71" ) cos(2n + 1)71"/

exp (2~71") + exp (2n ~ 1)71") > 2,

contradicting thus the uniform continuity of f on (0,1).

Example 5.36. Prove that the function f(x)

Isinxl
= --,
x ~ 0, is uniformly con-
x
tinuous on each of the intervals (-1,0) and (0,1), but is not uniformly continuous
on their union (-l,O)U(O,l).

Solution. The function f equals on the interval (0,1) to the function

1, x = 0;

Fl(X) =<
Sinx
x
°< x < 1;
sin 1, x = 1.
Clearly, FI is continuous on the closed interval [0, 1], hence it is uniformly continuous
there. This also gives the uniform continuity of f on the interval (0,1). Analogously,
the function F2 given by

- sin 1, x = -1;
SlllX
F2(X) =< x
-1 < x < 0;

-1, x = 0.
is uniformly continuous on the closed interval [-1,0]' and equals to f on (-1, 0).
Hence f is uniformly continuous on (-1,0).
210 CHAPTER 5.

However, we shall prove next that f is not uniformly continuous on the union

A:= (-1,0) U(O, 1).

1 - 1
Let Xn = --- and x~ = - - for n E N. These two sequences are from A and
n+l n+l
both tend to zero as n --+ 00. For given 8> 0, there exits n = no such that

IXn - Xn'1 -__2 1 < 8.

n+

But for every n E N it holds

If(xn) - f(x~)1
Isin (~)I_lsin (~)I
1 1
n+l n+l

2(n+l) I
1
. lll-- 1
S
n+l
~2(n+l). (2- .n+l
7r
-1)
- 4 =-.
7r

Hence for (say) E = 1, it holds If(xn) - f(x~)1 > E, proving that f is not uniformly
continuous on the set (-I,O)U(O,I).

Example 5.37. If the function f : I --+ R is bounded, monotone and continuous

on the open interval I, then f is also uniformly continuous on I if

a) I=(a,b); b) I = (a, +(0).

Solution.

a) Let us prove first that a bounded, monotone and continuous function on the open
interval ( a, b) has a right-hand limit at a. Assume, say, that f is monotonically
increasing on (a, b). Since f is bounded on (a, b), it follows that the range of
the function f,

B := f ((a, b» = {y E RI (::Ix E (a, b)) y = f(x)}

is bounded from below. By Example 1.50 a), it follows that the set B has an
infimum, which we shall denote by L. Let us show that

lim f(x) = L. (5.19)

x~a+

Let (Xn)nEN be an arbitrary sequence of numbers from (a, b) such that

lim Xn a.
n->oo

By passing (if necessary) to a subsequence, we can suppose that (Xn)nEN

is a monotonically decreasing sequence. By assumption, then the sequence
CONTINUITY 211

(f(Xn))nEN is also a monotonically decreasing sequence. Since f(xn) ~ L for

every n E N, there exists the limit

L':= n-+oo
lim f(xn). (5.20)

Clearly, either L' > L or, as we claim, L' = L.

If L' > L, then there exists an element Yo E B such that L' > Yo > L (because
L is the infimum of B). Since f is continuous on the interval (a, b), we can
apply Theorem 5.32, giving us the existence of a unique number Xo E (a, b)
such that f(xo) = Yo. From n-+oo
lim Xn = a it follows that there exists an no such
that it holds
n > no =? a < Xn < Xo.
This implies
n > no =? f(xn) ::::: f(xo) = Yo·
Since Yo < L', it follows that for infinitely many indices n it holds f(xn) < L',
which in view of the decrease of f(xn) is contradicting the definition of L' in
(5.20). Hence necessarily L = L', implying L = n-+oo
lim f(xn). Since (Xn)nEN was
an arbitrary sequence tending to a, it follows that (5.19) holds.
In an analogous way one can prove that the left-hand limit of f at the point
b exists as well; let us put
K:= lim f(x). (5.21)
x-+-b-

Let us introduce next the following function:

L, x = a;
F(x) := { f(x), a < x < b; (5.22)
K, x =b.

By the choice of Land K, the function F is continuous on the compact set

[a, b], hence by Theorem 5.30 it is uniformly continuous on [a, b]. But then
trivially follows that f is uniformly continuous on (a, b).

b) In the same way as in a), one proves that there exists the

lim f(x)
x-+a+
= L.

Let us assume, say, that f is monotonically increasing. Then from the bound-
edness of f it follows that the set

B:= f((a,+oo)) = {y E RI (:Ix E (a, +(0)) y = f(x)}

is bounded from above. Hence by Example 1.50 a), it follows that B has a
supremum, which we shall denote by K. Similarly as was done in a), one can
prove that
lim f(x) = K.
x-+oo
212 CHAPTER 5.

In order to prove the uniform continuity of f on (a, +00 ), let us put

L x = a;
F(x):= { f(x), x> 0.

Let 6 > °be given. Then there exists a number T > a, depending on 6, such
that
6
Vx >T =? IF(x) - KI < 4. (5.23)
In view of a), F is uniformly continuous on the set [a, TJ, hence f is uniformly
continuous on (a, T]. This means that there exists a 5 > such that °
(VXI,X2 E (a,T]) (lxI - x21 < {j =? If(XI) - f(X2)1 < ~). (5.24)

We have still to prove that f is uniformly continuous on the set [T, +00) as
well. To that end, from (5.23) it follows that for Xl > T and X2 > T it holds
If(xt} - f(X2)1 < If(XI) - K +K - f(X2)1
6 6 6 (5.25)
::; If(xt} - KI + IK - f(X2)1 ::; 4 + 4 = 2·
Hence IXI - x21 < {j =? If(XI) - f(X2)1 < 6/2, and from relations (5.24) and
(5.25) it follows that f is uniformly continuous on the set [T, +00).
Remark .
• The function F from (5.22), continuous on [a,b] and equal to f on (a,b), is
called the continuous extension of f from ( a, b) to [a, b] .
• The assumption on monotonicity of f in Example 5.37 is essential. Namely,
the function
g(x) = sin(l/x), x E (0,1)
is bounded and continuous on (0,1), but can not be continuously extended to
[0,1] (see Example 5.15 c)).
Exercise 5.38. Show that the function f is uniformly continuous on the set X
given below.
a) f(x) = sinx, X = R; b) f(x) = cosx, X = R;

c) f(x)=x 2, X=(-2,2). d) f(x) =..jX, X = [1,+00).

Exercise 5.39 Examine the uniform continuity of the following functions on their
domains.
slnx
a) f(x) = - , x E (0,7r); b) f(x) = x sinx, x E [0, +00);
x
x 1
c) f(x)=9_x 2' xE(-2,2); d) f(x) = ..jX' x> 0.

Answers. In a), b) and c), the functions are uniformly continuous on their do-
mains, while in d), f is not uniformly continuous, but only continuous.
Chapter 6

Derivatives

6.1 Introduction
6.1.1 Basic notions
Definition 6.1. Let f be a real function defined on an open interval (a, b) and let
Xo E (a, b). Then the following limit

f'(xo) = lim f(xo + h) - f(xo) (6.1 )

h ....O h
(provided it exists) is called the first derivative of f at the point Xo.

Besides the notation f'(xo) for the first derivative of the function f at Xo one also
uses the notation ix(xo).
Occasionally, instead of f'(xo) we shall write f~(xo) to emphasize the variable x
in which the derivative is found.
The number h in (6.1) is called the increment of the independent variable
x at the point Xo, while the difference f(xo + h) - f(xo) is called the increment of
the dependent variable at the point Xo. Thus relation (6.1) can be interpreted
as the limit of the quotient of the increments of the dependent and the independent
variable, when the latter tends to zero.
One can also define the one sided derivative of a function at some point Xo
from its domain as the one sided limit of the quotient f(xo + h~ - f(x o). The right-
hand side and the left-hand side derivative of f at Xo are denoted by f~(xo), and
f~ (xo). In the next table let us give the first derivatives of the most commonly used
elementary functions, together with the largest subsets of R on which they exist.

Table of first derivatives

1. (xn)' nx n - 1, nEZ, x E R.

2. (x")' O:X,,-l, 0: =f. 0, x > o.

213
214 CHAPTER 6.

3. (sinx)' cosx, x E R.

4. (cosx), = -sinx, x E R.

+ l)il k E Z}.
1
5. (tan x)' , x E R \ {(2k
cos 2 X
-1
6. (cot x)' = -'-2-'
SIn x
x E R \ {hi k E Z}.

7. (ax)' = aX. Ina, a> 0, x E R, (eX)' = eX, xER.

1
8. (loga x)' = -~-,
X· na
a> 0, ai-I, x> 0, (lnx)'
x
,x> O.

1
9. (arcsin x)' = .Jf=X2' Ixl < l.

-1
10. (arccos x)' = .Jf=X2' Ixl < 1.

1
11. (arctan x)' = 1 + x2 ' x E R.

-1
12. (arccot x)' x E R.
1 + x2 '

If the functions f and 9 are defined on an interval (a, b) and have first derivatives
at the point x E (a, b), then

• the first derivative of the sum, resp. difference of functions f and 9 is

(J(x) ± g(x))' = f'(x) ± g'(x);

• the first derivative of the product of functions f and 9 is

(J(x) . g(x))' = f'(x)g(x) + f(x)g'(x),

and, in particular,
(A· f(x))' = A· f'(x),
where A is an arbitrary real constant;

• the first derivative of the quotient of functions f and 9 is

f'(x)g(x) - f(x)g'(x)
( f(X))' provided that g(x) i- O.
g(x) g2(X)
DERIVATNES 215

Assume that the function 9 : (a, b) -+ (c, d) has a first derivative at Xo E (a, b)
and assume that the function f : (c, d) -+ R has a first derivative at the point
g( xo) E (c, d). Then the first derivative at Xo of the composite function
h = fog, h: ( a, b) -+ R,
exists and the so called chain rule holds:

h'(xo) = f; (g(xo)) . g'(xo). (6.2)

If the surjective function f : (a, b) -+ (c, d) satisfies the following three conditions
(i) the function f has a first derivative at the point Xo E (a, b);

(ii) the function f is monotone on the interval (a, b);

(iii) the number f'(xo) is different from zero,
then there exists an inverse function
r 1 : (c, d) -+ (a, b)
to the function f.
Then the first derivative of the inverse function at the point Yo = f(xo) is

(f-l)'(yO) = f'(~o) or (f-l)'(yO) = _ _1 (6.3)

Assume that the equation

F(x,y)=O (6.4)
defines uniquely a function
y = f(x), x E (a, b).
Then we say that f is given implicitly with equation (6.4).
Assume that for some Xo E (a, b), the derivatives of F from (6.4) exist and,
moreover, F~(xo, Yo) #- 0, with Yo = f(xo). Then from the chain rule (6.2) it follows
that the first derivative of the implicitly given function f at the point Xo is

f'(xo) = _ F~(xo, Yo) (6.5)

F~(xo, Yo)

Assume that two differentiable functions y = y(t) and x = x(t) of the same vari-
able ("parameter") t E (a, (3) are given. Then the function y = f(x), determined as
the mapping f that assigns to x = x(t) the number y = y(t) is called a parametric
function.
The first derivative of the parametric function y = f( x) at the point Xo is
given by
y'(xo) = y~(to) to E (a, (3) (6.6)
xWo) ,
where x(to) = Xo. Clearly, formula (6.6) has meaning only at those points to, where
x and y have derivatives and it holds x~(to) #- o.
216 CHAPTER 6.

Definition 6.2. If the increment b.y of the function f : (a, b) -+ R at the point
Xo E ( a, b) can be written in the form
b.y = f(xo + h) - f(xo) =D .h + r(h) . h,
for some number D (independent from h), and it holds
lim r(h) = 0,
h--+O

then for the function f we say that it is differentiable at the point Xo.

Theorem 6.3. A function f : (a, b) -+ R is differentiable at the point Xo E (a, b)

if and only if it has a first derivative at that point. 1
Hence from Definition 6.2 it follows D = f'(xo).
The differential of a differentiable function f at a point x is the linear function
df(x) : R -+ R of the increment h,
df(x)(h) = D· h, hER,
or, in view of Theorem 6.3, and by omitting h,
df(x) = f'(x) dx. (6.7)
Using Definition 6.2 and Theorem 6.3, we have the approximate formula
f(xo + h) ~ f(xo) + f'(xo) . h, (6.8)
provided that f is differentiable at Xo and h is "small".
Assume that a function f has a first derivative at every point x E ( a, b). Then the
function f' : (a, b) -+ R, is called the first derivative of f. If the first derivative of
the function f' at the point Xo E (a, b) exists, then it is called the second derivative
of the function f at the point Xo and will be denoted by f"(xo).
One defines analogously the third, fourth, ... , n-th derivative of a function f at
the point Xo, and they are denoted respectively by f"'(xo), f(4)(XO), ... , f(n)(xo).
If a function f : (a, b) -+ R has a first derivative at the point Xo E (a, b), then
the line
y - Yo = f'(xo)(x - xo), (6.9)
where Yo = f(xo), is called the tangent line of the graph of the function f at the
point T (xo, f(xo)). If, additionally, it holds f'(xo) -:F 0, the line
1
y - Yo = - - - ( x - xo) (610)
f'(xo) .
is the perpendicular line ofthe graph ofthefunction f at the point T (xo, f(xo)).

Assume a function f has a first derivative at a point Xo. If 0 ::; 0: < 7r is the angle
between the tangent line at the point Xo and the positive direction of the x-axis,
then it holds
tan 0: = f' (xo). (6.11 )
In other words, the slope of the tangent line of the graph f at some point is exactly
the value of the first derivative of f at that point.
lThis theorem is not true for functions of two or more variables.
DERIVATIVES 217

6.1.2 Examples and exercises

Example 6.4. Find by definition the first derivatives of the following functions at
the given points.

a) f(x) = x(x -1)2(X - 2)3, X E R, at the points Xo = 0, Xl = 1, X2 = 2;

b) f(x) = V1+X, X 2': -1, at the points Xo = 1, Xl = 0, and also the right-hand
side derivative at the point X2 = -1;

c) f(x)=x+(x-1)arcsinj X , x>O, at the pointxo = 1;

x+l

d) f(x) = \Ix + 1, x E R, at the point Xo = -1.

Solutions.

a) Using Definition (6.1), we have

/,(0) = lim h(h - 1)2(h - 2)3 -

h .....O h
°= -8,
/,(1) = lim (1 + h)(1 + h - 1)2(1 + h - 2)3
h ..... O h
°= 0,

/,(2) = lim (2 + h)(2 + h - 1)2(2 + h

h ..... O h
2)3 °= 0.
b) We have

VI + I + h - y'2 r I _ _1_
/'(1) = l~ h = h~ v'2+7i" + y'2 - 2y'2'

v'f+h - v1 1
1'(0) = hlim
..... O
h = -2'

f +' (-1) = hlim

F+ (-1+ h)h - Jl + (-1) = lim _1
h ..... O+ Vh
= +00.
..... O+

Geometrically, the last equality means that the vertical line x -1 is the
tangent line to the graph of f at the point (-1,0).

c) It holds

1 + h + (1 +h - 1) arcsin fl+h - 1
/,(1) = lim
h ..... O h
VI+h+l = 1 + 7r4·
218 CHAPTER 6.

d) In this case we have

1'( -1) = lim ~-1

h_O
+ hh + 1 - 0
= +00.
Exercise 6.5. Find by definition the first derivatives of the following functions.

a) f(x) = -;, at the point Xo = 1;

x

b) f(x) = \1(1 + x)2, at the points Xo = 0, Xl = -1;

c) f(x) = 3·12 + xl, at the point Xo = -3.

Answers.

a) 1'(1) = -3. b) 1'(0) ==~; 1'(-1) does not exist.

c) 1'(-3) = -3.
Example 6.6. Find by definition the first derivatives of the following functions.
a) f(x)=x 2 +2x, xER; b) f(x) = <Ix - 1, x E R;

c) f(x) = In(x + 1), x> -1; d) f(x) = cos(2x), x E R,

at the point x from its domain.
Solutions.
f '() 1· (x+h)2+2(x+h)-X2_2x
a) x = h-+O
1m h = 2x + 2.
b) For x i= 1, it holds
f'(x) = lim <Ix
h-+O
+h - 1-
h
rx=-r
. 1
hm - - - ; = = = = = - - - - - ; = = = = = = - r = = = =
h-+O \j(x -1 + h)2 + \j(x -1)(x -1 + h) + \j(x -1)2

1
= 30x- 1)2

At the point x = 1 it holds

,
f+(I) = hm
. \Ih-<IO
1 = +00,
h-+O+

f~(I) = lim
h-+O-
\Ih -h <10 = +00,
which means that the tangent line of the graph of f at x = 1 is parallel with
the y-axis.
DERIVATNES 219

c) It holds for x > -1

llh
f '( x ) = 1·1m In(x+h+l)-ln(x+l) = 1·1m In ( 1 + -h- )
h--+O h h--+O x +1

1 (
--lnlim 1 + - -
h) (x+1)/h I 1
x +1 h--+O X +1 x+l

d) We have

f'(x) = lim cos(2(x + h)) - cos(2x) = lim -2sin(2x + h)· sinh = -2sin(2x).
h--+O h h--+O h

Exercise 6.7. Find by definition the first derivatives of the following functions. at
the point x E R.

a) f(x) = xcos2x; b) f(x) = \1(1 + x)2; c) f(x) = 32+x.

Example 6.8. Find the first derivatives of the following functions.

a) f(x)={ll+x4, xER;

b) f(x) = F+ ,/1 + ~1 + x4, x E R;

7r
c) f(x) = 32tan3x , X # (2k + 1)2' k E Z;

d) f(x) = lnsinarccot eX, x E R;

2e2x
e) f(x)=ln\4J , xE(O,7r/2);
1 + cos x

f) f(x) = log3logslog7 x, x> 7.

Solutions. We leave to the reader to find the natural domains of the given functions
and the largest subsets of R on which their first derivatives exist.

a) Let us put u(x) = 1 + X4. Then from the chain rule it follows that

f'(x)
1
= -3 (u(X)t 2/3u'(X) = \I 4x
3 (1 + x )2
3
' 3

4
x E R.

3
b) f'(x) = x . 1 1
6/1 + \/1 + ~1 + x4 \/(1 + ~1 + x4)2 . (V/f+X4)3' x E R.
220 CHAPTER 6.

3 3 tan 2 X 7r
C) f'(x)=32tanxln3·(2tan3x)'=61n3·32tanx_-, x-=l-(2k+1)-, kEZ.
COS 2 X 2

d) In this case we have

1
f'(x) = sm arccot eX (sin arccot eX)'

1
sm arccot eX cos arccot eX. (arccot eX)'

_eX e2x
cot(arccot (e X ))--2- = ---2-' X E R.
l+e x l+e x

e) Since it holds

1 2x 1 2 X X 1 X
f(x) = -In(e ) - -lncos - = - - -lncos-
4 4 2 2 2 2'

we obtain
f'(x) = ~ + ~tan~, X E (0,7r/2).

. Inx .
f) Smce loga X = -1-, for X > 7, a > 0 and a -=I- 1 (prove that!), It follows that
na

1
f'(x) n 31og 5 1og 7 x (log5 log7 x)'

1 1 1 1 1
-----_.---.-.
log5 log7 X log7 X x In 31n 5 In 7 x In 31n x In log7 x .

Exercise 6.9. Find the first derivatives of the following functions.

a) f(x) = sinhx, x E R; b) f(x) = cosh x, x E R;

c) f(x) = tanh x, x E R; d) f(x) = cothx, x -=I- 0;

e) f(x) = In(x + vfx2+l), x E R; f) f(x) = In(x + JX2="T), x E R;

Answers.
1
a) f'(x) = coshx. b) f'(x) = sinhx. c) f'(x) = cosh2 X

-1 1
d) f'(x) = sinh2 x e) f'(x) = 'x2 +1 f) f'(x) =~
DERIVATIVES 221

Exercise 6.10. Find the first derivatives of the following functions.

a) f(x) = (Vx - ~r, x # 0;

b) f(x) = JX2+"1- x + yIX"2TI + x, x E R;

JX2+"1 + x JX2+"1 - x
c) f( x) = cos cos cos cos 2x, x ER;

1 2x
d) f(x) = -arctan--2 , x E (-1,1);
2 1- x
x ex
e) f(x)=ex+ee +ee , x E R;

f) f(x) = In(ln 2 (ln 3 (x 2 »), x # 0;

1 . a sin x + b ( 7l' 7l' )
g) f(x) = ~arcsm b' , Ibl < a; x E - - - .
a 2 - b2 a + sm x 2' 2 '

h) f(x) = 1 In b+ asinx - ~. cos x

Vb2 _ a2 a + bsinx ' lal < Ibl·

Answers. a)f'(x)=~(~- ~r(~+ ~). b) f'(x) = 8x.

c) f'{ x) = 2 sin cos cos cos 2x . sin cos cos 2x . sin cos 2x . sin 2x.

d) f'(x) = _1_2 , e) f'(x) = eX +exeex +exeexeeex •

l+x
1
f) f'(x) = 12 .
g) f'(x) = a + bsmx
Ixl ·In x 2 .In(ln3 (x 2 ))
h) f'(x) = 1. (compare with g».
a + bsmx
Example 6.11. Find the first derivatives of the following functions.
a) f(x) = xx, x> 0; b) f(x) = x Vx , x> 0;

c) f(x) = (Jxy, x> 0; d) f(x) = xsinx, x> 0;

e) f(x) = (sinx)COSX, sin x > 0; f) f(x) = (2 + sinxy, x E R.

30lutions.
'l) The function f can be written as f(x) = exlnx , x> 0, whose first derivative is
f'(x) = exlnX(lnx + 1) = xX(lnx + 1).
222 CHAPTER 6.

This first derivative can be found also by taking the logarithm of the function
y = XX and then differentiating the obtained expression.
In fact, from In y = x In x it follows
y'
- = In x
y
+ 1 =} y' = y(1n x + 1).
Putting now y = f (x) = XX in the last equality gives again

y' = 1'(x) = xX(1nx + 1).

b) From the expression
f(x) = ev'X lnx , x> 0,

vx)
it follows
1'(x) =ev'X lnx (lnx + =xv'X.lnx+2
2VX x 2VX .

c) Since it holds f(x) = e(xlnx)/2, it follows that

1'(x) = e(xlnx)/2 C \lnx).

d) The first derivative of f is

l' (x) = esm

.
XIn X ( cos x In x smx)
+ -x- .

e) From f(x) = ecosxlnsinx it follows that

f '()
X = ecosxlnsinx ( - Sln
. x 1n sm
. x + -cos.-2 x) .
slnx

f) From f(x) = exln(Hsinx) we obtain

1'(x) = exln(2+sinx) (In(2 + sin x) + x co~ x ) .

2+S111X

Example 6.12. Prove that a differentiable function f (a,b) ----7 R at the point
Xo E (a, b) is also continuous at that point.

Solution. By Definition 6.2 the increment of f at Xo can be written in the form

f(xo + h) - f(xo) = D· h + r(h) . h,

where D does not depend on h, while the remainder r( h) satisfies the condition
lim r(h) = O. Hence r(h) is bounded by for example IDI + 1 for Ihl sufficiently small.
h.... O
Thus there exists 151 > 0 such that the following implication holds

(V hER) Ihl < 151 =} Ir(h)1 < IDI + 1.

DERNATNES 223

But then for given c > 0there exists 0 := ~ min {2ID~ + 1,01} such that for Ihl < 0
it holds
c
If(xo + h) - f(xo)1 ::; (IDI + Ir(h)l)lhl < (21DI + 1)21DI + 1 = c.
Example 6.13. The function f is defined by f(x) = lxi, x E R. Check whether

a) f is continuous at the point x = OJ

b) f has a first derivative at the point x = OJ

c) f is differentiable at the point x = O.

Solutions.

a) Let us prove that f is continuous at O. Namely, for given c > 0 it holds

If(x) - f(O)1 = Ilxl-IOII = Ix - 01 < c,

provided that Ix - 01 < 0 := c.
b) The right and left derivative of f at 0 are

. -Ihl-
h- 0 -_ ~
h = 1, f~(O) = lim Ihl- 0
I +, (0) = lIm
h-+O+ h-+O- h = h-h = -1.
Since these one sided derivatives are different, f has no first derivative at O.
Remark. This example shows that a continuous function at some point does not
necessarily have a first derivative at that point.

c) Once we have proved in b) that f has no first derivative at the point 0, it follows
at once from Theorem 6.3 that f is not differentiable at O. Still, we shall prove
that f is not differentiable at 0 without using that theorem.
To that end let us write the increment of f at the point x = 0 in the form
f(h) - f(O) = D· h + r(h)· h ~ Ihl = (D + r(h»h,
where h is the increment of the independent variable. We shall show that
lim r( h) does not exist, whatever value one chooses for the constant D. In
h-+O
fact, it holds
h>0 '*
r(h) = 1 - D
and
h<0 '* r(h) = -1 - D.
Since 1 - D "# -1 - D for every D, the limit at zero of the remainder r( h)
does not exist, giving us the non differentiability of f at O.
224 CHAPTER 6.

Example 6.14. Find the largest sets on which the first derivative of the function f
lS

(i) continuous;

(ii) differentiable;

(iii) continuously differentiable.

a) f (x) = x . Ix I, x E R; b) f(x) = In lxi, x i= 0;

c) f(x) = I(x + 2)2(X - 1)31, x E R; d) f(x) = I cos 3 xl, x E R;

. 1 1
e) f(X)={
x sIn-,
x
x i= 0; f) f(x) = {
2'
x sm;, x i= 0;
0, x = 0; 0, x = o·,
. 1
g) f(x) = {
x"" sIn-, x i= 0; were
h ex . a poSt't'we parame t er.
lS
x
0, x = 0,

Solutions.

a) (i) The function f is a product of two continuous functions on R, hence it is

continuous on the whole set R,

(ii) Since Ixl= X· sgnx, x E R, we can write x ·Ixl = x 2 . sgn x, x E R. The

last function clearly has first derivative in each x i= 0, and for such x it holds

J'(x) = 2x· sgnx = 21xl.

Let us show next that the first derivative of f at 0 also exists.

1'(0) = lim h2 • sgn h - 0

h ..... O h = O.

(iii) Moreover, the function f' : R ~ R is also continuous at 0, since

lim J'(x) = lim(2lxl) = 0 = 1'(0).

x--+O x--+O

b) (i) The function f is continuous on the set R \ {O}. Note that one can not discuss
the continuity of f at 0, since f is not defined there.

(ii) It holds f(x) = In(xsgnx) for x i= 0, hence

J'(x) =
X·
1
sgnx
sgnx =
x
x .!., i= o.
DERIVATNES 225

(iii) Clearly, the function l' defined by

1
1'(x) = -, xi- 0
x

is continuous on its domain. (In view of Example 5.34, it is not uniformly

continuous there.)

c) (i) The function is continuous on R.

(ii) Since f(x) = I(x + 2)2(X - 1)31 = (x + 2)2(X - 1)3sgn(x - 1), x E R, the
first derivative of f forx i- 1 is

1'(x) = 2(x + 2)(x - 1)3sgn(x -1) + (x + 2)23(x - 1)2sgn(x - 1)

(x + 2)(5x 2 - X - 4)lx - 11-

The function f has also a first derivative at the point x 1. Namely, its
left-hand side derivative at 1 is given by

f~(I) = lim 1((1 + h) + 2)2((1 + h) - 1)31_ 0 = lim (3 + h)21 h 13 = 0,

h-+O- h h-+o- h

and, similarly, the right-hand side derivative at the point 1 is also o. This
means that 1'(1) = o.

(iii) The function l' is continuous at the whole set R. In particular, it holds

lim1'(x) = lim(x
x--+I x--+l
+ 2)(5x 2 - X - 4)lx -11 = 0 = 1'(1).

d) (i) The continuity of f on R follows from the following statement.

If a function f : A C R ----+ R is continuous on its domain, then so is the

function If I given by Ifl(x) := If(x)l, x E A.

We suggest to the reader to prove this theorem, and also to find a coun-
terexample that will show that the opposite is not true in general.

(ii) As in c), we have

f'(x) = 3cos 2 x(-sinx)· sgn(cosx) = -~sin(2x)1 cos xl,

for every x E R.

(iii) The function l' is continuous on R.

e) (i) In Example 5.15 d) we showed that f is continuous on R.

226 CHAPTER 6.

(ii) and (iii) For x =I- 0 it holds

f '() . -1 -
x = SIn 1
- COS - .
1
x x x

However, since the limit

. 1
h sm - - 0 1
lim
h--+O
~, = h--+O
lim sin -h

does not exist, f has no first derivative at the point O.

(iii) The first derivative is continuous on the set R \ {O}.

f) (i) The function f is continuous on R.

(ii) It holds for x =I- 0

f '( x ) = 2x sm
. -1 1
- cos -,
x x
while for x = 0 we have

•
h 2 sm 1
- - 0 1
1'(0) = h--+O
lim hh
•
= h--+O
lim h sin -h = 0, ( 6.12)

which means that f' exists on the whole set R.

(iii) The function f' is continuous on R \ {O}. Since the limit

lim I' (x) = lim (2x sin ~ - cos ~)

",--+0 ",--+0 X X

does not exist, the function f' has a second order discontinuity at the point O.

Remark. The function f' obtained as a derivative of some function f, can never
have first order discontinuities, although, as we just saw, it might have second order
discontinuities.

g) Similarly as in e) and f), it follows that f is continuous for a > 0 on R, the

derivative f' exists for a > 1 on R, and, finally, f' is continuous on R for
a> 2.
DERIVATIVES 227

Example 6.15. Find the largest sets on which the first derivatives of the following
functions exist.
2- x x < 2;
a) f(x) = { (2 - x)(3 ~ x), 2 ~ x ~ 3;
-(3 - x), x> 3;

arctan x, Ixl ~ 1;
-
{ 7r Ixl- 1
b) f(x)- "4sgnx+ 2 ' Ixl > 1;

x3 + 1, x ~ 0;
c) f(x) = { e- 1jx + 1, x> 0;

d) f(x) = { x 2 /sin;/, x =1= 0;

0, x = 0;

X, x E Q
{
e) f(x) = 0, x E R \ Q.

Solutions.
a) From the limits
lim f(x)
x--+2-
= x--+2+
lim f(x) = 0 = f(2)
and
lim f(x)
x->3-
= x->3+
lim f(x) = 0 = f(3)
it follows that f is continuous at the points 2 and 3 respectively, thus it is
continuous on R. Let us see what happens with the first derivatives of f at
the points 2 and 3. We have

f~(2)= lim f(2+h)-f(2) = lim 2-(h+2)-0=_I;

h->O- h h->O- h

I' (2) = lim f (2 + h) - f (2) = lim (2 - (2 + h)) (3 - (2 + h)) - 0) = -1

+ h->O+ h h->O+ h '
implying that /,(2) exists and equals to -1. Similarly we prove that

f~(3) = f~(3) = 1,
hence 1'(3) = 1. Thus the function I' (the first derivative of f) exists on the
whole R and it holds
-I, x ~ 2;
f'(x) = { 2x - 5, 2 < x < 3;
1, x::::: 3.
228 CHAPTER 6.

b) We leave to the reader to check that f is continuous on R.

Clearly, the derivatives of f at the points -1 and 1 are to be examined. We
have

lim f( -1 + h) - f( -1) . arctan( -1 + h) + ~

f~( -1) h-.O+ h = h-.O+
hm 4

. 1 -1+h+1 1
hm -h arctan 1 - ( -1 + h)
h-.O+
-,
2

lim =-f(~-_l_+_h...:....) - f ( -1 )
f!..( -1)
h-.O- h

1r
-sgn(-l+h)+
I - 1 + hi - 1 1r
-(--) 1
lim 4 2 4 - __
h-.O- h 2'
hence f has no first derivative at -1. In a similar way we obtain

f~(l) = f~(l) = ~,
hence l' (1) exists and equals ~.
So we get
1
1 + x2 ' - l',
-1 < x <

1
1'(x) = { x < -1;
-2'
1
x>l.
2'
c) For later purposes, let us show first that

. e -I/x
11m --=0
x-.O+ xm
1
for every m = 0,1, ... Putting t:= - gives that t ----+ +00 when x ----+ 0+ .
x
Hence
-I/x
lim _e__ = lim tme- t = O. (6.13)
x-.O+ xm t-.+oo
(Note that the easiest way to prove the last equality is to apply m-times the
L'Hospital's rule, see Section 6.4.)
On the set R \ {O} the function f is continuous. From (6.13) for m = 0 it
follows that
lim f(x) = lim e- 1/x + 1 = 1,
x-.O+ x-.O+
DERIVATIVES 229

and since lim

x----+o-
f (x) 1 = f(O), it follows that f is continuous also at the
point x = O.
Let us find the right-hand side first derivative of f at zero

I' (0) = lim f(O + h) - f(O) = lim (e- 1/ h + 1) - 1 = lim e- 1/ h = 0

+ h--->O+ h h--->O+ h h--->O+ h

where we used (6.13) for m = 1. Since also

f~(O) = 0,

f has a first derivative in O. Thus we obtain

3X2, x ::; 0;
{
f'(x) = e- 1 / x
x> O.
X2 '

Notice that I' is a continuous function on R.

d) It is easy to show that f is continuous on R (do that). Its first derivative is

2x Isin ~I- 7r cos ~ sgn (sin~), x -I- 0, x -I- 11k, k E Z \ {OJ;

f'(x) = { x x x
0, x = O.
At the points x = 11k, k E Z \ {OJ, f has no first derivative.

e) This function is continuous at the point 0, since it holds If(x)1 ::; Ixl for every
x E R. In other real points f is not continuous (one can use the method from
Example 5.10 b)).
Clearly, the only point where f might be differentiable is the point O. In fact,
we shall prove that f is not differentiable at O. Let us examine the quotient

Q(h) := f(h) - f(O) _ f(h)

I - -
h
We have
1, hE Q;
Q(h)= { 0, hER\Q.
Hence
1'(0) = lim Q(h)
h--->O

does not exist. This means that the set on which f' exists is empty.

Exercise 6.16. Assume that the domain A of the function f is a symmetric neigh-
bourhood of zero. Then
230 CHAPTER 6.

• If f is an odd continuous function, then it holds f(O) = o.

• The first derivative of an odd (resp. even) function is even (resp. odd).

Example 6.17. Find the first derivative of the inverse function f- 1 for the given
function f, where

a) f(x) = 2x + 1, x E R; b) f(x) = JX + 2, x> 0;

c) f(x) = x 2 - 2x, x> -1; d) f(x) = cosx, x E (0,7r);

e) f(x) = sinh x, x E R; f) f(x) = cosh x, x> O.

Solutions.

a) • First method. Putting y = f(x) = 2x + 1, it follows x = y; 1. Thus

the function f- 1 (x) = x ; 1, x E R, is the inverse function for the given

function f. The first derivative of f- 1 is
1
U- 1 )'(x) = 2' x E R.

• Second method. Since f'(x) = 2 and f'U- 1 (x)) = 2 from the formula
(6.3) it follows

U- 1 )'(x) = __. 1
1
2' x E R.

b) In this case we shall use the Second method from a). To that end, we have

f'(x) = 2~' x> 0 and r 1 (x) = (x - 2?,

hence the first derivative of the inverse function f- 1 to the given function f is
U- 1 )'(x) = 2.) f-l(x) = 2(x - 2), x> 2.

c) From the equation x = f(y) = y2 - 2y, it follows f'(y) = 2y - 2. Thus the first
derivative of the inverse function f- 1 to f in variable y is
1 1
U- 1 )'(y) = 2y _ 2 = -,
Since (y - 1)2 = X + 1, we get finally

U- 1 )'(x) = ~ ~, x>-1.
DERIVATNES 231

d) From l' (x) = - sin x, x E (0, 7r), it follows

1 1
(f-l)'(X) (arccos x)' = ---.-.
sin( arccos x)

1 1
Jl - cos 2 ( arccos x) Vf=X2'
XE(-I,I).

e) From f'(x) = cosh x it follows for x E R

(f-l)' = (arcsinhx), 1 1
cosh(f-l(x)) cosh(arcsinh x)

1 1
VI + sinh (arcsinh x)
2 ~.

1
f) (arccosh x)' = .../x 2 _ l' x>1.

Example 6.18. Find the domains and the first derivatives of the inverse functions
of the following functions.

x2
a) f(x)=x+lnx, x>O; b) f(x) = --2' X
l+x
< o.
Solutions.

a) For x> 0 it holds f'(x) = y~ = x + 1 > 0, hence there exists a unique function
x
f- 1 whose domain B is the range of f, and is inverse to f. Clearly, B = R;
unfortunately, it is impossible to find the analytic formula for f- 1 explicitly.
Still, we shall find its first derivative. Namely by relation (6.3) it follows:

1 1 X
(f-l )' (y) = y~ = x + 1 = x + 1 '
x

where x = x(y) is the solution in x of the equation y = x + In x.

?x
b) Since for x < 0 it holds f'(x) = (1 ~ x 2)2' it follows that there exists a unique
inverse function f- 1 to f with domain (0,1) = f (( -00,0)). Its derivative is

(f-l)'(y) = (1 + X2)2 x3
2x 2y2 '

2
where x = x(y) in the solution in x of the equation y = ~.
1+x
232 CHAPTER 6.

Example 6.19. Prove that there is but one differentiable function on R such that

y3 + 2y = x, y(3) = 1, (6.14)

and find its derivative y~.

Solution. The existence of the function y = y( x) follows from the Implicit function
theorem, which is somewhat out of the scope of this book. However, in this case we
can prove its existence in the following way.
The function
F(X,y)=y3+2y-x

has all possible partial derivatives .

• We differentiate the equation F( x, y) = 0, with respect to y, giving us

x~ = 2 + 3y2 > O.
Notice that we observed x as a function of y. Now the function x = x(y) is
monotone for y E R. Hence the inverse function of x = x(y) exists and is also
differentiable on R. Then it holds (see (6.3))

I 1 __ 1 •
Yx = '7 - 2 + 3y2
y

Let us assume that there are two solutions of equation (6.14) and let us denote
them by Yl (x) and Yz (x). Then it holds

yi + 2Yl = x and y~ + 2yz = x.

This implies

yi - y~ + 2(Yl - Yz) = 0 ::::} (Yl - yz)(y; + YlYZ + y~ + 2) = o.

Since yr + YIYZ + y~ + 2 > 0 for any Yl and Yz, it follows that Yl(X) = yz(x)
for all x E R.

• The other method of finding y~ was explained in the introduction.

Let us differentiate in x the equation (6.14), assuming that y = y(x). Then we
obtain
3yZ . y~ + 2 . y~ = 1.

Solving by y~, we again obtain

1
y~ = 2 + 3y z'
DERIVATIVES 233

Example 6.20. Find the first derivatives of the following functions y = y( x) given
implicitly.
a) x 2 + y2 = 4; b) 2x-3y+3=x 2 +2y-6x;

c) y'x+V'Y=5x; d) X4 + 4x 2y2 - 3xy 3 + 3x = 0;

e) (y2 _ 9)3 = (2 X 3 + 3x - 1)2; f) (2+xy)2=3x 2 -7.

Solutions.
a) Observe that (y2)~ = 2yy', hence differentiating the given equation by x it
follows:
2x + 2yy' = 0 :::} y' = -x/yo
b) From the equation 2 - 3y' = 2x + 2y' - 6 it follows y' = -2x + 8

1 y'
c) It holds that r; + -- = 5, hence
2yx 2V'Y

y' = 10V'Y - ~.
d) Firstly we have
4x 3 + 8xy2 + 8x 2yy' - 3y 3 - 9xy2y' +3 = 0,
which implies
, -4x3 - 8xy2 + 3 y 3 - 3
y =
8x 2y - 9xy2

e) In this case we have

6(y2 _ 9)2yy' = 2(2 x 3 + 3x - 1)(6x2 + 3),
which implies
, (2 X 3 + 3x - 1)(2x2 + 1)
y =
y(y2 - 9)2

f) Since 2(2 + xy)(y + xy') = 6x, it follows

, 3x - 2y - xy2
Y =
x(2 + xy)
Example 6.21. Find the first derivatives of the following parametric functions.
a) x = t 2 + 2t, y = 2t 3 - 6t, t E R;

b) x = 2(t - sint), y = 2(1- cost), t E R;

c) x=2cos 3 t, y=sin 3 t, tE (o,i).

234 CHAPTER 6.

Solutions.

a) Since it holds x~ = 2t + 2, y~ = 6t 2 - 6, it follows from relation (6.6)

6t 2 - 6 3t 2 - 3 t i- -1.
y~ = 2t +2 - t +1 '

b) From the relations x~ = 2(1 - cos t), y~ = 2 sin t, t E R, it follows

y' _ sin t
- (1 _ cos t)' t E R \ {2k1r1 k E Z}.

c) We have x~ = 6 cos 2 t( - sin t), y~ = 3 sin 2 t cos t, thus

. 2
3 sm
y' = t cos t -1
-6 cos 2 t sin t - 2cot t' t E (0, 7r /2).

Example 6.22. Find the differentials of the following functions given below.

a) f(x) = 2JCOS ~ + In(x + 1); 2

b) f(x) = 5arctg(2x + 7)e3x +1 + 12x;

sinx + tgx .
c) f (x) = x3 + 3X + 3x '
5 + 4x 2 X

d) y = arcsin(5x 4 + 2) + exp(sin x + cos x)

Solutions.

a) Since the first derivative of f is

. 1

R+
sm-
x 2x
1'(x) =
x 2 cos - 1+x 2 '
X

it follows (see relation (6.7)) that the differential of f at the point x from the
domain of f is

df (x) = 1'( x) dx = ( R1 )
x2
sin -
1
cos-
X
2x
+ 1 +x2
d
X.

X
DERIVATNES 235

b) Using relation (6.7), it follows

df(x) = (5. 1 + (2!+7)2e3"'+1 + 15· arctg(2x + 7)e3"'+1 + 12'" In 12) dx.

c) It holds

df(x) = (cosx + 1/ cos 2 x)(x 3 + 3'" + 3x) - (sinx + tgx)(3x 2 + 3"'ln3 + 3) dx.
(x3 + 3'" + 3x)2

d) The differential of f at the point x from the domain of f is

df (x) = ( 20x 3 + (5'" In 5 + 8x) - (.5'" + 4x 2)(cos x - sin x») dx.

)1- (5x 4 + 2)2 exp(smx + cos x)

Example 6.23. If f(x) = \ju 2 + v 2, where mEN, m > 1, while u = u(x) and
v = v(x) are differentiable functions of the variable x, find the differential df(x).

Solution. Using relation (6.7), we have

df(x) d (V'u 2 + v 2) = ~ (u 2 + v2)(1-m)/md(u 2 + v 2)

2
-( u 2 + v 2 )(1-m)/m( U du + v dv).
m

Example 6.24. Find the approximate value of the number A, whose exact value is

a) A = J4.0003; b) A = In(1.001); c) A = .ij1.0003 .

Solutions.

a) For the function f(x) = ,;x, x ~ 0, its first derivative is f'(x) = 2~' x> 0,
hence for x = 4, h = 0.0003, it follows from formula (6.8)
1 0.0003
J4.0003 ~ v'4 + 1A0.0003 = 2 + - - = 2.000075.
2v 4 4

b) Let us put h = 0.001, x = 1; then it holds

1
In(1.001) ~ In 1 + 10.001 = 0.001.

c) In this case we have h = 0.0003, x = 1, hence

.ijl.0003 ~ V'1 + !110.0003 = 1.0001 .

3v1
236 CHAPTER 6.

Example 6.25. Prove the approximate formula

x
yfa n + x ::::J a + n an- 1' (6.15)

where a > 0, n E N, while the number x satisfies Ixl ~ an (i.e., Ixl is much smaller
than an).

Solution. If we put f(x) := yfa n + x, then we have

1'(0) = !.(a n + x)(1-n)/nl = _1_.

n x=O na n 1
-

Since f is differentiable at 0, it follows

x
f(x) - f(O) - - 1 + X· rex),
= -nan (6.16)

where the remainder r satisfies lim rex) = 0, see Definition 6.2 and Theorem 6.3.
x->o
Thus neglecting the addend x . rex) in (6.16), we obtain the approximate formula
(6.15), provided that Ixl is much smaller than an.

Exercise 6.26. Prove the following approximate formulas for "small" Ixl.
1 x
a) ~::::Jl+-; b) sin x ::::J x; c) cos X ::::J 1.
1- x 3

Example 6.27. Assume that the functions u = u(x) and v = vex) are n times
differentiable on an interval (a,b), where n E N. Then using the usual convention
u(Oj(x) := u(x), the following, so-called Leibniz formula, holds:

(u(x) . v(x))(n) = ~ (; )uUl(x)v(n-j)(x), x E (a, b). ( 6.17)

Solution. For n = 1 the relation (6.17) reduces to the well known formula of the
product of differentiable functions.
Assume that relation (6.17) is true for n = k; we have to show that then it is
true for n = k + 1. It holds for x E (a, b)

(u(x) . v(x n"H) ~ ((u(x) . v(x)i'»)' ~ (t, G) u0> (x)v"-i) (X))'

= t (~)UU+1)(X)v(k-j)(x) + t (~)uU)(X)v(k-j+l)(x)
)=0 J )=0 J

= (~)u'(x)v(k)(x)+G)UIl(X)v(k-1)(X)+. '+(k: l)u(k)(X)V'(X)+G)u(k+l)(x)v(x)

+ (~) u(x )v(k+l)(x )+ G) u'(x )v(k)(x)+ .. + (k : 1) u(k-1)(X )v"(x G) u(k)(x )v'(x)
)+
DERIVATNES 237

= ~ (k ~ 1)UU)(X)V(k+ 1 -j)(X).
3=0 J
Thus we obtained relation (6.17) for n = k + 1, once we assumed its correctness
for n = k. By the principle of the mathematical induction, it follows that (6.17) is
correct for all n EN.
The reader should note that we used the equality

G) + C~l) G:D, =

which is true for all kEN and 0 ~ j ~ k - 1, see Example 1.28 a).

Example 6.28. Assume that the function f is three times differentiable on its do-
main. Find g" and gill, if the function 9 is given by:
a) g(x) = f(x 2 ); b) g(x) = f(1/x);

c) g(x) = f(e X ); d) g(x) = f(lnx).

Solutions.
a) Using the rule for the derivative of the composite function, it follows that
g'(x) = 2xf'(x 2 ), g"(x) = 4x 2 1"(X 2 ) + 2f'(x 2 ),

glll( x) = 8x 3 1'" (x 2 ) + 12x 1"( x 2 ).

b) We have

g' (x) = - ~
x
f' ( 1/ x ), g" ( x) = 23 f' ( 1/ x)
x
+ ~ I" (1/ x ) ,
X

glll(X) = -~f"'(l/x) - ~f"(l/x) - ~f'(l/x).

6 x 5 x X4

c) g"(x) = e2x f"(e X) + eX f'(e X),

gill ( x) = e3x f'''( eX) + 3e 2x 1"( eX) + eX f'( eX).

d) g"(x) = ~ (J"(lnx) - f'(lnx)) ,
x
glll(X) = --; (J11I(lnx) - 31"(lnx) + 2f'(lnx)).
x
Exercise 6.29. Prove that the coefficients aj, 0 ~ j ~ n, of the polynomial
Pn(x) = anx n + an_lx n- 1 + ... + alx + aD, x E R,
satisfy the following condition:

a- =
p~j)(O)
-., '
0 < j ~ n.
_ (6.18)
3 J.
238 CHAPTER 6.

Hint. Putting x = 0 in the equality

p~j)(x) = ann(n-l)··· (n-j+l)x n- j +an_l(n-l)(n-2) ... (n_j)x n- j - 1 +. ·+ajj!,

we obtain (6.18) for 0 < j :::; n.
Example 6.30. Find the j -th derivative (j E N) of the functions given below.

a) f (x) = eX, x E R; b) f (x) = 2x , x E R;

c) f (x) = sin x, x E R; d) f(x) = cosx, x E R;

1
e) f(x) = 1 _ x' Ixl < 1; f) f(x) = In(1 + x); Ixl < 1.
In particular, find f(j)(O) for j EN.
Solutions.
a) Since f'(x) = eX and f"(x) = eX for every x E R and using the mathematical
induction on j E N it follows that for all j EN, f(j) (x) = eX. Hence f(j) (0) =
1, for all j E N.

b) Similarly as in a), we have f(j) (x) = 2X Inj 2, for every x E Rand j EN. Thus
f(j)(O) = ln j 2, j EN.

c) If f(x) = sinx, x E R, then f'(x) = cosx, x E R, f"(x) = -sinx, x E R,

f"'(x) = -cosx, x E Rand j(4)(X) = sin x, x E R. Since
f(x) = f(4)(X) = sinx, x E R,

it follows for every x E R and every j E No that

f(4 j )( x) = sin x, f(4j+l)(X) = cos x,

j(4 j+2)(X) = -sinx, f(4 j+3)(X) =- cos x,

or we can write f(n)(x) = sin (x + n21l") n E N. Thus we have for j E No

f(4 j )(0) = f(4 j +2)(0) = 0, j(4 j +1)(X) = 1, j(4j+3)(X) = -1.

d) Analogously as in c) we have for every x E R and every j E No

f(4 j )(X) = cos x, f(4 j +1)(X) = -sinx,

j(4j+2)(X) = -cosx, j(4 j+3)(X) = sinx,

or we can write j(n)(x) = cos (x + n21l") n E N. In particular,

f(4 j )(0) = 1, f(4j+2)(0) = -1, f(4 j+1)(0) = j(4 j+3)(0) = O.

DERIVATNES 239

e) We leave to the reader to check that for every j EN, it holds

.,
j {j)( x ) -- (1 _J.X )j+l' IX I < 1.

Hence for every j EN it holds j{j)(O) = j!.

f) The first and the second derivative of the function j(x) = In(I + x), Ixl < 1, are

f'(x) = _1_, Ixl < 1, and f"(x) =( -1 \ry' Ixl < l.

I+x I+x
By the mathematical induction it follows then for every j E N

j{j)(x) = (_I)j-l(j -:- I)! Ixl < 1

(I+x)J '

(check!) and thus j{j)(O) = (_I)j-l(j - I)!.

Example 6.31. Find the n-th derivative of the following functions.

1
a) f(x) =ry ... ", x ~ {I,2}; b) f(x) = sin 2 x, x E R;

+ cos 4 x, a + bx
c) f(x) = sin 4 x x E R; d) f(x) = In --b-' a2
a- x
- b2 x 2 > o.
Solutions.

a) Since it holds for x f. 1, x f. 2,

1 1
f(x) = ._1
x-2 - x-I'
it follows that

f(n)(x) = (-Itn! ex _I2)n+l - (x _II)n+l) .

b) Since we have cos 2 x = ~ + ~ cos 2x, it follows that

f(n)(x) = 2n- 1 cos(2x + n1r/2)

Compare to Example 6.30 c), d).

c) Since f(x) = ~ + ~COS4X, it follows for n E N.

j(n)(x) = 4n- 1 cos(4x + n1r/2).
240 CHAPTER 6.

b b
d) From f'(x) = - - + - - we get
a + bx a - bx
n (n-l)!b n
f (n)( x ) -_ (-I)n-l(n-l)!b + .
(a + bx)n (a - bx)n

Exercise 6.32. Find the n-th derivative of the following functions.

1
a) f(x) = - - , Ixl < 1; b) f(x) = In(1 - x), Ixl < 1;
l+x

c) f(x) = sin3 x, x E R; d) f(x)=sin 2 2xcos3x, xER.

Answers.
( -1)nn! b) f(n)(x) = -(n - I)!
a) f(n)(x) = (1 + x)n+1 (1 - x)n .

3 . 1.
c) f(x)=-smx--sm3x () 3n . ( 3 x + - .
3 . ( x + - --sm
fn(x)=-sm n1l') n1l')
4 4' 4 24 2
1 - cos 4x cos 3x 1
d) f(x)= 2 cos3x=-2--4'(cos7x+cosx),

n
3 cos ( 3x+T
f(n)(x)=T n1l') -4'1 ( 7ncos ( 7x+T
n1l') +cos (n1l'))
x+T .

Example 6.33. Prove that the function

e-l/X, x> 0;
{
f(x) = 0, x::; O.

is infinitely differentiable on R, and, in particular, it holds f(n)(o) = 0 for all n E N.

Solution. Clearly, only the point x = 0 is to be discussed. Firstly, f is continuous
at 0, because it holds

lim f( x) = lim e- 1 / x = lim e- t = 0

X-+O+ X-+O+ t-++oo

lim f(x) = lim 0 = 0 = f(O).

x-+o- x-+o-

For x > 0 we have

f'(x) = ~e-l/X
x2 '
1 2) e -1/
f "( X ) -_ ( - - -3
x4 x
x
'

f ill (
X
) -
-
(
---+-
1
x6
6
x5
6) e -1/x .
X4

We leave to the reader to check that it holds

f(n)(x) = Q2n(1/x)e- 1 / X,
DERIVATIVES 241

where Q2n(X) is a polynomial of degree 2n such that Q2n(0) = o. Thus (see Section
6.4) we have

lim f(x) = lim Q2n(1/x)e- 1 / x = lim Q2n(t)e- t = O.

x-+O+ x-+o+ t-++CX)

Since obviously f{ n) (x) = 0 for all x < 0, it follows that

lim f{n)(x)
x-+o-
= x-+o-
lim 0 = O.

Thus the right and the left-hand side limits of f{n) at 0 are both equal to zero.
We still have to prove the existence of f{n)(o). Of course, In)(O) = 0, for all
n EN. Further on we have

f~(O) = lim f(O + h) - f(O) e- 1 / h

h--->O+ h = -h- = o.
If we assume that f~n) (0) = 0 for some n EN, then it holds

f~n+1)(O) = lim f{n)(o + h) - f{n)(o)

h--->O+ h

r hQ2n(1/h)e-l/h
h2.rr+ 1 = O.

Hence by the principle of the mathematical induction it follows that f~n)(O) = 0 for
all n E N, which implies the existence of f{n) on R.
Remark. This example shows that there exist infinitely differentiable functions
whose Maclaurin's polynomial of any degree is identically equal to zero, even though
the function is not identically equal to zero.

Example 6.34. Find the equations of the tangent and perpendicular lines at the
given point T(xo, f(xo)) on the graph of the following functions.

a) f(x) = y'X, Xo = 4; b) f(x) =e x2

-\ Xo = 1;

c) f(x) = arctanx 2 , Xo = 0; d) f(x) = arCSIn

+ 2) ,Xo = o.
. (X-2-
Solutions.

a) The equation of a line with slope k passing through the point T(xo, Yo) has the
form
y-yo=k·(x-xo).
The geometric interpretation of the first derivative at the point Xo is the tan-
gent line of the graph of f at its point T(xo, Yo), i.e., k = f'(xo) (see relation
(6.11)) together with f(xo) = Yo).
242 CHAPTER 6.

The first derivative of f at the point x is

f'(x) = 2~'
hence putting Xo = 4 gives us f' (4) = ~. The value of f at Xo = 4 is Yo =
f(4) = 2. Thus the equation of the sought tangent line has the form
1
y-2=4(x-4), or 4y-x=4.

The equation of the perpendicular line of the graph f at its point T(xo, Yo) is
of the form (see equation (6.10))
y - Yo = k1 . (x - xo),

where kl= - ~, provided, that the slope k( = f' (xo)) is nonzero. Hence
k1 = -4, which gives the equation of the perpendicular line at the point
T(4,2),
y + 4x = 18.

b) From the first derivative f'(x) = 2xex2 - 1 it follows that the slope k of the
tangent line is k = f'(I) = 2. Hence the equation of the sought tangent line
has the form
y -1 = k(x -I):::} y -1 = 2(x -1),
or finally
y - 2x +1 = O.
The equation of the perpendicular line of the graph f at its point T(I, I) is of
the form
2y + x - 3 = O.

c) From y' 2x 4 it follows that y'(O) o. Thus the tangent line has the
l+x
equation
y- 0 = 0 . (x - 0) hence y = O.
The sought tangent line is, in fact, the x-axis. Clearly, the perpendicular line
of the graph f at the point T(O,O) is the y-axis, whose equation is x = o.
d) The function f has for its natural domain the interval [-4,0), the absolute value
of the argument of the arcsin function must be less than or equal to one. It
is easy to show that f is continuous from the left-hand side at the point O.
However, f has no left sided derivative at the point 0, since

lim f(O + h) - f(O) = lim arCSIn

. (h+2)
-2- -
.
arCSIn I
h_O- h h_O- h

Iim t - 7r /2 1· z 1. z
= 1m = 1m = +00.
t_tr/2- 2(sint -1) z_O- 2(cosz -1) z_O- -4(sin 2 (z/2»)
DERIVATIVES 243

This shows that the sought tangent line is the y-axis, whose equation is x = 0.
7r
The equation of the sought perpendicular line at the point T(O, 2") is y = 0,
which is, in fact, the x-axis.

Example 6.35. Determine the parameter k so that the line y = kx + 1 becomes the
tangent line of the parabola

{(x,y)1 y2 = 4x, x ~ O},

and find their common point.

Solution. The given curve is the union of graphs of two functions, namely of

fl(X) = 2y'X, x ~ 0, and h(x) = -2y'X, x ~ 0.

Let us denote by T(xo, Yo), Xo > 0, the common point of the tangent line and the
parabola. Then
1
Yo = ±2JXo and k = ± Fa'

This gives
1
Yo = kxo + 1 =} 2JXo = --Xo + l.
Fa
So we obtain Xo = 1 and its common point with the parabola and the graph of It is
T(1,2). The equation of the sought tangent line is y = x + l.
Notice that there is no tangent line to the graph of h, since
1
-2JXo = ---Xo + 1 =} JXo = -1 < 0,
Fa
a contradiction. In fact, at the point x = 0, the graph of the function h and also
the given parabola have a vertical tangent line.
Example 6.36. Find the tangent lines to the parabola f( x) = x 2 - 3x + 1 through
the point A(2, -2) and determine the common point of each tangent line and the
parabola.
Solution. Let us denote by (xo, f(xo)) the common point of the tangent line and
the given parabola. The slope of the tangent line is k = f'(xo) = 2xo - 3, while its
equation is
y - f(xo) = k(x - xo), or
-2 - (x~ - 3xo + 1) = (2xo - 3)(2 - xo).
This implies the quadratic equation

x~ - 4xo + 3 = 0, whose solutions are x~ = 1, x~ = 3.

Thus the common points are B(l, -1) and C(3, 1), and the slopes of tangent lines
through the point A are kl = -1 and k2 = 3. The equations of the tangent lines are

y +x = ° and y = 3x - 8.
244 CHAPTER 6.

Example 6.37. The parametric representation of the central ellipse is

x=acost, y=bsint, 0:=;t:=;27f,
where a > 0 and b > 0 are the so-called half-axes of the ellipse. Prove that the area
of the triangle AO B is not less than the product ab, where A and B are those points
on the x- and y-axis in which the tangent line at an arbitrary point (x, y) of the
ellipsis intersects the x- and the y-axis respectively, while 0(0,0) is the origin.
Solution. Since the ellipse is symmetric both to x- and to the y-axis, it is enough
to analyze the case 0 < t < 7f /2; clearly in our case the values t = 0 and t = 7f /2
have no geometric meaning. The tangent line of the ellipse at the point (x(t),y(t))
has the equation
t
y - b sin = ( - ~ cot t) (x - a cos t).

The intersection of this line with the coordinate axes are the points A (_a_,
cos t
0) and

B (o,~)
smt
.Hence the area A~AOB of the triangle AOB satisfies
1 a b -~>ab.
A~AOB = "2 cos t sin t - 2 sin 2t -

6.2 Mean value theorems

6.2.1 Basic notions
Theorem 6.38. Rolle's Theorem.
If a function f : [a, b] --+ R is
• continuous on a closed interval [a, b],

• differentiable on the open interval (a, b),

• f(a) = f(b),
then there exists at least one number eE (a, b) such that 1'(0 = o.
Theorem 6.39. The Lagrange Theorem.
If a function f : [a, b] --+ R is
• continuous on a closed interval [a, b],
• differentiable on the open interval (a, b),
then there exists at least one number eE (a, b) such that
f'(e) = f(b) - f(a) (6.19)
b- a .
Geometrically, Theorem 6.38 (resp. 6.39) means that the tangent line at the point
eis parallel to x-axis (resp. to the line y - f(a) = f(b~ - f(a) (x - a) ).
DERIVATNES 245

6.2.2 Examples and exercises

Example 6.40. Show that the function f( x) = x( x-l)( x - 2) satisfies the con-
ditions of Rolle's theorem on the intervals [0,1], [1,2] and [0,2] and determine the
corresponding values of ~.

Solution. The given function is polynomial, hence it is continuous and has first
derivatives at every point of these intervals. Since it holds f(O) = f(l) = f(2) = 0,
the function f satisfies the conditions of Rolle's theorem on all three intervals [0,1],
[1,2] and [0,2].
We have I'(x) = 3x 2 - 6x + 2, and

I'(x) = ° for X12, = 1 ± v'3

3 .

So the points in which the first derivative of the considered function is equal to zero
are
6 =1- 3
v'3 E [0,1], and ~2 = 1 + 3v'3 E [1,2].

On the interval [0,2] the function I' has two real zeros, ~1 and ~2'

Example 6.41. On the intervals (-1,1) and (1,2) find the points in which the
tangent lines of the graph of the function f(x) = (x 2 -l)(x - 2) are horizontal.

Solution. The considered function is continuous and has a first derivative at every
point of these two intervals and it holds that f(-l) = f(l) = f(2) = 0. Hence f
satisfies the conditions of Rolle's theorem on the intervals [-1,1] and [1,2], and also
on the interval [-1,2].
From 1'( x) = 3x 2 - 4x - 1 it follows that 1'( x) = for °
~1 = 2 - n v'7 E [-1,1], ~2 = 2 + v'7
n
E [1,2].

The tangent lines of the graph of the function f(x) = (x 2 - 1)(x - 2) in the
points 6 and 6 are horizontal (1'(6) = 1'(6) = 0).

Example 6.42. Check whether the following functions satisfy the conditions of
Rolle's theorem.
1 - ij;2
a) f(x) = 2 on the interval [-1,1];

b) f(x) = Ix -11 on the interval [0,2];

n
c) f(x) = I: ~i sin(ix) on the interval [0, 7r];
i=1 z

d) f(x) = t ~i
i=1 z
sin(ix) on the interval [0, 7r/2].
246 CHAPTER 6.

Solutions.
a) No, because the function f has no first derivative at the point x = 0, as seen
from either of the following one-sided derivatives at 0:

lim f (0 + h) - f (0) ~ W 1
h-.O+ h I'1m 2
h-.O+
- 2h -2 = -00',

lim f(O
h-.O-
+ h) - f(O) =
h
lim
h-.O-
-w
2h
= +00.

b) No, because the function f has no first derivative at the point x = 1 (see Example
6.13).

c) Yes, and from Rolle's theorem we obtain that the equation

n

L ai cos (ix) = 0 (6.20)

i=l

has a solution on interval [0, 7r]. Namely, the first derivative of the considered
function is just the left side of the equation (6.20).

d) No, because it is not true that f(O) = f(7r/2). Namely, the value f(7r/2) depends
on n and on the coefficients ai, i = 1,2, ... , n.

Example 6.43. If the polynomial

Pn(x) = anx n + an _1 xn - 1 + ... + alx + ao, an -=f 0,

with real coefficients has only real roots, then its derivatives, P~, P::, ... , p~n-l), have
real roots as well.

Solution. Let us suppose first that all roots are simple. Then from Rolle's theorem
it follows that there exist n - 1 real roots of the polynomial P~ (x). Every root
of the polynomial P~(x) is located between the two roots of Pn(x). Similarly we
obtain that the polynomial P::(x) has n - 2 real roots, and so on. Note that the
(n - l)-th derivative
p~n-1)(X) = ann! x + an-1(n - I)!

has one real root while, the n-th derivative

p~n)(x) = ann!

is a constant.
If a number Xo is a multiple real root of order m > 1 of a polynomial Pn (x),
then it is also a root of the polynomial's derivative, as follows immediately from the
representation
Pn(x) = an(x - xo)mQn_m(x).
DERIVATNES 247

Example 6.44. Prove that the roots of the Legendre polynomial,

1 dn
Pn(x) = 2nnl dxn ((XZ - l)n) , (6.21 )

are real and located in the interval ( -1, 1).

Solution. The polynomial Rn(x) = (x Z - l)n has on the interval [-1,1] exactly
2n real roots, namely

Xl = Xz = ... = Xn = 1 and Xn+l = Xn+Z = ... = XZn = -1.

Since
R~n)(x) = 2 nnlPn (x), (6.22)
it follows from the previous example and Rolle's theorem that on the interval ( -1, 1)
there exist n real roots of the n- th derivative of the polynomial Rn (x). Hence, from
(6.22) it follows that there exist n real roots of the Legendre polynomial Pn(x).
Example 6.45. If a function f has a finite derivative l' at every point of the finite
or infinite interval (a, b) and it holds

lim f(x) = lim f(x), (6.23)

x~a+ x~b-

then there exists at least one point c E (a, b) satisfying f'(c) = O. Prove.

Solution. First we shall assume that (a, b) is a finite interval and let us denote by
C the limits in relation (6.23). Then the function

F(x) = { f(x), xE(a,b);

C, x E {a,b},
satisfies the conditions of Rolle's theorem, since it is continuous on the closed interval
[a,b], has finite first derivative on the open interval (a, b) and it holds F( a) = F( b).
Hence there exists a point c E (a, b) such that F'(c) = f'(c) = O.
If the interval (a, b) is infinite, but the limits in (6.23) are finite,

lim f(x) = lim f(x) = C,

x-+a+ x-+b-

then we consider the lines y = C + c, y = C - c. At least one of them intersects the

curve f in (at least) two points with apscisses ab az, provided that c is sufficiently
small. The function f satisfies the conditions of Rolle's theorem on lab az] C (a, b)
and this implies that there exists a point c E (a, b) such that f'(c) = O.
Similarly one can treat the case when the limits in relation (6.23) are +00 or
-00.

Example 6.46. Prove that

a) Laguerre's polynomial,
dn
Ln(x) = eX dxn (xne- X ),

has n positive roots;

248 CHAPTER 6.

b) the Chebychev-Hermite polynomial,

2 dn 2
Hn(x) = (_l)ne X dxn (e- X ),

has n real roots;

c) the equation,
dn
(1 + x2)n dxn ((1 + X2)-I) = 0,

has n real roots.

Solutions.

a) The function f(x) = xne- X satisfies

f(O) = lim f(x) = O.

x--++oo

From the previous example there exists a point el

E (0, +(0) such that 1'(ed =
o and also lim 1'(x) = O. Applying again the same procedure one obtains
x--++oo
two points 6 E (0,6) and 6 E (e2, +(0) such that 1"(6) = 1"(6) = 0, and
also 1"(0) = 0, and lim 1"(x) = O. Proceeding with this procedure we
X--++CXJ

obtain that the function f(n-l)(x) has n real roots including the point x = 0,
while f(n)(x) has n roots also, because f(n)(o) -=I- O. Therefore the polynomial

Ln(x) = eXf(n) (x)

has n positive zeros.

b) Let us consider the function g( x) = e- x2 on the interval (-00, +(0) and

similarly as in the previous example we obtain that g(n)(x) has n real roots.
So the polynomial
Hn(x) = (-lte x2 g(n)(x)
has n real roots.

c) In this case we consider the function h( x) = - 112' x E R, and apply the

+x
same procedure as in b).

Example 6.47. A differentiable function f is a constant function on the interval

[a, b] if and only if its derivative l' is equal to zero at each point of the open interval
(a, b).

Solution. It is trivial that if a function is a constant on the interval [a, b], then its
derivative is zero on ( a, b).
So we have to prove the opposite, i.e., if 1'(x) = 0 for every x E (a, b), then it
holds that f(XI) = f(X2) for each xI, X2 E [a, b].
DERIVATNES 249

From Lagrange's theorem it follows that there exists a point eE (Xl, X2), hence
eE (a, b), such that
f(xt} - f(X2) = f'(O(XI - X2).
From the supposition f'(x) = 0 and the last relation we obtain f(xt} = f(X2) for
each pair Xl, X2 E [a, bj.

Example 6.48. If the derivatives of two functions FI (x) and F 2 ( x) are equal on
some interval, i. e., it holds

F;(x) = F~(x), Xl, X2 E (a, b),

then the difference of these functions is a constant.

Solution. If we denote by f(x) = FI(x) - F2(x), then it holds that f'(x) = 0 on

the interval (a,b). So from Example 6.47 we obtain that on the same interval f(x)
is a constant.

Example 6.49.

a) If the first derivative of a function f is positive (resp. nonnegative) on an interval

(a, b), then this function is monotonically increasing (resp. nondecreasing) on
(a, b).

b) If the first derivative of the function g is negative (resp. nonpositive) on an inter-

val (a, b), then this function is monotonically decreasing (resp. nonincreasing)
on (a, b).

Solutions.

a) Assume that f'(x) > 0 for all X E (a, b), (resp. f'(x) :2: 0) for all X E (a, b). If
Xl, X2 are two points from the given interval and Xl < X2, then by Lagrange's
e
theorem there exists a point E (Xl, X2) such that

f(X2) - f(XI) = f'(e)(X2 - Xl).

Since by assumption the first derivative of the function f is positive (resp.

nonnegative) on the whole interval (a,b), from the last relation we obtain

f(X2) - f(xd > 0 (resp. f(XI) - f(X2) :2: 0).

This means that the function f is monotonically increasing (resp. nondecreas-

ing) on (a, b).

b) Analogous to a).

Example 6.50. The Cauchy theorem.

Assume that f and g are

• continuous on a closed interval [a, bj,

250 CHAPTER 6.

• differentiable on the open interval (a, b).

If, additionally, g'(x) i= 0 for all x E (a, b), then there exists at least one number
eE (a, b) such that
f(b) - f(a) 1'(0
(6.24)
g(b) - g(a) g'(O'
(If g(a) i= g(b), then the condition g'(x) i= 0, x E (a, b), can be replaced by the
weaker condition f(X)'2 + g(X)'2 i= 0, x E (a, b).)
Solution. If g'(x) i= 0, x E (a, b), then by Example 6.49 it follows that the
function 9 is monotone on (a, b). Assume that 9 is monotonically increasing on
(a,b). Then there exists an inverse function x = x(t), defined and differentiable on
the interval [a,,8], where a = g(a), ,8 = g(b). Note that g(a) < g(b).
The function f(x) can be written as the function f(x) = f(x(t)), depending on
t E [a, ,8]. Thus f is a differentiable function, as the composition of differentiable
functions. Applying the Lagrange's theorem and the chain rule we get

f(x(,8)) - f(x(a)) = f:(x(c))x~(c)

,8-a
for some c E (a, ,8), or
f(b) - f(a) 1'(x(c))
(g(b) - g(a)) g'(x(c)) .
Taking x(c) = e, we obtain the statement of the theorem.
The other cases are left to the reader.
Remarks. The Lagrange's theorem is a special case of Cauchy's theorem, when the
function 9 is given by g( x) = x.
Example 6.51. Check can one apply Cauchy's theorem to the following functions:
a) f(x) = x 2, g(x) = x 3 , X E [-1,1];
b) f(x) = x 2 + 2x, g(x) = x3 , X E [-1,1].
Solutions.
a) No, we can not apply Cauchy's theorem to these functions. Namely, we have
g( -1) = -1 i= 1 = g(l), but at the point x = 0 it holds 1'(0)2 + g'(0)2 = O.
Hence the condition 1'(X)2 + g'(X)2 i= 0, for all x E [-1,1], is not satisfied.
b) Yes, because in this case we have g( -1) = -1 i= 1 = g(l), 1'(x) = 2x + 2 and
g'(x) = 3x 2, so it holds U'(X))2 + (g'(X))2 i= 0, for all x E [-1,1].
Let us find the point e satisfying
f(l) - f( -1)
1'(e) eE (-1,1).
g(l) - g(-l) g'(e)'
From
3+1 2e +2
3r'
2
1+1 = or 3e - e - 1 = 0,

1 + v'I3 1- v'I3
we obtain 6 = ~ E(-1,1)ande2= c E(-l,l).
DERIVATNES 251

Example 6.52. Using Lagrange's theorem show the following inequalities.

a) I sinax - sinayl ::; lal'lx - yl, a -=J 0;

b) I arctan x - arctan yl ::; Ix - yl, x, y E R;

x-y x x-y
c) - - < I n - < - - , 0< y < x;
x y y
x
d) - - < In(1 + x) < x, x> 0;
l+x

e) eX> 1 + x, x E R;

f) eX > ex, x > 1;

g) n(b - a)a n- 1 < bn - an < n(b - a)b n- 1 , 0 < a < b, n E N;

a
h) -+1
n'"
< (1
( ) - -1),
n - 1 '" n'"
n E N, a> O.

Solutions.
a) Applying Lagrange's theorem to the function f(t) = sin at, which is continuous
and differentiable on any interval [y, x], we obtain that there exists a point
e E (y,x), such that it holds
sin ax - sin ay = a( x - y) cos ae, wherefrom

I sin ax - sinayl = lal·lcosael·lx - yl::; lal'lx - yl·

b) The function f (t) = arctan t is continuous and differentiable on the interval

[y, x], so there exists a point e E (y, x) with the property
x-y
arctan x - arctan y = 1 + e.
So we obtain
Ix -yl
I arctan x - arctanyl = 1 + (e)2 ::; Ix - yl·
c) The function f (x) = In x is continuous and differentiable on the interval [y, x],
o < y < x, hence, there exists a point e E (y, x) such that
x-y
lnx -lny = - - e .
From the relations

Ilnx-Inyl =-e
Ix - yl
and
1
-<-<-
x e
1 1
y'
we obtain the given inequality.
252 CHAPTER 6.

d) Using the function f(t) = In(1 + t) on the interval [0, x], and the relation

In(1 + x) -In(1 + 0) 1 1 1
- - t ' for ~ E (O,x), hence - - < - - t < 1,
x 1+., l+x 1+.,
we obtain the inequality.

e) The function f(x) = eX on the interval [0, x], x> ° (resp. on [x, 0], x < 0),
satisfies the conditions of Lagrange's theorem,

eX _eo
- - = e E =} eX - 1 = xe E.
x

• If x > 0, hence ~ > 0, then ee > 1 and eX > 1 + x .

• If x < 0, hence ~ < 0, then ee < 1, but xee > x, thus the inequality is
true.

f) Taking the same function as in the previous case and the interval [1, x], we have
eX _ e l
- - - = eE > e, for some ~ E (l,x).
x-I

g) lEnt. Use the function y = t n and the interval [a, b].

h) Taking the function f (x) = ~ on the interval [n - 1, n], we obtain the given
x"
inequality.

Example 6.53. If a derivative of a function f is bounded on an interval (a, b)} then

f is uniformly continuous on (a, b).

°
Solution. Assume that there exists M > such that 1f'(x)l::::; M. Then applying
the Lagrange theorem, for every pair of points Xl> X2 E (a, b) we can write

If(xt} - f(X2)1 = 1f'(OllxI - x21 ::::; MlxI - x21, ~ E (Xl, X2)'

Hence for given E > 0, we put 6:= ~ and obtain

(VXI' X2 E (a, b)) IXI - x21 < 6 =} If(xd - f(X2)1 < E.

6.3 Taylor's formula

6.3.1 Basic notions
Theorem 6.54. Taylor's theorem.
Let f be a continuous function and n a positive integer such that all derivatives up
to the n-th are continuous on an interval [a, b] and there exists a finite f(n+l) on the
DERIVATNES 253

interval (a, b). Then there exists eE ( a, b) such that for every x E ( a, b) the following
Taylor's formula holds
(X a)2 (x a)3
f(x) = f(a) + (x - a)f'(a) + 2! f"(a) + 3! f"'(a)

+ ... + (x - a)n f(n)(a) + (x - a)n+l f(n+l)(e).

n! (n+l)!
The following polynomial is called Taylor's polynomial of order n at the point a
for a function f.

Pn(x) = f(a)+(x-a)f'(a)+ (x _,a)2 f"(a) + (x _,a)3 f"'(a)+ ... + (x _,a)n f(n)(a).

2. 3. n.
Thus the Taylor's formula can be written as

f(x) = Pn(x) + Rn(x), (6.25)

where the remainder Rn is given by

(x at+ 1
Rn(x)= {-, 1 \ ' f(n+l)(e), e=a+().(x-a),O<()=()(x)<l.

Using Definition 4.54 we can write relation (6.25) as

f(x)=Pn(x)+o((x-at), x-ta.
For a = 0 we obtain Maclaurin's formula
2 3
f(x) = f(O) + xf'(O) + ~! J"(x) + ~! f"'(O)
x n+ 1
+ ... + -
xn
n!
f(n)(o) +
(n+l)!
f(n+1)(e), eE (a ' b).
The polynomial
x2 xn
Pn(x) = f(O) + xf'(O) + 1"1"(0)
2.
+ ... + ,.
n.
f(n)(o),

is called Maclaurin's polynomial of order n for the function f.

3.3.2 Examples and exercises

Exercise 6.55. Show that the Taylor's polynomial for a polynomial Pn (x) of degree
is just the polynomial itself.
"1

Example 6.56. Develop the polynomial

f(x) = x4 - 5x 3 - 3x 2 + 7x + 6
~nto powers of x - 2.
254 CHAPTER 6.

Solution. From f(2) = -16 and

1'(x) = 4x 3 - 15x 2 - 6x + 7, 1'(2) = -33
1"(x) = 12x2 - 30x - 6 1"(2) = -18;
1"'(x) = 24x - 30, 1"'(2) = 18, f(4)(X) = 24, f(4)(2) = 24,

it follows

X4 - 5x 3 - 3x 2 + 7x +6 -16 - 33(x _ 2) _I8(x - 2)2

2

+ 18(x-2)3 (X-2)4
31 +24---'-
. 4!
Thus
X4 - 5x 3 - 3x 2 + 7x +6 -16 - 33(x - 2) - 9(x - 2)2

+ 3(x-2)3+(x-2)4.
In this case the remainder is equal to zero, because the fifth derivative of f is
identically equal to zero.
Example 6.57. Approximate the function f(x) = eX using the Maclaurin's poly-
nomial of the

a) first degree; b) second degree; c) third degree,

and determine the corresponding remainder.
Solutions. From 1'(x) = eX = 1"(x) = 1"'(x) = f(4)(X), and f(O) = 1'(0) =
1"(0) = 1"'(0) = 1, it follows that
2
X e
a) eX=I+x+o(x), x~O and Rl = T e .

x2 3
x €
b) eX=I+x+T+o(x2), x~O and R2 = 3Te .
x2 x3 X4 e
c) eX = 1 + x + -2 + I" + o( x 3 ), X ~ 0 and R3 - -4!e ,
3. -

where in all three cases e

= ()x, 0 < () < 1, and 1 < ee < e.
Note that for x E [0,1], the remainders can be estimated by
x n +1 e
)Ie:::;( n+I.
Rn:::;( n+I. )1' n=I,2,3.

Example 6.58. Apply the Maclaurin's formula to the functions given below.

a) f(x) = eX; b) f(x) = sinx; c) f(x) = cos x;

1
d) f(x) = 1 _ x; e) f(x) = In(I - x); f) (1 + x )'\ a E Q.
DERIVATIVES 255

Solutions. Using Example 6.30 we obtain, when x -t 0:

n Xk n (_1)k x 2k+l
a) eX = 2:, + o(xn); b) sinx = 2:
k=O (2k + I)!
+ O(X 2n+2).
'
k=O k.
n
c) cosx- - Ln (_1)kx2k
(k)' + ox
( 2n+1)., 1
d) - - = xk L + o(xn);
k=o 2. 1- X k=O

e) In(l - x) = -
n xk
Lk
k=l
+ o(xn); f) (1 + x)" = 'to (~)xk + o(xn).
In f), 0: E R \ {O} and by definition
(0:) _ (0:) _ 0:(0: - 1) ... (0: - (k -
o - 1, k - k!
1))
' kEN.

Example 6.59. Find Maclaurin's formula for the functions given below.

a) f(x) = e5x+\ b) f(x) = vr=x;

1

c) f(x) = sin(2x + 7r/3); d) f(x) = In(x + e).

Solutions. In this example we use the following simple statement. If b i- 0 and
n n
f(x) =L ak xk + o(xn), then f(bx) =L akbkxk + o(xn), X -t o.
k=O k=O
a) From e 5x +1 = e· e 5x we obtain
n 5k k
e 5x+l = e ""' X
L...J -k' + 0
(
x n) , X -t o.
k=O .
b) From Example 6.58 f) we have

+ (_1)x)-1/2,
vr=x
_1_ = (1

hence

-1= = E( _l)k
n (
- 1/2)
k xk + o(xn), X -t 0, where for kEN

(
-1/2)= (-~).(-~-1) ... (-~-(k-1)) =(_ )d 2k - 1)!!
k k! 1 2kk!·
Thus
1 ~(2k-1)!! k n
--= = 1
1- X
+ k=l
L...J
2
kk'
.
x + o(x), x -t O.
256 CHAPTER 6.

c) From f(k)(x) = 2k sin(2x + rr/3 + h/2) and f(k)(O) = 2ksinC~·(2+3k»),

kEN, and from Example 6.58 b) we have

n 22k+lX2k+l (rr )
sin(2x+rr/3)=E(-1)k (2k+1)! sin 6(2+3k) +o(x2n+2), x---+O.

d) From In(x + e) = 1 + In(l + x/e) and Example 6.58 e) we get

n (_l)kxk
In(x + e) = 1 + E ~J..l + o(xn), X ---+ O.
k=l

Example 6.60. Apply the Maclaurin's formula to the following functions.

a) f(x) = (x 2 + 5)e3x ; b) f(x) = (x 2 ~x3eX);

e

c) f(x) = (x + 2)JI+X; d) f(x) = cos x + Ixl s ;

2 - 3x
e) f( x) = In 3 + 2x ; f) f(x) = In(6 + llx + 6x 2 + x 3);

1 - 2X2 x2 + 2
g) f(X)=2+ x-x 2; h) f(x)=x 3 +x2 +x+l

Solutions.

a) Using the equality (x 2 + 5)e3x = x 2e3x + 5e3x we have

n- 2 3kxk n 3kxk
f(x) x 2 E ~ + o(xn) + 5 E ~ + o(xn)

E(k - 2)! + 5 E~ + 5 + 15x +

n 3k-2xk n 3kxk
o(x n )

5+15x+ E (k-2)!
n (3k-2 3k)
k! +5 xk+o(xn)

n (3k-2 )
5+15x+E ~(k(k-l)+5.9) xk+o(xn), x---+O.
DERIVATIVES 257

b) In this case we have

n-2 ( 2x)k ) n ( x)k
f(x) x 2e- 2x + 3e- = x 2 ( L - ,
X
+ o(x n- 2) + 3 L ~ + o(xn)
k=O k. k=O k.

n (_2)k-2Xk n (_l)kxk n
{; (k-2)! +3-3x+3{; 7.' +O(X)

n ( l)kxk
3 - 3x + L(3 + k(k - 1)2 k- 2) - 71 + o(xn), X - t O.
k=2
c) Using Example 6.58 f) for 0: = 1/2, we get

V
~ x
1- x = 1- - - x
L
n (2k - 3)!! k
+ o(x n ) X -t O.
2 k=2 2kk! '
Hence
x2 (2k - 5)!! k
+ 2)yt'f=X
n n
(x x- 2 - L n
ol-_l 1 \,x + o(x )
k=3

n (2k - 3)!! k n
+ 2-x-2L kk' x +o(x )
k=2 2 .

3x 2 ~ 6(2k - 5)!!(k - 1) xk
2 -4- -L...J 2kk!
+ o(xn), X - t O.
k=3
d) Let us consider the function h(x) = Ix15. It has only four derivatives on Rand
it holds
h(O) = h'(O) = h"(O) = h"'(O) = h(4)(0) = O.
Now the function h has a fifth derivative on the set R \ {OJ, so we can write
h(x) = P4(X) + o(x 4) = O(X4), X - t 0,
since P4(X) == o. Thus we have
x2 X4
cos X + Ixl 5 = 1 - - + , + o(X4),
2 4.
X - t O.

e) Since
2 - 3x
In--=ln2-ln3+ln ( 1 -3X)
- -In ( 1 +2X)
-
3 + 2x 2 3 '
it follows from Example 6.58 e) that
n 3kxk n (_2)kxk
f(x) = In2-ln3-L~k +o(xn)+L k +o(x n )
k=l 2 k=l 3 k

n (( 4)k gk)xk
In2-ln3+L - l-~. +o(xn), x-tO.
k=l 6
258 CHAPTER 6.

f) From
In(6 + llx + 6x 2 + X3) = In(3 + x) + In(2 + x) + In(l + x),
it follows
n (_l)k-lXk n (_l)k-lXk n (_l)k-lXk n
f(x) = ln6+ L 3k k + L 2kk + L 1. +O(X)
k=l k=l k=l

n ( l)k-lXk
ln6 + L - C!,.L (2k + 3k + 6k ) + O(Xn), x -+ O.
k=l

g) From
1 - 2x2 -7 -1
--- - 2+ + --,---,-
2+x-x - 2 3(2-x) 3(1+x)'
it follows for x -+ 0
7 n xk 1 n 1 n ((_2)k+l_7)x k
f(x)=2--'Lk"--'L(-x)k+ o(x n )=-+L Ie ... , +o(x n ).
6 k=O 2 3 k=O 2 k=l 3.2
h) In this case we have
x2 + 2 1 ( 1- x 3)
f(x) = x3+x2+x+l ="2 1+x2+1+x

~ (E( _1)k x 2k - E( _1)k x 2k+1 + 3 ~{_l)kXk) + o(x2n)

n (3+(_1)k)x 2k n-l((_1)k+ 1 _3)x 2k+1 2n

L 2 +L () + o(x ), x -+ O.
k=O k=O
Example 6.61. Apply the Maclaurin's formula to the functions below.
a) f(x) = x 2 cos 2 x; b) f(x) = sin 2 xcos 2 x;

c) f(x) = sin4 + cos 4 x; d) f(x) = arctan x;

e) f(x) = arcsin x; f) f(x) = sinh x;

g) f(x) = coshx; h) f(x) = sinh x . cosh2x.

Solutions.
1 + cos2x .
a) Since x 2 cos 2 x = x 2 2 ,It follows from Example 6.58 c)
x2 1 n-l (_1)k22kx2k+2
f(x) = 2 + "2 L 1M\! + O(X2n+l)
k=O
n (_1)k-122k-3X2k
X
2
+~
'"
(2k _ 2)! + 0 ( x 2n+l) , X -+ 0 .
DERIVATNES 259

b) Since it holds
sin 2 x cos 2 x = ~(1 - cos 4x),
we have
n (_1)k-124k-3X2k
f(x) = L ("'\I + o(x2n+1), X ----t O.
k=l
c) From
. 1
sm 4 x + cos 4 X = 4(3 + cos 4x),
we have
n (_1)k24k-2x2k
f( x) = 1 + '"
L...,.; (01_\1
+ o(x2n+l) ,
X ----t O.
k=l
d) Since
1
L( _1)kx2k + o(x2n+1),
n
(arctan x)' = --2 =
l+x k=O
it follows
n x2k+1
arctan x = L( _l)k _ _ + o(x2n+2), X ----t O.
k=O 2k + 1
e) Since
(arcsin x)' = 1
v'f=X2 = 1+~
~ (2k2kk!
- I)!! x 2k + o(x2n+1),

it follows
.
arcsm x = x +L
n(2k-1)11.. x 2k +1 + o(x2n+2)
k=l 2kk!(2k + 1) .
eX _ e- X
f) From sinh x = " ,we obtain
n x2k+1
sinhx=L( k )1+ o(X 2n +2), x----tO.
k=O 2 + 1 .

eX + e- x
g) From cosh x = " ,we obtain
n 2k
cosh x = L (xk)1 + o(x2n+1), X ----t O.
k=O 2 .

e 3x _ eX + e- x - e- 3x 1 1
h) Since sinh x . cosh 2x = , = 2" sinh 3x - 2" sinh x, it follows
1 (n 32k+1x 2k +1 n x2k+1)
sinh x . cosh2x = - L
(k
2 k=O 2 + 1.
)1 - (k )1
k=O 2 + 1 .
L + o(x2n+2)

n x2k+1
2k
~ 2(2k + I)! (3 +1 - 1)
'" + 0 (2n+2)
x , x ----t O.
260 CHAPTER 6.

Example 6.62. Determine

. 2Vl + 2 tan x - 2e X + 2X2

a) 11m . . ;
x-+o arCSIn x - sm x

3x 2 2
b) lim x +2- - I---;; + exp( arctan x) +1
x-+o 1- x
I n - +2x
l+x

c)
. cosh(sin x)
hm
x-+o
-11 _
x
--
2
3x
5
2 2

tanh x - x

Solutions.

a) In this case we shall consider Maclaurin's formulas for the corresponding func-
tions. Since
t t2 3) x
t3 x3
V ~+ t
1 T L
= 1+---+-+0(x
2 8 16 ' --+ 0, t = tanx = x+ 3 +0(x3), x --+ 0,

we get

2tanx (2tanx)2 (2tanx)3 ( 3)

VI + 2tanx 1 + -2- - 8 + 16 +0 tan x ,

x2 5x 3
l+x-"2+T+o(x3), x--+O.

Also, from the equalities

x2 x3
eX = 1 + x + "2 + ""6 + 0(x 3),

x3
sin x = x -""6 + 0(x3), X --+ 0

x3
arcsin x = x + ""6 + 0(x 3), X --+ 0, hence

3
arcsinx-sinx=~ +0(x 3), x--+O,

we obtain

4 2 3
. 2Vl + 2 tan x - 2ex + 2X2 . -3 x + o(x )
hm . . = hm = 4.
arCSIn x - sIn x x-+o 1 3 (3)
x-+o
3"x +0 x
DERIVATNES 261

b) Using the following relations as x ~ 0:

x3
arctanx = x - "3 + o(x 3 ),

x ) 2 2
+ 2x +""6
x
+ o(x 3 ) = 1 + x + 2x -""6
x
3 3 3
exp(arctanx) = 1+ (x -"3 + o(x 3 ),

2 3
1- x
In - - = -2x - -x + o(x 3)
1+x 3 '

2
- - = 2 + 2x + 2X2 + 2x 3 + o(x 3 ),
I-x
we obtain
2
x + -3x - -2- + exp(arctanx) + 1 13
__ x 3 + o(x3 )
lim 2 1- x = lim 6 = 13/4.
2 3
x-+o 1 1- x
n -- + 2x x-+o --x + 0 (3)
x
l+x 3
c) From the relations

cosh(sin x) 1 + -1 ( x-=-- 3)2 + -1 ( x-=-- 3)4 + o(x4)

2 6 24 6 '

x2 X4
1+- - -8 + o(x 4) X ~ o·,
2 '

(1 _~2r/5 x2
1 ___
10
4
50 =-- + (x 4) ,0 x ~ 0;

x3
tanh x x-"3+0(x4) x~o,

we obtain

11m
·
cosh(sinx) -
tanh x
1- 2
- x
R
5 2
-
3x2
_
5 = lim 20~
x-+o __
x3
-21x4
__ + ox
( 4)

+ 0 (x 4)
= o.
_0

Example 6.63. Determine

a) lim (cos(xe X) -In(l - x) _ X)1/x3;
x-+o

b) !~ (cos(Sinx) + ~2) 1/(X (v'1+2X-1)); 2

c . (1
) 1x-+o
1m
sin x arctan x
-
tan x arcsin x
1).
262 CHAPTER 6.

Solutions.

a) In this case we use the following Maclaurin's formulas:

xeX = x + x 2 + O(X2), X -+ 0, hence

cos(xeX ) = 1 - x 2/2 - x 3 + o(x 3), X -+ 0,

cos(xeX ) -In(1 - x) = 1 + x - 2x 3/3 + o(x 3), x -+ °

and we obtain

lim(cos(xeX ) -In(l - x) - X)1/x3 = lim

X~O X~O
(1 _ 2x 3/3 + O(X3))1/x3 = e- 2/ 3 •

b) From

x2 5
sin x = x - x 3/6 + O(X4), cos (sin x) = 1 - "2 + 24 X4 + o(x 3), X -+ 0,

1 1
VI + 2x = 1 + x + o( x ), x2(Vl+2x-l) - X +0(x 3)'
3
x -+ 0,

we obtain
2) 1/(x (-/1+2x-1))
hm (COS(SlllX) + -
2 3 3
. . X . l/(x +o(x ))
x~o 2
= hm (1 + 5x 4/4! + O(X4))
x~o
= l.

c) In this case we have

I. (1
1m
x~osin x arctan x
-
1 ) I'
tan x arcsin x
= 1m
x~o
X4
X4
+ o( x 4 )
+ o( x 4) = 1
.

6.4 L'Hospital's Rule

6.4.1 Basic notions
Let c be a real number, or one of the symbols +CXl or -CXl.
The expression ~~:j is said to have undetermined form "0/0" at x = c, if

limf(x) = 0,
X'~c
limg(x) = 0.
x~c

The expression ~~:j is said to have undetermined form "CXl/CXl" at x = c if

limf(x)
x---+c
= CXl, limg(x)
x---+c
= CXl.
DERIVATNES 263

Theorem 6.64. L'Hospital's Rule

Suppose the functions f and 9 are differentiable at every point, except possibly at c in
an interval (a, b). If g'(x) i= 0 for x i= c, and if ~~;j has either the undetermined
o
form "-" or "-" at x
CXJ
= c' then
J' 0 CXJ

lim f(x) _ lim f'(x) (6.26)

g(x) - x---+c g'(x)'
x---+c

prom'd ed -
f'(x)
(-) h as a l"t
zmz as x approac h es c.
g' x

The statement of Theorem 6.64 remains true also if lim f'((x)) = CXJ, and then
x---+c g' x

. f(x)
hm -(-)
x---+c 9 X
= CXJ.

Moreover, the point c can be replaced with

6.4.2 Examples and Exercises

Example 6.65. Determine the following limits by using L 'Hospital's rule.

a) lim cos x + 3x - 1 · eX + e- x
b) 11m - 2 cos x.
. ,
x---+o 2x x---+O X sIn 2x

. In cos ax x a _ aX
c) hm 1 b' a, b i= 0;
x---+o n cos x
d) x-+a
lim
aX - aa
, a> 0, a i= 1;

e) hm
. (a + xr - aX
a> 0; f)
.
11m XX - X

x-+o x
0 ,
x---+l1n x - x +1
Solutions.
a) The functions f( x) = cos x + 3x - 1 and g( x) = 2x are differentiable in an
interval containing the point O. The given expression has the undetermined
o
form "-" when x 0 and
o --t
'
. cos x
11m + 3x - 1 l' - sin x
= 1m
+3 = /
3 2.
x---+o 2x x---+o 2

Clearly the conditions of the L'Hospital rule are fulfilled in the cases b)-f), and
therefore we have the following.
. eX + e- X - 2 cos x l' eX - e- X + 2 sin x
b ) hm = 1m -.---~--
x---+o x sin 2x x---+o sm 2x + 2x cos 2x
lim e X +e- x +2c.08X =4/4=1.
4 cos 2x - 4x sm2x
x---+o
264 CHAPTER 60

-asmax
0 In cos ax o cos ax _ 0 a sin ax 0cos bx
c ) 11m---- = hm bOb - hm
x->o In cos bx x->o - SIn x x->o b SIn bx COS ax 0

COS bx
ax
2 sIn
a - - - COS bx 0 a2
lim ax
x->o b2sin bx b2 0
- - 0cos ax
bx
x a - aX ax a- 1 - aX In a 1 -Ina
d) lim = lim - - - - -
x--+a aX - aa x--+a aX In a In a

x x exJn(a+x) (In(a + x) + _X_) - aXIna

(a+x) -a 1 a+x
e ) 11m = 1m
0 0

x->O 2 x x->o 2x

exJn(a+x) ((In(a + x) + _X_)2 + _1_ + a ) _ a Xln2 a

lim a+ x a+ x (a + x F 1
x->o 2 a
0
f) 11m XX - x 10 XX (In x + 1) - 1 10 xx+! (In x + 1) - x
= 1m = 1m -----'----,-------'---
x->1 In x - x + 1 x->1 1 x->1 1- x
- -1
x
xX+l(lnx + 1) (1 +..!:. + Inx) + xX-l
lim x = -20
x->1 -1
Example 6.66. Determine

a) 11m 3x2
0 + 2x - 2 b) 11m 21nx
-- b> 0;
0

x2 - 1
0

x->+oo ' x->+oo xb '

xb d) lim In(ln x)
c) lim -ax, b> 0, a E R;
x--++oo e x->+oo In x 0

Solutions. The expressions in this example have the undetermined form" 00", and
00
the conditions of the L'Hospital rule are fulfilled 0 So we have the followingo

a) lim 3x 2 + 2x - 2 = lim 6x + 2 = ~ = 3.
x->+oo x2- 1 x->+oo 2x 2
2
b) 0 -
11m 21n
- = 1m - - = 101m -=00
x 10 x 2
x->+oo xb x->+oo bX b- 1 x->+oo bx b
xb
c) Clearly, if a ::::; 0, then lim - ax = +000
x--++oo e
Let us assume that a > 00 Then we can find kEN such that k > band
applying the L'Hospital rule k-times we obtain
k
11m x
0 k!
11m --=00
0

x--++oo e ax x->+oo ake ax

DERIVATIVES 265

xb xk
Since 0 < -eax -< - ,
e ax
x > 0, using Exercise 4.31 it follows that

xb
lim - =
x-++oo e ax
o.
1
d) lim In(lnx) = lim xlnx - lim _1_ = o.
x-->+ooIn x x-->+oo1 In x
x-->+oo
x
Example 6.67. Determine
a) lim x( elfx - 1); b) lim sin x In cot x;
x---++oo x-->o+

c) lim x'" Inf3

x-->o+
(~),
X
0'., (3 > 0; d) x~~oo x (7r - 2 arcsin ( ~) ) .

Solutions.
-1
e l/x - 1 el / X (_)
a) lim x(e l / x - 1) = x---++oo
lim --1 lim x2 lim el / x = l.
x---++oo x-++oo -1 X-++<XJ

x x2
-1
. SIn x
· Slnx
. 1ncot x = 1·1m ---
In cot x t
lim - - - = o.
2
b) 11m lim co xsm x
x-->o+ - cos x
= x-->o+ cos 2 x
x-->O+ x-->O+
sinx sin 2 x
c) Let us find first for kEN the limit

lim x"'/k In
",-->0+
(~)
X
.

Applying the L 'Hospital rule, we get

lim x"'/k
1)
In (- = lim
--1 x ~ lim x"'/k = o. (6.27)
x-->o+ X x-->o+ 0'. -",/k-l x-->o+
-IX 0'.

Thus also
lim
x-->o+
x'" Ink (~)
X
= o.
Choosing kEN such that k > (3, it holds

Os;: x'" Ilnf3 xl s;: x'" lInk xl, x> 0,

which, in view of (6.27), implies finally

lim
x-->o+
x'" In f3 (~)
X
= o.
266 CHAPTER 6.

lim x ( , _ 2 ",,,m
d) "-+00 . () x ) _.' - 2
V x- ,. 1
1.
~+
- x-++oo
hm
.",,,,,
· (
.
x
JX2+T )
x
2X2

JX2+T - 2JX2+T 1

lim -2
x-++oo
J1 - 1 :;>+i
"' 1
x
2
+1 _ lim
"-+00
-2-
x'1+ 1
x'
~ 2.

x2
Example 6.68. Determine
a) lim (~ _ _1_) . b) lim (_1__ ~).
x-->o x 2 sin 2 x ' x-->o efC - 1 x
Solutions. In these two examples the undetermined forms "(Xl - (Xl" appear, so
firstly, we have to perform some transformations in order to obtain limits which can
be found by using the L'Hospital rule.
· (1
a) 11m - 2 - - -2 -
1) 1·
·2
sm x - x
= 1m ----,;,----
2

x-.O x sin x x-.o x 2 sin 2 x

. - sin 2 x + cos 2 x-I

= 11m -...."...----------:-:----::-
x-.o sin x + 4x sin x cos x + x 2 ( cos 2 X - sin 2 x)
2

. 2
_r - 2 sm x
- x~ sin 2 x + 4x sin x cos x + X2(coS 2 X - sin 2 x)

-2 1
= lim - - - - - - - - - - - ; ; - - - - - - - - -
x-.o x2 3
1 + 4-.- cos x + -.-2-(COS 2 X
X
- sin 2 x)
sm x SIn x

. (1
b) hm - - - - = lim
x--->O eX - 1
1)
x
x-ex +l
x-.o xe X - x
I-ex
= lim - - - - -
x-.o xe X + eX - 1
_eX
= lim = -1/2.
X-'O xe X + 2e X
Example 6.69. Determine
a) lim(1 + 2X)I/x; b) lim x 1 /(I-x).
X-'O x--+ 1 '

lim( cos x)l/x2 . . Cinx) l/x 2

c) d) hm - -
x--+o ' x-.o x

e)
. 2
hm(-arccosx)I/X;
X-+O 7r
f) hm . Cl +
x--->o e
x)l/xf 1x

g) lim
X-.+=
(~arctan
7r
x) X ; h) lim(1
x--->O
+ tanhx)l/x.
DERIVATNES 267

Solutions. In these cases we have the undetermined forms "1 =".

a) The function y = (1 + 2X)I/x can be written in a form
1 .
In y = -In(I
x
+ 2x), wherefrom It holds

2
limlny = lim In(I + 2x) = lim 1 + 2x = 2.
x->o x->O X x->o 1
Therefore we have

lim(I
x~o
+ 2x)1fx = exp (limln y) = e2 • x~o

b) Let us consider the function y = exp(1n u), where u = x 1 /(l-x). It holds

lim exp(1n
x->1
u) = exp (lim
x->1
In u) = exp (lim _I_In x) .
x->1 1 - x
So we obtain

( lnx
lim - -
x->1 1 - X
= x->1
lim..L = -1
-1
1)
-
:::} lim x 1 /{I-x) _ -1
x->O - e .

c) In this case we have

SIn X)
lim(cosx)l/x2 = exp (lim lncosx) = exp ( lim -~ = e- 1 / 2
x->o X->O x2 x->o 2x

sinx l/x 2 sin-

1n - x) ( x
-.--. x cos x -2 sin x )
d) lim ( - - )
X->O X
= exp ( x->o
lim
x2
x = exp lim sm x
x->o 2x
x

exp ( -11·1m xcosx-sinx)

. = exp (11·
- 1m -xsinx) = e -1/6 .
2 x->o x 2 sm x 2 x->o 2x sin x + x 2 cos X

e) lim
x->o
(~ arccos x)
1["
l/x = exp (lim
x->o
_In- ,-(.:. :.~X_a_r_c_co_s_x-,),- )
= exp (lim (_ 1 . 1 )) = e-2/1f.
x->o arccos x v'f=X2
. ((1 +
f) hm
x->o
X)1fx)I/X
e
= exp
(.
hm - In
x->o X
1 ((1 + e
X)I/X))

=
. In(l + x) - x
exp (hm
x->o x2
) = exp ( 1)
+hm
.
=
x->O
~-1
2x
e- 1/ 2 .
268 CHAPTER 6.

g) lim
X-++OO
(~arctan
7r
x) x = exp ( lim
X-++OO
xIn (~ arctan x))
7r

exp
r
( x-!~oo ~ . arct
1
1
an x )
= e
-2/7r
.
x2
1 1
h) lim(1
x-+o
+ tanhx)l/x = exp ( lim 1 + tanh x .
X-+O 1
~) -- e.

Example 6.70. Determine

a) lim (eX - I)X; b) lim xx;
X-+O+ X-+O+

c) lim x1/ln sinh x. d) lim (arcsinx)tanx.

X-+O+ ' X-+O+

Solutions. These examples are of the undetermined form "00".

a) For the function y = (eX - I)X we have In y = x In (eX - 1) . Hence
eX
· In(eX-l)
11m l' eX-I
lim Iny =lm--
X-+O+ X-+O+ 1 x-+o+-l
x x2

. _x 2 e x
hm - - -
. -(2xe + xe
hm
X 2 x)
=0.
X-+O+ eX - 1 X-+O+ eX
Therefore we have
lim (eX _ I)X lim In y = e = 1.
X-+O+ = e x-+O 0

b) From the equalities

1
· -1-
11m In x = l'1m -1
-; = l'1m ( -x ) = 0 ,
X-+O+ X-+O+ - X-+O+
x x2
we obtain
lim Iny 0
lim XX = ex-+ O = e = 1.
X-+O+

c) In this case we have

1
Inx . x lim xl/In sinh x
lim = hm --1- = 1, hence X-+O+ = e.
X-+O+ In sinh x X-+O+ _ _ cosh x
sinh x
DERIVATIVES 269

d) From
1 1
Iim In( arcsin x) =lm
l' arc~ . ~ 0
=,
X->O+ cot x X->O+ 1
- sin 2 x
we obtain lim (arcsinx)tanx = l.
X->O+

Example 6.71. Determine

a) lim Xl/x;
X~+OO
b) lim (3x 2
x.....,. +00
+3 X )*.

Solutions. These examples are of the undetermined form "00°".

a) From
Inx 1
lim - - = lim - = 0,
x->+oo X x->+oo X

we get lim xl/x

x---++oo
= eO = 1.
b) From
.
11m In(3x2 + 3 = l'1m 6x + 3x In 3 = 1n3
X
)

x->+oo X x->+oo 3x + 3x ' 2

we obtain lim (3x 2

x---++oo
+3 X )1fx = 3.

Example 6.72. Determine

3 . 1
2x - sinx x sm-
a) lim .; b)lim . 2 X.
x->+oo 2x + SIn x x->O SIn x

Solutions.

a) For x > 2, we have

2x - 1 2x - sin x 2x + 1 d 2x - 1 2x + 1
---< <--- an
·
11m --- = l'1m --- = 1
2x + 1 - 2x + sin x - 2x - 1 2x + 1
X---++OO x---++oo 2x - 1 '

hence
·
11m 2x - sin x 1
= .
2x + sin x
x---++oo

In this case we can not apply L'HospitaI's rule, because the following limit

·
11m (2x-sinx), l' 2-cosx
. = 1m ,
(2x + sm x)'
x---++oo x---++oo 2 + cos x

does not exist, namely there is no lim cos x.

x---++oo
270 CHAPTER 6.

b) Again, the following limit of the quotient of the derivatives

· f'(x)
11m d .
- (-) oes not eXIst,
x_o g' X

so we can not apply L'Hospital's rule. Still, we can calculate the limit:

3' 1 . 1
x SIn - x SIn-
lim . x = lim -._x_ = O.
x_o SIn 2 x x_O SIn 2 x

x2

6.5 Local extrema and monotonicity of functions

6.5.1 Basic notions
Theorem 6.73. Let f be a function that is continuous on a closed interval [a, b] and
differentiable on the open interval ( a, b).

• If f'(x) > 0 for all x E (a, b), then f is increasing on [a, b].
• If f'(x) <0 for all x E (a, b), then f is decreasing on [a, b].

For a proof of this theorem, see Example 6.49.

Theorem 6.74. Let f be a function that is continuous on a closed interval [a, b] and
differentiable on the open interval ( a, b).

• If f is increasing on [a, b], then f'(x) 20 for all x E (a, b),

• If f is decreasing on [a, b], then f'(x) ::; 0 for all x E (a, b).

Definition 6.75. A number c E (a,b) is a critical number (or: critical point)

of a function f: [a, b]- R if either f'(c) = 0 or f'(c) does not exist.

A necessary condition for the existence of extrema gives the following.

Theorem 6.76. If a function f : [a, b] _ R is continuous on [a, b] and has a local

extremum (maximum or minimum) at a number c E (a, b), then c is a critical number
of f.

Sufficient conditions for extrema are given in the following two statements.

Theorem 6.77. The first derivative test.

Suppose c is a critical number of a function f and (a, b) is an open interval containing
c. Suppose further that f is continuous on [a,b] and differentiable on (a,b), except
possibly at c.

• If f'(x) > 0, for x E (a, c) and f'(x) < 0 for x E (c, b), then f(c) is a local
maximum of f.
DERIVATNES 271

• If f'(x) < 0, for x E (a, c) and f'(x) > 0 for x E (c, b) then f(c) is a local
minimum of f.

• If either f'(x) < 0, or f'(x) > 0 for all E (a, b), except possibly in c, then f(c)
is not a local extremum of a function f.

Theorem 6.78. The second derivative test.

Suppose that f is a two times differentiable function on ( a, b), containing c and let
f'(c) = o.

• If f"(c) < 0, then f has a local maximum at c.

• If f"(c) > 0, then f has a local minimum at c.

Theorem 6.79. Assume a function f has at the point c all derivatives up to the
order n > 2, and it holds

f'(c) = f"(c) = ... = f{n-l}(c) = 0, but f{n}(c) #- o.

• If n is an even number, then c is an extremum point.

• If n is an odd number, then c is not an extremum point.

6.5.2 Examples and exercises

Example 6.80. Determine the critical numbers of the function

f(x) = (x + 5?{lx - 4, x E R.

Solution. The first derivative of f for x#-4 is

f'(x) 2(x + 5)(x - 4)1/3 + (x + 5)2-.-1

(x + 5)2 + 6(x + 5)(x - 4) (x + 5)(7x - 19)

3(x - 4)2/3 3(x - 4)2/3

Clearly we see that f'(x) = 0 for x = -5 and x = 19/7. Also the first derivative
does not exist for x = 4, hence the given function has three critical numbers,
namely Xl = -5, X2 = 19/7, X3 = 4.
Example 6.81. Determine the local extrema of the following functions and the in-
tervals on which they are monotonically increasing or decreasing.
x x
a) f(x) = x 2 -6x+16' xER; b) f(x) = "3 - {/X, x E R;

c) f(x) = X 2 / 3 (X 2 - 8), x E R; d) f (x) = x - sin x, x E R;

e) f(x) = sin 2 x, x E R; f) f(x)=x+lnx, x>o.

272 CHAPTER 6.

Solutions.
16 - x 2
a) From f'(x) = ( it follows that f'(x) = 0 for Xl = 4 and
X - 6x + 16
2 )2'

X2 = -4, and the critical points are A(4, 1/2), B( -4, -1/19). Using Theorem

6.78, from

11/( X ) = 2x 3 - 96x + 192 " 11/(4) < 0 1"(-4) > 0,

(x 2 -6x+16)3

it follows that the function 1 has a maximum at the point A, and it has a
minimum at B. Further on, we have

f'(x) > 0 {:::=:? 16 - x 2 > 0,

(resp. !,(x) < 0 {:::=:? 16 - x 2 < 0),

hence

• the function 1 increases for x E (-4,4);

• the function 1 decreases for x E (-00, -4) U (4, +00 ).

b) In this case we have

f'(x) ~
3
_ ~X-2/3 = ~
3 3
(1 __1_)
X 2/ 3

X 2/ 3 - 1 X2 - 1
3X 2 / 3 3X 2 / 3( X 4 / 3 + X 2 / 3 + 1) .

From f'(x) = 0, for Xl = -1 and X2 = 1 we obtain that the function has a

maximum at the point A( -1,2/3), and a minimum at B(l, -2/3).

• The function decreases for x 2 - 1 < 0, i.e., for x E (-1,1).

• The function 1 increases for x E (-00, -1) U (1, +00).

Note that x = 0 is a critical point of 1, but it has no extremum there.

c) From
!,(x) = 8(x 2 - 2) xi- 0,
3X I / 3 '

it follows that the critical points of 1 are Xl = V2" X2 = -V2, and X3 = o.

From the second derivative of 1 it follows that 1 has local minimums at Xl

and at X2.
From the sign of f' it follows that 1 has a local maximum at the point X3 = 0,
even though the first derivative does not exist at o.
Thus we obtain
DERIVATIVES 273

• the function 1 increases when f'(x) > 0, i.e., for

x E (-v'2, 0) U (v'2, +00) ;

• the function 1 decreases for x E (-00, -v'2) U (0, v'2) .
d) Since f'(x) = 1 - cos x, the function 1 increases on R \ {2k7r1 k E Z}. Also,
from
1"(2k7r) = sin(2k7r) = 0, and 1111(2k7r) = cos(2k7r) = 1 1= 0,
it follows that the function 1 increases on the whole R. The function has no
extrema points.

e) From f'(x) = 2sinxcosx = sin2x, it follows that 1'(x) = 0 for Xk = k27r,

k E Z. Since f"(x) = 2cos(2x) and f"(Xk) = 2(-1)\ we obtain that at
the points X2j, the function 1 has minimums, and at the points X2j+1l the
function 1 has maximums, for j E Z. Further on,

• the function 1 increases if x E U (k7r, (2k + 1)7r) ;

kEZ 2

• the function 1 decreases if x E U ((2k + 1)7r , (k + 1)7r) .

kEZ 2

1 1 +x . . .
f) From f'(x) = 1 + - = --, It follows that the functlOn has no extrema pomts.
x x
The function 1 increases on its whole domain, i.e., on (0, +00).

Example 6.82. Examine the extrema point of the following functions.

a) 1(x) = cosh x + cosx; b) 1(x) = x 3 + X4.

Solutions.

a) The function is differentiable for every x E R. We also have f'(x) = sinhx-sinx

and it holds
f'(x)=O for x=o.
Thus

1"(x) = cosh x-cos x, 1"(0) = 0, and 1"'(x) = sinhx+sinx, 1"'(0) = 0,

but
1(4)(X) = cosh x + cos x, 1(4)(0) = 21=0.
This means that the given function has an extremum point at x = O. Since
1(4)(0) > 0, at the point x = 0 1 has a minimum.
274 CHAPTER 6.

b) The critical points are Xl = °and X2 = -3/4, since l' (x) = 3x 2 + 4x 3 .

From

f"(x) = 6x + 12x2, 1"(0) = 0, and 1"'(x) = 6 + 24x, flll(O) = 6,

it follows that this function has no extremum at Xl = 0. Actually, A(O,O) is a

point of inflection, see Subsection 6.6.1.
The function has a minimum at X2.

Example 6.83. Examine the monotonicity and find the extrema points for the func-
tion y = f( x) given parametrically by

_ t3 t 3 - 2t2
X -
t2-+-
1' Y= t2 + l ' t E R.

Solution. From

t 2 (t 2 +3) , t(t-1)(t2+t+4)
'- ----,-
x - (t2+1)2
t
and
Yt= (t 2 + 1)2

and the derivative of a parametric function (see Subsection 6.1.1), we have

, =
Yx
y~
x~
= (t - 1) (t 2 + t
t(t2+3)
+ 4)
,
t ..J.
r·
°
Thus it follows that y~ = 0 for t = 1, i.e., for x = 1/2. In fact the function f has
°
two critical points. Namely, Xl = is obtained from t = 0, and the other, X2 = 1/2,
from t = 1.

• If t < 0, then
X E (-00,0).
X < 0, and from y~ > ° it follows that the function increases for

°
• If t E (0,1), then x E (0,1/2), and from y~ < it follows that the function
decreases for x E (0,1/2). Therefore, at the point Xl = the function has a
maXImum.
°
• Ift E (1, +(0), then X E (1/2, +(0) and from y~ > it follows that the function °
increases for x E (1/2, +(0). Therefore at the point x = 1/2 the function has
a ffilnImum.

Example 6.84. Let us consider the following functions.

f(x) ={
e- I / X2 , xi 0; g(x) = {
xe- I / X2 xi 0;
0, x = 0, 0, x = 0.
Prove the following.

a) f(n)(o) = g(n)(o) = 0.
DERIVATIVES 275

b) The function f has a minimum at x = 0, but the function 9 has no extrema at

x = o.

Solutions.

a) It is clear that

!,(x) = ~e-l/x2 f"(x) = (~

x 6 _~)
X4 e _1/x , ... , f(n)(x) = R 3n
2
(~) e- 1 / x2
,
x3 '

where R 3n (~) is a polynomial of ~ and of order 3n. Applying the L'Hospital

rule k-times, we get
-1/x2
lim ~k = 0, kEN.
x-+o x
So we obtain that 1'(0) = o. Analogously, we can show that g'(O) = o.
b) The function f has a minimum at x = 0, because it holds f(x) > 0, x i= o. The
function 9 has no extrema at x = 0, because we have

g( x) > 0, x > 0, and g( x) < 0, x < O.

Example 6.85. Determine the biggest term of the sequence given by

a) fn = vn
I "1 (\I'\,.. ; b) fn =
n2
., ..... nn;

_n
12
I" _,
C
)
In - en d) fn = y'n.

Solutions.

a) The function

-"fX ., 1995 - x
f(x) = x + 1995 wIth f (x) = 2-"fX(x + 1995)2

has its maximum at the point x = 1995, because f'(x) > 0 for x < 1995
(the function is increasing), and f'(x) < 0, for x > 1995 (the function is
decreasing). So n=1995.
2
b) The function f(x) = x 3 : 200 has a critical point at x = ?'400 ~ 7.37. Since
f(7) ~ 9.02.10- 2 > f(8) ~ 8.99.10- 2 , it follows that n = 7.
X 12 12xll _ X 12
c) From f(x) = - , f'(x) = , it follows that n = 12.
eX eX
276 CHAPTER 6.

d) In this case we consider the function f(x) = xl/x, whose derivative is

f'(x) = X(I/X)-2(1 -lnx).

Thus the maximum of this function is for x = e. Since f2 = v'2 ~ 1.41, f3 =

"y3 ~ 1.44, it follows that n = 3.

Example 6.86. If a function <jJ is a monotonically increasing differentiable function

and 1f'(x)1 :::; <jJ'(x) for x;::: Xo, then it holds

If(x) - f(xo)1 :::; <jJ(x) - <jJ(xo) for x;::: Xo·

Solution. The functions f and <jJ satisfy the conditions of the Cauchy theorem 6.50,
so we can write
f(X)-f(x o)I=If'(e)I<I, eE(xo,x).
I<jJ(x) - <jJ(xo) <jJ'W-
Since the function <jJ is increasing we obtain

If(x) - f(xo)1 :::; 1<jJ(x) - <jJ(xo) I = <jJ(x) - <jJ(xo) for x;::: Xo·

Example 6.87. If

(i) the functions f and 9 are n-times differentiable;

(ii) f(k)(xo) = g(k)(xo) for k = 0,1, ... ,n -1;

(iii) pn)(x) > g(n)(x) for x> xo,

then prove f(x) > g(x), x> Xo.

Solutions. The function given by <jJ(n-I)(x) = f(n-l)(x) - g(n-l)(x) satisfies the

conditions
<jJ(n-I)(XO) = °
and <jJ(n)(x) > 0, for x> Xo.
Applying the Lagrange theorem to the function <jJ(n-I)(x) on the interval [xo, x], we
get
<jJ(n-I)(X) - <jJ(n-I)(xo) = <jJ(n)(o(x - xo), e
E (xo,x),

hence <jJ(n-l}(x) > 0. This implies

f(n-I)(X) > g(n-l)(x) for x> Xo.

Similarly we obtain
f(n-2)(X) > g(n-2)(x) for x> Xo,
and continuing this procedure finally we get

f(x) > g(x) for x> Xo.

DERIVATIVES 277

Example 6.88. Prove the following inequalities:

x2 x
a) x - 2 < In(l + x) < x, x> 0; b) In(1 + x) > --1'
x+
x> 0;

c) x a - 1 > a(x - 1), a> 2, x> 1; d) \IX - \fO, < -\Ix - a, n> 1, x> a > O.

Solutions.
x2
a) Let us denote by f(x) = x - 2' g(x) = In(1 + x) and h(x) = x. It holds that
f(O) = g(O) = h(O) = 0 and

j'(x) = 1 - x < g'(x) = _1_ < h'(x) = 1, x> o.

l+x
From Example 6.87 we get f(x) < g(x) < h(x), x> 0, l.e.,

x2
X - 2 < In(l + x) < x, x> O.

x
b) In this case we denote f (x) In(1 + x) and g(x) Then from
x+l
f(O) = g(O) = 0 and
1 , 1
j'(x) = - - > 9 (x) = ~ \~,
l+x
we get
. x
f(x) > g(x), l.e., In(l + x) > - -
x+l
for x> o.
c) Denoting by f(x) = x a - 1 and g(x) = a(x - 1), in view of f(l) = 0 = g(l) and

j'(x) = ax a - 1 > a = g'(x), x> 1,

we obtain the given inequality.

d) If we take f (x) = \IX - \fO, and g( x) = -\Ix - a, then this functions satisfy the
conditions of Example 6.87 and the given inequality follows.

Example 6.89. Prove the following inequalities.

x2
a) cosx2::1-2, xER; b) arctan x :S x, x 2:: 0;

2 .
c) -x < smx,
7r
0< X < 2;
7r
d) sinx + tanx > 2x, 0< X
7r
< 2.

Solutions.
278 CHAPTER 6.

x2
a) Taking f(x) = cosx, g(x) = 1- 2' we have f(O) = g(O) = 1 and f'(x) =
- sin x, g'( x) = -x. The function h( x) = x - sin x was considered in Example
6.81 d). Since this function is a monotonically increasing one and h(O) = 0, it
holds
x - sin x > 0, x> 0, i.e., x > sin x, x> 0.
From f'(x) > g'(x), x> 0, it follows that

x2
f(x) > g(x), x> 0, i.e., cos x > 1 - 2' x> 0.

The functions f and 9 are even, which means that the last inequality holds
for x < 0, also.

. 1
b) Puttmg f(x) = x - arctan x, x 2 0, we get f'(x) = 1 - 1 + x 2 > 0, hence
f(x) 2 f(O) for x 2 0, which implies the given inequality.

c) It is enough to show that the function f given by

f(x)=sinx- 2;, XE[O,%],

has exactly two zeros on [0, %] . Let us suppose that f has a zero E(0, %) , Xl

besides the zeros Xo = ° and X2 = %. Then from Rolle's theorem it follows

that there exist at least two points X3 E (0, xt} and X4 E (Xb ~) , such that

f'(X3) = f'(X4) = 0. But the equation f'(x) = 0, i.e., cos x - ~7r = 0, has only

one solution on the interval (0, %). So the function f is of the same sign on
this interval. Since it holds (for instance)

f (~)
3 = V3 2 > 0,
2 -"3

we get sm 2x £or x E (7r)

. x > -:;- 0, '2 .

d) The functions f(x) = sinx + tan x and g(x) = 2x satisfy the conditions of the
Example 6.87, since f(O) = g(O) = 0, f'(0) = 2 = g'(O), and

f "( x ) - ' 2 sin x _ sin x(2 - cos 3 x)

- - sm x + - 3- -
cos X 3
>
cos X
°-_"()
9 x, x E(0, %) .

Therefore it holds f(x) > g(x), x E(0,%).

DERIVATIVES 279

6.6 Concavity
6.6.1 Basic notions
Definition 6.90.

• A function f : (a, b) -+ R is concave upward on ( a, b) if for every pair

Xl,X2 E (a, b) and for every a E (0,1) it holds

f(axl + (1 - a)x2) < af(xl) + (1 - a)f(x2)'

• A function f : (a, b) -+ R is concave downward on (a, b) if for every pair

XI, X2 E (a, b) and for every a E (0, 1) it holds

f(axl + (1 - a)x2) > af(xl) + (1 - a)f(x2)'

Theorem 6.91. If a function f is two times differentiable on an open interval (a, b),
then f is

• concave upward if for all X E (a, b) it holds f"(x) > 0;

• concave downward if for all x E (a, b) it holds f"(x) < 0.

Theorem 6.92. Let f be a differentiable function on an interval (a, b).

Then the function f is concave upward iff the graph of f is above the tangent
line through every point of an interval (a, b);
The function f is concave downward iff the graph of f is below the tangent line
through every point of an interval (a, b).

Definition 6.93. Let f be a continuous function on an interval (a,b), differentiable

at the point c E (a, b). If the function f changes its concavity at c, then c is called
inflection point of f.
Then the point A( c, f( c)) is called inflection point of the graph of f.

A necessary condition for the existence of the point of inflection gives the follow-
ing theorem.

Theorem 6.94. If c is a point of inflection of a function f, then either

f"(c) = 0, or f"(c) does not exist.

Sufficient conditions for the existence of the point of inflection give the following
two theorems.

Theorem 6.95. Let a function f be differentiable at a point c, and two times dif-
ferentiable on an open interval (a, b) containing c, except possibly at c. Then c a
point of inflection of f, if one of the following assumptions holds:
280 CHAPTER 6.

a) f"(x) > 0 if a < x < c and f"(x) < 0 if c < x < b; or

b) f"(x) < 0 if a < x < c and f"(x) > 0 if c < x < b.

Theorem 6.96. If a function f has at a point c all derivatives up to the order n,

n > 2, and it holds

J"(c) = flll(c) = ... = f(n-l}(c) = 0, while f(n}(c) i- 0,

then if n is an odd number, c is a point of inflection, while if n is an even number,
c is not a point of inflection.

6.6.2 Examples and exercises

Example 6.97. Find the points of inflection and the intervals of concavity upward
and downward of the following functions.

a) f(x)=x 4 -6x 2 +5x+3, xER; b) f(x) = 15xXVX

- 51 ' x> 0;

c) f(x) = 3 exp( qx), x E R; d) f(x) = xsin(lnx), x> O.

Solutions.

a) From f"(x) = 12(x 2 - 1), it follows that J"(x) > 0 for x E (-00, -1) U(I, +00),
where the function f is concave upward, and for x E (-1, 1) the function is
concave downward.
The points A(I, 3), B( -1, -7) are the points of inflection, because f"(x) = 0,
for Xl,2 = ±1 and from f"'(x) = 24x, it follows that f"'(x) i- 0, for x = ±l.

b) The function f is defined on the interval (0, +00) and can be written as
X -1
f(x) = 5. { xvx' x ~ 1;
I-x
xvx' 0 < x < 1,

and is differentiable at every point x > 0 except at x = 1. The second derivative

has the form
x-5

~~
x> 1;
x3VX'
f"(x) {
5-x
X3VX' °< x < 1,
and it holds 1"(5) = 0, and

J"(x) > 0, x E (0,1), J"(x) < 0, x E (1,5), J"(x) > 0, x E (5,+00).

DERIVATIVES 281

Therefore, on the intervals (0,1) and (5, +=) the function is concave up-
ward and on the interval (1,5) the function is concave downward. The point
A(5, 4/0) is a point of inflection, but the point B(I, 0) is not, because the
function f has no first derivative at x = 1.
e) In this case we have

f'(x) = exp( ijX) f"(x) = exp( ijX) . Vx( Vx - 2)

if;} , ... 0

for x #- 0. The function f satisfies xlim

..... o
f'(x) = +=, but

lim f"(x) = += and lim f"(x) = -=.

X"'" 0- X"'" 0+

Also, we have f"(x) = 0, for x = 8. From

f"(x) > 0, x < 0, f"(x) < 0, 0< x < 8, f"(x) > 0, x> 8,
we get that the function f has two points of inflection, A(0,3) and B(8, 3e2 ),
and it is concave upward for x E (-=,0) U(8, +=) and concave downward
for x E (0,8).

d) The domain of this function is the interval (0, +=) and

f"(x) = (sin(lnx) + cos(1nx))' = .!.(cos(lnx)-sin(1nx)) = V2 cos (lnx +~).

x x 4
The function is concave upward (i.e., f"(x) > 0) if
3~ ~
- 4 + 2h <lnx< 4"+2h, k=0,±I,±2,... ,
or
-3~ ~
exp (-4- + 2h) < x < exp (4 + 2h),
and is concave downward if
~ 5~
4" +2h < lnx < 4 +2h, k = 0,±1,±2, ... ,

or
~ 5~
exp(4" + 2h) < x < exp( 4 + 2h).
The inflection points are

Ak
_
-
(~
exp (4"+2h),
exp (~+ 2h))
V2 ' k=0,±1,±2,... ,

Bk = (exp e: + 2h), -exp (¥V2 +2h)) ' k = 0,±1,±2, ....

282 CHAPTER 6.

Example 6.98. Determine the inflection points of the function

. 1
x = 3 + cot t, y = -2smt + -.-, 0<t<1r.
smt
Solution. From
1 cost(2sin 2 t + 1)
X~ = - sin 2 t' y~ =
sin 2 t
and I
I Yt " (')' 1
Yx = x~' Yxx Yx t· x"
t

we obtain
y~ = cos t(2 sin 2 t + 1), (y~)~ = 3 cos 2t sin t,
and finally
Y:x = -3 sin 3 t cos 2t.

Therefore Y:x = 0
for t = ~, and t = 3:. Since

• Y;x < 0 for t E (0, ~), the function is concave downward there;

7r 37r) the function is concave upward there;

• Y;x > 0 for tE(4'4'

• Y;x < 0 for t E (3: ,7r), the function is concave downward there,

it follows that the points t = i and t = 3;, or A( 4,0) and B(2, 0), respectively, are
the points of inflection for the function y = y(x).

Example 6.99. Prove that if f"(x) > 0 for all x E (a, b), then f is concave upward
on (a,b). I.e., prove then that for every pair Xl,X2 E (a,b) and arbitrary numbers
0:1, 0:2 such that
0:1 > 0, 0:2 > 0, 0:1 + 0:2 = 1 (6.28)
the following inequality holds

f(O:lXl + 0:2 X2) < o:d(xd + o:d(X2). (6.29)

(See Theorem 6.91')

Solution. Let us suppose that 1"( x) > 0 for all x E (a, b) and let the arbitrary
numbers 0:1> 0:2, satisfy the conditions given by relation (6.28). If Xl, X2 belong to
the interval (a, b), and Xl < X2, then O:lXl + 0:2X2 E (a, b) also. Using Lagrange's
theorem we obtain

f(O:lxl + 0:2 X2) - f(xd = f'(6)0:2(X2 - Xl),

f(X2) - f(O:lXl + Q:2 X2) = f'(6)0:1(X2 - xd,

DERIVATIVES 283

where Xl < 6 < alxl + a2X2, and alxl + a2x2 < 6 < X2. Multiplying the last two
equalities with a!, a2 respectively, we get

ad(alXl + a2x2) - arJ(xl) = 1'(6)ala2(x2 - xd,

a2!(X2) - a2!(alxl + a2 x2) = 1'(e2)ala2(x2 - xd,

wherefrom

arJ(xd + a2!(x2) = (al + (2)f(alxl + a2 x2) + ala2(x2 - Xl)(1'(e2) - 1'(6))·

The function f has a second derivative on the interval (a, b) so we can apply La-
grange's theorem again, giving us

1'(e2) - 1'(ed = (e2 - el)J"(e3), e3 E (el, 6)·

Therefore, we can write

arJ(xd + a2!(x2) = f(alxl + a2 x2) + ala2(x2 - Xd(e2 - 6)1"(6)).

From the last relation, in view of f"(x) > 0, follows relation (6.29).
Example 6.100. Prove the following inequalities.

a)
an+b
...
n> (a+b)n
-2- , a > 0, b > 0, a i- b, n E N;

b) e X ;eY >e(X+Y)/2, xi-Yo

Solutions. If al = a2 = ~, then from Definition 6.90 we obtain the following

inequalities for Xl, X2 E ( a, b) :
f(Xl)+f(X2) >fCI;X2). (6.30)

for the concave upward function on the interval (a, b), and

f(XI) + f(X2) < f (Xl; X2) , (6.31)

for the concave downward function on the interval ( a, b).

Since the functions

a) f(x) = x n , x> 0, n E N; b) f (x) = eX, X E R,

are concave upward (J"(x) > 0, for X in their domains), relation (6.30) holds.
284 CHAPTER 6.

Example 6.101. If f is a concave upward function on the interval (a, b) and the
points Xl, X2, ... , Xn , belong to (a, b) and all a2, ... , an are nonnegative numbers such
that al + a2 + ... + an = 1, then the so-called Iensen's inequality holds:
f(alXI + ... + anxn) ::; ad(xd + ... + anf(x n ). (6.32)
Prove.
Solution. If n = 2, then we have inequality (6.29). In this relation we have equality
if one of al or a2 is equal to zero. If n > 2 then we apply mathematical induction.
.
We can denote by j3 = a2 + ... + an > and It holds
a2
°
+ ... +
'a n
= 1. /i 73
Let us suppose that
a2
/i an )
f( X2 + ... + 73 xn ::;
a2 ( )
f X2 + ... +
an
7i
f(xn). 73
Then we have

f (alxl + ... + anxn) = f (aIXI + j3 (~ X2 + ... + ~ xn) )

::; ad(xd + j3 ( af(x2 /i 73

2 ) + ... + an f(xn) ) ::; ad(xd + ... + anf(xn).

Example 6.102. Prove

a) alln Xl + ... + an In Xn ::; In( alxl + ... + anxn),

Xi ::::: 0, ai::::: 0, i = 1,2, ... n, I: a i=l;

i=l

Xl + ... + Xn .
b) ylXI ... Xn ::; , Xi::::: 0, z = 1, ... , n.
n
Solutions.
a) The function f(x) = lnx is concave downward on (0,+00) and for this function
it holds
f(alXI + ... + anxn) ::::: ad(xd + ... + anf(xn). (6.33)
Therefore we can write
alln(xd + ... + an In(x n ) ::; In(alxl + ... + anxn). (6.34)

b) From relation (6.34) we obtain

Xr' ... x~n ::; alxl + ... + anx n ,
wherefrom for al = ... = an = .!.n we obtain
Xl + ... + Xn
.
ylXI···Xn::; , Xi::::: 0, z = 1, ... ,n.
n
The last relation is the well known connection between the arithmetic and the
geometric mean, see Example 1.33.
Chapter 7

Graphs of functions

Example 7.1. Determine and sketch the graph of the function

f(x) = (x + 2)2(X - 1)3.

Solution. The graph of f is sketched in Figure 7.1.

:x:

Fig. 7.1. f(x) = (x + 2)2(x - 1)3

• The domain of the given function is R.

• This is neither an odd nor an even function.

• The zeros of the function f are found from f( x) = 0, hence

(x + 2)2(X - 1)3 =0 {=} x E {I, -2}.

• The first derivative of f is the function

f'(x) = (x + 2)(x - 1?(5x + 4), x E R,

. 4
wIth zeros Xl = 1, X2 = -2 and X3 = -S.

285
286 CHAPTER 7.

- The critical points of fare A(l, 0), B( -2,0), C (-0.8, -8.40).

- The function f is increasing when 1'(x) > 0, namely when
4
(x + 2)(5x + 4) > 0 <¢=} x E (-00, -2) U( -5' +00).

- The function f is decreasing when

(x + 2)(5x + 4) < 0, i.e., for x E ( -2, -~) .

• The second derivative of the function f is
f"(x) = 2(x - 1)(10x2 + 16x + 1),
hence 1"(x) = 0 for Xl = 1, X4 ~ -0.07 and Xs ~ -1.53.

- Since 1"( -2) = -54 < 0, the critical point B( -2,0) is a maximum.
- From 1"( -0.8) > 0, it follows that the function f has a minimum at the
point C (-0.8, -8.40).
Let us remark that the first derivative does not change its sign at x = 1,
and though it holds 1'(1) = 0, it has no extremum at x = 1.
- The function f is concave upward if 1"(x) = 2(x -1 )(10x2 + 16x + 1) > 0,
i.e., for x E (xs,x4)U(1,+00).
- The function f is concave downward if 1" (x) = 2( x-1)( 10x 2 + 16x + 1) <
0, i.e., for x E (-00, xs) U(X4, 1).
- From the previous conclusions it follows that the function f has three
points of inflection A(l,O) D(x4,f(x4)) E(xs,f(xs)). It holds f(X4) ~
-4.56 and f(xs) ~ -3.58.

• The given function has no line asymptotes.

Example 7.2. Determine and sketch the graph of the function

f(x) = x+1.
x2 +1
Solution. The graph of f is sketched in Figure 7.2.

• The domain of the given function is R.

• This is neither an odd nor an even function.

• The function f is equal to zero for x = -1.

• The first derivative of f is the function

_x 2 - 2x + 1
f'(x) = (x2 + 1)2 .
GRAPHS OF FUNCTIONS 287

The zeros of the first derivative are Xl = -1 + V2, X2 = -1 - V2, so

the critical points of fare

A ( -1 + V2, V2+1)
2 and B ( -1 - V2, - V2-1)
2 .

- The function f is increasing for _X2 - 2x + 1 > 0, i.e., for

x E (-1- V2,-1 + V2).
- The function f is decreasing for _x 2 - 2x + 1 < 0, i.e., for
x E (-00,-1- V2)U(-l + V2,+(0) .
• The second derivative of the function f is
f"(x) = 2x 3 +6x 2 -6x-2 =2(x-1)(x 2 +4x+1).
(x2 + 1)3 (x 2 + 1)3

- Since 1"( -1 + V2) < 0, the function has a maximum at the critical point

A ( -1 + V2, V22+ 1) .

1" (-1 - V2) > °the function has a minimum at the critical point
1)
- Since

B ( -1- V2, V22- .

- The function f is concave upward if (x -1)(x 2 +4x + 1) > 0, and this is

satisfied for x E (-2 - V3, -2 + V3) U(l, +(0).
- The function f is concave downward if 1"(x) < 0, i.e.,
x E (-00, -2 - V3) U( -2 + V3, 1).
- From the previous conclusions it follows that the function f has three
points of inflection, namely

C(l,l), D (-2 + V3, 1 +4V3) and E (-2 - V3, 1 -4 V3) .

• Asymptotes.

- The function has no vertical asymptotes.

The function has a horizontal asymptote y = °both when x ----t +00 and
x ----t -00, because

lim f(x) = lim x2 + 1 = 0.

x---+±oo x---+±oo x +1
288 CHAPTER 7.

- The function has no slanted asymptote.

iY
A

E B ·1 10 1

x+1
Fig. 7.2. f(x) = x2 +1
Example 7.3. Determine and sketch the graph of the function
x-5
f(x) = x2 - 9·

Solution. The graph of f is sketched in Figure 7.3.

• The domain of the given function is (-00, -3) U (-3,3) U (3, +00).

• This is neither an odd nor an even function.

• The function f is equal to zero for x = 5.

• The first derivative of f is the function

f'(x) = _x 2 + lOx - 9
(x 2 - 9)2
- The zeros of the first derivative are x = 1 and x = 9. The critical points
of fare A(l, 1/2) and B(9, 1/18).
- The function f is increasing for f'(x) > 0, i.e., for x E (1,3) U (3,9).
- The function f is decreasing for x E (-00, -3) U (-3,1) U (9, +00).

• The second derivative of the function f is

f"(x) = 2(x 3 - 15x 2 + 27x - 45)
(X2 - 9)3

- Since 1"(1)> 0, the function has a minimum at the critical point A(l, 1/2).
- Since 1"(9) < 0, the function has a maximum at the critical point
B(9,1/18).
- In this case we can not order directly the zeros of the numerator of I" (x).
From 1"(13) < 0 and 1"(14) > 0 it follows that for some Xo E (13,14) it
holds I"(xo) = O.
GRAPHS OF FUNCTIONS 289

- The function f is concave upward if

rex) > 0 ~ x E (-3,3) U (xo,+oo),

because

g(x) = 2(x 3 - 15x 2 + 27x - 45) < 0 for x < Xo

and x2-9<O for XE(-3,3).Also,wehave

g(x»O for x>xo, and x2-9>O for xE(xo,+OO).

- The function is concave downward for x E (-00, -3) U (3, xo).

• Asymptotes.

- The function has two vertical asymptotes x = -3 and x = 3, SInce

x -5 lim x-5
lim - - - - -00·
x ..... -3- x2 - 9 - , x ..... -3+ x2 _ 9 = +00;
x-5 .
11m x - 5
lim - - - = +00; x2. _-9 = -00.
x ..... 3+
x2 - 9
x .....3-

Function has a horizontal asymptote y 0 both when x ~ +00 and

x ~ -00, because

lim f(x) = lim x2- 5 = O.

x ..... ±oo x ..... ±oo X - 9

- The function has no slanted asymptote.

•
";I
II y

,
B
x

x-5
Fig. 7.3. f(x) = x 2 -9 Fig. 7.4. f(x) = xv'x+3
290 CHAPTER 7.

Example 7.4. Determine and sketch the graph of the function

f(x) = xVx+3.
Solution. The graph of f is sketched in Figure 7.4.

• The domain of the given function is [-3, +00).

• The zeros of the function f are Xl = 0, X2 = -3.

• The first derivative of f is the function
, X ~ 3x+6
f(x)= Vx+3+ vx + 3 = Vx+3'
2 x+3 2 x+3
- The first derivative has a zero for 3x + 6 = 0, i.e., X = -2, so the critical
point of f is A( -2, -2).
The function f is increasing for 3x + 6 > 0, i.e., for x > -2, because
2Vx+3 > 0 for every x from the domain.
The function f is decreasing for -3 < x < -2.
• The second derivative of the function f is
f"(x) = 3(x + 4) .
4(x + 3)3/2
- From 1"( -2) > 0 it follows that A( -2, -2) is a minimum for the function.
- Since fl/(x) > 0 for every x from domain, the function is concave upward.

• There are no line asymptotes.

Example 7.5. Determine and sketch the graph of the function

x
f(x) = {!x2 _ l'

Solution. The graph of f is sketched in Figure 7.5.

• The domain of the given function is (-00, -1) U( -1,1) U(l, +00).

• This is an odd function, because

-~
f(-x) = . = -f(x).
f/(-x)2 -1

• The zero of the function f is x = o.

• The first derivative of f is the function
x2 - 3
f'(x) = 3f/(x 2 _ 1)4
GRAPHS OF FUNCTIONS 291

- The zeros of the first derivative are XlV3 and X2 = -V3. The critical
=
. of f are A
pomts (V3)
V3'..y2 , B ( -V3, -..y2
V3) .
- The function f is increasing for X E (-00, -V3) U( V3, +00).
- The function f is decreasing for X E (-V3, -1) U( -1,1) U(I, V3).

• The second derivative of the function f is

f"(x)= 2x(9-x 2 )
9~(X2 - IF'

hence f"(x) = °for X3 = 0, X4 = 3 and X5 = -3.

- Since 1"( V3) > 0, the function has a minimum at A ( V3, ~) .

- Since 1"( -V3) < 0, the function has a maximum at B ( -V3, - ~) .

- The function f is concave upward if x E (-00, -3) U( -1,0) U(I, 3).
- The function is concave downward for x E (-3, -1) U(O, 1) U(3, +00).

- The points of inflection are 0(0,0), C(3,3/2) and D( -3, -3/2).

• Asymptotes.

The function has two vertical asymptotes, namely x 1 and x - I ,

because

x
lim f(x)
x-+-l-
= x-+-I-
lim 3~
'\ X
= -00, lim
x .... -l+
x
31"::21
yx- - 1 = +00 '

x = -00, .
11m x
lim 3~
x-+l- V X - - l x .... l+ V'x2 _ 1 = +00.

- The function has no slanted asymptote, because

x
·
11m -
x-++oo
- --
f(x)
x
1·1m ~ --
x-++oo x
° .
292 CHAPTER 7.

L
'31

"IL
";'
~ 31
II'
,,
II'
(a=2) ,
"

I
: A
I _------

,
... :.--~
"
--i_iS
:.,.....:~.)",

x
Fig. 7.5. f(x) = \fx2 _ 1 Fig. 7.6. f(x) = IT
V;-=-;;:
Example 7.6. Determine and sketch the graph of the function

f(x) =
fx3 ,
V;-=-;;: a> O.

Solution. The graph of f is sketched in Figure 7.6, for a = 2.

• The domain of the given function is (-=,0] U( a, +=), because

f(x)= [1;
-=Ixl' ~
3
x-a
-, x-a
hence x(x-a»OVx=O.

• The function f is equal to zero for x = 0, where it has a minimum.

• The first derivative of f is the function

f'(x) = (x - 32a) J(x ~ a)3·

The zero of the first derivative is x 32a The critical point of f is

A(3a2' 3a2v'3).
- The function f is increasing for x E (32a, +=).

The function f is decreasing for x E (-=,O)U(a, 32a).

GRAPHS OF FUNCTIONS 293

• The second derivative of the function f is

~ 3a 2 x 3a 2
f"(x) = V~· 4(x - ap = 4Jx(x _ a)S'

and it has no zeros.

Since f"(32a) > 0, the function has a minimum at the point

A (3a 3v'3a )
2' 2 .

- The function f is concave upward on the whole domain.

• Asymptotes.

- The function has a vertical asymptote x = a, but not at x = 0, because

lim
x->a+
g -- =
X -
3

a
+00, lim f(x) = lim
x---+o- x---+o-
J x3
X - a
= 0.

The function has slanted asymptotes y x +~ when x -+ +00 and

a
y = -x - 2" when x -+ -00, because

lim f(x) g x_ lim

IT
V~
x->+=-
X lim
x->+= xX- a = 1, x->-= x = -1,

lim (f(x) - x)
x->+=
= x->+=
lim (J x
X -
3
a
- x) ~,2

lim
x->-=
(IT +x)
V~
= -~.2
Example 7.7. Determine and sketch the graph of the function

f(x) = VlX2 - 1- Y9.

Solution. The graph of f is sketched in Figure 7.7.

• The domain of the given function is R.

• This is an even function.

• There are no zeros for this function.

294 OHAPTER 7.

• The first derivative is

f'(x) = ~~- {j(x 2 -1)2

3 fIX{j(x 2 - 1)2
- The critical points are

A(-I,-I), B(- ~,-w), 0(0,-1), D(~,-V'4) andE(I,-I),

because the first derivative has two zeros XI,2 = ~, but is not defined
at the points X3 = 0, X4 = -1 and X5 = 1.
- The function f is increasing for x E (- ~,O)U(~,oo).
- The function f is decreasing for x E (-00, - ~) U(O, ~).
- The function has minimums at the points

B (- ~, -w), D (~, -w) .

- The function has a maximum at the point 0(0, -1), even though it does
not have a horizontal tangent line at x = O. Namely, we have
2 ~ - {j(x 2 - 1)2
lim f'(x) = lim - = +00,
x-+o- x-+o- 3 fIX {j (x 2 - 1)2
and
2 ~ - {j(x 2 - 1)2
lim f'(x) lim - = -00.
x-+o+ x-+o+ 3 fIX{j(x 2 - 1)2

Note that the tangent lines through the points A and E are parallel with the
y-aXIS .

• The second derivative is

f"(x) = ~
9
(_1_ _{j(x3+2 _x2
~ 1)5
)
.

- The function is concave upward for x E (-1, 0) U (0, 1).

- The function is concave downward for x E (-00, -1) U (1, +00).
- From

lim f"(x) = .
bm -2 ( - 3
1- -
3 + x 2 ) = +00,
x-+-H x-+-H 9 ~ {j(x2 -1)5
lim f"(x)
x~-l-
= -00, lim f"(x)
x~l-
= +00 and lim f"(x)
x--+l+
= -00,
it follows that the function changes its concavity for x = 1 and x = -1.
GRAPHS OF FUNCTIONS 295

• Asymptotes.
- The function has no vertical asymptote.
- It has a horizontal asymptote y = °both when x -t +00 and x -t -00,
because

-1 = 0.
xl!~)V'X2 -1 - V9") = ",l!~oo {j(x2 -1)2 + {jX2(X2 -1) + W
- There is no slanted asymptote.

')I
1

ty
1
:t
D B

:Ie

B -2 01 1

= V'X2 - 1 - ifi2 12-xl

Fig. 7.7. f(x) Fig. 7.8. f(x) = 3VX2 +4
Example 7.8. Determine and sketch the graph of the function

12-xl .
f(x) = 3vx2 + 4

Solution. The graph of f is sketched in Figure 7.8. The given function can be
considered as the following two functions
2-x
fl(X) = 3y'X2TI' for x ~ 2,

x-2
f2(X) = 3.JX2+4' for x> 2.
Let us remark that it holds h(2) = lim h(x) and therefore this is a continuous
"' ..... 2+
function on R.

• The domain of the given function is R.

• The function fl(X) has a zero at the point A(2,O), while the function h(x)
has no zeros.
296 CHAPTER 7.

• The first derivatives are

I~ (x) = _ ~ x+2 2 x+2

+ 4)3/2' x> 2.
3 (X2 + 4)3/2' x::; 2, and I~(x) = 3 (x2

The first derivative of II has a zero at x = -2.

The first derivative of the function 12 has no zeros.
The function II is increasing in x E (-00, -2).
The function !I is decreasing in x E (-2,2).
The function II has a maximum at x B = - 2.
The function II has a minimum at x A = 2.
The function h is increasing in x E (2, +(0).

• The second derivatives are

I~'(x) = ~ . x 2 + 3x - 2 and I~'(x)=_~.x2+3x-2

3 (x2 + 4)5/2 ' x::; 2, 3 (x2+4)5/2' x> 2.

- The function !I has two points of inflection, namely

C (v17-3) 2 ,OA1 and D (3+v17)

- 2 ,OA5.

The function II is concave upward if

I{'(x) > 0, i.e., for x E (-00, XD) U (xc, 2).

The function !I is concave downward for x E (XD'XC).

The function h has no points of inflection.
The function h is concave downward if I~'(x) < 0, i.e., for every x E
(2, +(0).
• Asymptotes.

The function has no vertical asymptote.

The function has a horizontal asymptote y = 1/3 both when x -'> -00
and x -'> +00, because

lim 2- x lim x-2

x .... -oo 3v'x2 + 4 = 1/3 x .... +oo 3v'x2 + 4 = 1/3.

There is no slanted asymptote.

Example 7.9. Determine and sketch the graph of the function

I(x) = 2sinx + cos2x.

GRAPHS OF FUNCTIONS 297

Solution. The graph of f is sketched in Figure 7.9.

• The domain is (-00, +00).

• This function is periodic with basic period 21r, and therefore it is enough to
examine this function on the interval [0,21r].

• This is neither an odd nor an even function.

• The zeros are calculated from

f(x) =0, i.e., 2sinx+cos2x=0.

After the transformation cos 2x = cos 2 X - sin 2 x, we obtain

2 sin x + cos 2 X - sin 2 x = 0, i.e., 1 + 2 sin x - 2 sin 2 x = 0.

Taking t = sin x, we obtain the equation 1 + 2t - 2t2 = 0, with the solutions

t l ,2 = 1 ± V3
2

The value tl =
1+V3
2
.
> 1 can not be equal to sm x for any x E R, and
1- V3
therefore we are taking t2 = 2 .
Thus the zeros of f are Xl ~ 3.42, and X2 ~ 5.91.

• The first derivative is

f'(x) = 2cosx - 2sin2x = 2cosx - 4sinx cos x = 2cosx(1- 2sinx),

and it has zeros when

cos x = ° or 1 - 2 sin x = 0.

Thus the critical points of f on the interval [0,21r] are

A(~,l), B(3;,_3), CC;,3/2) and DC~,3/2).

• The second derivative of the function is f" = -2 sin x - 4 cos 2x and it holds

!"(~) = 2, 1"(3;) = 6, !"(i) = -3, 1"(561r) = -3,

wherefrom

- the point A (~, 1) is a local minimum;

31r
the point B( 2' -3) is a local minimum;
298 CHAPTER 7.

- the point C (i, 3/2) is a local a maximum;

- the point D(i,3/2) is a local maximum.

• The given function has no asymptotes.

'y

l"Y

Fig. 7.9. f(x) = 2sinx + cos2x Fig. 7.10. f(x) = sinxsin3x

Example 7.10. Determine and sketch the graph of the function

f (x) = sin x sin 3x.

Solution. The graph of f is sketched in Figure 7.10. The function can be written
as
f() cos 2x - cos 4x . 2 ( • 2 )
X = 2 = sm x 3 - 4 sm x .

• The domain is (-00, +oo).

• This function is periodic with basic period 7f.

• This is an even function.

• The zeros of f over [0,7f] are obtained from

7f 27f
sin x sin 3x = 0, which holds for Xl = 0, X2 ="3 and X3 = 3.

• The first derivative is

f'(x) = sin2x(4cos2x -1).

- The point 0(0,0) is a local minimum.
- The point B( 7f /2, -1) is a local minimum.
- The point A (arccos(1/4}/2, 9/16) is one of the two local maximums.

• The second derivative of the function is 1" = 16 cos 2 2x - 2 cos 2x - 8.

GRAPHS OF FUNCTIONS 299

• The inflection points are at

X4 = ±~ arccos Jf29 + 1 Xs = ±~2 (7r - arccos J12916 -1) .

---~
2 16'

• The given function has no line asymptotes.

Example 7.11. Determine and sketch the graph of the function

f(x) = sin3 x + cos3 X.

Solution. The graph of f is sketched in Figure 7.11.

• The domain is (-00, +(0).

• This function is periodic with basic period 27r.

• This is an even function.

• The zeros are obtained from the equation

. 3
sm x + cos3 x = ( sm
.
x + cos x ) ( sin2 -
1- - 2X) = 0,
. 77r
whose solutIOns are Xl = -37r and X2 = -.
4 4
• The first derivative is

f'(x) = ~ sin 2x(sin x - cos x).

- The points A(O,I), B (%,1) and C C:, -~) are local maximums.

. 7r J2 37r ..
- Thepomts D(4"'T)' E(7r,-I), and F(2,-I)arelocalmlmmums.

• The second derivative ofthe function is 1" (x) = ~(sin x + cos x ) (sin 2x - ~) .
• The inflection points are for

37r 77r d arcsin(3/2)

Xl = 4' X2 =4 an X3 = 2 :::::J 0.36,

and also for X4:::::J 1.21, Xs:::::J 3.51 and X6:::::J 4.35.

• The given function has no asymptotes.

300 CHAPTER 7.

'Y

- ...B'_-- A'... - .,
,~
I.!
E F

Fig. 7.11. f(x) = sin3 x + cos 3 X Fig. 7.12. f(x) = xe- x / a

Example 7.12. Determine and sketch the graph of the function

f(x) = xe- x / a , a i= 0.
Solution. The graph of f is sketched in Figure 7.12, both for a =2 and a = -2.

• The domain is (-00, +(0).

• This function is neither odd nor even.

• The zero of the function is at x = 0.

• The first derivative is

f'(x) = e- x / a (1 -~).
If a > 0, then the point A (a,~) is a maximum, and if a < 0, then the

point A (a,~) is a minimum of the function.

- If a> 0, then the function is increasing in (-00, a) and if a < 0, then the
function is increasing in (a, +00 ).
If a > 0, then the function is decreasing in (a, +(0) and if a < 0, then
the function is decreasing in (-00, a).

• The second derivative is

f"(x) = xe- x / a (~- ~) •

2 a a

The point of inflection is B(2a, 2ae- 2 ), for every a i= 0.

The function is concave upward in (2a, +(0).
The function is concave downward in (- 00, 2a).
GRAPHS OF FUNCTIONS 301

• Asymptotes.
The function has a horizontal asymptote y = 0 when x -+ += if a > 0
(resp. when x -+ -= if a < 0), because

lim xe- x / a = 0 for a > 0, and lim xe- x / a = 0, for a < O.

x----++oo x---+-oo

Example 7.13. Determine and sketch the graph of the function

f(x) = (1 - x2)e- 2x .
Solution. The graph of f is sketched in Figure 7.13.

• The domain of the function is the set R.

• The function is neither odd nor even.

• The zeros of the function are Xl = 1, Xl = -1.

• The first derivative is

f'(x) = (-2 - 2x + 2x2)e- 2x .

- The critical points are

A (1 - (-1 +
v's
2'
v's)e-(l-Vs))
2
and B (1 + (-1 -
2
v's, v's)e-(HVs))
2 .

.
- Th e functwn . for
f"IS IncreasIng X E ( -=, 1-2v's) U (1 +2v's ,+=.)
- The function f is decreasing for X E(1 -2v's ' 1+2v's) .
• The second derivative is f"(x) = (2 + 8x - 4x2)e- 2x .

- Since f" (1 +2 v's) > 0, it follows that the minimum is at the point

B (1 + (-1 -
v's
2'
v's)e-(HVs))
2 .

- Since f" ( 1 -2 v's) < 0, it follows that the maximum is at the point

A (1 - v's (-1
2'
+ v's)e-(l-Vs))
2 .
302 CHAPTER 7.

- The inflection points are

- -3 + 2V6e ,
C (1 -V6
2 ' 2
-2+v'6) D (1 + ~, -3 ~ 2V6 e- 2 -v'6) .
- The function is concave upward for

x E(1 _V62 ' 1+ V6)

2
.

- The function is concave downward for

xE (-00,1- ~)U(l+ ~,+oo).

• Asymptotes.
- The function has a horizontal asymptote when x --t +00, because
lim (1 - x2)e- 2x = O.
x--++oo

Note that
lim (1 - x2)e- 2x = -00.
x-+-oo

'Y
A 1~!'~
",.
,
/""
"
:It

1 B D

Fig. 7.13. f(x) = (1 - x 2)e- 2x . Fig. 7.14. f(x) = xe- 1jx2

Example 7.14. Determine and sketch the graph of the function

f( x) = xe- 1jx2 .

Solution. The graph of f is sketched in Figure 7.14.

• The domain of the function is (-00,0) U(O, +00).

• This is an odd function.

GRAPHS OF FUNCTIONS 303

• It has no zeros.

• The first derivative is

J'(x) = (X 2 + 2)e- 1 / x2
x2
Since J' has no real zeros, the given function does not have any critical
points.
This is an increasing function for every x E (-00,0) U(O, +00).

• The second derivative is

J"( x ) __ -2x 2
5
+ 4 e -1/x2
x
- The inflection points are

A( V2, V2e- 1 / 2 ) and B( -V2, _V2e- 1 / 2 ).

- The function J is concave upward for x E (-00, -J2) U (0, J2).

- The function J is concave downward for
x E (-V2,0) U (V2,+00).

• Asymptotes.

Though the function is not defined at x = 0, it has no vertical asymptote

there, because

lim xe- 1 / x2 - 0 lim xe- 1 / x2 = O.

x~o- - , x--->o+

It has no horizontal asymptote.

It has a slanted asymptote y = x both when x -+ +00 and when x -+ -00,
because
xe- 1 / X2
k= lim = 1, and
x--->±oo X

-1/x 2 1
n = lim (xe- 1 / x2
x--->±oo
- x) = lim x(e- 1 / x2
x--->±oo
- 1) = lim e
x--->±oo
v1 = O.
x·-
x2

Remark. This function can be continuously extended to the point x = 0, if one

defines J(O) = o.

Example 7.15. Determine and sketch the graph oj the Junction

J(x) = -(x + 2)e1 / x .

304 CHAPTER 7.

Solution. The graph of f is sketched in Figure 7.15.

• The given function is defined for x E (-00,0) U(O, +00).

• The function is neither an odd, nor an even function.

• The only zero is x = -2.

• The first derivative is

f'(x) = x+2e 1/ X _e 1/ X = 2+x_x2 l/x

x2 x2 e.

• The critical points are A(2,-4e1 / 2), B(-I,-e- 1 ).

- The function f is decreasing for x E (-00, -1) U(2, +00).

- The function f is increasing for x E (-1, 0) U( 0, 2).

• The second derivative of the function f is

f "( x ) -_ + 2 e 1/ x
- 5x__
X4 •

Since
f"(2) = - ~: e1 / 2 < 0 and f"( -1) = 3e- 1 > 0,

the following holds.

- The function f has a maximum at A(2, _4e 1 / 2 ).

- It has a minimum at B( -1, _e- 1 ) .

. fectlOn
- Th em l· . .1S C (-S'
pomt 2 -Se
8 -5/2) .
- The function is concave downward for x E (-2/5,0) U (0, +00).
- The function is concave upward for x E (-00, -2/5).

• Asymptotes.

- The vertical asymptote is x = 0, and it holds

lim f(x)
X---+O+
= X--+O+
lim (_x_2)e 1 / x = -00; lim f(x)
x---+o-
= x-+o-
lim (_x_2)e 1 / x = O.

- Since
lim f(x)
x---++oo
= lim (-x - 2)el/x
x---++oo
= -00,
the function has no horizontal asymptotes.
GRAPHS OF FUNCTIONS 305

- The slanted asymptote is y = -x - 3 when x ~ ±CXJ, because

k= lim f(x) = lim -x - 2 e1 / x = -1 and

x-+±oo x x-+±oo x '

n= lim (J(x) - kx) = lim ((-x - 2)e 1 / x

x-+±oo x-+±oo
+ x)

lim (x(l - e1 / x )) - lim 2e 1 / x

x-+±oo x-+±oo

1 - e1 / x 1 - et
lim - 2 = lim - - - 2 = -3.
x-dO<) 1 t-+O± t
x

',:6 10 2 " 1
...U
,
,, II
,,
,,

I
_______________ J.
I

,, I
iJt
-I

Fig. 7.15. f(x) = -(x + 2)e1 / x Fig. 7.16. f(x) = e2x /(1-x 2 )

Example 7.16. Determine and sketch the graph of the function

f(x) = e2x /(l-x 2 ).

Solution. The graph of f is sketched in Figure 7.16.

• The domain of the function is (-CXJ, -1) U( -1,1) U(l, +CXJ).

• The function is neither odd nor even.

• There are no zeros for the function.

• The first derivative is

f'(x) = 2(x 2 + 1)e2x /(1-x 2 )

(1 - x 2 )2
- The first derivative has no zeros.
- The function f is increasing for every x from the domain.
306 CHAPTER 7.

• The second derivative is

f"(x) = 4e2x/(1-x2) _x 5 + X4 - 2x 3 + 2X2 + 3x + 1
(1 - X 2 )4

It has a zero xA E (1,2), and there it has an inflection point.

The function is concave upward for x E (-00, -1) U( -1,1) U(l, xo) and
concave downward for x E (xo, +00).

• Asymptotes.
- The vertical asymptotes are x = -1 and x = 1, because
lim-~2x/(1-x2) = +00, lim e 2x /(1-x 2)_
x--+-l- x ..... -l+ - 0,

lim e 2x /(1-x 2) = +00, lim e2x /(1-x 2)

x--+I- x"'" 1 + = O.
The horizontal asymptote is y = 1 when x --t ±oo, because
lim e 2x /(1-x2) = l.
x-+±oo

There are no slanted asymptotes.

Example 7.17. Determine and sketch the graph of the function

f(x) = lnx
x
Solution. The graph of f is sketched in Figure 7.17.
• The domain of the function is (0, +00).

• The point x = 1 is a zero of the function.

• The first derivative is

f '()
x = 1- 2
x
In x' hence f '( x ) = 0 for x = e.

- The function f is increasing for x E (0, e).

The function f is decreasing for x E (e, +00 ).
The point A(e, lie) is a maximum of this function.

• The second derivative is

f"(x) = 2lnx - 3
x-

• The point of inflection is B( e3/2, 3e- 3 / 212).

• The function is concave upward for x E (e3/ 2, +00), and concave downward for
x E (0, e3 / 2 ).
GRAPHS OF FUNCTIONS 307

• Asymptotes.
· lnx
The vertica1 asymptote is x = 0, b ecause 11m - - = -00.

°when x
X-+O+ X

The horizontal asymptote is y = ~ +00, because

.
11m lnx
-=0.
x-++oo X

'Y
y
A B

o "

1
o ,... %
,II
II

Fig. 7.17. f(x) = lnx Fig. 7.18. f(x) = Vi

x ln 2 x
Example 7.18. Determine and sketch the graph of the function

f(x) = 1Vi
2 •
n x
Solution. The graph of f is sketched in Figure 7.18.

• The domain of the function is (O,l)U(l,+oo).

• The first derivative is

lnx - 4
f'(x) = 2Viln3 x' f'(x) = 0, for x = e 4 •

- The function f is increasing for x E (0,1) U( e4, +00).

- The function f is decreasing for x E (1, e4 ).
- The point A( e\ e 2 /8) is a minimum of this function.

• The second derivative is

24 -In 2 x
f"(x) = 4#ln4 x

• The points of inflection are at Xl = eV24 , X2 = e- V24

308 CHAPTER 7.

• The function is concave upward for x E (e- V24 ,l)U(l,e V24 ), and concave
downward for x E (0, e- V24 ) U( eV24 , +00).

• Asymptotes.

The vertical asymptote is x = 1, because lim v:

x_l± In x
= +00.
There are no horizontal or slanted asymptotes.

Example 7.19. Determine and sketch the graph of the function

f(x) = (1 + x?/x = exp C(l x+ X)) .

n

Solution. The graph of f is sketched in Figure 7.19.

• The domain of the function is (-1,0) U(O, +00).

• The first derivative is

!,(x) = (1 + X)l/X ( 1 _ In(l + x))

x(x+1) x2 .
- From the inequality
x
- - < In(l +x) < x,
x+1
we obtain that the function f is decreasing on its entire domain.

• The second derivative is

f"(x) = (l+X)l/x (( 1 _ In(l + X))2 + ~ (2In(x + 1) _ 2x + 3X2)) .

x(x+1) x2 x3 (l+xF
The function
¢(x) = 2ln(x + 1) _ 2x + 3x 2
(l+xp
has the first derivative
2X2
¢'(x) = (1 + x)3 > 0, for -1 < x < +00.

From ¢(O) = 0, it follows that

¢(x)<O if -l<x<O, and ¢(x»O for x>O,

wherefrom we get

¢(x) >
x3
° for x E (-l,O)U(O,+oo).
The first addend in the second derivative is grater than zero, and since the
second is also grater then zero, it follows f"(x) > °
for every x from the
domain. This means that
GRAPHS OF FUNCTIONS 309

- the function is concave upward on the whole domain.

• Asymptotes.
- The vertical asymptote is x = -1, because

lim (1 + X)l/X = +00 and lim (1 + x?/x = lim (1 + x)* = e.

x-+-l+ x-+o- x-+o+

- The function has a horizontal asymptote y = 1 when x --+ +00, because

lim (1
x---++oo
+ X)l/x = l.

Remark. The point A(O, e) does not belong to the graph of f, since the point 0 is

't
not in its natural domain. However, f can be continuously extended to x = 0 if one
defines f(O) = e (see the Remark after Example 5.37).

,'Y
'Y A
..,'
i ' :
k:
,,,
: I

, x
,
: A
y-l: "--
-----~---------- --x
,0I 1 8

2x
Fig. 7.19. f(x) = (1 + X)l/X Fig. 7.20. f( x) = arcsin 1 + x 2

Example 7.20. Determine and sketch the graph of the function

2x
f(x) = arcsin --2.
l+x
Solution. The graph of f is sketched in Figure 7.20.

• The domain of the function is (-00, +00), because (x - 1)2 :::: 0 and thus
2x
-1 < - - < l.
- 1 +x2 -

• This is an odd function.

• The function has a zero at x = O.

• The first derivative is

f'(x) = 2sgn(1 - x 2 )
1 + x2 ,x =f. ±l.
310 CHAPTER 7.

It has no zeros and it holds

lim f'(x) = -1, lim f'(x) = 1,

x-+-1- x--+-l+

lim f'(x) = 1, lim f'(x) = -l.

x-+l- x--+l+

The function f is increasing for x E (-1, 1).

- The function f is decreasing for x E (-oo,-I)U(I,+oo).

- It has a minimum at B (-1, -i-) .

- It has a maximum at A (I,~) .
• The second derivative is

f "( x ) = (1 + X 2 )2 sgn ( 1 - x 2) ,
-4x x ~ ±I

and it has a zero at x = 0 and this is the inflection point 0(0,0).

• The function is concave upward for x E (-1,0) U(I, +00) and concave down-
ward for x E (-00, -1) U(O, 1).

• From the previous conclusions it follows that the critical points

A(I,71)2) and B(-I,-7r/2)

are extrema points and also points in which the graph changes its concavity.

• Asymptotes.
There is no vertical asymptote.
The horizontal asymptote is y = 0 when x ---) ±oo, because
. . 2x
hm arCSIn -1--2
x--+±oo + X = O.
- There is no slanted asymptote.

Example 7.21. Determine and sketch the graph of the function

1 _x 2
f( x) = arccos --2.
I+x
Solution. The graph of f is sketched in Figure 7.21.

• The domain of the function is (-00, +00), because it holds

2 1 - x2
- 1 < -1 + - -2 = -- < l. (7.1 )
1+x 1 + x2
GRAPHS OF FUNCTIONS 311

• This is an even function.

• The function has no zeros, as follows from relation (7.1).

. .. 2· sgnx
• The first denvatIve IS 1'( x) = 2' x =I- O.
1+x
It has no zeros and it holds

lim f'(x) = -2, lim f'(x) = 2.

x->o- X->O+

So the function has a minimum at 0(0, b).

- The function f is increasing for x E (0, +00).
- The function f is decreasing for x E (-00,0).

• The second derivative is

-41xl =I- O.
f"(x) = (1 + x 2 )2' x

• The function has no inflection points, because it holds f"(x) < 0, for x =I- O.

• The function is concave downward for every x in the domain.

• Asymptotes.

There is no vertical asymptote.

The horizontal asymptote is y = 7r when x -+ ±oo, because

1 - x2
lim arccos - - - 2 =
1+x
7r.
x~±oo

- There is no slanted asymptote.

y
y=1t y = 1t12

.2-' B
~
'.p

x x
o o

1 - x2
Fig. 7.21. f(x) = arccos 1 + x 2 Fig. 7.22. arctan eX - In ( J :2:2X)
1
312 CHAPTER 7.

Example 7.22. Determine and sketch the graph of the function

f(x) = arctan eX -In ( J 1 :2:2X) .

Solution. The graph of f is sketched in Figure 7.22.

• The domain of the function is (- 00, +00 ).

• The function has no zeros.

• The first derivative is

eX - 1
l' (x) = 1 + e2x .

• The critical point is A (0, ~ + ~ In 2) .

- The function f is increasing for x E (0, +00).
- The function f is decreasing for x E (-00,0).

- So the function has a minimum at A (0, ~ + ~ In 2) .

• The second derivative is

1"(x) = eX ( _e 2x + 2ex + 1)
(1 + e2x )2
• It has a zero at XB = In(l + .J2), which is the point of inflection.

• The function is concave downward for x E (-00, In(l + .J2)),

• The function is concave upward for x E (In(l + .J2), +00).
• Asymptotes.

- The horizontal asymptote is y = 7r /2 when x -+ +00, because

lim (arctan eX - In (.
x-->+oo V~) +
1 e 2;'- )
= ~.2
The slanted asymptote is y = -x when x -+ -00.
Bibliography

[1] Adnadjevic, D., Kadelburg, Z., Mathematical analysis, Naucnaknjiga, Belgrade

1989 (in Serbo-Croatian).

[2] DeLilo, N.J., Advanced calculus with applications, Macmillan Publ. Co., New
York 1982.

[3] Demidovic, B.P., Collection of exercises in mathematical analysis, Moscow 1977

(in Russian).

[4] Fihtengoljc, T.M., Course on differential and integral calculus, Vol. I, Nauka,
Moscow 1970 (in Russian).

[5] Fridy, J.A., Introductory analysis, Hartcourt Brace Jovanovich Publ., New York
1987.

[6] Gajic, Lj., S. Pilipovic, Collection of exercises from Analysis I, Second part,
University of Novi Sad, Institute of Mathematics, Novi Sad 1990 (in Serbo-
Croatian).

[7] Gaskill, H.S., Narayanaswami, P.P., Foundation of analysis, Harper and Row,
Pub., New York 1989.
[8] Krall, A.M., Applied analysis, D. Reidel Publ. Co., MA 1986.

[9] Kudriavcev, L.D., Course on mathematical analysis, Vol. I, Vis. skola, Moscow
1981 (in Russian).

[10] Kurepa, S., Mathematical analysis, Vol. 2, Tehnicka knjiga, Zagreb 1977 (in
Serbo-Croatian)

[11] Lang, S., Analysis I, Addison-Wesley Publ., MA 1968.

[12] Lasko, LL, Boyarcuk, A.K., Gai, Ya.G., Golovac, G.P., Handy textbook on
mathematical analysis, Vol. 1, Visa skola, Kiev 1978 (in Russian).

[13] Loomis, L. H., Sternberg, S., Advanced calculus, John and Bartlett Publ., MA
1990.

[14] Mardesic, S., Mathematical analysis in the n-dimensional space, Skolska knjiga,
Zagreb 1974 (in Serbo-Croatian).

313
314 BIBLIOGRAPHY

[15J Milicic, P.M., Uscumlic, M.P., Collection of exercises from higher mathematics,
Vol. I, Naucna knjiga, Belgrade 1988 (in Serbo-Croatian).

[16J Pownal, Malcolm, W., Real analysis - a first course with foundations, Wm. C.
Brown Publ., IO 1994.

[17J Protter, M.,H., Morrey C.,B., A first course in real analysis, Springer-Verlag,
New York Inc. 1977.

[18J Ruckle, W. H., Modern analysis, PWS-Kent Publ. Co., MA 1969.

[19J Rudin, W., Principles of mathematical analysis, McGraw Hill, New York 1964.

[20J Stankovic, B., Elements of functional analysis, Naucna knjiga, Belgrade 1975
(in Serbo-Croatian).

[21 J Swokowski, E. W., Calculus with analytic geometry, Second edition, Prindle,
Weber and Schmidt, Boston, Massachusetts 1979.

[22J Zoric, V.A., Mathematical analysis, Nauka, Moscow 1981 (in Russian).

[23J TakaCi, Dj., TakaCi, A., Collection of exercises from Analysis I, First part,
University of Novi Sad, Institute of Mathematics, Novi Sad 1989 (in Serbo-
Croatian).

[24J TakaCi, Dj., TakaCi, A., Collection of exercises from differential and integral
calculus, University of Novi Sad, Institute of Mathematics, Novi Sad 1992 (in
Serbo-Croatian) .
Index

Abelian group, 2, 30 Cantor theorem, 5

Absolute value, 2 Cauchy sequence, 84
Accumulation point Cauchy's theorem, 249
of a sequence, 83 Chain rule, 215
of a set, 38 Chebychev-Hermite polynomial, 248
Additive identity element, 1 Closed
Additive inverse, 1 interval, 3
Antisymmetry of ::;, 2 set, 38
Archimedes theorem, 5, 22 Closure of a set, 38, 40
Associative law Codomain of a function, 45
for addition, 1 Commutative group, 2
for multiplication, 1 Commutative law
Asymptote for addition, 1
horizontal, 171 for multiplication, 1
slanted, 171 Completeness of R, 2, 84
vertical, 170 Composite function, 46
Asymptotic equivalence first derivative of, 215
of functions, 174 Concave
of sequences, 135 downward, 48, 279
Asymptotic relations, 174 upward, 48, 279
Concave function, 48
Basic elementary function, 48 Concavity, 279
Bernoulli's inequality, 14 Continuity
Big oh at a point, 181
of a function, 174 of composite function, 182
of a sequence, 135 on a set, 182
Bijection, 46 uniform, 204
Binomial coefficient, 12 Continuous extension of a function, 212
Binomial formula, 12 Continuous function, 181
Bolzano-Weierstrass theorem, 38 from the left side, 181
Boundary point of a set, 38 from the right side, 182
Bounded Convex function, 48
function, 47 Countable set, 38
interval, 3 Covering of a set, 38
sequence, 83 Critical number, 270
Bounded set, 3, 38 Critical point, 270
from above, 3 Cut
from below, 3 in Q, 29

315
316 INDEX

in R, 37 at a point, 181
irrational, 29 on a set, 182
rational, 29 differentiable, 216
elementary, 48
Decreasing function, 270 graph of a, 46
Dedekind's theorem, 35 injective, 45
Derivative inverse, 46
first, 213 irrational, 75
higher order, 216 rational, 74
left-hand side, 213 surjective, 45
right-hand side, 213 tends to infinity, 143
second, 216 uniformly continuous, 204
third, 216 Function concave
Differentiable function, 216 downward, 48, 279
Differential, 216 upward, 48, 279
Dirichlet's function, 59, 185, 199 Fundamental theorem of algebra, 73
Discontinuity
first order, 182 Global
removable, 182 maximum, 47
second order, 182 minimum, 47
Distance, 3, 44 Graph
Distributive law of multiplication over perpendicular line of the, 216
addition, 1 tangent line of the, 216
Divisor of a number, 4 Graph of a function, 46
Domain Greatest integer part, 23
natural, 46 Group, 2
of a function, 45
Harmonic sequence, 100
Elementary function, 48 Hausdorff topological space, 39
Equality of functions, 45 Heine-Borel property, 39
Euler's constant, 110 Higher order derivative, 216
Even function, 47 Horner scheme, 74
Exponential function, 48
Extremum, 270 Iensen's inequality, 284
Implicit function, 215
Field first derivative of, 215
complete totally ordered, 2 Increasing function, 270
of rational numbers, 5 Increment
of real numbers, 1 of the dependent variable, 213
totally ordered, 2 of the independent variable, 213
Finite set, 38 Infimum of a set, 3, 24
First derivative, 213 Infinite set, 38
Function, 45, 46 Inflection point, 279
basic elementary, 48 Injection, 45
bijective, 46 Integer, 4
continuous even, 4
INDEX 317

odd, 4 Maclaurin's
Integers formula, 253
set of, 4 polynomial, 253
Interior point of a set, 38 Mathematical induction principle, 4
Interval Maximum of a function
bounded, 3 global, 47
closed, 3 local, 47
open, 3 Maximum of a set, 3
unbounded, 3 Mean value theorems, 244
unbounded to the left, 3 Metric space, 44
unbounded to the right, 3 Minimum of a function
Inverse global, 47
function, 46 local, 47
trigonometric function, 48 Minimum of a set, 3
Inverse function Monotone sequence, 102
first derivative of, 215 Monotonically
Irrational decreasing function, 46
cut, 29 decreasing sequence, 102
number, 4 increasing function, 46
Irrational function, 75 increasing sequence, 102
Isolated point of a set, 38 non decreasing function, 47
nonincreasing function, 47
L'Hospital's Rule, 263 Monotonicity of functions, 270
Lagrange Theorem, 244 Multiplicative
Laguerre's polynomial, 247 identity element, 1
Largest common divisor, 107 inverse, 1
Left-hand limit, 142
Natural domain of a function, 46
Limes
Natural number, 4
inferior of a sequence, 117
prime, 4
superior of a sequence, 117
Neighborhood of a point, 37
Limit
Number
left-hand , 142
irrational, 4
of a sequence, 83
natural, 4
right-hand, 142
rational, 4
Limit of a function
real, 1
at a point, 141
Number e, 104, 107
at minus infinity, 142 Numbers
at plus infinity, 142 set of natural, 4
Local
extremum, 270 Odd function, 47
maximum, 47, 270 One sided derivative, 213
minimum, 47, 270, 271 Open
Local extrema of functions, 270 interval, 3
Logarithmic function, 48 set, 37
Lower bound of a set, 3 Open covering, 38
318 INDEX

Parametric function, 215 tends to minus infinity, 83

first derivative of, 215 tends to plus infinity, 83
Partial fraction, 74 Set
Period bounded, 3
basic, 47 bounded from above, 3
of a function, 47 bounded from below, 3
Perpendicular line of the graph, 216 closed, 38
Point compact, 38
critical, 270 inductive, 4
interior, 38 interior of a, 38
of closure, 38 of integers, 4
of inflection, 279 of natural numbers, 4
Polynomial, 72 of rational numbers, 4
Polynomial function, 72 open, 37
Power function, 48 symmetric, 47
Prime natural number, 4 unbounded, 3
Small oh
Range of a function, 45 of a sequence, 136
Rational of a function, 174
cut, 29 Stationary subsequence, 117
function, 74 Stirling's formula, 136
number, 4 Stolz theorem, 132
Real function of a real variable, 46 Sub covering, 39
Real number, 1 Subsequence, 116
field of, 1 stationary, 117
Relatively prime integers, 4 Supremum of a set, 3, 24
Riemann's function, 200 Surjection, 45
Right-hand limit, 142 Symmetric set, 47
Rolle's Theorem, 244
Root of a function, 46 Tangent line of the graph, 216
Root of a polynomial, 73 Taylor's
formula, 253
Scale of growths for sequences, 137 polynomial, 253
Second derivative, 216 theorem, 252
Sequence, 83 Topological space, 41
bounded, 83 Total ordering of :S;, 2
asymptotic relation, 135 Transitivity of :S;, 2
convergent, 83 Trigonometric function, 48
divergent, 83
general term of a, 83 Unbounded
limit of a, 83 interval, 3
monotone, 102 set, 3
monotonically decreasing, 102 Undetermined form
monotonically increasing, 102 0/0, 262
monotonically nondecreasing, 102 100 , 267
monotonically nonincreasing, 102 00 - 00,266
INDEX 319

00/00,262
Uniform continuity, 204
Uniformly continuous function, 212
Upper bound of a set, 3

Variable
dependent, 45
independent, 45

Zero of a function, 46
Zero of a polynomial, 73
complex, 73
rational, 73
real, 73
Kluwer Texts in the Mathematical Sciences
1. A.A. Harms and D.R. Wyman: Mathematics and Physics of Neutron Radiography.
1986 ISBN 90-277-2191-2
2. H.A. Mavromatis: Exercises in Quantum Mechanics. A Collection of lllustrative
Problems and Their Solutions. 1987 ISBN 90-277-2288-9
3. V.1. Kukulin, V.M. Krasnopol'sky and J. Horacek: Theory of Resonances. Principles
and Applications. 1989 ISBN 90-277-2364-8
4. M. Anderson and Todd Feil: Lattice-Ordered Groups. An Introduction. 1988
ISBN 90-277-2643-4
5. J. Avery: Hyperspherical Harmonics. Applications in Quantum Theory. 1989
ISBN 0-7923-0165-X
6. H.A. Mavromatis: Exercises in Quantum Mechanics. A Collection of lllustrative
Problems and Their Solutions. Second Revised Edition. 1992 ISBN 0-7923-1557-X
7. G. Micula and P. Pavel: Differential and Integral Equations through Practical
Problems and Exercises. 1992 ISBN 0-7923-1890-0
8. W.S. Anglin: The Queen of Mathematics. An Introduction to Number Theory. 1995
ISBN 0-7923-3287-3
9. Y.G. Borisovich, N.M. Bliznyakov, T.N. Fomenko and Y.A. Izrailevich: Introduction
to Differential and Algebraic Topology. 1995 ISBN 0-7923-3499-X
10. J. Schmeelk, D. Takaci and A. Takaci: Elementary Analysis through Examples and
Exercises. 1995 ISBN 0-7923-3597-X

KLUWER ACADEMIC PUBLISHERS - OORDRECHT / BOSTON / LONDON