‫قســــــــــــــــــــــــــم الرياضيـــــــــــــــــــــــــــــــات‬

1 Course title Real Analysis

2 Course Code M 305

3 Instructor Dr. fathi

Total Credit hours 3

4 Credit hours (Theory) 3

Credit hours (Practical) -

5 Prerequisites M 102

6 Program title Mathematics

7 Faculty Science

8 Department Mathematics

9 Level Third

10 Semester (s) First Semester

11 Language of Instruction English

-1-
 CONTENTS

Course Description 3

Learner Objectives 3

Real Numbers 4

Limits and Continuity 29

Infinite Sequences 54

Infinite Series 73

Power Series 125

References 152

-2-
Course Description:

This course covers properties of integers, sets, real numbers, limits of

functions, continuity of functions, infinite series, infinite sequences,
and power series.

Learner Objectives

During this course you will be engaged in addressing the following course
objectives:

1. Demonstrate knowledge and understanding of real analysis.

2. Demonstrate knowledge and understanding of the concept of limits

of functions.

3. Demonstrate knowledge and understanding of continuity of

functions.

4. Demonstrate knowledge and understanding of infinite series.

5. Demonstrate knowledge and understanding of the concept of

infinite sequences.

6. Demonstrate knowledge and understanding of the concept of power

series.

-3-
 Real Numbers
1.1: Real ( Line ) Numbers Set

We assume the reader is familiar with the geometric representation of the real
numbers set, denoted by R, by means of points on a straight line as in the
following figure [ Fig. 1.1 ]:

Note that (1.1.1):

A point , called the origin, is chosen to represent 0 and another point,

usually to the right of 0, is chosen to represent 1. Then there is a natural

way to pair off the points on the real line and the real numbers set R, that
is each point will represent a unique real number and each real number will
be represented by a unique point. For this reason we refer to R as the real
line and use the words point and number interchangeably. The set R can
be characterized by the statement that R is a complete, Archimedian,
ordered field. The algebraic properties of R are summarized in the
following axioms, which state that ( R, +,  ) is a field.

1.1.1: Definition ( Field axioms ): Consider the two binary operations on R:

+: R R → R
( x, y) → x + y,

: R  R → R
( x, y) → x  y  xy,  x, y  R,

-4-
where +,  denote the usual operations of addition and multiplication

respectively on R. Then ( R, +,  ) is a field if it satisfies the following

axioms:

( A1 ) Addition is associative: x + ( y + z ) = ( x + y ) + z,
( A2 ) Additive identity: x + 0 = x = 0 + x,
( A3 ) Additive inverse: x + ( − x ) = 0 = ( − x ) + x,
( A4 ) Addition is commutative: x + y = y + x,
(M1 ) Multiplication is associative: x ( y z ) = ( x y ) z,
(M 2 ) Multiplicative identity: x1 = x = 1 x,
(M 3 ) Multiplicative inverse: x x−1 = 1 = x−1 x, x  0,
(M 4 ) Multiplication is commutative: x y = y x,
(D ) Multiplication is distributive over addition: ( x + y) z = xz + yz,
for each x, y, z  R.
The real numbers set R, have a natural order relation that is compatible
with their algebraic structure. We visualize the ordered real numbers as the
real line, with smaller numbers to the left and larger numbers to the right.

Here, we define an order relation in R, using the concept of

positiveness:

-5-
1.1.2: Definition ( Ordered axioms): Let x, y, z  R. Then:

(O1 ) Either x  y, x = y or x  y,
(O2 ) If x  y and y  z, then x  z,

(O3 ) If x  y, then x + z  y + z,

(O4 ) If x  y and z is positive, then xz  yz,

(O5 ) If x  y and z is negative, then xz  yz.

Thus, (R, +,  , ) is an ordered field.

In the following we define open, closed and half-(open) closed intervals on

R:
1.1.3: Definition ( Intervals): Let a, b  R with a  b, the open, closed

and half-(open) closed intervals on R are given as:

] a, b [ = {x R: a  x  b },
] a,  [ = {x R: x  a },
• Open intervals : 
] −, b [ = {x R: x  b },
] −,  [ = {x R: −  x   } = R.

• Closed interval : [ a, b ] = { x : a  x  b }.

] a, b ] = {x R: a  x  b },
• Half − (Open ) closed intervals : 
[ a, b [ = {x R: a  x  b }.

-6-
Note that (1.1.2):

The set R of real numbers is totally ordered by the relation

x  y, read x is less than or equal to y, means x  y or x = y.

Here, we state some important examples of subsets on R:

1.1.4: Definition ( Subsets of R ): (i ) The natural numbers set, denoted by
N , and is given as:

N = {1, 2, 3, }.

(ii ) The integer numbers set, denoted by Z , and is given as:

Z = {, − 3, − 2, −1, 0, 1, 2, 3, }.

(iii ) The quotient ( rational ) numbers set, denoted by Q, and is given as:

x 
Q =  : x, y  Z , y  0 .
y 

One can easily show that (Q, +,  ) is a field. The rational numbers

are linearly ordered by their standard order, and this order is compatible
with the algebraic structure of Q. Thus, (Q, +,  , ) is an ordered field.

Moreover, this order is dense, (cf. for more details: 1.6: dense sets )
meaning that if r1, r2  Q, and r1  r2 , then there exists a rational

number r Q between them with r1  r  r2. For example, we can take:

1
r= (r1 + r2 ).
2

-7-
The fact that the rational numbers are densely ordered might suggest that
they contain all the numbers we need. But this is not the case: they have a
lot of " gaps, " which are filled by the irrational real numbers, such as

2.

(iv ) The irrational numbers set, denoted by Qc , are those real numbers

which are not rational: thus Qc , the complement ( relative to R ) of the

set Q of rational numbers, denotes the set of irrational numbers. Thus, it

is obvious that:

(1) N  Z  Q  R,

(2) R = Q  Qc.

Why there exist irrational numbers set Qc ? The following theorem

shows that 2 is irrational number, that is 2  Qc. In particular, the

length of the hypotenuse of a right-angled triangle with sides of length one
is not rational. Thus, the rational numbers are inadequate even for
Euclidean geometry; they are yet more inadequate for analysis. The

irrationality of 2

was discovered by the Pythagoreans of Ancient Greece in the 5th century

BC, perhaps by Hippasus of Metapontum.

Theorem 1.1.1 ( Irrational numbers ) : There is no rational number

xQ such that x 2 = 2.

-8-
 a
Proof : Suppose for contradiction that x2 = 2 and x= where
b
a, b  N . By cancelling common factors, we can assume a and b are
relatively prime (that is, the only integers that divide both a and b are

 1). Since x2 = 2, we have:

a 2 = 2 b2 ,

which implies that a 2 is even. Since the square of an odd number is odd,
a = 2 n, n  N must be even. Therefore:

2n2 = b2,

which implies that b2 is even, then b = 2 m, m  N is even. Hence a

and b have a common factor of 2, which contradicts the initial assumption
that there are no common factors. #

1.2: Absolute Value Function

Real numbers satisfy many inequalities. Firstly, we introduce the

definition of absolute value function, then we illustrate a simple, but more
fundamental, examples on the inequalities:

1.2.1: Definition ( Absolute value function ): The absolute value function is

x, x  0
defined as: f ( x ) = | x | := x2 = 0, x = 0
− x, x  0.


-9-
The geometrical meaning (interpretation) of | x |, as the distance between

the point x and the origin point 0. That is:

d ( x, 0 ) = | x − 0 | = | x |  0,  x  R.

In the following we introduce some simply inequalities, for some positive

real number (   0 ):

• | x|    −   x  .

• | x − c|    c −   x  c + .

• 0  | x − c |   , ( x  c )  c −   x  c or c  x  c +  .

• | x |  , (  0 )  x   or x  −.

We come now to one of the fundamental inequalities of calculus:

• | x + y |  | x | + | y |,  x, y  R.

This is called the triangle inequality in analogy with the geometric

observation that: “ In any triangle the length of each side is less than or
equal to the sum of the lengths of the other two sides ”.

Proof of the triangle inequality: The key here is to think of

| x | := x2 .

Note first that:

( x + y )2 = x2 + 2 xy + y 2  | x |2 + 2 | x || y | + | y |2 = ( | x | + | y | )2 .
Comparing the extremes of the inequality and taking square roots, we have:

( x + y ) 2  | x | + | y |.

- 10 -
The result follows from observing that:

( x + y ) 2 = | x + y |.
#

Here are a variant of the triangle inequality that also come up in

calculus:

• | x − y |  | x | + | y |,  x, y  R.

• | x − y |  | x − z | + | z − y |,  x, y, z  R.

• | | x | − | y | |  | x − y |  | x | + | y |,  x, y  R.

The proof of the last inequalities is obvious and follows directly from the
fundamental triangle inequality. On the other hand, we give the proof of the
left-hand side of the last inequality:

Proof: The key here is, also to think of | x | := x2 .

| | x | − | y | |2 = ( | x | − | y | )2 = | x |2 − 2 | x || y | + | y |2
 x2 − 2 xy + y 2 = ( x − y )2 = | x − y |2 .

By using:

| x − y | = ( x − y )2 .

Comparing the extremes of the inequality and taking square roots. Hence
the result follows, that is:

| | x | − | y | |  | x − y |  | x | + | y |,  x, y  R.
- 11 -
 Exercises (1.2):

x x +1
(1) Given that: x  0, compare: and .
x +1 x +2

1 1
( 2 ) Suppose that: x y  0. Show that if x  y, then  .
y x

( 3 ) Given that: x, y  0, show that if x2  y 2 , then x  y .

( 4 ) Show that if: 0  x  y, then x y.

(5) Show that: | x + y | = | x| + | y|  x y  0.

x y
(6) Show that if: 0  x  y, then  .
x +1 y +1

( 7 ) Let x, y, z be nonnegative real numbers. Show that:

x y z
if x  y + z, then   .
x +1 y +1 z +1

( 8 ) Show that if: x, y are real numbers, then:

1 2
xy  ( x + y2 ).
2

(9) Show that if: x, y are real numbers and x  y, then

x+ y x+ y
x   y. The number . is called the arithmetic mean of
2 2
x, y.

- 12 -
 x+ y
(10 ) Given that: 0  x  y, show that 0  xy  . The
2
number xy . is called the geometric mean of x, y.

(11 ) Show that if: x, y are real numbers then: | x − y|  |x| − | y| .

1.3: Bounded Sets

Next, we use the ordering properties of R to define the supremum

and infemum of a set of real numbers. These concepts are of central

importance in analysis. In particular, in section: ( 1.5: The Completeness
Axiom) we use them to state the completeness property of R. First, we
define upper and lower bounds:

1.3.1: Definition ( Bounded sets ): The real subset, denoted by A R is said to

be bounded above if there is a real number N R such that:

x  N ,  x  A. The subset A  R is said to be bounded below if there is a

real number M R such that: x  M ,  x  A. If A is both bounded

below and bounded above, we say that A is bounded.

Thus A is bounded iff ( if and only if )  A  [ M , N ]  R, that is A

is a subset of some closed interval [ M, N ] ( of finite length ). The set of all

natural numbers, denoted by N = {1, 2, 3, ... } is bounded below but not

above, hence N is not bounded. The open intervals ] 0, 1[, ]M , N [ and the

closed interval [ M , N ], where M , N  R are bounded, while they are

infinite sets.

- 13 -
1.3.2: Definition ( Upper bound and lower bound ): If A R is bounded above

then N is called an upper bound for A if x  N ,  x  A. If A R

is bounded below then M is called a lower bound for A if

x  M ,  x  A.

We often abbreviate upper bound and lower bound by u.b. and l.b.
respectively.

 n −1 
Thus, − 7 is an l.b. for N, and 1 is an u.b. for the set B =  : n  N .
 n 
Note that (1.3.1):

infinitely many numbers greater than − 7 are lower bounds for N , but that
there is no number less than 1 which is an upper bound for B. This leads up to
the concept of least upper bound and greatest lower bound.

1.3.3: Definition ( Least upper bound and greatest lower bound ): Let
  A R be bounded above. The number L is called the least upper

bound ( supremum) for A if ( 1 ) L is an upper bound for A, and (2) no

number smaller than L is an upper bound for A. Similarly, G is called the

greatest lower bound (infemum) for the set A bounded below, if (i ) G is a

lower bound for A, and ( ii ) no number greater than G is a lower bound for

A.

We abbreviate least upper bound as L = sup ( A ) = l. u. b. ( A ) and

greatest lower bound as G = inf ( A ) = g.l.b. ( A ).

- 14 -
 Now, we discuss that a non-empty subset   A R can have no more

than one l. u. b. ( if one exists ). For if L = sup ( A ) = l. u. b. ( A ) for A and

ML then by (2) M is not an upper bound for A. Moreover, if

M  L. then M cannot be a l. u. b. for A since L is an u.b. for A and

L  M . Similarly, no set can have more than one g. l. b. ( if one exists ).

Ex.1.3.1: Discuss the boundedness, of the set of reciprocal of the natural

numbers set:

1   1 1 
A =  : n  N  = 1, , , ...,
n   2 3 

then illustrate the L = sup ( A ) = l.u.b. ( A ) and G = Inf ( A ) = g.l.b. ( A ),

( if one exists ).

Sol. : It is obvious that A is bounded. With: L = sup ( A ) = l.u.b. ( A ) = 1,

which belongs to A, and G = Inf ( A ) = g.l.b. ( A ) = 0, which does not

belong to A. #

Theorem 1.3.1 ( Least upper bound ) : If

L = sup ( A ) = l. u. b. ( A ), and   0, then there is at least one

number a A such that:

L −   a  L  a  ] L − , L ].

Proof : Let   0. The condition aL is satisfied by all numbers

a A. All we have to show therefore is that there is some number a A

such that:

- 15 -
 L −   a.

Suppose on the contrary that there is no such number in A. We then

have:

x L −   x  A,

so that L − becomes an upper bound for A. But this cannot happen,

for it makes L − an upper bound that less than L and by

assumption, L is the least upper bound for A.

#

Note that (1.3.2):

from the theorem we just proved, we have: If

L = sup ( A ) = l. u. b. ( A ),

then A has at least one element in every half- open interval of the form:
]L − , L ],   0.

Similarly, the following theorem is obvious.

Theorem 1.3.2 (Greatest lower bound) :

If G = inf ( A ) = g.l.b. ( A), and   0, then there is at least one

number a A such that:

G  a  G + . #

- 16 -
 Exercises (1.3):

Discuss the boundedness of the following sets then illustrate the

L = sup ( A ) = l.u.b. ( A ) and G = inf ( A ) = g.l.b. ( A ), ( if one exists ):

( 1 ) The open intervals ] 0, 1[, ]M , N [ and the closed interval [ M , N ].

 n −1 
(2) B =  : n  N . ( 3 ) N = {1, 2, 3, ... }.
 n 

 ( −1) n
( 4 ) A1 = 
n

: n N . 
( 5 ) A2 = e − n : n N . 
 

 
( 7 ) A4 = ln n: n N.
1
( 6 ) A3 =  1 + : n N .
 n 

 ln (n + 2)   n +1 
( 8 ) A5 =  : n N . ( 9 ) A6 =  : n N .
 n+2   n 


(10 ) A7 = (0.9)n : n N .   n
(11) A8 = 
 n +1

: n N .


n 
(13 ) A10 = n n : n N .
 1
(12 ) A9 =  n : n N .
e   

- 17 -
1.4: Bounded Functions

1.4.1: Definition ( Bounded functions ): The real-valued function:

f : A → f [ A ],

is said to be bounded if its range f [ A] is bounded, as we have seen in the

boundedness of the set (cf.: 1.2: Bounded Sets ). That is: there exists a non-
negative real number M  0 , such that:

| f ( x )|  M ,  x  A = Dom ( f ).

Thus the functions: sin x, cos x are bounded, since they have the ( same )

range [−1, 1 ] , while the functions: ln x, e x , tan x, sinh x, cosh x are

unbounded.

Exercises (1.4):

Discuss the boundedness of the following functions:

( 1 ) sec x, csc x, tanh x, coth x, sech x, csch.

1 −1 x2
(2) , , , 1 + x 3, x 2 , | x | .
1 + x 2
1 + x2 1 + x2

1.5: The Completeness Axiom

Here, we illustrate the completeness axiom. The rational numbers Q and

real numbers R have similar algebraic and order properties (they are both

- 18 -
densely ordered fields). The crucial property that distinguishes R from Q is its
completeness. We begin with an example that illustrates the difference between
Q and R .

Ex.1.5.1: Define AQ by:


A = x  Q: x 2  2 . 
Then A is bounded from above by every M  Q+ such that M 2  2.
Nevertheless, A has no supremum in Q because 2 is irrational: for every

upper bound MQ there exists M ' Q such that 2 M'  M, so M

isn't a least upper bound of A in Q. On the other hand, A has a supremum in
R, namely;

L = sup ( A ) = l.u.b. ( A ) = 2. #

1.5.1: Definition ( Completeness axiom ): Every nonempty subset of real numbers

(   A  R) that is bounded from above has a supremum.

It follows easily from this axiom that: Every nonempty subset of real numbers
(   A  R) that is bounded from below has an infemum.

As a first application of this axiom, we prove that R has the Archimedean

property, meaning that no real number is greater than every natural number.

Theorem 1.5.1 ( Archimedean property ) : If x R, then there

exists n  N, such that x  n. ( That is N is unbounded above. )

- 19 -
Proof : Suppose, for contradiction, that there exists x R such that

x  n,  n  N . Then x is an upper bound of N, so N has a supremum

M = sup ( N ) = l.u.b. ( N )  R. Since n  M ,  n  N, we have

n −1  M −1,  n  N , which implies that n  M −1,  n  N . But

then M −1 is an upper bound of N , which contradicts the assumption that M

is a least upper bound. #

From Archimedean property, one easily gets the following results:

Corollary 1.5.1 (Reciproca ls of Archimedean property):

1
x  .such that n N , there exists  x  0,
n

1
That is, if x R such that x  ,  n  N, then x  0, while if
n
1
x  0, such that x  ,  n  N, then x = 0. Thus, we have,
n
directly, the following corollary:

Corollary 1.5.2

( 1 ) If x  y + ,    0  x  y,

( 2 ) If | x |  ,    0  x = 0.

Corollary 1.5.3

If    0, there exists n  N such that n − 1   n.

- 20 -
1.6: Dense Sets

1.6.1: Definition ( Dense Sets ): A set A is said to be dense ( or dense in R ) if

every open interval ] a, b [ contains a point of A.

Theorem 1.6.1 ( The rational numbers are densely ) : The set Q of

rational numbers is dense.

Proof : Let x y and consider the open interval ] x, y [. We must to find a

rational number belongs to ( inside ) this interval. By the reciprocals of

Archimedean property, Corollary 1.5.1 there is a positive integer n such that:

1
n  .
y−x

This means that:

ny  nx + 1.

Let m be chosen as the integer just less than nx +1, more precisely, we find an

integer mZ such that:

m  nx + 1  m + 1.

Thus accordingly, 1.1.2: Definition ( Ordered axioms) O4 :

m − 1  nx  ny, n  Z + ,

again using 1.1.2: Definition ( Ordered axioms) O4 and the initial assumption

that x  y, we directly obtain the result:

- 21 -
 m 1
x  x +  y,
n n

m
thus exhibiting a rational number in the interval. #
n
Corollary 1.6.1 (The irrational numbers are densely):

The set Q c of irrational numbers is dense.

Proof : Let x y and consider the open interval ] x, y [. Then, from the

last theorem: Theorem 1.6.1 ( The rational numbers are densely ) ,

we must have two rational numbers r1, r2  Q ( say ), such that:

x  r1  r2  y,

such that:

r2 − r1 r −r
  0  r1  r1 + 2 1  r2.
2 2
(The last ineq. Is left as an exercise to the reader.)

Thus, we get the result directly as

r2 − r1
x  r1  r1 +  r2  y,
2
such that:

r2 − r1
r1 + Qc .
2

- 22 -
 r2 − r1
thus exhibiting an irrational number r1 + Qc in the interval. #
2
1.7: Countable and Uncountable Sets

1.7.1: Definition ( Equivalent sets ): A set A is called equivalent to a set B, written

A ~ B, if there exists a function: f: A →B which is one-one and onto, that

is f: A →B is a one-to-one correspondence between the sets A and B.

1.7.2: Definition ( Denumerable ): A set A is called denumerable iff it is

equivalent to N, that is A ~ N.

1.7.3: Definition ( Denumerable and countable sets ): A set A is said to be:

( 1 ) Finite if it is the empty set or A ~ 1, 2, 3, ..., n , for some n  N,

( 2 ) Countably infinite ( or denumerable ) if A ~ N,

( 3 ) infinite if it is not finite,

( 4 ) Countable if it is finite or countably infinite,

( 5 ) Uncountable if it is not countable.

Ex.1.7.1: Discuss the equivalence of the natural numbers N , the odd

numbers N1 and the even numbers N2 then, determine whether or not

the sets N1, N2 are countable ( or denumerable ).

f : N1 → N2
Sol. : Consider the function:
: n → f (n) = n + 1,

where,

- 23 -
 N1 = 1, 3, 5, ..., N2 = 2, 4, 6, ....

One can easily show that f : N1 → N2 is a one-to-one correspondence,

thus:

N1 ~ N2.

g : N → N2
Similarly, the function:
: n → g (n) = 2n,

is a one-to-one correspondence, thus:

N ~ N2 ,

that is, all the subsets, N , N1 and N2 are equivalence, then they are

countable ( or denumerable ). ( Of course, N ~ N, and any subset

A N is also countable ( or denumerable ). The proof is left as an

exercise to the reader. ) #

Ex.1.7.2 : Determine whether or not, the integer numbers Z is

countable ( or denumerable ).

f: Z →N
Sol. : Consider the function: 2n, n  0,
: n → f (n) = 
1 − 2n, n  0,
where,

Z = 0, 1,  2, 3, ..., N = 1, 2, 3, 4, ....

- 24 -
One can easily show that f: Z→N is a one-to-one correspondence,

thus:

Z ~ N.

And hence, Z is countable ( or denumerable ). #

Theorem 1.7.1 ( Asubset of countable ) : Every subset of countable

set is countable.

Proof : Let B  A, where A is countable. If B is finite then it is immediately

countable. Let us now consider that B is infinite subset of A, that is A is

infinite countable ( denumerable ). Thus, there is a one-to-one correspondence
function:

f : N → A  N ~ A.

Since f: N →A is a one-to-one correspondence function, then,

f −1 : A → N exists and is also a one-to-one correspondence function. On the

other hand:

f −1[ B ]  f −1[ A ]  N ,

which means that f −1[ B ] is countable.

But:

f : f −1[ B ] → B,

that is,

f −1[ B ] ~ B,

- 25 -
which means that B is countable. #

Next, we prove some results about countable sets. The following

theorem states a useful necessary and sufficient condition for a set to be
countable.

Theorem 1.7.2 ( Countable and onto function ) :. A non-empty

set A is countable iff there is an onto function: f : N → A.

Proof : ( is omitted ). #

Theorem 1.7.3 ( Countable of A  B ) : If A, B are countable sets,

then A B is also countable set.

Proof : Firstly, Let A, B are countable infinite ( or denumerable ). Then

there exist two one-to-one correspondence function:

f : A → N, f : B → N.

We consider the function:

h: A  B → N,
: ( x, y ) → h( x, y ) = 2 f ( x) 3g ( y).

One can easily show that h is a one-one function into N, that is:

h : A  B → h( A  B) = Rang( h )  N .

Here:

A  B ~ h( A  B) = Rang( h )  N .
- 26 -
that is, A B is also countable.

Finally, suppose that one of the two sets A, B is finite, say, A is finite and is
given as:

A = a1, a2 , a3 , ..., an ,

then, the set:

C = a1, a2 , a3, ..., an , n +1, n +1, ... ~ N ,

which means that C countable infinite ( denumerable ), thus C B countable

infinite ( denumerable ). Since:

A B  C B

which means that A B is countable. #

Theorem 1.7.4 ( Union of countable) :. A countable union of

countable sets is countable.

Proof : Let X n : n  N . be a countable collection of countable sets.

Then, there is an onto function: f n : N → X n.

Consider the following function:

g : N  N →  n N X n
: ( n, k ) → f n ( k ).

Then g is also onto.

Since:

- 27 -
 N ~ N  N.

Then there exist two one-to-one correspondence function:

h: N → N  N,

and it follows that:

g  h : N →  nN X n

is onto, this implies that the union of the Xn is countable. #

Theorem 1.7.5 . The rational numbers set is countable.

Proof : Consider the function:

f : Q → Z  Z,
m
: → ( m, n ),
n
With no common factors between m, n. One can easily show that f is

one-one, then:

Q ~ f [ Q ] = Rang( f )  Z  Z ,

where Z Z is countable, then Q is also a countable set. #

- 28 -
 Chapter 2

Limits and Continuity

2.1: The Limit of Function

2.1.1: Definition ( The limit of a function ): The function f (x) has the limit
L as x tends to a, and we write:

for each  o there exists  =  ()  o


 then
lim f ( x) = L  such that: if 0  | x − a |    | f ( x) − L | .
x→a
 whenever
or: | f ( x) − L |   0  | x − a |  

Examples:

Ex. 2.1.1: Show that ( discuss, or prove, or study ):

lim ( 2x − 1) = 3
x→2

Sol. : “ Finding a  =  ()  0 ”

Let   0, and | x − 2|   . We seek a number   0 such that:

if 0  | x − 2 |   , then | (2 x − 1) − 1|  .

What we have to do first establish a connection between:

| (2 x − 1) − 3| and | x − 2|   .

- 29 -
The connection is simple:

| (2 x − 1) − 3| = 2 | x − 2 |  2 = .


This suggests that we choose:  =  () =  0. #
2
Ex.2.1.2 : Discuss: lim | x | = | a |.
x→a

Sol. : “ Finding a  =  ()  0 ”

Let   0, and |x − a|  . We seek a number   0 such that:

| f ( x) − L |    0  | x − a |  

That is: if 0  | x − a |   , then | | x | − | a | |  .

Since: | | x | − | a | |  | x − a |   =  () = . #

Ex. 2.1.3 : Discuss: lim x = a.

x→a

Sol. : “ Finding a  =  ()  0 ”

Let   0, and |x − a|  . We seek a number   0 such that:

if 0  | x − a |   , then | x − a |  .

Obviously we can choose  =  () = . #

Ex. 2.1.4 : For each constant c, discuss: lim c = c.

x→a

Sol. : Here we are dealing with the constant function f ( x ) = c.

- 30 -
“ Finding a  =  ()  0 ”. Let   0, and |x − a|  . We

seek a number   0 such that:

if 0  | x − a |   , then | c − c |  .

Since | c − c | = 0, we always have:

| c − c |  ,

no matter how  is chosen; in short, any positive number will do for

. #

Ex. 2.1.5 : Discuss: lim x 2 = 9.

x→3

Sol. : “ Finding a  =  ()  0 ”

Let   0, and | x − 3|   . We seek a number   0 such that:

if 0  | x − 3|   , then | x2 − 9 |  .

The connection between | x − 3| and | x2 − 9 | can be found by

factoring:

x2 − 9 = ( x + 3 )(x − 3 ),

and thus;

| x 2 − 9 | = | x + 3| | x − 3| .

At this point, we need to get an estimate for the size of | x + 3| for x

close to 3. For convenience, we will take x within one unite of 3.

- 31 -
 If | x − 3| 1, thin | x2 − 9 |  7 | x − 3|  7  = .

This means that we can let


 = min . (1, ). #
7

Ex. 2.1.6 : Discuss: lim x = 2.

x→4

Sol. : “ Finding a  =  ()  0 ”

Let   0, and | x − 4|   . We seek a number   0 such that:

if 0  | x − 4 |   , then | x − 2 |  .

To be able to form x , we need to have x  0. To ensure this, we must

have   4.

Remembering that we must have   4, let us move on to find a

connection between | x − 4 | and | x − 2 | . With x  0, we can

form x and write:

x − 4 = ( x )2 − 4 = ( x + 2 )( x − 2 ).

Thus, we have;

| x − 4 | = | x + 2 | | x − 2 |.

Since: x + 2  2  1, it follows that:

- 32 -
 | x − 2 |  | x − 4 |   = .

But remember the requirement   4. We can meet both requirements

on

 by setting  = min . (4,  ). #

Ex.2.1. 7 : Discuss:

−1, x  0,
if : f ( x) = 
 1, x  0,

then lim f ( x ) does not exist.

x→0

Proof : Suppose on the contrary that:

lim f ( x ) = L,
x→0

then there exist  =  ()  0, such that:

if 0  | x − 0 |   , then | f ( x ) − L |  ,    0.

Set  = 1( say ) :

if x  0, then | f ( x ) − L | = |1− L |  = 1  L  0, (positive ),

if x  0, then | f ( x ) − L | = | −1− L |
= |1 + L |  = 1  L  0, (negative ).

Clearly no such number L exists. #

- 33 -
 1
Ex. 2.1.8 : Prove that: lim x sin =1.
x→0 x
Sol. : “ Finding a  =  ()  0 ”

Let   0, and |x |  . We seek a number   0 such that:

1
if 0  | x |   , then | x sin |  .
x

1 1
Since: | x sin | = | x | | sin |  | x |   =  () = . #
x x

1
( What about: lim x sin =1. )
x→ x
Theorem 2.1.1 ( Uniqueness of the limit ) : If the limit exists

then it is unique, that is:

If : lim f ( x ) = L, and lim f ( x ) = M  L = M .

x→a x→a


Proof : (Using the  −  tecnique ). Let   0   0, and
2
assume that:


for each  o there exists 1 = 1()  o,
lim f ( x ) = L   such that:
x→a
  whenever
 | f ( x) − L |   0  | x − a |  1.
 2

- 34 -
Also:


for each  o there exists  2 =  2 ()  o,
lim f ( x ) = M   such that:
x→a
  whenever
 | f ( x) − M |   0  | x − a | 2
 2

Thus we have both:

 
| f ( x) − L |  , | f ( x) − M |   0  | x − a |  min .(1,  2 )
2 2
Using the triangle inequality. It follows that:

| L − M | = | L − f ( x) + f ( x) − M |  | f ( x) − L | + | f ( x) − M |
 
 + = .
2 2
In short we have found that:

| L − M |  .

But since  is an arbitrary, | L− M | is a nonnegative number which is

strictly less than every positive number. Think about it, and you will see
that | L− M | must be zero, that is;

|L− M | = 0  L = M.

Hence if the limit exists then it is unique. #

- 35 -
Theorem 2.1.2 ( Properties of the limits ) :

If : lim f ( x ) = L, and lim g ( x ) = M , then :

x→a x→a

( i ) lim ( f ( x ) + g ( x ) ) = L + M ,
x→a
( ii ) lim ( a f ( x ) ) = a L,
x→a
( i ii) lim ( f ( x ) g ( x ) ) = L M ,
x→a
f ( x) L
( iv ) lim = , M  0.
x→a g( x ) M

Proof : ( The proof is omitted ). #

Here are the appropriate −  definition of the two - one sided limits:

The right-hand limit, is denoted by: lim f ( x) ( or lim f ( x) ), and

x → a+ x a

left-hand limit, is denoted by: lim f ( x) ( or lim f ( x) ).

x → a− xa

2.1.2: Definition ( Right – hand limit ): Let f be a function defined at least

on an interval of the form a, c , then:

for each   o there exists  =  ()  o
lim f ( x) = L   then
x → a+ such that: if a  x  a +   | f ( x) − L |  .

2.1.3: Definition ( Left – hand limit ): Let f be a function defined at least

on an interval of the form d , a .

- 36 -
 for each   o there exists  =  ()  o
lim f ( x) = L   then
x → a− such that: if a −   x  a  | f ( x) − L |  .

−1, x  0,
Ex. 2.1.9 : If : f ( x) = 
 1, x  0,

then: lim f ( x) = 1  − 1 = lim f ( x),

x → 0+ x → 0−

although, as you saw before, lim f ( x) does not exist.

x→0

Ex. 2.1.10 : Here we consider the square-root function. Discuss:

lim x = 0.
x → 0+

Sol. : “ Finding a  =  ()  0 ”

Let   0, and | x |   , x  0. We seek a number   0 such

that:

if 0  | x |   , then | x |  .

Since:

0  | x |   , x  0  0  x    0  x   =  () = . #

Since x is not defined for x  0, we have no basis for taking a left-

hand limit at 0.

- 37 -
Ex. 2.1.11: Discuss:

2
lim −1 x
= 2.
x→0 1 + e
+

Sol. : “ Finding a  =  ()  0 ”

Let   0. Here, x  0. We show that:

2 whenever
−1 x
− 2   0  x   ().
1+ e

2 2 − 2 − 2e −1 x 2 2
 −1 x
− 2 = −1 x
= = 
1+ e 1+ e 1+ e1x
1+ e1x

2 2 1 2
 1+ e1 x   e1 x  −1   ln ( −1)
  x 
1
 0 x  =  (), 0  2. #
2
ln ( −1)

Ex. 2.1.12 : An interesting function to look at when dealing with one-
sided limits is the greatest integer function ( step function ):

The greatest integer function ( step function ), denoted by f ( x) = [ x]

and is given as: f ( x) = [ x] := greatest integer  x.

It is easy to see that for each integer n:

lim [ x] = n, lim [ x] = n −1.

x → n+ x → n−

- 38 -
For example: lim [ x] = 2, lim [ x] = 1.
x → 2+ x → 2−

If c is a number which is not an integer, then c lies between consecutive

integers n and n + 1. For such c, we have both:

lim [ x] = n = lim [ x].

x → c+ x → c−

For example: lim [ x] = 1 = lim [ x] ,

3+ 3−
x→ x→
2 2

and: lim [ x] = 2 = lim [ x]. #

9+ 9−
x→ x→
4 4

One-sided limits give us a simple way of determining whether or not a (

two-sided ) limit exists, namely:

iff
lim f ( x) = L  lim f ( x) = L = lim f ( x) .
x→a x → a+ x → a−

−1, x  0,
Ex. 2.1.13 : For : f ( x) = 
 1, x  0,

we have: lim f ( x) = 1  − 1 = lim f ( x).

x → 0+ x → 0−

Since the two one-sided limits are not equal, then:

lim f ( x) does not exist. #

x→0

- 39 -
 
 x2, x  0,
Ex. 2.1.14 : With : f ( x) = 

 x, x  0,

we have both: lim f ( x) = 0 = lim f ( x).

x → 0+ x → 0−

Consequently: lim f ( x) = 0. #
x→0

2.1.4: Definition ( Infinity ): We write:

for each M  0 there exists  =  (M )  0

(1) lim f ( x) =    then
x→a
such that: if 0  | x − a|    f ( x)  M .

for each M  0 there exists  =  (M )  0

(2) lim f ( x) = −    then
x→a
such that: if 0  | x − a|    f ( x)  − M .

for each   0 there exists K = K ()  0

(3) lim f ( x) = L   whenever
x→
such that: | f ( x) − L |    x  K ().

for each   0 there exists K = K ()  0

(4) lim f ( x) = L   whenever
x → −
such that: | f ( x) − L |    x  − K ().

for each M  0 there exists K = K (M )  0

(5) lim f ( x) =    whenever
x→
such that: f ( x)  M  x  K (M ).

- 40 -
Ex. 2.1.15 : Discuss:

1
lim = .
x → 1 ( x −1 ) 4

Sol. : “ Finding a  =  (M )  0 ”

Let M  0, we show that:

1
 M   0  | x −1|   (M ).
( x −1 ) 4

Since:

1 2 ]2  1  0  ( x −1 )2  1
 M   0  [( x − 1 )
( x −1) 4 M M

1
 0  | x −1|  =  (M ). #
4
M
Ex. 2.1.16 : Discuss:

1
Lim 2
= .
x→0 x
Sol. : “ Finding a  =  (M )  0 ”

Let M  0, we show that:

1
2
 M   0  | x |   ( M ).
x
Since:
- 41 -
 1 2  1  0  | x |2  1
 M   0  x
x2 M M

1
 0 | x|  =  ( M ). #
M

Ex. 2.1.17 : Discuss:

x +1 x +1
(i) lim = 1, (ii) lim = 1.
x→ x −1 x→− x −1

Sol. : ( i ) “ Finding a K = K ()  0 ”

Let   0, we show that:

x +1
− 1    x  K ().
x −1

Since:

x +1 2 2
−1 =   x −1  ,
x −1 x −1 

and:

2 2
x − 1  | x | + 1  | x | + 1 x −1   | x |  − 1 = K ().
 
Similarly, for (ii).

- 42 -
Ex. 2.1.18 : Discuss:

1 1
(i) lim = , (ii) lim = − .
+
x→0 x −
x→0 x

1
(Thus: lim , does not exist.)
x→0 x

1
Sol. : (i) lim = ,
+
x→0 x

“ Finding a  =  (M )  0 ”

Let M  0. Here, x  0. We show that:

1
 M  0  x   (M ).
x
Since:

1 1 1
 M   0 x   0  | x|  =  (M ).
x M M

1
(ii) lim = − . “ Finding a  =  (M )  0 ”
x → 0− x

Let M  0. Here, x  0. We show that:

1
 − M  −  (M )  x  0 .
x
Since:

- 43 -
 1 1
 −M  0  x  − =  (M ) . #
x M

Ex. 2.1.19 : Discuss:

1 1
(i) lim = 0, (ii) lim = 0.
x→ x x → − x

1
Sol. : (i) Lim = 0. “ Finding a K = K ()  0 ”
x→x

Let   0. We show that:

1
   x  K ().
x

Since:

1 1
   | x | = x  = K ().
x 

1
(ii) lim = 0. “ Finding a K = K ()  0 ”
x →− x

Let   0. We show that:

1
   x  − K ().
x

Since:

- 44 -
 1 1 −1
   |x| = − x   x  = − K (). #
x  

1
Ex. 2.1.20 : Prove that: lim x sin =1.
x→ x

1
Sol. : Set: u =  u → 0+ as x → ,
x

1 sin u
then: lim x sin = lim = 1.
x→ x x→0 + u

1 sin u
Also: lim x sin = lim = 1. #
x→− x x→0 − u

Ex. 2.1.21: Discuss:

(i) lim a x = , a  1, (ii) lim a x = , a  1.

x→ x → −

Sol. : (i) lim a x = , a  1. “ Finding a K = K (M )  0 ”

x→

Let M  0, we show that:

a x  M  x  K (M ).

Since:

ln M
a x  M  x ln a  ln M  x  = K ( M ), a  1.
ln a

Similarly, for (ii). #

- 45 -
Ex. 2.1.22 : Discuss, the following limits of the sequences:

 
 n +1  (−1)n 
(i)   , (ii)   .
 n n = 1  n n = 1

n +1 (−1) n
(i) lim = 1, (ii) lim = 0.
n→ n n→ n

n +1
Sol. : (i) lim = 1. “ Finding a N = N ()  0 ”
n→ n
Let   0. We show that:

n +1
− 1    n  N ().
n

Since:

n +1 1 1
− 1       n  = N ().
n n 

(−1)n
(ii) Lim = 0. “ Finding a N = N ()  0 ”
n→ n
Let   0. We show that:

(−1)n
   x  N ().
n

Since:

- 46 -
 (−1)n 1 1
      n  = N (). #
n n 

Some important limits: Our purpose here is to familiarize us with

some limits that are particularly important in calculus and to give us more
experience with limit arguments which we need in real analysis:

1
(i ) lim x n = 1, x  0.
n→

Proof : Since:

1
n=
1 1
ln lim x lim ln x n = ln x lim =0
n→ n→ n→ n
1
 lim x n = e0 =1, x  0. #
n→

(ii ) lim xn = 1, | x |  1.
n→

Proof : Take | x |  1 and observe that | x | 

n 
n =1 is a decreasing seq.,

where: | x |n +1 = | x | | x |n  | x |n , | x |  1

Now let   0. Then from (i ) we have:

1
lim  n =1.
n→

Thus there exists a positive integer k such that:

- 47 -
 1
| x|   k  | x |k = | xk |  .

That is:

| xk |    nk #

xn
(iii ) lim = 0,  x  R.
n →  n!

Proof : ( is omitted ).

1
(iv ) lim = 0,    0.
n →  n

Proof : Since   0, there exists an odd positive integer p such that:

1
 .
p

Then:

 1
1 1 1 p
0  
=      .
n n n
Since:

1
1 1 p 1
lim = 0  lim   = 0  lim = 0. #
n→n n→ n n →  n
 
n
 x
( v) lim 1 +  = ex.
n→
 n

- 48 -
Proof : For x = 0, the result is obvious. For x  0, we have:

 x
ln 1 + 
n
 x  n
ln lim 1 +  = x lim
n→ n→ x
 n
n
n
 x
= x  lim 1 +  = e x. #
n→
 n

Exercises (2.1):

(1) Finding the limit, then discuss it: lim ( 3x − 5 ), lim ( 8 − 4x )

x →1 x→−3

lim ( 2x − 1) = 3
x→2

(1) Evaluate the limits that exists:

(i) lim [2x ], (ii) lim [2x ], (iii) lim [2x ],

1+ 1− x→
1
x→ x→ 2
2 2

(iv) lim [2 x ], (v) lim [2 x ], (vi) lim [2 x ].

3+ 3− x→
3
x→ x→ 4
4 4

- 49 -
2.2: Continuity

2.2.1: Definition ( The continuity at a point ): Let f (x) be a function

defined at least on an open interval a −  , a +  . The function f (x)

is said to be continuous at the point a iff:

lim f ( x) = f (a),
x→a

that is, the necessary and sufficient conditions of continuity of a function


(i) f (a) exists  f is defined at a,
(ii) lim f ( x) = L, exists,
f (x) at a are:  x→a

(iii) lim f ( x) = L = f (a).
 x→a

It is obvious that (iii) guarantee (i), (ii).

2.2.2: Definition ( −  technique of continuity at a ) : The

function f (x) is continuous at a and we write:

for each  o there exists  =  ()  o


 then
lim f ( x) = f (a)  such that: if | x − a |    | f ( x) − f (a) | .
x→a
 whenever
or: | f ( x) − f (a) |   | x − a |  

Note that (2.2.1):

, if g(x) is continuous at a and f (x) is continuous at g (c), then the

composition ( f  g )( x) = f ( g ( x)) is continuous at a.

- 50 -
 If the domain of the function f (x) contains an interval

a −  , a +  , then f (x) can fail to be continuous ( discontinuous ) at

a only for two reasons: either f (x) does not have a limit as x tends to

a ( in which case a is called an essential discontinuity), or it does have a

limit, but the limit is not f (a). In this case a is called a removable

discontinuity; the discontinuity can be removed by redefining f (x) at a.

If the limit is L define f (x) at a to be L.

Most of the functions that we have encountered so far are continuous

at each point of their domains. In particular, this is true for polynomials,
rational functions ( quotient of polynomials ), and absolute-value functions.

With f ( x) and g ( x) continuous at a, we have:

lim f ( x) = f (a), and lim g ( x) = g (a),

x→a x→a

and thus, by the limit theorems:

( i ) lim [ f ( x) + g ( x) ] = f (a) + g (a),

x→a
( ii ) lim (  f ( x) ) =  f (a),
x→a
( i ii) lim [ f ( x) g ( x) ] = f (a) g (a),
x→a
f ( x) f (a)
( iv ) lim = , g (a)  0.
x→a g ( x) g (a)

This tells us that, if the functions f and g are continuous at a, then

f + g,  f , f g , and f / g , g ( a )  0 are also continuous

at a.

- 51 -
Ex. 2. 2.1: Show that f ( x) = x2, is continuous ( on R ) at x = 2.

Sol. : (i ) It is obvious that, the above conditions of continuity are

satisfied, namely; lim x2 = 4 = f (2).

x→2

(ii) −  technique of continuity at a = 2 : discuss:

lim x 2 = 4 = f (2).
x→2

“ Finding a  =  ()  0 ”

Let   0, and | x − 2|   . We seek a number   0 such that:

if 0  | x − 2 |   , then | x 2 − 4 |  .

if 0  | x − 2 |    −   x − 2   , thus by choosing
  1  1  x  3  3  x + 2  5  | x + 2 |  5.

Hence: | x2 − 4 | = | x + 2 | | x − 2 |  5 = . #

2.2.3: Definition ( One − sided continuity) : A function f (x) is

called continuous from the right at a iff: lim f ( x) = f (a).
x → a+

It is called: continuous from the left at a iff: lim f ( x) = f (a).

x → a−

Thus, it is clear that a function f (x) is continuous at a iff it is

continuous there from both sides; that is:

- 52 -
 lim f ( x) = f ( a ) = lim f ( x) .
x → a+ x → a+

At each integer the greatest-integer function: f ( x) = [ x],

is continuous from the right but discontinuous from the left.

2.2.4: Definition ( Continuity on an int erval ) : A function f (x) is

said to be continuous on the open interval ]a, b[ if it is continuous at each

point x0  ]a, b[. On particular; f (x) is said to be continuous on the

closed interval [a, b] iff:

(i) lim f ( x) = f ( xo ), a  xo  b, )
x → xo

(ii) lim f ( x) = f (a), is continuous from the right at a,

x → a+
(iii) lim f ( x) = f (b), is continuous from the left at b.
x → b−

- 53 -
 Chapter 3

Infinite Sequences
3.1: Definition ( Infinite sequences ): An infinite sequence is a function
denoted by:

sn  s(n)  f (n),

whose domain is the set of natural numbers N into R that is:

sn : N → R
: n → sn  R.

Here, sn is called the nth term of the given sequence, which plays a very
important role to study ( monotonicity, convergent, boundedness, … )
infinite sequences.

Often, we write an infinite sequences, as: sn n = 1, where:

sn n = 1 = s1, s1, s1, s1, ..., sn , ....
It is obvious that, the range of an infinite sequence is an ordered infinite set
which may be repeated.

Now we give some examples on infinite sequences to illustrate:


1  1 1 1 1  1
(1)   =  1, , , , ..., , ..., sn = , (Harmonic seq.)
 n n = 1  2 3 4 n  n

- 54 -
   = −1,1, −1, 1, ..., (−1) , ..., s = (−1) , (Oscillating seq.)
(2) (−1) n

n =1
n
n
n


 1   1 1 1 1  1
(3)  p  = 1, p , p , p , ..., p , ..., p  R, sn = p , ( P - seq.)
 n n = 1  2 3 4 n  n
 
(4) x n

n =1
= x, x 2
, x 3
, ..., x n

, ... , sn = xn , (G. seq.)

 (−1)n   1 1 1 (−1)n  (−1)n
(5)   =  −1, 2 , − 3 , 4 , ..., n , ..., sn = .
 n n = 1   n


  −1 n   −1  −1 2  −1 3  −1 
n   −1 
n
(6)     =  2 ,  2  ,  2  , ...,  2  , ..., sn =  2  .
   n = 1
2          

 
(7) en

n =1
= e, e 2
, e 3
, ..., e n

, ... , sn = en .


 sin n   sin 2 sin 3 sin n  sin n
(8)   =  sin 1, , , ..., , ... n
, s = .
 n n = 1  2 3 n  n

(9) cos n  n = 1 = −1,1, −1, ..., cos n  , ..., sn = cos n  (−1)n .


 −2 n   −2  −2 2  −2 3  −2 
n   −2 
n
(10)    =  ,   ,   , ...,   , ..., sn =   .
 3  n = 1  3  3   3   3   3

 n +1   3 4 n +1  n +1
(11)   =  2, , , ..., , ... n
, s = .
 n n = 1  2 3 n  n

3.2: Definition ( Bounded sequences ): An infinite sequence sn n = 1, is

said to be bounded, if:
- 55 -
 | sn |  M , M  0,  n  N .

3.3: Definition ( Monotonic sequences ): An infinite sequence sn n = 1, is

said to be:

(i) increasing iff sn  sn + 1,

(ii) nondecreasing iff sn  sn + 1,
(iii) decreasing iff sn  sn + 1,
(iv) nonincreasing iff sn  sn + 1,  n  N .

If any of these four properties holds, the sequence is said to be monotonic.

If otherwise it said to be not monotonic.


 n +1   3 4 n +1 
Ex. 3.1: The sequence:  n  =  2 3
2, , , ..., , ..., is
 n = 1  n 
decreasing. Also, it is bounded below by 1 and above by 2.

n +1 1 1
Sol. : Since: sn = =1+ 1+ = sn +1. #
n n n +1


 n  1 2 3 n 
Ex. 3.2 : The sequence:  n +1  = 2 3 4
, , , ..., , ..., is
 n = 1  n +1 

increasing. Also, it is bounded below by

1 and above by 1 .
2

- 56 -
Sol. : Since:

n 1 1
sn = =1− 1− = sn +1. #
n +1 n +1 n+2


 2n   23 2 4 2n 
Ex. 3.3 : The sequence:  n!  =  2, 2, 3! , 4! , ..., n! , ..., is
 n = 1  
nonincreasing and is bounded below by 0 and above by 2. It decreases for

n  2.

Sol. : Since: The first two terms are equal. On the other hand, for n2
the

sequence decreases, where:

2n 2 2n 2n + 1
sn =  = = sn +1,  n  2. #
n! ( n + 1) n! ( n + 1)!

Ex. 3.4 : The oscillating sequence

  −1,1, −1,1,..., (−1) , ...,

n 
(−1) n = 1 = n

is not monotonic and is bounded below by −1 and above by 1 .

Sol. : Since: sn  or  sn +1. #


1  1 1 1 1 
Ex. 3.5 : The harmonic sequence   =  2 3 4
1, , , , ..., , ...,
 n n = 1  n 

- 57 -
is decreasing and is bounded below by 0 and above by 1 .

1 1
Sol. : Since: sn =  = sn +1. #
n n +1

Exercises (3.1):

Determine the boundedness and the monotonicity of the indicated sequences:

 
 (−1)    −1 n 
 
n
( 1) 

 , (2)     , (3) n n =1
,
 n n = 1   2  n = 1

  
 n2   n +1   4n 
( 4)   , (5)  2  , (6)   ,
 n + 1 n = 1  n n = 1  4n + 1 n = 1
2


( 7) (0.9)n  
n =1
,  
(8) 2n

n =1
,  
(9) e −n

n =1
,

 


( 10) (−1)n n  
n =1
,
 n2 +1 
(11) 
3n + 2

 , (12) ln

 2n  

 n + 1
  ,
 n = 1
 n = 1

 
 4n 
( 13)  n  ,
 2n 
(14)  n  ,  
(15) e−n

n =1
,
 2 + 100 n = 1  4 + 1 n = 1

 
 2n −1  1 2n − 1 
( 16)  n  , (17) cos n n = 1, (18)  sin  ,
 2 n = 1 n 2 n = 1

- 58 -
   
 n + 1   ln( n + 2)    n + 1  
( 19)   , (20)   , ( 21)  ln    ,
 n n = 1  n + 2 n = 1   n  n = 1

  
 1 1 1  1010 n 
( 22)  4 −  , (23)  −  , (24)   ,
 n 
n =1
 n n +1 n = 1  n + 1 n = 1

  
n   n 
(26 )  ,
1
( 25 )  n  , n
n
 ( 27)  2  .
 e n =1  n =1  n +1 n = 1

3.4: Definition ( Convergent sequences ): An infinite sequence sn n = 1,

that has a limit L is said to be convergent to L. That is the seq.

sn n = 1, is convergent to L if:

Lim s = L  R.
n→ n

If otherwise, it said to divergent.

Note that (3.1):

A sequence cannot converge to more than one limit, for the uniqueness of
the limit, if it exists.


1
Ex. 3.6 : Discuss the convergence of the harmonic sequence: n .
 n = 1

1
Sol. : Since: lim sn = lim = 0 R,
n→ n→ n
- 59 -
 
1
thus the harmonic sequence: n is convergence. Let us now study
 n = 1
1
the previous limit: lim = 0.
n→ n
Let   0. “ Finding a N = N ()  0 ”. We show that:

1
   n  N ().
n

Since:

1 1 1
      n  = N (). #
n n 

Ex. 3.7 : Discuss the convergence of the sequence: sn n = 1, sn = 1.



Sol. : Since: lim s = lim 1 = 1 R,

n→ n n→

thus the sequence: sn n = 1, sn = 1 is convergence. Let us now examine

that: lim 1 = 1.
n→

Let   0. “ Finding a N = N ()  0 ”. We show that:

1 − 1    n  N (). ( *)

Since:

1−1 = 0  ,   0  n  1.
- 60 -
Thus, in this case, for any  0 we can make (*) hold by taking

N = 1, that is all positive numbers will do for N. ( This is one of the rare

cases where N does not depend on  ). #

Ex. 3.8 : Discuss the convergence of the sequence: n n = 1 .



Sol. : Since: lim s = lim n = ,

n→ n n→

thus the sequence: n n = 1 is divergence. We shall prove this seq. does not
have a limit. Assume the contrary. Then:

lim s = lim n = L,
n→ n n→

Then  0 there is an N = N ()  0 such that:

sn − L    n  N ().

Set  = 1 ( say ) that is:

sn − L  1  n  N ().

This is equivalent to:

−1 sn − L  1  n  N (),

−1 n − L  1  n  N (),

L −1 n  L + 1  n  N ().

- 61 -
The last statement says that all values of n greater than N () lie between
L −1 and L + 1. This is clearly false and the contradiction shows that

n n = 1 does not have a limit. #

The last example shows that a seq. whose terms get “ too big ” cannot
have a limit. This is not the only kind of seq. that does not have a limit.
Consider the following (oscillating) seq.

Ex.3.9 : Discuss the convergence of the oscillating seq.: (−1) 

n 
n =1
.

−1, n is an odd,
Sol. : Since: lim sn = lim (−1)n = 
n→ n→
 1, n is an even.
That is, the limit depends on the path ( n even or odd) thus the limit does

not exist. Hence the seq. is divergent. Now let us prove this seq. does not
have a limit. Assume the contrary. Then:

lim sn = lim (−1)n = L  R,

n→ n→

Then  0 there is an N = N ()  0 such that:

(−1)n − L    n  N ().

1
Set  = ( say ), that is, for n even we have:
2

1 1 3
1− L   L  .
2 2 2
While, for n odd we have:
- 62 -
 1 −3 −1
−1− L = 1 + L   L .
2 2 2

This is contradiction, which shows that  n 


(−1) n = 1 does not have a limit.
On the other hand, one can easily gets a contradiction from the last two
inequalities as:

1 1
2 = | 2 | = 1 + L + 1 − L  1 + L + 1 − L  + = 1,
2 2
which is also, another contradiction. #


 1
Ex. 3.10 : Discuss the convergence of the sequence: ln  .
 n n = 1

1
Sol. : Since: lim sn = lim ln = − ,
n→ n→ n

 1
thus the seq.: ln n  is divergent. Let us now discuss the limit:
 n = 1

1
lim sn = lim ln = − ,
n→ n→ n
Let M  0. “ Finding a N = N ( M )  0 ”. We show that:

1
ln  − M  n  N ( M ).
n
Since:

- 63 -
 1
ln = − ln n  − M  ln n  M  n  eM = N (M ). #
n

Ex. 3.11: Discuss the convergence of the geometric sequence: x n   

n =1
.

 , if x  1,
 0, if | x |  1,
Sol. : Since: lim sn = lim xn = 
n→ n→
 1, if x = 1,
does not exist, if x  −1,

thus the geometric sequence  

xn

n = 1 convergent to
0 if | x |  1, and

convergent to 1 if x = 1. If otherwise, it diverges. #

Theorem 3.1 ( Convergent and boundedness ) : Every

convergent sequence is bounded. That is:

If : lim s = L  R  sn  M , M  0,  n N.
n→ n

Proof : Assume that:

lim s = L  R.
n→ n

Then  0 there is an N = N ()  0 such that:

sn − L    n  N ().

This means that:

- 64 -
 sn = sn − L + L  sn − L + L   + L  n  N ().

Set:

m = max.(| s1 | + | s2 | + | s3 | + ... + | sn | ),

and, consequently,

sn   + L + m ,  n  N. #

Since every convergent sequence is bounded, a sequence which is not bounded

cannot be convergent. Namely, every unbounded sequence is divergent. The
sequence:

 
n  n 2 
  ,   , n ln n n = 1,
 2 n = 1  n + 1 
n =1

are all unbounded. Each of these sequences is therefore divergent.

Boundedness does not imply convergence. As a counterexample, consider the

oscillating seq.:

 
(−1) n + 1 n = 1 = 0, 2, 0, 2, ... .


This seq. is certainly bounded ( above by 2 and below by 0 ), but obviously it

does not converge: for, it has no a limit.

On the other hand, boundedness, together with monotonicity does imply

convergence.

- 65 -
Theorem 3.2 ( Boundedness, monotonicity and convergent ) :

A bounded nondecreasing sequence converges to the least upper bound of its

range, and a bounded nonincreasing sequence converges to the greatest lower
bound.

Proof : Suppose thatsn n = 1 is bounded and nondecreasing sequence. If

L = sup sn n =1 = l.u.b.sn n =1 is the least upper bound of the range of
the seq. Then:

sn  L,  n  N.

Now let   0, be an arbitrary positive number. Then there exists sk such that:

L −   sk ,

where, k = k (). Since the seq. is nondecreasing, we have:

sk  sn  n  k.

It follows that:

L −   sn  L  0  L − sn    0  | sn − L |  ,  n  k ().

Hence:

lim s = L  R.
n→ n

In a similar manner, we show that, a bounded nonincreasing sequence converges

to the greatest lower bound:

- 66 -
Suppose that sn n = 1 is bounded and nonincreasing sequence. If
G = inf ( A ) = g.l.b. ( A ). is the greatest lower bound of the range of the seq.

Then:

sn  G,  n  N.

Now let   0, be an arbitrary positive number. Then there exists sk such that:

sk  G + ,

where, k = k (). Since the seq. is nonincreasing, we have:

sk  sn  n  k.

It follows that:

G  sk  G +   0  sn − G    0  | sn − L|  ,  n  k ().

Hence:

lim s = G  R. #
n→ n

When a continuous function is applied to a convergent seq., the result is itself

a convergent seq. More precisely, we have the following theorem.

Theorem 3.3 ( Continuous function and convergent seq. ) :

Suppose that the seq., sn n = 1 is convergent to L, that is:

Lim s = L  R,
n→ n

- 67 -
and that, sn ,  n  N is in the domain of f (x). If f (x) is continuous

at L, then the seq.  f ( sn )n = 1 is convergent to f ( L ) that is:

lim f ( sn ) = f ( L )  R. :
n→

Proof : Assume that f is continuous at L and take   0. From the

continuity of f at L, hence there exists  =  ()  0 such that:

if x − L    f ( x) − f (L)  .

Since:

lim s = L R,
n→ n

that is,  0 there is an N = N ()  0 such that:

if n  N ()  sn − L  .

It follows therefore that:

if n  N ()  f (sn ) − f (L)  . #


 
Ex. 3.12 : Since:   is convergent to L = 0, and the function:
 n = 1
n

f ( x) = cos x is continuous at 0, thus the seq.


  
f ( sn ) 
=
n =1 
cos 
 n n = 1
- 68 -
is convergent to:

f ( L ) = f ( 0 ) = cos 0 = 1. #


 n 
Ex. 3.13 : Since:   is convergent to L = 1, and the
 n + 1 n = 1
function:

f ( x) = ln x is continuous at 1, thus the seq.


f ( sn )n = 1 = ln n n+ 1
 n = 1

is convergent to:

f ( L ) = f (1) = ln 1 = 0. #


1
Ex. 3.14 : Since:   is convergent to L = 0, and the function:
 n n = 1

 1 
0, thus the seq. f ( sn )

f ( x) = e x is continuous at = e n
n =1 
 n = 1

is convergent to:

f ( L ) = f ( 0 ) = e0 = 1. #

3.5: Definition ( Cauchy sequence ): An infinite sequence sn n = 1, is said

to be a Cauchy sequence iff:

for each  0 there exists a positive integer N = N ()  0 such that:

- 69 -
 sm − sn    m, n  N ().

Theorem 3.4 ( Cauchy seq. ) : Every convergent seq. is a Cauchy seq.

Proof : Suppose that the seq., sn n = 1 is convergent to L, that is:
lim sn = L  R,
n→

then,  0 there is an N = N ()  0 such that:


sn − L   n  N ().
2

This means that:

sm − sn = sm − L − (sn − L)  sm − L + sn − L    m, n  N ().

Now we illustrate by an example that a convergent seq. is a Cauchy

seq.


 cos n 
Ex.3.15: Show that the convergent seq.   is a Cauchy
 n n = 1
seq.

 
 cos n   (−1) n 
Sol. : It is obvious that, the seq.   =   is
 n n = 1  n n = 1
convergent to 0, we now show that it is a Cauchy seq.:

- 70 -
 (−1) m (−1) n 1 1 m+n
sm − sn = −  + = .
m n m n mn

Let m  n. Hence:

m+n 2m 2 2
sm − sn   =    m  n  N () =
mn mn n 
That is:

2
sm − sn    m, n  N () = . #

Exercises (3.2):

State whether or not the seq. converges and, if it does, find the limit:
 
 (−1)n
( 1) 

 ,
  −1 n 
(2)     , (3)  n 
n =1
,
 n n = 1   2  n = 1

  
 n  2
 n +1   4n 
( 4)   , (5)  2  , (6)   ,
 n + 1 n = 1  n n = 1  4n + 1 n = 1
2


( 7) (0.9)n 

n =1
,  
(8) 2n

n =1
,  
(9) e −n

n =1
,

 

   n +1    2n  
2

( 10) (−1)n n n =1
, (11)   , (12) ln    ,
 3n + 2 n = 1   n + 1  n = 1

- 71 -
  
 4n 
( 13)  n  ,
 2n 
(14)  n  ,  
(15) e−n

n =1
,
 2 + 100 n = 1  4 + 1 n = 1

 
 2n −1  1 2n −1 
(16)  n  , (17) cos n n = 1, (18)  sin  ,
 2 n = 1  n 2 n = 1

  
 n +1   ln(n + 2)    2n  
(19)   , (20)   , (21) ln    ,
 n n = 1  n + 2 n = 1   n + 1  n = 1

  
 1 1 1  1010 n 
( 22)  4 −  , (23)  −  , (24)   ,
 n 
n =1
 n n +1 n = 1  n + 1 n = 1

 
 1  
   n +1 
2n
(26) 

( 25) (−1) n n n = 1,  , (27) 1 +   .
 2 n 
n =1  n  
n = 1

- 72 -
 Chapter 4

Infinite Series
4.1: Infinite Series


4.1.1: Definition ( Infinite series ): An infinite series, denoted by  sn , is
n =1

the sum of an infinite seq. sn n = 1. That is:


 sn = s1 + s2 + s3 + ...+ sn + ...
n =1

Here, sn is called the n th term of the given series, which plays a very

important role to study the convergent of an infinite series.

Now we give some examples on infinite series to illustrate:


(1) n
 ax , a, x R, sn = axn, (Geometric series)
n=0

 1 1
(2)  p
, p R, sn = p
, ( P - series)
n = 1n n

 1 1
(3)  , sn = , ( Harmonic series)
n = 1n n

(4) n
 an x , an = an (n) , x R, sn = an xn, (Power series)
n=0

- 73 -
  (−1) (−1)n
n
(5)  , sn = , ( An alternating series)
n =1 n n
 n
(6) e , sn = en.
n =1

 1 1
(7)  , sn = .
n = 1 n(n + 1) n(n + 1)

 sin nx sin nx
(8)  , sn = .
n =1 n n

4.1.2: Definition ( Partial sum ): The partial sums, of the series  sn
n =1

denoted by, Sn or sumn and is given as:

S1 = s1,

S2 = s1 + s2,

S3 = s1 + s2 + s3,
.
.
.

and in general its n th partial sum, is given as:

n
Sn =  sn = s1 + s2 + s3 + ...+ sn.
n =1

- 74 -
These partial sums form a new sequence Sn n = 1, which may or may not
have a limit. If:

n
lim Sn = lim  sn = S  R (exists),
n→ n → n =1

as a real number, then we call S the sum of the infinite series.

That is, here, also, the partial sum Sn plays a very important role to

study the convergent of an infinite series, as we formulate in the following

definition.


4.1.3: Definition ( Convergent series ): An infinite Series  sn , is said to be
n =1

convergent to a real number S , if and only if, the sequence Sn n = 1 of

partial sums is convergent and the limit of the n th partial sum exists as the
real number S, that is:
n
lim Sn = lim  sn = S  R,
n→ n → n =1

in which case we call S the sum of the series and write:


 sn = S.
n =1

If otherwise ( the limit does not exist ), the series is said to be divergent.

 1
Ex. 4.1.1: Show that the series  is convergent, and find its
n = 1 n(n + 1)

sum.
- 75 -
Sol. : Since, its partial sum is given as:

n 1 1 1 1 1
sumn = Sn =  = + + + ... + ,
n = 1 n(n + 1) 1(2) 2(3) 3(4) n(n +1)

which can be simplified, by using the partial fraction decomposition as:

1 1 1
sn = = − ,
n(n + 1) n n + 1

Thus we have:

n 1 n 1 1 
sumn = Sn = =    − 
n + 1 

n = 1 n(n + 1) n =1  n

 1  1 1 1 1 1 1  1
= 1 −  +  −  +  −  + ... +  −  =1 − ,
 2   2 3   3 4   n n + 1  n +1

and so:

1
lim Sn = lim (1 − ) = 1  R.
n→ n→ n +1

Hence, the given series is convergent and:

 1
 =1. #
n = 1 n(n + 1)

Ex. 4.1.2 : Show that the series


( )
 −1 is divergent.
n
n =1

Sol. : Since, its partial sums alternate between 1 and 0 , that is:

- 76 -
 lim S , does not exist.
n→ n

Hence, the given series is divergent. #

 1
Ex. 4.1.3 : Discuss the convergent of the harmonic series  .
n = 1n

Sol. : One can easily write:

 1 1 1 1 1 1 1 1 1 1
 = 1 + + + + ... + + ... 1 + +  +  +  +...+  + ...
n = 1n 2 3 4 n 2  4 4 8 8
1 1 1 1
1 + + + + ...+ + ...
2 2 2 2
Thus the partial sum is given as:

n
sumn = Sn 1+  lim Sn =   R, (does not exist).
2 n→

Hence, the given series is divergent. ` #

Theorem 4.1.1 (The necessary and insufficie nt condition of convergent ) :


If the series  sn is convergent, then lim sn = 0.
n =1 n→


Proof : Let the series  sn be convergent, that is:
n =1

n
lim Sn = lim  sn = S  R.
n→ n → n =1

Since

- 77 -
 n
Sn =  sn = s1 + s2 + s3 + ...+ sn.
n =1

Then:

sn = Sn − Sn −1,

also

lim s = lim Sn − lim Sn − 1 = S − S = 0 #

n→ n n→ n→

 1
Observe that for the harmonic series  , in the previous Ex. 4.3,
n = 1n

we have:

1
lim sn = lim = 0,
n→ n→n

but we showed it is divergent. Hence, the converse of Theorem 4.1.1 is

not true in general, that is, if lim sn = 0 we cannot conclude that the
n→


series  sn is convergent (namely, the condition is insufficient).
n =1

On the other hand, the necessary and insufficient condition we have just
proved in Theorem 4.1.1 implies the following sufficient test for

divergence of a series:

• If lim s does not exist or if lim s  0, (*)

n → n n→ n

- 78 -
 
then the series  sn is divergent. Here, ( * ) is called the test for divergent.
n =1

n
  1
Ex. 4.1.4 : Show that the series  1 +  is divergent.
n = 1 n

Sol. : Since:

n
 1
lim sn = lim 1 +  = e  0.
n→ n→
 n

So, the series diverges by the test for divergence. #

 n
Ex. 4.1.5 : Show that the series  is divergent.
n = 1n +1

Sol. : Since:

n
lim sn = lim = 1  0.
n→ n →n + 1

So, the series diverges by the test for divergence. #

 (−1) n
n
Ex. 4.1.6 : Show that the series  is divergent.
n =1 n + 1

Sol. : Since:

(−1)n n
lim s = lim does not exist.
n→ n n→ n +1

So, the series diverges by the test for divergence. #

- 79 -
  2
Ex. 4.1.7 : Show that the series  n is divergent.
n =1

Sol. : Since:

lim sn = lim n2 =   0.
n→ n→

So, the series diverges by the test for divergence. #

2
 n
Ex. 4.1.8 : Show that the series  is divergent.
n = 1 3n
2
+4

Sol. : Since:

n2 1
lim sn = lim 2 =  0.
n→ n →  3n + 4 3

So, the series diverges by the test for divergence. #

 1
Ex. 4.1.9 : Show that the series  is divergent.
n =1 n
Sol. : Since:

 1 1 1 1 1
 = + + + ...+ + ...,
n =1 n 1 2 3 n

we have:

1
lim sn = lim = 0,
n→ n→ n

- 80 -
but, since:

n 1 1 1 1 1 n
Sn =  = + + + ...+  = n,
n =1 n 1 2 3 n n

that is, the sequence of partial sums Snn = 1 is unbounded, and therefore
the series diverges. #

An important example of an infinite series is the geometric series

which will be illustrated in the following definition and the theorem.

4.1.4: Definition ( The geometric series [G.S.] ): The general form of the
geometric series is given as:

 n
 ax , a, x R, sn = axn.
n=0

Namely,


 ax
n
= a + ax + ax2 + ax3 + ...+ axn + ...
n=0

sn +1
Here, a is called the first term of the series and x= is the base of the
sn
series. In the following theorem we discuss the convergence of the geometric
series.

Theorem 4.1.2 ( The convergent of the geometric series) : The

geometric series:

- 81 -
  a
 Convergent to , if x 1,
 ax , a, x R, 
n
1− x
n=0
Dinvergent if otherwise, x 1.

 n
Proof : The n th partial sum, of the geometric series  ax takes the
n=0

form:

Sn = a + ax + ax2 + ...+ axn −1.

Multiplication by x gives:

xSn = ax + ax2 + ...+ axn −1 + axn .

Subtracting the second equation from the first, we find that:

(1 − x)Sn = a(1− xn ),

This gives:

a(1− x n )
Sn = , x 1
1− x

If x = 1 then:

Sn = a + a + a + ...+ a = na  lim Sn = lim na =  .

n→ n→

Since the limit does not exist the geometric series diverges in this case.

If | x |  1 then:

- 82 -
 a(1− x n ) a
lim Sn = lim = .
n→ n →  1− x 1− x

a
Thus, when | x |  1 the geometric series is convergent to .
1− x

If x  −1 or x  1 then:

Lim S does not exist.

n→ n

Therefore, the geometric series diverges in those case. #

Ex. 4.1.10 : Determine whether the series is convergent or not:

 1
 .
n=0 (2)n
1
Sol. : It is a geometric series with a = 1, −1  x =  1, then the
2
series is convergent and its sum is:

 1 1 1
 = = .
(2)n #
n=0 1 2
1−
2
 −n
Ex. 4.1.11: Determine whether the series is convergent or not: e .
n=0

1
Sol. : It is a geometric series with a = 1, −1  x =  1, then the
e
series is convergent and its sum is:

- 83 -
  −n 1 e
 e = = . #
n=0
1−
1 e −1
e
Ex. 4.1.12 : Determine whether the series is convergent or not:
 n
 (0.9) .
n=0

Sol. : It is a geometric series with a = 1, −1  x = 0.9  1, then the

series is convergent and its sum is:

 1
 (0.9)
n
= = 10. #
n=0 1 − 0.9

Ex. 4.1.13 : Determine whether the series is convergent or not:

 n
 (0.9) .
n =1

Sol. : It is a geometric series with a = 0.9, −1  x = 0.9  1, then

the series is convergent and its sum is:

 0.9
 (0.9)
n
= = 9. #
n =1 1 − 0.9
 n
Ex. 4.1.14 : Determine whether the series is convergent or not: 2 .
n=0

Sol. : It is a geometric series with a = 1, x = 2  1, then the series is

divergent. #

- 84 -
 Whenever we have two convergent series we can add them, subtract
them, and multiply them by constants, to make other convergent series. The
next theorem gives the details.

Theorem 4.1.3 (Some basic results on convergent series) :

 
If  sn and  un are convergent series, then so are the series:
n =1 n =1

 
(i)  c sn = c  sn , c R,
n =1 n =1

  
(ii)  (sn  un ) =  sn   un.
n =1 n =1 n =1

Proof : Let:

n  n 
Sn =  sn , S=  sn , Un =  un , U=  un .
n =1 n =1 n =1 n =1

Using the properties of the limits. We easily have:

n n n
lim  csn = lim c  sn = c lim  sn = cS,
n → n = 1 n → n =1 n → n = 1

and

n n n
lim  (sn  un ) = lim (  sn   un )
n → n = 1 n → n = 1 n =1
n n
= 
lim  n lim  un
s = S  U. #
n → n = 1 n → n = 1

- 85 -
 In the following theorem we show that every nonzero multiple of a
divergent series diverges.

Theorem 4.1.4 (Multiple of a divergent series) :


If  sn diverges, and if c  0 is any real number different from 0, then
n =1


the series of multiples  csn diverges.
n =1


Proof : Suppose, to the contrary, that  csn converges. Then, from
n =1

Theorem 4.1.4, we find:

1   1 
 c sn =  csn =  sn ,
c n = 1 n = 1 c n =1

 
converges. That is,  csn cannot converge unless  sn also converges.
n =1 n =1

 
Thus, if  sn diverges, then  csn must diverge. #
n =1 n =1

 5
Ex. 4.1.15 : Discuss the convergence of the series:  n
.
n = 03

 5  1
Sol. : Since:  n
= 5  n
,
n = 03 n = 03

- 86 -
 1
which is a geometric series multiplied by 5 with a = 1, x =  1, then
3
1 15
the series is convergent to 5 = . #
1 2
1−
3

 1
Ex. 4.1.16 : Discuss the convergence of the series:  .
n = 1 2n

 1 1  1
Sol. : Since:  =  ,
n = 1 2n 2n = 1n

1
which is the harmonic series multiplied by , then the series is divergent,
2
for the divergent of the harmonic series #

Exercises (4.1):

Determine whether each series converges or not:

( 2) ,
n
 1   1   n
(1)  ln , (2)   , (3) 
n =1 n n = 1 2 n =1

 cos n
( 4)

 (−1)n +1 3
n
, (5)

 cos n , (6)  n
,
n =1 2 n=0 n=0 5

2
+1

( 7)  e−2n , (8)
 n
 , (9)

( )
 −1
n +1
n,
n=0 n =1 n n =1

- 87 -
 n
 2 −1   n −1 
n

(10)  (−1) n +1 2
, (11)  , (12)    ,
n = 1 n 
n n
n =1 10 n=0 3

 n!  1  7
(13)  , (14)  , | x |  1, (15)  ,
n = 3 (n −1)(n − 2)
n n
n = 0 1000 n =1x

 2
n +1 
  5 1 n
 2
(16)   n − n , (17)   n , (18)  n ,
n = 0 2 3  n = 0 5  n=05

  5  2 3 n
3  e
(19)   + ,
n
(20)   + n , (21)  n +1 ,
n = 1 n (n + 3) 4  n = 1 n 5  n = 13


( 22)  2
2
,
 n +1
(23) 
2
(
,
) (24)
 −1
 tan n,
n = 1n + 4n + 3 n = 1 n(n + 2) n =1

 
n
 4  n
( 25)  2 , (26)  , (27)  n +1 ,
n = 1 n −1 n = 1n +5 n = 13

 
n
 100  15
( 28)  n , (29)  n
, (30)  n + 1 ,
n =1 2 n = 1100 n = 13

    n 
 ln  .
n
( 31)  sin n, (32)  cos 1, (33) 
n =1 n =1 n = 1  2n + 5 

4.2: Tests for Convergence of Series with Nonnegative Terms

In general, it is difficult to find the exact sum of a series. We were able

to accomplish this for a few series, such as, geometric series, arithmetic

- 88 -
  1
series and  because in each of those cases we could find a
n = 1 n(n + 1)

simple formula for the n th partial sum Sn. But usually it is not easy to

compute lim S . On the other hand, if we find that lim s  0, we

n→ n n→ n


know that  sn is divergent. If we find that lim sn = 0, we know
n =1 n→


nothing about the convergence or divergence of  sn . Again, remember
n =1


that: If lim sn = 0, the series  sn might converge or it might diverge.
n→ n =1

Therefore, in the this section we develop several tests that enable us to

determine whether a series is convergent or not without explicitly finding
its sum. Here we direct our attention to series with nonnegative terms;
sn  0,  nN.

Theorem 4.2.1 (The integral test) : Suppose the function f (x) is a

continuous, decreasing, and positive function on [1, [ and let

sn = f (n). Then the series  sn is convergent if and only if the improper
n =1


integral  f ( x) dx is convergent. In other words:
1

 
(i)  f ( x) dx is convergent   sn is convergent,
1 n =1
 
(ii)  f ( x) dx is divergent   sn is divergent.
1 n =1

- 89 -
 b
Proof : Remember that a definite integral  f ( x) dx is said to be an
a

improper integral of the 1st kind if a = −  or b = , in this case if

the improper integral has a finite value it said to be convergent, if otherwise
it said to be divergent. Here, in this theorem we have that kind.

We start with the assumption that f (x) is a decreasing function with

f (n) = sn,  n  N. This leads us to observe that the rectangles in

Fig.[4.2....(a)], which have areas: a1, a2, a3, ..., an , collectively enclose

more area than that under the curve y = f (x) from

x = 1 to x = n + 1. That is,

n +1
 f ( x) dx  a1 + a2 + a3 + ...+ an.
1

In Fig.[4.2....(b)] the rectangles have been faced to the left instead of to

the right. If we momentarily disregard the first rectangle, of area a1, we

see that,

n
a2 + a3 + ... + an   f ( x) dx.
1

If we include a1, we have:

n
a1 + a2 + a3 + ... + an  a1 +  f ( x) dx.
1

Combining these results gives:

n +1 n
 f ( x) dx  a1 + a2 + a3 + ...+ an  a1 +  f ( x) dx.
1 1

- 90 -
 
If the integral  f ( x) dx is finite (convergent) the right-hand inequality
1
n 
shows that  sn is also finite (convergent). But if  f ( x) dx is infinite
n =1 1

n
(divergent) the left-hand inequality shows that  sn is also infinite
n =1

 n
(divergent). Hence the integral  f ( x) dx and the  sn are both finite
1 n =1

(convergent) or both infinite (divergent). #

We now, introduce a very important example of an infinite series, as a

general application on the integral test. Besides, it used to study the
convergence of many of other series, as we will show in the following.

4.2.1: Definition ( The P- series [P-S.] ): The general form of the P- series is
given as:

 1 1  1
 p
, p  R, sn
= p
, or  x
, x R.
n =1 n n n =1 n
Namely,

 1 1 1 1
 p
= 1 + p
+ p
+ ... + p
+ ...
n =1 n 2 3 n
In the following theorem we discuss the convergence of the P- series.

- 91 -
Theorem 4.2.2 ( The convergent of the P - series) : The P- series:

 1 Convergent, if p  1,
 , p  R, 
n =1 np Dinvergent if otherwise, p  1.

Proof : Let:

1
f ( x) = p
, x  [1,  [.
x

1
Then, if p 1, the function f ( x) = p
, is a continuous, decreasing, and
x
positive on [1,  [. Thus we have by using the integral test:

1−p b
  1  x 1
 f ( x) dx =  p dx =  x − p dx = lim = ,
1 1x 1 b → 1 − p p −1
1

 1
which is finite (convergent). Hence the P- series  is finite
n =1 np
(convergent) if p  1.

If p = 1, we have, the so called a harmonic series:

 1 1 1 1 1
 = 1+ + + + ... + + ...,
n = 1n 2 3 4 n

- 92 -
which we already know diverges. Again, by using the integral test, where

1
the function f ( x) = , is a continuous, decreasing, and positive on
x
[1,  [. Then:

 1 b
 f ( x) dx =  dx = lim ln x = ,
1 1x b→1

 1
and, since the integral diverges the series  likewise.
n =1 n

 1
Finally, p  1, then the terms of the P- series  are greater than
n =1 np
or equal the corresponding terms of divergent harmonic series. Hence the
 1
P- series  diverges, by the basic comparison test (the next test).
n =1 np
#

Ex. 4.2.1: Using the integral test to discuss the convergence of the series:

 1
 .
n = 1n +
2
1

1
Sol. : The function f ( x) = , is a continuous, decreasing, and
x +1
2

positive on [1,  [, so we use the integral test:

- 93 -
 1 b 1 b
 2 dx = lim  2 dx = lim tan −1 x
1 x +1 b →  1 x +1 b→ 1

b→
(
= lim tan −1 b − tan −1 1 = )  −  = .
2 4 4

 1
Thus,  dx is a convergent integral and so, by the integral test, the
1 x +1
2

 1
series  is convergent. #
n =1n + 1
2

Ex. 4.2.2 : Using the integral test to discuss the convergence of the series:

 1
 .
n = 1 5n + 1

1
Sol. : The function f ( x) = , is a continuous, decreasing, and
5x +1
positive on [1,  [, so we use the integral test:

 1 b 1 1 b
 dx = lim  dx = lim ln ( 5 x + 1 )
1 5x + 1 b→ 1 5 x + 1 5b →  1

= lim ln 5b = .
b→

 1
Thus,  dx is a divergent integral and so, by the integral test, the
1 5 x + 1
 1
series  is divergent. #
n = 1 5n + 1

- 94 -
Ex. 4.2.3 : Using the integral test to discuss the convergence of the series:

 1
 .
n = 1n ln(n +1)

1
Sol. : The function f ( x) = , is a continuous, decreasing, and
x ln( x +1)
positive on [1,  [, so we use the integral test, firstly, note that:

 1  1 b 1
 dx   dx = lim  dx
1 x ln( x +1) 1 ( x +1) ln( x +1) b →  1 ( x +1) ln( x +1)

= lim ln (ln( x +1)) =  − ln ln 2 = .

b

b→ 1

 1
Thus,  dx is a divergent integral and so, by the integral test,
1 x ln( x + 1)
 1
the series  . is divergent. #
n = 1 n ln(n + 1)

The convergence and the divergence of a series with nonnegative

terms is usually deduced by comparison with a series of known behavior.


Theorem 4.2.3 (The basic comparison test) : Let  sn be a
n =1

series with nonnegative terms, then:

 
(i)  sn converges if there exists a convergent series  u n with then
n =1 n =1

nonnegative terms such that:

sn  un ,  n sufficiently large.
- 95 -
  
(ii)  sn diverges if there exists a divergent series  t n with then
n =1 n =1

nonnegative terms such that:

sn  tn ,  n sufficiently large.

Proof : The proof is just a matter of noting that, in the first instance,


the partial sums of  sn are bounded, while in the second instance,
n =1

unbounded. The details are as follow:


In part (i), the partial sums of  sn are bounded from above by:
n =1


M = s1 + s2 + s3 + ...+ sn +  un .
0 n =1

Therefore, they form an increasing sequence that has a limit L that is less

than or equal M.


In part (ii), the partial sums of  sn are not bounded from above: if they
n =1


were, the partial sums of  t n would be bounded by
n =1


K = t1 + t2 + t3 + ...+ tn +  an .
0 n =1

- 96 -
  
That would imply convergence of  t n . Therefore, divergence of  t n
n =1 n =1


implies divergence of  sn . #
n =1

Note that (4.2.1):

To apply the basic comparison test to a series, we do not have to )i(

include the early terms of the series. We can start the test with any
index N, provided we include all the terms being tested from
there on.

To apply the basic comparison test we need to have on hand a list )ii(
of series that are known to converge and a list of series that are
known to diverge.

Ex. 4.2.4 : Using the basic comparison test to discuss the convergence of
the series:

1 2  1
5+ + +  .
7 3 n = 1 n!

Sol. : By ignoring the first three terms and comparing the reminder of the

 1
series with the convergent geometric series:  n −1
,
n = 12

where,

- 97 -
 1 1 1 1 1 1 1 1 1
s1 = =1= 0 , s2 = = = 2 −1 , s3 = =  3 −1 = ,
1! 2 2! 2 2 3! 6 2 4
1 1 1 1 1 1
s4 = =  4 −1 = , ..., sn =  n −1 = un.
4! 24 2 8 n! 2
Hence, accordingly, the basic comparison test the series

1 2  1
5+ + +  is convergent.
7 3 n = 1 n!
#

Ex. 4.2.5 : Using the basic comparison test to discuss the convergence of
the series:

 1
 .
n = 1 7n + 1

Sol. : This series diverges by comparison with the divergence series:

 1 1  1
 =  ,
n = 1 7(n +1) 7 n = 1 n + 1

where,

1 1
 . #
7(n +1) 7n +1

Ex. 4.2.6 : Using the basic comparison test to discuss the convergence of
the series:

 n3
 .
n =1n
5
+ 3n 4 + 13
- 98 -
Sol. : This series converges by comparison with the convergence series:

 1
 2
,
n = 1n

where,

n3 n3 1
 5 = 2. #
n + 3n +13 n
5 4
n

Ex. 4.2.7 : Determine whether the series:

 8
 ,
n = 1 3n + 4n +3
2

convergent or not.

Sol. : This series converges by comparison with the convergence series:

 8 8  1
 2
=  2
,
n = 1 3n 3 n
n = 1

where,

8 8
 . #
3n + 4n +3 3n
2 2

Ex. 4.2.8 : Test the series:

 ln n
 ,
n =1 n

- 99 -
for convergence or divergence.

Sol. : Since: ln n  1, n  3, and so:

ln n 1
 , n  3.
n n
Thus the given series is divergent by the basic comparison test. #

We come now to a somewhat more sophisticated comparison theorem.


Theorem 4.2.4 (The limit comparison test) : Suppose that  sn
n =1


and  u n are series with positive terms. If:
n =1

sn
lim = L R + , that is L] 0,  [,
n → un

Then either both series converge or both diverge.

Proof : Let m, M be positive numbers such that:

m  L  M.

sn
Because is close to L for large n, then there is an integer N such
un
that:

- 100 -
 sn
m  M, n  N,
un

and so:

m un  sn  M un , n  N.

  
So, it is obvious that, if  u n converges, so does  Mun . Thus,  sn
n =1 n =1 n =1


converges, by using the basic comparison test. While, if  u n diverges, so
n =1

 
does  mun . Thus,  sn diverges, by using also the basic comparison
n =1 n =1

test.

Ex. 4.2.9 : Test the series:

 1
 ,
n =12 − 1
n

for convergence or divergence.

Sol. : Using the limit comparison test with:

1 1
sn = , un =
2 −1
n
2n

and so:

- 101 -
 1
sn 2 n −1 2n
lim = lim = lim n =1 = L R +.
n →  un n → 1 n →  2 −1
2n
 1
Since the limit exists, L = 1 0, and the series  n
, is a convergent
n = 12

geometric, the given series converges by the limit comparison test. #

Ex. 4.2.10 : Test the series:

 2n + 3n
2
 ,
n =1 5 + n 5

for convergence or divergence.

Sol. : Using the limit comparison test with:

2n 2 + 3n 2n 2 2
sn = , un = 5
=
5 + n5 n 2 n

and so:

2n 2 + 3n
5 3
sn 5 + n5 2n 2 + 3n 2
lim = lim = lim =1 = L  R + .
n → un n →  2 n →  2 5 + n5
n

- 102 -
  1
Since the limit exists, L = 1 0, and the series  , is divergent ( p-
n =1 n
1
Series with p = 1, ), the given series diverges by the limit comparison
2
test. #

Ex. 4.2.11: Determine whether the series:

 
 sin ,
n =1 n
converges or diverges.

Sol. : Using the limit comparison test with:

sn = sin
, un =
,
n n
and so:


sin
sn
lim = lim n = 1 = L R + .
n → un n →  
n
   1
Since the limit exists, L = 1 0, and the series  =  is
n =1n n =1n

diverges ( p-Series with p = 1 ), the given series diverges by the limit

comparison test. #

- 103 -
 Exercises (4.2.1):

Determine whether each series converges or diverges:

n
 ln n   3  7
(1)  , (2)    , (3)  ,
n =1 n n = 1 4  n =1 (n +1)(n + 2)(n + 3)

 1  1  1
( 4)  , (5)  , (6)  ,
n = 0 1 + 2 ln n
2
n =1 3 n = 2 n ln n
n

 −5n  2n +1  2n +1
( 7) e , (8)  , (9)  ,
n=0 n =1 n +1
3 n =1 n +1
5

3
 7n + 2  n 2
 2n + 5
(10)  , (11)  , (12)  ,
n = 1 2n + 7 5n + 3n
5 5 3 2
n=0 =
n 2 + 2n −1 n 1

−n
  3  1  ln n
(13)    , (14)  , (15)  ,
n = 0 4  n = 11 +
3
2 ln n n=2 n

 1  n2  1
(16)  , (17)  4 , (18)  −n
,
n = 2 n (ln n)
2
n =0n − n +1
3
n = 0 2+3

 2n +1  2n +1  1
(19)  , (20)  , (21)  n +1
,
n =1 n 4 +1 n =1 n 3 +1 n = 12 −1

 2n +1  2  cos n
( 22)  , (23)  , (24)  ,
2n(n +1) n =1n +4 n = 1 n +1
3 2
n =1

- 104 -
  n  1  1
( 25)  , (26)  , (27)  ,
n = 1 n −1 n = 14 n +1
3 n = 1n− n

 5  1  n!
( 28)  , (29)  , (30)  n ,
n = 12+3 + n +1
n 2
n = 1n n = 1n

 n −1  2 + (−1) n+2
n

( 31)  , (32)  , (33)  .
n = 1n4
n
n =1 n n n =1 (n +1)n

 1+ 2
n
 1  2n
( 34)  sin , (35)  , (36)  .
n = 11+ 3 n = 11 + 3
n n
n =1 n

It is not always possible to tell whether a particular series converges by

using the basic comparison test or the limit comparison test. We might not
be able to find a series to compare it with. Nor is it always possible to use
the integral test to answer whatever questions of convergence then remain
unanswered. The terms of the series might not decrease as n increases, or

we might not find a formula for the nth term that we can integrate. What

we really need is an intrinsic test can be applied directly to the terms of the

given series  sn . The next two tests are intrinsic and they are easy to
n =1

 n
apply. Comparison with the geometric series  a x and with the P-series
n=0

 1
 p
, leads to the important tests for convergence: the root test and the
n = 1n

ratio test.

- 105 -
 
Theorem 4.2.5 (The root test) : Let  sn be a series with nonnegative
n =1

terms and suppose that:

  s converges if L  1,
n= 1 n

n  
sn = L    sn diverges if L  1,
lim
n →
n = 1
the test fails (is inconclusive or no information) if L =1.



Proof : We suppose first that L  1 and choose  so that:

L    1.

Since:

lim n sn = L,
n →

we have:

  n sn ,  n sufficient ly large.

Thus

 n  sn ,  n sufficient ly large.

Since:

 n

n =1

- 106 -
converges (a geometric series with 0    1), we know by the basic

comparison test that  sn converges.
n =1

We suppose now that L  1 and choose  so that:

L    1.

Since:

lim n sn = L,
n →

we have:

  n sn ,  n sufficient ly large.

Thus

 n  sn ,  n sufficient ly large.

Since:

 n

n =1

diverges (a geometric series with   1), the basic comparison test tells us

that  sn diverges.
n =1

To see the inconclusiveness of the root test when L = 1, we give a

counterexample, note that:

- 107 -
 lim n sn = L =1,
n →

For both the two series:

 1  1
 2
,  ,
n = 1n n = 1n

yet the first series converges while the second series diverges. Thus, some
other test for convergence must be used.

Ex. 4.2.12 : Applying the root test for the series:

 1
 .
n=2 (ln n)n

Sol. : Since:

1 1
lim n sn = lim = = 0 1.
(ln n) n →  ln n
lim
n n
n → n →

Then the series converges. #

Ex. 4.2.13 : Applying the root test for the series:

n
 2
 3
.
n =1n

Sol. : Since:

- 108 -
 3
n  1  1
n
n sn = lim n
2
= 2    = 2 1.
lim lim
n → n → n3 n →  n  
 
Then the series diverges. #

Ex. 4.2.14 : Applying the root test for the series:

n
  1
 1 −  .
n = 1 n

Sol. : Since:

n
n s = 
n 1 −
1  1
lim n lim  = lim 1 −  = 1.
n → n →  n  n → n 

Here the root test (fails) is inconclusive. It is also unnecessary: since

n
 1 1
lim 1 −  =  0.
n → n e

Hence the series diverges. #

Ex. 4.2.15 : Applying the root test for the series:

 1
 n
.
n = 1n

- 109 -
Sol. : Since:

1 1
lim sn = lim
n n = lim = 0 1.
n → n → nn n → n

Then the series converges. #


Theorem 4.2.6 (The ratio test) : Let  sn be a series with positive
n =1

terms and suppose that:

  s converges if L  1,
n= 1 n

sn + 1  
= L    sn diverges if L  1,
lim
n →  sn
n = 1
the test fails (is inconclusive or no information) if L =1.



Proof : We suppose first that L  1 and choose  so that:

L    1.

Since:

sn + 1
lim = L,
n → sn

we know that there exists n0  0 such that:

- 110 -
 sn + 1
if n  n0 then  .
sn

This gives:

sn   sn , sn  sn   2 sn ,
0 +1 0 0 +2 0 +1 0

and more generally:

sn +j
  j
sn .
0 0

For n  n0 , and setting j = n − n0 , we have:

sn0
sn   n − n0 sn =  n. (*)
 0
0 n

Since   1,

 sn0 sn0 
  =n

n
n =1  0 
n n0
n =1

converges (a geometric series with 0    1). Recalling (*), we can see


by the basic comparison test that  sn converges.
n =1

We suppose now that L  1. Then, from some index n0 on, we have:

sn + 1
  or sn  sn  sn  ...
sn 0 0 +1 0 +2

- 111 -
Hence, the terms of the series do not approach 0 as n becomes infinite,
and the series diverges, by the necessary condition for convergence.

Finally, to see the inconclusiveness of the ratio test when L = 1, we

give a counterexample, note that:

sn + 1
lim = L =1,
n → sn

For both the two series:

 1  1
 2
,  ,
n = 1n n = 1n

yet the first series converges while the second series diverges. Thus, some
other test for convergence must be used.

Ex. 4.2.16 : Applying the ratio test for the series:



(n!)2 .
n =1 (2n)!
Sol. : Since:

sn + 1
= lim
((n +1)!)2 ((2n)!)
=
(n +1)(n +1)
n →  (n!) (2n + 2)(2n +1)((2n)!) n →  (2n + 2)(2n +1)
lim 2 lim
n → sn
n +1 1
= lim = 1,
n →  4n + 2 4

Then the series converges. #

- 112 -
Ex. 4.2.17 : Applying the ratio test for the series:

 1
 .
n =1 (n)!
Sol. : Since:

sn + 1 n! 1
= lim = lim = 0 1
n →  (n +1)!
lim
n → sn n →  n +1

Then the series converges. #

Ex. 4.2.18 : Applying the ratio test for the series:

n
 n
 .
n =1 n!()
Sol. : Since:

(n +1) n +1(n!)
n
sn + 1  1
lim = lim n = lim 1+  = e 1.
n →  sn n →  n ((n +1)!) n →  n

Then the series diverges. #

Ex. 4.2.19 : Applying the ratio test for the series:

 n
 n
.
n = 110

Sol. : Since:

- 113 -
 sn + 1 (n +1)10n 1 n +1 1
lim = lim n +1
= lim = 1.
n → sn n →  n10 10 n →  n 10

Then the series converges. #

Ex. 4.2.20 : Applying the ratio test for the series:

 1
 .
n = 1 2n +1

Sol. : Since:

sn + 1 2n +1
lim = lim = 1.
n → sn n →  2n + 3

Here the ratio test (fails) is inconclusive. Therefore we have to look further.
Comparison with harmonic series shows that the series diverges, where:

1 1
 ,
2n +1 2(n +1)

 1
and the series  , diverges. #
n =1 2(n + 1)

Note that (4.2.2):

These tests can be applied to settle questions about the convergence or

divergence of series of nonpositive or negative terms. Just factor −1
from the series in question, and test the resulting series of nonnegative or
positive terms.

- 114 -
 Exercises (4.2.2):

Determine whether each series converges or diverges:

n
 n!   3  n!
(1)  , (2)  n  , (3)  ,
n =1  4
n 4n
n = 1100 n = 110

2
 1  ln n  1
( 4)  n
, (5)  , (6)  n
,
n =1 n n=2 n n = 2 ln n

 n +2
2 2
 n  1
( 7)  n , (8)  , (9)  ,
n = 1 n + 6n n = 11 +
3
n =1e n

(10) 
n
 2 n!
,
( ) (11)


n!
, (12)


n!
,
n = 1 n +1 n = 1 (n + 2)
n
n =1 n

 1  1  ln n
(13)  , (14)  , (15)  2
,
n = 2n ln n n = 2 n −1
2 n=2 n

n
 1  n!   n 
(16)  , (17)  , (18)    ,
n = 1 (n + 2)! n = 0  n + 100 
3
n=2 2
n ln n

 1  2
n
+5  1− n
(19)  n −1 , (20)  , (21)  ,
n = 13 +1 n =1 3n n = 1 n2
n

 n  1  2 −n
( 22)  , (23)  , (24) n e ,
n = 1 n +1
2 n
n =1 2 n =1

- 115 -
  1  1  1
( 25)  , (26)  , (27)  ,
n =1 n +2
3 n = 14 n +13 n = 1n− n

 5  (n + 1)(n + 2)  ln n
( 28)  , (29)  , (30)  ,
n = 12+3 n +1
n
n =1 n! n = 1n

 1− n
2
 n  n
( 31)  , (32)  , (33)  n .
n =1n2
n
n =1 (3n +1)n
n = 12


( 34) 
1
,
 n+3!
(35) 
( ,
)  sin
(36) 
2
n
.
n =1 n n ( )
n = 1 3! n! 3
n
n =1 2
n

4.3: Absolute Convergence

We now extend to series that have both positive and negative terms the
techniques that we have developed for answering questions about the convergence
of series of nonnegative terms. The extension is made possibly by a theorem that
says that; if a series converges after all its negative terms have been made
positive, then the unaltered series converges also.


Theorem 4.3.1 (The absolute convergenc e) : Let  sn be a series
n =1

with positive terms and suppose that:

 then 
If  sn converges   sn converges.
n =1 n =1

Proof : For each n we have:

− sn  sn  sn ,
- 116 -
so that:

0  sn + sn  2 sn ,

 
If  sn converges, then  2 sn converges and, by the basic comparison
n =1 n =1

test, the nonnegative series


 sn ( + sn )
n =1

Converges. On the other hand:

sn = sn + sn − sn .


Then, we can express the series  sn as the difference of two convergent series:
n =1


 sn =

 sn( + sn )−

 sn .
n =1 n =1 n =1


Therefore, the series  sn converges. #
n =1


4.3.1: Definition (Absolutely convergent): A series  sn is said to be absolutely
n =1


convergent if  sn , converges.
n =1

An obvious corollary of Theorem 4.3.1 (which is also called the

contraposition form of the theorem) is:

- 117 -
 
Corollary 4.3.1 (The absolute divergence ) : Let  sn be a series
n =1

with positive terms and suppose that:

 then 
If  sn diverges   sn diverges.
n =1 n =1

As we will show presently, the converse is false. There are convergent series
that are not absolutely convergent. Such series are called conditionally convergent.


4.3.2: Definition (Conditionally convergent): A convergent series  sn that is not
n =1

absolutely convergent, it is called conditionally convergent. That is:

 
If  sn converges, while  sn diverges.
n =1 n =1

 sin nx
Ex. 4.3.1: Show that the series:  3
,
n =1 n

is absolutely convergent.

sin nx 1
Sol. :  3
 3
.
n n

 sin nx
Using the basic comparison test,  converges, since the series
n =1 n3
 1
 3
converges (accordingly, P-series with p = 31). Then the given
n = 1n

sin nx
series  3
, is absolutely convergent. #
n =1 n

- 118 -
Ex. 4.3.2 : Show that the series:

1 1 1 1 1 1 1
1 − − 2 + 3 − 4 − 5 + 6 − 7 + ...
2 2 2 2 2 2 2
is absolutely convergent.

Sol. : If we replace each term by its absolute value, we obtain the series

 1 1 1 1 1 1 1 1
 n
= 1 + + 2 + 3 + 4 + 5 + 6 + 7 + ...
n = 02 2 2 2 2 2 2 2
This is a convergent geometric series. The initial series is therefore
absolutely convergent. #

Ex. 4.3.3 : Show that the series:

1 1 1 1 1 1 1 1
1− 3
− 3 − 3 − 3 + 3 + 3 − 3 + 3 ...
2 3 4 5 6 7 8 9
is absolutely convergent.

Sol. : If we replace each term by its absolute value, we obtain the series

 1 1 1 1 1 1 1 1 1
 3
=1+ 3
+ 3 + 3 + 3 + 3 + 3 + 3 + 3 + ...
n = 1n 2 3 4 5 6 7 8 9
This is a p-series with p = 31. It is therefore convergent. This means that

the initial series is absolutely convergent. #

4.3.3: Definition (Alternating series): Series in which the consecutive terms have
opposite signs are called alternating series. It has the general form:
- 119 -
 
( )
 −1
n+ 1
sn , sn  0,  n  N.
n =1

Theorem 4.3.2 (Alternati ng series " Leibniz test ") : If the alternating
series:


( )
 −1
n+ 1
sn , sn  0,  n  N.
n =1

satisfies:

(i) nonincreasing, that is sn  sn +1,  n  N ,

(ii) lim sn = 0.
n→

Then the series is converges.

Proof : If n is an even integer, n = 2m, then the alternative series:


( )
 −1
n+ 1
sn , sn  0,  n  N ,
n =1

can in general be written as:


( )
 −1
n +1
sn = s1 − s2 + s3 − s4 + ...+ s2m −1 − s2m + s2m +1 − s2m + 2 + ...
n =1

Now, let us firstly, look at the even partial sums, S2m. Since:

S2m = s1 − s2 + s3 − s4 + ...+ s2m −1 − s2m

= (s1 − s2 ) + (s3 − s4 ) + ...+ (s2m −1 − s2m )  0, where sn  sn +1.

On the other hand:

- 120 -
 S2m + 2 = S2m − s2m + 1 + s2m + 2, s2m + 1  s2m + 2,

we have:

S2 m + 2  S2 m .

This means that the sequence of even partial sums S2mm = 1 is decreasing. Being
bounded below by 0, S2m  0, it is convergent: that is the limit of the partial

sum exists as a real number, accordingly the definition of the convergent

sequence, namely:

lim S2m = L  R.
m→

Now, for the odd partial sums, S2m+1. Since:

S2m + 1 = S2m + s2m + 1, s2m + 1  0.

Since:

lim s2m + 1 = 0,
m→

we also have:

lim S2m +1 = lim S2m + lim s2m +1 = L.

m→ m→ m→

Since both the even and the odd partial sums tend to L, the sequence of all
partial sums Snn = 1 tends to L, we have:

- 121 -
 lim Sn = L,
n→

and so the alternative series:


 −1( )
n+ 1
sn , sn  0,  n  N ,
n =1

is convergent. #

Ex. 4.3.4 : From this theorem (Leibniz test), we can easily see that the

following series all converge:

(i)

 (−1)n+1 1 , (ii)

 (−1)n+1 1
, (iii)  (−1)n+1 .
 1
n =1 n n =1 n n =1 n!
The first two series converge only conditionally; the third converges absolutely. #

Ex. 4.3.5 : Discuss the convergence of the alternative series:

n2
(i)

 (−1)
n+ 1 3n
, (ii)

 (−1)n+ 1
.
n =1 4n −1 n =1 n +1
3

Sol. : For the series: (i)


 (−1)n+1 3n
,
n =1 4n −1

3n 3
lim sn = lim =  0.
n→ n →  4n −1 4

So condition (ii) is not satisfied. Instead, we look at the limit of the nth term
of the series:

- 122 -
 (−1)n 3n ,
lim sn = lim
n→ n→ 4n −1

This limit does not exist, so the series diverges by the test for divergence.

For the series:

n2
(ii)

 (−1) n+ 1
.
n =1 n +1
3

n2
lim sn = lim 3 = 0.
n→ n →  n +1

Thus the given series converges absolutely. Therefore, the series

n2
(ii)

 (−1) n+ 1
. converges. #
n =1 n +1
3

Exercises (4.3):

Test these series for; absolute convergence or conditional convergence:


( )
(1)  −1 n ,

(2) 
(−1)n , 
(3) 
(−1)n n ,
n =1 n =1 2n n =1 n +1


( 4) 
(−1)n n , 
(5) 
(−1)n ln n , (6) 
 −1 ( ) n
,
n =1 ln n n=2 n n = 2 n ln n


( 7) 
(−1)n n2 , (8)
 −1

( )
n
,

(9) 
(−1)n (n!)2 ,
n =1 e n
n = 1 2n + 1 n =1 (2n)!

- 123 -
 
(10)  (−1)  − 1 ,
n 1 
(11) 
n! 
, (12) 
(−1)n
,
n =1  n n!  n =1 (− 2)n
n = 1 n(n + 1)

  n       n
(13)  sin  , (14)  sin  2 , (15)  ,
 4   4n  n = 1n +
2
n =1 n =1 1


(16) 
(−1)n n , 
(17) 
( )n
−1
, (−1)  −
 1 1 

(18)  , n
n =1 2n n = 1n−2 n =1
n +1 
n  n

(−1)n sin  ,
n
  n
n
(20)  (−1) , (21)   −  ,
nn
 
(19) 
n =1 n n =1 n! n = 1 5 

n
   cos n 
sin
( 22)  cos , (23)  , (24) 
2 ,
3
n =1 n n =1 4
n n =1 n!


( 25)  (−1) 3n −1,
n 
(26)  (−1) n 1
,
 −1
(27) 
n
( ) n
,
n =1 2n +1 n =1 4n2 +1 n = 1 1+ 2 n
1
n n +1 n

( 28)  (−1) ,
ne 
(29)  (−1) n
,

(30)  ( ) n 10
−1 10 ,
n =1 n n =1 n +1 n =1 n

( 31)

 (−1)n ln1 + 1 , (32)

 (−1)n 3 n +1
, (33)  (−1)n 2 ,
 sin n
n =1  n n =1 n +1 n =1 n

(34)

 (−1)n 1+2n , (35)

 (−1)n 1+ n , (36)

 (− 2)n 1
.
n =1 n n =1 3+ n n =1 n+55

- 124 -
- 125 -
 Chapter 5

Power Series
5.1: Convergence of Power Series

5.1.1: Definition ( A power series ): A power series, is a series of the form:

 n
 an x , (*)
n=0

where x is a variable and the an are constants called the coefficients of the
series. For each fixed x, the series (*) is a series of constants that we can
test for convergence or divergence. A power series may converge for some
values of x, and diverge for other values of x. The sum of the series is a
function:


f ( x) =  an x
n
= a0 + a1x + a2 x2 + a3x3 + ...+ an xn + ...,
n=0

whose domain is the set of all x for which the series converges. Notice that
f resembles a polynomial. The only difference is that f has infinitely
many terms.

For instance, if we take an = const. = a (say),  n  N , the power

 n
series becomes the geometric series  ax , which converges when
n=0

x  1,

- 126 -
and diverges otherwise, that is if x  1.

More generally, series of the form:


 an ( x − x0 )n , (**)
n=0

is called a power series in (x − x0) or a power series centered at x0 or a

power series about x0. Notice that in writing out the term corresponding to

n = 0 in (*) and (**) we have adapted the convention that

( x − x0 )0 =1 even when x = x0. Notice also that when x = x0 all of the

terms are 0 for n  1 and so the power series (**) always converges
when

x = x0.

Our problem is to find for a given power series all values of 𝒙 for
which the series converges. In finding this set of values, the ratio test or the
root test to the series of absolute values will be used, as in the following
examples:

Examples:

Ex. 5.1.1: For what values of 𝒙 is the series ∑∞ 𝒏

𝒏=𝟎 𝒏! 𝒙 convergent?

Sol. : We use the ratio test. If 𝒙 ≠ 𝟎, we have:

(𝒏 + 𝟏)! 𝒙𝒏+𝟏
𝐥𝐢𝐦 | | = |𝒙| 𝐥𝐢𝐦 (𝒏 + 𝟏) = ∞.
𝒏→∞ 𝒏! 𝒙𝒏 𝒏→∞

- 127 -
By the ratio test, the series diverges when 𝒙 ≠ 𝟎. Thus, the given series
converges only when 𝒙 = 𝟎. #

(𝒙−𝟑)𝒏
Ex. 5.1.2 : For what values of 𝒙 does the series ∑∞
𝒏=𝟏 converge?
𝒏

Sol. : Using the ratio test, we have:

𝒏(𝒙 − 𝟑)𝒏+𝟏
𝐥𝐢𝐦 | | = |𝒙 − 𝟑| < 𝟏.
𝒏 → ∞ (𝒏 + 𝟏) (𝒙 − 𝟑)𝒏

By the ratio test, the given series is absolutely convergent, and therefore
convergent, when: |𝒙 − 𝟑| < 𝟏 and divergent when: : |𝒙 − 𝟑| > 𝟏.

Now:

|𝒙 − 𝟑| < 𝟏 ⟺ −𝟏 < 𝒙 − 𝟑 < 𝟏 ⟺ 𝟐 < 𝒙 < 𝟒,

so the series converges when: 𝟐 < 𝒙 < 𝟒. On the other hand the ratio test
gives no information when: |𝒙 − 𝟑| = 𝟏, that is 𝒙 = 𝟐 and 𝒙 = 𝟒. If we put
𝟏
𝒙 = 𝟒 in the given series, it becomes ∑∞
𝒏=𝟏 , the harmonic series, which is
𝒏
(−𝟏)𝒏
divergent. If 𝒙 = 𝟐, the series is ∑∞
𝒏=𝟏 , which converges by the
𝒏

alternating series test. Thus, the given power series converges for: 𝟐 ≤ 𝒙 <
𝟒. #

(−𝟏)𝒏 𝒙𝟐𝒏
Ex. 5.1.3: For what values of 𝒙 does the series ∑∞
𝒏=𝟎 converge?
𝟐𝟐𝒏 (𝒏!)𝟐

Sol. : Recall that, the given series is Bessel function of order 𝟎, that is we

can rewrite it as:

∞
(−𝟏)𝒏 𝒙𝟐𝒏
𝑱𝟎 (𝒙) = ∑ 𝟐𝒏 .
𝟐 (𝒏!)𝟐
𝒏=𝟎

- 128 -
Anyway, using the ratio test, we have:

𝟐𝟐𝒏 (𝒏!)𝟐 (−𝟏)𝒏+𝟏 𝒙𝟐(𝒏+𝟏) 𝟐

𝟏
𝐥𝐢𝐦 | 𝟐(𝒏+𝟏) | = 𝒙 𝐥𝐢𝐦 = 𝟎 < 𝟏, ∀ 𝒙
𝒏→∞ 𝟐 ((𝒏 + 𝟏)!)𝟐 (−𝟏)𝒏 𝒙𝟐𝒏 𝒏 → ∞ 𝟒(𝒏 + 𝟏)𝟐

∈ 𝓡.

Thus, by the ratio test, the given series converges for all values of 𝒙. In
other words, the domain of the Bessel function of order 𝟎, 𝑱𝟎 (𝒙) is ]−∞,
∞ [ = 𝓡. #

For the power series that we have looked at so far, the set of values of 𝒙
for which the series is convergent has always turned out to be an interval ( a
finite interval for the geometric series and the series in Ex. 5.1.2 ), the

infinite interval ]−∞, ∞ [ = 𝓡 in Ex. 5.1.3 , and a collapse interval

[𝟎, 𝟎] = {𝟎} in Ex. 5.1.1. The following theorem, says that this is true in

general:

Theorem 5.1.1 (The convergenc e for power series) : If a power

series:

∑∞ 𝒏 𝟐 𝒏
𝒏=𝟎 𝒂𝒏 𝒙 = 𝒂𝟎 + 𝒂𝟏 𝒙 + 𝒂𝟐 𝒙 + ⋯ + 𝒂𝒏 𝒙 + ⋯, (𝟓. 𝟏. 𝟏)

converges for 𝒙 = 𝒄 ( 𝒄 ≠ 𝟎 ), then it converges absolutely for all |𝒙| <

|𝒄|. If the series diverges for 𝒙 = 𝒅, then it diverges for all |𝒙| > |𝒅|.

Proof : The series ∑∞ 𝒏

𝒏=𝟎 𝒂𝒏 𝒄 being convergent, its general term tends to

zero, that is:

𝐥𝐢𝐦 𝒂𝒏 𝒄𝒏 = 𝟎,
𝒏→∞

therefore the set of values of these terms is bounded, that is there exists a
positive number 𝑴 such that:

- 129 -
 |𝒂𝒏 𝒄𝒏 | < 𝑴, ∀ 𝒏 > 𝑵. (𝟓. 𝟏. 𝟐)

Let us rewrite the power series (𝟓. 𝟏. 𝟏) as:

𝒙 𝒙 𝟐 𝒙 𝒏
∑∞ 𝒏 𝟐 𝒏
𝒏=𝟎 𝒂𝒏 𝒙 = 𝒂𝟎 + 𝒂𝟏 𝒄 ( 𝒄 ) + 𝒂𝟐 𝒄 ( 𝒄 ) + ⋯ + 𝒂𝒏 𝒄 ( 𝒄 ) + ⋯,

and form the series of the absolute values of its terms:

𝒙 𝒙 𝟐 𝒙 𝒏
∑∞ 𝒏 𝟐 𝒏
𝒏=𝟎|𝒂𝒏 𝒙 | = |𝒂𝟎 | + |𝒂𝟏 𝒄| | | + |𝒂𝟐 𝒄 | | | + ⋯ + |𝒂𝒏 𝒄 | | | + ⋯
𝒄 𝒄 𝒄

By the inequality (𝟓. 𝟏. 𝟐), every term of the latter series is less than the
𝒙
corresponding term of the geometric series with ratio | |, that is:
𝒄

∞
𝒏|
𝒙 𝟐
𝒙𝟐 𝒏
𝒙𝒏
∑ |𝒂𝒏 𝒙 = |𝒂𝟎 | + |𝒂𝟏 𝒄| | | + |𝒂𝟐 𝒄 | | | + ⋯ + |𝒂𝒏 𝒄 | | | + ⋯
𝒄 𝒄 𝒄
𝒏=𝟎

𝒙 𝒙 𝟐 𝒙 𝒏
< 𝑴 + 𝑴| | + 𝑴| | + ⋯ + 𝑴| | + ⋯
𝒄 𝒄 𝒄

For:

𝒙
|𝒙| < |𝒄| ⇒ | | < 𝟏,
𝒄

thus the geometric series on the right-hand side, namely:

𝒙 𝒙𝟐 𝒙𝒏
𝑴 + 𝑴| | + 𝑴| | + ⋯ + 𝑴| | + ⋯
𝒄 𝒄 𝒄

is convergent, therefore the series of the absolute values of the terms of the
given power series is also convergent, and hence the power series (𝟓. 𝟏. 𝟏) is
itself absolutely convergent. This proves the first half of the theorem.

The second half of the theorem involves nothing. For if the series
diverges at 𝒙 = 𝒅 and converges at |𝒙𝟎 | > |𝒅|, then accordingly, the first
half of the theorem, it is absolutely convergent at 𝒅. But the series cannot

- 130 -
both converge absolutely and diverge at one and the same time; namely,
which contradicts the hypothesis. Hence if it diverges at 𝒅 it also diverges
for all |𝒙| > |𝒅|. #

5.1.2: Definition ( Radians of convergence ): The number 𝒓 such that

power series:

∑∞ 𝒏 𝟐 𝒏
𝒏=𝟎 𝒂𝒏 𝒙 = 𝒂𝟎 + 𝒂𝟏 𝒙 + 𝒂𝟐 𝒙 + ⋯ + 𝒂𝒏 𝒙 + ⋯,

is convergent for all 𝒙 satisfying the condition |𝒙| < 𝒓 and divergent for
all 𝒙 satisfying the inequality |𝒙| > 𝒓 is called the radius of convergence of
the power series. The interval ]−𝒓, 𝒓[ is called the interval of convergence
of the power series.

Examples:

Ex. 5.1.4 : The series:

∞ ∞ ∞ ∞
𝒏
𝒙𝒏 (−𝒙)𝒏 𝒙𝒏
∑𝒙 , ∑ , ∑ , ∑ 𝟐,
𝒏 𝒏 𝒏
𝒏=𝟎 𝒏=𝟏 𝒏=𝟏 𝒏=𝟏

all have radius of convergence 𝒓 = 𝟏, but while the first ( geometric )

series converges only on the open interval ]−𝟏, 𝟏[, ( which is the interval
of convergence ), the second series converges on the half open (closed)
interval [−𝟏, 𝟏[, the third series converges on the half open (closed)
interval ]−𝟏, 𝟏], and the fourth series converges on the closed interval
[−𝟏, 𝟏].

( The proof is left as an exercise to the reader.) #

Ex. 5.1.5: Discuss the interval of convergence and the radius of

convergence of the series:

- 131 -
 ∞
𝒙𝒏
∑ .
𝒏!
𝒏=𝟎

Sol. : We use the ratio test. We have:

𝒏! 𝒙𝒏+𝟏 𝟏
𝐥𝐢𝐦 | | = |𝒙| 𝐥𝐢𝐦 = 𝟎 < 𝟏.
𝒏 → ∞ (𝒏 + 𝟏)! 𝒙𝒏 𝒏→∞𝒏 + 𝟏

Thus, by the ratio test, the given series converges for all values of 𝒙. In
other words, it is convergent for all finite values 𝒙, that is the interval of
convergence is ]−∞, ∞ [ = 𝓡. Accordingly, the radius of convergence is
𝒓 = ∞. #

5.2: Representation of Functions as Power Series

Here we learn how to represent certain types of functions as sums of

power series by manipulating geometric series or by differentiating or
integrating such a series.

5.2.1: Representation of functions as power series by using the geometric

series:

We start with an equation that we have seen before. From the

geometric series:

∞
𝟏
= ∑ 𝒙𝒏 = 𝟏 + 𝒙 + 𝒙𝟐 + 𝒙𝟑 + ⋯ + 𝒙𝒏 + ⋯ , ∀ 𝒙
𝟏−𝒙
𝒏=𝟏

∈ ]−𝟏, 𝟏[. (𝟓. 𝟐. 𝟏)

𝟏
We regard (𝟓. 𝟐. 𝟏) as expressing the function 𝒇(𝒙) = as a sum of a
𝟏−𝒙

power series.

- 132 -
 Examples:
𝟏
Ex. 5.2.1: Express 𝟏 + 𝒙𝟐
as sum of a power series, then deduce its radius

of convergence.

Sol. : From the geometric series, or exactly, by using (𝟓. 𝟐. 𝟏), replacing 𝒙

by −𝒙𝟐 , we have:

𝟏 𝟏
=
𝟏 + 𝒙𝟐 𝟏 − (−𝒙𝟐 )
∞

= ∑(−𝒙𝟐 )𝒏 = 𝟏 − 𝒙𝟐 + 𝒙𝟒 + ⋯ + (−𝒙𝟐 )𝒏 + ⋯
𝒏=𝟎

Because this is a geometric series, it converges when |−𝒙𝟐 | < 𝟏, that is,
𝒙𝟐 < 𝟏, or |𝒙| < 𝟏. Therefore, the interval of convergence is ]−𝟏, 𝟏[, and
hence the radius of convergence is 𝒓 = 𝟏. #

𝟏
Ex. 5.2.2 : Find a power series representation for 𝟐+𝒙
, then deduce its

radius of convergence.

Sol. : From the geometric series, or exactly, by using (𝟓. 𝟐. 𝟏), we first
−𝒙
factor a 𝟐 from the denominator, then replacing 𝒙 by , we have:
𝟐

∞ ∞
𝟏 𝟏 𝟏 −𝒙 𝒏 (−𝟏)𝒏 𝒏
= = ∑ ( ) = ∑ 𝒏+𝟏 𝒙 .
𝟐 + 𝒙 −𝒙 𝟐 𝟐 𝟐
𝟐 (𝟏 − ( )) 𝒏=𝟎 𝒏=𝟎
𝟐

−𝒙
This series converges when | | < 𝟏, that is, |𝒙| < 𝟐. So the interval of
𝟐

convergence is ]−𝟐, 𝟐[, and hence the radius of convergence is 𝒓 = 𝟐. #

- 133 -
 𝒙𝟑
Ex. 5.2.3: Find a power series representation for 𝟐+𝒙
, then deduce its

radius of convergence.

Sol. : Since the function is just 𝒙𝟑 times the function in Ex. 5.2.2, all we

have to do is to multiply that series by 𝒙𝟑 , that is:

∞
𝒙𝟑 (−𝟏)𝒏 𝒏+𝟑
= ∑ 𝒏+𝟏 𝒙 .
𝟐 + 𝒙 𝟐
𝒏=𝟎

Another way of writing this series is as follows:

∞
𝒙𝟑 (−𝟏)𝒏−𝟏 𝒏
=∑ 𝒙 .
𝟐 + 𝒙 𝟐𝒏−𝟐
𝒏=𝟑

Thus, as in Ex. 5.2.2, the interval of convergence is ]−𝟐, 𝟐[, and hence

the radius of convergence is 𝒓 = 𝟐. #

5.2.2: Differentiation and integration of power series

The sum of a power series is a function:

∞

𝒇(𝒙) = ∑ 𝒂𝒏 𝒙𝒏 = 𝒂𝟎 + 𝒂𝟏 𝒙 + 𝒂𝟐 𝒙𝟐 + ⋯ + 𝒂𝒏 𝒙𝒏 + ⋯,
𝒏=𝟎

whose domain is the interval of convergence of the series. We would like to

be able to differentiate and integrate such functions, and the following
theorems say that we can do so by differentiating or integrating each
individual term in the series, just as we would for a polynomial. This is
called term-by-term differentiation and integration.

Theorem 5.2.1 (The term − by − term differenti ation) : If the

power series:

- 134 -
 ∞

𝒇(𝒙) = ∑ 𝒂𝒏 𝒙𝒏 = 𝒂𝟎 + 𝒂𝟏 𝒙 + 𝒂𝟐 𝒙𝟐 + ⋯ + 𝒂𝒏 𝒙𝒏 + ⋯ , (𝟓. 𝟐. 𝟐)
𝒏=𝟎

has radius of convergence 𝒓 > 𝟎, then:

𝒇′(𝒙) = ∑ 𝒏𝒂𝒏 𝒙𝒏−𝟏 = 𝒂𝟏 + 𝟐𝒂𝟐 𝒙 + 𝟑𝒂𝟑 𝒙𝟐 + ⋯ + 𝒏𝒂𝒏 𝒙𝒏−𝟏 + ⋯ , (𝟓. 𝟐. 𝟑)

𝒏=𝟎

also has radius of convergence 𝒓 > 𝟎.

Proof : For the sake of simplicity, let us suppose that the limit of the ratio
𝒂𝒏+𝟏
exists as 𝒏 tends to infinity, then the radii of convergence of (𝟓. 𝟐. 𝟐)
𝒂𝒏

and (𝟓. 𝟐. 𝟑) can be found with the aid of the ratio (D'Alembert's) test, as:

𝒂𝒏+𝟏 𝒙𝒏+𝟏 𝒂𝒏+𝟏

𝐥𝐢𝐦 | | = |𝒙| 𝐥𝐢𝐦 | |,
𝒏→∞ 𝒂𝒏 𝒙𝒏 𝒏 → ∞ 𝒂𝒏

and:

(𝒏 + 𝟏)𝒂𝒏+𝟏 𝒙𝒏+𝟏 𝒏+𝟏 𝒂𝒏+𝟏 𝒂𝒏+𝟏

𝐥𝐢𝐦 | | = |𝒙| 𝐥𝐢𝐦 𝐥𝐢𝐦 | | = |𝒙| 𝐥𝐢𝐦 | |,
𝒏→∞ 𝒏𝒂𝒏 𝒙𝒏 𝒏→∞ 𝒏 𝒏 → ∞ 𝒂𝒏 𝒏 → ∞ 𝒂𝒏

which indicates the equality of the limits of the ratios of two successive
terms for both series. Hence, their radii of convergence coincides. #

Theorem 5.2.2 (The term − by − term integratio n) : If the power

series:
∞

𝒇(𝒙) = ∑ 𝒂𝒏 𝒙𝒏 = 𝒂𝟎 + 𝒂𝟏 𝒙 + 𝒂𝟐 𝒙𝟐 + ⋯ + 𝒂𝒏 𝒙𝒏 + ⋯,
𝒏=𝟎

has radius of convergence 𝒓 > 𝟎, then:

- 135 -
 ∞
𝒂𝒏 𝒏+𝟏 𝒂𝟏 𝟐 𝒂 𝟐 𝟑 𝒂𝟑 𝟒
∫ 𝒇(𝒙)𝒅𝒙 = ∑ 𝒙 = 𝒂𝟎 𝒙 + 𝒙 + 𝒙 + 𝒙 +⋯
𝒏+𝟏 𝟐 𝟑 𝟒
𝒏=𝟎

𝒂𝒏 𝒏+𝟏
+ 𝒙 + ⋯ + 𝑪,
𝒏+𝟏

also has radius of convergence 𝒓 > 𝟎. Here 𝑪 is the constant of integration.

Proof : ( The proof is left as an exercise to the reader. (Hint: It is exactly

as the previous theorem.)) #

Note that (5.2.1): Although Theorem 5.2.1, and Theorem 5.2.2 say
that the radius of convergence remains the same when a power series is
differentiated or integrated, this does not mean that the interval of
convergence remains the same. It may happen that the original series
converges at an end point, whereas the differential series diverges there. As
in the following examples:

Examples:

Ex. 5.2.4 : The series:

∞
𝒙𝒏
∑ 𝟐,
𝒏
𝒏=𝟏

that we have mentioned in Ex. 5.1.4 converges on the closed interval

[−𝟏, 𝟏], while its derivative:

∞
𝒙𝒏−𝟏
∑ ,
𝒏
𝒏=𝟏

converges on the half open (closed) interval [−𝟏, 𝟏[.

- 136 -
( The discussion is left as an exercise to the reader.) #

Ex. 5.2.5: The geometric series:

∑ 𝒙𝒏 = 𝟏 + 𝒙 + 𝒙𝟐 + 𝒙𝟑 + ⋯ + 𝒙𝒏 + ⋯,
𝒏=𝟎

converges on ]−𝟏, 𝟏[, thus all the series:

∞ ∞
𝒅 𝒏
∑ 𝒙 = ∑ 𝒏𝒙𝒏−𝟏 = 𝟏 + 𝟐𝒙 + 𝟑𝒙𝟐 + 𝟒𝒙𝟑 + ⋯ + 𝒏𝒙𝒏−𝟏 + ⋯,
𝒅𝒙
𝒏=𝟎 𝒏=𝟏

∞ ∞
𝒅𝟐 𝒏
∑ 𝟐 𝒙 = ∑ 𝒏(𝒏 − 𝟏)𝒙𝒏−𝟐 = 𝟐 + 𝟔𝒙 + 𝟏𝟐𝒙𝟐 + 𝟐𝟎𝒙𝟑 + ⋯
𝒅𝒙
𝒏=𝟎 𝒏=𝟏

+ 𝒏(𝒏 − 𝟏)𝒙𝒏−𝟐 + ⋯,

∞ ∞
𝒅𝟑 𝒏
∑ 𝟑 𝒙 = ∑ 𝒏(𝒏 − 𝟏)(𝒏 − 𝟐)𝒙𝒏−𝟑 = 𝟔 + 𝟐𝟒𝒙 + 𝟔𝟎𝒙𝟐 + ⋯
𝒅𝒙
𝒏=𝟎 𝒏=𝟏

+ 𝒏(𝒏 − 𝟏)(𝒏 − 𝟐)𝒙𝒏−𝟑 + ⋯,

all also converges on ]−𝟏, 𝟏[. #

Ex. 5.2.6 : Find the sum of the power series:

∞
𝒙𝒏
∑ , ∀ 𝒙 ∈ ]−𝟏, 𝟏[.
𝒏
𝒏=𝟏

Sol. : By setting:

∞
𝒙𝒏
𝒇(𝒙) = ∑ , ∀ 𝒙 ∈ ]−𝟏, 𝟏[,
𝒏
𝒏=𝟏

- 137 -
and noting that:
∞
𝟏
𝒇′(𝒙) = ∑ 𝒙𝒏 = , (𝐭𝐡𝐞 𝐠𝐞𝐨𝐦𝐞𝐭𝐫𝐢𝐜 𝐬𝐞𝐫𝐢𝐞𝐬 ) ∀ 𝒙 ∈ ]−𝟏, 𝟏[.
𝟏−𝒙
𝒏=𝟏

With:

𝟏
𝒇′(𝒙) = and 𝒇(𝟎) = 𝟎,
𝟏−𝒙

we can conclude that:

𝒙𝒏 𝟏
𝒇(𝒙) = ∑∞
𝒏=𝟏 = 𝐥𝐧 , ∀ 𝒙 ∈ ]−𝟏, 𝟏[. #
𝒏 𝟏−𝒙

𝟏
Ex. 5.2.7 : Express (𝟏−𝒙)𝟐
as a power series, then deduce its radius of

convergence.

Sol. : From the geometric series:

∞
𝟏
= ∑ 𝒙𝒏 = 𝟏 + 𝒙 + 𝒙𝟐 + 𝒙𝟑 + ⋯ + 𝒙𝒏 + ⋯ , ∀ 𝒙 ∈ ]−𝟏, 𝟏[,
𝟏−𝒙
𝒏=𝟏

By differentiating each side we get:

∞
𝟏
𝟐
= ∑ 𝒏𝒙𝒏−𝟏 = 𝟏 + 𝟐𝒙 + 𝟑𝒙𝟐 + 𝟒𝒙𝟑 + ⋯ + 𝒏𝒙𝒏−𝟏 + ⋯
(𝟏 − 𝒙)
𝒏=𝟏

Accordingly, to Theorem 5.2.1 the radius of convergence of the

differentiated series is the same as the radius of convergence of the original
series, namely 𝒓 = 𝟏. #

Ex. 5.2.8: Find a power series representation for 𝐥𝐧 (𝟏 − 𝒙), then deduce

its radius of convergence.

- 138 -
Sol. : Since:

𝒅 −𝟏
𝐥𝐧 (𝟏 − 𝒙) = . (𝟓. 𝟐. 𝟒)
𝒅𝒙 𝟏−𝒙

So, we integrate both sides of (𝟓. 𝟐. 𝟒), it follows directly:

𝒅𝒙
−𝐥𝐧 (𝟏 − 𝒙) = ∫ = ∫(𝟏 + 𝒙 + 𝒙𝟐 + 𝒙𝟑 + ⋯ + 𝒙𝒏 + ⋯ )𝒅𝒙
𝟏−𝒙

𝒙𝟐 𝒙𝟑 𝒙𝒏+𝟏
= 𝒙 + + + ⋯+ + ⋯ +𝑪
𝟐 𝟑 𝒏+𝟏
∞ ∞
𝒙𝒏+𝟏 𝒙𝒏
= ∑ +𝑪 = ∑ + 𝑪, |𝒙| < 𝟏.
𝒏+𝟏 𝒏
𝒏=𝟎 𝒏=𝟏

To determine the value of the integrating constant 𝑪, we put 𝒙 = 𝟎 in the

last equation, we obtain:

−𝐥𝐧 (𝟏 − 𝟎) = 𝟎 = 𝑪.

Thus,

𝒙𝟐 𝒙𝟑 𝒙𝒏+𝟏
𝐥𝐧 (𝟏 − 𝒙) = − 𝒙 − − − ⋯ − − ⋯
𝟐 𝟑 𝒏+𝟏

𝒙𝒏
= − ∑∞
𝒏=𝟏 , |𝒙| < 𝟏.
𝒏

Accordingly, to Theorem 5.2.2 the radius of convergence of the

integrated series is the same as the radius of convergence of the original

series, namely 𝒓 = 𝟏. #

𝟏
Note that (5.2.2): if we put 𝒙 =
𝟐
in the last example Ex. 5.2.8, it

happens:

- 139 -
 𝟏 𝒙𝟐 𝒙𝟑 𝒙𝒏+𝟏
𝐥𝐧 = −𝐥𝐧 𝟐 = − 𝒙 − − − ⋯ − − ⋯
𝟐 𝟐 𝟑 𝒏+𝟏

That is:
∞
𝟏 𝟏 𝟏 𝟏
𝐥𝐧 𝟐 = + + + ⋯ = ∑ 𝒏.
𝟐 𝟖 𝟐𝟒 𝒏𝟐
𝒏=𝟏

Ex. 5.2.9 : Find a power series representation for 𝐭𝐚𝐧−𝟏 𝒙, then deduce its

radius of convergence.

Sol. : Since:

𝒅 𝟏
𝐭𝐚𝐧−𝟏 𝒙 = .
𝒅𝒙 𝟏 + 𝒙𝟐

𝟏
So, we integrate both sides of the power series for found in
𝟏+𝒙𝟐

Ex. 5.2.1, it follows directly:

𝒅𝒙
𝐭𝐚𝐧−𝟏 𝒙 = ∫ 𝟐
= ∫(𝟏 − 𝒙𝟐 + 𝒙𝟒 + ⋯ + (−𝒙𝟐 )𝒏 + ⋯ ) 𝒅𝒙
𝟏 + 𝒙

𝒙𝟑 𝒙𝟓 𝒙𝟕
=𝑪+ 𝒙− + − + ⋯
𝟑 𝟓 𝟕

To determine the value of the integrating constant 𝑪, we put 𝒙 = 𝟎 in the

last equation, we obtain:

𝐭𝐚𝐧−𝟏 𝟎 = 𝟎 = 𝑪.

Thus,

- 140 -
 ∞
−𝟏
𝒙𝟑 𝒙𝟓 𝒙𝟕 𝒏
𝒙𝟐𝒏+𝟏
𝐭𝐚𝐧 𝒙 = 𝒙 − + − + ⋯ = ∑ (−𝟏) .
𝟑 𝟓 𝟕 𝟐𝒏 + 𝟏
𝒏=𝟎

According, to Theorem 5.2.2 the radius of convergence of the integrated

series is the same as the radius of convergence of the original series, namely
𝒓 = 𝟏. #

Ex. 5.2.10 : Evaluate:

𝒅𝒙
∫ ,
𝟏 + 𝒙𝟕

as a power series.

Sol. : The first step is to express the integrand as the sum of a power series.
As in Ex. 5.2.1, it follows directly:
∞ ∞
𝒅𝒙 𝒏 𝟕𝒏 𝒏
𝒙𝟕𝒏+𝟏
∫ = ∫ (∑(−𝟏) 𝒙 ) 𝒅𝒙 = ∑(−𝟏)
𝟏 + 𝒙𝟕 𝟕𝒏 + 𝟏
𝒏=𝟎 𝒏=𝟎

𝒙𝟖 𝒙𝟏𝟓 𝒏
𝒙𝟕𝒏+𝟏
=𝑪+ 𝒙− + − ⋯ + (−𝟏) + ⋯
𝟖 𝟏𝟓 𝟕𝒏 + 𝟏
𝟏
Because the series for converges when |−𝒙𝟕 | < 𝟏, that is, |𝒙| < 𝟏.
𝟏+𝒙𝟕

Therefore, the interval of convergence is ]−𝟏, 𝟏[, and hence the radius of
convergence is 𝒓 = 𝟏. Thus, accordingly, to Theorem 5.2.2 the radius of

convergence of the integrated series is the same as the radius of

convergence of the original series, namely 𝒓 = 𝟏. #

5.3: Expanding Functions into Power Series:

In the preceding section we were able to find power series

representations for a certain restricted class of functions. Here we
- 141 -
investigate more general problems to represent functions accordingly to
Taylor-Maclaurin (expansion) series and the special case of the Taylor-
Maclaurin series which is called the binomial (theorem) series.

5.3.1: Taylor-Maclaurin (expansion) series:

The replacement of a function by a simple expression of the type of a

polynomial proves extremely convenient in various problems of
mathematical analysis such as evaluating integrals, solving differential
equations and the like.

Thus, we shall assume that the given function 𝒇(𝒙) is infinitely

differentiable in a neighbourhood of a point 𝒙𝟎 . Suppose that it is
representable as the sum of a power series convergent within an interval
containing the point 𝒙𝟎 , (|𝒙 − 𝒙𝟎 | < 𝒓 ), that is:

𝒇(𝒙) = 𝒂𝟎 + 𝒂𝟏 (𝒙 − 𝒙𝟎 ) + 𝒂𝟐 (𝒙 − 𝒙𝟎 )𝟐 + ⋯ + 𝒂𝒏 (𝒙 − 𝒙𝟎 )𝒏 + ⋯ , |𝒙 − 𝒙𝟎 |
< 𝒓,

(𝟓. 𝟑. 𝟏)

where 𝒂𝟎 , 𝒂𝟏 , 𝒂𝟐 , ⋯ , 𝒂𝒏 , ⋯ are constant coefficients

We shall show that the general properties of power series make it

possible to find the constant coefficients 𝒂𝟎 , 𝒂𝟏 , 𝒂𝟐 , ⋯ , 𝒂𝒏 , ⋯, considered as
undetermined coefficients, by using the known values of the function 𝒇(𝒙)
and of its derivatives at the point 𝒙𝟎 , namely:

𝒇(𝒏) (𝒙𝟎 )
𝒂𝒏 = .
𝒏!

Let us put 𝒙 = 𝒙𝟎 in (𝟓. 𝟑. 𝟏). This yields:

𝒂𝟎 = 𝒇(𝒙𝟎 ).

- 142 -
Now, let us differentiate the power series and substitute 𝒙 = 𝒙𝟎 into the
differentiated series, we directly obtain:

𝒇′ (𝒙) = 𝒂𝟏 + 𝟐𝒂𝟐 (𝒙 − 𝒙𝟎 ) + ⋯ + 𝒏𝒂𝒏 (𝒙 − 𝒙𝟎 )𝒏−𝟏 + ⋯ , |𝒙 − 𝒙𝟎 | < 𝒓.

This result in:

𝒂𝟏 = 𝒇′(𝒙𝟎 ).

The repeated differentiation gives us:

𝒇′ (𝒙) = 𝟐𝒂𝟐 + 𝟔𝒂𝟑 (𝒙 − 𝒙𝟎 ) ⋯ + 𝒏(𝒏 − 𝟏)𝒂𝒏 (𝒙 − 𝒙𝟎 )𝒏−𝟐 + ⋯ , |𝒙 − 𝒙𝟎 | < 𝒓,

whence, on substituting 𝒙 = 𝒙𝟎 we obtain:

𝒇′′(𝒙𝟎 )
𝒂𝟐 = .
𝟐!

Proceeding in this way we obtain, after the 𝒏𝒕𝒉 differentiation:

𝒇(𝒏) (𝒙) = 𝒏! 𝒂𝒏 + ⋯ , |𝒙 − 𝒙𝟎 | < 𝒓,

where the terms indicated by dots involve the factor (𝒙 − 𝒙𝟎 ) and hence
turn into zero for 𝒙 = 𝒙𝟎 . Hence, putting 𝒙 = 𝒙𝟎 we receive:

𝒇(𝒏) (𝒙𝟎 )
𝒂𝒏 = .
𝒏!

We thus consequently determine the coefficients 𝒂𝟎 , 𝒂𝟏 , 𝒂𝟐 , ⋯ , 𝒂𝒏 , ⋯ of

expression (𝟓. 𝟑. 𝟏).

On substituting the expressions found into (𝟓. 𝟑. 𝟏) we arrive at the

series:

𝒇′′ (𝒙𝟎 )
𝒇(𝒙) = 𝒇(𝒙𝟎 ) + 𝒇′ (𝒙𝟎 )(𝒙 − 𝒙𝟎 ) + (𝒙 − 𝒙𝟎 )𝟐 + ⋯
𝟐!
𝒇(𝒏) (𝒙𝟎 )
+ (𝒙 − 𝒙𝟎 )𝒏
𝒏!
- 143 -
 𝒇(𝒏) (𝒙𝟎 )
+ ⋯ = ∑∞
𝒏=𝟎 (𝒙 − 𝒙𝟎 )𝒏 , |𝒙 − 𝒙𝟎 | < 𝒓,
𝒏!

(𝟓. 𝟑. 𝟐)

The series (𝟓. 𝟑. 𝟐) is called the Taylor (expansion) series of the function
𝒇(𝒙) at 𝒙𝟎 .

For the special case 𝒙𝟎 = 𝟎, the Taylor series becomes:

𝒇′′ (𝟎) 𝒇(𝒏) (𝟎)

𝒇(𝒙) = 𝒇(𝟎) + 𝒇′ (𝟎)𝒙 + 𝒙𝟐 + ⋯ + 𝒙𝒏 + ⋯
𝟐! 𝒏!

𝒇(𝒏) (𝟎)
= ∑∞
𝒏=𝟎 𝒙𝒏 , |𝒙| < 𝒓.
𝒏!

This case arises frequently enough that it is given the special name
Maclaurin series.

Examples:

Ex. 5.3.1: Find the Taylor series of the function 𝒇(𝒙) = 𝐜𝐨𝐬 𝐱, about the
point 𝒙𝟎 = 𝟐𝝅 then deduce its radius of convergence.

Sol. : If 𝒇(𝒙) = 𝐜𝐨𝐬 𝒙, then:

𝒇(𝟐𝒏) (𝒙) = (−𝟏)𝒏 𝐜𝐨𝐬 𝒙, 𝒇(𝟐𝒏+𝟏) (𝒙) = (−𝟏)𝒏+𝟏 𝐬𝐢𝐧 𝒙

and so, 𝒇(𝟐𝒏) (𝟐𝝅) = (−𝟏)𝒏 , 𝒇(𝟐𝒏+𝟏) (𝟐𝝅) = 𝟎.

Hence, the required series is:

(−𝟏)𝒏 (𝒙−𝟐𝝅)𝟐𝒏
𝒇(𝒙) = 𝐜𝐨𝐬 𝒙 = ∑∞
𝒏=𝟎 (𝟐𝒏)!

Using the ratio test, we can easily deduce the radius of convergence as:

(𝟐𝒏)!
𝒓 = |𝒙 − 𝟐𝝅| 𝐥𝐢𝐦
𝒏 → ∞ (𝟐𝒏+𝟐)!

- 144 -
 (𝟐𝒏)!
= |𝒙 − 𝟐𝝅| 𝒍𝒊𝒎 = 𝟎 < 𝟏.
𝒏 → ∞ (𝟐𝒏+𝟐)(𝟐𝒏+𝟏)(𝟐𝒏)!

So, by the ratio test, the series converges for all 𝒙 and the radius of
convergence is 𝒓 = ∞. #

Ex. 5.3.2 : Find the Maclaurin series of the function 𝒇(𝒙) = 𝐜𝐨𝐬 𝐱, then
deduce its radius of convergence.

Sol. : Similarly, if 𝒇(𝒙) = 𝐜𝐨𝐬 𝒙, then:

𝒇(𝟐𝒏) (𝒙) = (−𝟏)𝒏 𝐜𝐨𝐬 𝒙, 𝒇(𝟐𝒏+𝟏) (𝒙) = (−𝟏)𝒏+𝟏 𝐬𝐢𝐧 𝒙

and so, 𝒇(𝟐𝒏) (𝟎) = (−𝟏)𝒏 , 𝒇(𝟐𝒏+𝟏) (𝟎) = 𝟎.

Hence, the required series is:

(−𝟏)𝒏 𝒙𝟐𝒏
𝒇(𝒙) = 𝐜𝐨𝐬 𝒙 = ∑∞
𝒏=𝟎 (𝟐𝒏)!

𝒙𝟐 𝒙𝟒 (−𝟏)𝒏 𝒙𝟐𝒏
= 𝟏− + − ⋯+ + ⋯
𝟐! 𝟒! (𝟐𝒏)!

So, by the ratio test, the series converges for all 𝒙 and the radius of
convergence is 𝒓 = ∞. #

Ex. 5.3.3: Find the Maclaurin series of the function 𝒇(𝒙) = 𝐬𝐢𝐧 𝒙, then

deduce its radius of convergence.

Sol. : Similarly, if 𝒇(𝒙) = 𝐬𝐢𝐧 𝒙, then:

𝒇(𝟐𝒏) (𝒙) = (−𝟏)𝒏 𝐬𝐢𝐧 𝒙, 𝒇(𝟐𝒏+𝟏) (𝒙) = (−𝟏)𝒏 𝐜𝐨𝐬 𝒙

and so, 𝒇(𝟐𝒏) (𝟎) = 𝟎, 𝒇(𝟐𝒏+𝟏) (𝟎) = (−𝟏)𝒏 .

Hence, the required series is:

- 145 -
 (−𝟏)𝒏 𝒙𝟐𝒏+𝟏
𝒇(𝒙) = 𝐬𝐢𝐧 𝒙 = ∑∞
𝒏=𝟎 (𝟐𝒏+𝟏)!

𝒙𝟑 𝒙𝟓 (−𝟏)𝒏 𝒙𝟐𝒏+𝟏
= 𝒙− + − ⋯+ + ⋯
𝟑! 𝟓! (𝟐𝒏+𝟏)!

So, by the ratio test, the series converges for all 𝒙 and the radius of
convergence is 𝒓 = ∞. #

Ex. 5.3.4 : Find the Maclaurin series of the function 𝒇(𝒙) = 𝐞𝒙 , then

deduce its radius of convergence.

Sol. : If 𝒇(𝒙) = 𝐞𝒙 , then 𝒇(𝒏) (𝒙) = 𝐞𝒙 , and so, 𝒇(𝒏) (𝟎) = 𝟏.

Hence:

𝒇(𝒏) (𝒙𝟎 ) 𝟏
𝒂𝒏 = = ,
𝒏! 𝒏!

that is the Maclaurin series of the function 𝒇(𝒙) = 𝐞𝒙 is:

∞ ∞
𝒙
𝒇(𝒏) (𝟎) 𝒏 𝒙𝒏
𝒇(𝒙) = 𝐞 = ∑ 𝒙 = ∑
𝒏! 𝒏!
𝒏=𝟎 𝒏=𝟎

𝒙 𝒙𝟐 𝒙𝟑 𝒙𝒏
= 𝟏+ + + + ⋯+ + ⋯
𝟏! 𝟐! 𝟑! 𝒏!

Using the ratio test, we can easily deduce the radius of convergence as:

𝒂𝒏+𝟏 𝒙𝒏+𝟏 𝒂𝒏+𝟏 𝒏!

𝒓 = 𝐥𝐢𝐦 | | = |𝒙| 𝐥𝐢𝐦 | | = |𝒙| 𝐥𝐢𝐦
𝒏→∞ 𝒂𝒏 𝒙𝒏 𝒏 → ∞ 𝒂𝒏 𝒏 → ∞ (𝒏 + 𝟏)!

𝟏
= |𝒙| 𝐥𝐢𝐦 = 𝟎 < 𝟏.
𝒏 → ∞ 𝒏+𝟏

So, by the ratio test, the series converges for all 𝒙 and the radius of
convergence is 𝒓 = ∞. #

- 146 -
5.3.2: The binomial (expansion) series:

One of the most celebrated series of all times, the binomial series,
which is, exactly, the Maclaurin series of the function:

𝒇(𝒙) = (𝟏 + 𝒙)𝜶 , 𝜶 ∈ 𝓡. (𝟓. 𝟑. 𝟑)

Newton used it to estimate integrals and it can be used to give accurate

estimates of roots.

Theorem 5.3.1 (The binomial series) : If 𝜶 ∈ 𝓡, is any real

number and |𝒙| < 𝟏, then, the binomial series is given as:

𝜶(𝜶 − 𝟏) 𝟐 𝜶(𝜶 − 𝟏)(𝜶 − 𝟐) 𝟑

(𝟏 + 𝒙)𝜶 = 𝟏 + 𝜶𝒙 + 𝒙 + 𝒙 +⋯
𝟐! 𝟑!
𝜶(𝜶−𝟏)(𝜶−𝟐)⋯(𝜶−𝒏+𝟏)
+ 𝒙𝒏 + ⋯
𝒏!

𝜶 𝒏
= ∑∞
𝒏=𝟎 (𝒏) 𝒙 , |𝒙| < 𝟏, (𝟓. 𝟑. 𝟒)

where:

𝜶 𝜶(𝜶 − 𝟏)(𝜶 − 𝟐) ⋯ (𝜶 − 𝒏 + 𝟏) 𝜶
( ) = , 𝒏 ≥ 𝟏, ( ) = 𝟏,
𝒏 𝒏! 𝟎

is called the binomial coefficient.

Proof :

Using the Maclaurin series for:

𝒇(𝒙) = (𝟏 + 𝒙)𝜶 ,

one can easily deduce the binomial series as:

𝜶(𝜶 − 𝟏) 𝟐 𝜶(𝜶 − 𝟏)(𝜶 − 𝟐) 𝟑

𝒇(𝒙) = (𝟏 + 𝒙)𝜶 = 𝟏 + 𝜶𝒙 + 𝒙 + 𝒙 +⋯
𝟐! 𝟑!
- 147 -
 𝜶(𝜶−𝟏)(𝜶−𝟐)⋯(𝜶−𝒏+𝟏)
+ 𝒙𝒏 + ⋯
𝒏!

Note that (5.3.1): If 𝜶 is a non-negative integer, that is 𝜶 = 𝟏, 𝟐, 𝟑, 𝟒, ⋯ the

series terminates after (𝜶 + 𝟏) terms, because the coefficients from 𝒏 =
𝜶 + 𝟏 on are zero. But when 𝜶 is not an integer, the series is infinite. It
remains for us that the binomial series (𝟓. 𝟑. 𝟒), in general, converges when
|𝒙| < 𝟏. ( The proof is left as an exercise to the reader).

Examples:
𝟏
Ex. 5.3.5:Expand: (𝟏+𝒙)𝟐
as a power series.

Sol. : We use the binomial series with 𝒏 = −𝟐. The binomial coefficient is:

−𝟐 −𝟐(−𝟑)(−𝟒) ⋯ (−𝟐 − 𝒏 + 𝟏)
( ) =
𝒏 𝒏!

(−𝟏)𝒏 𝟐(𝟑)(𝟒) ⋯ (𝒏 + 𝟏)
= = (−𝟏)𝒏 (𝒏 + 𝟏)
𝒏!

and so, when |𝒙| < 𝟏, we have:

𝟏
𝟐
= (𝟏 + 𝒙)−𝟐
(𝟏 + 𝒙)

−𝟐 𝒏
= ∑∞
𝒏=𝟎 ( ) 𝒙 = ∑∞ 𝒏 𝒏 𝟐
𝒏=𝟎(−𝟏) (𝒏 + 𝟏)𝒙 = 𝟏 − 𝟏𝟐𝒙 + 𝟑𝒙 − ⋯, |𝒙| < 𝟏. #
𝒏

𝟏
Ex. 5.3.6 :Expand: 𝟏+𝒙
as a power series.

Sol. : We use the binomial series with 𝒏 = −𝟏. The binomial coefficient is:

- 148 -
 −𝟏 −𝟐(−𝟑)(−𝟒) ⋯ (−𝟏 − 𝒏 + 𝟏)
( ) =
𝒏 𝒏!

(−𝟏)𝒏 𝟐(𝟑)(𝟒) ⋯ (𝒏 + 𝟏)
= = (−𝟏)𝒏
𝒏!

and so, when |𝒙| < 𝟏, we have:

𝟏
= (𝟏 + 𝒙)−𝟏
𝟏+𝒙

−𝟏 𝒏
= ∑∞
𝒏=𝟎 ( ) 𝒙 = ∑∞ 𝒏 𝒏 𝟐
𝒏=𝟎(−𝟏) 𝒙 = 𝟏 − 𝒙 + 𝒙 − ⋯, |𝒙| < 𝟏. #
𝒏

Similarly, one can show that:

𝟏
= (𝟏 + (−𝒙))−𝟏
𝟏−𝒙

−𝟏
= ∑∞
𝒏=𝟎 ( ) (−𝒙)𝒏 = ∑∞ 𝒏 𝟐
𝒏=𝟎 𝒙 = 𝟏 + 𝒙 + 𝒙 + ⋯, |𝒙| < 𝟏. #
𝒏

Exercises 5:

(𝑨) Discuss the interval of convergence and the radius of convergence of the
following series:

𝒏𝒙𝒏 (𝟐𝒏)!𝒙𝒏 𝒏 𝒏 𝟏
(𝟏) ∑∞
𝒏=𝟎 𝒏
, (𝟐) ∑∞
𝒏=𝟎 , (𝟑) ∑∞ 𝒏
𝒏=𝟎 (𝟐) 𝒙 , (𝟒) ∑∞
𝒏=𝟎 𝒙𝒏 ,
𝟔 (𝟑𝒏)! (𝟐𝒏)!

(𝟐𝒙)𝒏 (−𝟏)𝒏 𝒙𝒏 𝒙𝒏 𝒏 𝒏
(𝟓) ∑∞
𝒏=𝟏 , (𝟔) ∑∞
𝒏=𝟎 , (𝟕) ∑∞
𝒏=𝟏 , (𝟖) ∑∞
𝒏=𝟎 ( ) 𝒙𝒏 ,
𝒏𝟐 √𝒏 𝒏𝟐 𝒏 𝟏𝟎𝟎

𝒏 𝟐 𝒙𝒏 𝟐 𝒏 𝒙𝒏 𝟏 𝒏−𝟏
(𝟗) ∑∞
𝒏=𝟎 𝟐
, (𝟏𝟎) ∑∞
𝒏=𝟏 , (𝟏𝟏) ∑∞
𝒏=𝟐 𝒙𝒏 , (𝟏𝟐) ∑∞
𝒏=𝟏 𝒙𝒏 ,
𝟏+𝒏 √𝒏 𝐥𝐧 𝒏 𝒏

𝟕 𝒏 𝒙𝒏 𝒙𝒏 𝒏
(𝟏𝟑) ∑∞
𝒏=𝟎 , (𝟏𝟒) ∑∞
𝒏=𝟏 , (𝟏𝟓) ∑∞
𝒏=𝟎 𝒙𝒏 , (𝟏𝟔) ∑∞ 𝒏
𝒏=𝟏 √𝒏𝒙 ,
𝒏! √𝒏 𝟏𝟎𝒏

- 149 -
 𝒄𝒐𝒔𝒏 𝒙 (𝟐𝒙−𝟓)𝒏 (𝒙−𝟐)𝒏 (𝟑𝒙+𝟔)𝒏
(𝟏𝟕) ∑∞
𝒏=𝟏 , (𝟏𝟖) ∑∞
𝒏=𝟏 , (𝟏𝟗) ∑∞
𝒏=𝟏 , (𝟐𝟎) ∑∞
𝒏=𝟏 ,
(𝒏)! 𝒏𝟐 𝒏𝟐 𝒏!

(𝒆𝒙)𝒏 𝒙𝟐𝒏+𝟏 𝒄𝒐𝒔𝒏 𝒙

(𝟐𝟏) ∑∞
𝒏=𝟎 , (𝟐𝟐) ∑∞
𝒏=𝟏 , (𝟐𝟑) ∑∞
𝒏=𝟏 , (𝟐𝟒) ∑∞ 𝒏
𝒏=𝟏(𝒏𝒙) ,
𝒏 +𝟏 𝟐𝒏+𝟏 𝒏𝒏

∞ 𝒏 ∞ ∞ ∞
𝒙𝟐 − 𝟏 𝒙𝒏 𝒙𝒏
(𝟐𝟓) ∑ ( ) , (𝟐𝟔) ∑ 𝒏 , (𝟐𝟕) ∑ 𝒏 𝟓 , (𝟐𝟖) ∑ √𝒏(𝒙
𝟐 𝒏𝟑 𝟓 𝒏
𝒏=𝟎 𝒏=𝟏 𝒏=𝟏 𝒏=𝟏

− 𝟏)𝒏 .

(𝑩) Find a power series representation for the function and determine the
interval of convergence:

𝟏 𝟑 𝒙𝟐 𝟒
(𝟏) , (𝟐) 𝟒
, (𝟑) 𝟐
, (𝟒) ,
𝟏+𝒙 𝟏−𝒙 (𝟏−𝟐𝒙) 𝒙𝟐 +𝒙−𝟐

𝟏+𝒙 𝒙 𝟏
(𝟓) 𝐥𝐧 ( ), (𝟔) , (𝟕) 𝐥𝐧(𝟑 + 𝒙), (𝟖) ,
𝟏−𝒙 𝟗 + 𝒙𝟐 𝟐𝟓 + 𝒙𝟐

𝒙 𝟏 𝒙
(𝟗) , (𝟏𝟎) 𝐭𝐚𝐧−𝟏 𝟐𝒙, (𝟏𝟏) , (𝟏𝟐) 𝐭𝐚𝐧−𝟏 ,
𝟏 + 𝟒𝒙 𝟏 − 𝒙𝟑 𝟑

𝟏 𝟏
(𝟏𝟑) , (𝟏𝟒) , (𝟏𝟓) 𝐥𝐧(𝟏 − 𝒙𝟐 ), (𝟏𝟔) 𝐬𝐢𝐧 𝒙𝟐 ,
(𝟏 − 𝒙)𝟐 (𝟏 − 𝒙)𝒏

𝟏+𝒙
(𝟏𝟕) 𝒙𝟐 𝐭𝐚𝐧−𝟏 𝒙, (𝟏𝟖) 𝒙 𝐬𝐢𝐧𝐡 𝒙𝟐 , (𝟏𝟗) 𝒙 𝐥𝐧(𝟏 + 𝒙𝟐 ), (𝟐𝟎) 𝐥𝐧 ( ).
𝟏−𝒙

(𝑪) Use differentiation to find a power series for:

𝟏
(𝟏) 𝒇(𝒙) = . What is the radius of convergence? Then, find a power
(𝟏+𝒙)𝟐

𝟏 𝒙𝟑
series for: 𝒈(𝒙) = (𝟏+𝒙)𝟑
, 𝒉(𝒙) = (𝟏+𝒙)𝟑
.

(𝟐) 𝒇(𝒙) = 𝐥𝐧 (𝟏 + 𝒙). What is the radius of convergence? Then, find a

power series for: 𝒈(𝒙) = 𝒙 𝐥𝐧 (𝟏 + 𝒙), 𝒉(𝒙) = 𝐥𝐧 (𝟏 + 𝒙𝟐 ).

- 150 -
(𝑫) Evaluate the indefinite integrals as a power series. What is the radius of
convergence?

𝒙 𝒅𝒙 𝐥𝐧 (𝟏 − 𝒙 )𝒅𝒙 𝒙 − 𝐭𝐚𝐧−𝟏 𝒙 𝒅𝒙
(𝟏) ∫ 𝟐
, (𝟐) ∫ , (𝟑) ∫ 𝟑
, (𝟒) ∫ 𝐭𝐚𝐧−𝟏 𝒙𝟐 𝒅𝒙.
𝟏−𝒙 𝒙 𝒙

(𝑬) Find the Maclaurin series for the given functions:

𝒙
(𝟏) 𝐜𝐨𝐬 𝝅𝒙, (𝟐) 𝒙 𝐭𝐚𝐧−𝟏 𝒙, (𝟑) 𝒆−𝟐 , (𝟒) 𝐬𝐢𝐧 𝒙𝟒 ,

(𝟓) 𝒙 𝐜𝐨𝐬 𝟐𝒙, (𝟔) 𝒙 𝐞−𝒙 , (𝟕) 𝐜𝐨𝐬 𝒙𝟑 , (𝟖) 𝒙𝟐 𝐜𝐨𝐬 𝒙𝟑 ,

(𝟗) 𝐥𝐧 (𝟏 + 𝒙), (𝟏𝟎) (𝟏 + 𝒙)−𝟑 , (𝟏𝟏) 𝒙 𝐞−𝒙 , (𝟏𝟐) 𝒙 𝐞−𝟐 .

- 151 -
 References

1− Richard R. Goldberg : Methods of Real Analysis,Oxford &

IBH Publishing Co. [ ] .

2− James Stewart: Calculus, Fifth Edition, Thomson Brooks/

Cole [2003].

3− S. L. Salas and Einar Hille : Calculus "One and Several

veriables", Fourth Edition, John Wiley & sons, Inc.[1982].

5− A. F. Bermant and J. G. Armanovich : Mathematical Analysis,

Fourth Printing, Mir Publishers Moscow [1989].

6− Frank Ayres, Jr. and Elliot Mendelson: Calculus , Fifth Edition ,

Schaum’s Outline Series, McGraw - Hill eBook, [2009].

- 152 -