You are on page 1of 4967

Lecture Notes On Advance d

Calculus II
Jie WuDepart ment of

Mathemati csNational University of Singapore

Contents
Chapter 1. Sequence s of Real Numbers 51 . S e q

u c 5 L o S n

e n e s 2 . i m i t s f e q u e c e s

5 3. Sequences which tend to

84 . Techniqu es For Computin

g Limits 9 5. The Least Upper Bounds and the Completeness Property of


R

126 . M o n o t o n e S e q u e n c e s 1 5 7 . S u b s e q u e

n c e s 1 6 8. The Limit Superior and Inferior of a Sequence 1 7 9. Cauchy

Sequences and the Completeness of


R

25C h a p t e r 2. Series of Real

Numbers 311 . S e r i e s 3 1 2 . T e s t s f o r P o s i t i v e

S e r i e s 3 4 3. The Dirichlet Test and Alternating Series 434. Absolute and

Conditiona l Convergen c e 4 9 5. Remarks on the various tests for convergence/di

vergence of series 50C h a p t e r 3. Sequences and Series of Functions

531 . P o i n t w i s e C o n v e r g e n c e 5 3 2 . U n i f o r m

C o n v e r g e n c e 5 7 3. Uniform Convergence of

{
F
n

}
a n d C o n t i n u i t y 6 3 4. Uniform Convergen ce and

Integratio n 655. Uniform Convergen ce and Dierentia tion 706 .

P o w e r S e r i e s 7 5 7. Dierent iation of Power

Series 838 . T a y l o r S e r i e s 8 7 B i b l i o

g h 9
3

r a y 7

CHAPTER 1

Sequences of Real Numbers


1. Sequences A sequence

is an ordered list of numbers. For example,1 , 2 , 3 ,

4 , 5 , 6The order of the sequence is important. For example,2

, 1 , 4 , 3 , 6

, 5is dierent from above sequence. An innite sequence is a list which does

not end.For example,1 , 1 / 2 , 1

/ 3 , 1 / 4 , 1

/ 5 ,

We are going to study innite

sequences. We denote by

{
a
n

}
the sequence

a
1

,a
2

,a
3

,a
n

Example

1.1
.

Here are some examples of innite sequences. (1). 1 , 12 ,

13 ,

, 1 n,


(2).13 , 13
2

, 13
3

, 13
4

(3). 1 ,


2 , 3 ,

4 , 5 ,

Can you nd a formula for

each of the above sequences? Answer: (1). a


n

=1

/n . (2). a
n

=1 / 3
n

. (3). (

1)
n 1

n .

2. Limits of Sequences
Definition

2.1
.

The limit of

{
a
n

}
is A

, and is written aslim


n

a
n

= A, if for any

> 0, there is a natural number N such that for every n>N , we have

|
a
n

< . Remark. 1. Some sequences do not satisfy the above. We call such sequences divergent

.2. Sequences which satisfy the above denition, i.e. A exists and is nite, are called convergent

sequences.
Example

2.2
.

Prove the following limits by using


N denition 1) lim
n

1 n

=0 .
5 6 1. SEQUENCES OF REAL NUMBERS

2) lim
n

n
2

n
2

+ 1= 1

. 3) lim
n

34

=0 .
Solution.

(1). Given any

> 0, we want to nd N such that

<

for n>N , i.e., n> 1 for n>N

. Choose N to be the smallest integer such that N

1 .( N is found now!) When n>N , then

n>N

1 or

<

. Thus lim
n

1 n = 0.(2). Given any >

0, we want to nd N such that

n
2

n
2

+1

< for n>N .Now

n
2

n
2

+1

<


n
2

+1

<

n
2

+1

n
2

+1

<

( n
2

+ 1)

n
2

+ 1( n +

n
2

+ 1)

<

n
2

+ 1( n +

n
2

+ 1)

<

n
2

+ 1( n +


n
2

+ 1) > 1

Observe that

n
2

+ 1( n +


n
2

+ 1) >n for n


1. Choose N to be the smallest integer such that N

. Then, for n>N ,

n
2

+ 1( n +

n
2

+ 1) >

N
2

+ 1( N

N
2

+ 1) >N

1 or

n
2

n
2

+1

< . Thus N

is found and hence the result.(3). Given any > 0, we want to nd N

such that

34

<

for n>N .Observe that

34

<

ln

34

< ln( )


n> ln( )ln(3 / 4)(

Note. ln(3 / 4) < 0!!) Choose N

to be the smallest positive integer such that N

ln(

)ln(3 / 4). When n>N , then n>N


ln( )ln(3 / 4)or

34

<

. The proof is nished.


Theorem

2.3
.

If

a
n

}
has a limit, then the limit is unique.
2 . L I M I T S O F

S E Q U E N C E S 7 Proof.

Let A and B be limits of

{
a
n

}
. Suppose that A

= B . Choose =

|
A

|
2. Then

> 0 because A = B

. By denition, there exists N


1

and N
2

suchthat

|
a
n

< for n>N


1

and

|
a
n

|
< for n>N

. For n> max

{
N
1

,N

}
, wehave

|
A

|
=

|
( A


a
n

)+( a
n

B )

||
A

a
n

|
+

|
a
n

|
< 2 =2

|
A

|
2=

|
A

|
,

which is a contradiction. Thus A = B .

Theorem

2.4 (Squeeze or Sandwich Theorem)


.

Given 3 sequences

{
a
n

}
,

b
n

}
,

{
c
n

}
such that (i) a
n

c
n

for every n and (ii) lim

a
n

= A = lim
n

c
n

, then lim
n

b
n

= A .
Proof.

For any > 0, there exists N

and N
2

such that

|
c
n

|
< for n>N

and

|
a
n

|
< for n>N
2

. Let N

= max

{
N
1

,N
2

. For n>N , we have

<c
n


A< and

<a
n


A< A

<c
n

<A

+ and A

<a
n

<A + . Thus A

<a

b
n

c
n

<A + or

|
b
n

|
< . By denition, we have lim

b
n

= A and hence the result.

Remark. The above theorem is still applicable if the inequality a


n


b
n

c
n

is true

eventually .
Example

2.5
.

Find limits 1) lim


n

1 + sin nn . 2)

3 n

14 n +1

.
Solution.

(1). Since0


1 + sin nn

2 n

and lim
n

2 n = lim
n

0 = 0, we havelim
n

1 + sin nn =0 .

(2). Since0

3 n


14 n +1

34

8 1. SEQUENCES OF REAL NUMBERS

and lim

34

= lim
n

0 = 0, we havelim
n

14 n +1

=0 .

3. Sequences which tend to

Definition

3.1
.

a
n

}
tends to +

if for each positive number k , there is an N such that a


n

>k for all n > N. Remark. For such sequences, we write as a

as n


orlim
n

a
n

=+


.
Example

3.2
.

The following sequences tend to +


1) a
n

ln

n .2) a
n

= (3 / 2)
n

.The sequences

ln n ,

n
2

and etc then tend to

.
Theorem

3.3 (Reciprocal Rule)


.

Consider a sequence

{
a
n

}
. (i) If a
n

>

0 for all n and lim


n

a
n

=0 , then lim
n

a
n

=+

. (ii) If lim
n

a
n

, then lim
n

1 a
n

=0 .
Proof.

We only prove (i). For each positive integer k , there exists N such that

a
n

<

1 k for n>N because lim


n

a
n

= 0. Then, for n>N , a


n

>k

because a
n

> 0. This nishes the proof.


Example

3.4

Since lim
n

= 0, we have lim
n

n =


. Similarly, sincelim
n

=+

, we have lim
n

n = 0.
4. TECHNIQUES FOR COMPUTING LIMITS 9

4. Techniques For Computing Limits


Theorem

4.1
.

Let f

be a continuous function. Then lim


n

f ( a
n

)= f ( lim
n

a
n

) .

Idea of Proof.

By the denition of continuity, when x

x
0

, f ( x )


f ( x
0

).Now lim
n

a
n

= A means that a
n

A when n

. Thus f (

a
n

f ( A

) when n

, that is,lim
n

f (

a
n

)= f ( A )= f

( lim
n

a
n

) .
Example

4.2
.

lim
n

sin

n 2 n +1

= lim
n

2+1 /n

= sin

=1 .
Theorem

4.3 (LHopitals Rule)


.

Suppose a
n

= f

( n ) , b
n

= g

( n ) . If lim
n

( n ) g ( n ) is of the form


or 00 , then lim
n

( n ) g ( n )= lim
n

f ( n ) g (

n ) . History Remark. Although the theorem is named after

Marquis de lHospital (1661 -1704), it should be called Bernoullis rule. The story is that in 1691, lHopitalasked

Johann Bernoulli (1667-1748) to provide, for a fee, lectures on the new subject of calculus. LHopital

subsequently incorporated these lectures into the rstcalculus text, LAnalyse des inniment petis

(Analysis of innitely small quantities) ,published in 1696. The initial version of what is now known as lHopitals

rule rstappeared in this text.


Example

4.4
.

Show that lim


n

1+ xn

= e
x

Proof.

lim
n

ln

1+ xn

= lim
n

n ln

1+ xn

= lim
n

ln

1+
xn

1 /n = lim
n

11+
xn

xn
2

1 n
2

= lim
n

x 1+
xn

= x. Thus lim

1+ xn

= e
x

Theorem

4.5
.

If lim
n

and lim
n

b
n

exist, then (1). lim

( a
n

+ b
n

) = lim

a
n

+ lim
n

b
n

(2). lim
n

ka
n

= k lim
n

a
n

, (3). lim
n

a
n

b
n

= lim
n

a
n

lim
n

b
n

, (4). lim
n

a
n

=lim
n

a
n

lim
n

, provided b
n

= 0 and lim
n

b
n

=0 .
10 1. SEQUENCES

OF REAL NUMBERS Proof.

omitted.
Example

4.6
.

Find the limit of

ln

n
2

+3 n

+ 22 + 4 n +2 n
2

+ cos

.
Solution.

lim
n

ln

n
2

+3 n + 22 + 4 n +2 n

+ cos

= lim
n

ln

( n
2

+3 n + 2) /n
2

(2 + 4 n +2 n
2

) /n
2

+ cos

= lim

ln

1+3 /n +2 /n
2

2 /n
2

+4 /n +2

+ cos

= ln

1 + 0 + 00 + 0 + 2

+ cos0 = ln

12

+1=1

ln2

.
Theorem

4.7 (Some Standard Limits)


.

Some standard limits are given as follows. 1. lim


n

1 n
p

=0 for any xed p> 0 . 2. lim


n

c
n

=0 for any xed c where

|
< 1 . 3. lim
n

c
1

=1 for any xed c> 0 . 4. lim

n
n

n =1 . 5. lim

n
p

c
n

=0 for any xed p

and c> 1 . 6. lim


n

c
n

n != 0 for any xed c . 7. lim


n

1+ xn

= e
x

for any xed

x . 8. lim
n

(ln n )
p

n
k

=0 for any xed k> 0 .


Proof.

Assertion 7 was proved in Example 4.4. 1. lim


n

n
p

lim
n

1 n

=0

=0 . 2. Case 1: When c = 0, the statement is

obvious.Case 2: When c> 0, we haveln

lim
n

c
n

= lim
n

ln c
n

= lim
n

n ln c =

. Thus, lim

c
n

= 0.Case 3: When c< 0, we have

|
c

|
n

|
c

|
n

for all n

. By Case 2, we havelim
n

|
c

|
n

) = 0 = lim
n

|
c

|
n

. Hence by Squeeze theorem, we also have lim


n

c
n

= 0. 3. lim
n

c
1

= c
lim
n

= c
0

= 1.

4. ln

lim
n

= lim

ln
n

n = lim
n

ln nn = 0 (by LHopitals rule) .


4. TECHNIQUES

FOR COMPUTING LIMITS 11

Thus, lim
n
n

= e
0

=1 . 5. Let k

be a xed positive integer such that p

k< 0. Thenlim

n
p

c
n

= lim
n

pn

p 1

c
n

ln c = lim p

( p

1) n
p 2

c
n

(ln c )
2

= lim
n

p ( p

1)


( p

k + 1)

n
p k

c
n

(ln c )

= lim
n

p ( p

1)

( p

+ 1) c
n

n
k p

(ln c

)
k

= 0by LHopitals rule. 6. Let a

= c
n

n != c

cn ( n

1)

1. Now x an integer M>c . Then for any n>M

,0 <a
n

= c


cn ( n

1)


( M + 1) a
M

<cna
M

. Note that a
M

is a xed number because M

is xed. Since lim


n

0 = 0 = lim
n

cna
M

,by the Squeeze theorem, lim


n

a
n

= 0. 8. Let

m = ln n . Then n = e
m

. By (5),lim
n

(ln n )
p

n
k

= lim
m

m
p

e
km

= lim
m

m
p

( e
k

)
m

=0 ,

where e
k

> 1 because k> 0.

Strategy: One can nd the limits of many sequences from those of the standardsequenc es.

Example

4.8
.

Find the limits 1) lim


n

8
n

+ (ln n )
10

+ n ! n

n ! . 2) lim
n

12 n

+1

3 n

.
Solution.

(1).lim
n

8
n

+ (ln n )
10

+ n ! n
6

!= lim
n

8
n

/n ! + (ln n )

10

/n !+1 n
6

/n !

1=0 + 0 + 10

1=

1 .

(2).lim
n


12 n +1

3 n

= lim
n

1+


12 n +1

2 n +1

n
2

n
+1

= lim
n

1+


12 n +1

2 n +1

32+1

/n

e
1

32

=1 e

e
12 1. SEQUENCES OF REAL NUMBERS

5. The Least Upper Bounds and the

Completeness Property of
R

5.1. From Natural Numbers to Real Numbers.

Starting with natural numbers


N

{
1

, 2 , 3 ,

, we obtain real numbers


R

by adding more and more new numbers in the following steps:

Step 1. By adding 0 and negative numbers, we have integers


Z

{
0 ,

1 ,


2 ,

3 ,

}
. Step 2. Then we have rational numbers
Q

pq

p,q

,q =0

. Step 3. Then, by adding irrational

numbers, we have all real numbers.Below we give some examples of irrational

numbers. Recall that a natural num-ber p> 1 is called prime if p

is NOT divisible by any natural numbers other than p and 1. For instance, 2, 3, 5, 7, 11,


are primes. Every natural number n> 1admits a unique (prime)

factorization n = p
1

p
2


p
k

, where each p
i

is prime. For instance, 20 = 2

5 and 66 = 2

11.
Example

5.1
.

If n is a natural number, and there is no natural number whose square is n

, then

n is NOT a rational number. In particular,


2 ,

3 ,


5 ,

are irrational numbers.


Proof.

Suppose that

is a rational number. We can write

n as ab

, where a,b

and b

= 0. Then

n = ab


n = a
2

b
2


a
2

= b
2

n.

Any prime occurring in the (unique) factorization of a will occur an even number of

times in the factorization of a


2

; similarly for b and b

. By a
2

= b
2

, any prime thatoccurs in the factorization of n must occur an even number of times, since all

primesoccurring in the factorization of b


2

n are exactly those occurring

in the factorizationof a
2

.Thus n can be written as

n =( p
1

p
2


p
k

)
2

, where, of course, the

p
i

s need not be distinct. This, however, is a contradiction tothe

hypothesis, since n is expressed as the square of a natural number.

5.2. Bounded Sets.


Definition

5.2
.

A set of real numbers S

is bounded above if there exists a nitereal number M such that x


M x

S.M is called an upper bound of S .


Definition

5.3

A set of real numbers S is bounded below

if there exists a nitereal number m such that m

x x


S.m is called a

lower bound of S .
5. THE LEAST UPPER BOUNDS AND THE COMPLETENESS PROPERTY OF

13 Definition

5.4
.

A set which is both bounded above and

below is called a bounded set . Remark. 1. Upper bounds and lower bounds

are not unique.2. Some sets only have upper bounds but not lower bounds.3. Some sets have only lower bounds

but not upper bounds.4. A set which is not bounded is called an unbounded set.
Example

5.5
.

Let S =

{
r

|
r is a rational number with r<

}
. Then S is bounded above.
Theorem

5.6
.

Every convergent sequence is bounded.


Proof.

Let

{
a
n

}
be a sequence convergent to A

. For = 1, there exists N suchthat

|
a

|
< 1 or A


1 <a
n

<A + 1 for n>N

. Choose M and m to be thelargest and smallest number of the nite numbers

a
1

,a
2

,...,a
N

,A +1 ,A


1 , respectively. When n


N , we have m

a
n


M because M ( m

) is the largest(smallest ) number of the above nite set. When n>N , we have m

1 <a
n

<A

+1

M. Thus, for all n , we have m


a
n

M and so

{
a
n

}
is bounded. The proof is nished.

Corollary

5.7 (Test for divergence)


.

If

{
a

}
is unbounded, then

{
a
n

}
diverges. Remark. 1. The converse may not be true, i.e., divergent sequence need

not be unbounded.2. The inverse may not be true, i.e., a bounded sequence may not be convergent.

Example. The sequence

{
1 ,

1 , 1 ,

1 ,

}
is bounded but NOT convergent. 5.3. Inmum and Supremum.

Recall that any nite set of real numbers hasa greatest element (maximum) and a least element (minimum).
Example

5.8
.

{
2 . 5 ,

3 . 1 ,

4 .

4 , 4 . 5 , 5

}
However, this property does not necessarily hold for innite sets.
Example

5.9
.

{
1 , 2 ,

3 , 4 ,

}
.
Definition

5.10
.

A real number M (

) is called the least upper bound

or supremum of a set E if (i) M is an upper bound of

E , i.e., x

M for every x


E , and( i i ) i f M <M , then

M is not an upper bound of E (i.e., there is an x


E such that M <x ).We write

M = sup E . Remark. (i) sup E

is unique whenever it exists.(ii) The main di erence between sup E and max E

is that sup E may not be anelement of E , whereas max E

must be an element of E if it does exist).(iii) If E has a maximum, then sup

E = max E .
14 1. SEQUENCES OF REAL NUMBERS Example

5.11
.

1. Let E =

{
r

|
0

}
. Then sup E =


2 but max E does not exist because

2 is not a rational number, that is, sup E

.2. Let E =

{
1 / 2

, 2 / 3 , 3 / 4

, 4 / 5 , 5 / 6

}
. Then sup E = 1 and max E

does not exist.3. Let E =

{
1 ,

1 / 2 , 1 / 3 ,

1 / 4 , 1 / 5 ,

}
. Then max E = 1 = sup E .
Definition

5.12
.

A real number m ( =


) is called the greatest lower bound or inmum of a set

E if (i) m is a lower bound of E , i.e., m


x for every x

, and( i i ) i f m >m , then m

is not a lower bound of E (i.e., there exists an x

E such that x<m ).We write m = inf E

. Remark. (i) inf E is unique whenever it exists.(ii) The

main di erence between inf E and min E is that inf E

may not be anelement of E , whereas min E must be an element of E

if it does exist.(iii) If E has a minimum, then inf E = min E

.
Example

5.13
.

1. Let E =

1 , 1 / 2 , 1 /

3 , 1 / 4 ,

}
. Then inf E = 0 but min E

does not exist.2. Let E =

{
r

|
0

}
. Then min E = inf

E = 0. 5.4. The Completeness of


R

. Consider the set

E =

{
r

|
r
2

< 2

}
. Then

E is a bounded subset of rational numbers, but sup E =


2 is NOT a rationalnumber. For considering sup and inf of bounded subsets of

rational numbers, wemay obtain irrational numbers. For bounded subsets of real numbers,

sup and inf are still real numbers. This is called completeness property of

. In details, wehave the following.


Theorem

5.14 (Completeness Axiom of

)
.

The following statement hold for subsets of real numbers: (i)

If E is bounded above, then sup E exists. (ii)

If E is bounded below, then inf E exists. Remark.

For assertion (i), it just means that if a subset of real numbers E isbounded above, then sup

E exists as a real number. Compare with rational case: if a subset of

rational numbers E is bounded above, then sup E exists only as a real

number, but it need not be a rational number. Remark. Richard Dedekind (a German

mathematician, 1831-1916), in 1872, usedalgebraic techniques to construct real number system


R

from
Q

. His basic ideas areas follows.Given a rational number r

, we can construct two sets U =

{
x

|
x

}
and L =

{
x

|
x<r

. (One can also construct U =

{
x

|
x>r

}
and

L =

{
x

|
x

}
.) The sets

U and L have the property that(1). U and L

are subsets of
Q

;(2). U

L =

;(3). U =

;(4). L

;(5).

L =

; and

6 . M O N O T O N E S E Q U E N C E S 1 5

(6). every element in U

is greater than every element in L .Such a paring ( U,L ) is called a Dedekind cut

. Then we can use inf U (or sup L ) todene a new

number. This is Dedekinds idea to construct all real numbers by usingrational numbers. For instance, let U

{
x

x
2

> 2

}
and L

{
x

x
2

> 2

}
.Then inf U

= sup L =

2. Another way to construct real numbers using

rationalnumbers was introduced by Georg Cantor (18451917). We will explain Cantorsideas in the section of

Cauchy sequences.Recal l that a set E is bounded if and only if it is bounded above and

boundedbelow. Thus the Completeness Axiom leads to


Corollary

5.15
.

If

E is bounded, then both sup E and inf E

exist. 6. Monotone Sequences


Definition

6.1
.

{
a

}
is called monotone increasing (decreasing) if

a
n

) a

n +1

for every n , that is, a


1

a
2

a
3

( a
1

a
3

).
Example

6.2
.

1. The sequence

{
1 /n

is monotone decreasing.2. The sequence

{
1 / 2

, 2 / 3 , 3 / 4

, 4 / 5 , 5 / 6

}
is monotone increasing.
Proposition

6.3
.

A monotone increasing (decreasing) sequence is bounded below (above).


Proof.

Let

{
a
n

}
be a monotone increasing

sequence, that is, a


1

a
2


a
3

. Then a

is a lower bound for

{
a
n

and hence the result.


Theorem

6.4 (Monotone Convergence Theorem)


.

Let

{
a
n

}
be a sequence. (i)

If

{
a
n

}
is monotone increasing and

bounded above, then

{
a
n

is convergent and lim


n

a
n

= sup
n

a
n

. (ii) If

{
a
n

}
is monotone decreasing and bounded below, then

a
n

}
is convergent and lim
n

a
n

= inf
n

a
n

.
Proof.

(i). Suppose

{
a
n

}
is monotone increasing and bounded above.

Thenby the Completeness Axiom of


R
n

, sup a
n

exists (nite). Now, given > 0, sincesup


n

a
n

< sup
n

a
n

, it follows that sup


n

is not an upper bound of

a
n

}
. In otherwords, there exists an integer N

such that a
N

> sup
n

a
n


. Then for all n>N , wehavesup
n

a
n


<a
N

a
n


sup
n

a
n

< sup
n

a
n

+ (since n>N ) .

16 1. SEQUENCES OF REAL NUMBERS

Equivalently,

sup
n

a
n

<

for all n>N and so lim


n

a
n

= sup
n

a
n

(exists).The proof of (ii) is similar.


Example

6.5
.

Let a
n

= nn + 1, that is,

a
n

}
=

{
1 /

2 , 2 / 3 , 3 /

4 ,

}
. Then a
n

ismonotone increasing and bounded above. Thussup


n

a
n

= lim
n

a
n

=1 .
Corollary

6.6
.

If

{
a
n

}
is monotone increasing

(decreasing), then either (i)

{
a
n

is convergent or (ii) lim


n

a
n

=+

) .
Proof.

Suppose

a
n

}
is monotone increasing, then either

a
n

}
is bounded aboveor not bounded above. Case (a)

: If

{
a
n

}
is bounded above, then by

the Monotone Convergence Theorem,

{
a
n

converges. Case (b) : If

{
a
n

is not bounded above, then

{
a
n

has no upper bounds. Thus forany given k> 0, k is not an upper bound of

{
a
n

}
. In other words, there exists

N such that a
N

> k. Since

a
n

}
is monotone increasing, it follows that for all

n>N , a
n

a
N

> k.

Therefore, lim
n

a
n

=+

.The proof for the case when

{
a
n

is monotone decreasing is similar.

7. Subsequences
Example

7.1
.

The following are the subsequences of

{
a
n

{
1 ,

, 1 ,

1 , 1

1 ,

}
.

{
a
2 n 1

}
=

{
1 , 1 , 1 ,

}{
a
2 n

}
=

{
1 ,

1 ,


1 ,

}
.In general, subsequences

of

{
a
n

}
are of the form

{
a
n
k

}
, k

=1 , 2 , 3 ,..., with n

<n
2

<n
3

<

Note. The rule is that we should choose a


n
1

rst and then a

n
2

with n
2

>n
1

andthen a
n

with n
3

>n
2

, so far and so on (up to innite). Thus

n
1

is at least 1, n
2

isat least 2, n
3

is at least 3,


.
Theorem

7.2
.

Suppose lim
n

a
n

= A . Then every subsequence of

a
n

}
also con-verges to A , that is,

lim
k

a
n
k

= A.
Proof.

For any given > 0, since lim


n

a
n

= A

, there exists N such that

|
a
n

|
< for all n > N. Then for all k>N

, we have n
k

k > N.
8. THE LIMIT SUPERIOR AND

INFERIOR OF A SEQUENCE 17

Hence

|
a
n
k

|
< for all k > N. Therefore, lim
k

a
n
k

= A .
Corollary

7.3

Suppose that

{
a
n

has two subsequences that converge to dier-ent limits. Then

{
a

}
is divergent.
Example

7.4
.

The sequence

{
1 ,

1 ,

1 ,

1 ,

is divergent because

{
a
2 n 1

}
=

{
1 , 1

}
converges to 1 and

a
2 n

}
=

{
1

1 ,

}
converges to


1 . 8. The Limit Superior and Inferior of a Sequence

Given a sequence

{
a
n

, we can form another sequence

{
b
n

given by b
n

= sup
k n

a
k

= sup

{
a
n

,a
n +1

,a

n +2

}
.
Example

8.1
.

Let

{
a
n

}
=

{
1 ,

1 ,

1 ,

1 ,

. Then b
n

= sup
k n

a
k

= sup

{
1 , 1

1 , 1

}
=1 .
Proposition

8.2
.

For any sequence

{
a
n

}
, the associated sequence

{
b
n

}
=

sup
k n

a
k

is always monotone decreasing.


Proof.

For each n , b

= sup

{
a
n

,a
n +1

,a
n +2

}
sup

a
n +1

,a
n +2

= b
n +1

.
Definition

8.3
.

The limit superior of

{
a
n

, denoted by limsup a
n

or limsup
n

a
n

or lim
n

a
n

is dened to be lim
n

b
n

, i.e.lim
n

a
n

= lim
n

b
n

= lim
n

sup
k n

.
Example

8.4
.

1. Let

{
a

}
=

{
1 ,


1 , 1 ,

1 , 1 ,

1 ,

}
.lim
n

a
n

= lim
n

b
n

= lim
n

1=1 . 2. Let

{
a
n

}
=

1 , 2 , 3 ,

. Then b
n

= sup
k n

a
k

= sup

{
n,n +1 ,

=+

and so lim
n

a
n

= lim

b
n

=+

.3. Let

{
a
n

}
=

1 ,

2 ,

3 ,

}
. Then b
n

= sup
k

a
k

= sup

{
n,

1 ,

n and so lim
n

a
n

= lim
n

b
n

Theorem

8.5
.

Given any sequence

{
a
n

}
, either (1). lim
n

a
n

exists (nite), or

18 1. SEQUENCES OF REAL NUMBERS

(2). lim
n

a
n

=+


, or (3). lim
n

a
n


.
Proof.

If

{
a
n

}
is not bounded above, then each b
n

is +


, and thuslim
n

a
n

= lim
n

b
n

=+

. If

a
n

}
is bounded above, then each b
n

is nite. Since

{
b
n

is monotone decreasing,by Corollary 1.7.3,

{
a
n

converges (to a nite limit), or lim


n

b
n


. Similarly, given any sequence

a
n

}
, we can form another sequence

c
n

}
given by c
n

= inf
k

a
k

= inf

{
a
n

,a
n +1

,a
n +2

.
Definition

8.6
.

The limit inferior of

a
n

}
, denoted by liminf a
n

or liminf

a
n

or lim
n

a
n

is dened to be lim
n

c
n

, i.e.lim
n

a
n

= lim
n

c
n

= lim
n

inf
k n

a
k

.
Example

8.7
.

1. Let

{
a
n

}
=

{
1 ,

1 ,

1 ,

1 , 1 ,


1 ,

}
.lim
n

a
n

= lim
n

c
n

= lim
n

(inf

{
1 , 1

1 , 1

) = lim
n

1=


1 . 2. Let

{
a
n

}
=

{
1 , 2

, 3 ,

}
. Then c
n

= inf
k n

a
k

= inf

n,n +1 ,

}
= n and so lim

a
n

= lim
n

c
n

=+


.3. Let

{
a
n

{
1 ,

3 ,

}
. Then

c
n

= inf
k n

a
k

= inf

{
n,

}
=

and so lim
n

a
n

= lim
n

c
n


.
Proposition

8.8
.

(i). As in Proposition 8.2,

for any given sequence

{
a
n

, the associated sequence

{
c
n

}
=

{
inf
k n

a
k

is always monotone increasing.(ii). As in Theorem 8.5, for any given

a
n

}
, lim
n

either exists (nite), or +

,or

). Remark. We always havelim


n

a
n


lim
n

a
n

because c

b
n

.
Proposition

8.9
.

(i). If lim
n

a
n

= B with

B =

, then given >

0 , there exists N such that a


n

<B +

for all n>N .(ii). lim


n

a
n

= C with C

=+


, then given > 0 , there exists N such that

a
n

>C

for all n>N

.
8. THE LIMIT SUPERIOR AND INFERIOR OF A SEQUENCE 19 Proof.

(i). If B =+


, the assertion is obvious and so we assume that B

isnite. Since lim


n

a
n

= B , given any

> 0, there exists N such that for all n>N ,

b
n

|
<


b
n

<B +


sup

{
a
n

,a
n

+1

}
<B + , i.e.

a
n

,a
n

+1 ,

<B

+ for all n>N . Proof of (ii) is similar.

Warning!!

Given a sequence

{
a
n

}
, lim

a
n

is a dierent concept from sup


n

. Fromthe denition, we have b


1

= sup
n

= sup

{
a
1

,a
2

}
b
n

= sup

{
a
n

,a
n +1

,a
n +2

with b
1

b
2

b
3

and lim
n

a
n

= lim
n

b
n

. Thus we have the relationlim


n

a
n

sup
n

a
n

b
1

, but lim
n

a
n

need not be equal to sup


n

a
n

in general. Similarly,lim
n

a
n

inf
n

a
n

c
1

, but need not be equal to in general.The correct

understanding is that lim is the largest subsequential limit of convergent subsequences, including those

possible subsequences tending to +

or

. Similarly, lim is the smallest subsequential limit . This is described in

thefollowing theorem.
Theorem

8.10
.

Let

{
a

}
be any sequence. Let B = lim
n

a
n

and let C = lim


n

a
n

. (i) Let

{
a
n
k

}
be any subsequence of

{
a
n

}
such that lim
k

a
n
k

exists, +

, or

. Then

C = lim
n

a
n

lim

a
n
k

lim
n

a
n

= B. (ii) There exists a subsequence

{
a
n
k

}
of

{
a
n

}
such that lim
k

a
n
k

= B. (iii) There exists a subsequence

{
a
m
k

}
of

{
a
n

}
such that lim
k

a
m
k

= C.
Proof.

Let b

= sup

{
a
n

,a
n +1

}
and let c
n

= inf

a
n

,a
n +1

}
.

(i). Since n
k

k , we have c
k

= inf

{
a
k

,a
k +1

}
a
n
k

b
k

= sup

{
a
k

,a
k +1

}
and so C = lim
k

c
k


lim
k

a
n
k

lim
k

b
k

= B.

(ii). We consider three cases B =+


or nite. Case I. B =

.
20 1. SEQUENCES OF REAL NUMBERS

Since b
n

= sup

{
a
n

,a
n +1

,a
n +2

}
a
n

and lim
n

= B =

, we havelim
n

a
n

= B

. In this case, we can choose

{
a
n

itself as a subsequence withthe desired property. Case II. B =+


. In this case we are going to construct a subsequence

a
n
k

}
of

{
a

}
such that lim
k

a
n
k

B =+

.Since b
1

b
2

b
n

= lim
n

b
n

=+

we have b
1

= b
2

=+

, that is b
n

=+


for all n . Since b
1

= sup

{
a
1

,a
2

=+

, there exists n
1

suchthat a
n

> 1 because 1 is NOT an upper bound of

{
a
1

,a
2

}
. Since b
n
1

+1

= sup

{
a
n
1

+1

,a
n
1

+2

,a
n
1

+3

}
=+

, there exists a
n
2

such that n
2

>n

and a
n
2

> 2 because 2 is NOT an upper boundof

{
a
n
1

+1

,a
n
1

+2

,a
n
1

+3

}
. Now, by induction, suppose that we

have constructed a
n
1

,a
n
2


,a
n
k

such that n
1

<n

<

<n
k

and a
n

>s for 1

. Since b
n
k

+1

= sup

{
a

n
k

+1

,a
n
k

+2

=+

, there exists a
n
k
+1

such that n
k +1

>n
k

and a
n
k

+1

>k + 1because k + 1 is NOT an upper bound of

{
a

n
k

+1

,a
n
k

+2

,a
n
k

+3

}
. The induction isnished and so we obtain a subsequence

a
n
1

,a
n
2

with the property that a


n
k

>k for any k . Since lim

k =+

, we havelim
k

a
n
k

=+

= B. Case III.

B is nite. We are going to construct a subsequence

{
a

n
k

}
of

{
a
n

}
suchthat lim
k

a
n
k

B .Since b
1

= sup

{
a
1

,a
2

}
,b
1

1 is not an upper bound of

{
a
1

,a
2

}
and so there exists a
n
1

such that a

n
1

>b
1

1 . Since

b
n
1

+1

= sup

{
a
n
1

+1

,a
n
1

+2

}
,b
n

+1

12is not an upper bound of

{
a
n

+1

,a
n
1

+2

and so there exists a


n
2

suchthat n
2

>n

and a
n
2

>b
n
1

+1


12 .
8. THE LIMIT SUPERIOR AND INFERIOR OF A SEQUENCE 21

Now, by induction, suppose that we have constructed a


n
1

,a

n
2

,a
n
k

such that n

<n
2

<

<n
k

and

a
n
s

>b
n
s

+1

1 s for 1

k . Since b
n
k

+1

= sup

a
n
k

+1

,a
n
k

+2

}
,b
n
k

+1

k + 1is not an upper bound of

{
a
n
k

+1

,a
n
k

+2

}
and so there exists

a
n
k
+1

such that n
k +1

>n
k

and a
n
k
+1

>b
n
k

+1

1 k +1 . The induction is nished and so we obtain a subsequence

{
a
n
1

,a
n
2

}
with theproperty that a
n
k

>b
n
k

+1

1 k for any k

. Consider the inequality b


n
k

+1

k< a
n
k

b
n
k

. Since

{
b
n

}
is convergent, we havelim
k

b
n
k

= lim
k

b
n
k

+1

= lim
n

b
n

= B andlim

( b
n
k

+1

1 k )= B

0= B.

Thus, by the Squeeze theorem, we havelim


k

a
n
k

= B = lim
n

a
n

(iii). The proof is similar to that of (ii). We leave it to you as a tutorial question.
Example

8.11
.

Find the limit inferior and limit superior of the following se-quences i)

2(


1)
n

n 3 n +2

, ii)

(1 + (

1)
n

)sin n 4

iii)

{
[1 . 5+(

1)
n

]
n

}
.
Solution.

(i). Note that

a
2 k

=1

2 k 3

2 k + 2lim

a
2 k

= lim
k

1 /k


46 + 2 /k =

46=


23 a
2 k 1

=1 + 2


(2 k

1)3

(2 k

1) + 2lim
k

a
2

k 1

= lim
k

1 /k +2


(2

1 /k )3


(2

1 /k )+2

/k =46=23Thus the subsequential limits are

23and solim
n

a
n

=23a n d l i m
n

a
n


23 .
22 1. SEQUENCES OF REAL NUMBERS

(ii). The sequence

(1 + (

1)
n

)sin n 4

0 , 2sin2

4= 2 , 0 , 2sin4 4= 0

, 0 , 2sin6 4=

2 , 0 , 2sin8 4= 0 ,

0 , 2sin10 4= 2 ,

. The subsequential limits are


2, 0 and 2. Thuslim
n

a
n

= 2 and lim

a
n

2 .

(iii). Note that a


2 k

=2 . 5
2 k

lim
k

a
2 k

= lim
k

2 . 5
k

=+

2 k 1

=0 . 5
2 k

lim
k

a
2 k 1

= lim

12

(2

k 1) /k

= lim
k

12

2 1 /k

=0

=0 . The subsequential limits are 0 and +


. Thuslim
n

a
n

=+


and lim
n

a
n

=0 .

The following theorem was originally proved by Bernhard Bolzano (17811848)and

modied slightly by Karl Weierstrass (1815-1897).


Corollary

8.12 (BolzanoWeierstrass)
.

Every bounded sequence has a conver-gent subsequence.


Proof.

Let

a
n

}
be a bounded sequence. Since

{
a

}
is bounded,

< inf

a
1

,a
2

}
= c

lim
n

a
n

lim
n

a
n

b
1

= sup

{
a
1

,a
2

}
< +

. Thus lim
n

a
n

is nite. By Part (ii) of Theorem 8.10, there exists a

convergent subse-quence

{
a
n
k

of

{
a
n

}
such thatlim
k

a
n
k

= lim
n

a
n

Theorem

8.13
.

lim
n

a
n

= lim
n

a
n

(nite, +

) if and only if lim


n

a
n

exists (nite), +

, or

.
Proof.

Suppose that lim


n

a
n

= lim
n

a
n

= A . Let b
n

= sup

{
a
n

,a
n +1

}
and let c
n

= inf

{
a

,a
n +1

}
. Then c

= inf

{
a
n

,a
n +1

}
a
n

b
n

= sup

{
a
n

,a
n +1

}
. By the assumption, we havelim
n

= lim
n

a
n

= lim
n

= lim
n

b
n

.
8. THE LIMIT SUPERIOR AND

INFERIOR OF A SEQUENCE 23

By the Squeeze theorem, the sequence

{
a
n

}
converges andlim
n

a
n

= lim

a
n

= lim
n

a
n

Conversely suppose that

{
a
n

converges, tends to +

, or tends to

. Let

A = lim
n

a
n

and let

a
n
k

}
be any subsequence of

a
n

}
. By Theorem 7.2, we havelim
k

n
k

= A . Thus the only subsequential limit of

a
n

}
is A . By Theorem 8.10, wehavelim
n

a
n

= lim
n

a
n

A = lim
n

a
n

Remark.

This theorem means that1) If lim


n

a
n

= lim
n

a
n

, thenlim
n

a
n

= lim
n

a
n

= lim
n

a
n

2) If lim
n

a
n

exists, +

or


, thenlim
n

a
n

= lim
n

a
n

= lim
n

a
n

.
Example

8.14
.

Let

{
a
n

be a sequence. Show that lim

|
a
n

=0 if and only if lim


n

|
a
n

|
=0 , if and only if lim
n

a
n

=0
Proof.

Suppose that lim

|
a
n

|
= 0. From

|
a
n

0, we have0 = lim
n

0 = lim0

lim

|
a
n

|
lim

a
n

|
=0 . Thuslim

a
n

|
= lim

|
a
n

|
= 0and so lim
n

|
a
n

|
exists andlim
n

|
a
n

|
= lim

|
a
n

= lim

|
a
n

|
= 0Conversely suppose that lim

|
a
n

|
= 0. Thenlim

|
a
n

|
= lim

a
n

|
= lim
n

a
n

|
= 0because

{
a
n

}
is convergent. Hence lim

|
a
n

|
= 0 if and only if lim
n

|
a

|
= 0.Now suppose that lim
n

|
a
n

|
= 0. From

a
n

|
a
n

|
a

|
, we have lim
n

a
n

= 0 by the Squeeze theorem.Conver sely suppose that lim


n

a
n

= 0. Thenlim
n

|
a
n

|
lim
n

a
n

|
=

|
0

= 0because the function f ( x )=

|
is continuous. Hencelim
n

a
n

|
=0

lim
n

a
n

=0 .
24 1. SEQUENCES OF REAL NUMBERS

Example

8.15
.

Let a
n

> 0 for all

n . Prove that lim


n

a
n


lim a
n +1

a
n

Proof.

Let B = lim a
n +1

a
n

. If B =+

, clearly lim
n

a
n

B =+

. So we mayassume that B< +

. Since a

> 0, we have a
n +1

a
n

>

0 and so B

0. Thus B is anite nonnegative

number. By Proposition 8.9, given any > 0, there exists N suchthat a

n +1

a
n

<B + for n>N

. Fixed any n with n>N , we have0 <a


n +1

a
n

,a
n +2

a
n +1

,a
n

+3

a
n +2

<B +

and so, for any k

1, we have0 <a
n +1

a
n

a
n +2

a
n +1


a
n +3

a
n +2

a
n + k

a
n + k 1

= a
n + k

a
n

( B + )
k


a
n + k

a
n

( B + )
k


a
1

n
+

n + k


a
1

n
+

B + )
kn
+

for any k


1 . Let k tends to


. We havelim
k

a
1

n
+


( B + )
kn
+

= lim
k

a
1

n
+

lim
k

( B + )
1

n/k
+1

= a
0 n

( B

+ )= B + . Thuslim
m

a
1

= lim
k

a
1

n
+

n + k

lim
k

a
1

n
+

( B +

)
kn
+

= lim
k

a
1

n
+

( B + )
kn

= B + . In other words,lim
n

a
n

+ for any > 0 and solim


n
n


a
n

= lim
0

lim
n
n

a
n

lim

( B + )=

B, that is,lim
n
n

a
n


lim
n

a
n +1

a
n

.
Exercise

8.1
.

Let a
n

>

0 for all n . Prove that lim a


n +1

a
n

lim
n

. By Example 8.15 and Exercise 8.1, we havelim a


n +1

a
n

lim
n

lim
n

a
n


lim a
n +1

a
n

Exercise

8.2
.

Let a
n

> 0 for all

n . Suppose that the limit lim


n

a
n +1

a
n

exists or +

. Prove that lim


n

a
n

exists and lim


n

a
n

= lim
n

a
n

+1

a
n

.
9. CAUCHY SEQUENCES AND THE COMPLETENESS OF
R

25

For instance,lim
n
n

n ! = lim
n

( n + 1)! n != lim
n

n + 1) = +

. 9. Cauchy Sequences and the

Completeness of
R

9.1. Cauchy Sequences.


Definition

9.1
.

{
a
n

}
is called a Cauchy sequence

if given any > 0, there existsa natural number N such that for all m ,

n>N , we have

|
a
n

|
< . Remark. Roughly speaking, a sequence is

Cauchy if the width of its tail

0 as n

.
Proposition

9.2
.

Every Cauchy sequence is bounded.


Proof.

Let

{
a
n

}
be a Cauchy sequence. Choose

= 1. There exists N such that

|
a

a
m

|
< 1 for

n,m > N . In particular,

|
a
n

N +1

|
< 1 or a
N +1


1 <a
n

<a
N +1

+ 1for

n>N . Let M = max

{
a
1

,a
2

,a
N

,a
N

+1

+1

}
m = min

{
a

,a
2

,a
N

,a

N +1

}
. For

N , we have m

a
n

M , and, for n>N , we have

a
N +1

<a
n

<a
N +1

+1

M.

Thus, for all n , we have m

a
n


M and so

{
a
n

}
is bounded.

9.2. Completeness of
R

. The following Criterion was formulated by Augustin-Louis Cauchy (17891857).


Theorem

9.3 (Cauchys criterion)


.

A sequence is a convergent sequence if and only if it is a

Cauchy sequence.
Proof.

, i.e., every convergent

sequence is Cauchy.Given that

{
a
n

is convergent, say lim


n

a
n

= A

. Then for any given > 0, thereexists N such that

a
n

|
<

2for all n > N. Now for any m,n > N ,

|
a

a
m

|
=

|
( a
n

A )


( a
m

A )

||
a
n

|
a
m

|
< 2+ 2= since both

m,n > N . Therefore,

{
a
n

is a Cauchy sequence.

=, i.e., every Cauchy sequence is

convergent.Give n that

{
a
n

is Cauchy. By Proposition 9.2,

{
a
n

is bounded. By the BolzanoWeierstrass Theorem (Corollary 8.12), there exists a

convergent subsequence

{
a
n
k

of

{
a
n

}
. Let A

= lim
k

a
n
k

. Given any > 0, since

{
a
n

}
is Cauchy, there exists N

such that

|
a
n

|
< 2for all n,m > N
1

26 1. SEQUENCES OF REAL NUMBERS

Since

{
a
n
k

}
converges to A , there exists K such that

|
a
n
k

|
< 2for all k > K. Let N = max

{
K,N
1

}
. Choose an n
k

such that k>N , for instance, choose n


k

to be n

N +1

. When n>N , by triangular inequality,

|
a

|
=

( a
n

a
n
k

+(

a
n
k

A )

||

a
n

a
n
k

|
a
n
k

|
< 2+ 2= because

n>N

N
1

, n
k


k>N

N
1

and k>N


K . Therefore

{
a
n

}
convergesto A by the denition.

Remark.

The statement that every Cauchy sequence in


R

converges

is often expressed by saying that


R

is complete. Note that our proof of the Cauchy

Criterionused the BolzanoWeierstrass Theorem and the proof of that one used the complete-ness axiom of

(Theorem 5.14). We did not prove Theorem 5.14, namely we treatedTheorem 5.14 as an axiom.

(Dedekind and Cantor proved the completeness axiomin 1872 independently.) Conversely, we can treat the

Cauchy Criterion as an axiom,and prove Theorem 5.14. In this sense, both Theorem 5.14 and the Cauchy

Criterionare regarded as the completeness of


R

.Cantor constructed real numbers from rational

numbers by using Cauchy se-quences. His ideas are as follows. Consider all of the Cauchy sequences

{
a
n

}
with a
n

. By the Cauchy criterion,

{
a
n

}
converges to a real number. Then he provedthat all real numbers can be obtained

as the limits of all of the Cauchy sequences

{
a
n

}
with a
n

. One can think that, by using our standard base 10 number sys-tem, any real number admits a (decimal)

expansion. For instance,

2=1 . 4142


,we can dene a sequence of rational numbers a
1

=1 ,a
2

=1 . 4 ,a
3

=1 . 41 ,a
4

=1 . 414

,a
5

=1 . 4142 ,

with lim
n

a
n

.
Example

9.4
.

Let s
n

= 1 +12
2

+13
2

+1 n
2

. Show that

{
s
n

}
is convergent.
Proof.

For each

1, we have

|
s
n +

s
n

|
=

1 +12
2

+1 n
2

+1( n + 1)
2


+1( n + k )
2

1 +12
2

+1 n
2

=1( n + 1)
2

+1(

n + 2)
2

+1( n

+ k )
2

1 n

( n + 1)+1( n + 1)( n + 2)+1( n

+ 2)( n + 3)+

+1( n +

1)( n + k )=

1 n

n +1

1 n +1

1 n +2

1 n

+ k

n + k

=1

1 n + k< 1

n. Given any > 0, choose N such that


1 N

< , that is, N> 1 . When m > n > N

,from the above,

|
s
m

s
n

|
< 1 n< 1 N<

9. CAUCHY SEQUENCES AND THE COMPLETENESS OF


R

27

Thus

s
n

}
is a Cauchy sequence and hence

s
n

}
converges by the Cauchy Criterion.

9.3. Contractive Sequences. In this subsection, we give an application of theCauchy

Criterion.A sequence

{
a
n

}
is called

contractive if there exists b ,0 <b< 1, such that(1)

a
n +1

a
n

|
a
n

a
n

|
for all n

2.

Theorem

9.5 (Contractive Theorem)


.

Every contractive sequence

converges. Furthermore, if

{
a
n

is contractive and A = lim


n

a
n

, then

(a).

|
A

a
n

|
b
n 1

|
a
2

a
1

|
, and (b).

|
A

a
n

|
b 1

|
a
n

a
n 1

|
,where b is the constant in above denition.
Proof.

From Inequality (1), we have

|
a
n +1

a
n

|
b

|
a
n


a
n 1

|
b
2

|
a
n 1

a
n

|
b
3

|
a
n

a
n 3

b
n 1

|
a
2

a
1

|
For each k

1, we have

|
a
n + k

a
n

|
=

|
( a
n +

a
n + k 1

)+(

a
n + k 1

a
n

+ k 2

)+

+( a
n

+1

a
n

||
a

n + k

a
n + k 1

|
+

|
a
n + k

a
n + k 2

|
+

|
a

n +1

a
n

|
b
n

+ k 1

|
a
2

a
1

|
+ b
n + k

|
a
2

a
1

|
+

+ b
n

|
a
2

a
1

|
=( b
n + k 1

b
n + k 2

b
n 1

|
a
2


a
1

|
= b
n

(1 + b + b
2


+ b
k

|
a

a
1

|
= b

n 1

(1

b
k

+1

)1

|
a
2


a
1

|
<b
n 1

|
a
2


a
1

|
. Since 0 <b<

1, we have lim
n

b
n 1

= 0 by the Standard limits and solim


n

b
n 1

|
a
2


a
1

|
=0 .

Therefore, given any > 0, there exists N such that b


n

|
a

a
1

|
< for

n>N and so, for all n>N and k

1,

|
a
n + k

a
n

|
<b
n 1

|
a
2

a
1

|
< . It follows that

{
a
n

}
is Cauchy and so it is convergent.Let A = lim
n

a
n

. From the inequality

|
a
n +

a
n

|
<b
n

|
a
2


a
1

|
, we have

|
A

a
n

lim
k

a
n + k

= lim
k

|
a
n + k


a
n

|
b
n 1

|
a
2


a
1

|
.
28 1. SEQUENCES

OF REAL NUMBERS

This is Inequality (a).For each k

1, we have

|
a
n + k

a
n

|
=

|
( a
n +

a
n + k 1

)+(

a
n + k 1

a
n

+ k 2

)+

+( a
n

+1

a
n

||
a

n + k

a
n + k 1

|
+

|
a
n + k

a
n + k 2

|
+

|
a

n +1

a
n

|
=

|
a
n +1

a
n

|
+

|
a
n +2


a
n +1

|
+

|
a
n + k


a
n + k 1

|
a
n

a
n

|
+ b
2

a
n

a
n 1

+ b
k

|
a

a
n 1

|
=(

b + b
2

b
k

|
a
n

a
n 1

|
= b (1 +

b + b
2

b
k 1

|
a
n


a
n 1

|
=

(1

b
k

)1

|
a
n

a
n 1

|
<b 1

|
a
n

a
n 1

|
. Thus

|
A

a
n

lim
k

a
n + k

= lim
k

|
a
n + k


a
n

|
b 1

|
a
n

a
n 1

|
. This is Inequality (b).

Example

9.6
.

Let c
1

(0 , 1) be arbitrary, and for n

1 set c
n +1

=13( c
2 n

+ 1) . 1) Prove that the sequence

{
c

}
is contractive. 2) Show that if c = lim
n

c
n

, then c is a root of the polynomial x


2


3 x +1 . 3) Let

c
1

=12 . Determine a value of n such that

|
c

c
n

< 10
3

.
Proof.

(1). First we show that 0 <c

< 1 for all n by induction. By theassumption 0 <c

< 1. Suppose that 0 <c


n

< 1. Then0

<c
n +1

=13( c
2 n

+ 1) <

13(1 + 1) =23 < 1 . The induction is nished and so 0 <c

< 1 for all n . Now from

|
c
n

+1

c
n

|
=

13(

c
2 n

+ 1)

13( c
2

n 1

+ 1)

=13

|
c
2 n

c
2 n

|
=13

|
c
n


c
n 1

||
c
n

+ c
n 1

|
13

|
c
n

c
n 1

|
(1 + 1) =23

|
c
n

c
n 1

|
, the sequence

{
c
n

}
is contractive with b

=
23

.(2). From c
n +1

=13( c
2

+ 1) ,c = lim
n

c
n +1

= lim
n

13( c
2 n

+ 1) =13

lim
n

c
2 n

+1

=13( c
2

+ 1)

. Thus c
2

3 c

+ 1 = 0, that is, c is a root of x


2

x + 1.(3). We use Inequality (a) of the Contractive Theorem.Note that c


2

=13( c
21

+ 1) =13

12

+1

=512 .b
n

|
c

c
1

|
=

23

23

512


12

<

10
3

9. CAUCHY SEQUENCES AND THE COMPLETENESS OF


R

29

= 3

23

n 1

112

< 110
3

23

n 1

< 410
3

( n

1)

ln

23

< ln4

3ln10=

1) > 3ln10

ln4ln3

ln2= 13 . 62 . Thus when

n = 15, we have

|
c

c
15

|
<b
n 1

|
c
2

c
1

|
< 10
3

Note.

From x
2

3 x

+ 1 = 0, we have x =3

3
2


42=3

52 . Since 0

1, c =3


52. CHAPTER 2

Series of Real Numbers

1. Series The expression a


1

+ a
2

a
3

written alternatively as

k =1

a
k

is called an

innite series .
Example

1.1
.

(1). 1 + 2 + 3 + 4+

.(2). 1 + 1 / 2+1 / 3+1 / 4+


(3). 1 + 1 / 2
2

+1 /

3
2

+1 / 4
2

.(4). 1 + 0 + 1 + 0+1+0+

.
Definition

1.2
.

Given a series

k =1

a
k

, its

n
th

partial sum S
n

is given by S
n

k =1

a
k

a
1

+ a
2

a
n

. The sequence

{
S
n

}
is called the sequence of partial sums of the series

k =1

a
k

.
Example

1.3
.

Consider the series 1

1+1

1+1

1+1

1+

. The S
2 n 1

= 1and S
2 n

= 0.
Definition

1.4
.

Consider the sequence of partial sums

S
n

}
of the series

k =1

a
k

.If this sequence converges to a number S , we say that the series

k =1

a
k

converges

to S and write

=1

a
k

= lim
n

S
n

S. If

{
S
n

diverges, then we say

k =1

a
n

diverges .
Example

1.5 (Geometric Series)


.

Let

a = 0. Consider the series

n =0

ar
n

a + ar + ar
2

+ ar

.
31 32 2. SERIES OF REAL NUMBERS

Then the partial sum S


n

= a + ar

+ ar
2

+ ar
n

= a (1 + r +

+ r
n 1

)=

a 1

rr =1

a ( n + 1) r = 1When

1 <r< 1, S
n

r as n

.When r> 1, S
n

diverges because r

as n


.When r = 1, S
n

a ( n + 1) diverges.When r =


1, S
n

= a [1

1)
n

]2diverges.Whe n

r<

1, S
n

diverges because

r
n

.Thus the geometric series

n =0

ar
n

converges if and only if

1 <r< 1 ,and,

n =0

ar
n

a 1

r for

1 <r< 1 .Remark. If

k =1

a
k

and

k =1

b
k

converges, then one always has(i)

=1

( a
k

+ b
k

)=

k =1

a
k

k =1

b
k

.(ii)

k =1

ca
k

k =1

a
k

.
Example

1.6
.

=1

14

15

n =1

14

n =1

15

14+

14

15+

15

=14

1 +14+

14

+15

1 +15+

15

=1411

14+1511


15=14

34+15

45=13+14=712

.
Theorem

1.7
.

If

k =1

a
k

converges, then

lim
k

a
k

=0 .

1 . S E R I S 3 3 Proof.

Recall that partial sum S


k

a
1

+ a
2

a
k

. We have S
k

S
k

=( a
1

+ a
2


+ a
k 1

+ a

( a
1

+ a

+ a
k 1

)= a
k

. Since the series

k =1

a
k

converges, the sequence

{
S
k

}
converges. Let

S = lim
k

S
k

.Thenlim
k

a
k

= lim
k

( S
k


S
k 1

) = lim
k

lim
k

S
k 1

= S

S =0 .

Corollary

1.8 (Divergence Test)


.

If lim
n

=0 (or does not exist), then

n =1

a
n

diverges.

Example

1.9
.

(1). The series

=1

1)
n

is divergent because the limit of the

n -thterm (

1)
n

does not exist.(2). The series

=1

n ! n
2

is divergent becauselim
n

n ! n
2

=1lim
n

n
2

n !=10= +

=0 . (3). The series

n =1

2 n + 13

n + 2is divergent becauselim


n

2 n + 13

n + 2= lim
n

2+1 /n 3+2 /n

=23 =0 . Remark. The divergence test is a one-

way test, i.e., lim


n

a
n

= 0 does NOT imply

n =1

a
n

converges.
Theorem

1.10 (Cauchy Criterion)


.

The series

k =1

a
k

converges if and only if given any > 0

, there exists N such that

= n +1

a
k

< for all m>n>N .

Proof.

The series

k =1

converges if and only if the sequence of its partial sums

{
S
n

}
converges, (by denition), if and only if

{
S
n

}
is Cauchy. The result follows from

|
S
m


S
n

|
=

( a
1

+ a
2

+ a
n

+ a
n +1


+ a
m

a
1

+ a
2

a
n

|
34 2. SERIES OF REAL NUMBERS

|
a
n

+ a
n +1

+ a
m

|
=

= n +1

a
k

.
Example

1.11 (Harmonic Series)

Show that the series

n =1

1 n diverges.
Proof.

Note that

|
S
2

S
n

|
=

1 +12+

+1 n

+1 n + 1+

+12 n

1 +12+

+1 n

=1 n + 1+1 n

+ 2+

+12 n

12

n +12 n +

+12 n

= n

12 n =12 .

Suppose that the series

n =1

1 n converges. Then

{
S
n

is Cauchy. Given =
12

. Thereexists N such that

|
S
m

S
n

|
< 12for all m>n>N . This contradicts to

the abovefact that

|
S
2 n

S
n

|
12, where m is chosen to be 2

n .

Note. The divergence of the harmonic series appears to have been

established byNicole Oresme (1323?1382) by showing that the sequence

=1

1 k

is NOT bounded.

Theorem

1.12
.

Suppose that eventually a


k

0 . Then

k =1

a
k

converges if and only if

{
S
n

}
is bounded above.
Proof.

We may assume that a


k


0 for all k . Since S
n +1


S
n

= a
n +1


0 , the sequence

{
S
n

}
is monotone increasing. Thus

k =1

a
k

converges if and only if

{
S
n

converges, if and only if

{
S
n

is bounded above (by the Monotone ConvergenceTh eorem).

Note. Suppose that

eventually a
k

0. This theorem means that the following.1) If

{
S
n

}
is bounded above, then

k =1

a
k

converges.2) If

{
S
n

}
is NOT bounded above, then

k =1

a
k

diverges.

2. Tests for Positive Series A series

=1

a
k

is called a ( eventually ) positive series if every term

a
k

is (even-tually) positive.
2. TESTS FOR POSITIVE SERIES 35

2.1. Comparison Test.


Theorem

2.1 (Comparison Test)


.

Consider 2 positive series

k =1

a
k

and

k =1

b
k

.Suppose that eventually 0

a
k


b
k

. (i) If

k =1

b
k

converges, then

k =1

a
k

converges.

(ii) If

k =1

a
k

diverges, then

k =1

b
k

diverges.
Proof.

Assertion (ii) follows immediately from (i). Let

A
n

=
n

k =1

a
k

, and B
n

=
n

k =1

b
k

. Then

A
n

B
n

for all n . Suppose that

k =1

b
k

converges, that is,

k =1

b
k

is a(nite) number. Then A


n

k =1

b
k

for all n and so A


n

is bounded above. By Theorem 1.12,

=1

a
k

is convergent.
Example

2.2
.

The series

k =1

2 k

13

k +2

converges because

2 k

13

k +2

23

and the geometric series

k =1

23

converges. Remark. 1. Suppose 0

a
n


b
n

and

n =1

b
n

diverges. Then NO conclusion can bedrawn.2. Similarly, suppose 0

b
n

and

n =1

a
n

converges. Then NO conclusion can bedrawn.


Example

2.3
.

12

2
n

3
n

. The series

n =1

3
n

diverges. Now

n =1

12

converges and

n =1

diverges.
36 2. SERIES OF REAL NUMBERS Corollary

2.4 (Limit Comparison Test)


.

Suppose that

k =1

a
k

and

k =1

b
k

are (eventually) positive series. (a). If lim


k

a
k

b
k

= L with 0 <L<


, then

k =1

a
k

converges if and only if

k =1

b
k

converges. (So

k =1

a
k

diverges if and only if

k =1

b
k

diverges.) (b). If lim


k

b
k

=0 , (that is, a
k

<< b
k

), and

k =1

b
k

converges , then

k =1

a
k

converges . (c). If lim


k

a
k

b
k

=0 , (that is, a
k

<< b

), and

k =1

diverges , then

k =1

b
k

di-verges . Remark. If lim


n

b
n

=+

, interchange a
n

and b
n

, and then apply assertions ( b )and ( c

).
Proof.

We may assume that

k =1

a
k

and

k =1

b
k

are positive series.(a). Sincelim


k

a
k

b
k

= L (

=0 ,


) ,

b
k

is bounded above, say, by

M . Thus0

a
k

Mb

for all k . Similarly, since lim


k

b
k

a
k

=1 L ( =0

, =

) ,

b
k

a
k

is bounded above, say, by M . Thus 0


b
k

M a

for all k .If

k =1

a
k

is convergent, then

k =1

M a
k

= M

k =1

a
k

is also convergent. By thecomparison test, it follows that

k =1

b
k

is also convergent because b


k

a
k

.If

k =1

b
k

is convergent, then

k =1

Mb
k

= M

k =1

b
k

is also convergent. By thecomparison test, it follows that

k =1

a
k

is also convergent because a


k

Mb
k

. Hence

k =1

a
k

is convergent if and only if

k =1

b
k

is convergent. Hence

k =1

a
k

is divergent if and only if

k =1

b
k

is divergent.

2. TESTS FOR POSITIVE SERIES 37

Now we prove assertions (b) and (c). lim


k

a
k

b
k

= 0 implies for every > 0, thereis an

N such that

b
k

<


k > N. We choose = 1. Then the above inequality is

a
k

<b
k

k > N. We get the result by

applying the comparison test.

Standard series used in comparison and limit

comparison tests. 1. The Geometric Series:

n =1

ar
n 1

converges if

|
r

|
< 1

, diverges if

|
r

|
1

. 2. The p-series: for a xed p ,

n =1

1 n
p

converges if p> 1 , diverges if p

1 . To be proved in the subsection on Integral Test.


Example

2.5
.

Determine the convergence or divergence: 1)

n =1

1 + cos nn
2

2)

n =1

ln n +

n
3

+8 n
4

2 n

+3
Solution.

(1). It is convergent, by the comparison test, because0

1 + cos nn
2

2 n
2

and the

p -series

n =1

1 n
2

is convergent.(2). It is divergent, by the limit

comparison test, becauselim


n

ln n + n
3

+8 n
4

2 n + 31

n = lim
n

n ln n +

n
4

+8 nn
4

2 n

+ 3= lim
n

ln nn
3

+ 1 +1 n
3

2 n
3

+3 n
4

=0 + 1 + 01

0 + 0= 1and the harmonic series

n =1

1 n diverges.

Example

2.6
.

Determine convergence or divergence of

n =2

1(ln n )

, where k is a constant.
38 2. SERIES OF REAL NUMBERS Solution.

Let b
n

=1(ln n )
k

and let

a
n

=1 n . Thenlim
n

a
n

b
n

= lim
n

1 n 1(ln n

)
k

= lim
n

(ln n )
k

n =0 . Since the harmonic series

n =2

1 n

is divergent, the series

n =2

1(ln n )
k

is divergent forany k .

2.2. Integral Test. Let f ( x

) be a realvalued function on [ a, +

) such that

f ( x ) is Riemann integrable, that is, the integral

ba

f ( x )

dx exists for every b>a .The following theorem is useful.


Theorem

2.7

Let f ( x ) be a real-valued function on

[ a,b ] . (a). If f (

x ) is continuous on [ a,b ] , then f

( x ) is Riemann integrable on [ a,b ]

. (b). If f ( x ) is monotone on

[ a,b ] , then f ( x )

is Riemann integrable on [ a,b ] .

The improper integral is dened by

f ( x ) dx = lim
b

ba

f ( x )

dx = area under f ( x ) over [ a,


) . Here we say that

f ( x

) dx converges if the limit lim


b

ba

f ( x )

dx exists (nite),i.e., the area under f ( x ) over [

a,

) is nite. We also say that

f ( x ) dx diverges

if the limit lim


b

ba

( x ) dx does not exist.


Theorem

2.8 (Integral Test)

Let f ( x ) be an ( eventually

) positive monotonedecrea sing function on [1 , +


) . Suppose we have a series

k =1

a
k

such that

a
k

= f ( k ) , then the series

k =1

a
k

and the integral

f ( x

) dx either both converge or both diverge .


Proof.

We may assume that f ( x ) is positive monotone decreasing on [1

, +

).Let a
n

f ( n ) for all n .
2. TESTS FOR

POSITIVE SERIES 39

From the graph, we see thatarea of the rectangles

area under f

( x ) over [1 ,n ] , i.e.,
n

k =2

f ( k

n 1

f ( x ) dx

f ( x

) dx. Thus, if

f ( x ) dx <

, then

k =2

a
k

k =2

f (

k )

f ( x ) dx <

, i.e., for all n ,


n

k =2

a
k

is bounded above by the nite number

f ( x

) dx . Since wealso have a


k

0, it follows from Theorem 1.12 that

=2

a
k

converges, and thus

k =1

a
k

also converges.Next we consider the following graph:From the graph, it is easy to see thatarea

under the rectangles

area under f ( x

) over [1 ,n ] , i.e.,
n 1

k =1

f ( k

n 1

f ( x ) dx. Thus, if

k =1

a
k

<


, then

>

k =1

a
k

n 1

k =1

=
n 1

=1

f ( k )

n 1

f ( x

) dx, i.e.,

n 1

f ( x ) dx is bounded above by the nite number

k =1

a
k

<


. Letting n

,it follows that we have

f ( x

) dx

k =1

a
k

<

40 2. SERIES OF REAL NUMBERS

In conclusion, we have

k =1

a
k

converges if and only if

( x ) dx converges, whichalso means that

k =1

a
k

diverges if and only if

( x ) dx diverges.
Example

2.9

Show that 1) the series

n =1

1 n
p

converges if and only if p> 1

. 2) the series

=2

1 n (ln n )
k

converges if and only if k> 1 . (1). If p


0, then1 n
p

does not tend to 0 and so, by divergence test,

n =1

1 n
p

diverges.Assum e that p> 0. Let f ( x ) =1

x
p

on [1 , +

). Then

f ( x ) is positive monotonedecre asing. Now

f ( x

) dx =

x
p

dx =


p +1 x
p +1

= 1ln(+


ln1 p = 1Thus

f ( x ) dx

converges if and only if p> 1 and so

n =1

1 n
p

converges if and onlyif p> 1.(2). Let

f ( x ) =1 x (ln x )

on [2 , +

). Then f

( x ) is positive. We check that f ( x ) is

eventually monotone decreasing. From f ( x

)=

x
1

(ln

x )
k

x
2

(ln x )

kx
1

(ln x

)
k 1

1 x =


x
2

(ln x )

k 1

(ln x + k ) ,

we have f ( x )

0 when ln x>

k . Thus f (

x ) is monotone decreasing when ln x>

and so f ( x ) is eventually monotone decreasing. Now

f ( x

) dx =

1 x (ln x )
k

dx = = = = = = = =

y = ln xdy =1 xdx

ln2

1 y
k

dy =


k +1 y
k

ln2

k = 1ln(+


ln(ln2) k = 1Thus

f ( x ) dx

converges if and only if k> 1 and so the series

n =2

1 n (ln

n )
k

con-verges if and only if k> 1. 2.3. Ratio Test.

2. TESTS FOR POSITIVE SERIES 41 Theorem

2.10 (Ratio Test)


.

Consider the positive series

n =1

a
n

. Suppose ( 2 ) l i m
n

a
n

+1

a
n

= . (i) If 0


< 1 , then

n =1

a
n

converges.

(ii) If 1 <

, then

n =1

a
n

diverges.

(iii) If =1 , then the test is inconclusive.


Proof.

We will prove (i) and (ii). Given any > 0, it follows from (1) that thereexists N

such that for all n>N ,

n +1

a
n

<

or

<a
n +1

a
n

< + . By repeating using the above inequalities, it

follows that for all m> 0,(3) a


N +1


)
m

<a
N +1+

<a
N +1

( +

)
m

. (i). If < 1, choose > 0 such that

+ < 1, then

m =1

a
N +1

+ )
m

converges(since it is a geometric series with

common ratio satisfying

|
r

|
=

+ < 1). Togetherwith the right-handside of (3), it

follows from the comparison test that

=1

a
N +1+ m

converges, and thus

n =1

a
n

converges.(ii). If > 1, choose > 0 such that


> 1, then by the left-hand-side of (3), wehave, for all m>

0, a
N +1+ m

a
N +1

)
m

>a

N +1

> 0 . In particular, lim


n

a
n

= 0 or does not exist. By the divergence test,

n =1

a
n

diverges.

Example

2.11
.

Determine convergence or divergence. 1)

n =1

n ! n

. 2)

n =1

( n !)
2

(2 n )! .

42 2. SERIES OF REAL NUMBERS

(1).lim
n

a
n +1

a
n

= lim
n

( n + 1)!( n

+ 1)
n +1

n ! n
n

= lim
n

( n + 1)! n !

( n + 1)
n

n
n

n + 1)= lim
n

1 +1 n

=1 e<

1 . Thus the series converges.(2).li m


n

a
n

+1

a
n

= lim
n

[( n + 1)!]

(2 n + 2)!( n !)
2

(2

n )!= lim
n

( n + 1)
2

(2

n + 2)(2 n + 1)= lim


n

( n

+ 1)
2

/n
2

(2 n + 2)(2 n

+ 1) /n
2

= lim
n

1+2 /n +1

/n
2

(2 + 2 /n )(2 + 1 /n )=1 + 0 + 0(2 + 0)


(2 + 0)=14 < 1 . Thus the series converges.

2.4. Root Test.


Theorem

2.12
.

Consider the series

n =1

a
n

with each

a
n

0 , and let (4)

= lim
n
n

a
n

(i) If 0

< 1 , then

n =1

a
n

converges.

(ii) If 1 <

, then

n =1

a
n

diverges.

(iii) If =1 , then the test is inconclusive.


Proof.

We will prove (i) and (ii).(i) Suppose that < 1. Then for all given >

0, it follows from (4) and Proposi-tion 8.9 of chapter 1, that there exists an N such that

a
n

< + for all

n>N .Now choose > 0 s.t. + <

1. Then( )

5 0

a
n

<

( + )
n

for all n > N.

Since

n =1

+ )
n

converges (as it is a geometric series with

common ratio satisfying

|
r

|
=

+ < 1), it follows from (5) and the comparison test that

n =1

a
n

converges.(ii). We are going to prove (ii) by contradiction. Given that > 1. Suppose that

n =1

a
n

converges. Then by Theorem 1.7, we have lim


n

a
n

= 0. In particular, there
3. THE DIRICHLET TEST AND ALTERNATING SERIES 43

exists N

such that 0

a
n

< 1 for all n>N

. Hence
n

a
n

< 1 for all n>N

, and itfollows that we must have

1, which is a contradiction. Hence

=1

a
n

diverges.
Corollary

2.13 (Simplied Root Test)


.

Consider the series

n =1

a
n

with each a
n

0 . Suppose that

lim
n
n

a
n

. (i) If 0

<

1 , then

n =1

a
n

converges. (ii) If 1 <


, then

n =1

a
n

diverges. (iii) If =1

, then the test is inconclusive.


Proof.

We will prove (i) and (ii). Recall from Theorem 8.13

of chapter 1 thatif lim


n
n

a
n

exists, then lim


n
n

a
n

= lim
n

a
n

. Then the Corollary follows

fromTheorem 2.12.
Example

2.14
.

Determine convergence or

divergence of the series

n =1

2
n

1 n

n
2

lim
n
n

a
n

= lim
n

2
n

1 n

n
2

1 n = lim
n

1 n

=2 e< 1 .

Thus the series converges.


Example

2.15
.

Determine convergence or

divergence of the series

n =1

(3 + sin n )
n


2 n

n
2

. lim
n
n

a
n

= lim
n

(3 + sin n

)
n

2 n

n
2

1 n = lim
n

(3 + sin n )


2 n


lim
n

2 n

=4 e
2

<

1 . Thus the series converges. 3. The Dirichlet Test and Alternating

Series3.1. The Dirichlet Test. The following theorem is due to Neils Abel (1802-1829).
44 2. SERIES OF

REAL NUMBERS Theorem

3.1 (Abel Partial Summation Formula)


.

Let

{
a
k

}
and

b
k

}
be se-quences, and let

{
A

}
k 0

be a sequence such that A


k


A
k 1

= a
k

for each k

1 . Then if 1

q ,
q

k = p

a
k

= A
q

b
q

p 1

b
p

+
q 1

k = p

A
k

b
k

b
k +1

) .

Proof.

Since a
k

= A
k

A
k 1

, a
p

b
p

+ a
p +1

b
p +1


+ a
q

b
q

=( A

A
p 1

) b
p

+( A
p +1

A
p

b
p +1

+( A
p +2

A
p +1

) b
p +2


+( A
q

A
q

) b
q

p 1

b
p

+ A
p

b
p

b
p +1

)+ A

p +1

( b
p +1

b
p

+2

)+

+ A
q 1

( b
q 1

b
q

)+ A
q

b
q

p 1

b
p

+ A
q

b
q

+
q 1

k =

A
k

( b
k

k +1

) .

Remark. There are two canonical choices of

{
A
n

}
(1). A
0

= 0, A
n

=
n

k =1

a
k

. Then A
n

A
n 1

= a
n

for all n

.(2). Suppose that

k =1

a
k

converges. Then we can also choose

{
A
n

}
by letting A
n

=
n

k =1

a
k

k =1

a
k

k = n

+1

a
k

, n

1, A

=1

a
k

. In this case, wealso have A


n

A
n 1

k = n

+1

a
k

k = n

a
k

= a
n

An application is to give the following theorem of Peter Lejeune Dirichlet (18051859).


Theorem

3.2 (Dirichlet Test)


.

Suppose that

{
a
k

}
and

{
b
k

are sequences of real numbers satisfying the following: (i). the sequence of the partial sums A

=
n

k =1

is bounded, (ii). b
1

b
2


b
3

0 , and (iii). lim

b
k

=0 .Then the series

k =1

a
k

b
k

converges.
3. THE DIRICHLET TEST AND ALTERNATING SERIES 45 Proof.

Since

A
n

}
is bounded, there exists a positive number M>

0 such that

|
A
n

|
<M for all

n . Also, since lim


n

b
n

= 0, given >

0, there exists N such that b


n

b
n

|
<

2 M for all n>N . Now, for m>n>N

, by Abel partial summationform ula,

= n +1

a
k

b
k

A
m

b
m


A
n

b
n +1

+
m

k = n +1

( b
k

b
k +1

|
A
m

||
b
m

|
+

A
n

||
b
n +1

+
m 1

k =

n +1

|
A
k

||
b
k


b
k +1

|
M


b
m

+ M

n +1

+
m 1

k = n +1

( b
k


b
k +1

) (because b
k


b
k +1

0 ,

|
A
k

|
M )= M


[ b
m

+ b
n +1

+( b
n +1

b
n +2

)+( b
n +2

b
n +3

)+

+( b
m 2


b
m 1

)+( b
m

b
m

)]= 2 Mb
n +1

< 2 M

2 M

= . Hence by the Cauchy Criterion, the series

k =1

a
k

b
k

converges.

3.2. Alternating Series Test. An alternating series is of the form

n =1

1)
n +1

a
n

= a
1


a
2

+ a
3

a
4

, or

n =1

1)
n

a
n

a
1

+ a
2

a
3

+ a

with each a
n

> 0.
Example

3.3
.

1+1

1+1


1+1

1+


12+13

14+

1+2


3+4

5+6

Theorem

3.4 (The Alternating Series test)


.

If

{
b
n

}
is a sequence satisfying (i) b
1

b
2

b
3

, and (ii) lim


n

b
n

=0 ,then

n =1

1)
n +1

b
n

( and

n =1

1)
n

b
n

) converge .
Proof.

Let

a
k

=(

1)
k +1

and let

A
n

=
n

k =1

a
k

. Then

|
A
n

1+1


1+

+(

1)

+(

1)
n +1

0 when n even1 when n oddThus

|
A

|
1 for all n , and the Dirichlet test applies.

46 2. SERIES OF REAL NUMBERS Example

3.5
.

Show that convergence or

divergence of the series

n =1

1)
n

1 n
p

convergence

p> 0 divergence p

0 .

Proof.

If p

0, then the n -th term (


1)
n

1 n
p

does not tend to 0. Thus the

seriesdiverges in this case by the divergence test.Assume that p> 0. Let a


n

=1 n
p

. Then a
n

>

0, monotone decreasing andlim


n

a
n

= 0. Thus the alternating

series converges in this case.In conclusion, we have that the series

n =1

1)
n

1 n
p

converges when p>

0 anddiverges when p

0.

Now we are going to give a theorem providing an estimate on the sum of (certain)series. For instance, let

S =

n =1

1)
n +1

1 n
2

. By taking the partial sum, we have S

S
1

=1 ,S

S
2

=1

12
2

=0 . 75 ,S

S
3

=1

12
2

+13
2


0 . 861 ,.... By using computer

program, we are able to compute much more, say S


1000000

. Amathematical problem is then

what is the error for estimating S by using the partialsum S


n

. In other words, how to estimate the remainder R


n

S
n

|
=

|
a
n +1

+ a
n +2

|
.
Theorem

3.6 (Alternating Series Estimation)


.

Let

{
b
n

}
be a sequence satisfying

(i) b
1

b
2

b
3

0 , and (ii) lim


n

b
n

=0 .Let S
n

=
n

k =1

1)
k +1

b
k

and S =

k =1

1)
k +1

b
k

Then the remainder R


n

|
S

S
n

|
b
n +1

for all n .
Proof.

R
n


1)
n +2

b
n +1

+(


1)
n +3

b
n +2

b
n +1

b
n +2

b
n +3

b
n +4

|
. Since b
n +1

b
n +2

+ b
n +3

b
n +4

= b
n

+1

( b
n +2

b
n +3

( b
n

+4

b
n +5

( b
n +6

b
n +7

b
n +1

and b
n

+1

b
n +2

+ b
n +3


b
n +4

=(

b
n +1

b
n +2

)+(

b
n +3

b
n +4

)+


0 , we have R
n

a
n +1

.
3. THE DIRICHLET TEST AND ALTERNATING SERIES 47

Example

3.7
.

Estimate

=1

1)
n +1

1 n

with error within 0 . 001 .


Solution.

From1(

n + 1)
4

10
3

, we have

n +1

6 or n

5. Thus

n =1


1)
n +1

1 n
4

12
4

+13
4


14
4

+15
4

with error within 0 .

001.

3.3. Trigonometric Series. Another application of Dirichlet test is

to conver-gence of trigonometric series. These series will be studied in much detail in the courseon Fourier series

with a lot of applications in physics and other sciences. (JosephFourier, 1768-1830.) We require the

following two identities.


Lemma

3.8
.

For t

=2 p , p

,
n

k =1

sin kt =cos12 t

cos

n +12

t 2sin12 t
n

k =1

cos kt =sin

n +12

sin12 t

2sin12 t
Proof.

We prove the second identity. A proof of the rst identity can

be foundfrom text book [ 2 , pp.297].Using the trigonometric identitysin x

cos y =12(sin( x + y ) + sin( x


y )) , we obtainsin12 t
n

k =1

cos kt =

k =1

sin12 t

cos kt =12
n

=1

sin

t 2+ kt

+ sin

t 2

kt

=12
n

k =1

sin

k +12

sin

12

=12

sin

1 +12

sin


12

sin

2 +12

sin

12

sin

3 +12

sin


12

sin

n +12

sin

12

=12

sin

n +12

sin12 t

.
Theorem

3.9 (Trigonometric Series Test)


.

Let

{
b
n

}
be a sequence satisfying
48 2. SERIES OF REAL NUMBERS

(i)

b
1

b
2

0 , and (ii) lim


n

=0 .Then (a). The series

k =1

b
k

sin

kt converges for all t

, and

(b). The series

k =1

b
k

cos kt converges for all t

, except perhaps t =2 p ,

.
Proof.

(a). Let a
k

= sin kt and let A


n

=
n

k =1

a
k

. If

t =2 p , then a
k

=sin2 kp

= 0 and so A
n

= 0 for t =2 p with

. If t

=2 p , then

|
A
n

=1

sin kt

cos12

cos

+12

t 2sin12 t

cos12

cos

n +12

sin12


1 + 12

sin12

=1

sin12 t

Thus

{|
A
n

|}
is bounded above, and the

Dirichlet test applies.(b). Let a


k

= cos kt and let A

=
n

k =1

. The Dirichlet test applies because

|
A
n

|
=

=1

cos kt

sin

n +12

sin12 t

2sin12 t

sin

n +12

sin12

sin12 t


1 + 12

sin12

=1

sin12 t

Example

3.10
.

Determine convergence or divergence of the series

k =1

cos ktk
q

, t

, q> 0

.
Solution.

When t =2 p with p

, the series

k =1

cos ktk
q

k =1

1 k
q

and so it converges for q> 1 and diverges for q

1.
4. ABSOLUTE AND CONDITIONAL CONVERGENCE 49

When t

=2 p , then the series converges by the trigonometric

series test because,for q> 0, the sequence1 k


q

is positive, monotone decreasing and lim


k

1 k
q

= 0. Inconclusion, the series

=1

cos ktk
q

is

convergent

q> 1 ,t

convergent 0 <q


1 ,t

,t

=2 p, for any p

divergent 0 <q

1 ,t =2

p for some p

4. Absolute and Conditional Convergence


Definition

4.1
.

A series

n =1

a
n

is called

absolutely convergent if

=1

|
a
n

|
converges.
Theorem

4.2
.

Every absolutely convergent series is convergent.


Proof.

Suppose that

n =1

a
n

converges absolutely, that is,

=1

|
a
n

|
converges bythe denition. Let

T
n

=
n

k =1

|
a
k

|
, S
n

=
n

k =1

a
k

. Since

{
T
n

}
converges,

{
T
n

}
is Cauchy.Thus, for any

> 0, there is a N such that

|
T
n


T
m

|
< for all

n,m > N . For any n,m > N , we may assume that m

n , say m = n + p

(as one of them should begreater than another). Then

|
S
n


S
m

|
=

S
n

( S
n

+ a

n +1

+ a
n +2

+ a
n + p

|
=

|
a
n +1

+ a
n +2

+ a
n + p

||
a
n +1

|
+

|
a
n +2

|
+

|
a
n + p

|
= T
m

T
n

|
T
n

T
m

|
< . Thus

{
S
n

}
is a Cauchy sequence and so it converges. Thus the series

n =1

a
n

convergesand hence the result.


Example

4.3
.

Determine convergence or

divergence of the series

n =2

sin n +12 n (ln n )


2

.
Solution.

Since

sin

n +12 n (ln n )
2

|
sin n

|
+12 n

(ln n )
2

1 +12 n

(ln n )
2

50 2. SERIES OF REAL NUMBERS

and

n =2

1 +12 n

(ln n )
2

=32

n =2

1 n (ln

n )
2

converges by Example 2.9, the series

n =2

sin

n +12 n (ln n )
2

converges. Thus the series

n =2

sin n +12 n (ln n )


2

converges.

Remark. If you are testing for absolute convergence, all the

techniques for thepositive series are applicable.Q: Is the converse of the Corollary true? I.e., if a series is

convergent, will it beabsolutely convergent?A: No, it is not necessarily true.


Example

4.4

The series

n =1


1)
n

1 n converges by Example 3.5,

but it is NOTabsolutely convergent by the p -series.


Definition

4.5

A series

n =1

is said to be conditionally convergent if

n =1

a
n

converges but

n =1

|
a
n

|
diverges.
Example

4.6
.

The series

n =1

1)
n

n is conditionally convergent.
Remark

4.7
.

Every series is either absolutely convergent, conditionally conver-gent or divergent.

Example

4.8
.

The series

n =1

1)
n

1 n
p

absolutely convergence p> 1 conditionally convergence 0 <p<

1 divergence p

0 5. Remarks on the various tests

for convergence/div ergence of series 1. n -th term test for divergence:

- a test for divergence ONLY, and it works for series with positive and negativeterms, e.g.

n =1

1)
n

.2. Comparison test/Limit Comparison test:

- when applying these tests, one usually compares the given series with a geometricseries or a

p -series.generally works for series which look like the geometric series or the p

-series,
5. REMARKS ON THE VARIOUS TESTS FOR CONVERGENCE/DI VERGENCE OF SERIES 51

e.g.

n =1

2+(

1)
n

4
n

n =1

2
1

.- when an oscillating factor/term appears, e.g.

n =1

2+(

1)
n

3
n

, try the Comparisontest rather than the Limit

Comparison test.3. Integral test: e.g.

n =2

1 n (ln n )
2

.4.

Ratio test: - generally works for series which look like the geometric series, series with n

!, andcertain series dened recursively,e.g.

=1

n
2

3
n

n =1

(2 n )!4

n !,

n =1

a
n

, where

a
1

= 1, a
n

= (12+1 n )

a
n 1

, n =2 , 3

.5. (Simplied) Root test:

- generally works for series where a


n

involves a high power such as the

n -th power,e.g.

n =1

n 3
n

n =1

2
n


1 n

n
2

.6. Alternating Series test: - works for alternating series only,e.g.

n =2

1)
n

ln nn . Remark.

In general, Tests 2 - 5 works only for

=1

a
n

, where a
n

0.

CHAPTER 3

Sequences and Series of Functions


1. Pointwise Convergence1.1.

Sequences of Functions. Let I be a (nonempty) subset of


R

, e.g. (


1 , 1) , [0 ,

1],etc. For each n

N
n

, let F

: I

be a function. Then we say

{
F
n

}
forms a se-quence of functions

on I .
Example

1.1
.

1. F
n

( x )= x
n

, 0 <x<

1. Then

{
F
n

}
forms a sequence of

functions on (0,1).2.

1+ xn

forms a sequence of functions on (

).Write out some terms: F


1

( x )=1+ xF

( x )=

1+

x 2

F
3

( x )=

1+

x 3

If we x the x

, and let n

,lim
n

F
n

( x ) = lim
n

1+ xn

= e

. So for each x , we can dene F ( x

) = lim
n

F
n

( x ) .

Definition

1.2
.

A sequence

{
F
n

}
is said to converge pointwise to a function F on

I if lim
n

F
n

( x )=

F ( x ) for each x

I,

i.e., for each x

I and given any >

0, there exists an N (which depends on x and

) such that

|
F
n

( x )


F ( x )

<

n > N. The function F is called the limiting function

of

{
F
n

}
. Remark.

The limiting function is necessarily unique.


Example

1.3
.

The sequence

{
x
n

}
converges pointwise on I

= (0 , 1) because thelimit F ( x ) = lim

x
n

= 0 exists for any 0 <x< 1.The sequence

{
x
n

}
does NOT converge on [


1 , 1] because the limit lim
n

doesnot exist when x =

1 , 1].

1.2. Series of Functions. A series of functions on a set I is of the form

n =1

f
n

x )= f
1

( x )+ f

( x )+

, where each

f
n

is a function on I .
53 54 3. SEQUENCES AND SERIES

OF FUNCTIONS Example

1.4
.

Below are some examples.1.

n =1

x
n 1

=1+ x + x
2

+ x
3

2.

k =1

sin kxk + x =sin x

1+ x +sin2 x 2+ x +sin3 x

3+ x +

, 0

1 .

As in chapter 2, we may form the partial sums S


n

( x

)=
n

k =1

f
k

( x )= f
1

( x )+

f
2

( x )+

f
n

( x ) . Then

{
S
n

}
forms a sequence of functions on

I .
Definition

1.5
.

The series

n =1

f
n

is said to

converge pointwise (to a function S ) on I if

{
S
n

}
converges pointwise (to S

) on I , (i.e. lim
n

S
n

( x

)= S ( x ) for each x

I .)
Example

1.6
.

What is the pointwise limit of

n =1

x
n 1

, where x

, 1) ? Does

=1

x
n 1

converge pointwise on [


1 , 1)
Solution.

Consider the partial sum

S
n

( x )=
n

i =1

x
i 1

=1+ x +

+ x
n

=1

x
n

x for

1 <x< 1. Thus

n =1

x
n 1

= lim
n

S
n

( x ) = lim
n

x
n

x =11

x for x

1 ,

1).Since the series

n =1

x
n 1

diverges when x =

1, the series of functions

n =1

x
n 1

does NOT converge pointwise on [

1 , 1).

1.3. Some Questions on Pointwise Convergence.

Suppose a sequence of functions

{
F
n

converges pointwise to a function F on the interval [ a,b

]. Also suppose a series of functions

=1

f
n

( x ) converges pointwise to a function

S ( x ) on [ a,b ].Among the questions we want to consider

the following. Some of these questions wereincorrectly believed to be true by many mathematicians prior to

nineteenth century,includin g the famous Cauchy. Cauchy in his text Cours dAnalyse proved a theo-rem to the

eect that the limit of a convergent sequence of continuous functions wasagain continuous. As

we will see, this result is false!


1. POINTWISE CONVERGE NCE 55

Question (a). If each F

is continuous at p

[ a,b ], is

F necessarily continuous at p ? Recall that F is continuous at p

if and only if lim


t p

F ( t )=

F ( p ) . Since F (

x ) = lim
n

F
n

( x ) for every

[ a,b ], what we really asking is doeslim

t p

lim
n

F
n

( t )

= lim
n

lim
t

F
n

( t )

? Question (a). For series of functions, we

can ask similar question. If each f


n

( x )is continuous at

p , is S ( x )=

n =1

f
n

x ) necessarily continuous at p ? Again what wereally asking is doeslim


t

n =1

( t )=

n =1

lim
t p

f
n

( t

)? Question (b). If each F


n

is dierentiable on [ a,b

], is F necessarily dierentiable on[ a,b ]? If so, does ddx

lim
n

F
n

( x ) = lim
n

ddxF
n

( x )? Question (b). If each

f
n

is dierentiable on [ a,b ], is S (

x )=

n =1

f
n

( x ) necessarilydier entiable on [ a,b

]? If so, does ddx

n =1

f
n

( x )=

n =1

ddxf
n

x )? Question (c). If each F


n

is Riemann integrable on [

a,b ], is F necessarily Riemannintegra ble on [ a,b ]? If so, does

ba

lim
n

F
n

( x ) dx = lim
n

ba

F
n

( x

) dx ? Question (c). If each f


n

is Riemann integrable on [ a,b ], is S ( x )=

n =1

f
n

x )necessarily Riemann integrable on [ a,b ]? If so, does

ba

n =1

f
n

( x ) dx =

n =1

ba

f
n

( x

) dx ?Without additional hypothesis the answer to all of these

questions is generally no . This additional hypothesis is so-called

uniform convergence , introduced byWeierstrass in 1850. For his many contributions to the subject

area, Weierstrassis often referred to as the father of modern analysis. The subsequent study on

thesubject together with questions from geometry and physics also lead to a new areacalled topology

in the end of 19th century. Poincare is often referred to as the fatherof topology. Topology together with

Riemann geometry provide the mathematicsfou ndation for Einsteins relativity theory. You

may learn some basic knowledge of topology and modern geometry in 4000 and 5000 modules.

56 3. SEQUENCES AND SERIES OF FUNCTIONS

Below we only give counterexamples to Questions (a)

and (c). You may read thetext book [ 2 ] for more examples.
Example

1.7 (Counterexample to Question (a))


.

Consider the functions F


n

( x )= x
n

,x

[0 , 1] . For each xed x

[0 , 1), we havelim
n

F
n

( x

) = lim
n

x
n

= 0 (since

|
x

|
< 1) . At x = 1, we havelim

F
n

(1) = lim
n

1
n

=1

. Thus

{
F
n

converges pointwise to the function F on the interval [0 , 1] given by

F ( x )=

, for x

[0 , 1)

, 1 ,x =1 . Each F
n

is continuous on the whole interval [0 , 1], but F is not continuous at

x = 1.This simple example gives a counter example to Cauchys (false) statement thatthe limit of continuous

functions is continuous, namely the limit of continuous func-tions need not be continuous.
Example

1.8 (Counterexample to Question (c))


.

Consider the functions F


n

( x )=

x, 0 <x<
1 n

, 2 n


n
2

x,
1 n

x<
2 n

, 0 ,
2 n


x< 1 . For each xed x


(0 , 1], one sees that F
n

x ) = 0 whenever n

2 x

, and hencelim
n

F
n

( x ) = lim
n

0=0

. Also, at x = 0, we havelim
n

F
n

(0) = lim

n
2

0=0 . Thus,

{
F
n

}
converges pointwise to the zero function

F ( x )

0 on the interval [0

, 1].For each n

1, we have

10

F
n

( x

) dx =

1 /n

n
2

xdx +

2 /n 1 /n

(2 n

n
2

x ) dx +

12

/n

0 dx = n
2

x
2

2 nx


n
2

x
2

n
1

+ 0=12+ (4

2)

(2

12) + 0 = 1 .

Thus we havelim
n

10

F
n

( x ) dx = lim
n

1=1 =0=

10

F ( x ) dx, i.e.
2. UNIFORM

CONVERGE NCE 57

lim
n

10

F
n

( x ) dx

10

lim
n

F
n

( x )

dx.

Reason: At dierent x , F
n

( x

) converges to F ( x ) at dierent pace (more specically,in the denition of

pointwise convergence, the choice of N depends on both and x

). 2. Uniform Convergence We dene a slightly dierent concept of convergence.

2.1. Uniform Convergence of Sequences of Functions.


Definition

2.1
.

{
F
n

}
is said to converge uniformly

to a function F on a set I if for every > 0, there exists an

N (which depends only on ) such that

|
F

( x )

F (

x )

|
< for ALL x


I whenever n>N .
Remark

2.2

If

{
F
n

converges uniformly to F on I , then

F
n

}
converges pointwise to F on

I . Conversely, if

{
F
n

converges pointwise to F on I , then

F
n

}
need not converge uniformly to F

on I . The inequality in the denition can be expressed as F

( x )

<F
n

x ) <F ( x )+ for all

I and n>N

. The geometric interpretation is that for n>N the graphof y = F

( x ) lies in the band spanned by the curves y =

F ( x )

and

y = F ( x )+

,that is, given any > 0 the graph of y = F


n

( x ) eventually lies in the band spannedby the curves y

= F ( x )

and y = F ( x )+

.
Example

2.3
.

The sequence of function F


n

x
n

converges pointwise to 0 on[0 , 1). But from the graph we

can see that the graph of F


n

= x
n

DOES NOTeventually lie in the band spanned by y =

and y =+

. The geometric reason alsotells us that

{
x
n

does not converge uniformly on [0 , 1). 2.2. Two Criteria for

Uniform Convergence of

{
F
n

}
.

The following the-orem is useful (computationall y) in determining whether a sequence of

functionsconver ges uniformly or not.


Theorem

2.4 ( T -test)
.

Suppose

{
F
n

}
is a sequence of functions

converging point-wise to a function F on a set I , and let T

= sup
x I

|
F
n

x )

F ( x )

|
. Then

{
F
n

}
converges uniformly to F on I if and only if

lim
n 0

T
n

=0 .
Proof.

First we prove the only if part. Suppose that

{
F
n

}
converges uniformlyto F on I

. Then for any given > 0, there exists N such that

F
n

( x )

( x )

|
< 2for all n>N

and x

I T
n

= sup
x I

|
F
n

x )

F ( x )

|
2 < for all n>N

58 3. SEQUENCES AND SERIES OF FUNCTIONS

|
T
n

|
= T
n

< for all n > N. Hence we have lim


n 0

= 0.Next we prove the if part. Suppose that lim


n 0

= 0. Then for any given > 0,there exists N such that

|
T
n

= T
n

< for all n>N

sup
x I

|
F
n

x )

F ( x )

|
< for all n>N

|
F

( x )

F (

x )

|
< for all n>N and

I. Hence

{
F
n

}
converges uniformly to F on I

. This nishes the proof of the theorem.


Theorem

2.5 (Cauchys Criterion)


.

A sequence of functions

{
F
n

converges uniformly on a set I if and only if given any > 0

, there exists a natural number N such that (6)

|
F

( x )

F
m

x )

|
< for all x


I and all m,n > N. Remark: Here N

does not depend on x .


Proof.

First we prove the only if

part. Suppose that

{
F
n

converges uniformlyto the function F on I . Then given any

> 0, there exists N such that

|
F
n

( x )

F ( x

|
< 2for all x

I and all n > N. Then for all x

and m,n > N ,

|
F
n

x )

F
m

( x

|
=

|
( F
n

( x )

F ( x

))

( F
m

( x

F ( x ))

||
F
n

( x )

F ( x )

|
+

|
F
m

( x )

F ( x )

|
< 2+

2= . This nishes the proof of the only if part.Next we

prove the if part. Suppose that equation 6 holds. Then for each xed point x

I ,

{
F
n

( x

}
is a Cauchy sequence of real numbers, and thus by Cauchyscriteri

on for sequences, the sequence of real numbers

{
F
n

( x )

}
converges. For each x


I , we denote the limit by F ( x

) = lim
n

F
n

( x ). Then

{
F ( x )

}
x

forms a functionon I , which we denote by F

. Given any > 0, by equation 6, there exists N suchthat

F
n

( x )

F
m

( x )

|
< 2for all x


I and all m,n > N. Then for each xed x


I and n>N , we have

F
n

( x )

( x )

|
=

F
n

( x )

lim
m

F
m

( x )

|
= lim

|
F
n

( x )


F
m

( x )

lim
m

2= 2 <

Thus

{
F
n

}
converges uniformly to

F , and this nishes the proof of the if part.


2. UNIFORM

CONVERGE NCE 59

2.3. Examples.
Example

2.6
.

Show that F
n

( x ) =sin
2

xn, x


, +

, converges uniformly.
Proof.

The limiting function F ( x

) is F ( x ) = lim
n

sin
2

xn = 0for all x

, +

). Since T
n

= sup
x

( , + )

|
F
n

( x )

F ( x

|
= sup
x ( , +

sin
2

xn


1 n

0as n


, thus the sequence of functions

{
F
n

( x

}
converges uniformly on (

, +

).
Example

2.7
.

Determine whether the following sequences of functions converge uniformly on the

indicated interval. (a) f


n

( x )=

n
2

ln xx
n

,x [1

, +

) ; (b) f

( x )= n
2

ln xx
n

,x [2 , +

) .
Solution.

Let f ( x ) = lim

f
n

( x ) = 0 for x


1.(a). T
n

= sup
x 1

|
f
n

( x )

|
= sup
x 1

n
2

ln xx
n

= sup
x 1

f
n

( x ). From f
n

( x

)= n
2

1 x

n
3

ln x

x
n 1

= n
2

n
3

ln xx
n +1

=0 , we have

n
2

n
3

ln x = 0 or

x = e
1

. Observe that f
n

x ) is monotone increasing for1

e
1

and monotone decreasing for x

e
1

. Thus T
n

= max
x 1

f
n

( x )= f
n

( e
1

)= n
2

1 n

e
1

= ne

0as n

and so

{
f
n

( x )

}
does NOT converge

uniformly.(b). Since e
1

2 for n


2, the function f
n

( x

) is monotone decreasing on[2 , +

) for n


2 and so T
n

= sup
x 2

|
f
n

( x )

f ( x )

|
= f

(2) = n
2

ln22
n

for n


2. Sincelim
n

T
n

= 0,

{
f
n

}
converges uniformly on [2 ,

).
Example

2.8
.

Show that

F
n

( x )= n
2

ln

x sin nxx
n

converges uniformly on [2 ,

) .
Solution.

F (

x ) = lim
n

F
n

( x ) = 0 for

2. Observe T
n

= sup
x

|
F
n

( x )


F ( x )

= sup
x 2

n
2

ln x

|
sin nx

|
x
n


n
2

ln22
n

for n


2. Since lim
n

T
n

= 0,

{
F
n

}
converges uniformly.

60 3. SEQUENCES AND SERIES OF FUNCTIONS

Remark. Let

F
n

}
be a sequence of functions on an interval I

. To see whether

{
F
n

}
is uniformly convergent, we

may try to do by the following steps.(1). Determine the limiting function F (

x ) = lim
n

F
n

( x ).(2). Determine

T
n

= sup
x I

|
F
n

( x )

F ( x

|
.(3). Check whether lim
n

T
n

= 0.If T
n

is dicult to be determined, then we may try to estimate an upper bound of

T
n

( a lower bound of T
n

if we guess that the sequence of

functions might not beuniformly convergent). 2.4. Uniform Convergence of Series of Functions.
Definition

2.9
.

n =1

is said to converge uniformly (to S ) on I

if the sequenceof its partial sums

{
S
n

converges uniformly (to S ) on I .


Theorem

2.10 (

T -test for Series of Functions)


.

Suppose

n =1

f
n

x ) is a series of functions converging pointwise on a set I

, and let T
n

= sup
x I

= n +1

f
k

( x )

. Then

n =1

f
n

x ) converges uniformly on I if and only if lim


n

T
n

=0 .
Proof.

Let S
n

( x )=
n

=1

f
k

( x ) and S (

x )=

k =1

f
k

( x ). Then

|
S
n

( x )

S ( x

|
=

= n +1

f
k

( x )

and so the T -test, Theorem 2.4, applies.

Theorem

2.11 (Cauchy Criterion)


.

A series of functions

n =1

f
n

converges uniformly on a set I if and only if given any > 0

, there exists a natural number N such that

= n +1

f
k

( x )

< for all x


I and all m > n > N. Remark: Here N

does not depend on x .


Proof.

The proof follows by applying the

Cauchy Criterion to the partial sums S


n

( x )=

k =1

f
k

x ).
2. UNIFORM CONVERGE NCE 61

The following test is very

useful in verifying that certain series of functions converge uniformly to some functions on an interval.

Theorem

2.12 (Weierstrass M -test)


.

Consider a series of functions

=1

f
k

on a set I . Suppose that (i)

f
k

( x )

|
M
k

for all x

I , k =1

, 2 ,

, and (ii)

k =1

M
k

converges.Then

k =1

f
k

converges uniformly (to some function) on I . Remark. Weierstrass

M -test only states that if a series of functions satises conditions (i) and (ii),

it converges uniformly on I . If a series of functions does not satisfythese two conditions, the

test fails, namely, no conclusion that you can claim fromthis test.
Proof.

Since

k =1

M
k

converges (by (ii)), by the Cauchy Criterion, given any > 0, there exists N

such that
m

k = n +1

M
k

= n +1

M
k

< for all m>n>N because

M
k

0. Then for all x

, we have

= n +1

f
k

( x )

k = n +1

f
k

( x )

|
m

k = n +1

M
k

< . Thus, by the Cauchy criterion,

k =1

f
k

converges uniformly on I .

2.5. Examples.
Example

2.13
.

Show that

n =1

cos
n

xn
2

+ x converges uniformly on (0 ,

) .
Proof.

Since

cos

xn
2

+ x


1 n
2

for all x

(0 ,

) and the series

n =1

1 n
2

is convergent by the p -series, the series of functions

n =1

cos
n

xn
2

+ x converges uniformly by the Weierstrass M -test and hence theresult.

62 3. SEQUENCES AND SERIES OF FUNCTIONS Example

2.14
.

Does the series of functions

n =1

n
2

x
n

sin nx

converge uniformly on the interval [0 ,


12

? Justify your answer.


Solution.

Note that

x
n

sin nx


n
2

2
n

for x


[0 ,
12

]. Sincelim
n

n + 1)
2

2
n +1

n
2

= lim
n

( n + 1)
2

n
2

=12 < 1 , the series

n =1

n
2

2
n

converges and so the series of functions

=1

n
2

x
n

sin nx convergesunifor mly on [0

,
12

] by the Weierstrass M -test.


Example

2.15

Show that the series

n =1

1)
n +1

x
n

converges uniformly on [0 , 1] . Note. The

M -test fails for this example because sup


x [0 , 1]


1)
n +1

x
n

=1

n but

n =1

1 n diverges. In this case, we use T -test or Cauchy Criterion.


Proof.

Let a
n

( x )= x
n

n . Then, for 0

1, we have

a
1

( x )

a
2

( x )

0 and lim
n

( x )=0 . Thus the series

n =1

1)
n +1

x
n

n converges pointwise on [0 ,

1] by the Alternating SeriesTest. By the Alternating Series Estimation, T


n

= sup
0 x 1

= n +1

1)
k +1


sup
0 x

x
n +1

n + 1=1 n +1

. Since lim
n

1 n + 1= 0, the series (


1)
n +1

x
n

converges uniformly by the T -test.


Example

2.16

Show that the series

n =1

x
n

n does not converge uniformly on [0 ,

1) .
Proof.

Observe that T
n

= sup
0

x< 1

|
S
n

( x )


S ( x )

= sup
0 x< 1

= n +1

x
k

= sup
0 <x< 1

k = n +1

x
k

k.
3. UNIFORM CONVERGENCE OF {
F n

} AND CONTINUITY 63

= sup
0 <x< 1

x
n

+1

n + 1+ x
n +2

n + 2+


sup

0 <x< 1

x
n +1

n + 1+ x
n +2

n + 2+

+ x
2 n

2 n

=1 n + 1+1 n

+ 2+

+12 n because x
k

k monotone increasing on (0 , 1)

12

n +12 n +

+12 n

= n

12 n =12 .

Thus the sequence

{
T
n

does not tend to 0 and so the series of functions

k =1

x
k

k doesNOT converge uniformly on [0 ,

1) by the T -test.

3. Uniform Convergence of

F
n

}
and Continuity In this section we will prove that the limit of

uniformly convergent sequence of continuous functions is again continuous.


Theorem

3.1
.

Let

{
F
n

be a sequence of continuous functions on an interval I .Suppose that

F
n

}
converges uniformly to a function F

on I . Then F is continuous on I .
Proof.

Fix any point x


0

I . We are going to show that

F is continuous at thepoint x
0

. Given any > 0, since

{
F
n

}
converges uniformly to F

on I , there exists N such that

|
F

( x )

F (

x )

|
< 3for all x


I and all n > N. Next we x an n>N (say,

n =[ N ] + 1). Since F
n

is continuous at x

, there exists > 0 (here depends on x


0

and ) such that for all x satisfying

x
0

|
< , we have

|
F
n

( x )

F
n

( x
0

|
<

3 . Then for all x satisfying

|
x


x
0

|
< , we have

|
F ( x )

F ( x
0

|
=

|
F ( x )

F
n

( x )+ F
n

x )

F
n

( x
0

)+ F
n

( x
0

F ( x
0

||
F

( x )

F
n

x )

|
+

|
F
n

( x )

F
n

x
0

|
+

|
F

( x
0

F (

x
0

|
< 3+

3+ 3= . Thus F is continuous at

x
0

. Since x
0

is arbitrary, it follows that F

is continuous on I .This nishes the proof of the theorem.


Example

3.2

Find the pointwise limit F of the sequence F


n

x )= x
2 n

1+ x
2 n

,x [0 , 1] .

Show using Theorem 3.1 that the convergence is not uniform.


Solution.

If 0


x< 1, we havelim
n

F
n

x ) = lim
n

x
2 n

1+ x

2 n

=01 + 0= 0 .
64 3. SEQUENCES AND SERIES OF FUNCTIONS

If

x = 1, then F
n

(1) =12. Thus the limiting function F

( x ) is F ( x )=

00


x< 1
12

x = 1Because each

F
n

( x ) is continuous but F (

x ) is not, the sequence of functions

{
F
n

( x )

}
does not converge uniformly on [0

, 1] by Theorem 3.1. It is possible that each F


n

and F ( x ) = lim
n

F
n

( x ) are continuous, but

{
F
n

}
doesnot converge uniformly to F ( x

).
Example

3.3
.

Let F
n

( x

)= nxe
nx
2

. Show that 1) Each F

( x ) is continuous on I = [0 ,

1] . 2)

{
F
n

x )

}
converges pointwise to a continuous function on

I . 3)

{
F
n

x )

}
does not converge uniformly on I

.
Proof.

Each F
n

( x

) is continuous because it is a well-dened elementary functionon [0 , 1]. Let F

( x ) = lim
n

F
n

( x

) on [0 , 1]. When x = 0, F
n

(0) = 0 for each

n and F (0) = lim


n

0 = 0, when x

= 0 (xed), F ( x ) = lim
n

F
n

( x ) = lim
n

nxe

nx
2

= lim
n

nxe
nx
2

= lim
n

xe
nx
2

x
2

= lim
n

1 xe
nx
2

=0 . Thus F

( x ) = 0 is continuous on [0 , 1]. Now T

= sup
0 x 1

|
F

( x )

F (

x )

|
= sup
0 x 1

F
n

( x ).From F
n

x )= ne
nx
2

+ nxe
nx

2 nx )=

ne
nx
2

(1

2 nx
2

)=0 , we have x =

12 n . Since F
n

(0) = 0 and F
n

(1) = ne
n

,sup
0

x 1

F
n

( x ) = max
0

x 1

F
n

( x ) = max

0 ,ne
n

,n

2 n

e
n
12

n 2 e.

Thus T
n

2 e . Since lim
n

T
n

=+


, the sequence

{
F
n

does not converge uniformly to F ( x ) on [0 ,

1].
Corollary

3.4
.

Suppose that

k =1

f
k

converges uniformly to a function S on an interval I . Suppose that each

f
k

is continuous on I . Then S is also continuous on

I .
Proof.

Consider the sequence of partial sums

S
n

}
on I , where we have S
n

=
n

k =1

f
k

. Then

{
S
n

}
converges uniformly to

S on I . If each f
k

is continuous on I

, theneach S
n

is also continuous on I . Then by Theorem 3.1,

S is also continuous on I .
Example

3.5

Is

n =1

xn

e
nx

,x (0 ,


) , a continuous function?
4. UNIFORM CONVERGENC E AND

INTEGRATION 65 Solution.

Let f
n

( x )=

xn
2

e
nx

. We nd an upper bound of f
n

x ). Observe that f
n

( x )=

xn
2

e
nx

xe
nx

n
2

= e
nx


nxe
nx

n
2

= e

nx

1 n

n. We obtain that

f
n

( x ) > 0 for 0

<x<
1 n

and f
n

( x

) < 0 for x>


1 n

. It follows that f
n

( x ) is monotone increasing on (0 ,
1 n

] and monotone decreasing on [


1 n

, +

). Thussup

0 <x< +

f
n

( x ) = max
0

<x< +

f
n

( x )= f

1 n

=
1 n

n
2

e
n
1

=1 en
3

. Let

M
n

=1 en
3

. Then

|
f

( x )

|
M
n

for

(0 ,

). Since

n =1

M
n

converges bythe

p -series, the series of functions

n =1

xn
2

e
nx

converges uniformly by Weierstrass

M -test on (0 ,

). According to Corollary 3.4, the function

n =1

xn
2

nx

is continuous on(0 ,

).

It is possible that each f


n

and

n =1

f
n

are continuous, but

n =1

f
n

does not convergeunifor mly.


Example

3.6
.

Consider the geometric series

n =0

x
n

=1+ x + x
2


=11

x for

1 <x< 1. In this case, each x


n

and

n =0

x
n

are continuous on I = (0 , 1). Weshow that

n =0

x
n

does not converge uniformly by T -test. Since T


n

= sup

1 <x< 1

x
n

+1

+ x
n +2


sup
0 x<

x
n +1

x
n +2

= sup
0 x< 1

x
n +1

+ x

n +2


sup
0 x< 1

x
n +1

=1 , the sequence

{
T
n

does not tend to 0 and so the series

=0

x
n

does not convergeunifor mly on (

, 1) by the T -test.

4. Uniform Convergence and Integration

Before we go on, we rst recall some facts about Riemann integrals (Reference:Our Text Book [

2 , Chapter 6, pp. 208-216]).


66 3. SEQUENCES AND SERIES OF FUNCTIONS

4.1. Review of Riemann Integration. Let f be a bounded function on a nite interval [

a,b ]. A partition P of [ a,b ] is a set of points

{
x
0

,x
1

,x
n

}
suchthat a = x
0

<x
1

<x
2

<

<x
n

= b. For such a partition P and i =1

, 2 ,

n , we denote M

( f ) = sup
x [ x
i

,x

f ( x ) m
i

f ) = inf
x [ x
i

,x
i

f ( x ) . The upper (Riemann) sum

of f with respect to the partition P is dened to be U (

P,f ) :=
n

i =1

M
i

( f ) x
i

Here x
i

= x
i

i 1

. Similarly the lower (Riemann) sum of f with respect to

P is dened to be L ( P,f ) :=
n

i =1

m
i

f ) x
i

. The upper and

lower integrals of f , denoted by

ba

f ( x ) ,dx and

ba

f ( x )

,dx ,respectively, are dened by

ba

( x ) dx = inf

{
U

( P,f )

|
P is a partition of[

a,b ]

}
,

ba

f ( x ) dx = sup

{
L ( P,f )

P is a partition of[ a,b ]

}
.

Remark. If f is bounded, then

ba

f ( x ) dx

ba

f ( x )

dx, see [ 2 , Theorem 6.1.4, p.211].A function f is said to be

Riemann integrable on [ a,b ] if

ba

f ( x )

dx =

ba

f (

x ) dx. The common value is called the Riemann integral

of f over [ a,b ], and it is denotedby

ba

f ( x )

dx =

ba

f (

x ) dx =

ba

f ( x ) dx . Remark.

In the following, (1) and (2) are from [ 2 , Theorem 6.1.8].1) Any continuous

function on a nite interval [ a,b ] is Riemann integrable.2) Any monotone function on a nite interval [

a,b ] is Riemann integrable.3) Any elementary function is continuous on its domain.

4. UNIFORM CONVERGENC E AND INTEGRATION 67

The elementary functions are the

polynomials, rational functions, power functions( x


a

), exponential functions ( a
x

), logarithmic functions, trigonometric and

inversetrigonom etric functions, hyperbolic and inverse hyperbolic functions, and all func-tions that can be

obtained from these by ve operations of addition, subtraction,mul tiplication, division, and composition.

4.2. The Theorems.


Theorem

4.1
.

Let

{
F

}
be a sequence of Riemann integrable functions on a nite interval

[ a,b ] . Suppose that

{
F
n

}
converges uniformly to a function F on [

a,b ] .Then F is Riemann integrable on [ a,b

] , and lim
n

ba

F
n

( x ) dx =

ba

F ( x )

dx i.e. lim
n

ba

F
n

( x ) dx =

ba

lim
n

F
n

( x )

dx.
Proof.

By denition, given any

> 0, there exists N such that

|
F
n

( x )

F ( x

|
< 2( b

a )for all n>N and x

a,b ]. Thus F
n

( x )


2( b

) <F ( x ) <F
n

x )+ 2( b

)for all x

[ a,b ] and n>N

. It follows that(7)

ba

F
n

( x )


2( b

dx

ba

F ( x )

dx

ba

( x ) dx

ba

F
n

( x )+ 2( b


a )

dx

for n>N . Since F


n

( x

) is Riemann integrable, we have

ba

F
n

( x )


2( b

dx =

ba

F
n

( x )

2( b

a )

dx =

ba

F
n

( x

) dx

2( b


a )

( b


a )=

ba

F
n

( x ) dx

2 ,

ba

F
n

( x )+

2( b

a )

dx =

ba

F
n

( x )+ 2( b


a )

dx

ba

F
n

x ) dx + 2( b


a )

( b


a )=

ba

F
n

( x )+ 2 .

Together with Inequality (7), we obtain(8)

ba

( x ) dx

ba

( x ) dx

ba

F ( x )

dx

ba

( x ) dx + 2for

n>N . It follows that0

ba

F ( x ) dx

ba

F ( x )

dx

.
68 3. SEQUENCES AND SERIES OF FUNCTIONS

Let

0, we have

ba

F ( x )

dx

ba

( x ) dx = 0and so F ( x

) is Riemann integrable. From Inequality (8), we have

ba

F
n

( x ) dx

ba

F ( x ) dx

ba

F
n

( x

) dx + 2 , that is,

ba

F
n

( x

) dx

ba

F ( x ) dx


2 < for n>N and hencelim

ba

F
n

x ) dx =

ba

F ( x ) dx.
Example

4.2

Compute, justifying your answer, lim


n

10

sin nxn + x

dx.
Solution.

Let F
n

( x ) =sin

nxn + x
2

. Then the limiting function F

( x ) = lim
n

F
n

( x

) = lim
n

sin nxn + x
2

= 0for any given 0

1, by the Squeeze theorem, because

1 n


sin nxn + x
2

1 n and lim
n

1 n =


lim
n

1 n = 0. Since T

= sup
0 x 1

|
F

( x )

F (

x )

|
= sup
0 x 1

sin

nxn + x
2


1 n, lim
n

T
n

= 0 by the Squeeze Theorem and so the sequence of functions

{
F

}
convergesunifor mly to F ( x

). Thuslim
n

10

sin

nxn + x
2

dx =

10

lim
n

sin

nxn + x
2

dx =

10

0 dx =0 .

Corollary

4.3
.

Suppose that

k =1

f
k

converges uniformly to a function S on an interval [ a,b ]

. Suppose that each f


k

is a Riemann integrable (bounded) function on

[ a,b ] . Then S is also Riemann integrable on [

a,b ] , and

ba

( x ) dx =

k =1

ba

f
k

( x ) dx, i.e.

ba

k =1

f
k

( x ) dx =

k =1

ba

f
k

( x

) dx.
4. UNIFORM CONVERGENC E AND INTEGRATION 69 Proof.

Consider the sequence of partial sums

{
S
n

on [ a,b ], where S
n

=
n

k =1

f
k

.Then

{
S
n

}
converges uniformly to S

on [ a,b ]. If each f
k

is Riemann integrable on[ a,b

], then each S
n

is also Riemann integrable on [ a,b ]. Then by Theorem 4.1,

S isalso Riemann integrable on [ a,b ], and

ba

S ( x )

dx = lim
n

ba

S
n

( x ) dx = lim
n

ban

=1

f
k

( x ) dx = lim
n

k =1

ba

f
k

( x

) dx =

=1

ba

f
k

x ) dx. By using this theorem, we have the following

amazing formula.
Example

4.4
.

Show that ln2 =

n =1

1 n


2
n

=12+12

2
2

+13

2
3

Proof.

Since

x
n 1

12

n 1

for 0

12and the geometric series

n =0

12

n 1

converges, the series

n =1

x
n 1

converges uniformly to11

x on [0 ,
12

] by the Weierstrass M

-test. Thus we have

12

11


x =

12

n =1

x
n

dx =

n =1

12

x
n 1

dx =

n =1

x
n

1 /

20

n =1

2
n

. Since

12

11

ln(1

x )

12

ln


12


ln

12

= ln2 , we obtain the formulaln2 =

n =1

1 n


2
n

=12+12

2
2

+13

2
3

Remark

4.5
.

Example 4.4 gives a way to estimate the number ln2

because theremainder R
n

k = n +1

1 k


2
k

=1( n + 1)2
n +1

+1( n + 2)2
n +2

<

1( n + 1)2
n +1

+1( n + 1)2
n

+2

+1( n + 1)2
n +3

70 3. SEQUENCES AND SERIES OF FUNCTIONS

=1( n + 1)2
n +1

1 +12+12
2

=1( n + 1)2
n +1


11

12

=1( n + 1)2

. For instance,ln2

12+12

+13

2
3

+110

2
10

with error less than111

2
10

=111264. 4.3. Remarks on Theorem 4.1. For completeness

we include the followingresult, which does not require uniform convergence

. The proof requires new theorycalled Lebesgue integration . There are applications in the area of

probability andstatistics.
Theorem

4.6 (Bounded Convergence Theorem)


.

Let

{
F
n

}
be a sequence of Rie-mann integrable

functions converging pointwise to F ( x ) on

[ a,b ] . Suppose that (i) F ( x

) is Riemann integrable, and (ii) there exists a positive constant M such that

|
F
n

( x )

M for all x

[ a,b ]

and n

.Then lim
n

ba

F
n

( x

) dx =

ba

( x ) dx =

ba

lim
n

F
n

( x ) dx 5. Uniform Convergence and Dierentiation

Theorem

5.1
.

Let

{
F
n

}
be a sequence of functions on [ a,b ] such that

(i) each F
n

exists and is continuous on [

a,b ] , (ii)

{
F
n

}
converges pointwise to a function F on [

a,b ] , and (iii)

{
F
n

}
converges uniformly on [ a,b ] .Then

F is dierentiable on [ a,b ] , and for all x


[ a,b ] , F

( x ) = lim
n

F
n

x ) , i.e. ddx

lim
n

F
n

( x )

= lim
n

ddxF
n

( x

. Remark.

Here the dierentiability and continuity at the endpoints a and b

refer tothe one sided derivatives and limits respectively.


Proof.

By (iii), there exists a function g such that

{
F

}
converges uniformlyto g on [ a,b

]. In particular, lim
n

F
n

( x

)= g ( x ) for all x

[ a,b ]. By (i), since each F


n

is continuous on [ a,b ], F
n

is also Riemann integrable on [ a,b ], and by the fundamentalthe orem of calculus,

xa

F
n

t ) dt = F
n

( x

F
n

( a ) for all

[ a,b ] . Letting

, we have, for all x

[ a,b ],( 9 )
n

l i m

xa

F
n

( t ) dt = lim

F
n

x )

F
n

( a

= F (

x )

F ( a )

.
5. UNIFORM CONVERGENCE AND DIFFERENTIATI ON 71

On the other hand, since

{
F
n

}
converges uniformly to

g on [ a,b ], it follows fromTheorem 4.1 thatlim


n

xa

F
n

t ) dt =

xa

lim
n

F
n

( t )

dt

xa

g ( t

) dt. Together with equation 9, it follows that F ( x

F ( a )=

xa

g ( t )

dt for all x

[ a,b ]

. By (i) and Theorem 3.1, g is continuous on [ a,b

]. Then by the fundamental theoremof calculus, we have ddx

xa

g ( t )

dt = g ( x ) .

Together with equation 9, it follows that F is also dierentiable on [ a,b

], and for all x

[ a,b ], ddx

F ( x )


F ( a )

= ddx

xa

g ( t )

dt = g ( x ) , i.e.

F ( x )= g ( x

) , i.e. ddx

lim
n

F
n

( x )

= lim
n

ddxF
n

( x )

. The proof is nished.


Remark

5.2
.

By inspecting the proof, Theorem 5.1 still holds when the closed interval

[ a,b ] is replaced by ( a,b ) ,

( a,b ] or [ a,b ) .

This theorem can be generalized as follows.


Theorem

5.3
.

Let

{
F
n

}
be a sequence of dierentiable functions on

[ a,b ] such that (a)

{
F

( x
0

}
converges for some

x
0

[ a,b ] , and

(b)

{
F
n

}
converges

uniformly on [ a,b ] .Then

F
n

}
converges uniformly to a function F

( x ) on [ a,b ] with,

F ( x ) = lim
n

( x ) , i.e. ddx

lim
n

F
n

( x )

= lim
n

ddxF
n

( x

The proof of this theorem is omitted, see [ 2 , Theorem 8.5.1, pp.340-341].


Corollary

5.4

Let

k =1

be a series of dierentiable functions on [ a,b ] such that

(a)

k =1

f
k

( x
0

) converges for some x


0


[ a,b ] , and (b)

k =1

f
k

converges

uniformly on [ a,b ] .
72 3. SEQUENCES

AND SERIES OF FUNCTIONS

Then

=1

f
k

converges uniformly to a function S (

x ) on [ a,b ] , and for all x


[ a,b ] , S

( x )=

=1

f
k

( x ) , i.e.

ddx

k =1

f
k

x )

k =1

ddxf
k

x ) .
Proof.

Consider the partial sums S


n

( x )=
n

=1

f
k

( x ) on [ a,b ]. By (a),

{
S
n

( x
0

}
converges, and, by (b),

{
S
n

( x )

}
converges uniformly on [ a,b

]. By Theorem 5.3,

{
S
n

( x

}
converges uniformly to a function S (

x ) on [ a,b ] and so

k =1

f
k

( x ) convergesunifor mly to

S ( x ) on [ a,b ] by the denition. Furthermore, by

Theorem 5.3, for all x

[ a,b ],

S ( x ) = lim
n

( x ) , that is, ddx

=1

f
k

( x )

= lim
n

=1

f
k

( x )

= lim
n n

k =1

f
k

x )=

k =1

ddxf
k

( x ) .

As an application, we give a proof of binomial series. Let a be any real number.The

binomial number, a chooses n , is dened by

an

= a ( a

1)( a


2)

( a

n + 1) n !for positive integers n . For instance,

12

=12 ,

12

=
12

12

2!=

18 ,

12

=
12

12

12

3!=
12


12


32

6=116 . We also use the convention that

a 0

= 1 for any a .
Theorem

5.5 (Binomial Series)


.

Let a be any real constant. Then (1 +

x )
a

=1+ ax + a (

1)2! x
2

=1+

n =1

an

x
n

n =0

an

x
n

for

|
x

|
<

1 .
Proof.

Let be any xed positive number with 0

<< 1 and let I =[

].Consider the series of functions1 +

=1

an

x
n

converges when x =0

, +

]. So it satises condition (a) of Corollary 5.4. Wecheck condition (b) of Corollary 5.4, namely, the

series of functions

n =1

an

nx
n 1

5. UNIFORM CONVERGENCE AND DIFFERENTIATI ON 73

converges uniformly on [

, + ]. Note that

an

nx
n 1

an

n
n 1

for

x [


, +

]. Let M
n

an

n
n 1

. Thenlim
n

M
n +1

M
n

= lim
n

an +1

n + 1)
n

an

n 1

= lim
n

|
a

||
a

||
a

|
( n + 1)!


( n + 1)

|
a

||
a

||
a

n +1

n !

= lim
n

|
a

||

||
a

|
n !

n + 1)

( n + 1)!

|
a

||
a

||
a

n +1

n = lim
n

|
a

|
n = lim
n

a/n

|
=

< 1 . Thus the series

n =1

M
n

converges and so the series of functions

n =1

an

nx
n 1

con-verges uniformly on [

, ], by the Weierstrass M

-test.By Corollary 5.4, the series 1+

=1

an

x
n

converges uniformly to a function

f ( x )on [

, ] with

f ( x )=

n =1

an

nx
n 1

. Next we are going to set up a dierential equation that(1 + x )

f ( x )= af ( x

) . ( Note. Since the goal is to show that f (

x ) = (1+ x )
a

, this equation is observed fromthat, if

y = (1 + x )
a

, then y

= a (1 + x )
a 1

and so (1 +

x ) y = (1 + x )
a

= y .)Now f ( x )=

n =1

an

nx
n 1

n =1

a ( a

1)


( a

n + 1)


nn ! x
n 1

n =1

a (

1)

( a


n + 1)( n

1)!

x
n 1

= a

n =1

1 n

x
n 1

and(1 +

x ) f ( x )= f

( x )+ xf ( x

)= a

n =1

1 n

x
n

+ a

=1

= a

n =0

1 n

x
n

+ a

n =1

1 n

x
n

= a

1+

n =1

1 n

= a

1+

n =1

an

x
n

= af ( x

) ,
74 3. SEQUENCES AND SERIES OF FUNCTIONS

where

1 n

1 n

=( a

1)(

2)

( a

n + 1) n

!+( a

1)( a

2)

( a


n + 2)( n

1)!=(

1)( a

2)


( a

n + 2) n !( a


n +1+ n )= a (

1)

( a


n + 1) n !=

an

. Let y = f ( x

). Then we obtain the dierential equation(1 + x ) dydx =

aydyy = adx 1+ x =

dyy =

adx 1+ x =


ln

|
y

|
=

a ln

|
1+ x

+ A = ln

|
1+ x

|
a

+ A = y

|
= e
ln | y |

e
A

|
1+ x

|
a

y = C

1+ x

|
a

, where C

e
A

is a constant. By putting x

= 0, C = (1 + 0)
a

= y (0) = 1 +

n =1

an

0
n

=1 .

Thus y =

|
1+ x

|
a

or(1 + x )
a

=1+

n =1

an

x
n

for

|
x

|
because 1+ x>

0 when

|
x

|
< 1. Since

is any number with 0 << 1,the formula(1 + x )

=1+

n =1

an

x
n

holds for all x

1 , 1). For instance,

1+

x = (1 + x )
12

k =0

12

x
k

= 1 +12

x +

12


12

2! x
2

x
3

= (1


x
3

)
12

k =0

12

=1

12 x
3

12


12

2! x
6


4 . 1=

4 +110

12

=2

1 +140

12

=2

k =0

12

140

.
Example

5.6
.

Evaluate

. 1 with error less than 0 . 001 .

6 . P O W E R S E R I E S 7 5 Solution.

4 .

1=

4 +110

12

=2

1 +140

12

=2

k =0

12

140

. =2+2

k =1

12

140

. Now

12

=
12

12

12

k +1

k != (

1)

k +112


12

12

k !for

2. Let b
k

=
12

12

12

k !

140

. Then, for k

2, we have b
k


0, b
k +1

b
k

12


12

12

12

( k + 1)!

140

k +112


12

12

k !

140

= k

12

40( k + 1)


1 , that is, b
2

b
3

0, and lim
k

b
k

= 0 by the Squeeze Theorem because0

b
k


12


( k

1) k

140

=12 k

40
k

for k

2 and lim
k

12 k

40
k

= 0. By the alternating

series estimation, from2

12

k +1

140

k +1

< 0 . 001

, we have k

1, because 2 b
2

=16400 < 0 . 001, and so

. 1

2+2

12

140= 2 .

025with error less than 0 . 001.

6. Power Series6.1. Power Series.

Definition

6.1
.

A power series in x is of the form

n =0

a
n

x
n

= a
0

+ a
1

x +

a
2

x
2

76 3. SEQUENCES

AND SERIES OF FUNCTIONS Example

6.2
.

Below are some examples1.

n =0

( n + 1)

x
n

=1+2 x +3 x
2

+4

x
3

+5 x
4

2.

n =0

x
n

!= 1 + x + x
2

2!+ x
3

3!+

Definition

6.3
.

A power series in

x
0

is of the form

n =0

a
n

x
0

)
n

= a

+ a
1

( x

)+ a
2

( x

)
2

Example

6.4
.

Here are some examples.1.

n =0

( x

1)
n

=1+( x


1) + ( x

1)
2


. 2.

n =1

n
2

( x + 2)
n

=(

x + 2) + 2
2

( x + 2)
2

+3
2

( x + 2)
3

Warning. Dont expand out the terms a


n

( x


x
0

)
n

in the power series

n =0

a
n

x
0

)
n

because, when you rearrange

terms in an (innite) series, you may get dierent values.(For partial sums, you can expand out, if it is

necessary, because there are only nitelymany terms.) Question: Given a power series

n =0

a
n

x
0

)
n

, when does it converge and

whendoes it diverge? In other words, what is the domain of the function

n =0

a
n

x
0

)
n

.We are going to answer this question. 6.2. Radius of Convergence.


Definition

6.5
.

Given a power series

n =0

a
n

( x

x
0

, the radius of conver-gence R is dened by R =1lim

n
n

|
a

|
. If lim
n
n

|
a
n

|
=

, we take R =0

, and if lim
n
n

|
a
n

|
=0 , we set R

. If lim
n

a
n +1

||
a
n

exists, R is also given by R =1lim


n

a
n +1

||
a
n

6 P S 7

. O W E R E R I E S 7

Remark. Recall thatlim

|
a

n +1

||
a
n

|
lim
n

|
a
n

|
lim
n

|
a
n

|
lim

a
n +1

||
a
n

. If lim
n

|
a
n +1

||
a
n

|
exists, thenlim

a
n +1

||
a
n

= lim

|
a
n +1

||
a

|
and so lim
n
n

|
a
n

|
exists andlim
n
n

|
a
n

|
= lim
n

|
a
n +1

||
a
n

|
.
Example

6.6
.

What is the radius of

convergence for the series 1+ x 3+ x


2

+ x
3

3
3

+ x
4

4
4

+ x
5

3
5

+ x

4
6

Solution.

Since a

14

2 k

n =2 k 13
2 k 1

n =2 k

1 , we have

|
a
n

|
=

14

n =2 k 13 n =2 k


1 . Thus lim
n
n

|
a
n

|
=13and so the radius of convergence R =1lim
n

|
a
n

|
=1
13

=3 .
Example

6.7

Find the radius of convergence of the power series

n =0

(4 x + 3)

n
3

Solution.

Observe that

n =1

(4 x + 3)
n

n
3

n =1

4
n

x +34

. Thus R =1lim

|
a
n +1

||
a

|
=1lim
n

4
n +1


n
3

( n + 1)
3


4
n

=1lim
n

1+
1 n

=14 .

78 3. SEQUENCES AND SERIES OF FUNCTIONS Theorem

6.8
.

Given any power series

k =0

a
k

x
0

)
k

with radius of convergence

R , 0

, then the series

k =0

a
k

( x

x
0

)
k

(i)

converges absolutely for all x with

|
x


x
0

|
<R , and (ii)

diverges for all x with

|
x

x
0

|
>R .
Proof.

By denition, the radius of convergence R =1lim

|
a

|
1

. Assertion (i)follows from the root test

because, fromlim

a
k

x
0

)
k

= lim

|
a

|
1

|
x

x
0

|
=

|
x


x
0

|
lim

a
k

|
1

|
x


x
0

|
1 R< R


1 R< 1 , the series

k =0

a
k

( x

x
0

converges.Next we are going to prove (ii) by

contradiction. Suppose that

k =0

a
k

( x

x
0

converges at a point x with

|
x


x
0

|
>R

. Then by Theorem 1.7, we havelim


k

a
k

( x


x
0

)
k

=0 . Let

= 1. Then there exists N such that

|
a

( x

x
0

)
k

|
< 1 for all k>N

|
a
k

( x

)
k

|
1

< 1 for all k>N

|
a
k

|
1

< 1

|
x

x
0

for all k>N

sup
n k

|
a
k

|
1

|
x

x
0

|
for all n>N

lim

a
k

|
1

|
x

x
0

1 R

|
x


x
0

|
< 1 R,

which is a contradiction. Hence we must have

k =0

a
k

( x

)
k

diverges at each x satisfying

|
x


x
0

|
>R .

6.3. Interval of convergence. In view of Theorem 6.8, for a power series

k =0

a
k

x
0

)
k

with radius of convergence

R , the set of points at which

=0

a
k

( x

x
0

)
k

is convergent form an interval called the interval of convergence

, which must beeither( x


0

R,x
0

R ) , ( x
0

R,x
0

+ R ] , [ x

R,x
0

+ R ) or [ x

R,x
0

+ R ] .

Example

6.9
.

Find the interval of convergence of the power series.(i)

n =1

( x


2)
n

n
2

(ii)

n =1

( x


2)
n

n (iii)

n =1

n ( x


2)
n

6 . P O W E R S E R I E S 7 9 Solution.

(i). First we nd the radius of convergence R =1lim


n

a
n +1

||
a
n

=1lim
n

n
2

( n + 1)
2

=1lim
n

1(1 +
1 n

)
2

=1

. Next we check the endingpoints x


0

R =2

1=1 , 3. When x

= 1, the seriesis

n =1


1)
n

n
2

, which is convergent by

Example 3.5. When x = 3, the series is

n =1

1 n
2

,which is convergent by the p -series. Thus the interval of convergence is [1

, 3].(ii). The radius of convergence is R =1lim


n

|
a
n +1

||
a
n

|
=1lim
n

n ( n + 1)=1lim

1(1 +
1 n

)= 1 .

Now we check the endingpoints x


0

=1 , 3. When x = 1, the series is

n =1

1)
n

n ,which is convergent by Example 3.5. When x

= 3, the series is

n =1

1 n , which isdivergent by the p -series. Thus the interval of

convergence is [1 , 3).(iii). The radius of convergence is R =1lim

|
a
n +1

||
a

|
=1lim
n

n +1 n

=1lim
n

1 +1 n

=1 .

Now we check the endingpoints x


0

=1 , 3. When x = 1, the series is

n =1

n (


1)
n

,and when x = 3, the series is

n =1

n . Both of these series are

divergent by thedivergence test. Thus the interval of convergence is (1 , 3).

6.4. Uniform Convergence of Power Series.


Theorem

6.10 (Abel Theorem)


.

Let

n =0

a
n

( x

x
0

)
n

be a power series, and let R

0 . (1).

If

n =0

a
n

R
n

converges, then the series of functions

n =0

a
n

x
0

)
n

converges uniformly on

[ x
0

,x
0

+ R ]

. (2). If

=0

a
n

R )
n

converges, then the series of functions

=0

a
n

( x

x
0

)
n

con-verges uniformly on [ x
0

R,x
0

] .
Proof.

We only prove assertion (1). The proof of

assertion (2) is similar to.


80 3. SEQUENCES AND SERIES OF FUNCTIONS

We may assume that

R> 0. Let t = x

R . Then

n =0

a
n

( x

x
0

n =0

R
n

t
n

. We are going to show that this series

converges uniformly on 0

1, that is,

x
0

x
0

+ R . Write a
n

for a
n

R
n

. By the assumption, the series

n =0

a
n

n =0

a
n

R
n

converges. Let A
n

=
n

k =0

a
k

k =0

a
k

k = n +1

. Then A
n

A
n 1

k = n +1

k = n

a
k

= a
n

By Abel Partial Summation Formula, for 0

n<m ,0

1,

= n +1

a
k

t
k

A
m

t
m


A
n

t
n +1

+
m

k = n +1

( t
k

t
k +1

|
A
m

|
t
m

|
A
n

|
t
n +1

+
m 1

k = n +1

A
k

|
t
k

(1

t ) . Since

n =0

a
n

converges, the remainders

k = n +1

A
n

tends to 0 and so lim


n

A
n

= 0.Given > 0, there exists N such that

|
A
n

|
< 2for

n>N . Now, for m>n>N and0


1,

= n +1

a
k

t
k

|
A
m

|
t
m

|
A
n

|
t
n +1

+
m 1

k = n +1

A
k

|
t
k

(1

t ) < 2 t
m

2 t
n +1

+
m 1

k = n +1


t
k

(1

t )= 2

t
m

+ t
n +1

+ (1

t )( t
n +1

+ t
n +2

+ t
m 1

= 2

t
m

+ t
n

+1

+( t
n +1

+ t
n +2

+ t
m 1


( t
n +2

+ t
n

+3

+ t
m

= 2


2 t
n +1

By the Cauchy Criterion, the series of functions

n =0

a
n

t
n

n =0

a
n

x
0

)
n

convergesunifor mly on 0

1, or on x
0

x
0

+ R

.
Theorem

6.11 (Uniform Convergence Theorem)


.

Let

n =0

a
n

x
0

)
n

be a power series of radius of convergence R> 0 . Let I

be the interval of convergence. Then


6 P S 8 . O W E R E R I E S 1

the series of functions

n =0

a
n

x
0

)
n

converges uniformly on any closed interval [ c,d ]


I . Remark. (1). I =[

x
0

R,x
0

+ R ], [

x
0

R,x
0

+ R ), (

x
0

R,x
0

+ R ], or (

x
0

R,x
0

+ R ).If

I =[ x
0

R,x
0

R ], then the power series converges on I . In other cases, since I

is not closed, the theorem says that the power series converges on any closed sub-interval of

I . For instance, if x
0

= 0, R = 1, and

I =[

1 , 1), then the power

seriesconverges uniformly on [

1 , 0 .

9], [0 , 0 . 9] and etc, but it need not converge uniformlyon [


1 , 1) or [0 , 1).(2). In any of the four cases,

the power series converges uniformly on any closedsubinterval of ( x

R,x
0

+ R ).
Proof.

There are three cases: (i). c

x
0

d , (ii). x
0

<c

, or (iii). c

d<x
0

. Case (i).

x
0

d . Since

n =0

a
n

x
0

)
n

and

n =0

a
n

x
0

)
n

(because c,d


I ),where d

x
0


0 and c

x
0


0, the power series

n =0

a
n

( x

)
n

converges uniformlyon c


x
0

] and x
0

d by the Abel theorem and so on the union [ c,d

] =[ c,x
0

[ x
0

,d ]. Case (ii). x
0

<c

d . Since

n =0

a
n

( d

x
0

converges, the power series

n =0

a
n

( x

x
0

converges uniformly on [ x
0

,d ] by the Abel Theorem and so

on the subinterval[ c,d ]

[ x

,d ]. Case (iii). c

d<x
0

. Since

n =0

a
n

( c

x
0

)
n

converges, the power series

n =0

a
n

( x

x
0

converges uniformly on [ c,x


0

] and so on [ c,d ]


[ c,x
0

].
Corollary

6.12

Let

n =0

( x

x
0

)
n

be a power series of radius of convergence R> 0 , and let I

be the interval of convergence. Suppose that [ c,d ] is a closed subinterval of

I . Then

dc

n =0

a
n

x
0

)
n

dx =

n =0

a
n

x
0

)
n +1


( c

x
0

n +1

n +1 .
Proof.

Since

n =0

a
n

x
0

)
n

converges uniformly on [

c,d ],

dc

n =0

a
n

x
0

)
n

dx =

n =0

dc

a
n

( x


x
0

)
n

dx =

n =0

a
n

x
0

)
n +1


( c

x
0

n +1

n +1 .
82 3. SEQUENCES AND SERIES

OF FUNCTIONS Corollary

6.13 (Abel)
.

Let

n =0

a
n

x
0

)
n

be a power series of radius of con-vergence R> 0 . (a). If

n =0

a
n

converges, then lim


x ( x
0

+ R )

n =0

a
n

x
0

)
n

n =0

a
n

R
n

. (b). If

=0

a
n

R )
n

converges, then lim


x ( x
0

R )
+

n =0

a
n

x
0

)
n

n =0

a
n


R )
n

.
Proof.

(a). By the Abel theorem, the power series

=0

a
n

( x

x
0

)
n

convergesunifor mly on [ x
0

,x
0

R ] and so the function f ( x )=

n =0

a
n

x
0

)
n

continuous on[ x

,x
0

+ R ]. Hence

n =0

a
n

R
n

= f ( x
0

+ R ) = lim

x ( x
0

+ R )

f (

x ) = lim
x ( x
0

+ R )

n =0

a
n

x
0

)
n

The proof of (b) is similar to that of (a).


Example

6.14
.

From the geometric series

n =0

x
n

=11

x for

|
< 1, we have11 + t =

n =0

1)
n

t
n

by letting x =

t . For any x

1 , 1), we haveln(1 + x )=

x 0

11 + tdt =

n =0

1)
n

x 0

t
n

dt =

n =0

1)
n

x
n +1

+1 . Since

=0

1)
n

x
n +1

n + 1converges when x = 1, we have

n =0

1)
n

1 n + 1= lim
x 1

n =0

1)
n

x
n +1

n + 1= lim
x 1

ln(1 + x ) = ln2 . In other words,ln2 = 1

12+13

14+

.
7. DIFFERENTIA

TION OF POWER SERIES 83 Example

6.15
.

From the geometric series

n =0

x
n

=11


x for

|
x

< 1, we have11 + t
2

n =0

1)
n

t
2 n

by letting x =


t
2

. For any x


1 , 1), we havearctan x =

x 0

11 + t
2

dt =

n =0

1)
n

x 0

t
2 n

dt =

n =0

1)
n

x
2 n +1

2 n +1

. Since

n =0

1)
n

x
2 n +1

2 n + 1converges when x =

1, it converges uniformly on [

1 , 1] andsoarctan x

n =0


1)
n

x
2 n +1

n + 1for all

|
x

|
1

. In particular, 4= arctan1 =

n =0

1)
n

12 n

+ 1= 1

13+15

17+

.
Example

6.16
.

Fromarctan x =

n =0

1)
n

x
2 n +1

2 n + 1for all

|
x

|
1 , we havearctan

x
2

n =0

1)
n

x
2(2 n +1)

2 n + 1=

=0

1)
n

x
4 n +2

2 n + 1for

|
x

1 and so

10

arctan x
2

dx =

n =0

1)
n

10

x
4 n +2

2 n +1 dx

n =0


1)
n

1(2 n + 1)(4 n

+ 3) . 7. Dierentiation of Power Series


Lemma

7.1
.

Let

{
a
n

}
and

{
b
n

}
be sequences such that a

0 , lim
n

exists with lim


n

b
n

=0 . Then lim a
n

b
n

= lim

a
n

lim
n

b
n

.
Proof.

Let B = lim
n

b
n

.Given any > 0, there exists N such that

|
b

|
< for

n>N , that is, B

<b
n

<B

+ for n > N. Thus, since a


n


0, a
n

( B

) <a
n

b
n

<a
n

B + ) for n>N and so( B


)lim a
n

= lim a
n

( B

lim a
n

b
n

lim a

( B + )=( B +

)lim a
n

.
84 3. SEQUENCES AND SERIES

OF FUNCTIONS

Now, by letting tend to 0, we have B


lim a
n

lim a

b
n

lim

a
n

. Thuslim a
n

b
n

lim a
n

= lim a
n


lim
n

b
n

Theorem

7.2
.

Suppose that

=0

a
n

( x

x
0

)
n

has radius of convergence R> 0 ,and f

( x )=

=0

a
n

( x

x
0

)
n

|
x

|
< R. Then (a). The power series

n =1

na
n

x
0

)
n 1

has radius of convergence R , and (b). f (

x )=

n =1

na
n

( x

x
0

n 1

for

|
x

x
0

|
<R .
Proof.

(a). By Lemma 7.1,lim

|
na
n

|
1

= lim
n

|
a
n

|
1

n
1

= lim
n

a
n

|
1

lim
n

n lim
n

a
n

|
1

1 = lim
n

|
a
n

|
1

=1

R. Thus the power series

=0

na
n

( x

x
0

)
n

n =1

na
n

( x

x
0

=( x

x
0

n =1

na

( x

x
0

)
n

has radius of convergence R and so has

n =1

na
n

x
0

)
n 1

.(b). For any with 0 <<R , the series of functions

n =1

na
n

x
0

)
n 1

convergesunifor mly on

|
x

x
0

|
by the Uniform Convergence Theorem because the closedinterval [

x
0

,x
0

+ ]


( x
0

R,x
0

+ R ). The result follows from Theorem 5.3.

Remark. The formula

f ( x )=

n =1

na
n

x
0

)
n 1

need not hold at the end points x = x


0

R in general even if the interval of convergence of

n =0

a
n

x
0

)
n

is[ x

R,x
0

+ R ].
Example

7.3
.

From Example 6.15,arctan x =

n =0

1)
n

x
2 n +1

2 n + 1for all

|
x

|
1 .
7. DIFFERENTIA

TION OF POWER SERIES 85

But11 + x
2

= (arctan x )

n =0


1)
n

x
2 n

only holds for

|
x

|
< 1 because when x

1, the right hand side diverges (and theleft hand side =12).

Corollary

7.4
.

Suppose that

=0

a
k

( x

x
0

)
k

has radius of convergence R> 0

with pointwise limiting function f ( x ) on

|
x

x
0

<R (i.e. f ( x )=

k =0

a
k

x
0

)
k

on

|
x

x
0

<R ), then f ( x ) has derivatives of all orders on

|
x

x
0

<R , and for each n , (10) f


( n )

( x )=

= n

k ( k

1)( k


2)

( k

n + 1) a
k

( x

x
0

)
k n

. In particular, (11)

a
k

= f
( k )

( x

) k ! for all k. (i.e. we have f

( x )=

=0

f
( k )

( x
0

k !( x

x
0

)
k

.)
Proof.

The result is obtained by successively applying the previous theorem to

f,f ,f , and etc. Equation

follows by setting x = x
0

in Equation 10, that is,

f
( n )

( x )= n !

a
n

+( n + 1) n

2 a
n +1

( x

)+( n + 2)( n + 1)

3 a
n +2

( x

)
2

.
Example

7.5
.

From the geometric series11

n =0

x
n

=1+ x + x
2

+ x
3

|
x

|
< 1

, we have1(1

x )
2

11

n =1

nx
n 1

=1+2 x +3 x
2

+4 x
3

for

|
x

< 1, and so x (1

x )

n =1

nx

= x +2 x
2

+3 x
3

+4 x
4

|
x

< 1 . By letting x =12, we have

n =1

n 2
n

=
12

12

=2
86 3. SEQUENCES AND SERIES

OF FUNCTIONS Example

7.6
.

Consider the function y =

n =0

x
n

!= 1 + x + x
2

2!+ x
3

3!+

. The radius of convergence R =1lim

|
a
n +1

||
a

|
=1lim
n

n !( n

+ 1)!=1lim
n

1 n + 1=10=

Thus for any x

( ,

), y =0+1+ x

+ x
2

2!+ x
3

3!+

= y =

dydx = y

dyy = dx =


dyy =

dx = ln

|
y

|
= x +

A = y

| |
=

e
x + A

= y =

Ce
x

,C =

e
A

constant

. Let x = 0. C = Ce
0

= y (0) = 1 + 0 + 0 +

=1 .

Hence we obtain the formula e


x

n =0

x
n

!= 1 + x + x
2

2!+ x
3

3!+

.
Definition

7.7
.

A real-valued function

f dened on an open interval I is said to be innitely dierentiable on

I if all (higher) derivatives f


( n )

( x

) , n

1 , exist.The set of innitely

dierentiable functions on I is denoted by C

( I

) . As a consequence, the functions f ( x

)=

n =0

a
n

( x

x
0

)
n

are innitely dierentiableon ( x


0

R,x
0

+ R ) if R> 0.
Corollary

7.8 (Uniqueness Theorem)

Suppose that

n =0

( x

x
0

)
n

and

n =0

b
n

( x

x
0

)
n

are two power series which converge for

|
x

x
0

|
<R with R> 0

. Then

n =0

a
n

( x

x
0

)
n

n =0

b
n

x
0

)
n

for

|
x

x
0

<R if and only if a


k

= b
k

for all

k =0 , 1 , 2 , 3

Proof.

Suppose that a
k

= b

for all k =0 , 1 , 2

, 3

. Then

n =0

a
n

x
0

)
n

n =0

b
n

x
0

)
n

.Conversely, suppose that

n =0

a
n

x
0

)
n

n =0

b
n

x
0

)
n

= f

( x ) for

|
x

x
0

|
<R .Then a
n

f
( n )

( x
0

) n

!and b
n

= f
( n )

x
0

) n !. Thus a
n

b
n

for all n =0 , 1 ,

2 , 3 ,

8 T S 8

. A Y L O R E R I E S 7

8. Taylor Series8.1. History Remarks.

The study of sequences and series of functions has itsorigins in the study of power series representation

of functions. The power seriesof ln(1 + x ) was known to Nicolaus Mercator (1620-1687) by

1668, and the powerseries of many other functions such as arctan x , arcsin x

, and etc, were discoveredarou nd 1670 by James Gregory (1625-1683). All these series were obtained withoutany

reference to calculus. The rst discoveries of Issac Newton (1642-1727), datingback to the early months of

1665, resulted from his ability to express functions interms of power series. His treatise on calculus,

published in 1737, was appropriatelyent itled A treatise of the methods of uxions and innite series

. Among his manyaccomplis hments, Newton derived the power series expansion of (1 +

x )
m/n

usingalgebraic techniques. This series and the geometric series were

crucial in many of his computations. Newton also displayed the power of his calculus by deriving

thepower series expansion of ln(1+ x ) using term-byterm integration of the expansion of 1

/ (1 + x ). Colin Maclaurin (1698-1746) and Brooks Taylor (1685-

1731) were amongthe rst mathematicians to use Newtons calculus in determining the coecients inthe power

series expansion of a function. Both realized that if a function f ( x

) hada power series expansion

n =0

a
n

( x

x
0

, then the coecients a


n

had to be given by f
(

n )

( x
0

) n !.

8.2. Taylor Polynomials and Taylor Series.


Definition

8.1
.

Let f

( x ) be a function dened on an open interval I , and let

x
0

I and n

1 . Suppose that f
( n )

( x )

exists for all x

I . The polynomial T

( f,x
0

)( x )=
n

k =0

f
( k )

( x
0

) k !( x


x
0

)
k

is called the Taylor polynomial

of order n of f at the point x


0

. If f is innitely dierentiable on I , the power series T

( f,x
0

)( x )=

n =0

f
( n )

( x
0

) n !( x


x
0

)
n

is called the Taylor series of

f and x
0

. For the special case x

= 0, the Taylor series of a function f is often referred toas Maclaurin series

. The rst few Taylor polynomials are as follows: T


0

( f,x

)( x )= f ( x
0

) ,T
1

( f,x
0

)( x

)= f ( x
0

)+ f

( x
0

)( x

x
0

) ,T
2

( f,x
0

)( x

)= f ( x
0

)+ f

( x
0

)( x

x
0

)+ f ( x
0

)2!( x


x
0

)
2

,T
3

f,x
0

)( x )= f ( x

)+ f ( x
0

)( x


x
0

)+ f ( x

)2!( x

x
0

)
2

+ f ( x
0

)3!( x


x
0

)
3

. The Taylor polynomial

T
1

( f,x
0

) is the linear approximation

of f at x
0

, that is the tangent line

passing through ( x
0

,f ( x
0

)) with slope f ( x
0

).
88 3. SEQUENCES

AND SERIES OF FUNCTIONS

In general, the Taylor polynomial T


n

of

f is a polynomial of degree less than orequal to n that satises the conditions T

( k ) n

( f,x
0

)( x
0

)= f
( k )

( x
0

)for 0

n . Since f
(

n )

( x
0

) might be zero, T
n

could very well be a polynomial of degree strictly less than n .


Example

8.2

Find the Maclaurin series of e


x

.
Solution.

Let

f ( x )= e
x

. Then f

( n )

( x )= e
x

. Thus

f
( n )

(0) = 1 and so theMaclaurin series of e


x

is1 + x + x
2

2!+

.
Example

8.3
.

Find the Taylor series of f

( x ) = sin x at x
0

.
Solution.

f ( x ) = sin xf

( x ) = cos xf ( x

)=

sin xf ( x

)=

cos x

( )=0 f ( )=


1 f ( )=0

f ( )=1 ,

T
3

( f, )( x )=0


( x

) + 0 +13!(

)
3

.T (

f, )( x )=

( x


)+13!( x

)
3

15!( x

)
5

k =0

1)
k +1

1(2 k + 1)!( x

)
2 k +1

.
Example

8.4

Find the Taylor series of f ( x ) =1 x

at x
0

=3 .
Proof.

1 x

=13

(3

x )=13


11

3 x 3

=13

n =0

x 3

=13

n =0

1)
n

3
n

( x

3)

n =0


1)
n

3
n +1

( x


3)
n

. Thus the Taylor series of 1 x

at x
0

= 3 is

n =0

1)
n

3
n +1

( x

3)
n

8 T S 8

. A Y L O R E R I E S 9

8.3. Taylor Theorem. In view of Theorem 7.4, we may ask the

following question: Question: Given a innitely dierentiable function f

( x ), does the equality(12) f ( x )=

T ( f,x
0

)( x )=

k =0

f
( k )

( x
0

) k !( x


x
0

)
k

hold for

x
0

|
<R ?(Here

R is the radius of the convergence of the Taylor series.)It turns out that in general, the answer is NO.

(See Example 8.5 for an example of afunction such that equation 12 does not hold.)However, as we have

seen, the above equality does hold for some elementaryfunct ions such as e
x

sin x , cos x , ln(1 + x ), (1 + x

)
a

, arctan x , and etc. We are goingto give certain hypothesis such

that the above equality holds for some functions.


Example

8.5 (Counterexample to the Question)

Consider the function f ( x )=

x
2

x =0 , 0

x =0 . Then we will show that f ( x

) = its Taylor series at x


0

= 0.

Proof.

First we compute f (0). By substituting y

=1 x
2

f (0) = lim
x 0

f ( x )

f (0)

0= lim
x 0

x
2

0 x = lim
x 0

1 xe
1

x
2

= lim
x 0

x
2

e
1

x
2

x = lim

ye
y

lim
x 0

x =0

0 (by LHopitals rule) = 0 .

For x

= 0, f (

x )= ddx

x
2

=2 x
3

x
2

. Thus, f (

x )=

x
3

x
2

=0 , 0 x =0 .

Next we compute f ( x ). By substituting

y =1 x
2

, f (0) = lim
x

f ( x )

f (0) x

0= lim
x

2 x
3

x
2


0 x = 2 lim
x 0

x
4

e
1

x
2

= 2 lim
y

y
2

e
y

= 0 (by LHopitals rule).Again, for x

= 0, f ( x )= ddx

2 x
3

x
2

)=(

6 x
4

+4 x
6

) e

x
2

. Similar calculations will lead to f (0) = f

(0) = f (0) = f
(3)

(0) = f
(4)

(0) =

=0 .
90 3. SEQUENCES AND SERIES

OF FUNCTIONS

Thus we have

k =0

f
( k )

(0) k ! x
k

k =0

k ! x
k

=0+0 x +0 x

=0 . Clearly, at any x

= 0, f ( x )= e

x
2

= 0. Therefore, f ( x

) =

k =0

f
( k )

(0) k !

x
k

. The remainder or error function

between f ( x ) and T
n

f,x
0

) is dened by R
n

( f,x
0

)(

x )= f ( x )

T
n

( f,x
0

)( x )

. Clearly f ( x ) = lim
n

T
n

( f,x
0

)( x ) = lim
n

k =0

f
(

k )

( x
0

) k !( x


x
0

)
k

k =0

f
( k )

( x
0

) k !( x


x
0

)
k

if and only if lim


n

R
n

( f,x
0

)( x )=0

. To emphasize this fact, we state it as a theorem


Theorem

8.6
.

Suppose that f is an innitely dierentiable function on an open interval I and

x
0

I . Then, for x

I , f ( x )=

n =0

f
( n )

( x
0

) n !( x


x
0

)
n

if and only if lim


n

R
n

( f,x
0

)( x )=0

. The remainder R
n

( f,x
0

) has been studied much and there are various forms of R


n

f,x
0

). We only provide one result called Lagrange Form of the Remainder

,attributed by Joseph Lagrange (1736-1813). But this result sometimes also referredto as

Taylors theorem.
Theorem

8.7 (Taylor Theorem)


.

Let f

be a function on an open interval I , x


0

I and n

. If f

( n +1)

( t ) exists for every t


I , then for any x

, there exists a between x


0

and x such that

(13) R
n

( f,x
0

)( x

)= f
( n +1)

( )( n

+ 1)!( x

x
0

)
n +1

. Thus f ( x )= f (

x
0

)+ f ( x
0

)(

x
0

)+ f (

x
0

)2!( x

x
0

+ f
( n )

( x
0

) n !( x


x
0

)
n

+ f
(

n +1)

( )( n + 1)!( x


x
0

)
n +1

, for some

between x and x
0

.
Proof.

Recall that R
n

( f,x
0

)= f

( x )

k =0

f
( k )

( x
0

) k !( x


x
0

)
k

.
8 . T A Y L O R

S E R I E S 9 1

Fixed x

I , let

M be dened by R
n

( f,x
0

)=

M ( x

x
0

)
n

+1

, that is, f ( x )=
n

k =0

f
( k )

( x
0

) k !( x


x
0

)
k

+ M (

x
0

)
n +1

( Note. M depends on x .) Our goal is to show that M

= f
n +1

( )( n

+ 1)!for some between x


0

and x

.We construct a function g ( t )= f (

t )

k =0

f
( k )

( x
0

k !( t

x
0

)
k


M ( t

x
0

)
n +1

= f ( t )


f ( x

)+ f ( x
0

)( t


x
0

)+

+ f

( n )

( x
0

) n !(

x
0

)
n


M ( t


x
0

)
n +1

By taking derivatives, we have g ( x


0

)=

g ( x
0

)= g (

x
0

)=

= g
( n

( x
0

)=0 . and(14) g
(

n +1)

( t )= f
( n +1)

( t )

( n + 1)!

M. For convenience, lets assume x>x


0

. By the choice of

M , g ( x ) = 0. By applyingthe

mean value theorem to g on the interval [ x


0

,x ], there exists

c
1

, x
0

<c
1

<x , suchthat0 =

g ( x )

g (

x
0

)= g ( c
1

)(

x
0

)= g

( c
1

)=0 . Since g

( x
0

)= g ( c
1

) = 0, by applying the mean value theorem to g on the interval[ x

,c
1

], there exists c
2

, x
0

<c
2

<c
1

, such that0 = g ( c

g ( x
0

)= g ( c
2

)( c
1


x
0

)= g

( c
2

)=0 . Continuing this manner, we obtain points

c
1

,c
2

,c
n

, x
0

<c
n

<c
n 1

<

<c
2

<c
1

<x , such that

g ( c
1

) = 0, g (

c
2

) = 0, g ( c
3

) = 0,


, g
( n )

( c

) = 0. By applyingthe mean value theorem once more to g


( n

on [ x
0

,c
n

], there exists ,

x
0

< < c
n

,such that0 = g
( n )

( c
n

g
( n

( x
0

)= g
( n +1)

( )( c
n

x
0

)= g
( n +1)

)=0 . From Equation (14),0 = g


( n +1)

)= f
( n +1)

( )


( n + 1)! M, that is, M

= f
( n +1)

( )( n

+ 1)!for some between x


0

and x (because

x
0

<<c
n

<x ).
Example

8.8

Show that sin x =

n =0

1)
n

x
2 n +1

(2 n + 1)!=

x
3

3!+ x
5

5!


x
7

7!+

|
x

|
<

.
92 3. SEQUENCES AND SERIES OF FUNCTIONS Proof.

First the right hand side is the

Maclaurin series of f ( x ) = sin x because

f ( x ) = sin xf ( x

) = cos xf ( x )=

sin xf ( x )=

cos x

f (0) = 0 f

(0) = 1 f (0) = 0 f (0) =


1 ,

Next let a
n

1)
n

x
2 n +1

(2 n + 1)!

. By the ratio testlim


n

a
n +1

a
n

= lim
n


1)
n +1

x
2 n +3

(2

n + 3)!

1)
n

x
2

n
+1

(2 n +1)!

= lim

x
2

(2 n + 3)(2 n + 2)= 0

< 1for all x . Thus R =+

.In the last step we show that the remainder tends to 0. Since f ( x

) = sin x , thehigher derivatives f


( n +1)

x ) is either

sin x or


cos x . Thus

|
f
(

n +1)

( x )

|
1 for all x

and all n . By the Taylor Theorem,

|
R
n

( f, 0)( x )

|
=

( n +1)

( )( n + 1) x

n +1

|
x

|
n +1

( n

+ 1)! . Since lim


n

|
x

|
n +1

( n + 1)!= 0, we have lim


n

|
R
n

( f, 0)( x

|
= 0, by the Squeeze Theorem, orlim
n

( f, 0)( x ) = 0 for all x . Hencesin

x =

n =0

1)
n

x
2 n +1

(2 n + 1)!= x

x
3

3!+ x
5

5!

x
7

7!+

|
x

|
<


) .

8.4. Some Standard Power Series.

Below is a list of Maclaurin series of someelementary functions. e


x

n =0

x
n

!= 1 + x + x
2

2!+ x
3

3!+

|
x

<

)sin x =

n =0

1)
n

x
2 n +1

(2 n + 1)!=

x
3

3!+ x
5

5!


x
7

7!+

|
x

|
<

. cos x =

n =0

1)
n

x
2 n

(2 n )!= 1

x
2

2!+

x
4

4!

x
6

6!+

|
x

|
<


) . ln(1 + x )=

n =1

1)
n +1

x
n

n = x


x
2

2+ x
3

1 <x

1) . 11

x =

n =0

x
n

=1+

x + x
2

+ x
3

|
x

|
<

1) . 11 + x =

n =0

1)
n

x
n

=1

x +

x
2

x
3

|
x

|
< 1)

8 T S 9

. A Y L O R E R I E S 3

arctan x =

=0

1)
n

x
2 n +1

2 n + 1= x

x
3

3+ x
5

1)(1 + x

)
a

n =0

an

x
n

=1+

a 1

x +

a 2

x
2

a 3

x
3

|
x

|
< 1)

, where

ak

= a

1)

( a


2)

( a

k + 1) k !for any real number a and integers k


1, and

a 0

=1 Remark. From these power series,

we can obtain Maclaurin series of various morecomplicate d functions by using operations such as substitution,

addition, subtraction,mult iplication, division, integrals, derivatives and etc. For the

Maclaurin series of sin x


2

can be obtained by replacing x by

x
2

in the Maclaurin series of sin x . By usingmultiplicat ion, we can

obtain the Maclaurin series of e


x

sin

x . By using long division,we can obtain the Maclaurin series of tan x =sin

x cos x . By taking integral, we can obtainthe Maclaurin series of non-

elementary functions, for instance f ( x )=

x 0

sin t
2

dt .As an application, we are going to compute number .

Computation of Step 1. Find the Maclaurin series of arcsin x for

|
x

|
< 1 .

For

|
x

|
< 1,arcsin

x =

x 0

t
2

dt =

x 0

1+(

t
2


12

dt =

x 0

k =0


12


t
2

)
k

dt =

k =0

x 0

12

1)
k

t
2 k

dt =

k =0

1)
k

12

x
2 k +1

2 k + 1= x +

k =1

1)
k

12

x
2 k +1

2 k + 1Note that

12


12


12


12

k +1

k !=


12


32


2 k


12

k != (


1)
k


(2 k

1) k

2
k

for k

1. Thus(15)arcsin x = x +

k =1

1)
k

1)
k

(2 k

1)2
k

k !

(2

k + 1) x
2 k +1

= x +

k =1

(2 k

1)2
k


k !

(2 k

+ 1) x
2 k +1

for

|
x

|
< 1. Step 2. Find a series expansion of

6 using 6= arcsin12 .

94 3. SEQUENCES AND SERIES OF FUNCTIONS

From Equation (15), we obtain the following formula.(16) 6=12+

k =1

(2 k

1)

k !

(2 k + 1)


2
3 k +1

. Step 3. Estimate the remainder.

The remainder of the formula 16 can be estimated as follows. R


n

|
S

S
n

k = n +1

(2 k


1)2
k

k !


(2 k + 1)

2
2

k +1

<

k =

n +1

1(2 k + 1)2
2 k +1

<

k = n +1

1(2 n

+ 3)2
2 k +1

=1(2 n + 3)2
2 n +3

1 +14+14
2

=1(2 n + 3)2
2 n

+3

14

=13(2 n + 3)2
2 n

+1

For instance, let n = 10, we have

12+
10

k =1

(1

(2 k

1)

k !

(2 k + 1)


2
3 k +1

with error less than6

13

23


2
21

=123

2
20

=124117248 . If we choose n = 20, we have

12+
20

k =1

(2 k


1)!! k !

(2

k + 1)

2
3 k +1

with error less than6

13


43

2
41

=143


2
40

=147278999994 368 < 10

13

. If n = 40, the error is less than1100340843

0280142215006 12608 < 10


26

. Remark.

There are several other methods for computing . For instance, we canalso use

4= arctan1 = 1

13+15

17+

, but one needs a huge number of terms to get enough accuracy. (So this method isno good for

computational purpose!) Another method is to use the formula of JohnMachin (16801751):4arctan15


arctan1239= 4 , see our text book [

1 , Problem 7, p.813] for details. Machin used his method in 1706to nd

correct to 100 decimal places. In 1995 Jonathan and Peter Borwein of SimonFraster University and Yasumasa

Kanada of the University of Tokyo calculated thevalue of to 4 ,

294 , 967 , 286 decimal places!

8 . T A Y L O R

S E R I E S 9 5

Another story on computing is a Chinese mathematician and astronomer

TsuChung Chi (430-501). He gave the rational approximation3 55113to

which is correctto 6 decimal places. He also proved that3 . 1415926 <<

3 . 1415927a remarkable result ( Note. He was a person lived 1500 years

ago!), on which it wouldbe nice to have more details but Tsu Chung Chis book, written with his son, is lost.(His

method is to cut o the circle by equal pieces to get his approximation to

.)Tsus astronomical achievements include the making of a new calendar in 463 whichnever came into use.

(According to the article of J J OConnor and EF Robertsonin http://wwwgroups.dcs.stand.ac.uk/

history/Mathema ticians/Tsu.html. ) Remark. Those, who are interested in more applications of

Taylor series, can try tonish the applied project, Radiation from the Stars , in our text book [

1 , pp.808-809].

Bibliography
[1] James Stewart,
Calculus

, 4th Edition, Brooks/Cole Publishing

Company press, 1999. (


TextBook

)[2] Manfred Stoll,


Introduction to Real Analysis

, 2nd Edition, Addison Wesley Longman, Inc.press, 2001. (

Text Book

)[3] Watson Fulks,


Advanced Calculus

, 3rd Edition, John Wiley & Sons, Inc. press, 1961.[4] Wilfred Kaplan,
Advanced Calculus

, 3rd Edition, Addison-Wesley

Publishing Companypress, 1984.[5] William R. Parzynski and Philip W. Zipse,


Introduction to mathematical analysis

, Inter-national Edition, McGraw-

Hill Book Company press,1987.[6] G. B. Thomas, Jr. and Ross L. Finney,


Calculus and analytic geometry

, 9th Edition,Internation al Student Edition,

Addison-Wesley Longman Inc. press, 1996.


97
Search Search History: Searching... Result 00 of 00 00 results for result for p.