NOTES ON

REAL VALUED FUNCTIONS OF A REAL VARIABLE

Sanjeev Agrawal

03: SEQUENCES 1
Limit of a sequence

Notation:
The set of real numbers - R;
The set of rational numbers - Q;
The set of natural numbers - N;
The set of integers - Z;
The set of complex numbers - C.

Given:
is a/the (order) complete (Archimedean) ordered field.

Convergence of sequences is central to the concept of ‘limits’. In this note, our aim is to
introduce and list some of the most important properties of real sequences.

and will both denote ‘strictly less than’.

................................................

1
 A Definitions:

Any function whose domain is N is called a sequence.

If the co-domain of the sequence is the set X, we call the sequence ‘a X sequence’ or ‘a sequence
in/of/from X’. For this note X R.
....
A function is completely determined by its domain, co-domain and its values.
As the domain and the co-domain of a real sequence are known ( N and R, respectively), it is
enough to explicitly give its values.
The notation usually used is a n (or a 1 , a 2 , . . . or a n ) where a n is the value that the sequence
takes at the natural number n.
....
A01 Bounded Sequences:
a n is said to be a bounded above sequence if and only if there exists a real number :
an , n N.
a n is said to be a bounded below sequence if and only if there exists a real number :
a n , n N.
a n is said to be a bounded sequence if and only if it is bounded above as well as below.
Then there there exist real numbers , :
an , n N.
Equivalently, there exists a real number M:
|a n | M, n N.
....
A02 Monotone Sequences:
a n is said to be a monotone increasing (or nondecreasing or increasing) sequence if and only if
a n a n 1 , n N.
a n is said to be a strictly monotone increasing (or strictly increasing) sequence if and only if
a n a n 1 , n N.
a n is said to be a monotone decreasing (or nonincreasing or decreasing) sequence if and only if
a n a n 1 , n N.
a n is said to be a strictly monotone decreasing (or strictly decreasing) sequence if and only if
a n a n 1 , n N.
a n is said to be a monotone sequence if it is either an increasing or a decreasing sequence.

a n is said to be a strictly monotone sequence if it is either a strictly increasing or a strictly

decreasing sequence.
....
A03 ‘Ultimately’:
A property P n , defined for natural numbers n and related to the sequence a n , is said to hold
ultimately for the sequence a n if there exists M : a n satisfies P n for all n M.

For example, 3, 2, 1, 2, 0, 0, 0, 0, . . . is ultimately constant.

2
 x n will denote the constant sequence where every entry (member/value) of the sequence is x.
....
We will use A, B, C, . . . to denote sequences. These are real valued functions with the set of
natural numbers as their domain. A a n denotes the function A n a n , n N.
....
A04 Subsequences:
Let A a n be a real sequence and N n k be a strictly increasing sequence of natural
numbers.
Then B A N is is called a subsequence of a n .

Thus, if a n is a real sequence and n k is a strictly increasing sequence of natural numbers, then
the sequence b k is called a subsequence of a n if b k a n k for each k.
....
A05 Note:

A05.01: We write a n k for b k .

A05.02: n k k, k.

A05.03: A subsequence of a n is a sequence obtained from it by ‘dropping’ some entries and

‘keeping the order in which the terms appear’.

A05.04: If n k k 1, the sequence is called the k-tail of a n .

....
A06 Cauchy Sequences:
a n is said to be a Cauchy sequence if and only if
Given 0, M : n, m M |a n a m | .
....
Some equivalent definitions are: a n is said to be a Cauchy sequence if and only if

Given 0, M : n m M |a n am | .

Given 0, M : n M&p 0 |a n p an | .

Given 0, M : n M |a n aM | .
....

3
 A07 Theorem:
If a n is a Cauchy sequence, then it is bounded.

Proof: Let a n be a Cauchy sequence.

Then M : n M |a n aM | 1.

Then n M |a n | |a n aM aM | |a n aM | |a M | 1 |a M |.

Thus |a n | max |a 1 |, |a 2 |, . . . , |a M |, 1 |a M | .
....
A08 The set of all sequences:
Let S denote the set of all real sequences.
A an , B bn , C c n , . . . are members of S.

A08.01: The following binary operations on S are well-defined:

A B : C, where c n a n b n , n N.

A. B : C, where c n an. bn, n N.

We also have a ‘scalar multiplication’ on S:

.A : an ; R, A S

For those familiar with linear algebra, it is trivial to see that S is a commutative algebra with
unity over R. Note:
A 1. A an ; A B an bn
....
A08.02: If a n 0, n, then 1 : 1
an .
A

Let A be an ultimately non-zero sequence.

Then there exists K : n K an 0.

Then it is customary to define 1 as

A
1 n K
1 :
A
1 n K
an
....

4
A09 Examples:
Some examples of real sequences:

an 0, n N.

1 n even
an
1 n odd

(Recursively defined) a 1 0, a 2 1; a n 1 an an 1, n 2 (Fibonacci).

a n is the natural number in the n th place (after the decimal) in the decimal expansion of .
....
Note: Given an ordered field F, we can replicate all the definitions and the results given above.
....

5
 Exercises N031.01:

(i) Show: a n is a sequence in an ordered field F. Then a n has a monotone subsequence.

(ii) Show: A is a monotone increasing sequence (in an ordered field F) if and only if A is a
monotone decreasing sequence.

(iii) Show: If A and B are Cauchy sequences (in an ordered field F), then A B is also a Cauchy
sequence.

(iv) Show: If A and B are Cauchy sequences (in an ordered field F), then A. B is also a Cauchy
sequence.

(v) Show: If A is a Cauchy sequence (in an ordered field F) & F, then . A is also a Cauchy
sequence.

(vi) Show: If A and B are momotone increasing sequences (in an ordered field F), then A B is
also a momotone increasing sequence.

(vii) Show: If A and B are momotone increasing sequences of non-negative members (in an
ordered field F), then A. B is also a momotone increasing sequence.

(viii) Show: If A is a Cauchy sequence (in an ordered field F) & B is a subsequence, then B is
also a Cauchy sequence.
................................................

6
B Limit of a sequence

B01 Limit:
Let a n be a real sequence.
We say, l is a limit of the sequence a n , if and only if
Given 0, M : n M |a n l| .
....
B02 Equivalent Definitions:
Some of the equivalent definitions of a limit of a sequence are:

l is a limit of the sequence a n , if and only if

B02.01: Given 0, M : n M l an l .
As, |a n l| l an l .

B02.02: Given 0, M : n M an l ,l .
As, |a n l| an l ,l .

B02.03: k 0 : Given 0, M : n M |a n l| k .
As, k : 0 : 0 .

B02.04: , 0 : Given 0, M 1 , M 2 : n M 1 l an & n M2 an l .

Take M max M 1 , M 2 & max , . Then n M |a n l| .

B02.05: Given 0, M : n M l an l .
‘Greater than’ implies ‘Greater than or equal to’. Other way, work with /2.

B02.06: A similar change in all other definitions will still imply that l is a limit of the sequence.
For Example, this works - Given 0, M : n M |a n l| .
.................................................
B03 Theorem - Limit is unique:
If l & m are limits of the sequence a n , then l m.

Proof: Let if possible l m. Let 3 l m. Then 0.

As l is a limit of the sequence a n , M 1 : n M1 |a n l| .

As m is a limit of the sequence a n , M 2 : n M2 |a n m| .

Let M max M 1 , M 2 .

Then |a M l| & |a M m| .

7
 But then,
3 l m |l m| |l aM aM m| |l aM | |a M m| 2 .
Contradiction.

.................................................
Theorem B03 states that - if a sequence has a limit, then it has only one (a unique) limit.
Thus the phrase ‘l is a limit of the sequence’ is generally written as ‘l is the limit of the
sequence’.

If the sequence a n has a limit, we say ‘the sequence a n converges’ or ‘ a n is a convergent

sequence’. If l is the limit of the sequence a n , we say ‘ a n converges to l’.

If the sequence a n does not have a limit, we say ‘the sequence a n diverges’ or ‘ a n is a
divergent sequence’.

Notation: All of the following notations are standard for the expression - l is the limit of the
sequence a n :
lim
n
a n l; lim a n l; a n l.
.....
A sequence is called a null sequence if its limit is zero.
.................................................
B04 Theorem - Convergent implies bounded:
If a n is a convergent sequence, then a n is a bounded sequence.

Proof: Let a n l.

Then M : n M |a n l| 1.

Thus |a n | |a n l l| |a n l| |l| l 1, n M.

But then |a n | max |a 1 |, |a 2 |, . . . , |a M |, l 1 , n.

.................................................
B05:
The converse of Theorem B04 is not true.
That is, there exist bounded sequences that do not converge.
Such a sequence is given in Example A09:
1 n even
an
1 n odd
We can show that this sequence does not converge as follows:

Let, if possible, a n l. Then M : n M |a n l| 1.

3

8
 Then
2 |1 1 | |a 2M a 2M 1 |
|a 2M l l a 2M 1 |
|a 2M l| |l a 2M 1 |
2.
3
Contradiction.
.................................................
B06 Theorem:
The sequence a n converges (to l) if and only if every subsequence of a n converges (to l).

Proof: As every sequence is a subsequence of itself, the fact that every subsequence converges
implies that the sequence converges.

Conversely let a n l and a n k be a subsequence of a n .

Let 0.
As a n l, M : n M |a n l| .
Now if k M, then as n k k (Note C02.02), we get |a n k l| .
Thus a n k l.
.................................................
B05 (Revisit):
1, 1, 1, 1, 1, . . . is not convergent as it has two subsequences a 2n & a 2n 1 converging to
different numbers, 1 & 1, respectively.
.................................................
B07 Theorem:
If a n is a convergent sequence, then a n is a Cauchy sequence.

Proof: Let a n l. Let 0. As a n l,

M:n M |a n l| .
2

Then, n, m M, we have
|a n am | |a n l l am |
|a n l| |l am |
................................................
B08 Theorem:
If a n is a ultimately a constant sequence (where a n k, n K), then it converges (to k).

Proof: Let 0. As a n k, n K, we have

n K |a n k| 0 .
.................................................

9
 B09 Theorem:
The sequence 1 is a null sequence.
n

Proof: Let 0.

As R is an Archimedean ordered field, there exists a natural number N such that N 1.

Then n N n 1.

Thus 1n 1
n
1
n 0 whenever n N.

Thus 1n 0.
........................................
B10 Note:

B10.01: As stated in earlier notes, 1n 0 is an equivalent characterization for an ordered field

to be Archimedean. The proof is almost immediate.

B10.02 (Exercise N031.02): As 1 is a monotone decreasing sequence, its limit, if it exists is

n
its infimum.

B10.03: As R is complete, every convergent sequence being bounded, has an infimum and a
supremum.

B10.04: But every bounded sequence is not convergent. So to guarantee convergence, a sequence
has to be a ‘little more’ than just being bounded.

B10.05: In the next section, we show that any bounded monotone sequence in R is convergent.
That is R has the MCT property. We note that ordered field F has the MCT property if and only if it
is a complete ordered field.

B10.06: In a later section, we show that a Cauchy sequence in R is convergent. This is that R is
Archimedean and has the CCC property. This is another equivalent characterization for an ordered
field to be a complete ordered field.

B10.07: In another section we show that R has the Bolzano - Weierstrass Property for
Sequences, that is, every bounded sequence has a convergent subsequence. Another equivalent
characterization for an ordered field to be a complete ordered field.

B10.08: We also indicate how the properties mentioned in B10.05 TO B10.07 are equivalent in
any ordered field.
................................................

10
 C R has the MCT property

C01 Theorem:
Let a n be a monotone increasing sequence in R.
Then, if a n is bounded above, then a n l where l sup a n .
Proof:
Let a n be bounded above.

As R is complete, sup a n exists. Let it be l.

We show a n l.

Let 0.

As l l, and l sup a n , l is not an upper bound of a n .

Thus there exists N : a N l .

Then n N
l a n as l sup a n
a N (as a n is a monotone increasing sequence)
l .
Hence a n l.
....
C02 Theorem:
Let a n be a monotone decreasing sequence. Then, if a n is bounded below, then a n l where
l inf a n .
Proof: Exactly similar to proof of Theorem C01.
....
C03 Theorem - R has MCT:
Every bounded monotone sequence in R converges.
....
C04 Note:

C04.01: We have already seen that convergent sequences are bounded but the converse is not
true.
For monotone sequences in R, (or in any complete ordered field) convergence is
equivalent to boundedness.

11
 C04.02: The existence of supremum (& infimum) is very crucial in the proofs of Theorems C01
& C02; and these follow from completeness of the ordered field R.

C04.03: Recall Exercise N011.03: Show that if MCT holds in an ordered field F, then F is
Archimedean.

C04.04: We shall see in a later section that if we assume that F ordered field where Theorem
C01 (and hence Theorems C02 (using algebra of limits) & C03) is true), then F is a complete
ordered field (has the SUP property).
................................................

12
 D R has the BWP for sequences

D01 Theorem (Exercise N031.01(i)):

a n is a sequence in R. Then a n has a monotone subsequence.
....
D02 Theorem
R has the Bolzano - Weierstrass Property for sequences.
That is,
Every bounded sequence in R has a convergent subsequence.

Proof: Follows from Theorems D01 & C03.

....
D03 Note:
Our proofs show the following:
Let F be an ordered field. Then

F has the SUP property

F has the MCT property

F has the BWP for sequences.

....
D04 Note: Recall Exercise N011.04: Show that if BWT SEQS holds in an ordered field F, then
F is Archimedean.
....
D05 Theorem:
Let F be an ordered field. Then, F has the BWP SEQS implies F is has the MCT property.
Proof: This follows from the following argument:
Let F be an ordered field.
Let F have the BWP property for sequences.
Let a n be a monotone increasing bounded above sequence in F.

Exercise N031.03: Show: a n is a monotone sequence in an ordered field F. If a n has a

convergent subsequence, then it is convergent.

As a n is a bounded sequence and F has BWP property for sequences, a n has a convergent
subsequence.
We are through.
....
D06 Note: The Bolzano - Weierstrass Theorem for sequences also guarantees the existence of
limit superior and limit inferior of sequences. Equivalent definitions and uses of limit superior and
limit inferior will be the matter for another set of notes.
................................................

13
 EA R is Archimedean and has the CCC (Cauchy Convergence Criterion)

EA01 Theorem (Theorem A07):

If a n is a Cauchy sequence in R, then it is bounded.
....
EA02 Theorem
If a n is a Cauchy sequence, then it converges if it has a convergent subsequence.

Proof: Exercise N031.04

....
EA03 Theorem
R has the CCC Property.
That is,
(Convergence of Cauchy Sequences) Every Cauchy sequence of real numbers converges.

Proof: Follows from Theorems EA01, EA02 & D02.

....
Aliter for Theorem EA03:
Proof: Follows from Theorems EA01, EA02 & C03.
....
Another proof is available in the note on limit superior and limit inferior.
....
EA04 Note:
Our proofs show the following:
Let F be an ordered field. Then

F has the SUP property

F has the MCT property

F has the BWP for sequences

SecF is Archimedean and has the CCC property

....
EA05 Note:
In this note, we shall be looking at one more characterization of completeness of an ordered field
- namely, the Cantor’s Intersection Property or CIP, in short. We shall do so in Section EB after we
have developed some tools for studying convergent sequences.

14
 We shall also indicate how to show the equivalence of these characterizations by completing the
cycle - we will outline a proof of - If the ordered field F is Archimedean and has the CIP, then F has
the SUP property.

We have shown that if F has any one of the four properties - SUP, INF, MCT or BWP SEQS,
then F is Archimedean.

But we cant drop this condition for the remaining two characterizations.

There exist non - Archimedean ordered fields which have CCC.

................................................

15
F Algebra of limits

F01 Theorem: Let a n l & l 0.

(a) Then M : a n 0, n M.
(b) If l 0, then M : a n 0, n M.
(c) If l 0, then M : a n 0, n M.
(d) If l 0, then a 0 & M : an a, n M.
(e) If l 0, then a 0 & M : an a, n M

Proof: Obviously, if (e) is proved, (d) will have a similar proof; and then all cases are done.

So let a n l&l 0.

Let l . Then 0.
2

As a n l, M : n M |a n l| .

Then l an l , n M.

Thus, n M an l l l l.
2 2

Take a l 0.
2
....
F02 Observations:

Theorem F01(a) says that if a sequence has a non-zero limit, then it is ultimately non-zero.

Theorem F01(b) says that if a sequence has a positive limit, then it is ultimately positive.

Theorem F01(c) says that if a sequence has a negative limit, then it is ultimately negative.

Theorem F01(d) says that if a sequence has a positive limit,

then it is ultimately positive and bounded away from zero.

Theorem F01(e) says that if a sequence has a negative limit,

then it is ultimately negative and bounded away from zero.

16
 Thus, if the limit of a sequence a n is non-zero, then we can talk of the sequence 1
a n , in the
sense of Note A08.02.

Theorem F01(d & e) imply that if a sequence has a non-zero limit, then
a 0 & M : |a n | a, n M.

Thus if a convergent sequence has infinitely many non-negative and infinitely many non-positive
terms, then it is a null sequence.
....
F03 Theorem - Algebra of limits:
Let A an , B b n , a n l, b n m, & R. Then,

(a) a n l.

(b) a n bn l m.

(c) a n b n lm.

(d) a n l
m if m 0.
bn

Proof: Let 0.

(a) As a n l, M : n M |a n l|
| | 1

Then n M
| an l| | ||a n l| | | .
| | 1
Thus a n l.
....
(b) As a n l, b n m, M :
n M |a n l| & |b n m|

Then n M
| an bn l m | |a n l| |b n m| 2 .
Thus, a n bn l m.
.....

17
(c) By Theorem B04, b n is a bounded sequence - let |b n | K, n.

As a n l, b n m, M : n M
|a n l| & |b n m|
2K 1 2|l| 1

Then n M
|a n b n lm| |a n b n lb n lb n lm|
|a n b n lb n | |lb n lm|
|a n l||b n | |b n m||l|
K|a n l| |b n m||l|
K |l|
2K 1 2|l| 1
.
Thus a n b n lm
.....
(d) We show 1 1
m if m 0. Then part (c) gives the required result.
bn

From Observations F02, b n is ultimately non-zero and bounded away from zero.
It also gives that a 0 & M : |b n | a, n M.

Then n M 1 1.
|b n | a

As b n m, M : n M |b n m| a|m|. (note that a|m| 0).

Then n max M, M
1 1 m bn
bn m mb n
|b n m|
|m||b n |
|b n m|
as n M
|m|a
a|m|
as n M
|m|a

Thus 1 1
m.
bn
....

18
F04 Theorem - Order on limits:
Let a n l & b n m. Then

(a) If a n is ultimately non-negative, then l 0.

(b) If ultimately, a n b n , then l m.

(c) If ultimately, a n , then l .

(d) If ultimately, a n , then l .

Proof: This is easy -

(a) follows immediately from Theorem F01(c).

For (b), apply Part (a) and the algebra of limits to b n an.

For (c), apply Part (b) taking b n to be the constant sequence n.

(d) is exactly similar to Part (c).

..............................................

19
 G The Sandwich Theorem

Arguably the most important result - it is the basis of almost all tests of convergence as well
as a handy tool in calculating limits.

G01 Theorem - Sandwich/Squeeze Theorem:

First Version: If k, K :
0 b n ka n , n K, and a n 0, then b n 0.
Second Version: If K :
c n b n a n , n K, and a n , cn , then b n .

Proof:
First Version: Let 0.

As a n 0, M : n M |a n | |a n 0|
|k| 1

Then n max M, K
|b n 0| |b n |
b n as n K bn 0
ka n as n K
|ka n 0|
|k||a n |
|k| as n M
|k| 1
.
Thus b n 0.

Second Version: Use the algebra of limits and apply the first version to the inequality
0 bn cn an cn, n K
....
G02 Note:

G02.01: If a n , then |a n | | |. This follows from the triangle inequality:

||a n | | || |a n |.

G02.02: The converse of the statement given above is false. Look at 1, 1, 1, 1, . . .

G02.03: The converse is true if 0, i.e., |a n | 0 an 0. This follows from |x| ||x||.

G02.04: Thus, a n an 0 |a n | 0.

20
 G02.05: If a n , & a n 0, then a n .
It is clear that ‘a n 0’ can be replaced by ‘ a n is ultimately non-negative’.

0, by Theorem F04(a).

Let 0.

First we take the case 0.

2
As a n 0 & an 0, M : n M an .

2
Then a n , n M.

So a n 0.

If 0, then 0, K : a n ; n K (exist as the limit is positive).

As a n , M:n M |a n | .

Then n max M, K
|a n | |a n |
| an |
an

G02.06: From a n & a n 0, all we can conclude is that 0 (and not 0). Look at the
sequence n . 1
Thus ‘after taking limits, strict inequalities may change to non-strict inequality’.
....
Exercise N031.05: A slightly different version of Theorem B06. Show
The sequence a n converges if and only if every subsequence of a n converges.
................................................

21
 EB R is Archimedean and has the CIP (Cantor’s Intersection Property)

EB01 Theorem CIP - Cantor’s Intersection Property:

Let
In an, bn : In I n 1 , &b n a n 0
then I n is a singleton.
Proof 1: Using R has the MCT property -
As I n In 1,
an an 1 & bn bn 1; n
&
a n b m ; n, m.

As R has the MCT property a n & b n both converge, to their supremum (a) and infimum (b),
respectively.
So
a n a b b m ; n, m.
Now
bn an 0 b a
Thus
a an, bn ; n
Also
x y, x, y In 0 y x bn an; n
which implies
y x
Thus I n is a singleton.
....
Proof 2: Using R is Archimedean and has the CCC -

This proof is clearer once the topology of R is understood.

The part of the proof that says I n is at most a singleton is as above.

To show that it contains at least one point, we argue -

Let x n I n an, bn .
(For example, take the mid-point in each case)

As I n I n 1 , we have - For each m,

xn Im; n m.

22
Thus
|x n xm | bk a k ; n, m k.
bn an 0 x n is a Cauchy sequence.

As R has the CCC, x n , and every tail of it, converges. Let the limit be x.
As each I k is a closed set, and the k-tail of x n lies in it, x I k for each k.
....
EB02 Note:
Our proofs show the following:
Let F be an ordered field. Then

F has the SUP property

F has the MCT property

F has the BWP SEQS

F is Archimedean and has the CCC property

F is Archimedean and has the CIP.

....
Exercise N031.06: Let F be an ordered field. Then show that

F is Archimedean and has the CIP

F has the SUP property

Hint: Try the following argument:

Start with (WLOG, why?) an infinite bounded above set .

Construct a sequence
In an, bn : In In 1
such that b n is an upper bound of the set
and a n is not.
I n 1 could be ‘half’ of I n .
Use the fact that F is Archimedean to assert that b n a n 0.
Use the fact that F has CIP to show that the supremum of the set exists.
....

EB03 Note:
The cycle of proof is now complete.

23
 We have shown that -

Let F be an ordered field. Then

F has the SUP property

F has the INF property

F has the MCT property

F has the BWP SEQS

F is Archimedean and has the CCC property

F is Archimedean and has the CIP

F has the SUP property.

....
Exercise N031.07: Show: a n is a monotone increasing sequence in an Archimedean ordered
field F. If a n is bounded above then it is a Cauchy sequence.
(Then, F is Archimedean and has the CCC property implies F has the MCT property.)
....................................................

24
 H Infinite limits and properly divergent sequences.

H01 Definition:
Let a n be a sequence such that
Given K, M : n M a n K.
We then say ‘a n tends to infinity ( )’ or ‘a n properly diverges to infinity ( )’ or ‘the limit of a n
is infinity ( )’.
We write lim a n or a n .

A similar definition: Let a n be a sequence such that

Given K, M : n M a n K.
We then say ‘a n tends to minus infinity ( )’ or ‘a n properly diverges to minus infinity ( )’ or
‘the limit of a n is minus infinity ( )’.
We write lim a n or a n .
....
H02 Note:
H02.01: a n an .

H02.02: If lim a n , then a n is bounded below but not above.

H02.03: If lim a n , then a n is bounded above but not below.

H02.04: a n has a subsequence which tends to if and only if a n is not bounded above.

H02.05: a n has a subsequence which tends to if and only if a n is not bounded below.

H02.06: lim a n if and only if every subsequence of a n tends to .

H02.07: lim a n if and only if every subsequence of a n tends to .

H02.08: Thus a n where R ,

if and only if
every subsequence of a n tends to .

H02.09: R , is called the set of extended real numbers and we denote it by R.

H02.10: It can be shown that a n does not have a limit in R if and only if there exist in R
which are limits of subsequences of a n . Such sequences are called oscillating sequences.
....
H03 Note on extension of the algebra of limits:
H03.01: Let a n & bn . If R or , then a n b n .

H03.02: Let a n & bn . If R or , then a n bn .

25
H03.03: If a n & bn , then we cannot predict the behaviour of a n bn .
If a n n , bn n, then a n b n .
If a n 2n, b n n, then a n b n .
If a n n, b n 2n, then a n b n .
If a n n 1 n, bn n, then a n b n oscillates.
...
This is why ‘ ’ is called an ‘indeterminate form’.

H03.04: Let a n & bn . If or R& 0, then a n b n .

H03.05: Let a n & bn . If or R& 0, then a n b n .

H03.06: Let a n & bn . If or R& 0, then a n b n .

H03.07: Let a n & bn . If or R& 0, then a n b n .

H03.08: If a n & b n 0, then we cannot predict the behaviour of a n b n .

If a n n, b n 1/n, then a n b n .
If a n n 2 , b n 1/n, then a n b n .
2
If a n n , b n 1/n, then a n b n .
If a n n, b n 1/n 2 , then a n b n 0.
If a n n, b n 1 n /n, then a n b n oscillates.
...
This is why ‘0. ’ is also an indeterminate form

H03.09: If a n , then |a n | .

H03.10: If a n , then 1/a n 0.

Let 0.
an
|a n |
K:
n K |a n | 1/
K:n K 1/|a n | .

Thus, if a n & bn , then b n /a n 0.

H03.11: But if both a n & b n , then we cannot predict the behaviour of b n /a n .

If a n n, b n n, then b n /a n ( 0).
If a n n 2 , b n n, then b n /a n 0.
If a n n, b n n 2 , then b n /a n .
...
This is why ‘ / ’ is an indeterminate form.

26
H03.12: If a n 0, then 1/a n may not even be defined.
Even if ultimately, a n 0, we cannot predict the behaviour of 1/a n .
If a n 1/n, then 1/a n .
If a n 1/n, then 1/a n .
If a n 1 n /n, then 1/a n oscillates.

But if a n 0, and is ultimately non-zero, then 1/|a n | .

Given K 0, M : n M
0 |a n | 1/K
Thus, M : n M 1/|a n | K.

H03.13: Thus, if a n 0, and is ultimately non-zero, & b n 0, then |b n /a n | .

Case (1): a n is ultimately positive: then
b n /a n if 0 (or )& if 0 (or ).
Case (2): a n is ultimately negative: then
b n /a n if 0 (or )& if 0 (or ).
Case (3): If a n has infinitely many positive & infinitely many negative terms, then
b n /a n oscillates but |b n /a n | .

H03.14: But if both a n & b n 0, then we cannot predict the behaviour of b n /a n .

If a n 1/n, b n /n, then b n /a n .
If a n 1/n 2 , b n 1/n, then b n /a n .
If a n 1/n, b n 1/n 2 , then b n /a n 0.
If a n 1 n /n, b n 1/n, then b n /a n oscillates.
...
So ‘0/0’ is also an indeterminate form.
....

27
Exercises N031.08:

Show
(i) a n bn; n K; & a n bn .

(ii) a n bn; n K; & b n an .

(iii) Let a n & bn . If R or , then a n bn .

(iv) Let a n & bn . If R or , then a n bn .

(v) Let a n & bn . If or R& 0, then a n b n .

(vi) Let a n & bn . If or R& 0, then a n b n .

(vii) Let a n & bn . If or R& 0, then a n b n .

(viii) If a n , then a1n 0.

(ix) If a n 0 & an 0; n K, then a1n .

(x) If a n 0 & an 0; n K, then a1n .

(xi) lim a n if and only if every subsequence of a n tends to .

(xii) lim a n if and only if every subsequence of a n tends to .

(xiii) a n has a subsequence which tends to if and only if a n is not bounded below.

(xiv) a n has a subsequence which tends to if and only if a n is not bounded above.

(xv) lim a n lim a n

................................................

28
 I Divergent Monotone sequences

I01 Theorem:
Let a n be a monotone increasing sequence. Then, if a n is not bounded above, then a n .

Proof:
Let a n be unbounded above and let K R.

Then as K is not an upper bound of a n , N : a N K.

Then n N an aN K.

Thus, a n .
....
I02 Theorem:
Let a n be a monotone decreasing sequence. If a n is not bounded below, then a n .

Proof: Exactly similar to proof of Theorem I01.

....
I03 Note:
Note that a monotone sequence cannot oscillate, it either converges (if it is bounded) or properly
diverges.
................................................

29
J Some convergent sequences

J01: lim p 1/n 1 where p 0.

Proof: First let p 1.

Let 0.

Let p 1 h. Then h 0.

Then a n p 1/n 1 h 1/n

1.

Let a n 1 h n . Then h n 0.

Then p a nn 1 hn n
1 nh n . . . 1 nh n .

p 1
Thus 0 hn n

By the sandwich theorem, h n 0

Thus a n 1 hn 1.

If p 1, then p 1/n 1 1.

If p 1, then as 1p 1, 11/n 1
p
1/n
1.
p

Thus p 1/n 1/ 1 1/1 1

p 1/n
....

30
J02: lim n 1/n 1.

Proof: Let 0.

Let a n n 1/n 1 hn .

Then n 2, h n 0.

Let a nn n 1 hn n.

Then n 2,
n
n hk nn 1
n 1 hn n
1 nh n h 2n .
k n 2
k 0

Thus 0 h 2n 2
n 1

By the sandwich theorem, h 2n 0. Thus, h n 0

Thus a n 1 hn 1.
....

31
J03:
We define (1)
an : 1 1 1 ... 1
1! 2! n!

& (2)

n
en : 1 1
n

Then (3)

n
en 1 1
n
n
n 1
r nr
r 0
n
1 nn 1 ... n r 1
nr r!
r 0
n
1 1 1 1 ... 1 r 1
r! n n
r 1

Therefore (4)

n 1
en 1 1 1 1 1 ... 1 r 1
r! n 1 n 1
r 1

As (5)
1 k 1 k
n 1 n; k n
&
1 1 1 ... 1 n 0
n 1! n 1 n 1

we get (6)
en en 1

Also. as (7)
1 k 1; k n
n

we get (8)

32
 n
en 1 1 1 1 ... 1 r 1
r! n n
r 1

1 1 1 ... 1
1! 2! n!
an.

Obviously, a n is a (strictly) monotone increasing sequence. (9)

Now, for k 2, (10)

k! k. k 1 . . . 2. 1
2. . . 2. 1
1
2k

Thus, (11)
an 1 1 1 ... 1
1! 2! n!
1 1 1 ... 1
1 22 1 2n 1
1 1 1 ... 1
2 2n 1
1 1
1 2n
1 1
2
3

Thus, both a n & e n (increasing sequences) are bounded above (by 3).
Hence they converge. (12)

Let (13)
lim a n a & lim e n e.

Then (14)
2 e a 3
................................

33
Fix k 1. (15)
For n k, (16)
n
en 1 1
n
n
n 1
r nr
r 0
k
1 nn 1 ... n r 1
nr r!
r 0
k
1 1 1 1 ... 1 r 1
r! n n
r 1

Let (17)
k
cn 1 1 1 1 ... 1 r 1
r! n n
r 1

As (18)
1 1 1 ... 1 r 1 1 . 1. 1. . . 1 1
r! n n r! r!

we have (19)
k
cn 1 1 1 1 ... 1 r 1
r! n n
r 1
k
1 1 ak
r!
r 1

From (16), we get (20)

e ak; k 1

Thus (21)
e a

Hence (22)
e a
Therefore (23)
n
e lim 1 1 lim 1 1 1 ... 1
n 1! 2! n!
.............................................................

34
Edited:
2005
26 Jan 2018
29 Jan 2018
11 Feb 2018
15 May 2019
14 Dec 2019
25 Jun 2022
26 Jun 2022
12 July 2022
13 July 2022

35