Sequences

Sequences
P. Sam Johnson
National Institute of Technology Karnataka (NITK)

Surathkal, Mangalore, India
P. Sam Johnson Sequences 1/190

Overview
We shall study a special class of functions (called sequences) whose

domain is the set of all natural numbers and range a set of real numbers.
Analysis is based on the notion of a limit, a concept that can be defined in

terms of sequences. Moreover, elementary functions, such as
trigonometric, exponential, and logarithm functions and many algebraic
functions, can be approximated by using sequences.
With modern computers, such approximations can be made accurate

enough for most practical purposes.
We shall discuss sequences of real numbers in a couple of lectures.

Sequences
A sequence is a list of numbers
a1 , a2 , a3 , . . . , an , . . .
in a given order. Each of a1 , a2 , a3 and so on represents a number. These

are the terms of the sequence.
The integer n is called the index of an , and indicates where an occurs in

the list. We can think of the sequence
a1 , a2 , a3 , . . . , an , . . .
as a function that sends 1 to a1 , 2 to a2 , 3 to a3 , and in general, the

function sends the positive integer n to the nth term an . This leads
to the formal definition of a sequence.

Infinite Sequence
Definition 1 (Infinite Sequence).
An infinite sequence of numbers is a function whose domain is the set of
positive integers.
Example 2.
The function associated to the sequence
2, 4, 6, 8, 10, 12, . . . , 2n, . . .
sends 1 to a1 = 2, 2 to a2 = 4, and so on. The general behavior of this

sequence is described by the formula
an = 2n.

Infinite Sequence
We can equally well make the domain, the integers larger than a given
number n0 , and we allow sequences of this type also.
The sequence
12, 14, 16, 18, 20, 22, . . .
is described by the formula an = 10 + 2n. It can also described by the
simpler formula bn = 2n, where the index n starts at 6 and increases.
To allow such simpler formulas, we let the first index of the sequence be
any integer. In the sequence above, {an } starts with a1 while {bn } starts
with b6 .
Order is important. The sequence 1, 2, 3, 4, . . . is not the same as the

sequence 2, 1, 3, 4, . . . .

Examples of Sequences
Sequences can be described by writing rules that specify their terms, such
as √ n−1
an = n cn =
1 n
bn = (−1)n+1 dn = (−1)n+1
n
or by listing terms, n√ √ √ √ o
{an } = 1, 2, 3, . . . , n, . . .
n 1 1 1 1 o
{bn } = 1, − , , − , . . . , (−1)n+1 , . . .
2 3 4 n
n 1 2 3 4 n−1 o
{cn } = 0, , , , , . . . , ,...
n 2 3 4 5 n o
{dn } = 1, −1, 1, −1, 1, −1, . . . , (−1)n+1 , . . . .
We also sometimes write
n√ o∞
{an } = n .
n=1

What is the difference between a set and a sequence?
Question : What is the difference between a set and a sequence?
Sequence always has a definite order of elements.
A set is a (well-defined) collection of distinct elements (that contains no

duplicate elements).
{1, −1, 1, −1, 1, −1, 1, . . .} is a sequence whose elements are from the set
{−1, 1}.

Convergence and Divergence
Sometimes the numbers in a sequence approach a single value as the index

n increases. This happens in the sequence
n 1 1 1 1 o
1, , , , . . . . , . . .
2 3 4 n
whose terms approach 0 as n gets large, and in the sequence
n 1 2 3 4 1 o
0, , , , , . . . , 1 − , . . .
2 3 4 5 n
whose terms approach 1.

On the other hand, sequences like

n√ √ √ √ o
1, 2, 3, . . . , n, . . .
have terms that get larger than any number as n increases, and sequences
like n o
1, −1, 1, −1, 1, −1, . . . , (−1)n+1 , . . .
bounce back and forth between 1 and −1, never converging to a single
value.

Converges, Diverges, Limit
Definition 3 (Converges, Diverges, Limit).
The sequence {an } converges to the number L if for every positive number
ε there corresponds a positive integer N such that for all n,
n > N =⇒ |an − L| < ε.
That is,
L − ε < an < L + ε for all n > N.
If no such number L exists, we say that {an } diverges.

Definition of Convergence
In some textbooks, the following definition is used for convergence.

Definition 4.
The sequence {an } converges to the number L if for every positive number
ε there corresponds a positive integer N such that for all n,
n ≥ N =⇒ |an − L| < ε.
The only difference is that we have n ≥ N instead of n > N. When n > N we mean, we are
concerned about the terms of the sequence from N + 1 onwards ; whereas n ≥ N we mean, we
are concerned about the terms of the sequence from N onwards. This does not affect the choice
of N. You will understand that the convergence/divergence of a sequence which does not
depend on first finitely many terms. A finite term can be a number (distance) from NITK to
any city in India, in milimeters.

The shortest math joke: Let epsilon be < 0.
Let {an } be a sequence converging to a real number, say L.
Now we apply the definition of convergence of {an } to L for ε. We will get

a natural number, say N1 , such that
n > N1 =⇒ |an − L| < ε.
Again we apply the definition of convergence of {an } to L for ε/99. We

will get a natural number, say N2 such that
n > N2 =⇒ |an − L| < ε.
Caution : We should not think that N2 is always more that 99N1 .

However, N2 is more than N1 .

Representing Sequences Graphically
The following figure shows two ways to represent sequences

graphically.
The first marks the first few points from a1 , a2 , a3 , . . . , an , . . . on the

real axis.
The second method shows the graph of the function defining the
sequence. The function is defined only on integer inputs, and the
graph consits of some points in the xy-plane, located at
(1, a1 ), (2, a2 ), . . . , (n, an ), . . ..

Sequences can be represented as points on the real line or as points in the

plane where the horizontal axis n is the index number of the term and the
vertical axis an is its value.

Sequences can be represented as points on the real line or as points in

the plane where the horizontal axis n is the index number of the term and
the vertical axis an is its value.

Visualizing Sequences
As discussed above, there are two ways to visualize a sequence of real

numbers.
In real line : We can have the terms of the sequence {an } on the
real line.
In the plane : Every sequence (of real numbers) is a function
(whose domain is the set of positive integers) and range (image) is in
the set of real numbers.

Which is good for better understanding?
Let us consider the constant sequence {10}∞

n=1 .
On the real line : We get the only one point, 10, on the real line.
Given ε > 0, we have an interval (10 − ε, 10 + ε) which contains
every element of the sequence {10}∞n=1 , hence any natural number N
will satisfy the following :
n > N =⇒ |an − L| = |10 − 10| < ε.
In the plane : We get points on the line y = 10 defined over the

set of positive integers. Given ε > 0, we get two lines y = 10 − ε and
y = 10 + ε. As every dot (n, an ) lies between 10 + ε and 10 − ε, we
can choose any natural number N which will satisfy the following :
n > N =⇒ |an − L| = |10 − 10| < ε.

Common Mistake
Some students may write the following incomplete line
L − ε < an < L + ε
for the definition of convergence which is not correct.
Given a seqence {an }, we guess some number L where the sequence might
converge.
To prove our guess L is correct, we start with ε > 0, we find a positive

integer N such that when n > N we must prove |an − L| < ε.
That is, given ε > 0, we shall find a positive integer N such that
n > N =⇒ |an − L| < ε.

Uniqueness of Limits
Theorem 5 (Uniqueness of Limits).
Every convergence sequence has a unique point.
That is, if {an } is a sequence of real numbers and, if L1 and L2 are

numbers such that
an → L1 and an → L2 ,
then
L1 = L2 .
In other words, a sequence cannot converge to more than one limit.

Outline of the proof
We would like to show that |L1 − L2 | < ε for every ε > 0.
Why is this useful ?

For every ε > 0, if we have |L1 − L2 | < ε, then |L1 − L2 | = 0, hence
L1 = L2 . [Reason : Suppose L1 6= L2 , for ε = |L1 −L
2
2|
, we get that
|L1 −L2 |
ε = 2 < |L1 − L2 | which contradicts to |L1 − L2 | < ε, for every ε > 0.]
How to show |L1 − L2 | < ε for every ε > 0 ?

We have
|L1 − L2 | = |L1 − an + an − L2 |
≤ |an − L1 | + |an − L2 |.
In order to show |L1 − L2 | < ε, we shall show that |an − L1 | < ε/2 and
|an − L2 | < ε/2 by applying the definition of convergence for ε/2.

Proof
Let {an } be a sequence of real numbers.
Suppose that {an } converges to two distinct numbers L1 and L2 .
Let ε > 0 be given.
Since an → L1 , there is a positive integer N1 such that
|an − L1 | < ε/2 for all n > N1 .
Similarly, as an → L2 , there is a positive integer N2 such that
|an − L2 | < ε/2 for all n > N2 .

Proof (contd...)
Now, for N > max{N1 , N2 },
|L1 − L2 | = |L1 − an + an − L2 |
≤ |an − L1 | + |an − L2 |
< ε/2 + ε/2 = ε.
Hence
0 < |L1 − L2 | < ε, for every ε > 0. (1)
|L1 −L2 |
But if ε is chosen to be 2 , we get a contradiciton with (1).
Thus, the sequence cannot converge to two limits.
Note : In the above proof, we considered a partition of ε as 2ε + 2ε . If we

ε 99ε
consider the partition as 100 + 100 , we may get the some natural numbers
M1 and M2 (may be different from N1 and N2 ) satisfying inequalities.

Converges, Diverges, Limit
The definition says that if we go far enough out in the sequence, by taking
the index n to be larger then some value N, the difference an and the limit
of the sequence becomes less than any preselected number ε > 0.
The definition is very similar to the definition of the limit of a

function f (x) as x tends to ∞. We will exploit this connection to
calculate limits of sequences.

If {an } converges to L, we write
lim an = L, or simply an → L,
n→∞
and call the limit of the sequence.
In the representation of a sequence as points in the plane, an → L if y = L

is a horizontal asymptote of the sequence of points {(n, an )}. In this
figure, all the an ’s after aN lie within ε of L.

Exercise
Exercise 6.
A formula for the nth term an of a sequence {an } is given. Find the values
of a1 , a2 , a3 , and a4 .
1−n
1. an = n2
2. an = 2 + (−1)n
2n
3. an = 2n+1
Note that when the index set for n is not explicitly given, it is
assumed that the index set is the set of natural numbers.

Exercise
Exercise 7.
The first term or two of a sequence along with a recursion formula for the
remaining terms are given. Write out the first ten terms of the sequence.
1. a1 = 1, an+1 = an + (1/2n )
2. a1 = −2, an+1 = nan /(n + 1)
3. a1 = 2, a2 = −1, an+2 = an+1 /an

Exercise
Exercise 8 (Finding a Sequence’s Formula).
Find a formula for the nth term of the sequence.
1. The sequence −1, 1, −1, 1, −1, 1, . . .
2. The sequence 1, − 41 , 19 , − 16
1 1
, 25 , . . .
3. The sequence −3, −2, −1, 0, 1, . . .
4. The sequence 0, 1, 1, 2, 2, 3, 3, 4, . . .
5. The sequence 1, 1, . . . , 1(10 times), 2, 2, . . . , 2(10 times), . . .
(Each positive integer is repeated 10 times.)
6. The sequence 1, 2, 2(2 times), 3, 3, 3(3 times), . . .
(Each positive integer n is repeated n times.)

Solution for Exercise 8
1. (−1)n , n≥1
(−1)n+1
2. n2
, n≥1
3. n − 4, n≥1
4. b n2 c, n≥1
n
5. d 10 e, n ≥ 1
j √ k
1+ 1+8n
6. 2 , n≥1

Sequence’s formula is not unique.
For example, one can find the following sequence’s formula

4 5 14 4 92 3 280 2 1936
an = n − n + n − n + n − 63
15 3 3 3 15
for the sequence {−1, 1, −1, 1, −1, 1, . . .}.
For n = 1, 2, 3, 4, 5, 6, we get −1, 1, −1, 1, −1, 1 respectively. But a7 = 63

(not −1).
A simple formula is an = (−1)n for n ≥ 1.
If we know the first finitely many terms of a sequence, say {an }, every one
in world can give a sequence’s formula for the sequence
{−1, 1, −1, 1, −1, 1, . . .} which satisfies an = (−1)n , 1 ≤ n ≤ 6, and a7 can
be any number of his/her choice.

Sequence’s formula is not unique.
Two students were asked to write an nth term for the sequence
1, 16, 81, 256, . . .
and to write 5th term of the sequence. One student gave the nth term as
an = n 4 .
The other student, who did not recognize this simple law of formation,
wrote
an = 10n3 − 35n2 + 50n − 24.
Which student gave the correct 5th term?

Using ε − N definition, prove the convergence of sequences.
Example 9 (Applying the ε − N Definition of Convergence).
Show that
1
1. lim =0
n→∞ n
2. lim k = k (any constant k).
n→∞

Solution
1. Let ε > 0 be given. We must show that there exists a positive integer
N such that for all n,
1
n > N =⇒ − 0 < ε.
n
This implication will hold if n1 < ε or n > 1/ε. If N is any integer
greater than 1/ε (or, choose N = d1/εe), the implication will hold for
all n > N. That is, n > N =⇒ | n1 − 0| = n1 < ε. This proves that
lim (1/n) = 0.
n→∞
2. Let ε > 0 be given. We must show that there exists a positive integer
N such that for all n,
n > N =⇒ |k − k| < ε.
Since k − k = 0, we can choose any positive integer for N (or, choose
N = 10 or N = 1000 billion) and the implication will hold. This
proves that lim k = k for any constant k.
n→∞
Bounded Sequence
Definition 10.
A sequence {an } is said to be bounded if there are real numbers m1 and
m2 such that
m1 ≤ an ≤ m2 for all n.
One can use the following equivalent statement for bounded: {an } is
bounded if there is a positive number M such that
|an | ≤ M, for all n.
Theorem 11.
Every convergent sequence is bounded.

Proof
Let {an } be a sequence converging to a limit L.
Let ε > 0 be a fixed number, say 100. Since an → L, there is a positive

integer N such that
|an − L| < 100, for all n > N.
That is,
L − 100 < an < L + 100, for all n > N.
Note that for n ≥ N + 1, all an ’s will lie in the interval (L − 100, L + 100).
What about the terms a1 , a2 , . . . , aN ?
Some of them may lie in the interval (L − ε, L + ε), or,
some of them may not lie in the interval.
Let m1 = min{a1 , a2 , . . . , aN , L − ε} and m2 = max{a1 , a2 , . . . , aN , L + ε}.

We get that for all n,
m1 ≤ an ≤ m2 .
Thus {an } is bounded. This completes the proof.
What is special about 100 in the above proof?
We could have used 1 in place of 100 and apply convergence of {an } to L

for ε = 1 to get some positive integer, say M.
Hence n > M =⇒ |an − L| < 1 and proceed.
Observations : We proved that every convergent sequence is bounded.

What about the converse of the above statement? It is not true. That is,
a bounded sequence is not necessarily convergent. For example, the
seqence {1, 2, 3, 4, 1, 2, 3, 4, . . .} is bounded, but not convergent.
Boundedness is a necessary condition for the convergence of a sequence

but not a sufficient condition. That is, if a sequence is not bounded, then
it cannot converge. For example, {n!} is not convergent because it is not
bounded.

Divergent Sequences
Example 12.
Show that the sequence {1, −1, 1, −1, 1, −1, . . . , (−1)n+1 , . . .} diverges.
Solution : Suppose the sequence converges to some number L. By

choosing ε = 1/2 in the definition of the limit, all terms an of the sequence
with index n larger than some N must lie within ε = 1/2 of L. Since the
number 1 appears repeatedly as every other term of the sequence, we must
have that the number 1 lies within the distance ε = 1/2 of L. It follows
that |L − 1| < 1/2, or equivalently, 1/2 < L < 3/2. Likewise, the number
−1 appears repeatedly in the sequence with arbitrarily high index. So we
must also have that |L − (−1)| < 1/2, or equivalently, −3/2 < L < −1/2.
But the number L cannot lie in both of the intervals (1/2, 3/2) and
(−3/2, −1/2) because they have no overlap. Therefore, no such limit L
exists and so the sequence diverges. Note that the same argument
works for any positive number ε smaller than 1, not just 1/2.

Divergent Sequences
√
The sequence { n} also diverges, but for a different reason. As n
increases, its terms become larger than any
√ fixed number. We describe the
behavior of this sequence by writing lim n = ∞.
n→∞
1. In writing infinity as the limit of a sequence, we are not saying that

the differences between the terms an and ∞ become small as n
increases.
2. Nor are we asserting that there is some number infinity that the
sequence approaches.
3. We are merely using a notation that captures the idea that an
eventually gets and stays larger than any n gets large.

Diverges to Infinity
Definition 13 (Diverges to Infinity).
The sequence {an } diverges to infinity if for every number M there is an
integer N such that for all n larger than N, an > M. If this condition holds
we write
lim an = ∞ or an → ∞.
n→∞
Similarly if for every number m there is an integer N such that for all
n > N we have an < m, then we say {an } diverges to negative infinity and
write
lim an = −∞ or an → −∞.
n→∞

Diverges to +∞ or −∞
The sequence diverges to ∞ The sequence diverges to −∞

because no matter what number M because all terms after some index
is chosen, the terms of the sequence N lie below any chosen number m.
after some index N all lie in the
yellow band above M.

Diverges to Infinity
Example 14.
A sequence may diverge without diverging to infinity or negative infinity.
The sequences
{1, −2, 3, −4, 5, −6, 7, −8, . . .}
and
{1, 0, 2, 0, 3, 0, . . .}
are also examples of such divergence.

Calculating Limits of Sequences
If we always had to use the formal definition of the limit of a sequence,

calculating with ε’s and N’s, then computing limits of sequences would
be a formidable task.
Fortunately we can derive a few basic examples, and then use these to
quickly analyze the limits of many more sequences.
We will need to understand how to combine and compare sequences.

Since sequences are functions with domain restricted to the positive
integers, it is not too surprising that the theorems on limits of functions
have versions for sequences.

Calculating Limits of Sequences
Theorem 15.
Let {an } and {bn } be sequences of real numbers and let A and B be real
numbers. The following rules hold if lim an = A and lim bn = B.
n→∞ n→∞
1. Sum Rule : lim (an + bn ) = A + B
n→∞
2. Difference Rule : lim (an − bn ) = A − B
n→∞
3. Product Rule : lim (an .bn ) = A.B
n→∞
4. Constant Multiple Rule : lim (k.bn ) = k.B (any number k)
n→∞
an A
5. Quotient Rule : lim = if bn 6= 0 for all n and B 6= 0
n→∞ bn B

Proof (Sum Rule)
Rough Work (not necessarily to be written in exam) :
We have to find a positive integer N such that
|(an + bn ) − (A + B)| < ε for all n > N.
Let ε > 0 be given. From the triangle inequality, we have
|(an + bn ) − (A + B)| ≤ |an − A| + |bn − B|.
Hence we can apply the defintion for ε/2, for the sequences {an }
and {bn } converging to A and B respectively.
Since an → A and bn → B, there are positive integers N1 and N2

respectively, such that
ε
|an − A| < for all n > N1
2
and
ε
|bn − B| < for all n > N2 .
2

Proof (contd...)
Thus for all n > N = max{N1 , N2 },
|(an + bn ) − (A + B)| ≤ |an − A| + |bn − B|

ε ε
< + = ε.
2 2
Hence {an + bn } converges to A + B.

Proof (Difference Rule)
|(an − bn ) − (A − B)| < ε for all n > N.
Let ε > 0 be given. From the triangle inequality, we have
|(an − bn ) − (A − B)| ≤ |an − A| + |bn − B|.
Hence we can apply the defintion for ε/2, for the sequences {an }

ε
2
and
ε
2

Proof (contd...)
|(an − bn ) − (A − B)| ≤ |an − A| + |bn − B|

ε ε
< + = ε.
2 2
Hence {an − bn } converges to A − B.

Proof (Product Rule)

We have to find a positive integer N such that |an bn − AB| < ε for all n > N.
From the triangle inequality, we have
|an bn − AB| = |an bn − an B + an B − AB| ≤ |an | |bn − B| + |an − A| |B|.
ε
Hence one may want to apply the defintion for ε as 2|B| and 2|aε | , for the sequences {an }
n
ε
Here there are some problems. The number 2|a | is depending on n ;
n
ε
when B is zero, what happens to the number 2|B| .
So, we are using boundedness of the convergent sequence {an } ;
and we are discussing a separate case when B = 0.
Since every convergent sequence is bounded, for the sequence {an }, there
exists a positive number C such that
|an | ≤ C , for all n.

Proof (contd...)
Case 1 : B is non-zero
ε
2|B|
and ε
2C
|an bn − AB| = |an bn − an B + an B − AB|
≤ |an | |bn − B| + |an − A| |B|
≤ C |bn − B| + |an − A| |B|
ε ε
< + = ε.
2 2
Hence {an bn } converges to AB when B 6= 0.
Proof (contd...)
Case 2 : B is zero
We have to find a positive integer N such that |an bn − 0| < ε for all n > N.
From the triangle inequality, we have |an bn − 0| = |an | |bn − 0|.
Hence one can apply the “ε − N” defintion for the sequence {bn } converging to 0.
Since bn → B, there is a positive integer N1 such that

ε
|bn − 0| < for all n > N1 .
C
Thus for all n > N1 ,
|an bn − 0| = |an | |bn − 0|
≤ C |bn − 0|
< ε.
Hence {an bn } converges to 0 when B = 0.
This completes the proof.

Proof (Constant Multiple Rule)
If k is zero, {kan } is a constant sequence, converging to the constant,

which is 0 here. So, let’s assume that k is non-zero.
We have to find a positive integer N such that |kan − kA| < ε for all n > N.
From the triangle inequality, we have |kan − kA| = |k| |an − A|.
Hence one can apply the “ε − N” defintion for the sequence {an } converging to A, for ε/|k|.
Since an → A, there is a positive integer N such that
ε
|an − A| < for all n > N.
|k|
Thus for all n > N,
|kan − kA| = |k| |an − A|

< ε.
Hence {kan } converges to kA.

Proof (Quotient Rule)
To prove the quotient rule, we use the following lemma.

Lemma 16.
If lim bn = B 6= 0, then there is a positive integer N such that
n→∞
|B|
|bn | > , for all n > N.
2
Proof of the lemma :
|B|
Since bn → B and ε = 2 , there exists a positive integer N such that
|B|
|bn − B| < , for all n > N.
2
This implies that
|B|
|B| − |bn | ≤ |bn − B| < , for all n > N
hence, 2
|B|
|bn | > , for all n > N.
2
Proof (contd...)
Let us apply the Lemma to prove the quotient rule.

an A |Ban −Abn | |Ban −BA−Abn +AB| |B|.|an −A|+|A|.|bn −B| |an −A| |A|.|bn −B|
− = |B|.|bn |
= |B|.|bn |
≤ |B|.|bn |
≤ |b + |B|.|b
bn B n| n|
|B|
Using lemma, there exists a positive integer N such that |bn | > 2 , for all n > N.
an A 2 2|A|.|bn −B|
We get bn
− B
< |B| |an − A| + |B|2
,
|B|ε |B|2 ε
Hence we can apply the defintion for ε as 4 and 4|A| , for the sequences {an } and {bn }
converging to A and B respectively. The case for A = 0, should be discussed separetely.
Case 1 : A is non-zero

|B|ε
4
Proof (contd...)
and
|B|2 ε
4|A|
By lemma, there exists a positive integer N3 such that
|B|
|bn | > , for all n > N3 .
2
Thus for all n > N = max{N1 , N2 , N3 },

an A |Ban − Abn | |Ban − BA − Abn + AB|
− = =
bn B |B|.|bn | |B|.|bn |
|B|.|an − A| + |A|.|bn − B| |an − A| |A|.|bn − B|
≤ ≤ +
|B|.|bn | |bn | |B|.|bn |
2 2|A|.|bn − B| ε ε
< |an − A| + < + = ε.
|B| |B|2 2 2
Hence { bann } converges to A
B when A 6= 0.
Proof (contd...)
Case 2 : A is zero
an
We have to find a positive integer N such that bn
−0 <ε for all n > N.
an
From the triangle inequality, we have bn
− 0 = |b1 | |an − 0|.
n
Hence one can apply the lemma for the sequence {bn } and
the “ε − N” defintion for the sequence {an } converging to 0.
By lemma, there exists a positive integer N1 such that
|B|
|bn | > , for all n > N1 .
2

Proof (contd...)
Since an → A, there is a positive integer N2 such that
|B|
|an − 0| < for all n > N2 .
2ε
an 1
−0 = |an − 0|
bn |bn |
2
= |an − 0|
|B|
< ε.
n o
an
Hence bn converges to 0 when A = 0.
This completes the proof.

Applying Theorem for Limits of Sequences
Example 17.

1 1
1. lim − = −1. lim = −1.0 = 0
n→∞ n n→∞ n

n−1 1 1
2. lim = lim 1 − = lim 1 − lim =1−0=1
n→∞ n n→∞ n n→∞ n→∞ n
5 1 1
3. lim 2 = 5. lim . lim = 5.0.0 = 0
n→∞ n n→∞ n n→∞ n
4 − 7n6 (4/n6 ) − 7 0−7
4. lim 6 = lim 6
= = −7
n→∞ n + 3 n→∞ 1 + (3/n ) 1+0

Algebra of Limits
While the sum, difference, product and quotient (under appropriate

conditions) of two convergent sequences is convergent, the converse may
not be true.
That is, if {an ± bn }, {an bn } or { bann } is convergent, the component

sequences {an } and {bn } may not be convergent. However, both these
sequences {an } and {bn } shall behave alike.
Equivalently, existence of lim (an ± bn ) does not necessarily imply the

n→∞
existence of the two limits lim an and lim bn . Analogously, for the
n→∞ n→∞
product and quotient.

Algebra of Limits
For example, consider the sequnces {an } and {bn } below:
When an = n2 and bn = −n2 ,

{an + bn } → 0 and { bann } → −1, but both {an } and {bn } are
divergent.
When an = bn = (−1)n ,
{an − bn } → 0, { bann } → 1 and {an bn } → 1, whereas both {an } and
{bn } oscillate finitely.
If an = (−1)n and bn = (−1)n+1 , then
{an + bn } → 0, { bann } → −1 and {an bn } → −1, while both {an } and
{bn } are not convergent (or oscillate finitely).

Algebra of Limits
If {an } converges and {bn } diverges, then {an + bn } diverges. But, when
both {an } and {bn } diverge, {an + bn } may converge, diverge or oscillate.
However, if {an } converges and {bn } diverges, then nothing can be said
about {an bn }. That is, {an bn } may converge, diverge or oscillate.
It follows from Constant Multiple rule that “every non-zero multiple of a

divergent sequence is divergent.” That is, for any c 6= 0, if {an } is
divergent, then {can } must also be divergent.
For suppose, to the contrary, that {can } converges for some number
c 6= 0. Then, taking k = 1/c in the “Constant Multiple Rule”, we see that
the sequence
n1 o
.can = {an }
c
converges. Thus, {can } cannot converge unless {an } also converges. If
{an } does not converge, then {can } does not converge.
Algebra of Limits
Exercise 18.
1. Give examples of sequences {an } and {bn } such that
(a) an → +∞, bn → −∞, but {an + bn } converges.
(b) an → +∞, bn → −∞, but {an + bn } diverges to −∞.
(c) an → +∞, bn → −∞, but {an + bn } oscillates.
2. Give examples of sequences {an } and {bn } such that
(a) an → +∞, {bn } converges, but {an bn } converges.
(b) an → +∞, {bn } converges, but {an bn } diverges to +∞.
(c) an → +∞, {bn } converges, but {an bn } oscillates.

1. (a) an = n and bn = −n
(b) an = n and bn = −2n
(c) an = n and bn = (−n + (−1)n )
1
2. (a) an = n and bn = n
(b) an = n and bn = (
1
1
n n is odd
(c) an = n and bn =
0 n is even

Sandwich / Squeeze Theorem
The next theorem is the sequence version of the Sandwich Theorem.

Theorem 19 (Sandwich / Squeeze Theorem for Sequences).
Let {an }, {bn } and {cn } be sequences of real numbers. If an ≤ bn ≤ cn
holds for all n beyond some index N, and if
lim an = lim cn = L,
n→∞ n→∞
then lim bn = L also.

n→∞

Proof of Sandwich Theorem
Since an → L and cn → L, there are positive integers N1 and N2

L − ε < an < L + ε for all n > N1
and
L − ε < cn < L + ε for all n > N2 .
Given that
an ≤ bn ≤ cn , for all n > N.
Thus for all n > N3 = max{N1 , N2 , N},
L − ε < an ≤ bn ≤ cn < L + ε for all n > N.
Hence {bn } converges to L.

Sandwich Theorem
An immediate consequence of the Sandwich Theorem for sequences is

that, if |bn | ≤ cn and cn →0, then bn →0 because −cn ≤ bn ≤ cn .
We use this fact in the next example.

Example 20.
Since 1/n → 0,
cos n
(a) n → 0 because − n1 ≤ cos n 1
n ≤ n
1
(b) 2n → 0 because 0 ≤ 21n ≤ n1
(c) (−1)n n1 → 0 because − n1 ≤ (−1)n n1 ≤ n1 .

An Application of Sequences - Bisection Method (Bolzano
Method)
The bisection method is based on the Intermediate Value Theorem which
states that if a function f (x) is continuous over an interval [a, b], then the
function takes on every value between f (a) and f (b).
Suppose f is a continuous function defined on [a, b], with f (a) and f (b) of
opposite sign. By the Intermediate Value Theorem, there exists a number
α in (a, b) with f (α) = 0.
Although the procedure will work when there is more than one root in the
interval (a, b), we assume for simplicity that the root in this interval is
unique. The method calls for a repeated halving of subintervals of [a, b]
and, at each step, locating the half containing α.

Algorithm
To begin, set a1 = a and b1 = b, and let x1 be the midpoint of [a, b].

That is,
b1 − a1 a1 + b1
x1 = a1 + = (first approximation).
2 2
If f (x1 ) = 0, then α = x1 , and we are done. If f (x1 ) 6= 0, then f (x1 ) has
the same sign as either f (a1 ) or f (b1 ).
1. When f (a1 ) and f (x1 ) have the same sign, α ∈ (x1 , b1 ), and we set
a2 = x1 and b2 = b1 .
2. When f (a1 ) and f (x1 ) have opposite signs, α ∈ (a1 , x1 ), and we set
a2 = a1 and b2 = x1 .
We then reapply the process to the interval [a2 , b2 ] to get second

approximation p2 .

How to apply bisection algorithm?
An interval [a, b] must be found with f (a).f (b) < 0.

As at each step the length of the interval known to contain a zero of
f is reduced by a factor of 2, it is advantageous to choose the initial
interval [a, b] as small as possible. For, example, if
f (x) = 2x 3 − x 2 + x − 1, we have both
f (−4).f (4) < 0 and f (0).f (1) < 0,
so the bisection algorithm could be used on either on the intervals

[−4, 4] or [0, 1]. However, starting the bisection algorithm on [0, 1]
instead of [−4, 4] will reduce by 3 the number of iterations required to
achieve a specified accuracy.

Bisection Method
Theorem 21.
Let f be a continuous function defined on [a, b] and f (a).f (b) < 0. The
bisection method generates a sequence {xn }∞n=1 approximating a zero α of
f with
b−a
|xn − α| ≤ , when n ≥ 1.
2n
By Sandwich Theorem, the sequence {pn } of approximations converges to

α with f (α) = 0.
The method has the important property that it always converges to a

solution, and for that reason it is often used as a starter for the more
efficient methods.
But it is slow to converge.

Number of Iterations Needed?
To determine the number of iterations necessary to solve
f (x) = x 3 + 4x 2 − 10 = 0
with accuracy 10−3 using a1 = 1 and b1 = 2 requires finding an integer N

that satisfies
|pN − α| ≤ 2−N (b − a) = 2−N < 10−3 .
A simple calculation shows that ten iterations will ensure an approximation
accurate to within 10−3 .
Again, it is important to keep in mind that the error analysis gives only a
bound for the number of iterations, and in many cases this bound is much
larger than the actual number required.

Continuous Function Theorem for Sequences
The following theorem states that applying a continuous function to a

convergent sequence produces a convergent sequence.
Theorem 22 (Continuous Function Theorem for Sequences).

Let {an } be a sequence of real numbers. If an → L and if f is a function
that is continuous at L and defined at all an , then f (an ) → f (L), that is,

lim f (an ) = f lim an .
n→∞ n→∞

Proof of “Continuous Function Theorem for Sequences”
Since f is continuous at L, there exists δ > 0 such that
|x − L| < δ =⇒ |f (x) − f (L)| < ε.
As an → L, by applying the defintion for ε = δ, there exists a positive

integer N such that
|an − L| < δ, for all n > N.
Thus for all n > N, we have
|f (an ) − f (L)| < ε.
This proves that {f (an )} converges to f (L).

How to use the Continuous Function Theorem for
Sequences?
Let {bn } be a sequence.
If we are able to find a sequence {an } converging to L and find a function

f which is continuous at L and defined of all an , then the limit of {bn } is
f (L).
Example 23.
q
n+1
By “Continuous Function Theorem for Sequences”, n → 1 because
n+1 √
n → 1 and f (x) = x is continuous at x = 1.

Continuous Function Theorem for Sequences
Example 24.
The sequence {1/n} converges to 0. By taking an = 1/n, f (x) = 2x , and
L = 0 in the “Continuous Function Theorem for Sequences”, we see that
21/n = f (1/n) → f (L) = 20 = 1. The sequence {21/n } converges to 1.
In the above example, if 2 is replaced by any positive real ‘a’, we still get
the same limit as 1:
Let a be a positive real number. By considering the continuous function

f (x) = ax , we can show that {a1/n } converges to 1.

An Application of Sequences - Sequential Criterion for
Continuity
Theorem 25 (Sequential Criterion for Continuity).
A function f : D ⊆ R → R is continuous at x = c if and only if every
sequence {an } in D that converges to c, the sequence {f (an )} converges
to f (c).

Exercise
Exercise 26.
Define g : R → R by
(
2x if x is rational
g (x) =
x +3 if x is irrational.
Find all points at which g is continuous.
To solve the exercise given above, we are using that fact that, in the set R
of real numbers, the set Q of rational numbers and the set R\Q (the
complement of Q) of irrational numbers are dense.
Definition 27.
A subset D of R is said to be dense if for every x ∈ R and every ε > 0,
there exists y ∈ D such that |x − y | < ε.

Note that for all x ∈ R, y = 2x and y = x + 3 are lines which intersect at

x = 3. It seems reasonable to suspect that g is only continuous at x = 3.
g is continuous at x = 3 :
Given ε > 0, we need to find δ > 0 such that
|x − 3| < δ =⇒ |g (x) − g (3)| < ε.
If x is rational, then |g (x) − g (3)| = |2x − 6| = 2|x − 3|. If x is irrational,

then |g (x) − g (3)| = |(x + 3) − 6| = |x − 3|.
So, given ε > 0, set δ = ε/2. Then for any real x (either rational or
irrational) satisfying
|x − 3| < δ =⇒ |g (x) − g (3)| < ε.
Hence g is continuous at x = 3.
Solution for Exercise 26 (contd...)
g is discontinuous at any x = c 6= 3 :
We don’t use “ε − δ” definition here (which you have studied in the last
semester). Instead, we use “Sequential Criterion for Continuity”.
Since the set Q of rational numbers and the set R\Q (the complement of
Q) of irrational numbers are dense, there exist sequences {qn } in Q and sn
in R\Q such that qn → c and sn → c.
However, f (qn ) = 2qn → 2c, whereas f (sn ) = sn + 3 → c + 3.
Since c 6= 3, we see that 2c 6= c + 3. This cannot occur if f is continuous

at x = c.

Finding Limits of Sequences Using Limits of Functions
In order to find the limit of {an }, sometimes we shall adopt the following
technique:
We may consider a function f such that f (n) = an , for all n ≥ N (for

some N).
If lim f (x) exists, say L, (L can be any number including ∞ or −∞),

x→∞
then we can say that the sequence {an } converges to L, or {an } diverges
to ∞, or −∞.
Why is this useful?
We can use L’Hôpital‘s Rule to find lim f (x).

x→∞

Recalling Indeterminate Forms and L’Hôpital‘s Rule
John Bernoulli discovered a rule for calculating limits of fractions whose

numerators and denominators both approach zero or +∞.
The rule is known today as L’Hôpital‘s Rule. He was French nobleman

who wrote the first introductory differential calculus text, where the rule
first appeared in print.
If the continuous functions f (x) and g (x) are both zero at x = a, then
f (x)
lim can be found by substituting x = a. The substitution produces
x→a g (x)
0/0, a meaningless expression, which we cannot evaluate.
We use 0/0 as a notation for an expression known as an indeterminate

form. There are seven indeterminate forms which are typically considered
in the literature:
0 ∞
, , 0 × ∞, ∞ − ∞, 00 , 1∞ , and ∞0 .
0 ∞
L’Hôpital‘s Rule
Theorem 28 (L’Hôpital‘s Rule (First Form)).
Suppose that f (a) = g (a) = 0, that f 0 (a) and g 0 (a) exist, and that
g 0 (a) 6= 0.
f (x) f 0 (a)
lim = 0 .
x→a g (x) g (a)
Sometimes after differentiation, the new numerator and denominator both

equal zero at x = a. In these cases, we apply a stronger form of
L’Hôpital‘s Rule.

Theorem 29 (L’Hôpital‘s Rule (Stronger Form)).
Suppose that f (a) = g (a) = 0, that f and g are differentiable on an open
interval I containing a, and that g 0 (a) 6= 0 if x 6= a. Then
f (x) f 0 (x)
lim = lim 0
x→a g (x) x→a g (x)
assuming that the limit on the right side exists.

To find
f (x)
lim
x→a g (x)
by L’Hôpital‘s Rule, continue to differentiate f and g , so long as we still

get the form 0/0 at x = a.
But as soon as one or the other of these derivatives is different from zero
at x = a we stop differentiating.
L’Hôpital‘s Rule does not apply when either the numerator or

denominator has a finite nonzero limit.

Indeterminate Form ∞/∞
Sometimes when we try to evaluate a limit as x → a by substituting x = a

we get an ambiguous expression like ∞/∞, ∞.0, or ∞ − ∞, instead of
0/0.
L’Hôpital‘s Rule applies to the indeterminate form ∞/∞ as well as

to 0/0. If f (x) → ±∞ and g (x) → ±∞ as x → a, then
f (x) f 0 (x)
lim = lim 0
x→a g (x) x→a g (x)
provided the limit on the right side exists.
In the notation x → a, a may be either finite or infinite. Moreover, x → a

may be replaced by the one-sided limits x → a+ or x → a− .

Indeterminate Form ∞.0 and ∞ − ∞
Sometimes these forms (∞.0 or ∞ − ∞) can be handled by using algebra

by converting them to 0/0 or ∞/∞ form.
Here again we do not mean to suggest that ∞.0 or ∞ − ∞ is a number.

They are only notations for functional behaviours when considering limits.

The next theorem enables us to use L’Hôpital‘s Rule to find the limits of
some sequences. It formalizes the connection between lim an and
n→∞
lim f (x).
x→∞
Theorem 30 (Using L’Hôpital‘s Rule).

Suppose that f (x) is a function defined for all x ≥ n0 and that {an } is a
sequence of real numbers such that an = f (n) for n ≥ n0 . Then
lim f (x) = L ⇒ lim an = L.

x→∞ n→∞

Proof
Suppose that lim f (x) = L.

x→∞
Then for each positive number ε there is a number M such that for all x.
x >M ⇒ |f (x) − L| < ε.
Let N be an integer greater than M and greater than or equal to n0 .
Then
n>N ⇒ an = f (n)
and
|an − L| = |f (n) − L| < ε.

Example 31.
Show that
ln n
lim = 0.
n→∞ n
The function lnxx is defined for all x ≥ 1 and agrees with the given
sequence at positive integers. Therefore, by L’Hôpital‘s Rule for
ln n ln x
sequences, lim will equal lim if the latter exists.
n→∞ n x→∞ x
A single application of L’Hôpital‘s Rule shows that
ln x 1/x 0
lim = lim = = 0.
n→∞ x x→∞ 1 1
ln n
We conclude that lim = 0.
n→∞ n

When we use L’Hôpital‘s Rule to find the limit of a sequence, we often

treat n as a continuous real variable and differentiate directly with
respect to n. This saves us from having to rewrite the formula for an as
we have done it in the example above.
Example 32.
Find
2n
lim .
n→∞ 5n
By L’Hôpital‘s Rule (differentiating with respect to n),
2n 2n . ln 2
lim = lim = ∞.
n→∞ 5n n→∞ 5

Example 33.
Find limit of each of the following sequences, if it exists.
√
(a) lim n n.
n→∞
(b) lim x 1/n , for a fixed positive real x.
n→∞
√
n
h1 i h 1 i
(a) lim n = lim exp ln n = exp lim ln n = exp(0) = 1.
n→∞ n→∞ n n→∞ n
(b) Let x > 0 be fixed.
h1 i h 1 i
lim x 1/n = lim exp ln x = exp lim ln x = exp(0) = 1.
n→∞ n→∞ n n→∞ n

Example 34.
Does the sequence whose nth term is
n+1 n

an =
n−1
converge? If so, find lim an .

n→∞
The limit leads to the indeterminate form 1∞ . We can apply L’Hôpital‘s

Rule if we first change the form to ∞ . 0 by taking the natural logarithms
of an
n+1 n

ln an = ln
n−1

n+1
= n ln .
n−1
Then,
h
n+1 i
lim ln an = lim n ln ∞.0 form
n→∞ n→∞ n−1

n+1
ln n−1 h0 i
= lim form
n→∞ 1/n 0
2
−2/(n − 1) h i
= lim By L’Hôpital‘s Rule
n→∞ −1/n2
2n2
= lim = 2.
n→∞ n2 − 1
Since ln an → 2 and f (x) = e x is continuous, by “Continuous Function

Theorem for Sequences”, we get that an = e ln an → e 2 . The sequence {an }
converges to e 2 .

Factorial Notation
The notation n! (“n factorial”) means the product 1.2.3. . . . n of the

integers from 1 to n. Notice that (n + 1)! = (n + 1).n!. Thus,
4! = 1.2.3.4 = 24 and 5! = 1.2.3.4.5 = 5.4! = 120.
We define 0! to be 1. Factorials grow even faster than exponential, as

the table suggests.
n e n (rounded) n!
1 3 1
5 148 120
10 22, 026 3, 628, 800
20 4.9x108 2.4x1018

√
Using ε − N definition, prove that lim n
n = 1.
n→∞
√
Let n
n = 1 + hn where hn ≥ 0. Then for all n, (since hn ≥ 0), we have
n(n − 1) 2 n(n − 1) 2
n = (1 + hn )n = 1 + nhn + hn + · · · + hnn > hn .
2 2
q
2 2
Hence hn2 < n−1 for all n ≥ 2, so |hn | < n−1 , for all n ≥ 2.
Let ε > 0 be given. Then

r
2
|hn | < < ε, when n > 1 + 2/ε2 .
n−1
For any positive integer N greater than 1 + 2/ε2 , we have

√
| n n − 1| = |hn | < ε, for all n > N.
√
Hence n
n → 1.
Using ε − N definition, prove that lim x 1/n = 1 (x > 0).
n→∞
Case : 0 < x < 1

√
Let y = 1/x. Since x < 1, y > 1. Let n y = 1 + hn , when hn > 0.
n(n + 1) 2
y = (1 + hn )n = 1 + nhn + hn + · · · + hnn
2
> 1 + nhn (since hn > 0), for all n.
y −1
So hn < n , for all n.
y −1 y −1
Let ε > 0 be given. Then |hn | < n < ε, when n > ε .
y −1 1/x−1
For any positive integer N greater than, ε = ε , we have
√
| n y − 1| = |hn | < ε, for all n > N.

Using ε − N definition, prove that lim x 1/n = 1 (x > 0).
n→∞
Case : x > 1
By the argument given in the above case, for any given ε > 0, there is an
integer N greater than x−1
ε , we have
√
| n x − 1| < ε, for all n > N.
Case : x = 1
The case is trivial, since {x 1/n } is the 1-constant sequence, converging to

1.
For given ε > 0, any positive integer N satisfying
n > N ⇒ |1 − 1| = 0 < ε.

Using ε − N definition, prove that lim x n converges only
n→∞
when −1 < x ≤ 1.
Exercise 35.
Prove that lim x n converges to 0 when |x| < 1.
n→∞
Proof. Case : |x| < 1

1 1 1
Let |x| = 1+h where h > 0. Therefore |x|n = (1+h)n ≤ 1+nh , for all n.
1 (1/ε−1)
Let ε > 0 be given. Then 1+nh <ε when n > h .
(1/ε−1)
For any positive integer N greater than h , we have
1
|x n − 0| = < ε, for all n > N.
1 + nh
Hence, {x n } converges to 0.
n→∞
when −1 < x ≤ 1.
Case : x = 1
When x = 1, evidently x n = 1. Therefore the sequence converges to 1.
Case : x > 1
Let x = 1 + h, h > 0. Then x n = (1 + h)n > 1 + nh, for all n.
Let M be a positive number (however large) such that 1 + nh > G .
For any G > 0, there is a positive integer N such that x n > G , for all
n > N. Hence the sequence diverges to ∞.
Case : x = −1
When x = −1, the equation {(−1)n } oscillates finitely.

n→∞
when −1 < x ≤ 1.
Case : x < −1
Let x = −y so that y > 1.
Thus we get the sequence {(−1)n y n } which have both positive and
negative terms.
The sequence is unbounded and the numerical values of the terms can be
made greater than any number (however large). Thus, it oscillates
infinitely.
Hence the sequence {x n } converges only when −1 < x ≤ 1.

x n
Using L’Hôpital‘s Rule, prove that lim 1+ = ex
n→∞ n
(any x).
x n x n
lim 1+ = exp lim ln 1 +
n→∞ n
n→∞
n x
= exp lim n ln 1 +
n→∞ n!
x
ln(1 + n ) h∞ i
= exp lim 1
form
n→∞
n
∞
1 −x
1+ xn · · n2
= exp lim −1
n→∞
n2
x
= exp lim x
n→∞ 1 +
n
= ex

xn
Using ε − N definition, prove that lim =0 (any x).
n→∞ n!
Let ε > 0 be given. First we choose a positive integer N such that
N ≥ 2|x|.
|x|N |x| 1
Let K = . Note that ≤ , for all K ≥ N.
N! K 2
Now we can choose a positive integer N1 > N such that
1 N1 −N
K < ε.
2
Then for all n ≥ N1 , we have
|x|n |x|n−N |x|N 1 n−N

= ≤ K < ε.
n! n(n − 1) · · · (n − N + 1) N! 2

Recursive Definitions
So far, we have calculated each an directly from the value of n.
But sequences are often defined recursively by giving
1. The value(s) of the initial term or terms, and

2. A rule, called a recursion formula, for calculating any later term
from terms that precede it.

Recursive Definitions
13th-century Italian mathematician Fibonacci posed the following problem,

concerning the breeding of rabbits :
Suppose that rabbits live forever and that every month each pair
produces a new pair which becomes productive at age 2 months.
If we start with one newborn pair, how many pairs of rabbits will we
have in the nth month ? The answer can be given in a recursion
formula.
The statements a1 = 1, a2 = 1, and an+1 = an + an−1 define the sequence

1, 1, 2, 3, 5, . . . of Fibonacci numbers. With a1 = 1 and a2 = 1, we have
a3 = 1 + 1 = 2, a4 = 2 + 1 = 3, a5 = 3 + 2 = 5, and so on.

Sequence Constructed Recursively
Example 36.
(a) The statements a1 = 1 and an = an−1 + 1 define the sequence
1, 2, 3, . . . , n, . . . of positive integers. With a1 = 1, we have
a2 = a1 + 1 = 2, a3 = a2 + 1 = 3, and so on.
(b) The statements a1 = 1 and an = n· an−1 define the sequence
1, 2, 6, 24, . . . , n!, . . . of factorials. With a1 = 1, we have
a2 = 2· a1 = 2, a3 = 3· a2 = 6, a4 = 4· a3 = 24, and so on.

Bounded Nondecreasing Sequence
The terms of a general sequence can bounce around, sometimes getting

larger, sometimes smaller. An important special kind of sequence is one
for which each term is at least as large as its predecessor.
Definition 37 (Nondecreasing Sequence).

A sequence {an } with the property that an ≤ an+1 for all n is called a
nondecreasing sequence.
Example 38 (Nondecreasing Sequence).

(a) The sequence 1, 2, 3, . . . , n, . . . of natural numbers.
(b) The sequence 12 , 23 , 34 , . . . , n+1
n
, . . ..
(c) The constant sequence {3}.
There are two kinds of nondecreasing sequences – those whose terms
increase beyond any finite bound and those whose terms do not.

Bounded, Upper Bound, Least Upper Bound
Definition 39.
A sequence {an } is bounded from above if there exists a number M such
that an ≤ M for all n. The number M is an upper bound for {an }. If M is
an upper bound for {an } but no number less than M is an upper bound
for {an }, then M is the least upper bound for {an }.
Example 40 (Applying the Definition for Boundedeness).

(a) The sequence 1, 2, 3, . . . , n, . . . has no upper bound.
(b) The sequence 12 , 23 , 34 , . . . , n+1
n
, . . . is bounded from above by M = 1.
No number less than 1 is an upper bound for the sequence, so 1 is the
least upper bound, from the following exercise.

The sequence {n/(n + 1)} has the least upper bound of 1.
Exercise 41.
Show that if M is a number less than 1, then the terms of {n/(n + 1)}
eventually exceed M.
Proof
M
Let 0 < M < 1 and N be an integer greater than 1−M .
Then
M n
n > N =⇒ n > =⇒ > M.
1−M n+1
n
Since n+1 < 1 for every n, this proves that 1 is the least upper bound for
n o
n
the sequence n+1 .

Bounded Nondecreasing Sequence
A nondecreasing sequence that is bounded from above always has the least
upper bound. This is the completeness property of the real numbers,
We will prove that if L is the least upper bound then the sequence
converges to L.
Theorem 42 (Nondecreasing Sequence Theorem).

A nondecreasing sequence of real numbers converges if and only if it is
bounded from above. If a nondecreasing sequence converges, it converges
to its least upper bound.
The above theorem implies that a nondecreasing sequence converges when

it is bounded from above. It diverges to infinity if it is not bounded from
above.

Monotonic Sequence Theorem
Theorem 43 (Monotonic Sequence Theorem).
If a sequence {an } is both bounded and monotonic, then the sequence
converges.

Proof of Nondecreasing Sequence Theorem / Monotonic
Sequence Theorem
Suppose we plot the points (1, a1 ), (2, a2 ), . . . , (n, an ), . . . in the xy-plane.
If M is an upper bound of the sequence, all these points will lie on below
the line y = M.
If the terms of a nondecreasing sequence have an upper bound M, they have a limit L ≤ M.

Proof of Nondecreasing Sequence Theorem / Monotonic
Sequence Theorem (contd...)
The line y = L is the lowest such line. None of the points (n, an ) lies
above y = L, but some do lie above any lower line y = L − ε, if ε is a
positive number. The sequence converges to L because
(a) an ≤ L for all values of n and

(b) Given any ε > 0, there exists at least one integer N for which
aN > L − ε.
The fact that {an } is nondecreasing tells us further that an ≥ an > L − ε,

for all n ≥ N. Thus, all the numbers an beyond the Nth number lie within
the ε-band of L. This is precisely the condition for L be the limit of the
sequence {an }.

Bounded, Lower Bound, Greatest Lower Bound
Definition 44.
A sequence of numbers {an } in which an ≥ an+1 for every n is called a
nonincreasing sequence. A sequence {an } is bounded from below if there
is number M with M ≥ an for every n. Such a number M is called a lower
bound for the sequence.
If m is a lower bound for {an } but no number greater than m is a lower
bound for {an }, then m is the greatest lower bound for {an }.
Theorem 45 (Nonincreasing Sequence Theorem).

A nonincreasing sequence of real numbers converges if and only if it is
bounded from below. If a nonincreasing sequence converges, it converges
to its greatest lower bound.
Proof is similar to the proof of Nondecreasing Sequence Theorem.

When to apply Nondecreasing Sequence Theorem /
Monotonic Sequence Theorem ?
To apply Nondecreasing Sequence Theorem / Monotonic Sequence
Theorem to the sequence {an }, we write first few terms of the sequence
{an } and observe the pattern.
Then using mathematical induction, we may prove that it is increasing /

decreasing and bounded from above / below.

Observations
1. The necessary and sufficient condition for the convergence of a

monotonic sequence is that it is bounded.
2. Every monotonic sequence either converges or diverges and is never
oscillatory.
3. Every monotonically increasing sequence which is not bounded above
diverges to +∞.
4. Every monotonically decreasing sequence which is not bounded below
diverges to −∞.

Exercise
Exercise 46.
Show that the sequence {an } defined by the recursion formula
√
an+1 = 3an , a1 = 1
converges to 3.

Solution
√
Clearly, 3 = a2 ≥ a1 = 1.
√ √
Suppose am+1 ≥ am =⇒ 3am+1 ≥ 3am =⇒ am+2 ≥ am+1 .
Thus by mathematical induction, an+1 ≥ an , for all n.

p √
Clearly, a1 < 3, a2 < 3, a3 < 3 3 < 3. Use can use powers of 3.
Again, by mathematical induction,
0 < an < 3, for all n.
Hence by nondecreasing sequence theorem, {an } converges.

Solution (contd...)
Suppose that {an } converges to L. Then {an+1 } also converges to L.

√ √
Since an+1 = 3an , as n → ∞ we get L = 3L.
So, L = 0, or, L = 3.
Since an ≥ 1, for all n, 0 cannot be a limit.
Thus {an } → 3, as n → ∞.

Examples
Example 47.
Some examples of convergent sequences are given below.
ln(n2 )
1. n = 2 lnn n → 2.0 = 0
√n 2
2. n = n2/n = (n1/n )2 → (1)2 = 1
√
3. 3n = 31/n (n1/n ) → 1.1 = 1
n
n
4. − 12 → 0
n n
5. n−2
n = 1 + −2n → e −2
100n
6. n! →0

Exercises
Exercise 48 (Finding Limits).
Which of the sequences {an } converge, and which diverge? Find the limit
of each convergent sequence.
1. an = 2 + (0.1)n
2n+1
2. an = √
1−3 n
1−5n4
3. an = n4 +8n3
1−n3
4. an = 70−4n2

1 1
5. an = 2 − 2n 3+ 2n

Solution
1. an = 2 + (0.1)n → 2, hence it converges.

2n+1
2. an = √
1−3 n
→ −∞, hence it diverges.
1−5n4
3. an = n4 +8n3
→ −5, hence it converges.
1−n3
4. an = 2 → ∞, hence it diverges.
70−4n
5. an = 2 − 21n 3 + 21n → 6, hence it converges.

Exercises
Exercise 49 (Finding Limits).
q
2n
1. an = n+1
sin2 n
2. an = 2 n
√
n
3. an = 10n
4. an = ln n − ln(n + 1)
n!
5. an = nn (Hint: Compare with 1/n.)

Solution
q
2n
√
1. lim an = lim n+1 = 2.
sin2 n 1
2. 0 ≤ 2n ≤
, by Sandwich theorem, {an } converges to 0.
n
2
√ √
n
3. lim an = lim 10 n n → 1.1 = 1, so {an } converges to 1.

n
4. lim an = lim ln n+1 → ln 1 = 0, so {an } converges to 0.
1.2.3···n
5. 0 ≤ lim n.n.n···n ≤ lim n1 , by Sandwich theorem, {an } converges to 0.

Exercise
Exercise 50.
n!
1. an = 2n .3n
1 1/(ln n)

2. an = n
n 1/n
x
3. an = 2n+1 ,x > 0
4. an = tanh n
5. an = √1
n
tan−1 n
(ln n)200
6. an = n
1√
7. an = √
n2 −1− n2 +n
1 n 1
R
8. an = n 1 x dx
Rn 1
9. an = 1 x p dx, p > 1

Solution
1. lim 2nn!.3n = lim 6n1/n! = ∞, diverges.

1 1/(ln n)
2. = lim exp ln1n ln( n1 ) = lim exp ln 1−ln n
= e −1 , converges.

lim n ln n
n 1/n 1/n
−2
3. x
lim 2n+1 1
= lim x 2n+1 = x exp lim (2n+1) 2 = xe 0 = x, x > 0, converges.
−e n −ne −1 2n
4. lim tanh n = lim ee n +e −n = lim e 2n +1 = 1, converges.
5. lim √1n tan−1 n = 0. π2 = 0, converges.

200 199
6. lim (ln n)
n = lim 200(lnn n) = · · · = lim 200!
n = 0, converges.
√ √
2 2
7. lim √n2 −1−1 √n2 +n = lim n −1+−1−n
n +n
= −2, converges.
n
8. lim n1 1 x1 dx = lim lnnn = 0, converges.
R
Rn h in h i
9. lim 1 x1p dx = lim p−1 1
x
1
p−1 = lim 1
p−1 n
1
p−1 − 1 1
= p−1 , p > 1,
1
converges.

Exercise
Exercise 51.
Assume that each of the following sequences is recursively defined. Find
its limit.
1. a1 = 1, an+1 = aann +2
+6
√
2. an+1 = 8 + 2an . Discuss the limit when a1 = −4 and a1 = 0.
3. 2, 2 + 21 , 2 + 1
2+ 12
,2 + 2+
1
1 ,···
2+ 1
2
r
√ p √ √ √
q p q p
4. 3, 3 + 3, 3 + 3 + 3, 3 + 3 + 3 + 3, · · ·

Solution
1. Since {an } converges, the limit is 2.

2. Since {an } converges, the limit is 4 for a1 = −4 and the limit is 0 for
a1 = 0.
3. By induction, we have an ≥ 2 for all n ≥ 1.
√ Since {an } is Cauchy, the
sequence {an } converges, the limit is 1 + 2.
Some observations : By looking at the terms of the sequence, we
have the following observations :
(a) odd terms form an increasing sequence
(b) even terms form a decreasing sequence
(c) all odd terms less than all even terms
So both odd term subsequence and even term subsequence converge
to the same limit.
4. By induction, an ≤ 3 for all n ≥√1 and an ≤ an+1 for all n ≥ 1, hence
{an } converges. The limit is 1+2 13 .

Exercise
Exercise 52.
The first term of a sequence is x1 = 1. Each succeeding term is the sum of
all those that come before it:
xn+1 = x1 + x2 + · · · + xn .
Write out enough early terms of the sequence to deduce a general formula
for xn that holds for n ≥ 2.

Solution
The terms are
1, 1, 2, 4, 8, 16, 32, . . . = 1, 20 , 21 , 22 , 23 , . . . .
=⇒ x1 = 1 and xn = 2n−2 , for n ≥ 2.

Exercise
Exercise 53.
A sequence of rational numbers is described as follows:
1 3 7 17 a a + 2b
, , , ,..., , ,....
1 2 5 12 b a+b
Here the numerators form one sequence, the denominators form a second
sequence, and their ratios form a third sequence. Let xn and yn be,
respectively, the numerator and the denominator of the nth fraction
rn = xn /yn .
(a) Verify that x12 − 2y12 = −1, x22 − 2y22 = +1 and, more generally, that
if a2 − 2b 2 = −1 or +1, then (a + 2b)2 − 2(a + b)2 = +1 or −1,
respectively.
(b) The fractions rn = xn /yn approach a limit as n increases. What is
that limit? (Hint: Use part (a) to show that rn2 − 2 = (1/yn )2 and
that yn is not less than n.)

Solution
Let f (a, b) = (a + 2b)2 − 2(a + b)2 = 2b 2 − a2 .
(a) x12 − 2y12 = −1 and x22 − 2y22 = +1 are verified for r1 = xy11 = 11 and
r2 = xy22 = 32 .
If a2 − 2b 2 = −1 or +1, then f (a, b) = +1 or −1, respectively.
2 2 −2b 2 )
(b) rn2 − 2 = a+2ba+b − 2 = − (a(a+b) a
2 . Note that b is the predecessor of
a+2b
a+b .
If a2 − 2b 2 = −1 or +1, then (a + 2b)2 − 2(a + b)2 is 1 or −1. Hence
(a2 − 2b 2 ) ±1
rn2 − 2 = − =
(a + b)2 yn
q
which implies that rn = 2 ± ( y1n )2 .
1
√
Verifty that yn ≥ n for all n, hence yn → 0 as n → ∞. Thus rn → 2.

Sequences Generated by Newton’s Method
Newton’s method, applied to a differentiable function f (x), begins with a

starting value x0 and constructs from it a sequence of numbers {xn } that
under favorable circumstances converges to a zero of f . The recursion
formula for the sequence is
f (xn )
xn+1 = xn − .
f 0 (xn )
For a > 0, one can easily show that the recursion formula for
f (x) = x 2 − a can be written as
a
xn + xn
xn+1 = .
2

Sequences Generated by Newton’s Method
Exercise 54.
The following sequences come from the recursion formula for Newton’s
method,
f (xn )
xn+1 = xn − 0 .
f (xn )
Does the sequence converge? If so, to what value? In each case, begin by
identifying the function f that generates the sequence.
1. x0 = 1, xn+1 = xn − [(sin xn − xn2 )/(cos xn − 2xn )]
xn2 −2 xn 1
2. x0 = 1, xn+1 = xn − 2xn = 2 + xn
tan xn −1
3. x0 = 1, xn+1 = xn − sec2 xn
4. x0 = 1, xn+1 = xn − 1.

Solution
1. The sequence converges to a solution of the equation sin x − x 2 = 0.

√
2. f (x) = x 2 − 2, the sequence converges to 2.
3. f (x) = tan x − 1, the sequence converges to 0.7853981 ≈ π/4.
4. f (x) = e x , the sequence diverges.

Exercise
Exercise 55.
1. Suppose that f (x) is differentiable for all x in [0, 1] and that f (0) = 0.
Define the sequence {an } by the rule an = nf (1/n). Show that
lim an = f 0 (0).
n→∞
Use the above result to find the limits of the following sequences:
(a) an = n tan−1 n1
(b) an = n(e 1/n − 1)
(c) an = n ln(1 + n2 )

Solution
f (∆x) f (0 + ∆x) − f (0)
1. lim nf (1/n) = lim = lim = f 0 (0),
n→∞ ∆x→0 ∆x ∆x→0 ∆x
where ∆x = n1 .
(a) lim n tan−1 n1 = f 0 (0) = 1, where f (x) = tan−1 x.
(b) lim n(e 1/n − 1) = f 0 (0) = e 0 − 1, where f (x) = e x − 1.
(c) lim n ln(1 + n2 ) = f 0 (0) = 2, where f (x) = ln(1 + 2x).

Pythagorean Triples
Exercise 56.
A triple of positive integers a, b, and c is called a Pythagorean triple if
2
a2 + b 2 = c 2 . Let a be an odd positive integer and let b = b a2 c and
2
c = d a2 e be, respectively, the integer floor and ceiling for a2 /2.
1. Show that a2 + b 2 = c 2 . (Hint: Let a = 2n + 1 and express b and c
in terms of n.)
2. By direct calculation, or by appealing to the figure here, find
2
ba c
lim a22 .
a→∞ d e
2

Solution
2 2
1. If a = 2n + 1, then b = b a2 c = 2n2 + 2n, c = d a2 e = 2n2 + 2n + 1,
hence a2 + b 2 = c 2 . (Verify!)
2. 2
b a2 c 2n2 + 2n
lim 2 = lim = 1.
a→∞ d a2 e n→∞ 2n2 + 2n + 1

Good Approximation for nth root of n!
Exercise 57.
Scottish mathematician James Stirling (1692-1770) showed that
n n √
n! ≈ 2nπ. (2)
e
Show that
lim (2nπ)1/(2n) = 1
n→∞
and hence, using Stirling’s approximation (2), that

√
n n
n! ≈
e
for large values of n.

Solution
ln 2nπ
lim (2nπ)1/(2n) = lim exp → e 0 = 1.
n→∞ n→∞ 2n
From (2), we get that

√
n n n
n! ≈ (2nπ)1/2n ≈ ,
e e
for large value n.

Exercise
Exercise 58.
1. Assuming that lim (1/nc ) = 0 if c is any positive constant, show that
n→∞
ln n
lim =0
n→∞ nc
if c is any positive constant.
2. Prove that lim (1/nc ) = 0 if c is any positive constant. (Hint : if
n→∞
ε = 0.001 and c = 0.04, how large should N be to ensure that
|1/nc − 0| < ε if n > N?)

Solution
ln n 1/n 1
1. lim = lim = lim = 0.
n→∞ nc n→∞ cnc−1 n→∞ cnc
2. For any ε > 0, there exists a positive integer N = e −(ln ε)/c such that
n > e −(ln ε)/c .
Hence ln n > − lnc ε =⇒ 1

nc < ε. Thus lim (1/nc ) = 0.
n→∞

The zipper theorem
Exercise 59.
Prove the ”zipper theorem” for sequences : If {an } and {bn } both
converge to L, then the sequence
a1 , b1 , a2 , b2 , . . . , an , bn , . . .
converges to L.

Solution
Define c2n = bn and c2n−1 = an , for n = 1, 2, . . ..
Since an → L and bn → L, there are positive integers N1 and N2

respectively such that
|an − L| < ε, for all n > N1
and
|bn − L| < ε, for all n > N2 .
If n > 1 + 2 max{N1 , N1 }, then
|cn − L| < ε, for all n > N.
Hence {cn } converges to L.

Exercise
Exercise 60.
Which of the following sequences converge and which diverge?
2 1
1. an = 2 − n − 2n
n+1
2. an = ((−1)n + 1)

n

Solution
2 1 2 1
1. Let’s assume an ≤ an+1 for all n. So 2 − n − 2n ≤2− n+1 − 2n+1
which gives that for all n
2 1
≥ − n+1 .
n(n + 1) 2
Hence {an } is non-decreasing. Also 2 is an upper bound. Thus {an }

converges to the lub.
2. (
0 if n is odd
an = 1
2(1 + n ) if n is even
so {an } diverges.

Exercise
Exercise 61.
The first term of a sequence is x1 = cos(1). The next terms are x2 = x1 or
cos(2), whichever is larger, and
x3 = x2 or cos(3),
whichever is larger (farther to the right). In general,
xn+1 = max{xn , cos(n + 1)}.

Solution
Verify that
xn = max{cos 1, cos 2, . . . , cos n}.
By the definition of maximum,
xn+1 ≥ xn
for all n.
Hence {xn } is non-decreasing and bounded from above by 1, so it

converges.

Exercise
Exercise 62.
Which of the sequences converge and which diverge?
√
1 + 2n
1. an = √
n
n+1
2. an =
n
4n+1 + 3n
3. an =
4n
4. a1 = 1, an+1 = 2an − 3

Solution
√
1. Verify that {an } is nonincreasing and bounded from below by 2. So
it converges to glb.
2. Verify that {an } is nonincreasing and bounded from below by 1. So it
converges to glb.
3. Verify that {an } is nonincreasing and bounded from below by 4,
hence it converges to glb.
4. Verify that an = −2n + 3 and {an } is nonincreasing but not bounded
from below, it diverges.

Uniqueness of least upper bounds
Exercise 63.
1. Show that if M1 and M2 are least upper bounds for the sequence
{an }, then M1 = M2 . That is, a sequence cannot have two different
least upper bounds.
2. Is it true that a sequence {an } of positive numbers must converge if it
is bounded from above? Give reasons for your answer.

Solution
1. Use the fact that the least upper bound is also an upper bound.
2. No. The sequence {1, 2, 3, 1, 2, 3, . . .} is bounded from above but it
does not converge. However, any increasing sequence which is
bounded from above will converge.

Cauchy Sequence
Definition 64.
A sequence {an } is said to be a Cauchy sequence if to every positive
number ε there corresponds an integer N such that for all m and n,
m > N and n > N =⇒ |am − an | < ε.
Equivalently, {an } is said to be a Cauchy sequence if given ε > 0, there

exists a positive integer m such that |am+p − am | < ε, for all p ≥ 0, p ∈ N.
Exercise 65.
Prove that every convergent sequence is a Cauchy sequence.
That is, prove that if {an } is a convergent sequence, then to every positive
number ε there corresponds an integer N such that for all
m > N and n > N =⇒ |am − an | < ε.

Solution
Let {an } be a convergent sequence, converging to some L.
As an → L, there is a positive integer N such that
|an − L| < ε/2 for all n > N.
Hence for all n, m > N, we have
|an − am | ≤ |an − L| + |am − L| < ε/2 + ε/2.
Thus {an } is a Cauchy sequence.

Cauchy’s general principle of convergence
We proved that every convergent sequence is a Cauchy sequence. What

about the converse ? The converse is also true, which is shown in the
following result (without proof).
Theorem 66 (Cauchy’s general principle of convergence).

A necessary and sufficient condition for the convergence of a sequence
{an } is that, for each ε > 0, there exists a positive integer N such that
|am − an | < ε, for all n, m > N.

How is “Cauchy’s general principle of convergence” useful?
We can apply definition to test convergence of a sequence to a given limit

L. But in case, the limit L is not known, nor can any guess be made
of the same, the above theorem can be used. It involves only the
terms of the sequence and is useful for determining whether a sequence
converges or not.
Exercise
Exercise 67.
Show that the sequence {an }, where
1 1
an = 1 + + ··· +
2 n
cannot converge.

Solution
Suppose that the sequence {an } converges. For ε = 12 , there is a positive

integer N such that
1
|am − an | < , for all n, m > N.
2
But
1 1 1 1 1
|am − a2m | = + + ··· + > m=
m+1 m+2 2m 2m 2
which contradicts that “every convergent sequence is a Cauchy sequence”.
Hence the sequence cannot converge.

Exercise
Exercise 68.
1 1
an = + ··· +
(n + 1)2 (2n)2
converges to 0.

Solution
n n
2
≤ an ≤ 2 for all n.
(2n) n
1 1
Hence 4n ≤ an ≤ n for all n.
By Sandwich theorem, {an } converges to 0.

Exercise
Exercise 69.
1 1 1
an = + + ··· +
n+1 n+2 2n
is convergent.

Solution
1
an+1 − an = 2(n+1)(2n+1) > 0 for all n.
So, the sequence {an } is nondecreasing.

1 1 1
Also 0 < an < n + n + ··· + n = 1, which implies that {an } is bounded.
Thus {an } converges.

Exercise
Exercise 70.
1 1 1
an = + + ··· +
1! 2! n!
is convergent.

Solution
1
an+1 − an = (n+1)! > 0 for all n.
So, the sequence {an } is nondecreasing.

1 1 1 1
Also 0 < an < 1 + 2 + 22
+ ··· + 2n−1
=2− 2n−1
< 2, which implies that
{an } is bounded.
Thus {an } converges.

A recursive definition of π/2
If we start with x1 = 1 and define the subsequent terms of {xn } by the rule
xn = xn−1 + cos xn−1 ,
we generate a sequence that converges to π/2.

Subsequences
If the terms of one sequence appear in another sequence in their given

order, we call the first sequence a subsequence of the second.
From a sequence {an }, if we pick up only the terms whose suffices are
n1 , n2 , n3 . . . and generate a new sequence, namely, {an1 , an2 , an3 , . . . }
mainitaining the same order as in {an }, then this new sequence {ank } is
called a subsequence of {an }.
Definition 71.
Let {an } be a given sequence. If {nk } is a strictly increasing sequence of
natural numbers (i.e., n1 < n2 < n3 < · · · ), then {ank } is called a
subsequence of {an }.

Subsequences
Example 72.
1. The sequences {2, 4, 6, . . . }, {1, 3, 5, . . . }, {1, 4, 9, 16, . . . } are some
subsequences of {n}.
2. The sequences {1, 313 , 315 , . . . }, { 312 , 314 , 316 , . . . } are some
subsequences of { 31n }.
3. The sequence {2n } is a subsequence of {n}, whereas the sequence
{4, 3, 10, 11, 12, . . .} is not a subsequence of {n}.

Properties of Subsequences
1. The terms of a subsequence occur in the same order in which they

occur in the original sequence.
2. Every sequence is a subsequence of itself.
3. The sequence {a2k } formed by taking the terms of even index from
{an } is called the even subsequence of {an } and the sequence
{a2k+1 } formed by taking the terms of odd index from {an } is called
the odd subsequence of {an }.
4. If {un } is a subsequence of {an } and {vn } is a subsequence of {un },
then {vn } is also a subsequence of {an }.
5. The interval between various terms of a subsequence need not be
regular.
6. Given a term am of the sequence {an }, there is a term of a
subsequence following it.

Theorem 73.
If a sequence {an } converges to A, then every subsequence of {an }
converges to A.
Proof : Let {an } converge to A and {ank } be a subsequence of {an }.

Claim: {ank } converges to A.
That is, to prove that given any ε > 0, there exists a positive integer N such that
|ank − A| < ε, ∀nk ≥ N.
Let ε > 0 be given. Then, since an → A as n → ∞, corresponding to this ε there exists N ∈ N

such that
|an − A| < ε, ∀n ≥ N. (3)
Choose nk0 ∈ N such that nk0 ≥ N and ank0 is an element of {ank }.
Then, for any nk ≥ nk0 , we have nk ≥ nk0 ≥ N =⇒ nk ≥ N.
Therefore, from (3) we have,
|ank − A| < ε, ∀nk ≥ N.
Since ε is arbitrary, it follows by definition that ank → A, as nk → ∞.

1. The converse of Theorem (73) is not true. That is, if a subsequence

or even infinitely many subsequences of a given sequence converge,
then it does not mean that the original sequence should converge.
For example, the sequence {(−1)n } does not converge. But, its odd
and even subsequences converge to −1 and 1 respectively.
2. However, if all subsequences of a given sequence {an } converge to the
same limit A, only then {an } converges to A.
3. Based on the above points, to prove that a given sequence is not
convergent, it is sufficient to show that two of its subseqeunces
converge to different limits.
4. Every sequence contains a monotonic (increasing / decreasing)
subsequence.

Exercise
Exercise 74.
Prove that if two subsequences of a sequence {an } have different limits
L1 6= L2 , then {an } diverges.
Solution : Let k(n) and p(n) be two order-preserving functions whose

domains are the sets of positive integers and whose range are subsets of
the positive integers.
Consider two subsequences ak(n) and ap(n) converging to L1 and L2

respectively. So
|ak(n) − ap(n) | → |L1 − L2 | > 0.
Hence, there is no positive integer N such that for all n, m > N such that
|an − am | < ε. Thus {an } is not a Cauchy sequence.
Since every convergent sequence is a Cauchy sequence, so {an } is not a

convergent sequence, thus it diverges.
Exercise
The following result says that to prove that a given sequence is

convergent, it is sufficient to show that its odd and even subsequences
converge to the same limit.
Exercise 75.
For a sequence {an }, the terms of even index are denoted by a2k and the
terms of odd index by a2k+1 . Prove that if a2k → L and a2k+1 → L, then
an → L.
Solution : Let ε > 0 be given. As a2k → L and a2k+1 → L, there are positive integers N1 and
N2 such that
|a2k − L| < ε for all 2k > N1
and
|a2k+1 − L| < ε for all 2k + 1 > N2 .
Then for any n > max{N1 , N2 },
|an − L| < ε whether n is even or odd.
Hence an → L.
Subsequences of Divergent Sequences
1. {an } diverges to +∞ if and only if every subsequence of {an }

diverges to +∞ if and only if both the odd and even subsequences of
{an } diverge to +∞.
2. {an } diverges to −∞ if and only if every subsequence of {an }
diverges to −∞ if and only if both the odd and even subsequences of
{an } diverge to −∞.
3. Knowing that some subsequence of {an } diverges to +∞ (or −∞)
does not necessarily imply {an } diverges to +∞ (or −∞).
For example, for the sequence {an = n(−1)n }, the subsequence {a2n }
diverges to +∞ whereas {a2n−1 } diverges to −∞. But, {an } neither
diverges to +∞ nor to −∞.

Other Significant Properties of Sequences
Exercise 76.
1. If sequence {an } converges to A, then {|an |} converges to |A|.
2. If a sequence {an } converges to 0 and {bn } is a bounded sequence, then {an bn }
converges to 0.
3. {an } converges to 0 if and only if {|an |} converges to 0.
4. If an → A and an ≥ k for all n, where k is any constant, then A ≥ k.
5. If an → A and an ≤ k for all n, where k is any constant, then A ≤ k.
6. If an → A, bn → B and an ≤ bn , for all n, then A ≤ B.
7. If sequence {an } diverges to +∞, then {an } is bounded below, but unbounded above.
[But the converse is not true. That is, a sequence which is bounded below and unbounded
above need not diverge to +∞.]
8. If sequence {an } diverges to −∞, then {an } is bounded above, but unbounded below.
[But the converse is not true. That is, a sequence which is bounded above and
unbounded below need not diverge to −∞.]

Extra Problems
Difficult Level - High

Applications of Sequences
According to a front-page article in the December 15, 1992, issue of the Wall Street Journal,
Ford Motor Company used about 7 14 hours of labor to produce stampings for the average
vehicle, down from an estimated 15 hours in 1980. The Japanese needed only about 3 12 hours.
Ford’s improvement since 1980 represents an average decrease of 6% per year. If that rate
continues, then n years from 1992 Ford will use about Sn = 7.25(0.94)n hours of labor produce
stampings for the average vehicle. Assuming that the Japanese continue to spend 3 12 hours per
vehicle, how many more years will it take Ford to catch up? Find out two ways:
(a) Find the first term of the sequence {Sn } that is less than or equal to 3.5.
(b) Graph f (x) = 7.25(0.94)x and use Trace to find where the graph crosses the line y = 3.5.

Applications of Sequences - Compound Interest, Deposits,
and Withdrawals
If you invest an amount of money A0 at a fixed annual interest rate r compounded m times per year, and if the constant
amount b is added to the account at the end of each compounding period (or taken from the account if b < 0), then the
amount you have after n + 1 compounding period is
r

An+1 = 1+ An + b. (4)
m
(a) If A0 = 1000, r = 0.02015, m = 12, and b = 50, calculate and plot the first 100 points (n, An ), How much money is
in your account at the end of 5 years? Does {An } converge? Is {An } bounded?
(b) Repeat part (a) with An = 5000, r = 0.0589, m = 12, and b = −50.
(c) If you invest 5000 dollars in certificate of deposite (CD) that pays 4.5% annually, compounded quarterly, and you make
no further investments in the CD, approximately how many years will it take before you have 20,000 dollars? What if
the CD earns 6.25%
(d) It can be shown that for any k ≥ 0, the sequence defined recursively by Equation (4) satisfies the relation
k
r mb mb

Ak = 1+ A0 + − . (5)
m r r
For the values of the constants A0 , r , m, and b given in part (a), validate this assertion by comparing the values of the
first 50 terms of both sequences. Then show by direct substitution that the terms in Equation (5) satisfy the recursion
formula in Equation (4).

Applications of Sequences
The size of an undisturbed fish population has been modeled by the formula
bpn
pn+1 =
a + pn
where pn is the fish population after n years and a and b are positive constants that depend on
the species and its environment. Suppose that the population in year 0 is p0 > 0.
(a) Show that if {pn } is convergent, then the only possible values for its limit are 0 and b − a.
(b) Show that pn+1 < (b/a)pn .
(c) Use part (b) to show that if a > b, then limn→∞ pn = 0; in other words, the population
dies out.
(d) Now assume that a < b. Show that if p0 < b − a, then {pn } is increasing and
0 < pn < b − a. Show also that if p0 > b − a, then {pn } is decreasing and pn > b − a.
Deduce that if a < b, then
lim pn = b − a.
n→∞

Applications of Sequences - Logistic Difference Equation
The recursive relation
an+1 = ran (1 − an )
is called the logistic difference equation, and when the initial value a0 is given, the equation defines the logistic sequences {an }.
Throughout this exercise we chose a0 in the interval 0 < a0 < 1, say a0 = 0.3.
(a) Choose r = 3/4. Calculate and plot the points (n, an ) for the first 100 terms in the sequence. Does it appear to
converge? What do you guess is the limit? Does the limit seem to depend on your choice of a0 ?
(b) Choose several values of r in the interval 1 < r < 3 and repeat the procedures in part(a). Be sure to choose some
points near the endpoints of the interval. Describe the behavior of the sequences you observe in your plots.
(c) Now examine the behavior of the sequence for values of r near the endpoints of the interval 3 < r < 3.45 The
transition value r = 3 is called a bifurcation value and the new behavior of the sequence in the interval is called an
attracting 2-cycle. Explain why this reasonably describes the behavior.
(d) Next explore the behavior for r values near the endpoints of each of the intervals 3.45 < r < 3.54 and
3.54 < r < 3.55, Plot the first 200 terms of the sequences. Describe in your own words the behavior observed in your
plots for each interval. Among how may values does the sequence appear to oscillate for c interval? The values
r = 3.45 and r = 3.54 (rounded to two decimal places) are also called bifurcation values because the behavior of the
sequences changes as r crosses over those values.
(e) The situation gets even more interesting. There is actually an increasing sequence of bifurcation values
3 < 3.45 < 3.54 < · · · < cn < cn+1 · · · such that for cn < r < cn+1 the logistic sequence {an } eventually
oscillates steadily among 2n values, called an attracting 2n -cycle. Moreover, the bifurcation sequence {cn } is bounded
from above by 3.57 (so it converges). If you choose a value r < 3.57 you will observe a 2n -cycle of some sort. Choose
r = 3.5695 and plot 300 points.
(f) Let us see what happens when r = 3.57. Choose r = 3.65 and calculate and plot the first 300 terms of {an }. Observe
how the terms wander around in an unpredictable, chaotic fashion. You cannot predict the value of an+1 from previous
values of the sequence.
(g) For r = 3.65 choose two starting values of an that are close together, say, a0 = 0.3 and a0 = 0.301. Calculate and plot
the first 300 values of the sequences determined by each starting value. Compare the behaviors observed in your plots.
How far out do you go before the corresponding terms of our two sequences appear to depart from each other? Repeat
the exploration for r = 3.75. Can you see how the plots look different depending on your choice of a0 ? We say that the
logistic sequence is sensitive to the initial condition a0 .

Few More Properties of Sequences
1. Bolzano-Weierstrass theorem : Every bounded sequence of real

numbers has a convergent subsequence.
2. The set of limit points of a bounded sequence has the greatest and
the least members.
3. Every bounded sequence with a unique limit is convergent.
4. A necessary and sufficient condition for the convergence of a sequence
is that it is bounded and has a unique limit point.
np
5. If a > 0, and p is real, then lim = 0.
n→∞ (1 + a)n
an
6. If {an } converges and {bn } diverges, show that lim = 0 and
n→∞ bn
{an + bn } is divergent.
7. Given that lim an = a, lim bn = b and {Sn } and {Tn } are two
n→∞ n→∞
sequences, where Sn = max{an , bn }, Tn = min{an , bn }. Show that
the sequences {Sn } and {Tn } are convergent and that
lim Sn = max{a, b} and lim Tn = min{a, b}.
n→∞ n→∞
Exercises
1. Cauchy’s First Theorem on Limits:

a1 + a2 + · · · + an
Show that if lim an = L, then lim = L.
n→∞ n→∞ n
2. Using Cauchy’s first theorem on limits, prove the following :
(a) If a sequence {an } of positive
n terms converges o to a positive limit L,
1/n
then so does the sequence (a1 a2 . . . an ) of its geometric terms.
an+1
(b) If {an } be a sequence such that lim = L, where |L| < 1, then
n→∞ an
lim an = 0.
n→∞
an+1
(c) If {an } be a sequence such that lim = L, where L > 1, then
n→∞ an
lim an = ∞.
n→∞
m(m − 1)(m − 2) · · · (m − n + 1) n
3. Show that lim x = 0, |x| < 1.
n→∞ n!
4. We proved that a sequence {an } converges to L if and only if every
subsequence converges to L. Show that lim an = ∞ (−∞) if and
n→∞
only if every subsequence of {an } tends to ∞ (−∞).
Exercises
1. If {an } is a sequence such that

s
ab 2 + an2
an+1 = , b>0 for all n
a+1
and a1 = a > 0, then show that the sequence {an } converges to b.

2. If {an } is a sequence of positive real numebrs such that
an = an−1 +a2
n−2
, for all n ≥ 2, then show that {an } converges.
Find lim an .
n→∞
3. Show that the sequence {an } defined by
1 a
an = an + , for all n ≥ 1
2 an
and a1 = 0 converges to 3.

Exercises
1. If a sequence {an } is defined by an = 1+abn−1 , where

b > 0, a1 > 0, n ≥ 2, then show that the sequence converges to the
positive root of the equation
x 2 + x − b = 0.
2. Two sequence {an } and {bn } are defined inductively by a1 = 21 and

b1 = 1 and
p 1 1h 1 1 i
an = an−1 bn−1 and = + n = 2, 3, 4, . . . .
bn 2 an bn−1
Prove that an−1 < an < bn < bn−1 , n = 2, 3, 4, . . . and deduce that
both the sequences converge to the same limit L, where 12 < L < 1.

Extra Problems
Difficult Level - Average

Exercises
Find a formula for the general term an (the nth term) of the sequence
1. 1, 12 , 12 , 31 , 13 , 31 , 14 , 41 , 14 , 41 , 15 , . . . .
2. 0.9, 0.99, 0.999, 0.9999, . . . .
1 1
3. 2, 2 + 41 , 12 + 1
4 + 18 , 21 + 1
4 + 1
8 + 1
16 , . . . .
n o
4. 0, 1, 0, 0, 1, 0, 0, 0, 1, . . .
assuming that the pattern of the first few terms continues.

Exercises
n √ o
1. The limit of the sequence ln(4(n − n2 − n)) is .
2. Show that the sequence defined by
1
a1 = 2, an+1 =
3 − an
satifies 0 < an ≤ 2 and is decreasing. Deduce that the sequence is
convergent and find it its limit.
3. Consider two sequences {an } and {bn } given by an = (1/n)(sin(1/n))
and bn = (sin(1/n))(1/n) . Choose the correct answer.
(a) an → 0, bn → 0 (c) an → 1, bn → 1
(b) an → 0, bn → 1 (d) an → 1, bn → 0

Exercises
an 2
1. Let a1 = 1 and an+1 = 2 + an , n ≥ 1. Find the limit of the sequence
{an } if it exists.
2. Let p be a prime number. For each k, 0 ≤ k ≤ p − 1, the sequence
{anp+k } converges to 2 as n → ∞. Show that {an } converges to 2.
3. Let c and a1 be two positive real numbers. Consider the sequence
{an } defined by
1 c
an+1 = an +
2 an
for all n ∈ N. Show that an+1 ≤ an ∀ n ≥ 2 and {an } is bounded from
below. Find the limit of the sequence {an }.

Exercises
1. Find the limit of the following sequences whose nth term is given by
the formula
(−1)n 2n 2n2 + 3
(i) (ii) 2 (iii) 2
n+1 3n + 1 3n + 1
(Ans: (i) 0, (ii) 0, (iii) 2/3).
2. Show that the sequences given above converge to the corresponding
limits by ε − N definition.
3. Discuss the convergence of the sequence (an ) defined recursively by
(i) a1 = 1, an+1 = 2 − 3an , n = 1, 2, . . . (ii) a1 = 1 and
an
an+1 = 1+a n
, n = 1, 2, . . ..
(Ans: (i) divergent (ii) convergent)
4. Let a1 = 2, an+1 = 12 (an + a2n ), n = 1, 2, . . . Show that {an } is
√
decreasing and bounded below by 2.

Exercises
1. Find the limit of the sequence

r q
n√ q √ √ o
2, 2 2, 2 2 2, . . . .
Ans: 2.
1 n 3n + 1 n

2. Find the limit of (i) an = 1 + (ii) an = .
n 3n − 1
(Ans: (i) e, (ii) e 2/3 ).
n
x
3. For any real number x, show that converges.
n!

4. Show that log n c
n
→ 0 for any c > 0.

Exercises
1. Give an example of a continuous function f (x) and a sequence (an )

such that f (an ) converges but (an ) diverges.
2. Discuss the convergence of √
2 √
(i) sin2n n (ii) 2nn!3n (iii) nn!n (iv) n1/n (v) n − n + 1.
(Ans: (ii) divergent. (i),(ii),(iv),(v) convergent.)
3. Give an example of a sequence (an ) of positive numbers which
converges but the sequence (bn ) diverges where bn = aan+1
n
.
4. Let a1 = a, a2 = f (a1 ), a3 = f (a2 ) = f (f (a)), . . . , an+1 = f (an ),
where f is a continuous function. If lim an = L, show that f (L) = L.
n→∞

Exercises
1. Define the sequences {an } and {bn } as follows:
an + bn p
0 < b1 < a1 , an+1 = and bn+1 = an bn for n ∈ N.
2
Show that {an } and {bn } both tend to the same limit. This limit is
called the arithmetic-geometric mean of a1 and b1 .
2. Let the sequence {an } be defined by an = lim [x]+[2x]+...+[nx]
n2
, where
n→∞
x is a real number. Is this sequence convergent? If so, what is the
limit? (Ans: x/2)
3. Show that the sequence {(1 + 1/n)n } is a monotone increasing
sequence, bounded above.

Exercises
1. Let {bn } be a bounded sequence which satisfies the condition

bn+1 ≥ bn − 21n , n ∈ N. Show that the sequence {bn } is convergent.
2. For c > 2, the sequence {pn } is defined recursively by
p1 = c 2 , pn+1 = (pn − c)2 , n > 1. Show that the sequence (pn )
strictly increases.
[Hint : By induction, first prove that pn > 2c.]

References
1. M.D. Weir, J. Hass and F.R. Giordano, Thomas’ Calculus, 11th

Edition, Pearson Publishers.
2. N. Piskunov, Differential and Integral Calculus, Vol I & II (Translated
by George Yankovsky).
3. S.C. Malik and Savitha Arora, Mathematical Analysis, New Age
Publishers.
4. R. G. Bartle, D. R. Sherbert, Introduction to Real Analysis, Wiley
Publishers.

Sequences

Uploaded by

Copyright:

Available Formats

Sequences

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sequences

Uploaded by

Copyright:

Available Formats

Sequences

National Institute of Technology Karnataka (NITK)

P. Sam Johnson Sequences 1/190

We shall study a special class of functions (called sequences) whose

Analysis is based on the notion of a limit, a concept that can be defined in

With modern computers, such approximations can be made accurate

We shall discuss sequences of real numbers in a couple of lectures.

P. Sam Johnson Sequences 2/190

A sequence is a list of numbers

in a given order. Each of a1 , a2 , a3 and so on represents a number. These

The integer n is called the index of an , and indicates where an occurs in

as a function that sends 1 to a1 , 2 to a2 , 3 to a3 , and in general, the

P. Sam Johnson Sequences 3/190

2, 4, 6, 8, 10, 12, . . . , 2n, . . .

sends 1 to a1 = 2, 2 to a2 = 4, and so on. The general behavior of this

P. Sam Johnson Sequences 4/190

Order is important. The sequence 1, 2, 3, 4, . . . is not the same as the

P. Sam Johnson Sequences 5/190

P. Sam Johnson Sequences 6/190

Question : What is the difference between a set and a sequence?

Sequence always has a definite order of elements.

A set is a (well-defined) collection of distinct elements (that contains no

P. Sam Johnson Sequences 7/190

Sometimes the numbers in a sequence approach a single value as the index

P. Sam Johnson Sequences 8/190

On the other hand, sequences like

P. Sam Johnson Sequences 9/190

n > N =⇒ |an − L| < ε.

If no such number L exists, we say that {an } diverges.

In some textbooks, the following definition is used for convergence.

P. Sam Johnson Sequences 11/190

Let {an } be a sequence converging to a real number, say L.

Now we apply the definition of convergence of {an } to L for ε. We will get

n > N1 =⇒ |an − L| < ε.

Again we apply the definition of convergence of {an } to L for ε/99. We

n > N2 =⇒ |an − L| < ε.

Caution : We should not think that N2 is always more that 99N1 .

P. Sam Johnson Sequences 12/190

The following figure shows two ways to represent sequences

The first marks the first few points from a1 , a2 , a3 , . . . , an , . . . on the

P. Sam Johnson Sequences 13/190

Sequences can be represented as points on the real line or as points in the

P. Sam Johnson Sequences 14/190

Sequences can be represented as points on the real line or as points in

P. Sam Johnson Sequences 15/190

As discussed above, there are two ways to visualize a sequence of real

P. Sam Johnson Sequences 16/190

Let us consider the constant sequence {10}∞

n > N =⇒ |an − L| = |10 − 10| < ε.

In the plane : We get points on the line y = 10 defined over the

n > N =⇒ |an − L| = |10 − 10| < ε.

P. Sam Johnson Sequences 17/190

Some students may write the following incomplete line

for the definition of convergence which is not correct.

To prove our guess L is correct, we start with ε > 0, we find a positive

n > N =⇒ |an − L| < ε.

P. Sam Johnson Sequences 18/190

That is, if {an } is a sequence of real numbers and, if L1 and L2 are

In other words, a sequence cannot converge to more than one limit.

P. Sam Johnson Sequences 19/190