SMA 2301 REAL ANALYSIS I

1 Countable and Uncountable Sets

1.1 Equivalence of Sets
Definition 1.1.1: Let A and B be sets. If there exists a bijection (one-to-one and onto
function) f : A −→ B, then A and B can be put into a one-to-one correspondence. In this
case, we say that A and B are equivalent or equinumerous or have the same cardinality, and
we denote this by A ∼ B.
Proposition 1.1.1: Equivalence of sets is an equivalence relation.
Proof :
(a) Reflexivity: The identity map ι : A −→ A, ι (a) = a for all a ∈ A is a one-to-one and
onto mapping from A to itself. Hence A ∼ A.
(b) Symmetry: Suppose A ∼ B. Then there exists a bijection f : A −→ B. But every
bijection has an inverse which is also a bijection. In this case, f −1 is a bijection and
maps B to A. Hence B ∼ A.
(c) Transitivity: Suppose A ∼ B and B ∼ C. Then there exists bijections f : A −→ B
and g : B −→ C. But composition of two bijections is also a bijection. So g ◦ f is a
bijection and maps A to C. Hence A ∼ C. @
Examples:
1. Let E = 2N, the set of all even natural numbers. Then the function f : N −→ E
defined by f (n) = 2n for all n ∈ N is both one-to-one and onto. Hence E ∼ N.
2. Show that N ∼ Z.
Solution: Define a function f : N −→ Z by
(
n
2
if n is even
f (n) = −(n−1) .
2
if n is odd

Then f is a one-to-one and onto function from N to Z. Hence N ∼ Z.

1.2 Types of Sets

Definition 1.2.1: A set A is finite if A = ∅, i.e., A is empty or A ∼ Nn = {1, · · · , n}, i.e.,
A is equivalent to the set of the first n positive integers. In other words, a set is finite if it
has n elements where n is finite. A set is said to be infinite if it is not finite.
Definition 1.2.2: A set is A countable if A is finite or A ∼ N (i.e., A is equivalent to the
natural number set). If A ∼ N, then A is said to be countably infinite or denumerable. A set
A is said to be uncountable if it is not countable.
Example:
From a previous example N ∼ Z. Hence Z is countable (countably infinite).
Theorem 1.2.1: A set is countable if it is equivalent to a subset of N.

1
Example:
We can also show that Z is countable by showing that there exists a bijection from Z to a
subset of N. For instance, define g : Z −→⊆ N by
(
2x if x ≥ 0
g (x) = .
3−x if x < 0

Theorem 1.2.2: Every subset of a countably infinite set is countable.

Proof : Suppose B is a subset of a countably infinite set A. If B is finite, then B is countable,
from Definition 1.2.2. On the other hand, let B be infinite and let A = {a1 , a2 , · · · }. Suppose
n1 is the smallest subscript for which an1 ∈ B, n2 the next smallest, and so on. Then
B = {an1 , an2 , · · · }. The elements of B are thus labelled with 1, 2, · · · , and so B is countable.
@
Theorem 1.2.3: Any subset of a countable set is countable.
Proof : Suppose A is countable and B ⊂ A. Since A is countable, there is a bijection
′ ′
f : A −→ N. But then f (B) = N is a subset of N, and f is a bijection between B and N .
′
Hence B is countable (since B is equivalent to a subset N of N). @
Theorem 1.2.4: If B ⊂ A and B is uncountable, then so is A.
Proof : If A is countable, then so is B (by Theorem 1.2.3). So A is uncountable as well. @
∞
Theorem 1.2.5: If A1 , A2 , · · · are countable sets, then ∪ An is countable, i.e., the countable
n=1
union of countable sets is countable.
Proof : Write

A1 = a11 , a12 , a13 , · · ·



A2 = a21 , a22 , a23 , · · ·



..
.
An = {an1 , an2 , an3 , · · · }
..
.

so that ajk is the k th element of Aj (j = 1, 2, · · · ). Define the height of ajk to be j + k.

Then a11 is the only element of height 2; likewise a12 and a21 are the only element of height
3; and so on. Since for any integer m ≥ 2 there are only m − 1 elements of height m
∞
namely a1m−1 , a2m−2 , · · · , am−1


1 , we may arrange (count) the elements of ∪ An according to
n=1
their heights as
a11 , a12 , a21 , a13 , a22 , a31 , a14 , · · ·
removing any ajk that has already been counted. This counting scheme eventually counts
∞
every ajk . Hence ∪ An is countable. @
n=1

Example:
Prove that the set Q of rational numbers is countable.

2
 ∞
Proof : Clearly, Q = ∪ En where En = {0/n, ±1/n, ±2/n, ±3/n, · · · }. Now, each En is equiv-
n=1
alent to Z and is thus countable (since Z is countable). Since Q is the countable union of
countable sets, it is countable.
Further Examples:
1. The set R of real numbers is uncountable.
Proof : Suppose R is countable, i.e., R = {x1 , x2 , · · · }. Let
 
1 1
I1 = x1 − , x1 +
4 4
 
1 1
I2 = x2 − , x2 +
8 8
..
.
In = xn − 2−(n+1) , xn + 2−(n+1)

..
.

The length of In is 2−n so that the sum of all the lengths of In ’s is 2−1 +2−2 +2−3 +· · · = 1.
∞ ∞
But xn ∈ In so that R = ∪ {xn } ⊂ ∪ In . This means that the whole real line (whose
n=1 n=1
length is infinite) would be covered by (contained in) a union of intervals whose lengths
add up to 1. This is a contradiction. Therefore, R must be uncountable.
2. Let A = (0, ∞). Then A ∼ R since the function f : R −→ A defined by f (x) = ex is
both one-to-one and onto. Since R is uncountable and A ∼ R, then A is uncountable.
3. In general, any interval of real numbers is uncountable.
Theorem 1.2.6: Let A1 , A2 , · · · , An be a finite family of countable sets. Then the Cartesian
product A1 × A2 × · · · × An is countable.
Examples:
1. Z × Z is countable. To prove this
 consider the fuction f : Z × Z −→⊆ N defined by

2a 3b if a ≥ 0, b ≥ 0
5−a 7b

if a < 0, b ≥ 0
f (a, b) = .


11a 13−b if a ≥ 0, b < 0
 −a −b
17 19 if a < 0, b < 0

In this case, f is both one-to-one and onto, and so Z × Z is countable.

2. Q ∼ Z × N ⊂ Z × Z. From Example 1 above, Z × Z is countable. Since Z × N is a
subset of a countable set, it is countable, by Theorem 1.2.3. Since Q is equivalent to a
countable set, it is countable.
3. The set N × N is countable. This is so because the function f : N × N −→ N defined
by f (m, n) = 2m 3n is both one-to-one and onto.
4. If A and B are countable, then A × B is countable.
Proof : Since A and B are countable, there exist bijections f : N −→ A and g : N −→ B.
Define h : N × N −→ A × B by h (m, n) = (f (m) , g (n)). The function h in this case

3
 is both one-to-one and onto. Since N × N is countably infinite, there exists a bijection
k : N −→ N × N. Then l : N −→ A × B defined by l = h ◦ k is one-to-one and onto
(since composition of bijections is also a bijection). So the set A × B is countable.

2 Open and Closed Sets

2.1 Metric Spaces
Definition 2.1.1: Let X be a non-empty set. A metric or distance function or simply
distance on X is a function d : X × X −→ R (i.e., d is real-valued) with properties that:
M1 : d (x, y) ≥ 0 for all x, y ∈ X, with d (x, y) = 0 if and only if x = y (i.e., d is positive
definite).
M2 : d (x, y) = d (y, x) for all x, y ∈ X (i.e., d is symmetric).
M3 : d (x, y) ≤ d (x, z) + d (z, y) for all x, y, z ∈ X (triangular property).
Remark 2.1.1 :
(i ) Condition M1 follows from the other two since

2d (x, y) = d (x, y) + d (y, x) ≥ d (x, x) = 0.

(ii ) The name triangular property comes from the property of triangles where the sum of
the lengths of any two sides of a triangle is always greater than or equal to the length
of the remaining side.
(iii ) The distance function d on X is neither one-to-one (since d (x, y) = d (y, x) for all
x, y ∈ X) nor onto (since negatives in R do not have pre-images, by M1 ).
Definition 2.1.2: Let d be a metric on X. The pair (X, d) is called a metric space. Elements
in a metric space are called points.
Remark 2.1.2 : If it is clear what the distance function d on X is, we write the metric space
(X, d) as simply X.
Examples:
1. The most important examples of metric spaces are the Euclidean spaces Rn especially
R (the real line) and R2 (the complex plane). The distance d : Rn × Rn −→ R on Rn is
defined as follows:
Let x = (x1 , x2 , · · · , xn ) and y = (y1 , y2 , · · · , yn ) be points in Rn . Then

d (x, y) = |x − y|
= |(x1 − y1 , x2 − y2 , · · · , xn − yn )|
q
= (x1 − y1 )2 + (x2 − y2 )2 + · · · (xn − yn )2 (positive square root).

For instance,
(a) In R, d −1, 32 = −1 − 3
= | − 52 | = 25 .


2

(b) Let x = (x1 , x2 ) , y = (y1 , y2 ) ∈ R2 . Diagramatically, we have

4
 y

b x = (x1 , x2 )

b
y = (y1 , y2 )

Then

d (x, y) = |x − y|
= |(x1 − y1 , x2 − y2 )|
q
= (x1 − y1 )2 + (x2 − y2 )2 (positive square root).

2. Let x = (x1 , x2 , · · · , xn ) , y = (y1 , y2 , · · · , yn ) ∈ Rn and let d : Rn × Rn −→ R be defined

by
d (x, y) = max {|xi − yi |} .
1≤i≤n

Then d is a metric on Rn .
Note 2.1.1 : Every subset Y of a metric space X is a metric space with the same distance
function since if conditions M1 , M2 , and M3 hold for all x, y, z ∈ X, they also hold if we
restrict x, y, z to lie in Y .
Definition 2.1.3: Let X 6= ∅. A topology on X is a collection T of subsets of X such that
(i ) ∅, X ∈ T .
(ii ) The union of any number of members of T is in T .
(iii ) Finite intersection of members of T is in T .
The elements of T in this case are called open (T -open) subsets of X.
Examples:
1. The set of all open intervals of all real numbers is a topology in R.
2. Let X = {a, b, c}. Then P = {∅, {a} , {b} , {c} , {a, b} , {a, c} , {b, c} , X}, the power set
of X, is a topology on X.
Definition 2.1.4: Let x ∈ Rn and let r > 0. Then
(a) the open ball with centre at x and radius r is the set of all y ∈ Rn such that |y −x| < r.
(b) the closed ball with centre at x and radius r is the set of all y ∈ Rn such that |y−x| ≤ r.
Examples:
1. In R, balls are intervals.

5
 r1 r1 r2 r2

b b

x1 x2
Open Interval Closed Interval

Suppose x ∈ R and δ > 0. Then X = (x − δ, x + δ) is an open ball centered at x and

having radius δ since if y ∈ X, then x − δ < y < x + δ implying that −δ < y − x < δ
so that |y − x| < δ.
2. In R2 , balls are discs.

b
b b
b

b b

b b
b
b b
b b b
b
b
b b
b
b b
b
b b
b b
b
b b
b
b b
b b
b b b
b b
b b
b
b
b
b b
b b
b
b b b
b b
b b
b
b b
b
b
b
b

b b
b b
b b
b
b
b
b b

b b
b b b b
b
b
b
b
b b
b
b b
b
b
b
b
b
b
b
b

Open Disc Closed Disc

3. In R3 , balls are spheres.

2.2 Neighbourhoods
Definition 2.2.1: Let N ⊆ R and let x ∈ R. Then N is a neighbourhood (written nbhd) of
x if there exist δ > 0 such that (x − δ, x + δ) ⊆ N.
Remark 2.2.1 : Clearly, if N is a neighbourhood of x, then x ∈ N. This implies that if
x∈
/ N, then N cannot be a neighbourhood of x.
Examples:
1. Consider the set A = (1, 4). In this case, A is a neighbourhood of 3 since if we take
δ = 0.1, then (3 − 0.1, 3 + 0.1) = (2.9, 3.1) ⊂ A. In general, if 1 < x < 4, then A is a
neighbourhood of x. On the other hand, A is not a neighbourhood of 1 since for every
δ > 0, (1 − δ, 1) ⊂ (1 − δ, 1 + δ) but (1 − δ, 1) ∩ A = ∅ so that (1 − δ, 1 + δ) * A. In a
similar manner, for any δ > 0, (4, 4 + δ) ⊂ (4 − δ, 4 + δ) but (4, 4 + δ) ∩ A = ∅ so that
(4 − δ, 4 + δ) * A. Therefore, A is not a neighbourhood of 4.
2. In general, the set A = (a, b) or (a, b] or [a, b) or [a, b] is a neighbourhood of every x ∈ A
such that a < x < b. However, A is not a neighbourhood of x if x ∈ / (a, b). For instance,
in Example 1 above, A is not a neighbourhood of 0, 1, 4, 5.5, etc.
Theorem 2.2.1: If U is a neighbourhood of x and U ⊆ V , then V is also a neighbourhood
of x.

6
Proof : Suppose U is a neighbourhood of x. Then there exists δ > 0 so that (x − δ, x + δ) ⊆ U.
But U ⊆ V . So (x − δ, x + δ) ⊆ V . Hence V is also a neighbourhood of x.
Exercise:
Consider the set A = (−1, 3) ∪ {8}. Show that:
(a) A is a neighbourhood of 0.
(b) A is not a neighbourhood of either 3 or 8.
(c) A is not a neighbourhood of 5.

2.3 Interior Points

Definition 2.3.1: A point x is an interior point of a set A if there exists δ > 0 such that
(x − δ, x + δ) ⊆ A. In other words, x is an interior point of A if A is a neighbourhood of x.
The interior of A, denoted by int(A) or Å, is the set of all interior points of A.
To prove that x ∈ int (A), we show that there exists δ > 0 such that (x − δ, x + δ) ⊆ A, and
/ int (A), we show that for all δ > 0, (x − δ, x + δ) * A, i.e., (x − δ, x + δ)
to prove that x ∈
contains points not in A.
Proposition 2.3.1: Int (A) ⊆ A.
Proof : Suppose x ∈ int (A). Then there exists δ > 0 such that (x − δ, x + δ) ⊆ A. Since
x ∈ (x − δ, x + δ), then x ∈ A. Therefore, int (A) ⊆ A. @
Remark 2.3.1 : If x ∈ int (A), then x ∈ A.
Examples:
1. Every point of (a, b) where a < b is an interior point of (a, b). To show this, let x ∈ (a, b)
and let
min {|x − a|, |x − b|}
δ= .
2
Then (x − δ, x + δ) ⊂ (a, b). On the other hand, any point outside of (a,

b) is not an

interior point. Therefore, int ((a, b)) = (a, b). For instance, int −1, − 12 = −1, − 12 .
2. The points a and b are not interior points of [a, b] for a < b. However, every other point
of the interval is an interior point. Therefore, int ([a, b]) = (a, b).
3
3. Let A = [1, 2). Then 2
∈ int (A) but 1 ∈
/ int (A).
4. Let B = [1, 2] ∪ {3}. Then int (B) = (1, 2).
5. The sets Q and Qc = R\Q have no interior points, i.e., int (Q) = ∅ and int (Qc ) = ∅.
Proof : Suppose x ∈ int (Q). Then there exists δ > 0 such that (x − δ, x + δ) ⊆ Q. But
this can’t happen since there is at least one irrational number between x − δ and x + δ
(from the fact that every interval of positive length contains both a rational and an
irrational number). So int (Q) = ∅. A similar argument works for Qc .
6. Let x ∈ R and let δ = 1. Then (x − 1, x + 1) ⊂ R. So every real number is an interior
point of R. Therefore, int (R) = R.
Definition 2.3.2: Let A be a set of real numbers. The exterior of A, denoted by ext (A), is
the interior of its complement, i.e., ext (A) = int (Ac ).

7
Example: If A = (a, b), then ext (A) = int ((−∞, a] ∪ [b, ∞)) = (−∞, a) ∪ (b, ∞).
Exercise:
Show that if A, B ⊆ R, then:
(a) int (A ∩ B) = int (A) ∩ int (B).
(b) it is not necessarily that int (A ∪ B) = int (A) ∪ int (B).

2.4 Isolated Points

Definition 2.4.1: A point x is an isolated point of a set A if there exists δ > 0 such that
(x − δ, x + δ) ∩ A = {x}, i.e., x is an isolated point of A if there exists an open interval
containing x which does not contain any point of A different from x.
Remark 2.4.1 : If x is an isolated point of A, then x ∈ A.
Example:
Let A = [1, 3) ∪ {8}. For δ = 1, (8 − δ, 8 + δ) ∩ A = (7, 9) ∩ A = {8}. Hence 8 is an isolated
point of A. Now, suppose x ∈ [1, 3). Then for all δ > 0, (x − δ, x + δ) ∩ (A\ {x}) 6= ∅, i.e.,
(x − δ, x + δ) ∩ A 6= {x} for all δ > 0. Hence x is not an isolated point of A. Therefore, 8 is
the only isolated point of A.
Exercise:
Consider the set A = [1, 2] ∪ {3} ∪ (4, 5). Show that 3 is the only isolated point of A.

2.5 Limit Points

Definition 2.5.1: Let A ⊆ R. A point x ∈ R is a limit point or an accumulation point or
a cluster point of A if for all δ > 0, (x − δ, x + δ) ∩ (A\ {x}) 6= ∅, i.e., x is a limit point of
A if every neighbourhood of x contains at least one point y ∈ A different from x. In other
words, x is a limit point of A if for any δ > 0, there exists at least one point y ∈ A, with
y 6= x, such that |y − x| < δ. The set of all limit points of a set A, denoted by A′ , is called
the derived set of A.
Remark 2.5.1 :
1. A point x is a limit point of A if there are points in A that are arbitrarily close to x.
2. A limit point of A need not be an element of A. Hence A′ need not be a subset of A.
3. From definition, an isolated point of A can never be a limit point of A. In fact, if x ∈ A
/ A′ , then x is an isolated point of A. On the other hand, if x ∈ A′ , then x is
and x ∈
not an isolated point of A.
Proposition 2.5.1: For a set A, if x ∈ A and x ∈ / A′ , then x is an isolated point of A.
/ A′ , then there exists δ > 0 such that (x − δ, x + δ) ∩ (A\ {x}) = ∅.
Proof : Let x ∈ A. If x ∈
Since x ∈ (x − δ, x + δ) and x ∈ A, this means that (x − δ, x + δ) ∩ A = {x} for some δ > 0.
Hence x is an isolated point of A.
Examples:
1. Let A = (a, b) and let x ∈ A. Then x ∈ A′ since for any δ > 0, the interval (x − δ, x + δ)
contains points of A other than x (infinitely many points in fact). Hence A ⊆ A′ .

8
 2. Consider the set A = (1, 3) ∪ {8}. Then for any δ > 0, (1, 1 + δ) ∩ (A\ {1}) 6= ∅. This
implies that (1 − δ, 1 + δ) ∩ (A\ {1}) 6= ∅ for any δ > 0. Hence 1 ∈ A′ . In a similar
manner, for any δ > 0, (3 − δ, 3) ∩ (A\ {3}) 6= ∅ so that (3 − δ, 3 + δ) ∩ (A\ {3}) 6= ∅
for any δ > 0, meaning 3 ∈ A′ . On the other hand, 8 ∈ / A′ since if we take δ = 1, then
(8 − δ, 8 + δ) ∩ (A\ {8}) = (7, 9) ∩ (A\ {8}) = ∅ (in fact, 8 is an isolated point of A).
Remark 2.5.2 : In general, if A = (a, b) or (a, b] or [a, b) or [a, b] for a < b, then a, b ∈ A′ .
This together with Example 1 above mean that [a, b] ⊆ A′ .
3. Let A = (a, b) or (a, b] or [a, b) or [a, b]. Then A′ = [a, b]. To show this, from Remark
2.5.2, we know that [a, b] ⊆ A′ . Hence it suffices to show that if x ∈ R and x ∈ / [a, b],
′
then x ∈ / A . Suppose x > b. Let δ < x − b. Then b < x − δ and b ∈ / (x − δ, x + δ) so
that (x − δ, x + δ) ∩ (A\ {x}) = ∅. Similarly, suppose x < a. Let δ < a − x. Then
a > x + δ and a ∈ / (x − δ, x + δ) so that (x − δ, x + δ) ∩ (A\ {x}) = ∅. In either case,
x∈/ A′ .
4. Z′ = ∅.
Proof : Suppose Z has a limit point x. Then for any δ > 0, (x − δ, x + δ)∩(Z\ {x}) 6= ∅.
But for any real number x, there exists n ∈ Z such that n < x ≤ n + 1. Now, if we let
δ < min {|x − n|, |x − n − 1|}, the interval n < x ≤ n + 1 contains no integers different
from x, i.e., (x − δ, x + δ) ∩ (Z\ {x}) = ∅. Hence no real number is a limit point of Z.
5. Q′ = R.
Proof : Suppose x ∈ R. Then for any δ > 0, the interval (x − δ, x + δ) contains infinitely
many points of Q other than x. Thus (x − δ, x + δ) ∩ (Q\ {x}) 6= ∅. Therefore, every
real number is a limit point of Q.
Exercise:
Show that if A = n1 : n ∈ N , then A′ = {0}.


Further Examples:
1. Prove that if A, B ⊂ R and A ⊆ B, then A′ ⊆ B ′ .
Solution: Let x ∈ A′ . Then for all δ > 0, (x − δ, x + δ) ∩ (A\ {x}) 6= ∅. If A ⊆ B, it
follows that (x − δ, x + δ) ∩ (B\ {x}) 6= ∅ for all δ > 0, and so x ∈ B ′ . Hence A′ ⊆ B ′ .
2. Prove that if A, B ⊂ R, then (A ∪ B)′ = A′ ∪ B ′ .
Solution: A ⊆ A ∪ B and, by Example 1 above, A′ ⊆ (A ∪ B)′ . Similarly, B ⊆ A ∪ B
and B ′ ⊆ (A ∪ B)′ . Hence A′ ∪ B ′ ⊆ (A ∪ B)′ .
Now, suppose x ∈ (A ∪ B)′ . Then
∀δ > 0, (x − δ, x + δ) ∩ [(A ∪ B) \ {x}] 6= ∅
⇒ ∀δ > 0, (x − δ, x + δ) ∩ [(A\ {x}) ∪ (B\ {x})] 6= ∅
⇒ ∀δ > 0, [(x − δ, x + δ) ∩ (A\ {x})] ∪ [(x − δ, x + δ) ∩ (B\ {x})] 6= ∅
⇒ ∀δ > 0, [(x − δ, x + δ) ∩ (A\ {x})] 6= ∅ or [(x − δ, x + δ) ∩ (B\ {x})] 6= ∅
⇒ x ∈ A′ or x ∈ B ′
⇒ x ∈ A′ ∪ B ′ .
Hence (A ∪ B)′ ⊆ A′ ∪ B ′ .

9
 Therefore, (A ∪ B)′ = A′ ∪ B ′ .
Theorem 2.5.1: Let x ∈ R and X ⊂ R. If x is a limit point of X, then any neighbourhood
of x contains infinitely many members of X.
Proof : Let A be a neighbourhood of x containing only a finite number of points of X, i.e.,
A ∩ X is finite. Then A ∩ X\ {x} is also finite. Suppose A ∩ X\ {x} = {y1 , y2 , · · · , yn }. We
show that there exists δ > 0 such that (x − δ, x + δ) ∩ A does not contain any member of
X\ {x}. Since both A and (x − δ, x + δ) are neighbourhoods of x, so is their intersection.
This will prove that there is a neighbourhood of x containing no element of X\ {x} hence
proving x is not a limit point of X. Let δ = min {|x − y1 |, |x − y2 |, · · · , |x − yn |}. Since
x 6= yi (i = 1, · · · , n) and δ > 0, then (x − δ, x + δ) ∩ A contains no point of X other than x,
hence the proof. @
Corollary 2.5.1:
(a) If x has a neighbourhood which only contains finitely many members of X, then x
cannot be a limit point of X.
(b) A finite point set has no limit points.

2.6 Open Sets and Closed Sets

Definition 2.6.1: Let A ⊆ R. Then:
(a) A is open if for any x ∈ A, there exists δ > 0 such that (x − δ, x + δ) ⊆ A, i.e., A is
open if it is (or contains) a neighbourhood of each of its points. In other words, A is
open if every point of A is an interior point of A.
(b) A is closed if every limit point of A is a point of A, i.e., if A′ ⊆ A.
Theorem 2.6.1: A set A is open if and only if its complement is closed.
Proof : Suppose A is open and suppose x is a limit point of Ac . Then for every δ > 0,
(x − δ, x + δ) ∩ (Ac \ {x}) 6= ∅, i.e., for every δ > 0, the interval (x − δ, x + δ) contains a
point y ∈ Ac such that y 6= x. So (x − δ, x + δ) * A for every δ > 0. Hence x ∈ / int (A).
Since A is open, then x ∈ / A, meaning that x ∈ Ac . It now follows that Ac is a closed.
Conversely, suppose Ac is a closed. Let x ∈ A. Then x ∈ / Ac and x is not a limit point
of Ac (since a closed set contains all its limit points). So there exists δ > 0 such that
(x − δ, x + δ) ∩ Ac = ∅, i.e., (x − δ, x + δ) ⊆ A. Thus x ∈ int (A) and it follows that A is
open. @
Corollary 2.6.1: A set A is closed if and only if its complement is open.
Theorem 2.6.2: Let A ⊂ R. Then:
(a) Å is open.
(b) A is open if and only if A = Å.
(c) If B is open and B ⊆ A, then B ⊆ Å.
Proof :
(a) We need to show that if x ∈ Å, then there exists δ > 0 such that (x − δ, x + δ) ⊆ Å.
Suppose x ∈ Å. Then there exists δ > 0 such that (x − δ, x + δ) ⊆ A. We show that
(x − δ, x + δ) ⊆ Å by showing that every point of (x − δ, x + δ) is an interior point of

10
 A. Let y ∈ (x − δ, x + δ) and let

min {|y − x + δ|, |y − x − δ|}

ε= .
2

Then (y − ε, y + ε) ⊂ (x − δ, x + δ) ⊆ A. So y ∈ Å by definition, and since y was

arbitrary, (x − δ, x + δ) ⊆ Å.
(b) Suppose A = Å. Then A is open since Å is open, from part (a) of the theorem.
Conversely, supose A is open. Then every point of A is an interior point of A. Hence
A ⊆ Å. But by definition, Å ⊆ A. Therefore, A = Å.
(c) Let x ∈ B. Then x is an interior point of of B since B is open. So there exists δ > 0
such that (x − δ, x + δ) ⊆ B. But since B ⊆ A, then (x − δ, x + δ) ⊆ A and it follows
that x ∈ Å. Therefore, B ⊆ Å. @
Remark 2.6.1 : Parts (a) and (c) of Theorem 2.6.2 show that Å is the largest open subset
of A.
Examples:
1. Let a, b ∈ R such that a < b. Then (a, b) is open since int ((a, b)) = (a, b). This proves
that every open interval is an open set. Similarly, the infinite intervals (−∞, a) and
(a, ∞) are open.
2. Let A = [a, b]. Then A is closed since it contains all its limit points (from a previous
example, A′ = A). Alternatively, it is closed since Ac = (−∞, a) ∪ (b, ∞) which is open.
3. Sets that are both open and closed:
From a previous example, int (R) = R. So R is open. But ∅c = R\∅ = R. So the empty
set ∅ is closed.
Now, since there are no elements in ∅, then for every x ∈ ∅, (x − δ, x + δ) ⊂ ∅ for any
δ > 0. Hence ∅ is open. But Rc = R\R = ∅. Hence R is closed.
Therefore, R and ∅ are sets that are both open and closed; in fact, these are the only
sets with this property.
4. Sets that are neither open nor closed :
(a) Let a, b ∈ R such that a < b. Consider the set A = [a, b). Then A is not open since
a ∈ A but a ∈ / Å. Similarly, A is not closed since b ∈ A′ but b ∈/ A. Alternatively,
consider A = (−∞, a) ∪ [b, ∞). In this case, Ac is not closed since a is a limit point
c

of Ac but a ∈ / Ac . So A is not open. Similarly, Ac is not open since b ∈ Ac but

b∈/ int (Ac ). So A is not closed either. Therefore, [a, b) is neither open nor closed.
The same is true for (a, b].
(b) The set Q is neither open nor closed. As seen earlier none of the points of Q is an
interior point. Thus Q is not open and so Qc is not closed. Similarly, none of the
points of Qc is an interior point. Thus Qc is not open and so (Qc )c = Q is not
closed. Therefore, Q and Qc are examples of subsets of R that are neither open
nor closed.
Note 2.6.1 : Being closed is not the opposite of being open, i.e., a set is not either open or
closed, it can be neither.

11
 5. Let x ∈ R. Then {x} is closed since {x}c = (−∞, x) ∪ (x, ∞) is open. Alternatively,
{x} is closed since it contains all its limit points, namely none. On the other hand, {x}
is not open since there does not exist δ > 0 such that (x − δ, x + δ) ⊆ {x}.
Theorem 2.6.3: Let {Gi } be an infinite collection of open sets and {Hi } an infinite collection
of closed sets. Then:
∞
(a) ∪ Gi is open (i.e., arbitrary union of open sets is open).
i=1
n
(b) ∩ Gi is open (i.e., finite intersection of open sets is open).
i=1
n
(c) ∪ Hi is closed (i.e., finite union of closed sets is closed).
i=1
∞
(d ) ∩ Hi is closed (i.e., arbitrary intersection of closed sets is closed).
i=1

Proof :
∞ ∞
(a) We prove that if x ∈ ∪ Gi , then x is an interior point of ∪ Gi , i.e., there exists δ > 0
i=1 i=1
∞ ∞
such that (x − δ, x + δ) ⊆ ∪ Gi . Now, suppose x ∈ ∪ Gi . Then x ∈ Gi for at least
i=1 i=1
one i, and since Gi is open, x is an interior point of Gi . So there exists δi > 0 such
∞
that (x − δi , x + δi ) ⊆ Gi . Therefore, (x − δi , x + δi ) ⊆ ∪ Gi .
i=1
n n n
(b) If ∩ Gi = ∅, then ∩ Gi is open since ∅ is open. On the other hand, suppose ∩ Gi 6= ∅
i=1 i=1 i=1
n
and let x ∈ ∩ Gi . Then x ∈ Gi for all i = 1, 2, · · · , n. But each Gi (i = 1, 2, · · · , n)
i=1
is open. So, for each i = 1, 2, · · · , n, there exists δi > 0 such that (x − δi , x + δi ) ⊆ Gi .
Now, let δ = min {δ1 , · · · , δn }. Since δi > 0 for all i = 1, 2, · · · , n, then δ > 0. Hence
∞
(x − δ, x + δ) ⊆ Gi for every i = 1, 2, · · · , n. Hence (x − δ, x + δ) ⊆ ∩ Gi . Therefore,
i=1
n n n
x is an interior point of ∩ Gi , and since x is an arbitrary point of ∩ Gi , then ∩ Gi
i=1 i=1 i=1
is open.
Remark 2.6.2 : The intersection of an infinite number of open sets need not be open.
Counterexample:

In R, let Gn = − n1 , n1 , n ∈ N, i.e., G1 = (−1, 1), G2 = − 12 , 12 , G3 = − 13 , 31 , · · · . Then



∞
Gn is open for each n = 1, 2, · · · . However, ∩ Gn = {0}, which is not open (by Example 5
n=1
above).
(c) Hi is closed means Hic is open, for each i= 1, 2, ·· · , n. So, by part (b) of the theorem,
n n c
∩ Hic is open, and by Theorem 2.6.1, ∩ Hic is closed. But by the De Morgan’s
i=1  c n i=1
n n n
law, ∩ Hic = ∪ (Hic )c = ∪ Hi . Therefore, ∪ Hi is closed.
i=1 i=1 i=1 i=1

Remark 2.6.3 : The union of an infinite collection of closed sets need not be closed.
Counterexample:
 n n
, n ∈ N, i.e., H1 = − 12 , 12 , H2 = − 32 , 23 , H3 = − 34 , 34 , · · · .
      
In R, let Hn = − n+1 , n+1
∞
Then Hn is closed for each n = 1, 2, · · · . However, ∪ Hn = (−1, 1), which is not closed.
n=1

12
 (d ) Hi is closed means Hic is open, for eachi = 1, 2,· · · . So by part (a) of the theorem,
∞ ∞ c
∪ Hic is open, and by Theorem 2.6.1, ∪ Hic is closed. But by the De Morgan’s
i=1  c ∞ i=1
∞ ∞ ∞
law, ∪ Hic = ∩ (Hic )c = ∩ Hi . Therefore, ∩ Hi is closed. @
i=1 i=1 i=1 i=1

2.7 Closure of a Set

Definition 2.7.1: Let E ⊂ R and let E ′ be the set of all limit points of E. Then the closure
of E is the set E = E ∪ E ′ . A point of the closure of E is called an adherent point of E.
Thus, if x is an adherent point of E, then x ∈ E or x ∈ E ′ .
Remark 2.7.1 : E is the intersection of all closed supersets of E.
Examples:
1. (a, b) = [a, b].
2. Q = R.
3. Qc = R.
4. ∅ = ∅.
5. {x} = {x}.
6. If E = [−1, 0), then E ′ = [−1, 0] and E = E ∪ E ′ = [−1, 0].
7. If F = (0, 1] ∪ {3}, then F ′ = [0, 1] and F = F ∪ F ′ = [0, 1] ∪ {3}.
Theorem 2.7.1: Let E ⊂ R. Then:
(a) E is closed.
(b) E = E if and only if E is closed.
(c) If F is closed and E ⊂ F , then E ⊂ F .
Proof :
(a) Since E is the intersection of all closed supersets of E, then E is closed, by Theorem
2.6.3.
(b) Suppose E = E. From part (a) of the theorem, E is closed. Therefore, E is closed.
Conversely, suppose E is closed. Then E ′ ⊆ E, and E = E ∪ E ′ = E.
(c) Suppose F is closed and F ⊃ E. Then F ⊇ F ′ and F ′ ⊃ E ′ . Hence F ⊃ E ′ . Therefore,
F ⊃ E ∪ E ′ = E. @
Remark 2.7.2 : By parts (a) and (c) of Theorem 2.7.1, E is the smallest closed subset of
R that contains E.
Example:
Every singleton set {x} is closed since {x} = {x}.
Exercise:
1. Show that [1, 2) ∪ {3} is neither open nor closed.
2. Let A, B ⊆ R.
(a) Prove that:
(i ) if A ⊆ B, then A ⊆ B.

13
 (ii ) A ∪ B = A ∪ B.
(iii ) A ∩ B ⊆ A ∩ B.
(iv ) if A is open, then A′ = A.
(b) Give an example such that A ∩ B 6= A ∩ B.

2.8 Boundary
Definition 2.8.1: Let A ⊆ R. The boundary of A, denoted bdry (A), is the set of points
x ∈ R such that every neighbourhood of x contains at least one point of A and at least one
point not in A. Alternatively, bdry (A) is the set A\Å, i.e., the set of points in the closure of
A not belonging to the interior of A. An element in the boundary of A is called a boundary
point of A.
Examples:
1. bdry ((0, 5)) = bdry ([0, 5)) = bdry ((0, 5]) = bdry ([0, 5]) = {0, 5}.
2. bdry (∅) = ∅.
3. bdry (Q) = bdry (Qc ) = R.
4. bdry (Z) = Z.
5. bdry (Q ∩ [0, 1]) = [0, 1].
Remark 2.8.1 :
1. bdry (A) is closed.
2. bdry (A) = bdry (Ac ).
Theorem 2.8.1: A set is closed if and only if it contains all its boundary points.
Examples:
1. bdry (Z) = Z and hence Z is closed.
2. bdry (Q) * Q and hence Q is not closed.
Corollary 2.8.1: A set is open if and only if it contains none of its boundary points.

2.9 Dense Sets

Definition 2.9.1: Let A ⊂ R. Then A is said to be dense (in R) if A ∩ (a, b) 6= ∅ for each
open interval (a, b). More generally, suppose A, B ⊂ R. If every open interval that intersects
B also intersects A, we say that A is dense in B.
Examples:
1. The set Q is dense in R since for any interval (a, b) in R, Q ∩ (a, b) 6= ∅.
2. Q is dense in Qc .
1 3


3. Z is not dense in R since ,
2 4
∩ Z = ∅.
4. N is not dense in R since (−1, 0) ∩ N = ∅.
5. N is not dense in Z since (−1, 1) ∩ Z 6= ∅ but (−1, 1) ∩ N = ∅.

14
Remark 2.9.1 : A set A is dense in a set B if every point of B is a limit point of A, or a
point of A, or both.
Exercise:
Prove that:
(a) if A is dense in B and C ⊂ B, then A is dense in C.
(b) A is dense in B if and only if A ⊃ B.
(c) every set A is dense in its closure A.
(d ) if A ⊂ B and A is dense in B, then A = B.
Definition 2.9.2: Let A ⊂ R. Then A is said to be nowhere dense in R provided that every
open interval I contains an open subinterval J such that A ∩ J = ∅. In other words, A is
nowhere dense if A contains no open interval.
Examples:
1. Any finite set is nowhere dense.
2. N is nowhere dense.
3. The set n1 : n ∈ N is nowhere dense.


4. Any finite union of nowhere dense sets is nowhere dense.

5. A countable intersection of nowhere dense sets need not be nowhere dense.
n
Theorem 2.9.1: Let A1 , A2 , · · · , An be nowhere dense in R. Then ∪ Ai is also nowhere
i=1
dense in R.
Proof : Let I be any open interval in R. We seek an open interval J ⊂ I such that J ∩ Ai = ∅
for each i = 1, 2, · · · , n. Since A1 is nowhere dense, there exists an open interval I1 ⊂ I such
that A1 ∩I1 = ∅. Now, A2 is also nowhere dense in R. So there exists an open interval I2 ⊂ I1
such that A2 ∩ I2 = ∅. Proceeding in this manner we obtain intervals I1 ⊃ I2 ⊃ I3 ⊃ · · · ⊃ In
such that Ai ∩ Ii = ∅ for i = 1, 2, · · · , n. Since In ⊂ Ii for i = 1, 2, · · · , n, then Ai ∩ In = ∅
for i = 1, 2, · · · , n. Thus
n  n n
∪ Ai ∩ In = ∪ (Ai ∩ In ) = ∪ ∅ = ∅,
i=1 i=1 i=1

which is what we sought to prove. @

2.10 Compact Sets

Definition 2.10.1: Let E ⊆ R. A collection T consisting of open sets is an open cover for
the set E if every element in E is in some set G from T , i.e., if E ⊂ ∪ G. The set E has a
G∈T
n
finite subcover from T if there exists G1 , G2 , · · · , Gn in T such that E ⊂ ∪ Gi .
i=1

Examples:
1. The collection T = 1 − n1 , 2 − n1 : n ∈ N is an open cover for the set E = [1/2, 2).


2. The collection T = {(−n, n) : n ∈ N} is an open cover for the set R = (−∞, ∞).
3. Let r ∈ R. Then T = {(r − 1, r + 1) : r ∈ R} is an open cover for the set R.

15
Remark 2.10.1 : The collections in Examples 1 and 2 above are countable while the col-
lection in Example 3 is uncountable.
Definition 2.10.2: The set E ⊆ R is a compact set if every open cover for E has a finite
subcover.
Example:
The set E = [a, ∞) is not compact since T = a − n1 , n : n ∈ N is an open cover for E,


with the property that no finite subcollection covers E. Observe that the set E in this case
is closed but not bounded.
Theorem 2.10.1: If E is a compact set, then E is bounded.
Proof : The collection T = {(−n, n) : n ∈ N} is an open cover for any set E ⊆ R. The union
of any finite subcollection will be an interval of the form (−n0 , n0 ). If E is compact, there
must be an interval (−n0 , n0 ), so that E ⊂ (−n0 , n0 ), and so E is bounded. @
Example:
The set E = [a, b) is not compact since the collection T = a − n1 , b− n1 : n ∈ N  is an


open cover for E, but the union of any finite subcollection has the form a − n11 , b − n12 , and
does not therefore cover E. Observe that the set E in this case is bounded but not closed.
Theorem 2.10.2: If E is a compact set, then E is closed.
Proof : If E is not
∞  closed,
 1 1
 ∞ be a limit point x of E with x ∈
there must / E. The collection
T = {Gn }n=1 = R\ x − n , x + n n=1 is an open cover for E. In order for E to be compact,
k
there must be a finite subcollection {Gni }ki=1 of T that covers E. Since ∪ Gni = Gn′ where
i=1
k
n′ = max {n1 , n2 , · · · , nk }, we have E ⊂ ∪ Gni = Gn′ = R\ x − n1′ , x + n1′ or equivalently,
 
i=1
E ∩ x − n1′ , x + n1′ = ∅. But E ∩ x − n1′ , x + n1′ = ∅ implies that x is an exterior point of
   

E, and therefore not a limit point of E. This contradiction means that E is not compact,
and the proof is complete. @
Remark 2.10.2 : Theorems 2.10.1 and 2.10.2 prove that a compact set need be both closed
and bounded. The next theorem asserts that these two conditions are also sufficient for
compactness.
Theorem 2.10.3 (Heine-Borel Theorem): The set E ⊆ R is compact if and only if E is
closed and bounded.

3 Upper Bounds and Lower Bounds

Definition 3.1.1: A real number m is called an upper bound (u.b) of a non-empty subset A
of R if x ≤ m for all x ∈ A. A real number n is called an lower bound (l.b) of a non-empty
subset A of R if n ≤ x for all x ∈ A.
The set of all upper bounds of a set A is denoted by UA while the set of all lower bounds of
A is denoted by LA .
Note 3.1.1 :
1. A lower bound or an upper bound of A need not be an element of A.

16
 2. A lower bound or an upper bound of A is not unique.
Definition 3.1.2: If A has an upper bound, then A is said to be bounded above. If A has a
lower bound, then A is said to be bounded below. If A has both an upper bound and a lower
bound, then we say that A is bounded. If A is neither bounded nor bounded above or below,
we say that A is unbounded. By definition, A is bounded if A ⊆ [n, m] for some interval
[n, m] of finite length.
Examples:
1. Let A = (1, 3]. Then A = {x | 1 < x ≤ 3}. If n ∈ R such that n ≤ 1, then n ∈ LA , and
if m ∈ R such that m ≥ 3, then n ∈ UA . Therefore, A is bounded.
2. The set N = {1, 2, 3, 4, · · · } is bounded below but not above. For instance, −10 is a
lower bound of N (in fact, every real number n ≤ 1 is a lower bound of N).
Remark 3.1.1 : Examples 1 and 2 above show that a bounded set need not be countable
and vice versa.
n
3. The number 1 is an upper bound of the set B = 12 , 43 , 78 , · · · , 2 2−1

n , · · · . On the other
1 1
hand, 2 is a lower bound of B. In fact, if x ∈ B, then x ≥ 2 and x < 1.
4. The set C = n1 : n ∈ N , i.e., C = 1, 21 , 13 , 41 , · · · is bounded since if x ∈ C, then
 

0 < x ≤ 1. Each real number k ≥ 1 is an upper bound of C while each real number
l ≤ 0 is a lower bound of C. Hence C has infinitely many lower bounds and infinitely
many upper bounds.
5. The set Z = {· · · , −3, −2, −1, 0, 1, 2, 3, · · · } is unbounded.
6. The set Z2 = {z 2 | z ∈ Z} = {0, 1, 4, 9, 16, 25 · · · } is bounded below while the set
−Z2 = {−z 2 | z ∈ Z} = {· · · , −25, −16, −9, −4, −1, 0} is bounded above.
7. The set Y = {sin x | 0 ≤ x < 2π} is bounded by −1 and 1. On the other hand, the set
K = {tan x | 0 ≤ x < 2π} is unbounded. However, K ′ = {tan x | 0 ≤ x ≤ π/2} is bounded
below.
Proposition 3.1.1: Let A ⊂ R.
(a) If m, m′ ∈ R with m′ > m and m ∈ UA , then m′ ∈ UA .
(b) If n, n′ ∈ R with n′ < n and n ∈ LA , then n′ ∈ LA .
From Examples 2 and 3 above, infinitely many real numbers greater than −10 are lower
bounds of N, but there is no number less that 1 which is an upper bound of B. This leads
us to the concept of least upper bound and greatest lower bound.
Definition 3.1.3: Let A ⊂ R. An element u ∈ R is called the least upper bound (l.u.b) of
A or supremum of A, written l.u.b A or sup A, if:
(i ) u ∈ UA (i.e., u is an upper bound of A)
(ii ) u ≤ m for all m ∈ UA .
Definition 3.1.4: Let A ⊂ R. An element l ∈ R is called the greatest lower bound (g.l.b) of
A or infimum of A, written g.l.b A or inf A, if:
(i ) l ∈ LA (i.e., l is a lower bound of A)
(ii ) l ≥ n for all n ∈ LA .

17
Some Consequences of the Definitions
1. If α is an upper bound of a set A ⊂ R and α ∈ A, then α is the supremum of A.
Proof : Suppose α, β ∈ A and α, β ∈ UA . Since α ∈ UA and β ∈ A, then α ≥ β. In a
similarly manner, since β ∈ UA and α ∈ A, then β ≥ α. The two inequalities hold
simultaneously if and only if α = β. So we cannot have two upper bounds in a set.
Therefore, if α ∈ A and α ∈ UA , then α is the supremum of A. @
2. If a set has an infimum, then the infimum is unique.
Proof : Let A be a non-empty subset of R. Suppose α = inf A and also β = inf A. Then
α, β ∈ LA . If α = inf A, then α ≥ l for all l ∈ LA . But β ∈ LA . So α ≥ β. Similarly, if
β = inf A, and α ∈ LA , then β ≥ α. The inequalities α ≥ β and β ≥ α hold simultane-
ously if and only if α = β. Therefore, inf A is unique. @
Remark 3.1.2 : If A, B ⊂ R where A ⊆ B, then inf B ≤ inf A and sup A ≤ sup B.
Examples:
1. Let A = (3, 4). If n ∈ LA , then n ≤ 3, and if m ∈ UA , then m ≥ 4. Hence inf A = 3
and sup A = 4.
n
2. Let B = 21 , 34 , 87 , · · · , 2 2−1 1

n , · · · . Then inf B = 2 and sup B = 1. In this case, inf B ∈ B

and sup B ∈
/ B.
3. Inf N = 1 but and sup N = ∞ since N has no upper bound.
4. Let C = {x} for x ∈ R. Then inf C = sup C = x.
5. If D = 0, 12 , 32 , 43 , 65 , 67 , · · · , then inf D = 0 and sup D = 1.


6. The empty set ∅ is bounded since ∅ ⊂ [n, m] for any interval [n, m] of finite length.
Thus every number m ∈ R is an upper bound of ∅ and so ∅ does not have a supremum.
Similarly, ∅ does not have an infimum.

Completeness Axiom of Real Numbers (Least Upper Bound Axiom)

Statement: Every non-empty set of real numbers that is bounded above has a least upper
bound in R.

4 Sequences and Series of Real Numbers

4.1 Sequences and Subsequences
Definition 4.1.1: A sequence of real numbers is a function from positive integers (natural
numbers) into the set of real numbers, i.e., f : N −→ R, f (n) = an . The sequence may be
denoted as {an } or {an }∞n=1 . We may also write the sequence as {a1 , a2 , · · · , an , · · · } where
a1 is the first term of the sequence, a2 the second, and so on.
Remark 4.1.1 : The sequence illustration should not be confused with set notation. In sets,
the order of elements does not matter, but in sequences, it does. Thus a sequence {1, 2, 3, 4}
is different from {1, 3, 4, 2}. In set notation, these would be considered equal.

18
Examples:
1. {1} = {1, 1, 1, 1, · · · }.
2. {2n}∞
n=1 = {2, 4, 6, 8, · · · }.
∞
3. (−1)n+1 n=1 = {1, −1, 1, −1, · · · }.


Definition 4.1.2: Let {an } be a sequence of real numbers and consider the sequence {ni }
of positive integers such that n1 < n2 < n3 < · · · . Then the sequence {ani } is called a sub-
sequence of {an }. If {an } = {a1 , a2 , a3 , a4 , · · · } is a sequence, then {ani } = {a21 , a42 , a63 , · · · }
is a subsequence.
Remark 4.1.2 : A subsequence {ani } of a sequence {an } is itself a sequence and the numbers
n1 , n2 , · · · themselves form a subsequence of the sequence of positive integers 1, 2, 3, · · · .
Examples:
∞
1. Consider the sequence {(−1)n }n=1 = {−1, 1, −1, 1, −1, 1, · · · }. Then −1, −1, −1, · · ·
∞
and 1, 1, 1, · · · are subsequences of {(−1)n }n=1 .
2. The sequence 1, 2, 1, 3, 1, 4, 1, 5, · · · has subsequences 1, 1, 1, 1, · · · and 1, 2, 3, 4, 5, · · · .

4.2 Convergent and Divergent Sequences

Definition 4.2.1: Let {an }∞ ∞
n=1 be a sequence of real numbers. We say that {an }n=1 has a
limit L ∈ R if for every ε > 0, there exists a positive integer N = N (ε) such that

|an − L| < ε for all n ≥ N.

In this case, we write lim an = L or an −→ L (as n −→ ∞).
n→∞

Remark 4.2.1 : In Definition 4.2.1,

(i ) the limit L must be a real number.
(ii ) the value of N will, in general, depend of the value of ε.
(iii ) the inequality |an − L| < ε must hold for all values of n except at most a finite
number - namely, n = 1, 2, · · · , N − 1.
(iv ) the proof that lim an = L consists, upon given ε > 0, of finding a value of N such
n→∞
that |an − L| < ε for all n ≥ N.
Definition 4.2.2: If a sequence {an }∞ ∞
n=1 of real numbers has a limit L, we say that {an }n=1
converges (or {an }∞
n=1 is convergent) to L.

Definition 4.2.3: If a sequence {an }∞ n=1 of real numbers does not have a limit, we say that
{an }n=1 diverges or {an }n=1 is divergent. A sequence {an }∞
∞ ∞
n=1 of real numbers diverges to
∞, written lim an = ∞ or an −→ ∞, if for any real number M > 0, there exist N ∈ N such
n→∞
that an ≥ M for all n ≥ N. A sequence {an }∞
n=1 of real numbers diverges to −∞, written
lim an = −∞ or an −→ −∞, if for any real number M > 0, there exist N ∈ N such that
n→∞
an < −M for all n ≥ N. If the sequence {an }∞
n=1 of real numbers diverges but lim an 6= ±∞, n→∞
we say that {an }∞
n=1 oscillates.

19
Theorem 4.2.1: A point x is a limit point of a set A if and only if there exists a sequence
{xn }∞
n=1 in A such that lim xn = x.
n→∞

Examples:
1. The sequence n1 = 1, 21 , 13 , 41 , · · · converges to 0.
 

Proof : Given ε > 0 we find N = N (ε) ∈ N such that

1
− 0 < ε for all n ≥ N (1)
n
or
1
< ε for all n ≥ N. (2)
n
If we choose N such that N1 < ε, then (1) and (2) will hold since n1 ≤ N1 if n ≥ N. Now,
1
N
< ε if and only if n > 1ε . So, we can choose N ∈ N such that N > 1ε proving that
1
n
−→ 0.
Note 4.2.1 : The limit 0 if the sequence in Example 1 above is not equal to any term of the
sequence.
2. The sequence {an }∞ ∞
n=1 = {n}n=1 = {1, 2, 3, · · · } diverges.
Proof : Suppose the contrary, i.e., an −→ L for some L ∈ R. Then for every ε > 0 there
exists a positive integer N = N (ε) for which
|n − L| < ε (n ≥ N) .
In particular, there is an N for which
|n − L| < 1 (n ≥ N)
or
−1 < n − L < 1 (n ≥ N)
or
L−1< n<L+1 (n ≥ N) .
This means that all values of n greater than or equal to N lie between L − 1 and L + 1,
a contradiction. Hence the sequence diverges.
Remark 4.2.2 : It is obvious that {an }∞ ∞
n=1 = {n}n=1 diverges to ∞, by Definition 4.2.3. For
given M > 0, choosing N ∈ N such that N ≥ M, then an = n ≥ M for all n ≥ N.
∞
3. The sequence {(−1)n }n=1 diverges.
Proof : The terms of the sequence are −1, 1, −1, 1, · · · . Suppose the sequence converges
1 1


to L for some L ∈ R. Then for ε = 2 there is an N = N 2 ∈ N such that
1
|(−1)n − L| < (n ≥ N) . (1)
2
For n even (1) becomes
1
|1 − L| < (n ≥ N) , (2)
2

20
 and for n odd (1) becomes
1
| − 1 − L| < (n ≥ N) . (3)
2
In this case, (2) implies that L is less than 21 unit away from 1 while (3) implies that L
is less than 21 unit away from −1, a contradiction. [Alternatively, the inequality (3) is
equivalent to |1 + L| < 12 . But then

1 1
2 = | (1 + L) + (1 − L) | ≤ |1 + L| + |1 − L| < + = 1,
2 2
which is a contradiction.] Hence the sequence does not have a limit.
n o∞
2n
4. The sequence n+4n 1/2 converges to 2.
n=1
Proof : Given ε > 0 we find (calculate) N ∈ N such that

2n
−2 < ε (n ≥ N) . (1)
n + 4n1/2

The ineqiality (1) is equivalent to

2n − 2n − 8n1/2
<ε (n ≥ N)
n + 4n1/2
or
8n1/2
<ε (n ≥ N) . (2)
n + 4n1/2
1
8n /2 8
Now the left hand side of (2) is less than n
= n1/2
. Hence (2) will be true if

8
<ε (n ≥ N) . (3)
n1/2
8 64
If we choose N such that N 1/2
< ε, that is, choose N > ε2
, then (3) will certainly be
8 8
true. (For n1/2
≤ N 1/2
if n ≥ N). We have thus shown that if N is any positive integer
64
such that N > ε2
, then (3) and hence (2) and finally (1) will be true. This proves that
2n
lim n+4n 1/2 = 2.
n→∞

5. The sequence {log (1/n)}∞

n=1 diverges to −∞.
Proof : Given M > 0 we find N ∈ N such that
1
log < −M (n ≥ N) . (1)
n
But this is equivalent to
log n > M (n ≥ N) ,
or
n > eM (n ≥ N) . (2)

21
 Thus if we choose N ≥ eM , the (2) and hence (1) will hold.
Exercise:
∞
Use definition to show that the sequence {an }∞ (−1)n+1

n=1 = n=1
has no limit.
Theorem 4.2.2: Let {an }∞
n=1 be a sequence of non-negative numbers. If lim an = L, then
n→∞
L ≥ 0.
Proof : Supose the contrary, i.e., L < 0. Then for ε = − L2 there exists N ∈ N such that

L
|an − L| < − (n ≥ N) .
2
In particular,
L
|aN − L| < − ,
2
which implies
L
aN − L < −
2
or
L
aN < < 0 for L < 0,
2
a contradiction since, by hypothesis, aN ≥ 0. Hence L ≥ 0. @
Theorem 4.2.3: If {an }∞n=1 converges, then its limit is unique.
Proof : Suppose an −→ L and an −→ M. We show that L = M. Suppose the contrary, i.e.,
L 6= M so that |M − L| > 0. Let ε = 12 |M − L|. By hypothesis an −→ L. So there exists
N1 = N1 (ε) ∈ N such that
|an − L| < ε (n ≥ N1 ) .
Similarly, since an −→ M there exists N2 = N2 (ε) ∈ N such that

|an − M| < ε (n ≥ N2 ) .

Let N = max {N1 , N2 }. Then N ≥ N1 and N ≥ N2 so that

|an − L| < ε (n ≥ N)

and
|an − M| < ε (n ≥ N) .
Thus

|M − L| = |(an − L) − (an − M)| ≤ |an − L| + |an − M| < 2ε = |M − L|.

This contradiction shows that M = L, which is what we wished to show. @

Definition 4.2.4: Let {ani } be a subsequence of a sequence {an } of real numbers. If {ani }
converges, its limit is called a subsequential limit of {an }.
Theorem 4.2.4: A sequence {an }∞ n=1 of real numbers is convergent to L if and only if every
∞
subsequence of {an }n=1 is also convergent to L.

22
Corollary 4.2.1: All subsequences of a convergent sequence {an }∞ n=1 of real numbers con-
verge to the same limit.
Proof : If the sequence {an }∞ ∞
n=1 converges to L, then, by Theorem 4.2.3, {an }n=1 converges
to no other limit. By Theorem 4.2.4, then, all subsequences of {an }∞
n=1 converge to L (and
to no other limit). @
Examples:
1. The sequence 1, − 21 , 13 , − 41 , 15 , − 61 , · · · converges to 0. By Corollary 4.2.1, all its subse-
quences converge to 0. For instace, the subsequences 1, 31 , 51 , · · · and − 12 , − 41 , − 16 , · · ·
both converge to 0.
∞
2. Consider the sequence {(−1)n }n=1 . Then −1, −1, −1, −1, · · · and 1, 1, 1, 1, · · · are sub-
∞
sequences of {(−1)n }n=1 which converge respectively to −1 and 1. Since the two sub-
∞
sequences converge to different limits, then, by Theorem 4.2.4, the sequence {(−1)n }n=1
diverges. This example shows that a divergent sequence may have a convergent subse-
quence.
3. The sequence {n}∞ n=1 shows that a divergent sequence need have no convergent subse-
quence.
4. Consider the sequence 1, 2, 1, 3, 1, 4, 1, 5, · · · . By Corollary 4.2.1, the sequence diverges
since it has the divergent subsequence 1, 2, 3, 4, 5, · · · . Moreover, the sequence does not
diverge to ∞ since there is no N ∈ N for which an > 2 for all n ≥ N. The sequence
obviously does not diverge to −∞. Hence it oscillates.
5. The sequence {an }∞ n=1 = {1, −2, 3, −4, 5, −6, · · · } has the subsequence 1, 3, 5, · · · which
diverges to ∞ and also has the subsequence −2, −4, −6, · · · which diverges to −∞. The
sequence {an }∞n=1 diverges, but lim an 6= ±∞; the sequence oscillates.
n→∞

4.3 Bounded Sequences

Since a sequence {an }∞n=1 of real numbers is a function from N into R, the range of the
sequence, namely {a1 , a2 , · · · }, is a subset of R. The range of a sequence may be finite or it
may be infinite.
Definition 4.3.1: A sequence {an }∞ ∞
n=1 is bounded above if the range of {an }n=1 is bounded
above, it is bounded below if its range is bounded below, and it is bounded if its range is
bounded. Thus {an }∞n=1 is bounded if and only if there exists M ∈ R such that

|an | ≤ M for all n ∈ N.

Example:
Consider the sequence {an } defined by
(
1 if n is even
an = ,
0 if n is odd

i.e., {an } = {0, 1, 0, 1, 0, 1, · · · }. Then the range of {an } is the set {0, 1} which is bounded.
Therefore, the sequence is bounded.

23
If a sequence diverges to ∞ or to −∞, then it is not bounded. A sequence that diverges to
∞ must, however, be bounded below (since such a sequence can have only a finite number
of negative terms). Similarly, if a sequence diverges to −∞, then it is bounded above. An
oscillating sequence may or may not be bounded.
Examples:
1. The sequence 1, −2, 3, −4, 5, −6, · · · oscillates and is neither bounded above nor bounded
below.
2. The sequence −1, 1, −1, 1, −1, 1, · · · oscillates between −1 and 1; it is bounded.
3. The sequence 1, 2, 1, 3, 1, 4, 1, 5, · · · oscillates and is bounded below but is not bounded
above.
Theorem 4.3.1: If the sequence {an }∞
n=1 of real numbers is convergent, then it is bounded.
Proof : Suppose an −→ L. Then given ε = 1, there exists N ∈ N such that

|an − L| < 1 for all n ≥ N.

Since |an | = |L + (an − L)| ≤ |L| + |an − L|, then

|an | < |L| + 1 for all n ≥ N. (1)

If we let M = max {|a1 | , |a2 | , · · · , |aN −1 |}, then

|an | ≤ M for all n = 1, · · · , N − 1. (2)

From (1) and (2) we have that

|an | < M + |L| + 1 for all n ∈ N,

which shows that {an }∞

n=1 is bounded. @

Remark 4.3.1 :
1. The converse of Theorem 4.3.1 is not necessarily true, i.e., a bounded sequence need
not be convergent.
2. From Theorem 4.3.1, an unbounded sequence is divergent.
Examples:
∞
1. The sequence {(−1)n }n=1 is bounded but not convergent.
2. The sequence 1, −2, 3, −4, 5, −6, · · · is divergent since it is neither bounded above nor
bounded below.
Theorem 4.3.2 (Bolzano-Weierstrass Theorem): Let {an }∞ n=1 be a bounded sequence
of real numbers. Then {an }∞
n=1 has a convergent subsequence.

4.4 Monotonic Sequences

As seen in Remark 4.3.1, a bounded sequence need not be convergent. In this section we
consider a condition which, together with boundedness, will ensure that a sequence is con-
vergent.

24
Definition 4.4.1: The sequence {an }∞ n=1 is non-decreasing if an ≤ an+1 for all n; in the
special case where an < an+1 for all n, the sequence {an }∞ n=1 is said to be increasing. The
sequence {an }∞ n=1 is non-increasing if an ≥ a n+1 for all n; in the special case where an > an+1
∞
for all n, the sequence {an }n=1 is said to be decreasing. A monotonic sequence is a sequence
which is either non-increasing or non-decreasing.
Remark 4.4.1 :
1. A non-decreasing sequence {an }∞
n=1 is always bounded below (by a1 ).

2. A non-increasing sequence {an }∞

n=1 is always bounded above (by a1 ).

Examples:
1 ∞
= 1, 1 21 , 1 34 , 1 87 , · · ·
 
1. The sequence 2 − 2n−1 n=1
is non-decreasing and bounded.
2. The sequence {n}∞
n=1 = {1, 2, 3, 4, · · · } is non-decreasing and bounded below (but not
bounded above).
Theorem 4.4.1: A non-decreasing sequence which is bounded above is convergent.
Proof : Suppose {an }∞
n=1 is non-decreasing and bounded above. Then the set A = {a1 , a2 , · · · }
is a non-empty subset of R which is bounded above. By the completeness axiom of real
numbers, this set has a least upper bound. Let

M = l.u.b {a1 , a2 , · · · } = l.u.b of A.

We will prove that an −→ M as n −→ ∞. Given ε > 0 the number M − ε is not an upper

bound for A. Hence, for some N ∈ N, aN > M − ε. But, since {an }∞
n=1 is non-decreasing,
this implies
an > M − ε for all n ≥ N. (1)
On the other hand, since M is an upper bound for A,

M ≥ an for all n ∈ N. (2)

Form (1) and (2) we conclude that

|an − M| < ε for all n ≥ N.

This proves that lim an = M. @

n→∞

Examples:  1 ∞
1. As seen in a previous example, the sequence 2 − 2n−1 n=1
is non-decreasing and
bounded. It converges to 2.
1 n
2. Consider the sequence {an }∞


n=1 where a n = 1 + n
. It can be shown that an ≤ an+1
∞
for every n ∈ N, i.e., {an }n=1 is non-decreasing, and that an < 3 for every n ∈ N, i.e.,
{an }∞ ∞
n=1 is bounded above (by 3). Hence, by Theorem 4.4.1, {an }n=1 is convergent. In
fact, lim an = e where the number e has decimal expansion 2.7182818 · · · .
n→∞

Remark 4.4.2 :
1. Theorem 4.4.1 gives a set of criteria that enables us to prove that a sequence converges

25
 without first guessing its limit.
2. A non-decreasing sequence does not oscillate, imlying that a non-decreasing sequence
that is not bounded above diverges to ∞.
Theorem 4.4.2: A non-decreasing sequence which is not bounded above diverges to ∞.
Proof : Suppose {an }∞
n=1 is non-decreasing but not bounded above. Given M > 0 we find
N ∈ N such that
an > M for all n ≥ N. (1)
Now, since M is not an upper bound for {a1 , a2 , · · · } there must exist N ∈ N such that
aN > M. Then, for this N, (1) follows from the hypothesis that {an }∞ n=1 is non-decreasing.
This proves the theorem. @
Theorem 4.4.3: Let {an }∞ n=1 be a non-increasing sequence of real numbers.
∞
(a) If {an }n=1 is bounded below, it is convergent.
(b) If {an }∞
n=1 is not bounded below, it diverges to −∞.
Proof : The proof of the theorem follows the proofs of Theorems 4.4.1 and 4.4.2 exactly,
with all upper bounds and least upper bounds replaced by lower bounds and greatest lower
bounds. @
Remark 4.4.3 : From Theorems 4.4.1 and 4.4.3, a bounded non-decreasing or non-increasing
sequence is convergent. Hence convergence and boundedness are equivalent for monotonic
sequences of real numbers.
Theorem 4.4.4: Let {an }∞ ∞
n=1 be a sequence of real numbers. Then {an }n=1 has a monotonic
subsequence.

4.5 Operations on Convergent Sequences

Remark 4.5.1 : Since sequences of real numbers are real-valued functions, the definition of
the sum, difference, product and quotient of sequences follows the properties of real-valued
functions.
If {an }∞ ∞ ∞ ∞ ∞
n=1 and {bn }n=1 are sequences of real numbers, then {an }n=1 +{bn }n=1 = {an + bn }n=1
and {an }∞ ∞ ∞ ∞ ∞
n=1 · {bn }n=1 = {an · bn }n=1 etc. Also, if c ∈ R, then c {an }n=1 = {can }n=1 .

Theorem 4.5.1: Let {an }∞ ∞

n=1 and {bn }n=1 be sequences of real numbers. If lim an = a and
n→∞
lim bn = b, then lim (an + bn ) = a + b.
n→∞ n→∞
Proof : Given ε > 0 we find N = N (ε) ∈ N such that

|(an + bn ) − (a + b)| < ε for all n ≥ N. (1)

Now,

|(an + bn ) − (a + b)| = |(an − a) + (bn − b)| ≤ |an − a| + |bn − b| .

Hence (1) holds if

|an − a| + |bn − b| < ε for all n ≥ N.

26
Thus we try to make both |an − a| and |bn − b| less than ε/2 by taking n sufficiently large.
Since an −→ a there exists N1 ∈ N such that |an − a| < ε/2 for all n ≥ N1 . Also, since
bn −→ b, there exists N2 ∈ N such that |bn − b| < ε/2 for all n ≥ N2 .
Let N = max {N1 , N2 }. Then |an − a| < ε/2 for all n ≥ N and |bn − b| < ε/2 for all n ≥ N so
that
|an − a| + |bn − b| < ε/2 + ε/2 = ε for all n ≥ N,
hence the proof. @
Theorem 4.5.2: Let {an }∞ n=1 be a sequence of real numbers, and let c ∈ R. If n→∞
lim an = a,
then lim can = ca.
n→∞
Proof : If c = 0, the theorem is obvious. Now, suppose c 6= 0. Given ε > 0 we find
N = N (ε) ∈ N such that

|can − ca| < ε for all n ≥ N. (1)

Now, since an −→ a there exists N ∈ N such that

|an − a| < ε/|c| for all n ≥ N.

But then
|c| · |an − a| < ε for all n ≥ N,
which is equivalent to (1). @
Theorem 4.5.3:
(a) If 0 < x < 1, then {xn }∞
n=1 converges to 0.

(b) If 1 < x < ∞, then {xn }∞n=1 diverges to ∞.

Proof :
(a) If 0 < x < 1, then xn+1 = x · xn < xn . Hence {xn }∞ n
n=1 is non-increasing. Since x > 0
for n ∈ N, {xn }∞ n ∞
n=1 is bounded below. By Theorem 4.4.3, {x }n=1 is convergent. Let
L = lim xn . From Theorem 4.5.2 (with c = x) it follows that lim x · xn = xL. That is,
n→∞ n→∞
∞ ∞
{xn+1 }n=1 converges to xL. But {xn+1 }n=1 is a subsequence of {xn }∞
n=1 . By Corollary
4.2.1, L = xL and so L (1 − x) = 0. Since x 6= 1, this shows that L = 0, and part (a)
is proved.
(b) If x > 1, then xn+1 = x · xn > xn so that {xn }∞
n=1 is non-decreasing. We will show that
{xn }∞n=1 is not bounded above. For if {xn ∞
} n=1 were bounded above, then by Theorem
n ∞
4.4.1 {x }n=1 would converge to some L ∈ R. But the same reasoning as in part (a)
would show that L = Lx, so that L = 0 = lim xn . But xn > 1 and so {xn }∞ n=1 obviously
n→∞
cannot converge to 0. This contradiction proves that {xn }∞
n=1 is not bounded above.
The conclusion follows from Theorem 4.4.2. @
Theorem 4.5.4: Let {an }∞ ∞
n=1 and {bn }n=1 be sequences of real numbers. If lim an = a and
n→∞
lim bn = b, then lim (an − bn ) = a − b.
n→∞ n→∞

27
Proof : Since lim bn = b, it follows from Theorem 4.5.2 (with c = −1) that lim (−bn ) = −b.
n→∞ n→∞
But then, using Theorem 4.5.1,

lim (an − bn ) = lim [an + (−bn )] = lim an + lim (−bn ) = a + (−b) = a − b,

n→∞ n→∞ n→∞ n→∞

which is what we wished to prove. @

Corollary 4.5.1: Let {an }∞ ∞
n=1 and {bn }n=1 be sequences of real numbers such that an ≤ bn
for all n ∈ N. If lim an = a and lim bn = b, then a ≤ b.
n→∞ n→∞
Proof : By Theorem 4.5.4, b − a = lim (bn − an ). But bn − an ≥ 0 for all n ∈ N. Hence by
n→∞
Theorem 4.2.2, b − a ≥ 0, which establishes the result. @
Remark 4.5.1 : Corollary 4.5.1 remains true even if an > bn for a finite number of values
of n.
Theorem 4.5.5: Let {an }∞ ∞
n=1 and {bn }n=1 be sequences of real numbers. If n→∞
lim an = a and
lim bn = b, then lim an bn = ab.
n→∞ n→∞
Proof : Given ε > 0 we find N ∈ N such that

|an bn − ab| < ε for all n ≥ N. (1)

Now,

|an bn − ab| = |(an bn − abn ) + (abn − ab)|

≤ |an bn − abn | + |abn − ab|
= |bn | · |an − a| + |a| · |bn − b| .

Hence (1) holds if

|bn | · |an − a| + |a| · |bn − b| < ε for all n ≥ N. (2)

Since {bn }∞
n=1 is convergent, then by Theorem 4.3.1, it is bounded. So there exists M > 0
such that |bn | ≤ M for all n ∈ N. Then (2) will certainly hold if

M |an − a| + |a| · |bn − b| < ε for all n ≥ N.

Now, choose N1 ∈ N so that

ε
M |an − a| < for all n ≥ N1 ,
2
and choose N2 ∈ N such that
ε
|a| · |bn − b| < for all n ≥ N2 .
2
ε ε
If N = max {N1 , N2 }, then M |an − a| < 2
for all n ≥ N and |a| · |bn − b| < 2
for all n ≥ N,
and it follows that
ε ε
M |an − a| + |a| · |bn − b| < + = ε for all n ≥ N. @
2 2

28
Lemma 4.5.1: Let {an }∞
n=1 be a sequences of real numbers such that lim an = a where n→∞
a 6= 0. Then lim (1/an ) = 1/a.
n→∞
Proof : Either a > 0 or a < 0. We prove the case a > 0 (the case a < 0 can be proved by
applying the first case to {−an }∞
n=1 ). Suppose a > 0. Given ε > 0 we find N = N (ε) ∈ N
such that
|1/an − 1/a| < ε for all n ≥ N
or
|an − a|
< ε for all n ≥ N (1)
|an a|
Now, there exists N1 ∈ N such that |an − a| < a2 for all n ≥ N1 . This implies that
a
an > for all n ≥ N1
2
so that
a2
for all n ≥ N1 .
an a >
2
In addition, there exists N2 ∈ N such that
a2 ε
|an − a| < for all n ≥ N2 .
2
Thus if N = max {N1 , N2 }, we have, for n ≥ N,
|an − a| 1 1 a2 ε
= · |an − a| < a2 · = ε.
|an a| |an a| /2 2
Hence (1) holds for this N. This completes the proof. @
Theorem 4.5.6: Let {an }∞ ∞
n=1 and {bn }n=1 be sequences of real numbers. If lim an = a and n→∞
lim bn = b where b 6= 0. Then lim (ab/bn ) = a/b.
n→∞ n→∞
Proof : Using Theorem 4.5.5 and Lemma 4.5.1 we have
1 1
lim an · =a· ,
n→∞ bn b
which is what we wished to show. @
2
Example: Prove that lim 3n −6n
2 = 3/5.
n→∞ 5n +4
Solution:
3n2 − 6n 3 − 6/n lim (3 − 6/n)
n→∞
lim = lim =
n→∞ 5n2 + 4 n→∞ 5 + 4/n2 lim (5 + 4/n2 )
n→∞
lim 3 + lim (−6/n) lim 3 − 6 lim 1/n
n→∞ n→∞ n→∞ n→∞
= = 2
lim 5 + lim 4/n2

n→∞ n→∞ lim 5 + 4 lim /n1
n→∞ n→∞
3 − 6 (0)
= 2 = /5.
3
5 + 4 (0)

29
4.6 Operations on Divergent Sequences
If {an }∞ ∞
n=1 is a divergent sequence, then {−an }n=1 is also divergent, and the sum of these
two sequences is clearly not divergent. Moreover, the product of the divergent sequence
∞
{(−1)n }n=1 with itself is not divergent. For sequences that diverge to ∞ we have the following
results:
Theorem 4.6.1: Let {an }∞ ∞
n=1 and {bn }n=1 be sequences of real numbers that diverge to ∞.
Then {an + bn }∞ ∞
n=1 and {an bn }n=1 diverge to ∞ also.
Proof : Given M > 0, choose N1 ∈ N such that

an > M for all n ≥ N1 ,

and choose N2 ∈ N such that

bn > 1 for all n ≥ N2 .
(The above is possible since both an −→ ∞ and bn −→ ∞ as n −→ ∞.) Then, for
N = max {N1 , N2 }, we have

an + bn > M + 1 > M for all n ≥ N,

and
an bn > M · 1 = M for all n ≥ N,
hence the proof. @
Theorem 4.6.2: Let {an }∞ ∞ ∞
n=1 and {bn }n=1 be sequences of real numbers. If {an }n=1 diverges
to ∞ and {bn }∞ ∞
n=1 is bounded, then {an + bn }n=1 diverges to ∞.
Proof : By hypothesis there exists Q > 0 such that

|bn | ≤ Q for all n ∈ N.

Given M > 0 choose N ∈ N such that

an > M + Q for all n ≥ N.

Then
an + bn ≥ an − |bn | > (M + Q) − Q = M for all n ≥ N.
That is,
an + bn > M for all n ≥ N
which shows that an + bn −→ ∞ as n −→ ∞. @
Corollary 4.6.1: If {an }∞ ∞ ∞
n=1 diverges to ∞ and {bn }n=1 converges, then {an + bn }n=1 di-
verges to ∞.
Proof : The proof follows directly from Theorems 4.3.1 and 4.6.2. @
Remark 4.6.1 : Theorems 4.6.1 and 4.6.2, and Corollary 4.6.1 hold if we replace ∞ with
−∞.

30
4.7 Limit Superior and Limit Inferior
Consider a sequence {an }∞ n=1 that is bounded above, say an ≤ M for all n ∈ N. Then for a
fixed n ∈ N, the set {an , an+1 , an+2 , · · · } is bounded above and hence has a least upper bound
Mn = l.u.b {an , an+1 , an+2 , · · · }. Moreover, Mn ≥ Mn+1 since Mn+1 = l.u.b {an+1 , an+2 , · · · }
is the least upper bound of a subset of {an , an+1 , an+2 , · · · }. Thus the sequence {Mn }∞ n=1 is
non-increasing and thus either converges or diverges to −∞.
Definition 4.7.1: Let {an }∞ n=1 be a sequence of real numbers that is bounded above, and
let Mn = l.u.b {an , an+1 , an+2 , · · · }.
(a) If {Mn }∞
n=1 converges, then lim sup an = lim Mn .
n−→∞ n→∞

(b) If {Mn }∞
n=1 diverges to −∞ then lim sup an = −∞.
n−→∞

Definition 4.7.2: If {an }∞

n=1 is a sequence of real numbers that is not bounded above, we
write lim sup an = ∞.
n−→∞

Examples:
1. Let an = (−1)n , n ∈ N. Then {an }∞
n=1 that is bounded above. In this case, Mn = 1 for
all n ∈ N and hence lim Mn = 1. Thus lim sup (−1)n = 1.
n→∞ n−→∞

2. Suppose an = −n, n ∈ N. Then Mn = l.u.b {−n, −n − 1, −n − 2, · · · } = −n. In this

case, Mn −→ −∞ as n −→ ∞ and so lim sup (−n) = −∞.
n−→∞

3. Consider the sequence an = 1, −1, 1, −2, 1, −3, 1, −4, · · · . Then Mn = 1 for all n ∈ N.
Hence lim sup an = 1.
n−→∞

4. The sequence {n}∞

n=1 is not bounded above. Hence, by Definition 4.7.2, lim sup n = ∞.
n−→∞

Theorem 4.7.1: If {an }∞

n=1 is convergent sequence of real numbers, then
lim sup an = lim an .
n−→∞ n→∞

Proof : Let L = lim an . Then given ε > 0 there exists N ∈ N such that
n→∞

|an − L| < ε for all n ≥ N,

or
L − ε < an < L + ε for all n ≥ N.
Thus if n ≥ N, then L + ε is an upper bound for {an , an+1 , an+2 , · · · } and L − ε is not a lower
bound. Hence
L − ε < Mn = l.u.b {an , an+1 , an+2 , · · · } ≤ L + ε,
and so, by Corollary 4.5.1,
L − ε < lim Mn ≤ L + ε.
n→∞
But lim Mn = lim sup an . Thus
n→∞ n−→∞

L − ε < lim sup an ≤ L + ε.

n−→∞

31
Since ε was arbitrary, this imlies lim sup an = L, which is what we wished to prove. @
n−→∞
∞
Now, suppose the sequence {an }n=1 of real numbers is bounded below. Then for a fixed
n ∈ N, the set {an , an+1 , an+2 , · · · } is bounded below and hence has a greatest lower bound.
Let mn = g.l.b {an , an+1 , an+2 , · · · }. Then {mn }∞ n=1 is non-decreasing sequence and hence
either converges or diverges to ∞.
Definition 4.7.3: Let {an }∞ n=1 be a sequence of real numbers that is bounded below, and
let mn = g.l.b {an , an+1 , an+2 , · · · }.
(a) If {mn }∞
n=1 converges, then lim inf an = lim mn .
n−→∞ n→∞

(b) If {mn }∞
n=1 diverges to ∞ then lim inf an = ∞.
n−→∞

Definition 4.7.4: If {an }∞

n=1 is a sequence of real numbers that is not bounded below, we
write lim inf an = −∞.
n−→∞

Examples:
n ∞
1. The sequence {an }∞
n=1 = {(−1) }n=1 is bounded below. In this case, mn = −1 for every
n ∈ N and hence lim inf (−1)n = lim mn = −1.
n−→∞ n→∞

2. Let an = n, n ∈ N. Then mn = g.l.b {n, n + 1, n + 2, · · · } = n. Hence lim mn = ∞ so

n→∞
that lim inf n = ∞.
n−→∞

3. The sequence an = 1, −1, 1, −2, 1, −3, 1, −4, · · · is not bounded below. Therefore, by
Definition 4.7.4, lim inf an = −∞.
n−→∞

4. The sequence {−n}∞

n=1 is not bounded below. Hence lim inf n = −∞.
n−→∞

Theorem 4.7.2: If {an }∞is convergent sequence of real numbers, then lim inf an = lim an .
n=1 n−→∞ n→∞
Proof : The proof of this theorem is similar to the proof of Theorem 4.7.1. @
Theorem 4.7.3: If {an }∞
n=1 is a sequence of real numbers, then lim inf an ≤ lim sup an .
n−→∞ n−→∞
Proof : If {an }∞
n=1 is bounded, then

mn = g.l.b {an , an+1 , an+2 , · · · } ≤ l.u.b {an , an+1 , an+2 , · · · } = Mn .

Thus mn ≤ Mn and so, by Corollary 4.5.1, lim inf an ≤ lim sup an . If {an }∞
n=1 is not bounded,
n−→∞ n−→∞
then lim sup an = ∞ or lim inf an = −∞ and lim inf an < lim sup an . @
n−→∞ n−→∞ n−→∞ n−→∞

Remark 4.7.1 : From Theorems 4.7.1 and 4.7.2 we see that if lim an = L, then we must
n→∞
have lim sup an = lim inf an = L.
n−→∞ n→∞

We now prove the converse of Remark 4.7.1.

Theorem 4.7.4: Let {an }∞
n=1 is a sequence of real numbers. If lim sup an = lim inf an = L
n−→∞ n−→∞
where L ∈ R, then {an }∞
n=1 is convergent and n→∞
lim an = L.
Proof : By hypothesis we have
L = lim sup an = lim l.u.b {an , an+1 , an+2 , · · · } .
n−→∞ n→∞

32
Thus given ε > 0 there exists N1 ∈ N such that

|l.u.b {an , an+1 , an+2 , · · · } − L| < ε for all n ≥ N1

implying that
an < L + ε for all n ≥ N1 (1)
Similarly, since lim inf an = L, there exists N2 ∈ N such that
n−→∞

|g.l.b {an , an+1 , an+2 , · · · } − L| < ε for all n ≥ N2

implying that
an > L − ε for all n ≥ N2 (2)
If N = max {N1 , N2 }, then from (1) and (2) we have

|an − L| < ε for all n ≥ N.

This proves that lim an = L. @

n→∞

Theorem 4.7.5: Let {an }∞

n=1 is a sequence of real numbers. If lim sup an = ∞ = lim inf an ,
n−→∞ n−→∞
then {an }∞
n=1 diverges to ∞.
Proof : Since lim inf an = ∞, given M > 0 there exists N ∈ N such that
n−→∞

mn = g.l.b {an , an+1 , an+2 , · · · } > M for all n ≥ N.

This implies that M is a lower bound (but not the g.l.b) for {an , an+1 , · · · } for all n ≥ N, so
that
an > M for all n ≥ N,
which establishes the required conclusion. @
Corollary 4.7.1: If {an }∞
n=1 is a sequence of real numbers diverging to ∞, then

lim sup an = ∞ = lim inf an .

n−→∞ n−→∞

Theorem 4.7.6: Let {an }∞ ∞

n=1 and {bn }n=1 be bounded sequences of real numbers. If an ≤ bn
for all n ∈ N, then lim sup an ≤ lim sup bn and lim inf an ≤ lim inf bn .
n−→∞ n−→∞ n−→∞ n−→∞
Proof : From the hypothesis an ≤ bn for all n ∈ N it is clear that

l.u.b {an , an+1 , an+2 , · · · } ≤ l.u.b {bn , bn+1 , bn+2 , · · · } ,

and
g.l.b {an , an+1 , an+2 , · · · } ≤ g.l.b {bn , bn+1 , bn+2 , · · · } .
Taking the limit as n −→ ∞ on both sides of these inequalities, and using Corollary 4.5.1,
we prove the theorem. @
Remark 4.7.2 :

33
 (i ) Theorems 4.7.6 remains true even if an > bn for a finite number of n.
(ii ) It is not always true that lim sup (an + bn ) = lim sup an + lim sup bn , even for bounded
n−→∞ n−→∞ n−→∞
sequences {an }∞ ∞
n=1 and {bn }n=1 .
n ∞ n+1 ∞
Example: Let {an }∞ ∞ 
n=1 = {(−1) }n=1 and {bn }n=1 = (−1) n=1
so that an + bn = 0 for
each n ∈ N. In this case, lim sup an = 1 = lim sup bn so that lim sup an + lim sup bn = 2.
n−→∞ n−→∞ n−→∞ n−→∞
However, lim sup (an + bn ) = 0.
n−→∞

Theorem 4.7.7: Let {an }∞ ∞

n=1 and {bn }n=1 be bounded sequences of real numbers. Then
(a) lim sup (an + bn ) ≤ lim sup an + lim sup bn
n−→∞ n−→∞ n−→∞

(b) lim inf (an + bn ) ≥ lim inf an + lim inf bn

n−→∞ n−→∞ n−→∞
Proof : We prove part (a) of the theorem; the proof of part (b) is much the same. Let

Mn = l.u.b {an , an+1 , an+2 , · · · } ,

and
Pn = l.u.b {bn , bn+1 , bn+2 , · · · } .
Then
ak ≤ Mn for all k ≥ n,
and
bk ≤ Pn for all k ≥ n,
so that
ak + bk ≤ Mn + Pn for all k ≥ n.
Thus Mn + Pn is an upper bound for {an + bn , an+1 + bn+1 , an+2 + bn+2 , · · · } so that

l.u.b {an + bn , an+1 + bn+1 , an+2 + bn+2 , · · · } ≤ Mn + Pn .

By Corollary 4.5.1 and Theorem 4.5.1,

lim l.u.b {an + bn , an+1 + bn+1 , an+2 + bn+2 , · · · } ≤ lim (Mn + Pn ) = lim Mn + lim Pn
n→∞ n→∞ n→∞ n→∞

or
lim sup (an + bn ) ≤ lim sup an + lim sup bn . @
n−→∞ n−→∞ n−→∞

Theorem 4.7.8: Any bounded sequence of real numbers has a converent subsequence.
Exercise:
1. Let {an }∞ ∞
n=1 be a sequence of real numbers and let {ani }n=1 be any subsequence of
{an }∞
n=1 . Prove that if lim sup an = M, then lim sup ani ≤ M.
n−→∞ n−→∞
2. Let {an }∞
n=1 be a bounded sequence of real numbers and let lim inf an = m. Prove that:
n−→∞
(a) there is a subsequence of {an }∞
n=1 which converges to m.
∞
(b) no subsequence of {an }n=1 converges to a limit less than m.

34
4.8 Cauchy Sequences
Definition 4.8.1: Let {an }∞ ∞
n=1 be a sequence of real numbers. Then {an }n=1 called a Cauchy
sequence if for any ε > 0 there exists N = N (ε) ∈ N such that

|am − an | < ε for all m, n ≥ N.

In other words, {an }∞

n=1 is Cauchy if am and an are close together when m and n are very
large.
Remark 4.8.1 : The Cauchy criterion is the most important criterion for proving that a
sequence converges without knowing its limit.
Theorem 4.8.1: Let {an }∞ ∞
n=1 be a convergent sequence of real numbers. Then {an }n=1 is
Cauchy.
Proof : Let L = lim an . Then, given ε > 0 there exists N ∈ N such that
n→∞

ε
|ak − L| < for all k ≥ N.
2
Thus if m, n ≥ N, we have
ε ε
|am − an | = |(am − L) + (L − an )| ≤ |(am − L)| + |(L − an )| < +
2 2
so that
|am − an | < ε for all m, n ≥ N,
which proves that {an }∞
n=1 is Cauchy. @

Lemma 4.8.1: If {an }∞ ∞

n=1 is a Cauchy sequence of real numbers, then {an }n=1 is bounded.
Proof : Given ε = 1, choose N ∈ N such that

|am − an | < 1 for all m, n ≥ N.

Then
|am − aN | < 1 for all m ≥ N. (1)
Hence, if m ≥ N, we have

|am | = |(am − aN ) + aN | ≤ |am − aN | + |aN |

and so, using (1)
|am | ≤ 1 + |aN | for all m ≥ N.
If M = max {|a1 | , · · · , |aN −1 |}, then

|am | < M + 1 + |aN | for all m ∈ N.

Hence {an }∞
n=1 is bounded.

35
Theorem 4.8.2: If {an }∞ ∞
n=1 is a Cauchy sequence of real numbers, then {an }n=1 is conver-
gent.
∞
Proof : By Theorem 4.4.4, {an }∞

n=1 has a monotonic subsequence anj j=1 . By Lemma 4.8.1,
∞ ∞
{an }∞
 
n=1 is bounded. Hence anj j=1 is bounded. Thus anj j=1 converges to some a ∈ R (by
∞
Remark 4.4.3). We will show that {an }∞

n=1 itself converges to a. Fix ε > 0. Since anj j=1
converges to a, there exists J ∈ N such that
ε
anj − a < for all j ≥ J. (1)
2
Since {an }∞
n=1 is Cauchy, there exists N ∈ N such that

ε
|am − an | < for all m, n ≥ N. (2)
2
We may choose N such that N ≥ J.
Now suppose k ∈ N and k ≥ N. Then k ≥ J, so (1) implies
ε
|ank − a| < .
2
Also, nk ≥ k ≥ N, so (2) implies
ε
|ak − ank | < .
2
Therefore,
ε ε
|ak − a| = |(ak − ank ) + (ank − a)| ≤ |ak − ank | + |ank − a| < +
2 2
so that
|ak − a| < ε for all k ≥ N,
and the theorem follows. @

4.9 Infinite Series

4.9.1 Convergence of Infinite Series
Definition 4.9.1: Let {an }∞ n=1 be a sequence of real numbers. An expression of the form
P ∞
P
a1 + a2 + · · · + an + · · · , denoted by an or an , is called an infinite series.
n=1
∞
P
Definition 4.9.2: Let an be an infinite series of real numbers and let sn = a1 +a2 +· · ·+an
n=1
be the sum of the first n terms (nth partial sum) of the sequence {an }∞ ∞
n=1 . Then {sn }n=1 is
∞
P
called the sequence of partial sums of the series an .
n=1
∞
an be an infinite series and let {sn }∞
P
Definition 4.9.3: Let n=1 be the corresponding
n=1
∞ ∞
an is said to converge if {sn }∞
P P
sequence of partial sums. Then n=1 converges, an diverges
n=1 n=1

36
 ∞
if {sn }∞ an oscillates if {sn }∞ ∞
P
n=1 diverges, and n=1 oscillates. If {sn }n=1 converges to s, then
n=1
∞
P ∞
P
s is called the sum of the series an and we write s = an or s = a1 + a2 + · · · + an + · · · .
n=1 n=1
∞
P n
P
In this case, an = s = lim sn = lim ai . If the series diverges to ∞, we write
n=1 n→∞ n→∞ i=1
∞
P ∞
P
an = ∞, and if it diverges to −∞, we write an = −∞.
n=1 n=1
∞
P ∞
P
Theorem 4.9.1: Let an and bn be convergent series.
n=1 n=1
∞
P ∞
P ∞
P ∞
P
(a) Then (an ± bn ) is also convergent and (an ± bn ) = an ± bn .
n=1 n=1 n=1 n=1
∞ ∞ ∞
(b) If c ∈ R, then
P P P
can is also convergent and can = c an .
n=1 n=1 n=1
Proof :
(a)
∞
X n
X
(an ± bn ) = lim (ai ± bi )
n→∞
n=1 i=1
n n
!
X X
= lim ai ± bi
n→∞
i=1 i=1
n
X n
X
= lim ai ± lim bi
n→∞ n→∞
i=1 i=1
∞
X ∞
X
= an ± bn .
n=1 n=1

∞ n n ∞
(b) If c ∈ R, then
P P P P
can = lim cai = c lim ai = c an . @
n=1 n→∞ i=1 n→∞ i=1 n=1
∞
P
Theorem 4.9.2: If an is a convergent series, then lim an = 0.
n=1 n→∞
∞
P
Proof : Suppose an = s. Then lim sn = s where sn = a1 + a2 + · · · + an . But then
n=1 n→∞
lim sn−1 = s. Since an = sn − sn−1 , then, by Theorem 4.3.4,
n→∞

lim an = lim (sn − sn−1 ) = lim sn − lim sn−1 = s − s = 0. @

n→∞ n→∞ n→∞ n→∞

Remark 4.9.1 : Theorem 4.9.2 gives a necessary but not sufficient condition that a series
be convergent.
∞
P
Corollary 4.9.1: If lim an 6= 0, then an diverges.
n→∞ n=1
Proof : The corollary is just the contrapositive of Theorem 4.9.2. @

37
Examples:
1−n
1. Suppose an = 1+2n . Then lim an = − 21 6= 0. Hence, by Corollary 4.9.1, the series
n→∞
∞
1−n
P
1+2n
diverges.
n=1
∞
(−1)n diverges since lim (−1)n does not exist.
P
2. The series
n=1 n→∞
∞
n2 n2 1
P
3. The series 3n2 +n
diverges since lim 2 = 6= 0.
n=1 n→∞ 3n +n 3

Exercise:
Show that the following series are divergent.
(i ) 21 + 32 + 43 + · · ·
(ii ) − 12 + 23 − 43 + · · ·
P∞
Theorem 4.9.3: Let an be a series of non-negative numbers. Then
n=1
∞
an converges if the sequence {sn }∞
P
(a) n=1 is bounded.
n=1
∞
an diverges if the sequence {sn }∞
P
(b) n=1 is not bounded.
n=1
Proof :
(a) Since an+1 ≥ 0 we have sn+1 = a1 + · · · + an + an+1 = sn + an+1 ≥ sn . Thus {sn }∞
n=1 is
non-decreasing and (by hypothesis) bounded. By Theorem 4.4.1, {sn }∞ n=1 converges,
∞
P
and thus an converges.
n=1
∞
{sn }∞ is not bounded, then by Theorem 4.3.1, {sn }∞
P
(b) If n=1 n=1 diverges. Hence an
n=1
diverges. @
Example:
∞
1
P
The harmonic series n
is divergent.
n=1
Proof : In this case sn = 1 + 12 + · · · + n1 . Consider the subsequence s1 , s2 , s4 , s8 , · · · , s2n−1 , · · ·
of {sn }∞
n=1 . We have
s1 = 1,
1
s2 = 1 + 2
= 32 ,
1 1 3 1 1
s4 = s2 + 3
+ 4
> 2
+ 4
+ 4
= 2,
1 1 1 1
s8 = s4 + 5
+ + +
6 7 8
> 2 + 81 + 81 + 1
8
+ 1
8
= 52 ,
..
.
In general, it may be shown by induction that s2n ≥ n+2
2
. Thus {sn }∞
n=1 contains a divergent
∞
subsequence and hence, by Corollary 4.2.1, {sn }∞ 1
P
n=1 diverges. Therefore, the series n
n=1
diverges. @

38
 ∞
P
Remark 4.9.2 : The divergence of the harmonic series shows that a series an may diverge
n=1
even if lim an = 0.
n→∞

4.9.2 Alternating Series

Definition 4.9.4: A series of the form a1 − a2 + a3 − a4 + · · · + (−1)n+1 an + · · · where an > 0
(or < 0) for all n is called an alternating series.
Examples:
(a) −1 + 2 − 3 + 4 − 5 + · · ·
(b) 1 − 12 + 13 − 14 + · · ·
(c) 1 − 21 + 1
4
− 18 + · · ·
are all alternating series.
Theorem 4.9.4 (Leibniz Test): If {an }∞ n=1 is a non-increasing sequence of positive numbers
∞
(−1)n+1 an is convergent.
P
such that lim an = 0, then the alternating series
n→∞ n=1

Examples:
∞
P 1
1 ∞
1. From a previous example n
is diverges. However, since n n=1
is non-increasing
n=1
∞
(−1)n+1
= 1 − 12 + 31 − 14 + · · ·
P
and lim an = 0, it follows from Theorem 4.9.4 that n
n→∞ n=1
converges.
 1 ∞
2. The sequence {an }∞
 1 1 1
n=1 = 2n−1 n=1 = 1, 2 , 4 , 8 , · · · is non-increasing and n→∞ lim an = 0.
∞
(−1)n+1 2n−1 1
= 1 − 21 + 41 − 18 + · · · converges.
P
Hence, by Theorem 4.9.4, the series
n=1

4.9.3 Absolute Convergence

∞
P
Definition 4.9.5: Let an be a series of real numbers.
n=1
∞
P ∞
P
(a) If |an | converges, we say that an converges absolutely.
n=1 n=1
∞
P ∞
P ∞
P
(b) If an converges but |an | diverges, we say that an converges conditionally.
n=1 n=1 n=1

Examples:
∞
an = 1 − 21 + 14 − 18 + · · · converges. Also, it is
P
1. From Example 2 above, the series
n=1
∞ ∞
|an | = 1 + 21 + 1
+ 81 + · · · converges. Hence
P P
easy to show that the series 4
an
n=1 n=1
converges absolutely.
∞
an = 1 − 21 + 13 − 14 + · · · converges. However,
P
2. From Example 1 above, the series
n=1

39
 ∞ ∞
|an | = 1 + 12 + 31 + 41 + · · · diverges. Hence
P P
from a previous example, the series an
n=1 n=1
converges conditionally.
∞
P ∞
P
Theorem 4.9.5: If an converges absolutely, then an converges.
n=1 n=1
Proof : Let sn = a1 + · · · + an . We wish to prove that {sn }∞
n=1 converges. By Theorem
∞
∞ P
4.8.2, it is enough to show that {sn }n=1 is Cauchy. By hypothesis |an | < ∞. Thus if
n=1
tn = |a1 | + · · · + |an |, then {tn }∞ ∞
n=1 converges. By Theorem 4.8.1, {tn }n=1 is Cauchy. Thus
given ε > 0 there exists N ∈ N such that

|tm − tn | < ε for all m, n ≥ N.

But (if m > n, say), |sm − sn | = |an+1 + · · · + am | ≤ |an+1 | + · · · + |am | = tm − tn . Thus

|sm − sn | < ε for all m, n ≥ N.

Hence {sn }∞
n=1 is Cauchy, which is what we wished to show. @

4.9.4 Tests for Convergence

In Subsubsection 4.9.1, we defined the convergence of a series in terms of the convergence of
its associated sequence of partial sums {sn }∞
n=1 . However, it is not always easy to find a simple
formula for the nth partial sums. Consequently, the technique of establishing convergence by
directly finding the limit of {sn }∞
n=1 does not always work. However, the situation is better
for a series whose terms are positive.
∞
P
Remark 4.9.3 : If an is a series of positive terms, then its sequence of partial sums
n=1
{sn }∞
n=1 is monotonic increasing, for sn+1 − sn = an+1 > 0 for all n ∈ N.
∞
P P∞
Theorem 4.9.6: If an is a series of positive terms, then an either converges to a
n=1 n=1
positive number or diverges to ∞.
∞
an and its sequence of partial sums {sn }∞
P
Proof : If the series n=1 is bounded, there exists
n=1
k > 0 such that k > sn+1 > sn ≥ a1 for all n ∈ N. Thus {sn }∞
n=1 being monotonic increasing
and bounded is convergent, by Theorem 4.4.1, and k ≥ lim sn ≥ a1 > 0.
n→∞
If {sn }∞
n=1 is unbounded, then {sn }∞
n=1 being monotonic increasing, by Remark 4.9.3, diverges
P∞
to ∞, by Theorem 4.4.2, i.e., an diverges to ∞. @
n=1
∞
P ∞
P
Theorem 4.9.7 (Comparison Test): Let an and bn be series of positive terms
n=1 n=1
and suppose that an ≤ bn for all n ∈ N.
∞
P ∞
P
(a) If bn converges, then an converges.
n=1 n=1
P∞ ∞
P
(b) If an diverges, then bn diverges.
n=1 n=1

40
Proof : Let sn = a1 + · · · + an and tn = b1 + · · · + bn be the corresponding nth partial sums.
Since an ≤ bn for all n ∈ N, then sn ≤ tn for all n ∈ N.
∞
bn converges, then {tn }∞ ∞
P
(a) If n=1 is bounded. Thus {sn }n=1 is also bounded. Since
n=1
∞
{sn }∞
P
n=1 is increasing and bounded, it is convergent (by Theorem 4.4.1). Hence an
n=1
converges.
∞
an diverges, then {sn }∞ ∞
P
(b) If n=1 is necessarily unbounded and thus {tn }n=1 is also
n=1
∞
unbounded and therefore {tn }∞
P
n=1 is divergent. Hence bn diverges. @
n=1
Examples:
∞ ∞
1 1
P P
1. Consider the series np
where p ≥ 2. It can be shown that n2
is convergent.
n=1 n=1
∞
1 1 1
P
Now, if p ≥ 2, then np
≤ n2
. Therefore, by comparison test, np
is also convergent.
n=1
∞
1
P
The series np
is called a p-series and actually converges for any p > 1.
n=1
∞
n+7
P
2. Determine whether the series n3 +1
converges or diverges.
n=1
Solution: We have that
n+7 n + 7n
3
≤ 3
n +1 n +1
8n
< 3
n
8
= 2.
n
∞ ∞
8 n+7
P P
The series n2
converges and thus, by comparison test, the series n3 +1
also
n=1 n=1
converges.
∞
P ∞
P
Theorem 4.9.8 (Limit Comparison Test): Let an and bn be series of positive
n=1 n=1
an
terms and let lim = L.
n→∞ bn
∞
P ∞
P ∞
P
(a) If L > 0, then an converges if and only if an converges (in other words, an
n=1 n=1 n=1
∞
P
and an converge or diverge together).
n=1
∞
P ∞
P
(b) If L = 0 and bn converges, then an converges also.
n=1 n=1
P∞ ∞
P
(c) If L = ∞ and bn diverges, then an diverges also.
n=1 n=1
Examples:
∞ ∞ ∞ ∞ 1
1 1 an n
P P P P n+1
1. Let an = n+1
and bn = n
. Then L = lim = lim 1 = lim = 1.
n=1 n=1 n=1 n=1 n→∞ bn n→∞ n n→∞ n+1
Since 0 < L < ∞, the two series converge or diverge together. But the harmonic series

41
 ∞ ∞
1 1
P P
n
diverges. Hence the series n+1
diverges also.
n=1 n=1
∞ ∞ ∞ ∞ 1
1 1 an n2 +1 n2
P P P P
2. If an = n2 +1
and bn = n2
, then L = lim = lim 1 = lim 2 +1 = 1.
n=1 n=1 n=1 n=1 n→∞ bn n→∞ n2 n→∞ n
∞ ∞
1 1
P P
Since 0 < L < ∞, and n2
is a convergent p-series, then n2 +1
converges also.
n=1 n=1
∞
P an+1
Theorem 4.9.9 (Ratio Test): Let an be a series of positive terms and let lim = l,
n=1 n→∞ an
finite or infinite. ∞
P
(a) If l < 1, then an converges.
n=1
P∞
(b) If l > 1, then an diverges.
n=1
(c) If l = 1, the series may converge or diverge and so, the test fails.
∞
an
P
Corollary 4.9.2: Let an be series of positive terms and let lim an+1 = l, finite or infinite.
n=1 n→∞
∞
P
(a) If l > 1, then an converges.
n=1
P∞
(b) If l < 1, then an diverges.
n=1
(c) If l = 1, the series may converge or diverge and so, the test fails.
Examples:
∞
2n
P
1. The series n!
converges since by the ratio test
n=1

an+1 2n+1 n! 2
lim = lim n
= lim = 0 < 1.
n→∞ an n→∞ (n + 1)! 2 n→∞ n + 1

12 ·22 22 ·32 32 ·42

2. The series 1!
+ 2!
+ 3!
+ · · · is convergent.
2 2
Proof : In this case, an = n ·(n+1)
n!
so that

an+1 (n + 1)2 · (n + 2)2 n!

lim = lim ·
n→∞ an n→∞ (n + 1)! n2 · (n + 1)2
(n + 2)2
= lim 2
n→∞ n (n + 1)
 
2 1
= lim 1 + ·
n→∞ n (n + 1)
= 0 < 1.

Hence by the ratio test, the series converges. @

∞
n!xn
P
3. Prove that for x > 0, the series nn
converges if x < e and diverges if x > e.
n=1

42
 n!xn
Proof : Since x > 0, the series is of positive terms and an = nn
for all n ∈ N. Thus

an+1 (n + 1)!xn+1 nn
= ·
an (n + 1)n+1 n!xn
(n + 1) n!xn x nn
= ·
(n + 1)n (n + 1) n!xn
 n
n
= x
n+1
x
=
n
1 + n1
an+1 x x x
and lim = lim 1 n = e . Hence, by the ratio test, the series converges if <1
n→∞ an n→∞ (1+ n ) e

or x < e, and diverges if xe > 1 or x > e.

Exercise:
3 2 4 3 5 4 n+1 n
Use the ratio test to show that the series 21 +





3
+ 5
+ 7
+· · ·+ 2n−1
+· · · converges.
∞
P √
Theorem 4.9.10 (Root Test): Let an be series of positive terms and let lim n an = l,
n=1 n→∞
finite or infinite. ∞
P
(a) If l < 1, then an converges.
n=1
P∞
(b) If l > 1, then an diverges.
n=1
(c) If l = 1, the series may converge or diverge and so, the test fails.
Examples:
∞ n
2−n+(−1) is convergent.
P
1. Show that the series
n=1
√ (−1)n
1
Solution: In this case, lim n an = lim 2−1+ n = 2
< 1. Therefore, by the root test,
n→∞ n→∞
the series is convergent.
∞ 2
nn
P
2. Show that the series n2
is convergent.
n=1 (n+1)
n2
Solution: In this case, an = n n2 > 0 for all n ∈ N. By the root test, the series
(n+1)
√ nn 1 1
converges since lim n an = lim (n+1) n = lim 1 n = e < 1.
n→∞ n→∞ n→∞ (1+ n )
∞
1
P
3. The series n pn
converges if p > 0 and diverges if p ≤ 0.
n=1
√
Proof : In this case, an = np1n > 0 for all n ∈ N and for all p ∈ R, lim n an = lim n1p .
√ n→∞ n→∞
Now, if p > 0, lim n an = 0 < 1 and the series converges. Since an = 1 for all n ∈ N
n→∞
∞ √
1
P
if p = 0, n pn
diverges. If p < 0, lim n an = ∞ > 1, and by the root test, the series
n=1 n→∞
diverges. @
2 3 2 4 3 5 4




4. Test for convergence of the series 1
+ 3
+ 5
+ 7
+ ···.

43
 n+1

n √ n+1 1
Solution: In this case, an = 2n−1 so that lim n an = lim = < 1. Hence, by
n→∞ n→∞ 2n−1 2
the root test, the series converges.
Exercise: ∞ 2
3n
P
1. Test for convergence of the series 43n
.
n=1
∞ ∞
e2n en
P P
2. Show that 3n
diverges while 3n
converges.
n=1 n=1
∞
2
1 −n
xn converges if x < e and diverges if x > e.
P
3. Show that for x > 0, 1+ n
n=1

Theorem 4.9.11 (Integral Test): Let f be a continuous, decreasing positive-valued func-

∞
P
tion defined on the interval [1, ∞) and consider the series an with an = f (n).
n=1
R∞ ∞
P
(a) If the improper integral 1
f (x) dx < ∞, then the series an converges.
n=1
R∞ P∞
(b) If the improper integral 1
f (x) dx = ∞, then the series an diverges.
n=1
Examples:
∞
1
P
1. The series np
converges if p > 1 and diverges if p ≤ 1.
n=1
1
Proof : Let f (x) = xp
.Suppose p > 1. Then
Z ∞ Z t
dx
p
= lim x−p dx
1 x t→∞ 1
 1−p t
x
= lim
t→∞ 1 − p
 1−p 1 
t 1
= lim −
t→∞ 1 − p 1−p
 
1 1
= lim p−1
+
t→∞ (1 − p) t p−1
1
= < ∞.
p−1
∞
1
P
Hence, np
converges if p > 1.
n=1
Now, suppose p < 1. Then
Z ∞ t
dx
Z
= lim x−p dx
1 xp t→∞ 1
 1−p t
x
= lim
t→∞ 1 − p
 1−p 1 
t 1
= lim −
t→∞ 1 − p 1−p
= ∞.

44
 ∞
1
P
Hence, np
diverges if p < 1.
n=1
Finally, if p = 1,
∞ Z t
dx dx
Z
= lim
1 x t→∞ 1 x

= lim [ln x]t1

t→∞
= lim [ln t − ln 1]
t→∞
= ∞.
∞
1
P
Hence, np
diverges if p = 1. @
n=1
∞
1
P
2. Determine the convergence of the series n2 +1
.
n=1
1
Solution: Let f (x) = x+1
. Then f satisfies the hypothesis of the integral theorem.
We now have that
∞ t
dx dx
Z Z
= lim dx
1 x + 1 t→∞ 1 x+1
t
= lim tan−1 x 1

t→∞
= lim tan−1 t − tan−1 1
 
t→∞
= lim tan−1 t − tan−1 1


t→∞
π π π
= − = .
2 4 4
R ∞ dx
Since 1 x+1 exists, the series converges.
Exercise:
Use the integral test to determine whether the following series converge or diverge.
∞
1
P
(a) n3
n=1
∞
n
P
(b) n2 +1
n=1
∞
1
P
(c) en
n=1
∞
ln n
P
(d ) n
n=1

45
5 Limits and Continuity of Functions
5.1 Limits of Functions
Definition 5.1.1: Consider a function f : A −→ R and let c be an accumulation point of
A. We say that f has a limit at c or f converges at c, if there exists a number L ∈ R such
that for any given ε > 0 there exists δ > 0 such that if x ∈ A and 0 < |x − c| < δ then
|f (x) − L| < ε. In this case, we write limf (x) = L or f (x) −→ L as x −→ c and say that
x→c
f converges to L at c or that f has a limit L at c. If f does not converge at c, we say that f
diverges at c.
Remark 5.1.1 : By definition, if limf (x) = L, then for any ε > 0 there exists δ > 0 such
x→c
that for all x ∈ (c − δ, c + δ) ∩ A not equal to c, it holds that f (x) ∈ (L − ε, L + ε).
Theorem 5.1.1: A function f : A −→ R can have at most one limit at c.
Proof : Suppose that f (x) −→ L and f (x) −→ M as x −→ c, and let ε > 0. Then
there exists δ > 0 such that |f (x) − L| < 2ε and |f (x) − M| < 2ε for all x ∈ A satisfying
0 < |x − c| < δ. If 0 < |x − c| < δ, then
|L − M| ≤ |f (x) − L| + |f (x) − M|
ε ε
< +
2 2
= ε.
Since ε > 0 is arbitrary, then |L − M| < ε implies that L = M. @
Examples:
1. Define the function f : R −→ R by f (x) = 5x + 3. Prove that lim f (x) = 13.
x→2
Proof : We have that
|f (x) − 13| = |(5x + 3) − 13|
= |5x − 10|
= 5 |x − 2| .

Now, if 0 < |x − 2| < 5ε , then |f (x) − 13| < 5 5ε = ε. Thus, given ε > 0 we let δ = 5ε

so that if 0 < |x − 2| < δ, then |f (x) − 13| < ε. Hence, by definition, lim f (x) = 13. @
x→2
x+1
2. Let f : R −→ R where f (x) = x2 +3
. Prove that lim f (x) = 21 .
x→1
Proof : We have that
1 x+1 1
f (x) − = 2 −
2 x +3 2
x2 − 2x + 1
=
2 (x2 + 3)
|x − 1|2
=
2 (x2 + 3)
< |x − 1|2 .

46
 √
Let ε > 0 be arbitrary and let δ = ε. Now, if 0 < |x − 1| < δ, then |x − 1|2 < δ 2 = ε.
Hence if 0 < |x − 1| < δ, then f (x) − 21 < ε. Therefore, lim f (x) = 12 . @
x→1

Theorem 5.1.2: Let f : A −→ R be a function and let c be an accumulation point of A.

Then limf (x) = L if and only if for every sequence {xn } in A converging to c (with xn 6= c
x→c
for all n ∈ N) the sequence {f (xn )} converges to L.
Corollary 5.1.1: Let f : A −→ R be a function. Let c be an accumulation point of A and
let L ∈ R. Then f does not converge to L at c if and only if there exists a sequence {xn } in
A converging to c with xn 6= c for all n ∈ N, and such that {f (xn )} does not converge to L.
Corollary 5.1.2: Let f : A −→ R be a function and let c be an accumulation point of A.
Suppose that {xn } and {yn } are sequences in A converging to c, with xn 6= c and yn 6= c for
all n ∈ N. If f (xn ) and f (yn ) converge but lim f (xn ) 6= lim f (yn ), then f does not have a
x→∞ x→∞
limit.
Example:
Prove that lim x1 does not exist.
x→0 1 ∞
Proof : Consider {xn }∞ n=1 = n n=1
, which clearly converges to c = 0 and xn 6= 0 for all
n ∈ N. Then {f (xn )} = {n} which is unbounded and thus does not converge. Thus, by
Corollary 5.1.1, lim x1 does not exist. @
x→0

Exercise:
Use the ε-δ definition of the limit of a function to prove that:
(a) lim (2x − 5) = −7
x→−1
(b) lim 2x+3 = 3
x→3 4x−9
2 −3x
(c) lim xx+3 =2
x→6

Theorem 5.1.3: Let f, g : A −→ R be functions and let c be an accumulation point of A.

Suppose that limf (x) = L and limg (x) = M. Then:
x→c x→c
(a) lim (f ± g) (x) = L ± M
x→c
(b) lim (f g) (x) = LM
x→c
 
(b) lim fg (x) = M L
, if M 6= 0.
x→c

Corollary 5.1.3: Let f1 , · · · , fk : A −→ R be functions and let c be an accumulation point

of A. If lim fi (x) exists for each i = 1, 2, · · · , k, then
x→c
Pk k
P
(a) lim fi (x) = limfi (x)
x→c i=1 i=1 x→c
Qk Qk
(b) lim fi (x) = limfi (x)
x→c i=1 i=1 x→c

Examples:
1. If f (x) = a0 + a1 x + a2 x2 + · · · + an xn is a polynomial function, then lim f (x) = f (c)
x→c
for every c ∈ R. If g (x) = b0 + b1 x + b2 x2 + · · · + bm xm is another polynomial function

47
 and g (x) 6= 0 in a neighbourhood of x = c and limg (x) = g (c) 6= 0, then lim fg(x)
(x)
= f (c)
g(c)
.
x→c x→c
2
2. Prove that lim xx−2
−4
=4
x→2
x2 −4
Proof : Observe that lim (x − 2) = 0. If x 6= 2, then x−2
= x + 2. Hence the functions
x→2
x2 −4
f (x) = x−2
and g (x) = x + 2 are equal at every point in R\ {2}. Since lim g (x) = 4,
x→2
it follows that lim f (x) = 4. @
x→2

Definition 5.1.2: Let f : A −→ R be a function. Then lim f (x) = L (i.e., f (x) −→ L as

x→∞
x −→ ∞) if given ε > 0, there exists M ∈ R such that |f (x) − L| < ε for all x > M.
Remark 5.1.2 : Definition 5.1.2 requires that the set A contains some interval of the form
(c, ∞).
Example:
Prove that lim x12 = 0.
x→∞
Proof : Given ε > 0, we need to find M ∈ R such that
1
− 0 < ε for all x > M. (1)
x2
Since (1) is equivalent to
1 √
< ε for all x > M,
x
it is clear that (1) will hold if we take M = √1ε . @
Definition 5.1.3: Let f : A −→ R be a function and let c be an accumulation point of A.
We say that f approaches L as x approaches c from the right, if given ε > 0, there exists
δ > 0 such that for all x ∈ A for which c < x < c + δ, then |f (x) − L| < ε. In this case, we
write lim+f (x) = L and the number L is called the right-hand limit of f at c. Similarly, we
x→c
say that f approaches M as x approaches c from the left, if given ε > 0, there exists δ > 0
such that for all x ∈ A for which c − δ < x < c, then |f (x) − M| < ε. In this case, we write
lim− f (x) = M and the number M is called the left-hand limit of f at c.
x→c

Remark 5.1.3 :
1. The statement lim+f (x) = L involves only values of f (x) for x to the right of c, while
x→c
lim− f (x) = M involves only values of f (x) for x to the left of c.
x→c

2. lim+f (x) and lim− f (x) may exist without being equal to each other.
x→c x→c

3. limf (x) = L if and only if lim+f (x) = lim− f (x) = L.

x→c x→c x→c

Example:(
x if 0 ≤ x < 1
If f (x) = , then lim− f (x) = 1 while lim+ f (x) = 2.
3 − x if 1 ≤ x ≤ 2 x→1 x→1

5.2 Continuous and Discontinuous Functions

Definition 5.2.1: Let f : A −→ R be a function. Then f is continuous at c ∈ A if:

48
 (i ) limf (x) exists,
x→c

(ii ) f (c) is defined, and

(iii ) limf (x) = f (c).
x→c

In other words, f is continuous at c ∈ A if for each ε > 0, there is some δ > 0 such that if
x ∈ A and |x − c| < δ, then |f (x) − f (c)| < ε. If f is not continuous at c, then we say that
f is discontinuous at c. The function f is continuous on A if f is continuous at each x ∈ A.
Remark 5.2.1 : In Definition 5.2.1, the number δ, in general, depends on ε and the point c.
Definition 5.2.2: Consider a function f : A −→ R and let c ∈ A. Then f is right-continuous
at c if lim+f (x) = f (c), f is left-continuous at c if lim− f (x) = f (c).
x→c x→c

Remark 5.2.2 : A function f : A −→ R is continuous at c if and only if f is both right-

continuous and left-continuous at c.
Theorem 5.2.1: The function f : A −→ R is continuous at c ∈ A if and only if for every
sequence {xn } in A such that lim xn = c, we have lim f (xn ) = f lim xn = f (c).
n→∞ n→∞ n→∞

Theorem 5.2.2: The function f : A −→ R is discontinuous at c ∈ A if and only if there

exists a sequence {xn } in A converging to c but f (xn ) does not converges to f (c).
Theorem 5.2.3: Let f, g : A −→ R be continuous functions at c ∈ A and let b ∈ R. Then
f + g, f − g, f g, and bf are also continuous at c. Furthermore, if g (c) 6= 0, then fg is also
continuous at c.
Proof : Since f and g are continuous at c we have lim f (x) = f (c) and limg (x) = g (c). By
x→c x→c
Theorem 5.1.3, lim [f (x) + g (x)] = f (c)+g (c) or, equivalently, lim (f + g) (x) = (f + g) (c).
x→c x→c
Hence f + g is continuous at c. The remainder of the theorem is proved similarly. @
Corollary 5.2.1: Let f, g : A −→ R be continuous functions on A and let b ∈ R. Then
f + g, f − g, f g, and bf are also continuous on A. Furthermore, if g (c) 6= 0, then fg is also
continuous on A.
Examples:
1. Consider the function f : R −→ R defined by f (x) = 5x + 3. In a previous example,
it was proved that lim f (x) = 13. Furthermore, f (2) = 13. Therefore, f is continuous
x→2
at x = 2.
2. Let f (x) = sinx x where x ∈ R\ {0}. Then f is not continuous at x = 0 since it is not
defined at x = 0, even though lim sinx x exists (and is equal to 1). However, the function
( x→0
sin x
x
if x 6= 0
g defined by g (x) = is continuous at x = 0 since lim g (x) = g (0).
1 if x = 0 x→0

3. Prove that f (x) = x is continuous on R.

Proof : Suppose c ∈ R. Let ε > 0 be arbitrary. Let δ = ε. If 0 < |x − c| < δ, then
|f (x) − f (c)| = |x − c| < δ = ε. @
√
4. The function f (x) = x is discontinuous at all c < 0, right-continuous at c = 0, and

49
 continuous at all c > 0. To prove that f is continuous at all c > 0, we have that
√ √
|f (x) − f (c)| = x − c
√ √
√ √ x+ c
= x− c · √ √
x+ c
|x − c|
=√ √
x+ c
1
≤ √ |x − c| .
c
√
Hence, given ε > 0, suppose 0 < |x − c| < cε. Then
1 1 √
|f (x) − f (c)| < √ |x − c| = √ cε = ε.
c c

5. If f (x) = a0 + a1 x + a2 x2 + · · · + an xn is a polynomial function, then limf (x) = f (c)

x→c
for every c ∈ R. Thus f is continuous on R. If g (x) = b0 + b1 x + b2 x2 + · · · + bm xm
is another polynomial function and g (c) 6= 0, then lim fg(x)
(x)
= fg(c)
(c)
. Hence, h (x) = fg(x)
(x)
x→c
is continuous at every c where g is non-zero.
(
1
if x 6= 0
6. Determine the points of continuity of the function f (x) = x
0 if x = 0.
1
Solution: Suppose that c 6= 0. Then limf (x) = c = f (c). Hence, f is continuous at
x→c
c ∈ R\ {0}. Now, let c = 0. The sequence xn = n1 converges to c = 0 but f (xn ) = n
does not converge. Hence lim f (x) does not exist. Thus, even though f (0) = 0 is
x→0
well-defined, f is discontinuous at c = 0.
Exercise: 
1
 1−x
 if x < 0
2
1. Show that the function f (x) = x + 1 if 0 ≤ x ≤ 2 is continuous at x = 0 but

x + 4 if x > 2

discontinuous at x = 2. (
1 if x ∈ Q
2. Show that the function f (x) = is discontinuous everywhere.
0 if x ∈ R\Q

5.3 Uniform Continuity of Functions

Definition 5.3.1: A function f : A −→ R is said to be uniformly continuous on A if
for each ε > 0, there exists δ > 0 such that for all x, y ∈ A such that |x − y| < δ, then
|f (x) − f (y)| < ε.
Remark 5.3.1 : In the case of uniform continuity of a function f : A −→ R, the number
δ depends only on ε and not on the points of A. This is unlike in the case of pointwise
continuity, where δ, in general, depends on ε and the points of A. Thus uniform continuity
is a global property whereas the pointwise continuity is a local property.

50
Examples:
1. Let k be any non-zero constant. Show that f (x) = kx is uniformly continuous on R.
Solution: We have that |f (x) − f (y)| = |kx − ky| = |k| |x − y|. Hence, for any ε > 0,
ε
we let δ = |k| , and thus if x, y ∈ R such that |x − y| < δ, then |f (x) − f (y)| < ε.
2. The function f (x) = x1 is uniformly continuous on the set [1, 3].
Proof : Let x, y ∈ [1, 3]. Then 13 ≤ x1 ≤ 1 and 31 ≤ y1 ≤ 1. Now, we have that

1 1
|f (x) − f (y)| = −
x y
y−x
=
xy
1 1
= |x − y|
|x| |y|
≤ |x − y| .

Given ε > 0, we can set δ = ε so that if |x − y| < δ, then |f (x) − f (y)| < ε. @
Exercise:
Show that each of the following functions is uniformly continuous on R:
(a) f (x) = sin x
(b) f (x) = sin 2x
1
(c) f (x) = 1+x 2

Theorem 5.3.1: Let f : A −→ R be a function. The following are equivalent:

(a) The function is not uniformly continuous on A.
(b) There exists ε > 0 such that for every δ > 0, there exists x, y ∈ A such that |x − y| < δ
but |f (x) − f (y)| ≥ ε.
(c) There exists ε > 0 and a pair of sequences {xn } and {yn } in A where lim (xn − yn ) = 0
x→∞
and |f (xn ) − f (yn )| ≥ ε.
Example:
1. The function f (x) = x1 is not uniformly continuous on A = (0, 1]. 
Proof : Let ε = 1. Consider the sequences {xn } = n1 and {yn } = 2n 1
in A. Then
 
1 1 1
lim (xn − yn ) = lim − = lim =0
x→∞ x→∞ n 2n x→∞ 2n

but
|f (xn ) − f (yn )| = |n − 2n| = n ≥ 1 = ε.
Hence, by Theorem 5.3.1, the conclusion follows. @
1
Remark 5.3.2 : In general, the function f (x) = x
is not uniformly continuous on the set
A = (0, ∞).
2. The function f (x) = x2 is not uniformly continuous on [1, ∞), since the function is
1

unbounded as x −→ ∞. We accordingly can set {xn } = {n} and {yn } = n + n so

51
 that  2
1 1
|f (xn ) − f (yn )| = n+ − n2 = 2 + > 2.
n n2
Exercise:
2
Show that the function f (x) = x
is not uniformly continuous on A = (0, 2].
Theorem 5.3.2: If the function f : A −→ R is uniformly continuous on A, then it is
continuous on A.
Proof : Suppose f is uniformly continuous on A. Then for ε > 0, there exists δ > 0 such
that for all x1 , x2 ∈ A such that |x1 − x2 | < δ implies |f (x1 ) − f (x2 )| < ε. Now, let x ∈ A.
Then on taking x1 = x, we find for ε > 0, there exists δ > 0 such that for fixed x2 ∈ A such
that |x − x2 | < δ implies |f (x) − f (x2 )| < ε. Hence f is continuous at every x2 ∈ A, i.e., f
is continuous on A. @
Theorem 5.3.3: If the function f : A −→ R is continuous on the closed interval A = [a, b],
then it is uniformly continuous on A.
Proof : Suppose that f is not uniformly continuous on A. Then by Theorem 5.3.1, there
exists ε > 0 and two sequences {xn } and {yn } in A such that lim (xn − yn ) = 0 and
x→∞
|f (xn ) − f (yn )| ≥ ε. Since {xn } is a bounded sequence, by Theorem 4.3.2, there is a sub-
sequence {xnk } of {xn } that converges to a point x0 ∈ [a, b]. But then the corresponding
subsequence {ynk } of {yn } also converges to x0 since
ε ε
|ynk − x0 | ≤ |ynk − xnk | + |xnk − x0 | < + = ε.
2 2
But since the sequences {xnk } and {ynk } both converge to x0 , and f is continuous at
x0 , it follows that the sequences {f (xnk )} and {f (ynk )} must both converge to f (x0 ) or
lim |f (xnk ) − f (ynk )| = 0 which contradicts the assumption that |f (xnk ) − f (ynk )| ≥ ε for
x→∞
all k. @
Example:
Consider the function f (x) = x1 . By Theorem 5.3.3, f is uniformly continuous on any closed
interval that does not contain 0. By Definition 5.3.1, f is uniformly continuous on any set of
the form [c, ∞) for c > 0, because if x, y ∈ [c, ∞) and |x − y| < δ = c2 ε, then
1 1 y−x
− =
x y xy
1 1
= |x − y|
|x| |y|
11 2
< cε
cc
= ε.
Exercise:
1. Let f : A −→ R and g : A −→ R be uniformly continuous functions on A. Prove that
f + g is uniformly continuous on A.
2. Prove that f (x) = x2 is not uniformly continuous on R.
3. Show that f (x) = sin x2 is not uniformly continuous on R.

52