Chapter 1

Topology

1.1 Interior of a set

When we say that “the interior of a house is nice”, we mean the
inside portion of the house is neat and tidy. The interior has, generally,
nothing to do with colour from outside or any other show pieces, which
are kept outside the house. The mathematical concept “interior” is
very much similar to whatever intuitive feeling that we have about the
word interior. Let us see the actual definition:
Definition 1. The interior in the metric space X of a subset A of X
is defined to be the union of the family of open subsets of X, which are
contained in A.
The definition been stated, the questions that naturally strike us
are these: Is it possible that such a family may be empty? What is
the guaranty that this family does contain at least one element? Well,
obviously such a family cannot be empty, for ∅ is an open subset of X
and it is contained in any A.

The interior of a set A is denoted by Ao .

Warning: Observe the definition carefully. When we say family of

open subsets, these should be open in X and not in A. Otherwise, A is
always open in A and therefore, interior of any A would turn out to be
A itself.
Example 1. X o = X, where X is any metric space. This should be
obvious, because X is open in X and any open subset of X cannot
contain X.
Example 2. Let A = [0, 1]. Consider A as a subset of R with Euclidean
metric. Then it is clear that Ao contains (0, 1/n) ∀n ∈ N. Also, if
(a, b) is contained in [0, 1] then it is not difficult to show that there
exist p, q ∈ N such that 0 < 1/p ≤ a < b ≤ 1/q ≤ 1. Therefore,
[0, 1]o = (0, 1).

1
Check Your Progress
Find interior of A in each of the following:

1. A = (0, 1), where A is a subset of R with Euclidean metric.

2. A = R − {0}, where A is a subset of R with Euclidean metric.

——————————————————————————————

Theorem: Suppose (X, d) is a metric space and suppose A ⊆ X.

Then we get the following:
(i) Ao ⊆ A
(ii) Ao is open
(iii) A is open if and only if A = Ao

Proof:(i) Let {Uλ }λ∈Λ be a family S of open subsets of X such that

o
Uλ ⊆ A, for each λ. Then A = λ∈Λ Uλ ⊆ A.
(ii) By definition of interior, it is union of open sets. Hence, Ao is open.
(iii) Suppose A = Ao . Then A is open, since Ao is open by (ii) above.
Conversely, suppose A is open. Since A ⊆ A, we S have A ∈ {Uλ }λ∈Λ .
Let A = Uλo , for some λo ∈ Λ. Then A = Uλo ⊆ λ∈Λ Uλ = Ao ⊆ A.
This shows that A = Ao .

From (ii) and (iii) above, it is clear that interior of interior of A is

interior of A. That is, (Ao )o = Ao .

Again, recall that the interior of A is the union of every open set,
which is contained in A. Thus, Ao is the largest open set contained in
A.

It should be noted that Ao may be empty. For example, if A = Q,

then Ao = ∅.

Definition 2. A point a is said to be an interior point of A if a ∈ Ao .

Using metric of X, it is easy to define an interior point and interior

of a set A.

Definition 3. Let A ⊆ X and a ∈ A. Then a is said to be an interior

point of A if there exists r > 0 such that B(a, r) ⊆ A.

Definition 4. Ao is the set of all interior points of A.

Since, the mathematical idea of interior is similar to our intuition,

it should not be surprising if it turns out that interior respects subset
relationship. We have a theorem in this regard:

Theorem: Suppose (X, d) is a metric space and suppose that A, B ⊆

2
X. Then A ⊆ B ⇒ Ao ⊆ B o .

Proof: Let A ⊆ B. We know that Ao ⊆ A. Therefore, Ao ⊆ B.

Thus Ao is an open subset of B. Now B o is the largest open subset of
B. Therefore, Ao ⊆ B o and the proof is complete.

Note that A ⊆ A ∪ B. Thus, as a consequence of the above theo-

rem, Ao ⊆ (A ∪ B)o and clearly, B o ⊆ (A ∪ B)o . Therefore, we get
Ao ∪ B o ⊆ (A ∪ B)o .

Suppose, in general it were true that Ao ∪ B o ⊇ (A ∪ B)o , then we

would have got a nice relation Ao ∪ B o = (A ∪ B)o , ∀A, B ⊆ X. How-
ever, let X = R and let A = Q and B = R \ Q. Then interiors of
both A and B are empty, but interior of their union is whole of R, since
A ∪ B = R. Thus, we can only say that Ao ∪ B o ⊆ (A ∪ B)o .

In case of intersection, however, the interior respects equality. That

is, we have,

Theorem: (A ∩ B)o = Ao ∩ B o .

Proof: We know that A ∩ B ⊆ A, B. Therefore, by the above the-

orem, it is obvious that (A ∩ B)o ⊆ Ao ∩ B o . Also, we have Ao ⊆ A
and B o ⊆ B. Thus, Ao ∩ B o ⊆ A ∩ B.

But Ao ∩ B o is an open subset of A ∩ B and (A ∩ B)o is the largest

open subset of A ∩ B. Therefore, (A ∩ B)o ⊇ Ao ∩ B o .

Thus, (A ∩ B)o = Ao ∩ B o .

1.2 Limit Points

When we say that the limit of a sequence xn is a, we mean that
the terms of the sequence xn move arbitrarily close to a as n increases.
The sequence xn may never have the term a as one of its terms. The
symbol xn → a indicates the proximity of a from xn . The concet of
limit point (also know as accumulation point) is related with this idea.
If we say that a is a limit point of a set A, we mean that a is arbitrarily
colse to A. Let us have a formal definition:
Definition 5. Let A ⊆ X. A point a ∈ X is called a limit point of A
if ∀r > 0, the set B(a, r) contains at least one point of A other than a.
Note that x is a point of the metric space X. It may or may not be
a member of A. The idea of the limit point will be clear through some

3
examples.

Example 3. Let A = R with Euclidean metric. Then every point of A

is a limit point of A, for if a ∈ A, then for any r > 0, the set B(a, r) does
contain at least one point of A, other than a. In particular, a+r/2 ∈ A.
Example 4. Let A = (0, 1) ⊆ R, with Euclidean metric. Then 0 and
1 are limit points of A. However, it is not difficult to show that any
a > 1 or a < 0 is not a limit point of A. Thus, the set of limit points
of (0, 1) under Euclidean metric is [0, 1].
Example 5. Let A = (0, 1) ⊆ R as before but with discrete metric.
Let a ∈ R. Then for 0 < r < 1, the set B(a, r) = {a}. Therefore, a
is not a limit point of A. In fact, even if A is any other subset of a
discrete metric space X, then also no member of X is a limit point of
A. Thus, the set of limit points of A is ∅.
Example 6. Let A = {1/n|n ∈ N} ⊆ R, under Euclidean metric.
Then the set of limit points of A is {0}.
The above examples illustrate some of the properties of limit points,
namely,
(i) a set may have no limit point,
(ii) a set may have finitely many limit points,
(iii) a set may have infinitely many limit points,
(iv) a limit point of a set may or may not be a member of the set,
(v) set of limit points depends upon the underlying metric of the space.

Check Your Progress

Prove the following:
1. Let A = {1/n|n ∈ N} ⊆ R under Euclidean metric. Then the set
of limit points of A is {0}.
2. Let A = N ⊆ R under Euclidean metric. Then the set of limit
points of N is ∅.
3. Set of limit points of Q = Set of limit points of R \ Q = R.
Definition 6. The set of all the limit points of a set A is called the
derived set of A and it is denoted by A0 .
Definition 7. A point of A is called isolated point of A, if it is not a
limit point of A. It should be clear that the set of all isolated points of
A is A \ A0 .
Example 7. Let A = (0, 1) ⊆ R, with Euclidean metric. Then A0 =
[0, 1].

4
Example 8. Let A = (0, 1) ⊆ R, with discrete metric. Then A0 = ∅
Example 9. Let A = {1/n|n ∈ N} ⊆ R, under Euclidean metric.
Then every point of A is an isolated point of A.
Observe the difference between the definitions of limit point and
isolated point. A limit point of a set may not be a member of the set.
However, an isolated point of a set (if exists) has to be a member of
that set.

So how to decide whether or not a point is an isolated point of a set?

Let A ⊆ X. Then a ∈ A is called an isolated point of A if there exists
r > 0, such that the set A ∩ B(a, r) is a singleton set {a}. Similarly,
we have a theorem about limit point of a set.

Theorem: If a is a limit point of A then every open ball with center

at a contains infinitely many points of A.

Proof: Suppose there exists r > 0 such that B(a, r) contains finitely
many points, say x1 , x2 , ..., xn of A.

Let r0 = min {(d(x1 , a), (d(x2 , a), (d(x3 , a), ..., (d(xn , a)}. Then B(a, r0 )
contains no point of A, other than a. Thus, a cannot be a limit point of
A. But this is a contradiction. Thus, if a is a limit point of A then every
open ball with centre at a must contain infinitely many points of A.

Note that this theorem allows no limit point for any finite set.

1.3 Closed Set

Definition 8. A subset A of X is said to be closed in X if the com-

plement of A is open in X.
Note the following:
1. A set may be closed, when viewed as a subset of, say, X and yet
may not be closed, when viewed as a subset of, say, Y , even if the
underlying metric is same in these cases.
2. A set may be closed in (X, d1 ) and yet may not be closed in
(X, d2 ).
3. The terms open set and closed set are not opposites of each
other. That is, a set may neither be open nor be closed. Similarly,
a set may be both open and closed. Such a set is, sometimes,
called clopen.

5
Example 10. The set (0, 1) is open in R with Euclidean metric. There-
fore, its complement (−∞, 0] ∪ [1, ∞) is closed in R with Euclidean
metric

Example 11. The set [0, 1] is closed in R with Euclidean metric, for,
its complement (−∞, 0) ∪ (1, ∞) is open in R with Euclidean metric.

Example 12. For any metric space (X, d), X and ∅ are both open.
Therefore, their respective complements ∅ and X are closed in (X, d).
Thus, in any metric space (X, d), both X and ∅ are open as well as
closed.

Example 13. Let X be a metric space with discrete metric. If A ⊆ X,

then A is open in X. Therefore, X \ A is closed in X. Thus, every
subset A of a discrete metric space is both open and closed.

Example 14. In case of Euclidean metric on R2 , both {(a, b)|a2 + b2 ≤

1} and {(a, b)|a2 + b2 = 1} are closed sets.

Check Your Progress

1. Show that Z is closed in R under Euclidean metric.

2. Show that [0, 1] ∪ [2, 3] is closed in R under Euclidean metric.

3. Is Q closed in R under Euclidean metric?

It is worth to note that corresponding to every statement about open

sets, there exists a statement about closed sets. Moreover, just like we
define open ball, we may define closed ball as follows:

Definition 9. Let (X, d) be a metric space and let a ∈ X. The set

B(a, r) = {x ∈ X|d(a, x) ≤ r} is called the closed ball with center a
and radius r.

Theorem: Let (X, d) be a metric space. Then,

(a) X and ∅ are closed.
(b) The intersection of arbitrary collection of closed sets is closed.
(c) The unition of finite collection of closed sets is closed.

Proof:
(a) Already we have proved that in any metric space X, d), X is open.
Therefore, by definition of closed set, the complement of X in (X, d)
must be closed. Thus, ∅ is closed in (X, d).
Also, we have shown that in any metric space X, d), ∅ is open. There-
fore, by definition of closed set, the complement of ∅ in (X, d) must be
closed. Thus, X is closed in (X, d).

(b) Let Fλ , λ ∈ Λ be a family of closed sets. We need to show that

6
T
TFλ
λ∈Λ S Clearly, X \ Fλ is open for each λ ∈ Λ. Note that
is closed.
X\ λ∈Λ Fλ = λ∈Λ (X \ Fλ ).

S
Now the set (X \ Fλ ) is open for each λ and therefore, λ∈Λ (X \ Fλ )
is open because of the property of open sets
T (that arbitrary union of
open sets is open). Thus, its complement λ∈Λ Fλ is closed.

(c) Let F1 , F2 , ..., Fn be closed sets. Therefore, each X \ Fi is open,

if 1 ≤ i ≤ n.

We have proved that finite intersection of open sets is open. Thus,

(X \ F1 ) ∩ (X \ F2 ) ∩ (X \ F3 ) ∩ ... ∩ (X \ Fn ) is open. Clearly its
complement is closed.

Now if x 6∈ (X \ F1 ) ∩ (X \ F2 ) ∩ (X \ F3 ) ∩ ... ∩ (X \ Fn ). Thus,

x 6∈ (X \ Fi ), for some i, 1 ≤ i ≤ n. Hence, x ∈ Fi , for some i,
1 ≤ i ≤ n. Thus, x ∈ F1 ∪ F2 ∪ F3 ∪ ... ∪ Fn .

Thus, the set F1 ∪ F2 ∪ ... ∪ Fn is the complement of (X \ F1 ) ∩ (X \

F2 ) ∩ ... ∩ (X \ Fn ). Therefore, we have shown that F1 ∪ F2 ∪ ... ∪ Fn is
closed.

We have shown in part (c) above that the union of finite collection
of closed sets is closed. It should be noted that union of an arbitrary
collection of closed sets may not be closed.

For example, consider the sets [ 13 , 23 ], [ 14 , 34 ], [ 15 , 45 ], ..., [ k1 , k−1

k
] (where
k ∈ N \ {1} in R), with Euclidean metric. Then each of these sets is
closed but, it is not difficult to show that, [ 13 , 23 ] ∪ [ 14 , 43 ] ∪ [ 15 , 45 ] ∪ .... =
(0, 1), which is open (and not closed) in R with Euclidean metric. Thus,
arbitrary union of closed sets need not be closed.

Check Your Progress

1. Let XS be an infinite metric space with discrete metric. Then
X = x∈X {x}. Each {x} is closed in this X and X too is closed
in X. Does this contradict result (c) above?
2. Can we say that (0, 1) is closed in (0, 1) with Euclidean metric?
Justify.
3. If F is a closed set and G is an open set in a metric space, then
show that F \ G is closed and G \ F is open in the same metric
space.
It is worth noting that the notion of closed set can be explained using
limit points. We have,

7
Theorem: Let (X, d) be any metric space. Suppose A ⊆ X. Then A
is closed in X if and only if each limit point of X lies in X.

Proof: Suppose A is closed. Let a be a limit point of A such that

a 6∈ A. Therefore, a ∈ X \ A, where X \ A is open, since A is closed.

Therefore, ∃ r > 0 such that B(x, r) ∈ X \ A. Thus, B(x, r) ∩ A = ∅.

This shows that a is not a limit point of A, which is a contradiction.
Thus, A contains each of its limit points.

Conversely, let A contain each of its limit points. Suppose A is not

closed. Therefore, X \ A is not open. Thus, ∃ x ∈ X \ A such that
∀ r > 0, the set B(x, r) * X \ A, that is, B(x, r) ∩ A 6= ∅. In fact
(B(x, r) \ x) ∩ A 6= ∅, since x 6∈ A. This shows that x is a limit point of
A but x 6∈ A. This is a contradiction. Therefore, A must be closed.

Example 15. The set of limit points of N ⊆ R with Euclidean metric

is empty and we know that ∅ ⊆ N. Thus, N is closed.

Example 16. In Euclidean metric space, every real number is a limit

point of Q, but R * Q. Therefore, Q is not closed in R with Euclidean
metric.

Example 17. After proving the theorem, “If a is a limit point of A

then every open ball with center at a contains infinitely many points of
A”, recall that we remarked that because of this theorem no finite set
can have a limit point. Therefore, every finite set in any metric space
is closed.

1.4 Closure of a set

Definition 10. The closure of a subset A in a metric space (X, d) is

the intersection of the family of all closed subsets of X that contain A.

The definition been stated, the questions that naturally strike us

are these: Is it possible that such a family may be empty? What is
the guaranty that this family does contain at least one element? Well,
obviously such a family cannot be empty, for X is a closed subset of X
and it does contain any such A.

Closure of A is denoted by A.

8
 Warning: Observe the definition carefully. When we say family of
closed subsets, these should be closed in X and not in A. Otherwise, A
is always closed in A and therefore, closure of any A would turn out to
be A itself.

Example 18. In R with Euclidean metric, (0, 1) = [0, 1] = (0, 1] =

[0, 1) = [0, 1].

Example 19. In R with Euclidean metric, Q = R \ Q = R.

Example 20. In R with discrete metric, (0, 1) = (0, 1), [0, 1] = [0, 1].
Also, (0, 1] = (0, 1] and [0, 1) = [0, 1).

One might have noticed a bit similarity between the concepts of

interior and closure of a set. We have a similar result also:

Theorem: Let (X, d) be a metric space. Suppose A ⊆ X. Then

(i) A ⊆ A
(ii) A is closed
(iii) A is closed if and only if A = A

Proof:(i) This should be obvious from the definition, since A is the

intersection of closed subsets of X, such that each of these contains A.

(ii) We have already showed that intersection of arbitrary family of

closed sets is closed. Therefore, A must be closed.

(iii) Already we have shown that A is closed. So, if A = A, then

A should be closed.

Coversely, suppose A is closed. We will show that A = A. Since,

we have already shown in (i) that A ⊆ A, it is enough to show that
A ⊆ A, if A is closed.

Let {Fλ }λ∈Λ be the family of closed subsets of X, such

T that each Fλ con-
tains A. Let A = Fλ0 , for some λ0 ∈ Λ. Then A = λ∈Λ Fλ ⊆ Fλ0 = A.
Thus, A ⊆ A and thus, theorem is proved.

It should be obvious that for any A in any (X, d), A = A, from (ii)
and (iii) above. Also, A (being intersection of closed sets of X, con-
taining A) is contained in every closed set of X, that contains A. Thus,
A must be the smallest closed set of X, which contains A. The reader
is suggested to compare these results with the similar ones in the topic
of interior of a set.

9
 Theorem: Suppose (X, d) is a metric space. Let A, B ⊆ X. Then
A ⊆ B ⇒ A ⊆ B.

Proof: Let A ⊆ B. But then we have B ⊆ B. Therefore, A ⊆ B.

Thus, A is contained in the closed set B.

But A is contained in A also, which is the smallest closed set, that

contains A. Therefore, A ⊆ B.

Although there are a few similarities in the concepts of interior and

closure of a set, there are, obviously, differences also. Recall that inte-
rior of intersection is the intersection of interiors, but for the union, we
don’t get this precise equality.

In case of closure, exactly opposite situation arises. That is, union

of closures equals closure of unions, but in case of intersection, we don’t
get such precise equality. We have,

Theorem: Let (X, d) be a metric space and let A, B ⊆ X. Then

A ∪ B = A ∪ B.

Proof: Note that A ⊆ A∪B and so A ⊆ A ∪ B. Similarly, B ⊆ A ∪ B.

As a result of which, A ∪ B ⊆ A ∪ B ..........(*)

To show the reverse part, observe that A ⊆ A and B ⊆ B. There-

fore, A ∪ B ⊆ A ∪ B. Both A and B are closed and therefore, their
union is closed.

But, A ∪ B is the smallest closed set, which contains A ∪ B. Therefore.

A ∪ B ⊆ A ∪ B. ..........(**)

From (*) and (**), it is clear that A ∪ B = A ∪ B.

Note what happens in case of intersection of two sets. We would expect

equality between A ∩ B and A ∩ B. However, in general, we can only
say that A ∩ B ⊆ A ∩ B.

To understand this, first note that A ∩ B ⊆ A and A ∩ B ⊆ B. There-

fore, by an earlier theorem, A ∩ B ⊆ A ∩ B. Now we need to show that
there are cases when A ∩ B * A ∩ B.

Let A and B be the sets of rationals and irrationals, respectively. Then

A = R and B = R. Thus, A ∩ B = R. However, A ∩ B = ∅ and so
A ∩ B = ∅. Thus, A ∩ B * A ∩ B.

10
Check Your Progress
1. Find the closure of (0, 1) in R with Euclidean metric.
2. Find the closure of [0, 1] in R with Euclidean metric.
3. Find the closure of {1/n|n ∈ N} in R with Euclidean metric.
The word ‘closure’ is self explanatory in the sense that if a point x
is in the closure of a set A, then roughly it means x is arbitrarily close
to the set A. To be precise, we have,

Theorem: Let A be a nonempty subset of X and x ∈ X. Then x

is in the closure of A (that is, x ∈ A), if and only if every non empty
open ball with center at x contains at least one point of A.

Proof: Suppose x ∈ A. Let, if possible, B(x, r) ∩ A = ∅ for some

r > 0. Then X \ B(x, r). Now, B(x, r) is open and so X \ B(x, r) is
closed. Therefore, A ⊆ X \ B(x, r).

Thus, x ∈ B(x, r) and x ∈ A ⊆ X \ B(x, r). Therefore, x ∈ B(x, r)

and x ∈ X \ B(x, r), which is impossible. Thus, B(x, r) ∩ A 6= ∅.

Conversely, suppose B(x, r) ∩ A 6= ∅ for all r > 0 and x ∈

/ A. Then
x ∈ X \ A.

Now X \ A is open. Therefore, ∃ r0 > 0 such that B(x, r0 ) ⊆ X \ A.

However, A ⊆ A. Hence, ∅ =
6 B(x, r0 ) ∩ A ⊆ B(x, r0 ) ∩ A, which is a
contradiction.

Example 21. Let X be a discrete metric space. Then any two distinct
points of X are at a distance of 1 unit from each other. Therefore, there
is no point of X, which is arbitrarily close to, say x, other than x itself.
Thus, the closure of {x} is {x} itself.
Example 22. From the above example, it should be clear that if A ⊆ X
under discrete metric, then A = A.
Definition 11. The points of A are called closure points of A.
Example 23. Note the difference between limit points and closure
points of a set. Treating N, Z as subsets of R under Euclidean metric,
we note that the sets of their limit points are empty, but N = N and
Z = Z. Also, the set {1/n|n ∈ N} has only one limit point, that is 0,
while it has infinitely many closure points, 1, 21 , 13 , ... and 0.

11
 The set of limit points of a set is a subset of the set of closure points
of the set and the strict inequality is possible, as can be seen from the
above example.

It’s the right time to observe how interior and closure of a set are
related to each other. We have a theorem in this regard,

Theorem: Let A ⊆ X. Then (i) X \ Ao = (X \ A) and (ii)

X \ A = (X \ A)o .

Proof: Left to the readers.

1.5 Boundary of a set

What do we understand from the word ‘Boundary’ ? If a person
is on the boundary of A and B, then we feel that he is neither in A
nor in B. Or sometimes we feel that he is equally close to A and B.
The mathematical concept Boundary has somewhat similar meaning
to these ideas. To be precise, we have,

Definition 12. Let A be a subset of X. Then the boundary of A is

defined to be the set A \ Ao . It is denoted by ∂A.

Example 24. Consider A = (0, 1) and B = [0, 1] as subsets of R with

Euclidean metric. Then Ao = A, A = B, B o = A and B = B. Also,
B \ A = {0, 1}. Therefore, ∂A = ∂B = {0, 1}.

Example 25. The interior and closure of R with Euclidean metric is

R itself. Therefore, ∂R = ∅.

Example 26. Let A be any subset of a discrete metric space X. Then

both Ao and A are A itself. Thus, ∂A = ∅.

Let us understand the properties of the boundary of a set.

Theorem: Let A be a subset of a metric space X. Then,

(i) ∂A = A ∩ X \ A = X \ (Ao ∪ (X \ A)o ).
(ii) A = Ao ∪ ∂A.
(iii) X = Ao ∪ (X \ A)o ∪ ∂A.

Proof: (i) By definition, ∂A = A \ Ao . If P and Q are any

two subsets of X, then by set theory, it can be shown easily that
P \ Q = P ∩ (X \ Q).

12
Therefore, A \ Ao = A ∩ (X \ Ao ).

By the theorem at the end of the last section, A∩(X\Ao ) = A∩(X \ A).

Thus, ∂A = A ∩ X \ A.

By the same theorem, X \ A = (X \ A)o . Thus, A = X \ (X \ A)o and

X \ Ao = (X \ A).

Therefore, A ∩ (X \ Ao ) = (X \ (X \ A)o ) ∩ (X \ Ao ). The R.H.S.

is equal to X \ {(X \ A)o ∪ Ao } = X \ {Ao ∪ (X \ A)o }. Thus, (i) is
established.

(ii) If Q ⊆ P then P = Q ∪ (P \ Q). We know that Ao ⊆ A ⊆ A.

That is, Ao ⊆ A.

Therefore, A = Ao ∪ (A \ Ao ) = Ao ∪ ∂A and (ii) is established.

(iii) We have proved in (i) that ∂A = X \ (Ao ∪ (X \ A)o ). After

restating this, we get X = Ao ∪ (X \ A)o ∪ ∂A.

Note what we have proved in the first part of (i) ∂A = A ∩ X \ A.

The boundary of a set A is made up of precisely those points in the
closure of A as well as in the closure of X \ A. Therefore, the points of
∂A are arbitrarily close to A as well as to the outside of A. What do
we mean by this statement? Consider any open ball with a point of the
boundary as its center. Then this ball must contain at least one point
of the set A, as well as at least one point of X \ A. From this point of
view, it is not difficult to understand that if A = {1/n|n ∈ N}, then its
boundary is A ∪ {0}.

Check Your Progress

1. Let A ⊆ R2 under Euclidean metric, be such that neither coordi-
nate of any point of A is zero. Find ∂A.

2. Under Euclidean metric, what is ∂A if A = {(x, y) ∈ R2 |n − 1 <

x2 + y 2 < n, n ∈ N}?

1.6 Dense set

From the basic set theory, we know that the cardinalities of N, Z
and Q are equal. Roughly speaking, this means that these sets have

13
same number of elements. In fact, the set of even numbers, set of odd
numbers and the set of primes all have equal cardinalities. However,
if we pick up a real number, say r, then can we guarantee that there
would exist integer/s which is/are arbitrarily close to r? The answer is
obviously no! Now let us ask the same question by replacing integers by
rational numbers. This time we get affirmative answer. The reason is
obvious. If r is rational, then there is nothing to show. If r is irrational,
then we know that the closure of the set of irrational numbers is R itself.
This suggests that even if the cardinalities of Z and Q are equal, the
spread of these sets on the real line is different. This calls for a new
term for the sets like Q.

Definition 13. If a subset A of a metric space X is such that A = X,

then A is said to be dense in X.

Again the reader is suggested to note that the denseness of a proper

subset A of X depends not only on X but also on the metric under
consideration. Through examples, this will be clear.

Example 27. A metric space X is, obviously, dense in X under any

metric.

Example 28. Under Euclidean metric, the set (0, 1) is dense in [0, 1],
but not in R.

Example 29. Under Euclidean metric, both the sets Q and R \ Q are
dense in R.

Example 30. A set is dense in its closure under any metric.

Example 31. Recall that under discrete metric, the closure of a set
A is A itself. Thus, neither Q nor R \ Q is dense in R, under discrete
metric.

Example 32. The set {(x, y)|x ∈ Q, y ∈ Q} is dense in R2 .

Example 33. A point in R2 is said to be a lattice point if both of its

coordinates are integers. The set of lattice points in R2 is not dense
under Euclidean metric.

Example 34. Under Euclidean metric, neither N nor Z is dense in R,

but both R \ N and R \ Z are dense in R.

Let us understand the definition of dense set. A subset A of X is

said to be dense in X if every point of X lies in the closure of A. Thus,
for every x ∈ X and for any r > 0, the open ball B(x, r) intersects A.
In fact, we have a theorem in this regard,

14
 Theorem: The necessary and sufficient condition for a subset A
of X to be dense in X is that every non empty open subset U of X
intersects A.

Proof: Let A be a dense set in X and let U be a non empty open

subset of X. Thus, for each x ∈ U , there exists r > 0 such that
B(x, r) ⊆ U . But by definition of the dense set, B(x, r) ∩ A 6= ∅.
Clearly, U ∩ A 6= ∅.

Conversely, suppose every open subset of X intersects A. Then as

B(x, r) is open for all x ∈ X and for all r > 0, it is clear that
B(x, r) ∩ A 6= ∅. Therefore, A is dense in X.

Check Your Progress

1. Give an example of A ⊆ X, such that neither A nor X \ A is
dense in X.

2. Let A ⊆ R2 be such that exactly one coordinate of the points of

A is integer. (That is, (1, 3.1), (3.1, 1) ∈ R2 but (1, 1), (3.1, 3.1) ∈
/
2 2
R ) Is A dense in R , under Euclidean metric?

3. Let A ⊆ B ⊆ X. If A is dense in X, then show that B is also

dense in X.

4. Let A be any finite subset of R. Show that R \ A is dense under

Euclidean metric.

5. Let A ⊆ B ⊆ X. If A is dense in B and B is dense in X, then is

A dense in X? Justify.

1.7 Equivalent metrics

By this time the reader must have realized that while talking about
the subsets A, B, again and again we mention the set X and also the
metric defined on it. It is to be noted that a particular set, say A, may
lose some of its properties (e.g. openness, closeness, denseness etc.) if
we change the set X, whose subset is A.

This seems quite natural, but at the same time, even if we maintain
X, and change the metric, then also there is a possibility that a given
subset may lose some of its properties. In fact, it is also possible that
the subset may gain some extra properties if we change the set X or
change the metric, defined on it.

15
However, there are cases where the change in the metric does not affect
the openness (similarly closeness) of any given subset. Since open sets
are important in Topology, such particular metrics have importance
and so they possess a separate terminology. Such metrics are known as
equivalent metrics.

There are various ways to define equivalent metrics. One way is to

use open sets in metric spaces, another way is to use open balls. There
is one more way of using convergence of sequences and so on! We will
start with open sets.

Definition 14. Two metrics d1 , d2 on the same set X are said to be

equivalent if each open set in (X, d1 ) is open in (X, d2 ) and vice versa.
It should be obvious, from this definition, that on R, the Euclidean
metric and discrete metric are not equivalent.

We may adopt the terminology of saying a d1 -ball and a d2 -ball for

some open balls in (X, d1 ) and (X, d2 ) respectively. With this, we may
have a theorem, which, at times, regarded as an equivalent definition
of equivalent metrics.

Theorem: Two metrics d1 , d2 on the same set X are equivalent if

for any x ∈ X, each d1 -ball centered at x, contains some d2 -ball cen-
tered at x and vice versa.

Proof: Left to the readers as an easy exercise.

There is a sufficient (but not necessary) condition to check whether

or not the metrics are equivalent. We present it as a theorem and
prove the same. The following theorem uses the preceding theorem to
prove its claim. Therefore, the readers are suggested to try to prove
the last theorem, before going ahead.

Theorem: Let X be a non empty set. Two metrics d1 and d2

on X are equivalent if there exist two positive numbers a, b such that:
ad1 (x, y) ≤ d2 (x, y) ≤ bd1 (x, y), ∀x, y ∈ X.

Proof: Assume that on a non empty set X, we have defined two

metrics d1 and d2 and that for two fixed positive numbers a, b, we are
getting ad1 (x, y) ≤ d2 (x, y) ≤ bd1 (x, y), ∀x, y ∈ X. Our aim is to show
that the two metrics are equivalent.

Since we have two different metrics, let us adopt the notation Bd1 (x, r)

16
to denote the open ball with center x and radius r in (X, d1 ). That is,
Bd1 (x, r) = {y ∈ X|d1 (x, y) < r}. Similarly, we use Bd2 (x, r).

Let  > 0 be any. We will show that there exists a δ > 0 such that
Bd1 (x, δ) ⊆ Bd2 (x, ) and conversely, for any η > 0, there exists a δ > 0
such that Bd2 (x, δ) ⊆ Bd1 (x, η).

Let δ = b . We will show that Bd1 (x, δ) ⊆ Bd2 (x, ).

Let y ∈ Bd1 (x, δ). Then d1 (x, y) < δ.

But, d2 (x, y) ≤ bd1 (x, y). Thus, d2 (x, y) ≤ bδ = . Hence y ∈ Bd2 (x, ).
This implies that Bd1 (x, δ) ⊆ Bd2 (x, ). Similarly, it can be shown that
Bd2 (x, δ) ⊆ Bd1 (x, η). Thus, the proof is complete.

Check Your Progress

1. Let (X, d) be any metric space. Define d∗ on X as follows:
d∗ (x, y) =min{1, d(x, y)}, ∀x, y ∈ X. Show that d∗ is a metric
on X.

2. In the above example, show that d and d∗ are equivalent metrics.

d(x,y)
3. Let (X, d) be a metric space. Define d∗ on X by d∗ (x, y) = 1+d(x,y)
.
Show that d∗ is a metric and is equivalent to d.

4. Taking the help of the above problem (3), show that the condition
given in the last theorem (about the equivalent metrics) is not
necessary.

17