4 Open sets and closed sets

Throughout this section, (X, %) is a metric space.

Definition 4.1. A set A ⊆ X is open if it contains an open ball about each

Lemma 4.2. An open ball in a metric space (X, %) is an open set.

Proof. If x ∈ Br (α) then %(x, α) = r − ε where ε > 0. If y ∈ Bε (x) then

%(α, y) ≤ %(α, x) + %(x, y) < r − ε + ε = r.

This implies that y ∈ Br (α) for all y ∈ Bε (x), that is, Bε (x) ⊆ Br (α).

Theorem 4.3. If (X, %) is a metric space then

1. the whole space X and the empty set ∅ are both open,

2. the union of any collection of open subsets of X is open,

3. the intersection of any finite collection of open subsets of X is open.

Proof. (1) The whole space is open because it contains all open balls, and
the empty set is open because it does not contain any points.
(2) Suppose
S {Ai : i ∈ I} is a collection of open sets, indexed by I, and
let A = i∈I Ai . Let x ∈ A be arbitrary. Then x belongs to at least one of
the sets Ai . Since this set is open, it contains an open ball about x; clearly,
this ball lies in A. But x ∈ A was chosen arbitrarily, and so A meets the
definition of an open set.
(3) Suppose A1 , A2 , . . . , An are open sets and let A = ni=1 Ai . Let x ∈ A
T
be arbitrary. Then x ∈ Ai for every i = 1, . . . , n. For each i, since Ai is
open, there exists ri > 0 such that Bri (x) ⊆ Ai . Let r = min{r1 , r2 , . . . , rn }.
Then Br (x) ⊆ Bri (x) ⊆ Ai for all i = 1, . . . , n, and hence Br (x) ⊆ A. But
x ∈ A was chosen arbitrarily, and hence A is an open set.

An infinite intersection of open sets is not necessarily open.

Example 4.4. Let An be the open T∞intervals (−1/n, 1/n) in R. Then An

10
Remark 4.5. In this course, we introduced the notion of metric, and then
used that notion to define open sets. But it is possible to define the open sets
directly to be any collection of subsets of X satisfying conclusions (1)–(3)
of Theorem 4.3. This leads to the more general notion of topological space.
Not every topological space is a metric space: there are collections of open
sets satisfying (1)–(3) that do not arise from a metric on X.

Lemma 4.6. A set is open if and only if it is equal to the union of a

Proof. According to Theorem 4.3(2) the union of any collection of open balls
is open. On the other hand, if A is open then for every
S point x ∈ A there
exists aSball B(x) about x lying in A. We have A = x∈A B(x). Indeed, the
union x∈A B(x) is a subset of A because every ball B(x) is a subset of A,
and the union contains every point x ∈ A because x ∈ B(x).

Definition 4.7. The interior of a set A is the union of all open sets con-
tained in A, that is, the maximal open set contained in A. The interior of
A is denoted by int(A).

The definion is legitimate because of Theorem 4.3(2).

Definition 4.8. If A ⊆ X then C(A) = X \ A denotes the complement of

Definition 4.9. A set is A ⊆ X is closed iff its complement C(X) is open.

Lemma 4.10. A closed ball in a metric space (X, %) is a closed set.

Proof. Consider the closed ball Br [α]. We need to show that C(Br [α]) is
open. Suppose x is any point in C(Br [α]). Since x is not in Br [α], it must
be the case that %(α, x) > r. Choose ε > 0 such that %(α, x) > r + ε.
We claim that Br [α] ∩ Bε (x) = ∅. Suppose to the contrary that Br [α]
and Bε (x) have a point z in common. Then, by the triangle inequality,
%(α, x) < %(α, z) + %(z, x) < r + ε, which contradicts the choice of ε. We
have shown that every point x ∈ C(A) is the centre of ball Bε (x) completely
contained in C(A), and hence C(A) is open, as required.

Definition 4.11. A point x ∈ X is called a limit point of a set A if every

Other terms for “limit point” are point of accumulation or cluster point.

11
Theorem 4.12. A set is closed if it contains all its limit points.
Proof. Suppose A is closed. Then, by definition, the complement C(A) =
X \ A is open. Take any point x ∈ C(A) lying outside A. Since C(A) is open,
there is an ε > 0 such that the ball Bε (x) ⊆ C(A) lies entirely within C(A).
Clearly, Bε (x) ∩ A = ∅, demonstrating that x is not a limit point of A. But
the choice of x ∈ C(A) was arbitrary, and so A must contain all its limit
points.
Suppose A contains all its limit points. Let x ∈ C(A) be arbitrary. Since
x is not a limit point, there is an ε > 0 such that the ball Bε (x) ∩ A = ∅.
Thus C(A) is open and A is closed.
Theorem 4.13. In a metric space (X, %)
1. the whole space X and the empty set ∅ are both closed,
2. the intersection of any collection of closed sets is closed,
3. the union of any finite collection of closed sets is closed.
Proof. The theorem follows from Theorem 4.3 and the definition of closed
set.
(1) C(X) = ∅ and C(∅) = X.
T Suppose {Ai : i ∈ I} is a collection of sets, indexed by I, and let
(2)
A = i∈I Ai . Then
\  [
C(A) = C Ai = C(Ai ).
i∈I i∈I
The r.h.s. is a union of a collection of open sets and hence open. Thus, by
definition, A is closed.
(3) Suppose A = kn=1 Ai is a finite union of closed sets. then
S
 [k  \n
C(A) = C Ai = C(Ai ).
i=1 i=1
The r.h.s. is a finite intersection of open sets and hence open. Thus, by
definition, A is closed.
Definition 4.14. The closure of a set A is the intersection of all closed sets
containing A, that is, the minimal closed set containing A. The closure is
denoted by cl(A) or A.
The definition is legitimate because of Theorem 4.13(2). Clearly, int(A) ⊆
A ⊆ A.
Theorem 4.15. A = A ∪ A0 .
The proof is omitted.

12