1.6 Connected Sets
Definition 1.6.1. Let \(A\) and \(B\) be subsets of \(\mathbb {R}\). Then \(A\) and \(B\) are said to be separated if neither has
a point in common with the closure of the other. i.e if \(A\cap \overline {B} = \emptyset \) or \(\overline {A} \cap B = \emptyset \).
Note. If \(A\) and \(B\) are separated that they are disjoint since \(\overline {A} \cap B \subset A \cap \overline {B} = \emptyset \). However, the converse is not necessarily
true.
Definition 1.6.2. Let \(A\) and \(B\) be subsets of \(\mathbb {R}\). Then the distance between \(A\) and \(B\) is defined by
\(d(A,B) = \inf \big \{\big | a - b\big |:\, a\in A\) and \(b\in B\big \}\).
Theorem 1.6.3. In order for two sets \(A\) and \(B\) to be separated, it is sufficient but not necessary,
that \(d(A,B)>0\).
Proof. Suppose \(d(A,B) = K > 0\), then the equation \(K = \inf \big \{\big |a - b\big |: \, a\in A\, , \, b\in B\big \}\) implies that \(\big | a - b\big | \geq K\) for every pair of points \(a\in A\) and \(b\in B\).
If \(a\in A\), it is not an adherent point of \(B\), since the \(bd\big (a - 1/2 K\, , \, a + 1/2 K\big )\) of \(a\) contains not point of \(B\). Therefore \(A \cap \overline {B} = \emptyset \). Similarly \(\overline {A} \cap B = \emptyset \).
Hence \(A\) and \(B\) are separated.
The condition is not necessary, since two sets \(A\) and \(B\) may be at zero distance apart, and yet be
separated. For example, the sets \(x< 0\) and \(x> 0\) are separated but are at zero distance. But on the other
hand, the sets \(x< 0\) and \(x\geq 0\) are disjoint but not separated.
□
Theorem 1.6.4. Let \(A\) and \(B\) be subsets of \(\mathbb {R}\) and \(A_1\subset A\, ,\, B_1\subset B\) where \(A\) and \(B\) are separated sets. Then \(A_1\) and \(B_1\) are
separated.
Proof. Since \(A\) and \(B\) are separated \(A \cap \overline {B} = \emptyset \). Thus \(d\big (A,\overline {B}\big ) > 0\). Taking \(d(A,B) = K > 0\), we have \(\big | a- b\big | = K > 0\) where \(a\in A\) and \(b\in \overline {B}\). Thus, for any point \(b\in \overline {B}\),
it is not an adherent point of \(A\) and \(A_1\), since any \(nbd \big (b - 1/2 K \, , \, b+1/2 K\big )\) of \(b\) does not contain a point of \(A\). Consequently, \(d(A,\overline {B})= K > 0\).
Therefore, \(A_1 \cap \overline {B}_1 = \emptyset \), since \(B_1\subset B \, \implies \, \overline {B}_1 \subset \overline {B}\).
Similarly, \(\overline {A}_1 \cap B_1 = \emptyset \). Hence, \(A_1\) and \(B_1\) are separated.
□
Theorem 1.6.5. Two closed subsets of \(\mathbb {R}\) are separated if and only if they are disjoint.
Proof. Suppose \(F_1\) and \(F_2\) are closed subsets of \(\mathbb {R}\) which are separated. Then \(F_1\cap \overline {F}_2 = \emptyset = \overline {F}_1\cap F_2\) and \(F_1 = \overline {F}_1\hspace {0.2cm},\hspace {0.2cm} F_2 = \overline {F}_2\). This means that \(F_1\cap F_2 = \emptyset . \implies F_1\)
and \(F_2\) are disjoints.
Conversely, suppose that \(F_1\) and \(F_2\) are closed and disjoint sets. Then \(F_1 = \overline {F}_1\hspace {0.2cm},\hspace {0.2cm} F_2 = \overline {F}_2\) and \(F_1\cap F_2 =\emptyset = F_1\cap \overline {F}_2 = \emptyset \). Similarly \(\overline {F}_1 \cap F_2 = \emptyset \). Hence \(F_1\) and
\(F_2\) are separated sets.
□
Theorem 1.6.6. If the open set \(O\) is the union of two separated sets \(A\) and \(B\), then \(A\) and \(B\) are open.
Proof. Suppose that \(A\) and \(B\) are not empty. For if \(A = \emptyset \) then \(B = O\), and \(A\) and \(B\) are both open. Suppose \(x\) is
any point of \(A\). Then there exist a \(nbd \big (x - r\, , \, x + r\big )\) of \(x\) consisting entirely of points of \(O\). Since \(x\) is not a limit point
of \(B\), there exist a \(nbd \big ( x-r_1\, , \, x + r_1\big )\) of \(x\) , where \(r_1 \leq r\), which contains no point of \(B\) and so consists entirely of points of \(A\).
Every point of \(A\) is thus an interior point of \(A\). Hence \(A\) is open similarly for \(B\).
□
Theorem 1.6.7. If the closed set \(F\) is the union of two separated sets \(A\) and \(B\), then \(A\) and \(B\) are
closed.
Proof. Since \(F= A\cap B\) is closed, we have \(A\cup B = \overline {A\cup B} = \overline {A} \cup \overline {B}\). Hence,
\[\overline {A} = \overline {A} \cap \big (\overline {A}\cup \overline {B}\big ) = \overline {A} \cap \big ( A\cup B) = \big (\overline {A} \cap A\big ) \cup \big ( \overline {A} \cup B\big ) = A\cup \emptyset = A\]
Thus, \(A\) is closed. Similarly \(B\) is closed.
□
Definition 1.6.8. A subset \(A\) of \(\mathbb {R}\) is said to be disconnected if it is a union of two non-empty
separated sets.
It should be noted that a subset is no special virtue in the word ‘two’ for if \(A = B\cup C\cup D\), where \(B, \, C\,\) and \(D\) are
not empty and each pair is separated, then \(A = B\cup \big ( C\cup D\big )\), where \(B\) and \(C\cup D\) are separated, is disconnected.
Another way of expressing that \(A(\subset \mathbb {R})\) is that \(A\) is disconnected if and only if \(A\subseteq O_1 \cup O_2\), where \(O_1\) and \(O_2\) are open
sets such that \(A\cap O_1\) and \(A\cap O_2\) are not empty, but \(A\cap O_1 \cap O_2\) is empty.
Similarly, another expression is that \(A\) is disconnected if and only if \(A\subseteq F_1\cup F_2\), where \(F_1\) and \(F_2\) are closed sets
such that \(A\cap F_1\) and \(A\cap F_2\) are not empty by \(A\cap F_1 \cap F_2\) is empty.
Definition 1.6.9. A subset \(A\) of \(\mathbb {R}\) is said to be connected if it is not disconnected i.e \(A\) is not a
union of two non-empty separated sets.
Theorem 1.6.10. A subset \(A\) of \(\mathbb {R}\) is connected if and only if for \(x,y \in A\) and \(x < z < y\), then \(z\in A\).
Proof. Suppose \(\, \exists \, x, \, y \in A\) and some \(z \in (x,y)\ni z\not \in A\). Then \(A = P \cup Q'\) where \( P = A \cap \big (-\infty \, ,\, z\big )\) and \(Q = A\cap \big (z\, , \, \infty \big )\). Since \(x\in P\) and \(y \in Q\, ,\, P\) and \(Q\) are non-empty. Since \(P\subset \big (-\infty \, ,\, z\big )\) and \(Q\subset \big (z\, , \,\infty \big )\),
they are separated. Hence \(A\) is not connected, being a union of two non-empty separated sets.
Hence \(z \in A\).
Conversely, suppose that \(A\) is not connected. Then there are non-empty separated sets \(P\) and \(Q\) such
that \(P\cup Q = A\). Take \(x\in P \) and \(y \in Q\) and assume (without loss of generality) that \(x<y\). Define \(z = \sup \big (P\cap \big [x,y\big ]\big )\). Then \(z\in P\) and hence \(z\not \in Q\).
In particular, \(x\leq z\leq y\). If \(z\not \in P\), it follows that \(x < z < y\) and \(z\in A\). If \(z \in P\), then \(z\not \in Q\), hence there exist \(z_1\ni z < z_1 < y\) and \(z_1 \not \in Q\). Then \(x < z_1 < y\) and \(z_1\not \in A\). This
contradicts that \(A\) is not connected. Therefore \(A\) is connected.
□
Proof. Suppose to the contrary that \(A\) is an interval. Then \(\, \exists \, a, \, b\in A\) with \(a< b\) and \(c\in \big (a,b\big ) \ni c\not \in A\). Let \(G_c = A \cap \big (-\infty \, , \, c\big )\) and \(H_c = A \cap \big ( c\, ,\, \infty \big )\). Note that both
\(H_c\) and \(G_c\) are non-empty. Also since \(c\not \in A\hspace {0.2cm},\hspace {0.2cm} A = G_c\cup H_c\). Since \(\big (-\infty \, , \, c\big ) \cap \big [c\, ,\, \infty \big ) = \emptyset \) and \(\big (-\infty \, ,\, c\big ] \cap \big [c\, , \infty \big ) = \emptyset \) then \(G_c\) and \(H_c\) are separated. Thus, \(A\) is not connected.
Therefore, if \(A\) is connected, then \(A\) is an interval.
Conversely, \(A\) is not connected and write \(A = G\cup H\) where \(G\) and \(H\) are non-empty separated sets. Without
loss of generality, let \(a\in G\, ,\, b\in H\) and \(a< b\). Note that \(G\cap [a,b]\) is bounded so that \(c = \sup G \cap [a,b]\) exists in \(\mathbb {R}\) Then \(c\in \overline {G \cap [a,b]}\subset \overline {G}\cap \overline {[a,b]}\). In other words \(c\in \overline {G}\).
Since \(\overline {G} \cap H = \emptyset , \, c\not \in H\). If \(c\not \in G\) then \(c\not \in A\), and so \(A\) is not connected by the previous theorem. Hence, \(A\) is not an interval,
by the previous theorem. Therefore, if \(A\) is an interval then \(A\) is connected.
□
Proof. Let \(A\) and \(B\) be two connected sets which are not separated. If \(A\cup B\) is not connected, \(A\cup B = C \cup D\) where \(C\)
and \(D\) are non-empty separated sets. Then \(A = A\cap (A \cup B) = A\cap (C\cup D) = (A\cap C) \cup (A\cap D)\) which expresses \(A\) as the union of two separated sets,
which is impossible since \(A\) is connected, unless one of \(A\cap C\) and \(A\cap D\) is empty. Suppose that \(S\cap D\, B\subseteq C\) or \(B\subseteq D\). But this
gives \(A \cap \overline {B} \subseteq C \cap \overline {D} = \emptyset \hspace {0.2cm},\hspace {0.2cm} \overline {A}\cap B \subseteq \overline {C} \cap D = \emptyset \). which is not possible. Hence \(A\cup B\) is connected.
□
Proof. Let \(A\) and \(B\) be two connected sets, whose intersection \(A\cap B\) is not empty since \(A\cap B \subseteq A\cap \overline {B}\) and \(A \cap B \subseteq \overline {A} \cap B\), the result
is an immediate consequence of the theorem.
□
Questions on this section
Stuck on something here? Ask below and it stays attached to this topic.