Connected sets
Definition 1.
For a metric space , a separation of is a partition of into two proper subsets and such that and are both open in .
Note that if and are both open in , it follows that and are both closed in . A set which is both open and closed in called a clopen set. So, and are clopen in .
Definition 2.
If has a separation, is called disconnected. If is not disconnected, is connected.
For example, is disconnected because .
Characterization of connected sets in R
Theorem 3.
A nonempty subset is connected (( and ) ), i.e, a connected set in is either a singleton, an interval, a ray, or .
Proof of
We will prove the contrapositive. Let with , . Let and . . ❏Proof of
Suppose , with and nonempty, disjoint and open in . Let , , and WLOG . We know . Let . Clearly, , since then every neighborhood of must contain a point of , making it impossible to find a neighborhood of in (Alternatively, being a closed set forces ). also leads to a contradiction, since B must contain a neighborhood of , which in turn would contain an element smaller than . ❏
Images of connected sets under continuous functions are connected
Theorem 4.
If is a connected metric space and is continuous, is connected.
Proof
Let , , open in . This implies . ❏
Intermediate value theorem
Rudin, 4.23
Theorem 5.
Let be continuous, WLOG . Then, such that .
Proof
From the previous theorem, we know is connected. From the characterization of connected sets in , we know that . ❏
Important
If satisfies the conclusion of the intermediate value theorem ( such that , such that ), need not be continuous. For example, consider defined by
Separated sets
Definition 6.
Subsets and of are called separated if and , where the closures are taken in .
Note that need not be empty.
Theorem 7.
and are separated in is a separation of .
Proof of
Closure of in = (Closure of in ) (), which is equal to , since (Closure of in ) = . Thus, is closed in . Similarly, is closed in . ❏Proof of
We have to show that no limit point of in is contained in (and vice versa). FTSOC, let be a limit point of , and . This implies , which implies , since is closed in . But, and are disjoint.
Similarly, no limit point of in is contained in . ❏
Theorem 8(Corollary).
and are separated in and are separated in .
Rudin’s definition of connectedness
Definition 9.
is disconnected if is a union of two nonempty separated sets.
If is not disconnected, is connected.
From the previous theorem (take to be ), this is equivalent to our definition in the beginning.
Some theorems
Theorem 10(Theorem 1).
If and form a separation of then any connected subset of is contained in or in .
Proof
Let be connected. Let be a separation of . FTSOC, assume and . Since and are open in , and are open in . We can write , i.e, is the disjoint union of two nonempty sets open in . But this means is disconnected! ❏
Theorem 11(Theorem 2).
If a family of connected subsets of have a point in common, then their union is connected.
Proof
Let be a family of connected subsets of , and let . FTSOC, assume is disconnected, i.e, there exist non empty, disjoint and such that and and are open in . WLOG, . Consider . There must exist such that . Since and are open in , it follows that and are open in . We can write , i.e, is the the disjoint union of two nonempty sets open in . But this means is disconnected! ❏
Theorem 12(Theorem 3).
Let . is connected is connected.
Proof
FTSOC, suppose , where and are disjoint nonempty open subsets of . Then . Note that and are open in as and are open in . Since is connected, one of or must be empty. WLOG, suppose , which implies .
This means contains only points from , and since is nonempty, it must contain at least one point from . However, this implies does not contain all its limit points (as points in are limit points of ), contradicting the assumption that is closed. ❏
Connected components
Define if points and belong to a connected subset of . This is an equivalence relation on :
- is reflexive, since every singleton set is connected.
- is symmetric.
- is transitive, because of theorem 2.
Theorem 13(Theorem 4).
The equivalence classes of are connected.
Proof
Pick any and consider its equivalence class . for all , i.e, for some connected for all . Note that for all . Thus, . Also note that . So, from theorem 2, is connected. ❏
The equivalence classes of are hence called connected components of .
Theorem 14(Theorem 5).
Any connected subset of is contained in one of its connected components.
Proof
Let be connected, with and and , where and are connected components of . Then, , implying is not connected. ❏
Theorem 15(Theorem 6).
Connected components of are closed in .
Proof
Let be a connected component of . From theorem 3, must be connected. From theorem 5, must then be contained in a connected component of . This forces . ❏
The components are also open in if their number is finite, since finite unions of closed sets are closed.
The connected components of are singletons, since for any interval one can find such that . Such a space is called totally disconnected.