Contents

3 The Quotient Topology and Quotient Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3.1 Quotient Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3.1.1 Two Equivalent Definitions of a Quotient Map. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3.1.2 Open and Closed Maps are Quotient Maps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3.2 The Quotient Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.3 Quotient Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.3.1 The Quotient Space as a Space of Equivalence Classes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.3.2 The Quotient Space as a Partition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.3.3 Quotient Spaces Induced by Maps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.4 Conditions for the Restriction of Quotient Map to Be a Quotient Map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.5 On Other Possible Pathologies of Quotient Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.5.1 Composites of Quotient Maps Are Quotient Maps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.5.2 Products of Quotient Maps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.5.3 Quotient Maps and Hausdorff Spaces. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.6 Universal Property of Quotient Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

4 Adjunction, or Attaching Space. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
5.1 A Map With a Continuous Right-Inverse is a Quotient Map; Retractions are Quotient Maps. . . . . . . . . . . . . 13
5.2 A Quotient Map That is Neither Open Nor Closed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
5.3 Practice With Quotients. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
5.4 A Quotient of a Compact Space is Compact. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
5.5 (Important) Constructing the Circle S1 As a Quotient Space of the Unit Interval I. . . . . . . . . . . . . . . . . . . . . 14
5.6 More Practice with Quotients. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
5.7 The Subspace Topology and Quotient Topology Are Categorical Duals in Top. . . . . . . . . . . . . . . . . . . . . . . . . 14

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

1
Chapter 3
The Quotient Topology and Quotient Spaces

3.1 Quotient Maps

3.1.1 Two Equivalent Definitions of a Quotient Map.

Definition (Quotient Map). Let X and Y be topological spaces and let p : X → Y be a surjective map. The map p is said
to be a quotient map provided a subset U of Y is open in Y iff p−1 [U ] is open in X. Equivalently, p is a quotient map
provided a subset F of Y is closed in Y iff p−1 [F ] is closed in X. The equivalence follows from p−1 [Y \ B] = X \ p−1 [B].

Remark. Thus, with quotient map, p : X → Y , we get more than just continuity. Indeed we have a sort of “strong
continuity” because we require a kind of correspondence between inverse images of sets that are open/closed in X and the
open/closed sets of Y .

Definition (Saturated Sets). Let X and Y be topological spaces and let p : X → Y be a surjective map. We say that
a subset C of X is saturated (with respect to the surjective map p : X → Y ) if C contains every set p−1 [{y}] that it
intersects—that is, C = p−1 [p[C]].

An Equivalent Definition of Quotient Maps in Terms of Open Saturated Sets.

Proposition 18. Let X and Y be topological spaces and let p : X → Y be a surjective map. Then p is a quotient map iff
p is continuous and maps saturated open sets to open sets of Y iff p is continuous and p maps saturated closed sets of X
to closed sets of Y .

Proof. Exercise. 

3.1.2 Open and Closed Maps are Quotient Maps.

Definition (Open and Closed Maps). Let X and Y be topological spaces and let f : X → Y be a function. We say
that f is an open map if for every open sets U of X, f [U ] is open in Y . We say that f is a closed map if for every
closed set F of X, f [F ] is closed in Y .

Remark. Let X and Y be topological spaces. By the preceding proposition, if p : X → Y is surjective and open or closed,
then p is a quotient map.

Example 2. Let X = [0, 1] ∪ [2, 3] with the subspace topology and let Y = [0, 2] with the subspace topology. The map
p : X → Y defined by p(x) = x1[0,1] (x) + (x − 1)1[2,3] (x) is readily seen to be surjective, continuous and closed. Therefore
it is a quotient map. It is not, however, an open map. The image of the open set [0, 1] of X is not open in Y .

3
4 3 The Quotient Topology and Quotient Spaces

3.2 The Quotient Topology

Construction of the Quotient Topology.

Definition. If X is a topological space and A is a set and if p : X → A is a surjective map, then there exists exactly one
topology T on A wrt which p is a quotient map; it is called the quotient topology induced by p.

Let the notation be as above. Thetopology T is of course defined

by letting it consist of those subsets U of A such that
p−1 [U ] is open in X. That is T = U ⊆ A : p−1 [U ] is open in X . It is easy to check that T is a topology: The sets Ø and
A are open because p−1 [Ø] = Ø, p−1 [A] = X and the other two conditions follow from the basic properties of preimages
of sets under functions, namely:
" #
[ [
−1
p Uα = p−1 [Uα ]
α∈J α∈J
" n
# n
\ \
−1
p Ui = p−1 [Ui ].
i=1 i=1

Example 3. Let A be the three-point set A = {a, b, c} and let p : R → A by


a if x > 0,

p(x) = b if x < 0,

c if x = 0.


You can check that the quotient topology on A induced by p is the one indicated by the picture below:

a b c

3.3 Quotient Spaces

3.3.1 The Quotient Space as a Space of Equivalence Classes.

Definition (Quotient Space of X). Let X be a topological space, let ∼ be an equivalence relation1 on X and X/ ∼ be
the space of equivalence classes (i.e., X/ ∼ is a partition of X into disjoint subsets whose union is X). Let p : X → X/ ∼
be the surjective map that carries each point of X to the element of X/ ∼ containing it. In the quotient topology induced
by p, the space X/ ∼ is called a quotient space of X.
1
A transitive, symmetric and reflexive relation.
3.3 Quotient Spaces 5

3.3.2 The Quotient Space as a Partition.

Given a partition of a set, there exists an equivalence relation on X whose equivalence classes correspond to the sets in
our partition and conversely. Thus, if X ∗ is a partition of a topological space X, one can think of X ∗ as having been
obtained by “identifying” each pair of equivalent points. Hence, we may reformulate the definition of a quotient space of
a topological space X in terms of partitions of X.
It is because of the equivalence class formulation that the quotient space X ∗ is often called the identification space,
or a decomposition space of the space X.

The Typical Open Set of a Quotient Space.

Let X be a topological space and X ∗ a partition of X and p : X → X ∗ the canonical quotient map taking each element
of x to the class containing it in X ∗ . We can describe the quotient topology of X ∗ in another way. A subset U ⊆ X ∗ is a
collection of equivalence classes, and the set p−1 [U ] is just the union of the equivalence classes belonging to U . Thus, the
typical open set of X ∗ is a collection of equivalence classes whose union is an open set of X. Thus, a set U ⊆ X ∗ is open
if and only if the union of the equivalence classes its elements belong to is open in X. This grants us a wide degree of
latitude in constructing open sets and also permits us to “force” certain sets to be open with respect to a specific topology
by passing to the quotient topology via the projection p.

Subspaces Do Not Behave Well Under Quotient Maps.

If p : X → Y is a quotient map and A is a subspace of X, then the map q : A → p(A) obtained by restricting p need not
be a quotient map. The following example shows this as well as the fact that projection maps need not be closed maps.
Example 4. Let π1 : R × R → R be the projection onto the first coordinate; then π1 is continuous and surjective and
open—this last contention is proved generally in the last result of the Product Topology subsection—and hence π1 is
a quotient map. However, π1 is not a closed map. Indeed, the subset C = {(x, y) : xy = 1} is closed—this is most easily
seen by observing that the map f : R2 → R defined by (x, y) 7→ xy is continuous and hence, f −1 [{1}] is closed as {1} is
closed in R. However, π1 [C] = R \ {0} = {0}c which is not closed in R and is in fact open in R!
Let A be the subspace A = C ∪ {0, 0} of R × R. Then the map q : A → R obtained by restricting π1 to A. This is
continuous (since π1 is continuous on R × R) and clearly surjective, but it is not a quotient map. Indeed, the one point
set {0, 0} is open in A and is saturated with respect to q, but its image is not open in R.

3.3.3 Quotient Spaces Induced by Maps.

One important class of quotient are those induced by maps and, in particular, continuous surjective maps. Considering
the case of simply a function, suppose X and Y are topological spaces and let f : X → Y be a map. Then we obtain a
partition of X comprised of preimages. Namely, X ∗ = f −1 [{y}] : y ∈ Y . In other words, we define ∼ on X by x ∼ x0 iff


∃y ∈ Y such that f −1 [{x}] = f −1 [{x0 }] so that X ∗ = X/ ∼.

Remark. The preceding discussions suggests that if p : X → Y is a quotient map, then Y is homeomorphic to the quotient
space of X identified by the partition X ∗ = {p−1 [{y}] : y ∈ Y }. This ends up being the case, and we shall prove this in
the final part of this section on quotient spaces.

A Worked Example.

Example 5 (2-Sphere). Let X = {(x, y) : x2 + y 2 ≤ 1} ⊆ R2 with subspace topology. Define ∼ on X by declaring

(x, y) ∼ (x0 , y 0 ) iff x2 + y 2 = x02 + y 02 = 1 or x = x0 and y = y 0 . This produces the ∗
 partition X2 = X/ ∼ of X consisting
of all the one-point sets {(x, y)} for which x + y < 1 along with the set S = (x, y) ∈ X : x + y 2 = 1 . Obviously S1
2 2 1

is disjoint from the one-point sets and the union of all one-point sets with S1 is X. The typical saturated open sets in X
(with respect to the canonical projection to the quotient) are (1) open subsets of S1 or (2) sets that are a nbhd of S1 in
X. Let p : X → X ∗ be the quotient map.
6 3 The Quotient Topology and Quotient Spaces

To see this, note that if U ⊆ X ∗ is open, p−1 [U ] is open in X and if S1 is not a subset of p−1 [U ], then p−1 [U ] = U .
Otherwise, U contains the equivalence class of S1 and—using brackets [ ] to denote equivalence classes of elements in
X ∗ —we may write U = [(0, 1)] ∪· U \ {[(0, 1)]} so that p−1 [U ] = S1 ∪· p−1 [U \ {[(0, 1)]}]. Since U \ {[(0, 1)]} ⊆ {(x, y) ∈
X : x2 + y 2 < 1}, p−1 [U \ {[(0, 1)]}] = U \ {[(0, 1)]}, so that p−1 [U ] = S1 ∪· (U \ S1 ). This is open if it is equal to V ∩ X
for some open set V of R2 and for such a set V , S1 ⊆ V , so clearly there exists an open ball about each point of S1 that
is contained in V and hence whose intersection would be contained in U . This is clear.
∗ 3
 It turns 2out 2that 2X is homeomorphic with the subspace of R called the unit 2-sphere defined by S2 =
1 ∗
(x, y, z) : x + y + z = 1 . Indeed, heuristically,
 we should expect
this! We can think of the set S ∈ X as a dis-
tinguished “pole” of S2 by pulling S1 ⊆ X = (x, y) : x2 + y 2 ≤ 1 ⊆ R2 “together” out of the paper.
Let us now proceed to define the desired quotient map q : X → S2 explicitly. We will show that p is surjective and is
a quotient map—we shall do this latter thing by verifying that our constructed q maps saturated open sets of X to open
sets of S2 . For whatever q we define, we shall demand q(x, y) = (0, 0, 1) for all x2 + y 2 = 1 (the choice of this point (0, 0, 1)
was arbitrary but convenient). Switching to spherical coordinates, let’s say (r, θ) 7→ ((1 − r) cos θ, (1 − r) sin θ, 2r − 1).
Switching back to rectangular coordinates, for x2 + y 2 > 0, we let
p p !
(1 − x2 + y 2 ) (1 − x2 + y 2 ) p
(x, y) 7→ p x, p y, 2 x2 + y 2 − 1 .
x2 + y 2 x2 + y 2
2
This is continuous for all (x, y) ∈ X with xp + y 2 > 0 as each component is a composite of continuous functions. For
√
x = y = 0, we begin by noticing
√ that |x| ≤ x2 + y 2 so that x((x2 + y 2 )−1/2 + 1) ≤ x(|x|−1/2 + 1) = sgn(x) x + 1 and
hence, as x, y → 0, sgn(x) x + 1 → 0 and thus by the squeeze theorem, so too does the original function. Analogous
reasoning holds for the second component. Thus, we are at liberty to define a continuous extension (0, 0) 7→ (0, 0, −1).
This constructed function q clearly surjects and is continuous. Moreover, for each x ∈ S2 , q is constant on the fiber
0
q −1 [{x}] and if |x| < 1, q −1 [x] = (x, y) where x = (x, y, z) whereas if |x| = 1, q −1 [{x}] = S1 . Therefore the partition X ∗
−1
induced by q comprised of the equivalence classes of preimages of singletons q [{x}] is the same partition as that induced
0 0
by p : X → X ∗ . That is, X ∗ = X ∗ and therefore X ∗ and X ∗ have the same quotient topologies as identical quotient
spaces of X. Finally, this means that the saturated open sets of q : X → S2 are the same as those of p : X → X ∗ and thus
q obviously maps saturated open sets to open sets in S2 . Therefore q is a quotient map. Thus with the quotient topology
on X ∗ , we claim that X ∗ is homeomorphic with S2 . This follows from a corollary we shall prove whereby a continuous
surjective g : X → Z which is a quotient map iff the quotient space induced by G on X is homeomorphic to Z.

A Classical Example.

Example 6 (Torus). Let X be the rectangle X = [0, 1] × [0, 1] with the subspace topology inherited from R2 . Define a
partition X ∗ of X as follows: we identify points (x, 0) with (x, 1) where 0 < x < 1 we identify points (0, y) with (1, y)
where 0 < y < 1 and we glue together (i.e., identify) the corners (0, 0), (0, 1), (1, 0) and (1, 1). The resulting space X ∗
has equivalence classes comprised of the two point sets {(x, 0), (x, 1)} where 0 < x < 1, {(0, y), (1, y)} where 0 < y < 1,
the four-point set {(0, 0), (0, 1), (1, 0), (1, 1)} and finally every point not lying in one of these equivalence classes has an
equivalence class unto itself.
Let p : X → X ∗ be the canonical quotient map. Noting that balls form a countable base in X since R2 is second
countable, the typical saturated open sets in X (with respect to p) are comprised of the four following types illustrated in
Figure 3.1.

Fig. 3.1 Typical Saturated Open Sets

3.4 Conditions for the Restriction of Quotient Map to Be a Quotient Map 7

Each—almost by definition—is an open set of X that equals a union of elements of X ∗ . The image of each of these sets
under p is an open set of X ∗ , as indicated in Figure 3.2 below.

Fig. 3.2 Image Under p

This description of X ∗ is just the mathematical way of saying what we expressed in pictures when we pasted the edges
of a rectangle together to form a torus. We leave the details of these manipulations to the reader.

3.4 Conditions for the Restriction of Quotient Map to Be a Quotient Map

We have seen that subspaces do not behave well under quotient maps: the restriction of a quotient map p : X → Y to a
subspace A ⊆ X, q : A → p[A], need not be a quotient map itself. However, we can salvage this somewhat.
Theorem 3.26. Let p : X → Y be a quotient map, A be a subspace of X that is saturated with respect to p and q : A → p[A]
be the map obtained by restricting p.
(1) If A is either open or closed in X, then q is a quotient map.
(2) If p is either an open map or a closed map, then q is a quotient map.
Proof. We shall prove both (1) and (2) together.
Step 1. We verify first the following two equations:

q −1 [V ] = p−1 [V ] if V ⊆ p[A];
p[U ∩ A] = p[U ] ∩ p[A] if U ⊆ X.

Since A is saturated wrt p, the first is trivial. For the second, note first that for any two subsets U and A of X, p[U ∩ A] ⊆
p[U ] ∩ p[A] by set-theoretic principles. To prove the reverse inclusion, suppose y = p(u) = p(a) for u ∈ U and a ∈ A. Since
A is saturated, A contains the set p−1 [p[{a}]], so that in particular, A contains u. Then y = p(u) where u ∈ U ∩ A.
Step 2. Now suppose A is open or p is open. We wish to show that q is a quotient map. To do so, we show that if
V ⊆ p[A] is such that q −1 [V ] is open in A, then V is open in p[A]. Suppose we are given such a V ⊆ p[A].
Suppose first that A is open. Since q −1 [V ] is open in A and A is open in X. Since q −1 [V ] = p−1 [V ], p−1 [V ] is open in
X and hence, V is open in Y because p : X → Y is a quotient map. In particular, V is open in p[A].
Now suppose p is open. Since q −1 [V ]= p−1 [V −1
 ] and q [V ] is open
 −1in A,
−1
 we have p [V ] = U ∩ A for some set U ⊆ X
−1
open in X. Now, since p is surjective, p p [V ] = V ; then V = p p [V ] = p[U ∩ A] = p[U ] ∩ p[A]. The set p[U ] is open
in Y because p is an open map; hence V is open in p[A].
Step 3. The case when A or p is closed the same mutatis mutandis. Specifically, we need only replace the word “open”
by “closed” throughout the second step. 
8 3 The Quotient Topology and Quotient Spaces

3.5 On Other Possible Pathologies of Quotient Maps

3.5.1 Composites of Quotient Maps Are Quotient Maps.

First observe that composites of maps behave nicely. It is easy to check

 that the composite of two quotient maps is a
quotient map. Indeed, this follows immediately from the equation p−1 q −1 [U ] = (q ◦ p)−1 [U ].

3.5.2 Products of Quotient Maps.

On the other hand, products of maps do not respect the property of being a quotient map. One needs further conditions
on either the maps or the spaces in order for this statement to be true. One such condition on the spaces is called
local compactness—that is, that every point has a compact neighborhood (i.e., a compact set containing an open nbhd
of the point in question)—we shall study local compactness later. A condition on the maps is the condition that both
the maps p and q be open maps. If p : X → Y and q : W → Z are quotient maps which are open, then their product
p × q : X × W → Y × Z where (x, w) 7→ (p(x), q(w)) is an open map [Pf?] and hence a quotient map as it is obviously
surjective.

3.5.3 Quotient Maps and Hausdorff Spaces.

Finally, the Hausdorff condition does not behave well; even if X is Hausdorff, there is no reason that the quotient space
X ∗ needs to be Hausdorff. There is a simple condition for X ∗ to satisfy the T1 axiom (i.e., finite point sets are closed ;
one simply requires that each element of the partition X ∗ be a closed subset of X. Conditions that ensure X ∗ is Hausdorff
are harder to find and we shall return to this question later, perhaps.

3.6 Universal Property of Quotient Maps

Recall that while we suspect a space Y to be homeomorphic to some quotient space of X if there exists a quotient map
p : X → Y , we do not yet know that this is true. We shall prove that here as a corollary of the universal property.
Theorem 3.27. Let p : X → Y be a quotient map. Let Z be a space and let g : X → Z be a map that is constant on all
fibers of p—that is, on each set p−1 [{y}] ⊆ X for y ∈ Y . Then g induces a unique map f : Y → Z such that f ◦ p = g.
Conversely, if f : Y → Z is continuous, f induces g = f ◦ p. Moreover, the map f satisfies the following properties:
(1) f is continuous iff g is continuous.
(2) f is a quotient map iff g is a quotient map.
Therefore, a function f : Y → Z is continuous iff f ◦ p is continuous.

X
g
p

Y f
Z

Proof. First we shall show such a map is unique and define the unique map f : Y → Z. Suppose h : Y → Z were
such that h(p(x)) = g(x) for all x ∈ X. Fix x ∈ X. Then, in particular, for all x0 in the fiber of p over p(x) (i.e., for all
x0 ∈ p−1 [{p(x)}]), h(p(x0 )) = h(p(x)) = g(x0 ) = g(x) as g is constant on this fiber. Let F = p−1 [{p(x)}]. Then h[F ] = g(x).
Since x was arbitrary, this is true for all x ∈ X. Since p surjects onto Y , this completely defines h and, in particular, if
y ∈ Y , then {h(y)} = h[{y}] = g p−1 [{y}] where we know that g p−1 [{y}] is a singleton since g is constant on each
fiber p−1 [{y}]. Since f must also satisfy f (p(x)) = g(x) for allx ∈ X, f is forced to be of the form of h and hence  is
unique and defined by letting f (y) be the single element of g −1 p−1 [{y}] or, abusing notation, f (y) = g −1 p−1 [{y}] .
Suppose f is continuous. Then g = f ◦ p is continuous as the composite of continuous  −1 functions. Conversely, suppose g
−1 −1 −1

is continuous. Let V be open in Z. Then g [V ] is open in X. But g [V ] = p f [V ] . Because p is a quotient map,
p−1 f −1 [V ] is open in X iff f −1 [V ] is open in Y , showing that f is continuous.
 
3.6 Universal Property of Quotient Maps 9

If f is a quotient map, then g is the composite of two quotient maps and hence is a quotient map. Conversely, suppose
g is a quotient map. Since g is surjective, so must f be surjective.  −1 Fix V ⊆ Z; we shall show that V is open in Z iff
f −1 [V ] is open in Y . Suppose f −1
[V ] is open in Y . Then p −1
f [V ] is open in X since p is continuous. By assumption
g = f ◦ p, so p−1 f −1 [V ] = g −1 [V ] and g is a quotient map, so this forces V to be open in Z. Conversely, if V is open in


Z, then since quotient maps are continuous, obviously f −1 [V ] is open in Y . 

Remark. We can rephrase part of this as follows: Quotient maps p : X → Y are characterized among the class of all such
surjective maps X → Y by the following property: if Z is any topological space and f : Y → Z any function, then f is
continuous iff f ◦ p is continuous. Hence, continuous maps g : X → Z which are constant on each fiber of p factor through
p : X → Y with an induced map f : Y → Z making the above diagram commute.

Corollary 3. Let g : X → Z be a surjective continuous map. Let X ∗ be the following collection of subsets of X:

X ∗ = g −1 [{z}] : z ∈ Z .


Give X ∗ the quotient topology via the canonical quotient map p : X → X ∗ .

(a) g is constant on all fibers p−1 [{[x]}] for [x] ∈ X ∗ .
(b) g induces a unique bijective map f : X ∗ → Z such that f ◦ p = g.
(c) g is a quotient map iff f : X ∗ → Z is a homeomorphism
X
g
p

X∗ f
Z

(d) If Z is Hausdorff, so is X ∗

Proof. (a) We have constructed X ∗ in such a way that g is constant on all fibers over p—indeed, the elements of X ∗
are precisely all fibers of g over z (i.e., sets of the form g −1 [{z}] where z ∈ Z) and hence by definition of the canonical
quotient map, p−1 [{g −1 [{z}]}] = g −1 [{z}]; whence the assertion. (b) Thus, by the previous
 −1 theorem,  g−1
induces a unique
∗
[{g −1 [{z}]}] =
 
map
 −1f : X  → Z such that f ◦ p = g given by (with the usual abuse of notation) f ( g [{z}] ) = g p
g g [{z}] = z. Since g surjects, f surjects and it is easy to see that it is precisely the surjectivity of g which makes f
an injection—hence, the unique induced map f is a bijection. Now we proceed to the “meat” of this theorem.
(c) Suppose the induced map f is a homeomorphism. Then both f is open and hence a quotient map. Since both f
and p are quotients maps, their composite g = f ◦ p is a quotient map. Conversely, suppose that g is a quotient map. By
the preceding theorem, f is a quotient map. Since f is bijective and a quotient map, a set V ⊆ Z is open in Z iff f −1 [V ]
is open. Hence, it is a bijective open map and thus a homeomorphism.
(d) Suppose Z is Hausdorff. Given distinct points of X ∗ , their images under f are distinct and thus possess disjoint
open neighborhoods U and V . Hence, f −1 [U ] and f −1 [V ] are disjoint open neighborhoods of these two given points in
X ∗ . Hence, any two points in X ∗ have disjoint open neighborhoods and so X ∗ is Hausdorff. 

Let us end by remarking that Munkres has an example showing that the product of quotient maps need not be a
quotient map on page 143. One take away from this example is that the two definitions of a quotient map: (1) p : X → Y
is surjective and V is open (resp. closed) in Y iff p−1 [V ] is open (resp. closed) in X and (2) p is a surjective map mapping
saturated open (resp. closed) sets in X (with respect to p) to open (resp. closed) sets in Y .
Chapter 4
Adjunction, or Attaching Space.

One particularly useful application of the quotient space construction is that of an adjunction space (or, an attaching space).
The idea is that one topological space is “glued” to the another. Now let’s make this concrete.
Let X and Y be topological spaces with A a subspace of Y . Let f : A → ` X be a continuous map (called the
attaching map). One forms the adjunction space X ∪f Y by identifying on X Y each a ∈ A with its image f (a) ∈ Y .
In other words, we define ∼ on X Y by saying that x ∼ x0 iff f (x) = x0 . In other words, we identify x with f (x). Even
`
more specifically, this is
z∼z ∀z ∈ X t Y
a ∼ f (a), f (a) ∼ a ∀a ∈ A
a ∼ a0 ∀a, a0 ∈ A s.t. f (a) = f (a0 )
Abusing notation, we often write
def
a
X ∪f Y = (X Y )/{f [A] ∼ A}.
` `
The topology` on this quotient space is the one induced by the resulting quotient map p : X Y → (X Y )/{f [A] ∼ A}
(i.e., p : X Y → X ∪f Y .
It turns out that adjunction space is a pushout in Top. That is, if X and Y are spaces, A ⊆ Y a subspace and
f : A → X continuous, then the triple
(X ∪f Y, p ◦ ιX , p ◦ ιY )
`
consisting
` of the adjunction space X ∪f Y = (X Y )/{f [A] ∼ A} along with the maps p ◦ ιX`and p ◦ ιY —where
p : X Y → X∪f Y is the quotient map and ιX and ιY are the canonical inclusions of X and Y into X Y , respectively—is
the universal object fitting into the diagram:

A Y

X
cont cont
By which we mean if (Q, j1 , j2 ) were another such triple where j1 : X −→ Q and j2 : Y −→ Q were another such pair fitting
into the above diagram then there is a unique continuous map (which ends up being a homeomorphism) h : X ∪f Y → Q
such that TFDC:

A Y

f p◦ιY

j2

X p◦ιX X ∪f Y

j1
Q

11
12 4 Adjunction, or Attaching Space.

Exercise. Let the notation be as above. If (Q, j1 , j2 ) is another such pushout as above in Top, then Q is homeomorphic
with X ∪f Y .
`
Exercise. Let X and Y be spaces, A` ⊆ Y a closed subspace and f : A → Y a continuous map. Let p : X Y → X ∪f Y
be the quotient map and ιX : X → X Y the canonical inclusion. Then p ◦ ιX is a closed embedding and p ◦ ιY | (Y \ A)
is an open embedding.
Chapter 5
Exercises

5.1 A Map With a Continuous Right-Inverse is a Quotient Map; Retractions are

Quotient Maps.

Exercise. (a) Let p : X → Y be a continuous map. If p has a continuous right-inverse, then p is a quotient map. That
cont
is, if there exists a continuous map f : Y −→ X such that p ◦ f = idY , then p is a quotient map.
cont
(b) If A ⊆ X, a retraction of X onto A is a continuous map r : X −→ A such that ∀a ∈ A, r(a) = a. Show that a
retraction is a quotient map.

Proof. (a) If p has a right-inverse, it must be surjective. Let U be open in Y , then p−1 [U ] is open in X by continuity of
p. Conversely, let U ⊆ Y and suppose p−1 [U ] is open in X—we show U is open in Y . Since p ◦ f = idY , (p ◦ f )−1 [U ] = U
−1 −1

but (p ◦ f ) [U ] = f p [U ] = U and since f is continuous and we assumed p1 [U ] is open in X, it must be that U is
1

open in Y . Hence, p is a quotient map.

(b) Let r : X → A be a retraction of X onto A—then r obviously surjects. Let X ∗ = {r−1 [{a}] : a ∈ A} be a partition of
X. Then as we have shown, r is constant on each fiber of the canonical quotient map p : X → X ∗ and there therefore exists
a unique continuous bijection f : X ∗ → A such that f ◦ p = r. As we know, it suffices to show that f is a homeomorphism
for in this case r is a quotient map and to this end it suffices to show that f is open. Let U ⊆ X ∗ be open. Then p−1 [U ]
is open in X. Each element [a] ∈ U ⊆ X ∗ has precisely one element of A in the fiber p−1 [{[a]}]. Then V = p−1 [U ] ∩ A
isthe setof all such elements of A. Then V is open in A as it is the intersection of an open subset of X with A. Clearly
r p−1 [U ] = V . Since r = f ◦ p, this shows that V = f [U ] and hence that f [U ] is open. Since U was arbitrary, f is an
open map. Since f is bijective and continuous, it is a homeomorphism. 

5.2 A Quotient Map That is Neither Open Nor Closed.


Exercise. Let π1 : R × R → R be the projection on the first coordinate and let A = (x, y) ∈ R2 : x ≥ 0 or y = 0 . Let
q : A → R be the map obtained by restriction π1 . Then q is a quotient map that is neither open nor closed.

5.3 Practice With Quotients.

Exercise. Define an equivalence relation ∼1 on R2 by (x0 , y0 ) ∼1 (x1 , y1 ) iff x0 +y02 = x1 +y12 . Let X ∗ be the corresponding
quotient space. It is homeomorphic to a familiar space; what is it? [Hint: Set g(x, y) = x + y 2 .] Repeat this with equivalence
relation ∼2 defined by (x0 , y0 ) ∼2 (x1 , y1 ) iff x20 + y02 = x21 + y12 .

5.4 A Quotient of a Compact Space is Compact.

Exercise. Let X be a compact space and X ∗ a quotient space of X. Then X ∗ is compact.

Proof. Let p : X → X ∗ be the resulting quotient map. Since X is compact and p is surjective, p[X] = X ∗ is compact, as
we know/shall see. 

13
14 5 Exercises

5.5 (Important) Constructing the Circle S1 As a Quotient Space of the Unit Interval I.

Notation. We denote the unit interval [0, 1] with its usual subspace topology by I = [0, 1]. We denote the unit circle
with its usual subspace topology (i.e., as a subspace of R2 or, equivalently as a subspace of C) by S1 .
Exercise. Prove that S1 is homeomorphic with a quotient of the unit interval I.
Proof. Define ∼ on I = [0, 1] by x ∼ y iff either x = 0 and y = 1 or x = 1 and y = 0. This defines a partition I ∗ = I/ ∼
into the equivalence classes comprised of the singletons in (0, 1) and the equivalence class [0] = {0, 1}. Let p : I → I/ ∼
def
be the resulting quotient map. The basic saturated open sets of I are therefore open sets in I ◦ = (0, 1) and unions of an
open nbhd of 0 and an open nbhd of 1.
Viewing S1 ⊆ C, define g : I → S1 by g(t) = e2πit . This is obviously continuous and its image is all of Z with g(0) = g(1).
Hence, in particular, g is constant on p−1 [{[x]}] for each [x] ∈ I ∗ . Now, since g is surjective and continuous, g induces a
unique, bijective and continuous map f : I/ ∼→ S1 such that f ◦p = g—namely, abusing notation, f ([x]) = g −1 p−1 [{[x]}]
satisfying f ◦ p = g.
If we can show that f is a continuous bijection from a compact space to a Hausdorff space, then by Theorem 1.44,
f is a homeomorphism. Since [0, 1] is compact and p : [0, 1] → I/ ∼ is continuous, I/ ∼ is compact. Since we know that f
is continuous and bijective, it remains to show that f is injective. This is clear because if f ([x]) = f ([x0 ]) but [x] 6= [x0 ],
then there exists x1 ∈ p−1 [{[x]}], x2 ∈ p−1 [{x0 }] (hence, x1 6= x2 ) such that f (p(x1 )) = f (p(x2 )) since p surjects. But then
since f ◦ p = g, g(x1 ) = g(x2 ). Since g is constant on the fibers of p, this is only possible if x1 ∼ x2 as described above,
but this contradicts the assumption that [x] 6= [x0 ] since we then must have p(x1 ) = p(x2 ). 

5.6 More Practice with Quotients.

Notation. Recall that S2 is the unit sphere viewed as a subspace of R3 .


Exercise. Let S+ be the closed upper hemisphere in S2 (i.e., S+ = (x, y, z) : x2 + y 2 + z 2 = 1 and z ≥ 0 ). Define an
equivalence relation ∼1 on S+ by
(
0 x = x0 , or
x ∼1 x if 0 0
x, x ∈ BdS2 (S+ ) and x = −x

and define an equivalence relation ∼2 on the entire 2-sphere S2 by

(
0 x = x0 , or
x ∼2 x if
x = −x0 .

Then the inclusion ι : S+ ,→ S2 induces a homeomorphism h : S+ / ∼1 → S2 / ∼2 .

[Hint: Let p1 : S+ → S+ / ∼1 and p2 : S2 → S2 / ∼2 be the corresponding quotient maps. Prove that our map h makes the
following diagram
ι
S+ S2
p2 ◦ι
p1 p2

S+ / ∼1 h
S2 / ∼2

commute.]

5.7 The Subspace Topology and Quotient Topology Are Categorical Duals in Top.

Definition (Subspace Topology). Let Y be a topological space, X a set and let ι : X → Y be an injection. We define
the subspace topology on X as the initial topology induced by this diagram.
Exercise. The subspace topology and the quotient topology are categorical duals in Top.
Index

adjunction space, 11 pushout, 11

attaching map, 11
attaching space, 11 quotient map, 3
quotient space, 4
closed map, 3
quotient topology, 4
decomposition space, 5
retraction, 13
identification space, 5
initial topology, 14 saturated, 3
subspace topology, 14
local compactness, 8

open map, 3 unit 2-sphere, 6

15