Topology (H) Lecture 6

Lecturer: Zuoqin Wang

Time: March 25, 2021

THE QUOTIENT TOPOLOGY

1. The quotient topology

¶ The quotient topology.
Last time we introduced several abstract methods to construct topologies on ab-
stract spaces (which is widely used in point-set topology and analysis). Today we will
introduce another way to construct topological spaces: the quotient topology.
In fact the quotient topology is not a brand new method to construct topology.
It is merely a simple special case of the co-induced topology that we introduced last
time. However, since it is very concrete and “visible”, it is widely used in geometry
and algebraic topology. Here is the definition:
Definition 1.1 (The quotient topology).
(1) Let (X, TX ) be a topological space, Y be a set, and p : X → Y be a surjective
map. The co-induced topology on Y induced by the map p is called the quotient
topology on Y . In other words,
a set V ⊂ Y is open if and only if p−1 (V ) is open in (X, TX ).
(2) A continuous surjective map p : (X, TX ) → (Y, TY ) is called a quotient map,
and Y is called the quotient space of X if TY coincides with the quotient
topology on Y induced by p.
(3) Given a quotient map p, we call p−1 (y) the fiber of p over the point y ∈ Y .
Note: by definition, the composition of two quotient maps is again a quotient map.
Here is a typical way to construct quotient maps/quotient topology: Start with a
topological space (X, TX ), and define an equivalent relation ∼ on X. Recall that this
means
• x ∼ x;
• x ∼ y =⇒ y ∼ x;
• x ∼ y, y ∼ z =⇒ x ∼ z.
Then one gets an abstract space consisting of all equivalence classes
Y = X/ ∼
and a natural projection map
p : X → X/ ∼, x 7→ [x].

1
2 THE QUOTIENT TOPOLOGY

So in this case, each fiber is an equivalence class. Note that the “quotient by a map”
description and the “quotient by an equivalence relation” description are equivalent:
Given any equivalence relation description, we have a natural projection map as shown
above; conversely given any quotient map f : X → Y , we can define an equivalence
relation by x ∼ y ⇐⇒ f (x) = f (y) and thus get an equivalence relation description of
the same quotient space.
Example 1.2 (The circle). One can regard the circle S 1 as a quotient space via
(1) S 1 = [0, 1]/{0, 1}: in other words, the only equivalence is 0 ∼ 1.
(2) S 1 = R/Z: in other words, we used the equivalence relation
x ∼ y ⇐⇒ x − y ∈ Z.
Example 1.3. Define an equivalence relation ∼ on R by
x ∼ y ⇐⇒ x − y ∈ Q.
Then what is the quotient topology on X = R/ ∼? Let U ⊂ X be an open set. Then
p−1 (U ) is open in R. In particular, there is an interval (a, b) ⊂ U . Since any real
number x ∈ R is equivalent to some number in (a, b), we must have p−1 (U ) = R. So
the quotient topology on X is the trivial topology.

¶ Universality.
According to the universality of the co-induced topology, namely Proposition 2.8
in Lecture 5 (whose proof is in your PSet), we have
Theorem 1.4 (Universality of quotient topology). Let X, Y, Z be topological spaces,
p : X → Y be a quotient map, and f : Y → Z be a map. Then f is continuous if
and only if g = f ◦ p is continuous. Moreover, the quotient topology on Y is the only
topology satisfying this property.

As a consequence, we have
Corollary 1.5. If p : X → Y is a quotient map, f : X → Z is a continuous map such
that f is constant on each fiber. Then the naturally induced map
f¯ : Y → Z, f¯(y) := f (p−1 (y))
is continuous.

¶ The real projective space.

Let’s give an important example of quotient space: the real projective space. We
can give two descriptions.
Example 1.6 (The real projective space).
 THE QUOTIENT TOPOLOGY 3

• On X = Rn+1 − {0} we can define an equivalence relation

x ∼ y ⇐⇒ ∃0 6= λ ∈ R s.t. x = λy.
The quotient space
RPn = Rn+1 − {0}/ ∼
(endowed with the quotient topology1) is called the real projective space. This
gives a geometric explanation of the real projective space:
RPn = the space of all lines in Rn+1 passing the origin 0.
• We can also start with the unit sphere S n ⊂ Rn+1 and define an equivalence
relation
x ∼ y ⇔ x = ±y.
n+1
Since every line in R passing the origin 0 intersect S n exactly at two antipo-
dal points, the resulting quotient space are the same.

Note that when n = 1, RP1 is in fact homeomorphic to S 1 , since according to the

second description, we may start with a half circle and identify the two end points.
However, the geometric picture, even in the case n = 2 where we may start the con-
struction with a hemisphere, is very complicated:

Figure 1. A “picture” of RP2

As one can see, there is a “self-intersection” in the picture. However, the intersec-
tion should not exist in a real “picture” of RP2 . In fact, there is no way to embed RP2
into R3 . It can only be embedded into R4 . Moreover, just like the Möbius band, RPn
is not orientable for any even number n.
Remark 1.7. Similarly, one can define a topology on the space of complex lines in Cn+1
(the complex projective space CPn ). More generally, one can define a topology on the
space of k-dimensional vector subspaces of a vector space V (the Grassmannian mani-
fold Gr(k, V )2). Note that RPn is just a special Grassmannian : RPn = Gr(1, Rn+1 ).
1
Here, we endow with Rn+1 − {0} the standard Euclidean topology.
2However, for k > 1, Gr(k, V ) can not be realized as a quotient space of V . Instead it can be
realized as a quotient space of a much larger space, e.g. GL(V ).
4 THE QUOTIENT TOPOLOGY

¶ Construction: Gluing one point to another in the same space.

In what follows we will introduce many very concrete geometric ways to get quotient
spaces from known spaces. The first is :
Gluing : Let X be a topological space, by gluing the points a and b in
X, we means: considering the quotient space obtained from the equiv-
alence relation which contains only one non-trivial equivalence: a ∼ b.
Similarly, we may glue a subset A to a subset B in X by identifying
each point in A to a specific point in B.
This is widely used in constructing surfaces topologically from planar polygons: just
glue boundary line segments using prescribed way.3

Figure 2. The cylinder

Figure 3. The Möbius band

Figure 4. The torus

Figure 5. The Klein bottle

3Note that the Klein bottle can not be embedded into R3 . It is a non-orientable surface.
 THE QUOTIENT TOPOLOGY 5

It turns out that any (compact) surface can be constructed by starting at a suitably-
chosen polygon and attaching its boundary edges in suitable way. Here is a more
complicated one:

Figure 6. The 2-torus

At the end of this semester, we will use such polygonal presentation to prove the
classification theorem of compact surfaces.

¶ Construction: Attaching space (adjunction space).

We may attach one space to another along a given map:
Attaching space : Let X, Y be topological spaces, and A ⊂ Y a sub-
space, and f : A → X a continuous map. Then the attaching space
X ∪f Y is formed by taking the disjoint union of X and Y and identi-
fying each a ∈ A with f (a) ∈ X, i.e.
X ∪f Y = X t Y /{a ∼ f (a)}.
There are two special cases that we are going to use later:
(1) (The wedge sum) Given two topological spaces X and Y , the wedge sum X ∨ Y
of X and Y is formed by attaching one point in X to one point in Y :
X ∨ Y = X t Y /{x0 ∼ y0 }.

W of spaces Xα , with a point xα ∈ Xα chosen, we

More generally, given a family
may form the wedge sum α Xα by attaching all Xα ’s at the point xα ’s.
6 THE QUOTIENT TOPOLOGY

Figure 7. S 1 ∨ S 1 Figure 8. S 1 ∨ S 1 ∨ S 1 ∨ S 1 ∨ S 1

Remark 1.8. When talking about wedge sum, we are really working on “pointed
space” (X, x0 ), namely a space with a point x0 chosen. The wedge sum of (X, y0 )
and (Y, y0 ) is again a pointed space (X ∨ Y, {x0 }). By this way, when studying
the wedge sum of many spaces, we are always attaching the marked points into
one point.
(2) (The connected sum) Given two geometric objects A and B that are locally
Euclidian (“manifolds”), the connected sum A#B is constructed as follows:
one can remove a small ball (disk) from each, and then glue the boundary
spheres (circles) so that they are “connected together”. 4

Figure 9. The connected sum A#B

In other words,
A#B = (A − D1 ) ∪f (B − D2 ),
where D1 , D2 are small disks on A and B respectively, and f is the attaching
map that identify ∂D1 with ∂D2 as shown in the picture.

¶ Construction: Squeeze a subset into one point.

Next let’s consider
4This
is widely used in constructing new surfaces from given surfaces, or more in constructing new
manifolds from given ones in manifold theory.
 THE QUOTIENT TOPOLOGY 7

Squeeze : Let X be a topological space, and Y ⊂ X. We may define

an equivalence relation on X by requiring and only requiring y1 ∼ y2
for any y1 , y2 ∈ Y . In other words, in the quotient space, we “squeeze”
all points in Y to one point. For simplicity we just denote the quotient
space by X/Y .
For example, we consider the unit disk D in R2 . We can squeeze its boundary
circle into one point. What do we get? A sphere S 2 ! Similarly we may squeeze the
boundary sphere S n−1 of the unit ball B(0, 1) in Rn to get Sn .

Figure 10. Squeeze the boundary circle to get a sphere

As another example: By regarding X as X × {y0 } and regarding Y as {x0 } × Y ,

we can view the wedge sum X ∨ Y as a subspace of X × Y . Then one may define the
smash product of X and Y , denoted by X ∧ Y , as
X ∧ Y = X × Y /X ∨ Y.
¶ Construction: the cone space and the suspension.
Given any topological space X, one may construct the cone space and the sus-
pension of X (used in algebraic topology), both as a quotient space of the cylinder
X × [0, 1]:
(1) The cone space of X, denoted by C(X), is formed by squeezing X × {0} in
X × [0, 1] into one point, namely, C(X) = X × [0, 1]/X × {0}:

Figure 11. The cone space C(X)

8 THE QUOTIENT TOPOLOGY

(2) The suspension of X, denoted by S(X), is formed by squeezing X × {0} to one

point, and also squeezing all points in X × {1} to another point.

Figure 12. The suspension S(X)

(3) More generally, given topological spaces X and Y , the join of X and Y , some-
times denoted by X ? Y , is defined as X ? Y = X × Y × I/ ∼, where ∼ is given
by
(x, y1 , 0) ∼ (x, y2 , 0), (x1 , y, 1) ∼ (x2 , y, 1), ∀x, x1 , x2 ∈ X; y, y1 , y2 ∈ Y.

Figure 13. The join X ? Y

¶ Construction: Mapping cylinder, mapping cone and mapping torus.

One may also study spaces associated to maps:
(1) Given a continuous map f : X → Y , the mapping cylinder of f , denoted by
Mf = (X × [0, 1]) tf˜ Y,
is by definition the attaching space of X ×[0, 1] and Y via the map fe : X ×{0} →
Y, f (x, 0) := f (x).
 THE QUOTIENT TOPOLOGY 9

Figure 14. Mapping cylinder

(2) Given a continuous map f : X → Y , the mapping cone of f , denoted by Cf , is

by definition to be the quotient space
Cf = (X × [0, 1]) tf˜ Y / ∼,
namely the quotient space of the mapping cylinder Mf with respect to the
equivalence relation
(x1 , 1) ∼ (x2 , 1), (x, 0) ∼ f (x), ∀x, x1 , x2 ∈ X.

Figure 15. Mapping cone

(3) Given a homeomorphism f : X → X, the mapping torus of f is defined to be

Tf := X × [0, 1]/(1, x) ∼ (0, f (x)).
Mapping tori of surface homeomorphisms play a key role in the theory of 3-
manifolds and have been intensely studied.

2. Quotient by a group action

¶ The homeomorphism group.
Symmetries play an essential role in all branches of mathematics. The mathemat-
ical language of symmetry is group.
Proposition 2.1. Let X be a topological space and let
Hom(X) = {f : X → X | f is a homeomorphism}.
10 THE QUOTIENT TOPOLOGY

Then under the usual composition of maps, Hom(X) is a group.

Proof. This is almost trivial:
• given two homeomorphisms f and g of X, the composition g ◦ f is again a
homeomorphism of X, and the associativity holds by definition,
• the identity map Id is the identity element in this group,
• the inverse map f −1 is a homeomorphism and is the inverse of f in this group.

So given any topological space X, we have a god-given group that describes sym-
metries of X in the category of topology:
Definition 2.2. Hom(X) is called the homeomorphism group of X.

Note that for any element f ∈ Hom(X), we may say f “acts” on the space X by
sending an element x ∈ X to its image f (x) ∈ X.

¶ The group action.

We define
Definition 2.3. Let G be any group, and X be a space.
(1) An (left) action 5 of group G on the space X is a map
α : G × X → X, (g, x) 7→ g · x,
so that for any x ∈ X and any g, h ∈ G,
• e · x = x for any x ∈ X,
• g · (h · x) = (gh) · x, for any g, h ∈ G and x ∈ X.
(2) In the case X is a topological space, an (left) action of the group G on the
topological space X is an action so that for any g ∈ G, the map
τg : X → X, x 7→ τg (x) := g · x
is a continuous map (and thus is a homeomorphism since (τg )−1 = τg−1 is also
continuous).
(3) In the case X is a topological space and G is a topological group, we say the
action is a continuous action if the map α is a continuous map.
Remark 2.4. So an action of G on a topological space X is a group homomorphism
τ : G → Hom(X) = {f : X → X | f is a homeomorphism},
i.e. associate to any g ∈ G a homeomorphism τg : X → X, such that
τg ◦ τh = τgh , ∀g, h ∈ G.
5There
is also a conception of right action, in which we use x · g instead, and replace the second
condition by (x · g) · h = x · (gh). The theory for right action is almost the same as left actions.
 THE QUOTIENT TOPOLOGY 11

Note that we may (and will always) assume τ is injective. Otherwise we can always
replace G by G/ker(τ ), which acts on X in the obvious way. Such an action is called
a faithful action.

¶ Orbits and the orbit space.

Definition 2.5. Given a group action of G on X, the orbit of x ∈ X is the set
G · x := {g · x | g ∈ G.}

We will see many examples below where the orbit is very simple. Here we give an
example where the orbit is very complicated:
Example 2.6. Consider S 1 acts on S 1 × S 1 by
√
eiα · (eiθ1 , eiθ2 ) := (ei(θ+α) , ei(θ+ 2α)
).
Then the orbit is a “dense curve” on the torus S 1 × S 1 .

We may define an equivalence relation ∼ on X by

x1 ∼ x2 ⇐⇒ ∃ g ∈ G s.t. x1 = g · x2 .
In other words, two elements in X are equivalent if and only if they lie in the same
orbit. It is easy to check that this is an equivalence relation.
Definition 2.7. Given a group action of G on a topological space X, the orbit space
is defined to be the quotient space X/G = X/ ∼.

So by definition, the orbit space is “the space of orbits”, endowed with the quotient
topology.
Example 2.8. Consider the R>0 (as a multiplicative group) action on R by multiplica-
tion, i.e.
a · x := ax.
Then there are three orbits: R>0 , {0}, R<0 . As a result, the orbit space consists of
three elements, {+, 0, −}, and the topology on the orbit space is
{∅, {+}, {−}, {+, −}, {+, 0, −}}.

¶ Examples.
We list several simple examples of orbit spaces which we will use when we study
covering spaces later this semester.
Example 2.9 (S 1 again). G = Z acts on X = R via
τ (n)(x) = n + x. (translation)
R/Z ' S 1 .
12 THE QUOTIENT TOPOLOGY

Example 2.10 (S 1 again and again). G = Zn acts on X = S 1 ⊂ C via

τ (k)(z) = e2πik/n z. (rotation)
S 1 /Zn ' S 1 .
f = S n via
Example 2.11 (RPn again). G = Z2 acts on X
τ (1)(x) = x and τ (−1)(x) = −x. (antipodal)
n
S n /Z2 ' RP .
Example 2.12 (n-torus). G = Zn acts on X = Rn via
τ (m1 , · · · , mn )(x1 , · · · , xn ) = (x1 + m1 , · · · , xn + mn ).
Rn /Zn ' Tn ' S 1 × · · · × S 1 .
Example 2.13 (Lens space L(p, q)). Let p, q be co-prime numbers. We define an action
of G = Zp = {1, e2πi/p , · · · , e2πi(p−1)/p } on X = S 3 ⊂ C2 via
τ (e2πik/p )(z1 , z2 ) = (e2πik/p z1 , e2πikq/p z2 ).
L(p; q) := S 3 /Zp is known as the lens space.
¶ Example: Hopf fibration.
Finally we give an example of a continuous group action (which is not properly
discontinuous).
Let’s regard the circle group S 1 as
S 1 = {z ∈ C | |z| = 1}
and regard the three dimensional sphere S 3 as
S 3 = {(z1 , z2 ) ∈ C2 | |z1 |2 + |z2 |2 = 1}.
Then we can define an action of S 1 on S 3 via
z · (z1 , z2 ) := (zz1 , zz2 ).
Then one has
(1) Each orbit S 1 · (z1 , z2 ) is homeomorphic to a circle.
(2) The orbit space S 3 /S 1 is homeomorphic to S 2 :
A sketch of proof: We let
X = {(z1 , z2 ) ∈ S 3 | |z1 | ≤ |z2 |}
and
Y = {(z1 , z2 ) ∈ S 3 | |z1 | ≥ |z2 |}.
Note that both X and Y (and thus X ∩ Y ) are invariant under the
S 1 -action. So S 3 /S 1 can be constructed as gluing X/S 1 and Y /S 1
along the “boundary” X ∩ Y /S 1 which is a quotient of the torus
X ∩ Y = (z1 , z2 ) ∈ S 3 | |z1 | = |z2 |}
 THE QUOTIENT TOPOLOGY 13

by S 1 , and thus is a circle. Now consider X/S 1 . We can define a map

1
f : D2 → X, z 7→ √ (z, 1)
2
and show that f is a homeomorphism which maps the boundary circle
of D2 to the boundary circle X ∩ Y /S 1 . Similarly Y /S 1 is homeomor-
phic to a disk whose boundary gets mapped to X ∩ Y /S 1 . As a result,
the quotient S 3 /S 1 is homeomorphic to the space obtained by gluing
two unit disks along their boundary, which is the sphere S 2 !
The quotient map p : S 3 → S 3 /S 1 ' S 2 is known as the Hopf fibration and plays an
important role in geometry and topology.