Quotient Spaces and Identification Topologies

Quotient spaces glue together points of a topological space according to an equivalence relation, equipping the resulting set of equivalence classes with the finest topology making the projection continuous. Key result: a set in the quotient is open if and only if its preimage under the projection is open. Applies to: constructing spheres, tori, Möbius bands, projective spaces, CW-complexes, and orbit spaces of group actions. Origin: crystallized in the early 20th century within general topology (Hausdorff, Alexandroff, Kuratowski).

Given a topological space $X$ and a surjection $q : X \to Y$ , the identification (quotient) topology on $Y$ declares a subset $U \subseteq Y$ open precisely when $q^{- 1} (U)$ is open in $X$ . When $Y$ is the set $X / \sim$ of equivalence classes of a relation $\sim$ on $X$ , the space $Y$ is called the quotient space. This construction is the universal way to "collapse" or "glue" parts of a space, and it underlies almost every concrete surface and complex built in topology. The surprising fact is that quotients can destroy desirable properties effortlessly: a quotient of a beautifully Hausdorff, metrizable space need not even be Hausdorff. Quotients are characterized not by an explicit formula but by a universal property that makes them the categorical dual of subspaces.

The Core Insight

A quotient topology is the largest possible topology on the target that keeps the projection continuous. Continuity from $X$ down to $Y$ is "free"; the real content is that a function out of $Y$ is continuous iff its composite with the projection is continuous. Quotients answer the question: what is the right topology on a set of glued-together points? — and the answer is forced by demanding maximal openness compatible with $q$ .

Formal Statement

Equivalence Relation and Quotient Set

An equivalence relation $\sim$ on a set $X$ is a relation that is reflexive ( $x \sim x$ ), symmetric ( $x \sim y \Rightarrow y \sim x$ ), and transitive ( $x \sim y, y \sim z \Rightarrow x \sim z$ ). The equivalence class of $x$ is

[x] = {y \in X : y \sim x} .

⟳directly unwind the membership condition; the three axioms guarantee the classes partition

X

The quotient set is the collection of classes,

X / \sim = {[x] : x \in X},

with the canonical projection $q : X \to X / \sim$ , $q (x) = [x]$ .

⟳

q

is automatically surjective since every class is

[x]

for some

x

Quotient / Identification Topology

Let $X$ be a topological space and $q : X \to Y$ a surjection. The quotient topology $τ_{q}$ on $Y$ is

τ_{q} = {U \subseteq Y : q^{- 1} (U) is open in X} .

⟳verify

\emptyset, Y

qualify and that

q^{- 1}

commutes with arbitrary unions and finite intersections, so

τ_{q}

is a topology.

When $Y = X / \sim$ and $q$ is the canonical projection, $(X / \sim, τ_{q})$ is the quotient space. A surjection $q$ carrying this topology is called an identification map (or quotient map): a continuous surjection such that

U open in Y ⟺ q^{- 1} (U) open in X .

⟳the forward direction is continuity; the backward direction is the maximality of

τ_{q}

Universal Property of the Quotient

Let $q : X \to Y$ be a quotient map. A function $f : Y \to Z$ (with $Z$ any topological space) is continuous if and only if the composite $f \circ q : X \to Z$ is continuous:

f continuous ⟺ f \circ q continuous .

⟳for

V

open in

Z

, note

(f \circ q)^{- 1} (V) = q^{- 1} (f^{- 1} (V))

, then apply the iff of equation [4].

Equivalently, a continuous $g : X \to Z$ that is constant on fibers of $q$ (i.e. $q (x) = q (x^{'}) \Rightarrow g (x) = g (x^{'})$ ) factors uniquely through a continuous $\overset{g}{ˉ} : Y \to Z$ with $g = \overset{g}{ˉ} \circ q$ .

Closed-Set Criterion

A subset $C \subseteq Y$ is closed in the quotient topology if and only if $q^{- 1} (C)$ is closed in $X$ :

C closed in Y ⟺ q^{- 1} (C) closed in X .

⟳take complements in [3] and use

q^{- 1} (Y ∖ C) = X ∖ q^{- 1} (C)

Why It Holds

Quotient Topology is the Finest Topology Making $q$ Continuous

We show $τ_{q}$ defined in [3] is the largest topology on $Y$ for which $q$ is continuous.

Step 1 — $τ_{q}$ is a topology.

q^{- 1} (\emptyset) = \emptyset

and

q^{- 1} (Y) = X

are open, so

\emptyset, Y \in τ_{q}

. For a family

{U_{i}}

τ_{q}

q^{- 1} (i ⋃ U_{i}) = i ⋃ q^{- 1} (U_{i}),

a union of open sets, hence open; thus

⋃_{i} U_{i} \in τ_{q}

. For finitely many

U_{1}, \dots, U_{n}

q^{- 1} (k = 1 ⋂ n U_{k}) = k = 1 ⋂ n q^{- 1} (U_{k}),

a finite intersection of opens, hence open; thus

⋂ U_{k} \in τ_{q}

. Step 2 — $q$ is continuous for $τ_{q}$ . By definition every

U \in τ_{q}

has

q^{- 1} (U)

open, which is exactly continuity. Step 3 — Maximality. Suppose

σ

is any topology on

Y

making

q

continuous. Then for each

U \in σ

q^{- 1} (U)

is open, so

U \in τ_{q}

by [3]. Hence

σ \subseteq τ_{q}

: every topology rendering

q

continuous is coarser than

τ_{q}

. Therefore

τ_{q}

is the finest such topology.

■

The Universal Property

( $\Rightarrow$ ) If $f : Y \to Z$ is continuous and $q$ is continuous, then $f \circ q$ is continuous as a composite.

( $\Leftarrow$ ) Suppose $f \circ q$ is continuous. Let $V \subseteq Z$ be open. Then

q^{- 1} (f^{- 1} (V)) = (f \circ q)^{- 1} (V)

is open in

X

. By the quotient criterion [4],

f^{- 1} (V)

is open in

Y

. Since

V

was arbitrary,

f

is continuous.

Uniqueness of the induced map. If

g

is constant on fibers, define

\overset{g}{ˉ} ([x]) := g (x)

; this is well-defined precisely because

g

does not depend on the representative. Then

\overset{g}{ˉ} \circ q = g

, and any

h

with

h \circ q = g

satisfies

h ([x]) = h (q (x)) = g (x) = \overset{g}{ˉ} ([x])

, so

h = \overset{g}{ˉ}

. Continuity of

\overset{g}{ˉ}

follows from the iff above applied to

\overset{g}{ˉ} \circ q = g

■

Felix Hausdorff & Pavel Alexandroff | 1868–1942, 1896–1982

Hausdorff's Grundzüge der Mengenlehre (1914) systematized point-set topology and the separation axioms that quotients so readily violate. Alexandroff, alongside Heinz Hopf, developed the combinatorial and constructive side — simplicial complexes, identification of cells, and the "gluing" viewpoint that became the language of CW-complexes (J.H.C. Whitehead, 1949). The quotient-as-universal-construction perspective matured with the categorical reformulation of topology in the mid-20th century.

Building Intuition

Think of the projection $q$ as a stamping machine that presses several points of $X$ onto a single point of $Y$ . The topology question is: which collections of stamped points should we call "open"? The rule is conservative and honest: a region downstairs is open exactly when the entire bundle of points that map into it was already open upstairs. Nothing is open in $Y$ "by fiat" — openness must be witnessed in $X$ .

A second mental picture: gluing the edges of paper. To form a cylinder you identify the left edge of a rectangle with the right edge; to form a torus you also glue top to bottom. The quotient topology tells you exactly how neighborhoods behave along the seam — a small disk straddling the seam is genuinely a disk, not two half-disks, because its preimage in the rectangle is open.

Forming a torus by identifying opposite edges of a square

Worked Examples

Circle as a Quotient of an Interval

Given:

X = [0, 1]

with the subspace topology from

R

, and the equivalence relation generated by

0 \sim 1

(all other points only equivalent to themselves). Goal: show

X / \sim

is homeomorphic to the circle

S^{1} = {(x, y) : x^{2} + y^{2} = 1}

. Step 1. Define

g : [0, 1] \to S^{1}

g (t) = (cos 2 π t, sin 2 π t) .

This is continuous and surjective, with

g (0) = g (1) = (1, 0)

and otherwise injective. Step 2.

g

is constant on fibers of

q

: the only nontrivial fiber is

{0, 1}

, and

g (0) = g (1)

. By the universal property [5], there is a unique continuous

\overset{g}{ˉ} : [0, 1] / \sim \to S^{1}

with

\overset{g}{ˉ} \circ q = g

. Step 3.

\overset{g}{ˉ}

is a continuous bijection. The domain

[0, 1] / \sim

is compact (continuous image of the compact

[0, 1]

) and

S^{1}

is Hausdorff. A continuous bijection from a compact space to a Hausdorff space is a homeomorphism. Answer:

[0, 1] / (0 \sim 1) ≅ S^{1}

(No units — this is a topological identification, dimensionless.)

Real Projective Line $\mathbb{RP}^1$

Given:

X = S^{1} \subseteq R^{2}

with the relation

x \sim - x

(antipodal points identified). Each class has exactly two representatives. Step 1. The projection

q : S^{1} \to S^{1} / \sim = RP^{1}

is a 2-to-1 continuous surjection. Parametrize

S^{1}

θ \in [0, 2 π)

; then

θ

and

θ + π

are identified. Step 2. Reparametrize: the map

ϕ (θ) = 2 θ (mod 2 π)

sends both

θ

and

θ + π

to the same value, since

2 (θ + π) = 2 θ + 2 π \equiv 2 θ

. So

ϕ

is constant on fibers and induces

\overset{ˉ}{ϕ} : RP^{1} \to S^{1}

. Step 3.

\overset{ˉ}{ϕ}

is a continuous bijection from a compact space (

RP^{1}

, image of compact

S^{1}

) to the Hausdorff

S^{1}

, hence a homeomorphism. Answer:

RP^{1} ≅ S^{1}

The "doubling map" wraps the circle around itself twice, and the antipodal quotient unwinds exactly that doubling. (Contrast: $RP^{2} = S^{2} / (\pm)$ is not homeomorphic to $S^{2}$ .)

Common Misconceptions

Quotients Need Not Be Hausdorff

The most common error is to assume a quotient inherits separation properties from $X$ . It does not. Even if $X$ is metrizable (hence Hausdorff, normal, etc.), $X / \sim$ can fail to be Hausdorff, $T_{1}$ , or even $T_{0}$ . The quotient topology guarantees only that $q$ is continuous and the universal property holds. A useful sufficient condition: if $X$ is compact Hausdorff and the relation $\sim$ is closed (i.e. ${(x, y) : x \sim y}$ is closed in $X \times X$ ), then $X / \sim$ is Hausdorff.

The Line with Two Origins

Take two copies of the real line, $X = R \times {0} ⊔ R \times {1}$ , and identify $(t, 0) \sim (t, 1)$ for every $t \neq = 0$ (leaving the two origins distinct). The quotient $Y$ is locally Euclidean and $T_{1}$ , but not Hausdorff: the two origins $0_{a}$ and $0_{b}$ cannot be separated by disjoint open sets, since every neighborhood of each contains a punctured interval $(- ε, ε) ∖ {0}$ that overlaps the other's. This shows separation can collapse under quotients even when no points are glued together at the troublesome spot.

Open Maps vs. Quotient Maps

A quotient map need not be open or closed. Conversely, every continuous open surjection and every continuous closed surjection is a quotient map — these are convenient sufficient conditions, but neither is necessary. Do not conflate "identification map" with "open map." Also note: the quotient of an injective continuous map can fail to be a quotient map unless it is an embedding.

Connections

Quotient topologies are the categorical dual of Subspace Topology: subspaces use the coarsest topology making the inclusion continuous (initial topology), while quotients use the finest topology making the projection continuous (final topology). The Product Topology is an initial construction; the disjoint union (coproduct) topology is a final construction, and gluing two spaces along a subspace (an adjunction space $X \cup_{f} Y$ ) is built as a quotient of a disjoint union. Quotients by group actions feed into Covering Spaces and Orbit Spaces, while iterated cell attachments via quotients build CW Complexes. The behavior of Connectedness and Compactness under continuous maps explains why those properties do survive quotients.

Real-World Applications

Surface topology / computer graphics. Tori, Klein bottles, and Möbius bands are constructed as edge-identification quotients of polygons; texture-mapping and "wrap-around" game worlds (Asteroids' screen-edge teleport) realize the torus quotient $R^{2} / Z^{2}$ .
Robotics configuration spaces. A robot arm with a rotating joint has configuration space $S^{1} = R /2 π Z$ ; multi-joint systems give tori as quotients, central to motion planning.
Crystallography. Periodic crystal lattices live on the quotient torus $R^{3} / L$ for a lattice $L$ ; the Brillouin zone in solid-state physics is exactly such a fundamental-domain quotient.
Projective geometry in vision. Camera images use $RP^{2}$ , the quotient of $R^{3} ∖ {0}$ by scaling, to model lines of sight (homogeneous coordinates).
Data analysis (TDA). Persistent homology and Mapper build simplicial complexes whose geometric realizations are quotients of disjoint simplices glued along faces.
Quotient manifolds in physics. Orbifolds and gauge-symmetry quotients ( $fields / gauge group$ ) in gauge theory and string theory are identification spaces.
Mesh processing. Closing a 3D scanned surface into a watertight manifold uses vertex/edge identifications — quotients in disguise.

Historical Development

1895–1904: Henri Poincaré's Analysis Situs introduces building manifolds by gluing faces of polyhedra — the prototype of identification spaces.
1906: Maurice Fréchet abstracts metric and limit notions, seeding general topology.
1914: Felix Hausdorff's Grundzüge der Mengenlehre establishes the axiomatic theory of topological spaces and separation axioms.
1922–1933: Kazimierz Kuratowski, Pavel Alexandroff, and the Moscow/Warsaw schools formalize continuous maps, closure operators, and the explicit notion of identification topology.
1932: Alexandroff & Hopf's Topologie presents systematic gluing constructions for simplicial complexes.
1935–1945: The categorical viewpoint emerges (Eilenberg–Mac Lane), framing quotients via universal properties and final topologies.
1949: J.H.C. Whitehead introduces CW-complexes, defined entirely by iterated quotient (cell-attaching) constructions.
1955–1960s: Standard textbook treatments (Kelley's General Topology, 1955) canonize the quotient/identification topology as taught today.

Subtleties & Limits

Quotients are notoriously ill-behaved with respect to products: the product of two quotient maps $q_{1} \times q_{2}$ need not be a quotient map. Whitehead's theorem rescues this when one factor is locally compact Hausdorff, but the general failure motivated the development of convenient categories of spaces (compactly generated weak Hausdorff spaces, used in modern algebraic topology) precisely so that quotients and products cooperate. Open practical questions in applied topology include: which equivalence relations on a manifold yield a quotient that is again a manifold? (answer: free, proper, properly discontinuous group actions give manifolds — but the precise interplay of smoothness, properness, and Hausdorffness remains a recurring source of subtle errors). The classification of all quotients of even simple spaces up to homeomorphism is, in general, intractable.