Altanis

2.5Subspace Topology

Updated 27 May 2026Chapter (PDF)

[2.5.1]Definition#

Let (X,TX)(X, \Tau_X) be a topological space, and let YXY \subseteq X. Then the collection

TY={YU:UTX}\Tau_Y = \{ Y \cap U : U \in \Tau_X \}

is the subspace topology on YY.

Proof.

Note that ,XTX\emptyset, X \in \Tau_X, and these sets in TX\Tau_X correspond directly to ,Y\emptyset, Y in TY\Tau_Y. Now suppose {Yλ}λΛ\{Y_\lambda\}_{\lambda \in \Lambda} is a collection of sets in TY\Tau_Y, where each Yλ=YUλY_\lambda = Y \cap U_\lambda. Then

λΛYλ=λΛYUλ=Y(λΛUλ)TX,\bigcup_{\lambda \in \Lambda} Y_\lambda = \bigcup_{\lambda \in \Lambda} Y \cap U_\lambda = Y \cap \underbrace{\left(\bigcup_{\lambda \in \Lambda} U_\lambda \right)}_{\in \Tau_X},

and so TY\Tau_Y is closed under arbitrary union. An entirely symmetric argument follows to show TY\Tau_Y is closed under finite intersection—thus TY\Tau_Y is a topology.

[2.5.2]Theorem#

Suppose (X,TX)(X, \Tau_X) is a topological space with basis B\mathcal{B}. Then, for any subspace (Y,TY)(Y, \Tau_Y), the collection

C={BY:BB}\mathcal{C} = \{ B \cap Y : B \in \mathcal{B} \}

forms a basis for TY\Tau_Y.

Proof.

Let yYy \in Y. Then note some OTXO \in \Tau_X is such that yOXy \in O \subseteq X. But then OYO \cap Y is a basis element, so y(OY)OXy \in (O \cap Y) \subseteq O \subseteq X. Thus C\mathcal{C} covers YY and locally refines every element, making it a basis.

Note that if XX is a topological space and YXY \subseteq X is endowed with the subspace topology, then any open subset in YY is simply the intersection of some open set in XX with YY.

[2.5.3]Corollary(Subspace Topology is a Subset of Parent Topology for Open Subspaces)#

Suppose (Y,TY)(Y, \Tau_Y) is an open subspace of (X,TX)(X, \Tau_X). Then TYTX\Tau_Y \subseteq \Tau_X. Equivalently, all open sets in YY are open in XX.

Proof.

Suppose OTYO \in \Tau_Y. Then OO is an arbitrary union of basis elements of TY\Tau_Y. Let B\mathcal{B} be a basis for TX\Tau_X. Then we may write OO as the arbitrary union of basis elements, specifically with the basis induced by B\mathcal{B}. Then

O=λΛYUλ=YTX(λΛUλ)TXTX,O = \bigcup_{\lambda \in \Lambda} Y \cap U_\lambda = \underbrace{Y}_{\in \Tau_X} \cap \underbrace{\left(\bigcup_{\lambda \in \Lambda} U_\lambda \right)}_{\in \Tau_X} \in \Tau_X,

where {Uλ}λΛ\{U_\lambda\}_{\lambda \in \Lambda} is a set of open sets in XX. Then the arbitrary union of each UλU_\lambda is open in XX, and since YY is open as well, the intersection of this arbitrary union with YY is open as well. Thus OO is open in XX, meaning TYTX\Tau_Y \subseteq \Tau_X.

[2.5.4]Theorem(Product Topology and Subspace Topology Coincide)#

Suppose X,YX, Y are topological spaces with subspaces UXU \subseteq X and VYV \subseteq Y, endowed with the subspace topology. Then the subspace topology on U×VX×YU \times V \subseteq X \times Y is exactly the product topology.

Proof.

Let TX,TY,TU,TV\Tau_X, \Tau_Y, \Tau_U, \Tau_V be topologies endowed on the respective sets. Let TS\Tau_S be the subspace topology for U×VX×YU \times V \subseteq X \times Y, and let TU×V\Tau_{U \times V} be the product topology on U×VU \times V. Now define these collections of sets as bases for the topologies:

BS={(OX×OY)(U×V):OXTX,OYTY},\mathcal{B}_{S} = \{ (O_X \times O_Y) \cap (U \times V) : O_X \in \Tau_X, O_Y \in \Tau_Y \},
BU×V={(OU×OV):OUTU,OVTV}.\mathcal{B}_{U \times V} = \{ (O_U \times O_V) : O_U \in \Tau_U, O_V \in \Tau_V \}.

Let OSTSO_S \in \Tau_S. Then note OSO_S is the arbitrary union of {BSλ}λΛ\{B_{S_\lambda}\}_{\lambda \in \Lambda}, and so

OS=λΛ(OXλ×OYλ)(U×V)=λΛ(OXλU)TU×(OYλV)TVTU×VTU×V.O_S = \bigcup_{\lambda \in \Lambda} (O_{X_\lambda} \times O_{Y_\lambda}) \cap (U \times V) = \bigcup_{\lambda \in \Lambda} \underbrace{ \underbrace{(O_{X_\lambda} \cap U)}_{\in \Tau_U} \times \underbrace{(O_{Y_\lambda} \cap V)}_{\in \Tau_V}}_{\in \Tau_{U \times V}} \in \Tau_{U \times V}.

Thus TSTU×V\Tau_S \subseteq \Tau_{U \times V}. Conversely, suppose OU×VTU×VO_{U \times V} \in \Tau_{U \times V}. Then

OU×V=λΛ(OUλ×OVλ)=(λΛ(OXλ×OYλ))(U×V)TS,O_{U \times V} = \bigcup_{\lambda \in \Lambda} (O_{U_\lambda} \times O_{V_\lambda}) = \left( \bigcup_{\lambda \in \Lambda} (O_{X_\lambda} \times O_{Y_\lambda}) \right) \cap (U \times V) \in \Tau_{S},

and so TSTU×V\Tau_S \supseteq \Tau_{U \times V}, completing the proof.

The product topology is nice in this regard, but note the order topology is not. That is, if (X,<)(X, <) is an ordered set and (Y,<)(Y, <) is a subspace with the same ordering, the order topology is not necessarily the inherited subspace topology. We provide an example where the subspace and order topology do and don't coincide.

[2.5.5]Example(Subspace Topology vs. Order Topology)#

For our first example, let Y1=[0,1]Y_1 = [0, 1] be a subset of R\bR. Consider its subspace and order topologies. The basis for the subspace topology takes the form

BS={(a,b)Y1:a,bR}={(a,b)a,bY1,[0,b)bY1,(a,1]aY1,Y or otherwise.\mathcal{B}_S = \{(a, b) \cap Y_1 : a, b \in \bR\} = \begin{cases} (a, b) & a, b \in Y_1, \\ [0, b) & b \in Y_1, \\ (a, 1] & a \in Y_1, \\ Y \text{ or } \emptyset & \text{otherwise.} \end{cases}

Note this coincides exactly with the basis for the order topology (excluding Y,Y, \emptyset, but they are redundant when generating the topology). Thus the subspace and order topologies are equivalent.

Now let Y2=[0,1){2}Y_2 = [0, 1) \cup \{2\}. The basis for the subspace topology takes the form

BS={(a,b)([0,1){2}):a,bR}={[(a,b)[0,1)][(a,b){2}]:a,bR}.\mathcal{B}_S = \{(a, b) \cap ([0, 1) \cup \{2\}): a, b \in \bR\} = \{[(a, b) \cap [0, 1)] \cup [(a, b) \cap \{2\}] : a, b \in \bR \}.

Namely, note that if we choose (a,b)=(1.5,2.5)(a, b) = (1.5, 2.5), we have that (a,b)Y={2}BS(a, b) \cap Y = \{2\} \in \mathcal{B}_S, and so it is a member of the subspace topology. But note that any element of the order topology that contains 22 must either be [0,1){2}[0, 1) \cup \{2\} or (a,1){2}(a, 1) \cup \{2\}, where 0<a<10 < a < 1. In any case, {2}\{2\} is not in the order topology, meaning the order topology and subspace topology do not agree for Y2Y_2.

[2.5.6]Definition(Convex Set)#

Suppose (X,<)(X, <) is an ordered set and (Y,<)(X,<)(Y, <) \subseteq (X, <). We say YY is convex in XX if, for any a<bYa < b \in Y, the open interval (a,b)XY(a, b)_X \subseteq Y.

For example, [0,1][0, 1] is a convex subset of R\bR, but [0,1){2}[0, 1) \cup \{2\} is not (since (1,2)R(1, 2)_\bR contains elements like 1.51.5).

[2.5.7]Theorem(Order Topology and Subspace Topology Coincide for Convex Subsets)#

Suppose (X,<)(X, <) is an topological space endowed with the order topology. Let (Y,<)(Y, <) be a subspace of XX, where YXY \subseteq X is convex. Then the subspace topology and order topology coincide.

Proof.

Without loss of generality, suppose YY has a maximum and minimum. Then a basis element for the subspace topology on YY takes the form

x(BS={(a,b)Y:a,bX})={(a,b)a,bY,[min(Y),b)bY,(a,max(Y)]aY,Y or otherwise,x \in (\mathcal{B}_S = \{(a, b) \cap Y : a, b \in X\}) = \begin{cases} (a, b) & a, b \in Y, \\ [\min(Y), b) & b \in Y, \\ (a, \max(Y)] & a \in Y, \\ Y \text{ or } \emptyset & \text{otherwise,} \end{cases}

which is exactly the order topology. Of course, note that the second and third forms of basis elements are removed if YY does not have a minimum or maximum, respectively. Nevertheless, this coincides perfectly with the order topology on YY, completing the proof.

2.5.1Exercises#

[2.5.8]Problem#

Suppose YY is a subspace of XX and AYA \subseteq Y. Show that the subspace topology AA inherits from YY is the same as the subspace topology AA inherits from XX.

Proof.

Let TST\mathcal{T}_{ST} denote the subspace topology endowed on set SS as a subspace of TT. Let TX\Tau_X be the topology on XX. We show TAYTAXTAY\mathcal{T}_{AY} \subseteq \mathcal{T}_{AX} \subseteq \mathcal{T}_{AY}.

First, suppose OTAYO \in \mathcal{T}_{AY}. Then note there exist a family of open sets {OYλ}λΛTYX\{O_{Y_\lambda}\}_{\lambda \in \Lambda} \subseteq \mathcal{T}_{YX} such that

O=λΛ(OYλA)=A(λΛ(OYλ))=A(λΛ(OXλY))=(AY)A(λΛOXλ)TAX.O = \bigcup_{\lambda \in \Lambda} (O_{Y_\lambda} \cap A) = A \cap \left( \bigcup_{\lambda \in \Lambda} (O_{Y_\lambda}) \right) = A \cap \left( \bigcup_{\lambda \in \Lambda} (O_{X_\lambda} \cap Y) \right) = \overset{A}{\cancel{(A \cap Y)}} \cap \left( \bigcup_{\lambda \in \Lambda} O_{X_\lambda} \right) \in \Tau_{AX}.

Conversely, let OTAXO \in \Tau_{AX}. Then note there exist a family of open sets {OYλ}λΛTX\{O_{Y_\lambda}\}_{\lambda \in \Lambda} \subseteq \mathcal{T}_{X} such that

O=λΛ(OXλA)=A(λΛ(OXλ))=(AY)(λΛ(OXλ))=A(λΛ(OXλY))TAY,O = \bigcup_{\lambda \in \Lambda} (O_{X_\lambda} \cap A) = A \cap \left( \bigcup_{\lambda \in \Lambda} (O_{X_\lambda}) \right) = (A \cap Y) \cap \left( \bigcup_{\lambda \in \Lambda} (O_{X_\lambda}) \right) = A \cap \left( \bigcup_{\lambda \in \Lambda} (O_{X_\lambda} \cap Y) \right) \in \Tau_{AY},

completing the proof.

[2.5.9]Corollary#

Suppose f:XYf: X \to Y is a function of sets, and Λ\Lambda is an index set. Show that

f(λΛAλ)=λΛf(Aλ).f\left( \bigcup_{\lambda \in \Lambda} A_\lambda \right) = \bigcup_{\lambda \in \Lambda} f(A_\lambda).
Proof.

Suppose yf(λΛAλ)y \in f\left( \bigcup_{\lambda \in \Lambda} A_\lambda \right). Then there is some μΛ\mu \in \Lambda such that xAμx \in A_\mu and y=f(x)y = f(x). Then, of course, yf(Aμ)λΛf(Aλ)y \in f(A_\mu) \subseteq \bigcup_{\lambda \in \Lambda} f(A_\lambda). Conversely, suppose yλΛf(Aλ)y \in \bigcup_{\lambda \in \Lambda} f(A_\lambda). There is some μΛ\mu \in \Lambda such that there is some xAμx \in A_\mu such that y=f(x)f(Aμ)f(λΛAλ)y = f(x) \in f(A_\mu) \subseteq f\left( \bigcup_{\lambda \in \Lambda} A_\lambda \right), completing the proof.

[2.5.10]Problem#

A map f:XYf: X \to Y is said to be an open map if, for every open set UXU \subseteq X, the image f(U)f(U) is open in YY. Show πX:X×YX\pi_X: X \times Y \to X and πY:X×YY\pi_Y: X \times Y \to Y are open maps.

Solution.

Let UU be an open subset of X×YX \times Y. Then note there exists an indexing set Λ\Lambda such that U=λΛ(OXλ×OYλ)U = \bigcup_{\lambda \in \Lambda} (O_{X_\lambda} \times O_{Y_\lambda}). Then

πX(U)=πX(λΛ(OXλ×OYλ))=λΛπX(OXλ×OYλ)=λΛOXλ,\pi_X(U) = \pi_X\left(\bigcup_{\lambda \in \Lambda} (O_{X_\lambda} \times O_{Y_\lambda})\right) = \bigcup_{\lambda \in \Lambda} \pi_X(O_{X_\lambda} \times O_{Y_\lambda}) = \bigcup_{\lambda \in \Lambda} O_{X_\lambda},

which is open in XX. An analogous procedure can be applied for πY\pi_Y.

[2.5.11]Problem#

Show that

B={(a,b)×(c,d):a<b,c<d and a,b,c,dQ}\mathcal{B} = \{(a, b) \times (c, d): a < b, c < d \text{ and } a, b, c, d \in \bQ\}

is a basis for the standard topology on R2\bR^2.

Solution.

Let OR2O \subseteq \bR^2 be open, and let xOx \in O, where x=(x1,x2)x = (x_1, x_2). Since OO is the arbitrary union of open boxes, there is some a<b,c<dRa' < b', c' < d' \in \bR such that x[(a,b)×(c,d)]Ox \in [(a', b') \times (c', d')] \subseteq O. This means a<x1<ba' < x_1 < b' and c<x2<dc' < x_2 < d'. By the density of the rationals, there exist a,b,c,dQa, b, c, d \in \bQ such that

a<a<x1<b<b,a' < a < x_1 < b < b',
c<c<x2<d<d.c' < c < x_2 < d < d'.

Then x[(a,b)×(c,d)][(a,b)×(c,d)]Ox \in [(a, b) \times (c, d)] \subseteq [(a', b') \times (c', d')] \subseteq O, meaning B\mathcal{B} covers R2\bR^2 and locally refines each open set, completing the proof.