Polish Bitopological Spaces

1. Introduction

A bitopological space is a set equipped with two topologies. The systematic study of such spaces was initiated by Kelly, who developed a theory in which the interaction between the two topologies replaces the role played by symmetry in ordinary metric topology [1]. In particular, pairwise separation, pairwise regularity, pairwise normality, and quasi-metrizability provide natural bitopological analogues of classical topological concepts.

On the other hand, Polish spaces occupy a central position in modern descriptive set theory. A Polish space is a separable topological space admitting a compatible complete metric. Standard examples include ${\mathbb{R}}^n$, the Cantor space, the Baire space, and separable Banach spaces [2,3]. Polish spaces are especially important because their Borel, analytic, and coanalytic structures have strong regularity properties.

The purpose of this paper is to combine these two theories. We study a bitopological space $(X,{\tau }_1,{\tau }_2)$ when both $(X,{\tau }_1)$ and $(X,{\tau }_2)$ are Polish.

The motivation is not merely terminological. In many natural bitopological constructions the two topologies encode two opposite directions of approximation, two asymmetric notions of convergence, or two semicontinuity structures. Kelly already observed that a quasi-metric and its conjugate naturally generate two topologies and recover important symmetric phenomena at the bitopological level [1]. Moreover, Ciesielski, Flagg, and Kopperman showed that Polish spaces have a natural connection with bitopological structures arising from computational and domain-theoretic considerations [4].

We introduce a basic notion of Polish bitopological space and establish several elementary permanence properties. We also distinguish the statement that both component topologies are Polish from the stronger statement that their join is Polish. This distinction is important: componentwise Polishness does not automatically imply that the topology generated jointly by the two topologies has the desired Polish properties.

2. Preliminaries

2.1. Bitopological Spaces

We begin with the standard definition.

Definition 1.  

Abitopological spaceis a triple $(X,{\tau }_1,{\tau }_2)$, where X is a nonempty set and ${\tau }_1,{\tau }_2$ are topologies on X. The pair $({\tau }_1,{\tau }_2)$ is called abitopologyon X.

For $i\in 1,2$, we write ${cl}_i\left(A\right)$ and ${int}_i\left(A\right)$ for the closure and interior of $A\subseteq X$ with respect to ${\tau }_i$.

The join of ${\tau }_1$ and ${\tau }_2$ is denoted by ${\tau }_1\vee {\tau }_2$.

It is the smallest topology containing both ${\tau }_1$ and ${\tau }_2$.

The intersection ${\tau }_1\cap {\tau }_2$ is also a topology on X and is the largest topology contained in both ${\tau }_1$ and ${\tau }_2$.

Definition 2.  

A bitopological space $(X,{\tau }_1,{\tau }_2)$ is calledpairwise Hausdorffif for every distinct $x,y\in X$ there exist $U\in {\tau }_1$ and $V\in {\tau }_2$ such that

$$x\in U,y\in V,U\cap V=⌀,$$

and the same condition holds after interchanging ${\tau }_1$ and ${\tau }_2$.

Pairwise separation axioms are among the characteristic features of bitopological spaces and should not be confused with the ordinary Hausdorff property of either individual topology [1,5].

2.2. Polish Spaces

We recall the classical definition.

Definition 3.  

A topological space $(X,\tau )$ is calledPolishif it is separable and admits a compatible complete metric.

Equivalently, there exists a complete metric d on X such that $\tau ={\tau }_d$ and X has a countable dense subset. This is the standard definition used in descriptive set theory [2,3].

Important permanence properties of Polish spaces include the following: closed subspaces of Polish spaces are Polish, countable products of Polish spaces are Polish, and countable disjoint unions of Polish spaces are Polish [2].

3. Polish Bitopological Spaces

We now introduce the main notion.

Definition 4.  

A bitopological space $(X,{\tau }_1,{\tau }_2)$ is called aPolish bitopological spaceif both $(X,{\tau }_1)$ and $(X,{\tau }_2)$ are Polish spaces.

Thus there exist complete compatible metrics $d_1$ and $d_2$ such that ${\tau }_1={\tau }_{d_1}$, ${\tau }_2={\tau }_{d_2}$, and both $(X,d_1)$ and $(X,d_2)$ are separable and complete.

Remark 1.  

The metrics $d_1$ and $d_2$ need not be equal and need not be topologically equivalent. In particular, a Polish bitopological space can encode two genuinely different Polish structures on the same underlying set.

Definition 5.  

A Polish bitopological space $(X,{\tau }_1,{\tau }_2)$ is calledjointly Polishif $(X,{\tau }_1\vee {\tau }_2)$ is Polish.

Clearly, joint Polishness implies that the join topology is Polish, but it does not follow merely from the definition that the join of two Polish topologies on the same set is Polish.

Definition 6.  

A Polish bitopological space $(X,{\tau }_1,{\tau }_2)$ is calledcommonly separableif there exists a countable set $D\subseteq X$ which is dense in both $(X,{\tau }_1)$ and $(X,{\tau }_2)$.

Proposition 1.  

Every second-countable bitopological space is commonly separable.

Proof. 

Let ${\mathcal{B}}_1$ and ${\mathcal{B}}_2$ be countable bases for ${\tau }_1$ and ${\tau }_2$, respectively. Choose one point from every nonempty member of ${\mathcal{B}}_1\cup {\mathcal{B}}_2$.

The resulting set is countable and intersects every nonempty basic open set in both topologies. Hence it is dense in both topologies. □

Corollary 1.  

Every Polish bitopological space is commonly separable.

Proof. 

Both component topologies are second countable because every Polish space is second countable. Apply the proposition. □

4. Borel Structures

For a topology $\tau$ on X, let $\mathcal{B}\left(\tau \right)$ denote the Borel $\sigma$-algebra generated by $\tau$.

Proposition 2.  

Given a bitopological space, there are therefore three naturally associated Borel structures:

$${\mathcal{B}}_1=\mathcal{B}\left({\tau }_1\right),{\mathcal{B}}_2=\mathcal{B}\left({\tau }_2\right),{\mathcal{B}}_{\vee }=\mathcal{B}({\tau }_1\vee {\tau }_2).$$

Since ${\tau }_1,{\tau }_2\subseteq {\tau }_1\vee {\tau }_2$, we have

$${\mathcal{B}}_1,{\mathcal{B}}_2\subseteq {\mathcal{B}}_{\vee }.$$

$$\sigma (\mathcal{B}\left({\tau }_1\right)\cup \mathcal{B}\left({\tau }_2\right)).$$

Proof. 

Since ${\tau }_1,{\tau }_2\subseteq {\tau }_1\vee {\tau }_2$, we have

$$\sigma (\mathcal{B}\left({\tau }_1\right)\cup \mathcal{B}\left({\tau }_2\right))\subseteq \mathcal{B}({\tau }_1\vee {\tau }_2).$$

Conversely, the topology ${\tau }_1\vee {\tau }_2$ is generated by finite intersections of sets from ${\tau }_1\cup {\tau }_2$. Such finite intersections belong to

$$\sigma (\mathcal{B}\left({\tau }_1\right)\cup \mathcal{B}\left({\tau }_2\right)).$$

Therefore every ${\tau }_1\vee {\tau }_2$-open set belongs to this sigma-algebra, and hence

$$\mathcal{B}({\tau }_1\vee {\tau }_2)\subseteq \sigma (\mathcal{B}\left({\tau }_1\right)\cup \mathcal{B}\left({\tau }_2\right)).$$

□

This gives a useful interpretation of the Borel structure of the bitopological join.

Definition 7.  

A Polish bitopological space is calledBorel compatibleif $\mathcal{B}\left({\tau }_1\right)=\mathcal{B}\left({\tau }_2\right)$.

Proposition 3.  

If $(X,{\tau }_1,{\tau }_2)$ is Borel compatible, then $\mathcal{B}({\tau }_1\vee {\tau }_2)=\mathcal{B}\left({\tau }_1\right)=\mathcal{B}\left({\tau }_2\right)$.

Proof. 

Observe $\sigma (\mathcal{B}\left({\tau }_1\right)\cup \mathcal{B}\left({\tau }_2\right))$. If $\mathcal{B}\left({\tau }_1\right)=\mathcal{B}\left({\tau }_2\right)$, this sigma-algebra is equal to the common Borel sigma-algebra. □

Thus two distinct Polish topologies can have exactly the same measurable structure.

5. A Fundamental Example from Quasi-Metrics

Quasi-metrics provide one of the most natural sources of bitopological spaces. Kelly’s work showed that an asymmetric distance can be used to construct two topologies which restore a degree of symmetry at the bitopological level [1].

Definition 8.  

A function $q:X\times X⟶[0,\infty )$ is aquasi-metricif $q(x,x)=0$ and $q(x,z)\le q(x,y)+q(y,z)$ for all $x,y,z\in X$. If $q(x,y)=q(y,x)=0$ implies $x=y$, then q is called a $T_0$ quasi-metric.

The conjugate quasi-metric is $q^{-1}(x,y)=q(y,x)$.

Definition 9.  

The bitopological space $(X,{\tau }_q,{\tau }_{q^{-1}})$ is called thequasi-metric bitopologyinduced by q.

Theorem 1.  

Let q be a quasi-metric on X. If both $(X,{\tau }_q)$ and $(X,{\tau }_{q^{-1}})$ are Polish, then $(X,{\tau }_q,{\tau }_{q^{-1}})$ is a Polish bitopological space.

Proof. 

This follows immediately from the definition of a Polish bitopological space. □

The theorem is elementary, but its importance is structural: it provides a large class of examples in which the two Polish structures are generated by a single asymmetric object.

6. A Concrete Example

Consider $\mathbb{R}$ with its usual topology ${\tau }_u$ and the lower-limit topology ${\tau }_{\ell }$ generated by half-open intervals $[a,b)$.

The lower-limit topology is generated by the metric

$$d_{\ell }(x,y)=\left\{\begin{array}{cc}y-x,\hfill & x\le y,\hfill \\ 1,\hfill & x>y,\hfill \end{array}\right.$$

after a standard bounded modification if necessary. The corresponding asymmetric structure can also be viewed through a quasi-metric.

The Sorgenfrey line is separable and completely metrizable with respect to a suitable compatible metric only when interpreted carefully; in particular, one must distinguish the topology from the usual Euclidean metric. Consequently, this example illustrates why componentwise verification is essential.

For a genuinely Polish example, let $X=\mathbb{R}$ and choose two compatible Polish metrics $d_1,d_2$ generating two possibly different Polish topologies ${\tau }_1,{\tau }_2$.

Then $(\mathbb{R},{\tau }_1,{\tau }_2)$ is a Polish bitopological space. If the two topologies are generated by different complete metrics but define the same topology, the bitopological structure is nontrivial as a pair of metrics but topologically diagonal.

7. Pairwise Separation

The interaction between the two topologies is more interesting when they are genuinely distinct.

Definition 10.  

A bitopological space $(X,{\tau }_1,{\tau }_2)$ ispairwise regularif, for $i\ne j$, whenever F is ${\tau }_i$-closed and $x\notin F$, there exist $U\in {\tau }_i$ and $V\in {\tau }_j$ such that

$$x\in U,F\subseteq V,U\cap V=⌀.$$

Theorem 2.  

Let $(X,{\tau }_1,{\tau }_2)$ be a Polish bitopological space. Suppose that ${\tau }_1$ and ${\tau }_2$ satisfy ${\tau }_1\subseteq {\tau }_2$.

Then the identity map $id:(X,{\tau }_2)⟶(X,{\tau }_1)$ is continuous.

Proof. 

Let $U\in {\tau }_1$. Since ${\tau }_1\subseteq {\tau }_2$, we have $U\in {\tau }_2$. Therefore ${id}^{-1}\left(U\right)=U$ is ${\tau }_2$-open. Hence the identity map is continuous. □

Remark 2.  

The reverse identity $id:(X,{\tau }_1)⟶(X,{\tau }_2)$ need not be continuous. If it is also continuous, then ${\tau }_1={\tau }_2$.

Thus a Polish bitopological space can naturally encode a continuous asymmetric relationship between two Polish structures.

8. Baire Properties

Every Polish space is a Baire space. Therefore a Polish bitopological space is Baire in each of its component topologies.

Theorem 3.  

Let $(X,{\tau }_1,{\tau }_2)$ be a Polish bitopological space. Then $(X,{\tau }_1)$ and $(X,{\tau }_2)$ are Baire spaces.

Proof. 

Both component spaces are Polish. Every completely metrizable space is a Baire space. Hence both component spaces are Baire. □

This permits a pairwise notion of residuality.

Definition 11.  

A subset $A\subseteq X$ is calledpairwise residualif it is residual with respect to both ${\tau }_1$ and ${\tau }_2$.

Proposition 4.  

If A and B are pairwise residual subsets of a Polish bitopological space, then $A\cap B$ is pairwise residual.

Proof. 

Since A and B are residual in $(X,{\tau }_i)$ for each $i=1,2$, their intersection is residual in each Baire space $(X,{\tau }_i)$. Hence $A\cap B$ is pairwise residual. □

9. Closed Subspaces

The usual permanence theorem for Polish spaces immediately yields a bitopological version.

Theorem 4.  

Let $(X,{\tau }_1,{\tau }_2)$ be a Polish bitopological space, and let $F\subseteq X$ be closed in both ${\tau }_1$ and ${\tau }_2$. Then $(F,{\tau }_1{|}_F,{\tau }_2{|}_F)$ is a Polish bitopological space.

Proof. 

Since F is ${\tau }_i$-closed and $(X,{\tau }_i)$ is Polish, the subspace $(F,{\tau }_i{|}_F)$ is Polish for $i=1,2$. Hence both component topologies on F are Polish. □

The condition that F be closed in both topologies is sufficient but not necessary. One topology may have F closed while the other recognizes F as a $G_{\delta }$ set.

Proposition 5.  

Let $(X,{\tau }_1,{\tau }_2)$ be a Polish bitopological space. If $F\subseteq X$ is a $G_{\delta }$ subset in both ${\tau }_1$ and ${\tau }_2$, then $(F,{\tau }_1{|}_F,{\tau }_2{|}_F)$ is Polish bitopological.

Proof. 

A $G_{\delta }$ subspace of a Polish space is Polish. Apply this theorem to $(X,{\tau }_1)$ and $(X,{\tau }_2)$ separately. □

10. Products

Let

$$(X,{\tau }_1,{\tau }_2)\mathrm{and}(Y,{\Sigma }_1,{\Sigma }_2)$$

be Polish bitopological spaces.

Define the product bitopology by

$${\tau }_1\times {\Sigma }_1\mathrm{and}{\tau }_2\times {\Sigma }_2.$$

Theorem 5.  

The product

$$(X\times Y,{\tau }_1\times {\Sigma }_1,{\tau }_2\times {\Sigma }_2)$$

is a Polish bitopological space.

Proof. 

The product of two Polish spaces is Polish. Therefore $(X\times Y,{\tau }_1\times {\Sigma }_1)$ and $(X\times Y,{\tau }_2\times {\Sigma }_2)$ are both Polish. Hence the resulting bitopological space is Polish. □

The same argument gives the following.

Corollary 2.  

A countable product of Polish bitopological spaces is Polish bitopological when each component topology is equipped with the corresponding product topology.

11. Continuous and Borel Maps

Let

$$(X,{\tau }_1,{\tau }_2)\mathrm{and}(Y,{\Sigma }_1,{\Sigma }_2)$$

be bitopological spaces.

Definition 12.  

A map $f:X⟶Y$ is calledpairwise continuousif $f:(X,{\tau }_i)⟶(Y,{\Sigma }_i)$ is continuous for $i=1,2$.

Proposition 6.  

Every pairwise continuous map between Polish bitopological spaces is Borel measurable with respect to each component Borel structure.

Proof. 

For each $i=1,2$, the map $f:(X,{\tau }_i)⟶(Y,{\Sigma }_i)$ is continuous. Continuous maps between Polish spaces are Borel measurable. Hence $f^{-1}\left(B\right)\in \mathcal{B}\left({\tau }_i\right)$ for every $B\in \mathcal{B}\left({\Sigma }_i\right)$. □

Definition 13.  

A map $f:(X,{\tau }_1,{\tau }_2)⟶(Y,{\Sigma }_1,{\Sigma }_2)$ is calledpairwise Borelif $f:(X,\mathcal{B}\left({\tau }_i\right))⟶(Y,\mathcal{B}\left({\Sigma }_i\right))$ is Borel measurable for $i=1,2$.

Thus

$$\mathrm{pairwise}\mathrm{continuous}⟹\mathrm{pairwise}\mathrm{Borel}.$$

The converse generally fails, just as it fails for ordinary Polish spaces.

12. Borel Compatibility

The two component Borel structures need not be different even when the topologies are different.

Theorem 6.  

Suppose $(X,{\tau }_1,{\tau }_2)$ is a Polish bitopological space and

$$\mathcal{B}\left({\tau }_1\right)=\mathcal{B}\left({\tau }_2\right).$$

Then the identity maps $id:(X,{\tau }_1)\to (X,{\tau }_2)$ and $id:(X,{\tau }_2)\to (X,{\tau }_1)$ are Borel isomorphisms.

Proof. 

The identity map is Borel in the first direction because for every ${\tau }_2$-Borel set B, ${id}^{-1}\left(B\right)=B\in \mathcal{B}\left({\tau }_1\right)$.

The same argument applies in the reverse direction. Both maps are therefore Borel measurable and are inverse to one another. □

Remark 3.  

This result shows that Polish bitopology can carry information that is invisible at the level of Borel sets. Two different Polish topologies may determine the same standard Borel space.

This observation is particularly relevant in descriptive set theory, where the Borel structure rather than the precise topology is often the primary measurable object [2,3].

13. Common Polish Refinements

A natural question is when a Polish bitopological space admits a single Polish topology refining both component topologies.

Definition 14.  

A topology τ on X is called acommon Polish refinementof ${\tau }_1$ and ${\tau }_2$ if ${\tau }_1\cup {\tau }_2\subseteq \tau$ and $(X,\tau )$ is Polish.

The strongest canonical candidate is the join ${\tau }_1\vee {\tau }_2$.

Proposition 7.  

If ${\tau }_1\vee {\tau }_2$ is second countable and admits a complete compatible metric, then $(X,{\tau }_1\vee {\tau }_2)$ is Polish.

Proof. 

Second countability implies separability for a metrizable space. Together with complete metrizability, this gives Polishness. □

Remark 4.  

The preceding proposition separates the two issues that arise in joint Polishness: separability and complete metrizability. Even though each component topology is Polish, one must verify these properties for the join separately.

14. A Criterion for Joint Polishness

We give a useful sufficient condition.

Theorem 7.  

Let $(X,{\tau }_1,{\tau }_2)$ be a Polish bitopological space. Suppose that there exists a complete metric d on X such that ${\tau }_1\vee {\tau }_2={\tau }_d$. Then $(X,{\tau }_1,{\tau }_2)$ is jointly Polish.

Proof. 

The topology ${\tau }_1\vee {\tau }_2$ is induced by the complete metric d. Since both ${\tau }_1$ and ${\tau }_2$ are second countable, the join generated by countably many bases is second countable. Consequently $(X,{\tau }_1\vee {\tau }_2)$ is separable and completely metrizable, hence Polish. □

Remark 5.  

The hypothesis is substantially stronger than componentwise Polishness. It explicitly requires a complete metric for the common refinement.

15. Polish Bitopological Groups

The theory becomes particularly interesting for algebraic structures.

Definition 15.  

APolish bitopological groupis a group G equipped with two Polish topologies ${\tau }_1,{\tau }_2$ such that the relevant group operations are compatible with the two topological structures.

A particularly strong condition is the following.

Definition 16.  

A bitopological group $(G,{\tau }_1,{\tau }_2)$ is calledstrongly pairwise Polishif both topological groups

$$(G,{\tau }_1)\mathrm{and}(G,{\tau }_2)$$

are Polish topological groups and inversion identifies the two topologies in the sense that $x\mapsto x^{-1}$ is a homeomorphism $(G,{\tau }_1)⟶(G,{\tau }_2)$.

This situation occurs naturally when the two topologies arise from opposite quasi-metric structures.

Proposition 8.  

Let G be a group with two Polish group topologies ${\tau }_1$ and ${\tau }_2$. If inversion is a homeomorphism $(G,{\tau }_1)⟶(G,{\tau }_2)$, then the two spaces have isomorphic Borel structures.

Proof. 

Inversion is a homeomorphism and hence a Borel isomorphism between the two Polish spaces. □

16. Relation with Descriptive Set Theory

The theory of Polish spaces is fundamental in descriptive set theory. In particular, analytic and coanalytic subsets are defined naturally inside Polish spaces [2,3].

For a Polish bitopological space we obtain two descriptive structures: $(X,{\tau }_1)$ and $(X,{\tau }_2)$. This suggests the study of subsets that have different descriptive complexities relative to the two topologies.

Definition 17.  

A subset $A\subseteq X$ is calledbi-Borelif $A\in \mathcal{B}\left({\tau }_1\right)\cap \mathcal{B}\left({\tau }_2\right)$. It is calledbi-analyticif it is analytic in both Polish spaces $(X,{\tau }_1)$ and $(X,{\tau }_2)$.

If $\mathcal{B}\left({\tau }_1\right)=\mathcal{B}\left({\tau }_2\right)$, then every Borel set is automatically bi-Borel.

17. Pairwise Zero-Dimensional Polish Spaces

A Polish space is zero-dimensional if it possesses a basis consisting of clopen sets. This leads to a natural bitopological analogue.

Definition 18.  

A Polish bitopological space $(X,{\tau }_1,{\tau }_2)$ is calledpairwise zero-dimensionalif both $(X,{\tau }_1)$ and $(X,{\tau }_2)$ are zero-dimensional Polish spaces.

Proposition 9.  

Every countable discrete bitopological space $(X,\mathcal{P}(X),\mathcal{P}(X\left)\right)$ is pairwise zero-dimensional Polish.

Proof. 

A countable discrete space is Polish under the discrete metric, and every subset is clopen. □

The Cantor space provides a canonical uncountable example. Let $\mathcal{C}=2^{\mathbb{N}}$ with its standard product topology. If two Polish zero-dimensional topologies are placed on the same underlying Cantor set, the resulting object is a natural candidate for a zero-dimensional Polish bitopological space.

18. Pairwise G_δ Sets

The $G_{\delta }$ property is particularly useful because it characterizes Polish subspaces of Polish spaces.

Definition 19.  

A subset $A\subseteq X$ is calledpairwise $G_{\delta }$if

$$A\in {\Pi }_2^0\left({\tau }_1\right)\cap {\Pi }_2^0\left({\tau }_2\right).$$

Theorem 8.  

Let $(X,{\tau }_1,{\tau }_2)$ be Polish bitopological and let $A\subseteq X$ be pairwise $G_{\delta }$. Then $(A,{\tau }_1{|}_A,{\tau }_2{|}_A)$ is Polish bitopological.

Proof. 

Since A is $G_{\delta }$ in each Polish space $(X,{\tau }_i)$, the subspace $(A,{\tau }_i{|}_A)$ is Polish for $i=1,2$. Hence the induced bitopological space is Polish bitopological. □

This gives a convenient descriptive-set-theoretic method for producing new Polish bitopological spaces.

19. Maps Preserving the Two Structures

We can strengthen pairwise continuity by requiring compatibility with the two topologies in opposite directions.

Definition 20.  

Let $f:(X,{\tau }_1,{\tau }_2)⟶(Y,{\Sigma }_1,{\Sigma }_2)$. We say that f iscross-continuousif $f:(X,{\tau }_1)\to (Y,{\Sigma }_2)$ and $f:(X,{\tau }_2)\to (Y,{\Sigma }_1)$ are continuous.

Cross-continuity is particularly natural in applications where ${\tau }_1$ and ${\tau }_2$ represent dual or opposite approximation structures.

Proposition 10.  

Every cross-continuous map between Polish bitopological spaces is Borel measurable from each component Borel structure into both target Borel structures.

Proof. 

Continuity in each of the two cross directions implies Borel measurability in each corresponding direction because the spaces are Polish. Thus for every $B\in \mathcal{B}\left({\Sigma }_1\right)\cup \mathcal{B}\left({\Sigma }_2\right)$, the appropriate inverse image belongs to the corresponding source Borel structure. □

20. Countable Bases and Bitopological Metrization

A Polish topology possesses a countable base. Consequently, a Polish bitopological space admits two countable bases ${\mathcal{B}}_1$, ${\mathcal{B}}_2$.

Proposition 11.  

Every Polish bitopological space has a countable family ${\mathcal{B}}_1\cup {\mathcal{B}}_2$ which generates the entire bitopological structure.

Proof. 

Choose countable bases ${\mathcal{B}}_1$ and ${\mathcal{B}}_2$ for the two Polish topologies. Their union generates both topologies and hence the bitopological structure. □

This observation suggests a bitopological version of classical metrization theory. One may ask whether the two metrics can be combined into a single metric or quasi-metric that simultaneously encodes both topologies.

A simple construction is $d(x,y)=d_1(x,y)+d_2(x,y)$.

This metric generates a topology finer than both ${\tau }_1$ and ${\tau }_2$. However, completeness of this sum metric requires separate verification and cannot simply be inferred from the completeness of $d_1$ and $d_2$ without further argument.

21. A Common Metric Construction

The preceding observation leads to the following useful result.

Theorem 9.  

Let $d_1$ and $d_2$ be compatible complete metrics for ${\tau }_1$ and ${\tau }_2$, respectively. Suppose that the identity maps

$$(X,d_1)⟶(X,d_2)\mathrm{and}(X,d_2)⟶(X,d_1)$$

are uniformly continuous. Then $d=d_1+d_2$ is a complete metric generating the join topology ${\tau }_1\vee {\tau }_2$. It follows that $(X,{\tau }_1,{\tau }_2)$ is jointly Polish.

Proof. 

The metric $d=d_1+d_2$ generates the join topology because a d-ball contains a suitable intersection of a $d_1$-ball and a $d_2$-ball, while every sufficiently small d-ball is contained in both component balls.

Now let $\left(x_n\right)$ be a d-Cauchy sequence. Since

$$d_1(x_n,x_m)\le d(x_n,x_m)$$

and

$$d_2(x_n,x_m)\le d(x_n,x_m),$$

the sequence is Cauchy in both $d_1$ and $d_2$. Completeness gives points $x_1,x_2\in X$ such that

$$x_n\to x_1\mathrm{in}{d}_1$$

and

$$x_n\to x_2\mathrm{in}{d}_2.$$

The assumed uniform compatibility implies that the two limits agree. Hence there exists $x\in X$ such that $d(x_n,x)\to 0$. Thus d is complete. Since both component spaces are separable, their join is second countable under the stated compatibility assumptions. Hence the join is Polish. □

22. Comparison with Classical Polish Spaces

The diagonal construction provides the simplest examples.

Example 1.  

Let $(X,\tau )$ be any Polish space. Then $(X,\tau ,\tau )$ is a Polish bitopological space.

More interestingly, suppose ${\tau }_1$ and ${\tau }_2$ are distinct Polish topologies on the same set X. Then $(X,{\tau }_1,{\tau }_2)$ contains information not present in either topology alone.

For example, one may have ${\tau }_1⊊{\tau }_2$, in which case the identity map $(X,{\tau }_2)\to (X,{\tau }_1)$ is continuous, while the reverse identity need not be continuous.

This asymmetric behavior is one of the main reasons for studying bitopological structures rather than replacing them immediately by a single topology.

23. Connection with Existing Bitopological Theory

The framework developed here is compatible with the classical theory initiated by Kelly. In particular, Kelly’s treatment of bitopological spaces shows that quasi-metric constructions can be expressed through two associated topologies and that classical metrization phenomena have bitopological analogues [1].

The relationship with Polish spaces has also appeared in the context of computational models. Ciesielski, Flagg, and Kopperman constructed bitopological structures associated with Polish spaces and showed that certain compact subtopologies can be used to obtain pairwise regular bitopological spaces [4]. Their work demonstrates that the interaction between Polish topology and bitopology is not merely formal.

The present approach differs in emphasis: rather than starting with a single Polish topology and constructing a secondary compact or computational topology, we begin with two Polish topologies and study the resulting descriptive and pairwise structure.

24. Conclusions

We introduced the notion of a Polish bitopological space as a bitopological space $(X,{\tau }_1,{\tau }_2)$ for which both component spaces are Polish. This definition combines the countability and completeness properties of Polish spaces with the two-directional structure of bitopology.

Several basic properties were established. In particular, Polish bitopological spaces are commonly separable, Baire in each component topology, stable under suitable closed and $G_{\delta }$ subspaces, and stable under countable products. Pairwise continuous maps are automatically pairwise Borel, while equality of the two Borel structures makes the two component spaces Borel isomorphic.

The distinction between componentwise Polishness and joint Polishness is fundamental. The existence of two Polish topologies does not by itself provide a complete metric for the join topology. Additional compatibility assumptions, such as suitable uniform compatibility of complete metrics, can guarantee joint Polishness.

The framework therefore provides a natural setting for further interaction between bitopology, quasi-metric geometry, descriptive set theory, domain theory, and topological algebra.