©

Houston Journal of Mathematics

2021 University of Houston
Volume 47, Number 4, 2021, Pages 931–944

COMPLETENESS PROPERTIES IN TOPOLOGICAL SPACES

HAVING A PAIR-BASE

SÜLEYMAN ÖNAL AND ÇETIN VURAL

Communicated by Yasunao Hattori

Abstract. We introduce the completeness of a pair-base and study the

topological spaces having such a base. We investigate, among the others,
Čech-complete spaces and subcompact spaces have a complete pair-base,
and we prove that if a topological space X has a complete pair-base then
X is domain representable. We establish that a paracompact p-space X
must be Čech-complete if it has a countable complete pair-base. We also
show that the property of having a complete pair-base is preserving under
retractions.

1. Introduction and terminology

Throughout this paper, all spaces are assumed to be regular at least, and κ, µ
denote infinite cardinal numbers.
Main goal of this paper is to introduce certain completeness properties in topo-
logical spaces having a pair-base and investigate the properties of the topological
spaces having such a base and also investigate which spaces have such a base.
One of the most known completeness property is the completeness of metric
spaces and the other one being of a topological space is complete in the sense of
Čech, more widely known as Čech-complete space. A space is Čech-complete if
it is homeomorphic to a Gδ -subset of a compact space. It is well known that a
metrizable space X is completely metrizable if and only if X is Čech-complete.
One of the generalisations of completeness of metric spaces is subcompactness.
A space X is subcompact if it has a base B of nonempty open subsets with the

2010 Mathematics Subject Classification. 54D35; 54H11; 22A05; 22A10; 54B05.

Key words and phrases. Čech-complete, P -complete, pair-base, subcompact, domain
representable.
The authors acknowledge the help of Dr. Hasan Gül in the preparation of the manuscript.
931
932 SÜLEYMAN ÖNAL AND ÇETIN VURAL

property that every regular filter base U in B has nonempty intersection, and such
a base B is called subcompact base for X. Recall that a family U of nonempty
open sets is a regular filter base if for any U, V ∈ U there is W ∈ U such that
cl (W ) ⊆ U ∩ V. It is known that, for metrizable spaces, subcompactness is equiv-
alent to Čech-completeness. Another completeness property is the concept of
domain representability. Bennett and Lutzer, in [1], called a topological space
domain representable if it can be represented as the space of maximal element of
some continuous directed complete partial order, namely domain, with the Scott
topology and they proved that Čech-complete spaces are domain representable.
They also proved, in [2], that subcompact regular spaces are domain representable.
Then Fleissner and Yengulalp, in [6], gave a simplified characterization of domain
representability. The authors, in [11], have proposed to use the name P -complete
space instead this simplified characterization of domain representable space. Let
us recall the definition of P -complete space. It is said that the triple (P, , ϕ)
represents the topological space X if
(P 1) The family {ϕ(p) : p ∈ P } of nonempty open sets is a base for X,
(P 2)  is transitive relation on P,
(P 3) for all p, q ∈ P, p  q implies ϕ (p) ⊆ ϕ (q) ,
(P 4) for all x ∈ X, {p ∈ P : x ∈ ϕ (p)} is downward directed,
T
(P 5) if F ⊆ P and (F, ) is downward directed, then p∈F ϕ(p) 6= ∅.
Such a triple (P, , ϕ) is called a domain representation of X. A topological
space X is called P -complete if there exists a domain representation (P, , ϕ) of
X.
In [6], Fleissner and Yengulalp introduced the concept of generalized subcom-
pactness and then Yengulalp, in [13], proved that generalized subcompactness is
equivalent to domain representability. In [6, Definition 2.3], the definition of gen-
eralized subcompactness as follows: A space X is generalized subcompact if there
are B and ≺ satisfying
(G1) B is a family of nonempty open sets which is a base for X,
(G2) ≺ is a transitive, antisymmetric relation on B,
(G3) B ≺ B 0 implies B ⊆ B 0 ,
(G4) if x ∈ X, then {B ∈ B : x ∈ B} is downward directed by ≺, and
T
(G5) if F ⊆ B and (F, ≺) is downward directed, then F 6= ∅.
Such a base B is called a generalized subcompact base for X.
Although it is known that Čech-complete spaces are P -complete(i.e., domain
representable) but it is not known whether they are subcompact. Essentially, it is
 COMPLETENESS PROPERTIES IN TOPOLOGICAL SPACES 933

still an open question whether subcompactness is hereditary with respect to dense

Gδ -subspaces. In this paper we show that the property of having a complete pair-
base is preserved by dense Gδ -subspaces. Therefore we obtain that Čech-complete
spaces have a complete pair-base. We also prove that subcompact spaces have a
complete pair-base and we show that any base of the topological space having a
complete pair-base is a generalized subcompact base, so it is P -complete. Hence
the property of having a complete pair-base is weaker than the subcompactness
and it seems like stronger than the P -completeness. But, we do not know yet
whether every P -complete space has a complete pair-base.

We will use the customary terminology.

2. Pair-collections and completeness of them

We have realised that to use a pair-base instead of a base in some completeness
properties which called Amsterdam properties(see, for example [3]) is performing
well than to use a base.

P ? (X) is the set of all nonempty subsets of the topological space X. A pair-
collection P = {(A, B) : cl (A) ⊆ B} in X is a collection of subsets of P ? (X) ×
P ? (X) together with the partial order “(A, B)  (C, D) if and only if B ⊆ C”.

Definition 1. Let P be a pair-collection.

(1) A subset F of P is called a filter base in P if for any (A, B) , (C, D) ∈ F

there exists (E, F ) ∈ F such that (E, F )  (A, B) and (E, F )  (C, D) .
(2) A subset O of P is called an order-centered family in P if for any
(A, B) , (C, D) ∈ O there exists (E, F ) ∈ P such that (E, F )  (A, B)
and (E, F )  (C, D) .
(3) A subset H of P is said to have the µ-intersection property if
T
(A,B)∈G B 6= ∅ for every subfamily G of H with |G| < µ. For µ = ω0 , H
is called centered in P, and for µ = ω1 , H is said to have the countable
intersection property.

It is obvious that every filter base in P is order-centered and every order-

centered family in P is centered.

Definition 2. Let P be a pair-collection.

934 SÜLEYMAN ÖNAL AND ÇETIN VURAL

T
(1) It is said that P is κ-complete if (A,B)∈F B 6= ∅ for every filter base F
in P with |F| < κ. For κ = ω1 , P is called countably complete. If P is
κ-complete for all infinite cardinal κ, then P is called complete.
T
(2) It is said that P is (κ, µ)-complete if (A,B)∈F B 6= ∅ for every filter base
T
F in P with |F| < κ whenever (A,B)∈C B 6= ∅ for every filter base C in
F with |C| < µ. If P is (κ, ω1 )-complete for all infinite cardinal κ, then P
is called L-complete.
T
(3) It is said that P is κ-oc-complete if (A,B)∈O B 6= ∅ for every order-
centered family O in P with |O| < κ. If P is κ-oc-complete for all infinite
cardinal κ, then P is called oc-complete.
T
(4) It is said that P is µ-centered complete if (A,B)∈H B 6= ∅ for every
subfamily H of P having the µ-intersection property. For µ = ω0 , P is
called centered complete.

Although we have stated the above definitions in a broad sense, hereafter we

will use only the notions countably completeness, L-completeness, oc-completeness
and centered completeness for fluency. It is easily seen that we have the following
implications;
centered complete ⇒ oc-complete ⇒ complete ⇒ L-complete,
and we also have if P is countably complete and L-complete, then P is complete.
We will see that subcompact spaces have a complete pair-base. It is easy to
see that co-compact spaces have a centered complete pair-base. So, the spaces
satisfying one of the four basic Amsterdam properties(co-compactness, regular co-
compactness, base compactness and subcompactness) have a complete pair-base.
(Amsterdam properties can be found in [3].)

Definition 3. Let X be a topological space and P = {(A, B) : cl (A) ⊆ B} be a

pair-collection in X. P is called a quasi-pair-base for the space X if for every open
subset U of X and every x ∈ U there is (A, B) ∈ P such that x ∈ int (A) and
B ⊆ U. A quasi-pair-base P is called a pair-base for the space X if A’s are open.
Essentially, the property of having a complete pair-base and the property of
having a complete quasi-pair-base are the same. Moreover, we can say that this
is also true the other completeness properties in Definition 2. Indeed, if Q =
{(A, B) : cl (A) ⊆ B} is a quasi-pair-base for a topological space X, then the
family P = {(int(A), B) : (A, B) ∈ Q} is a pair-base for X and it satisfies the
completeness property that satisfied by Q.
 COMPLETENESS PROPERTIES IN TOPOLOGICAL SPACES 935

Let P and Q be pair-collections. The composition of P and Q is defined by

P ◦ Q = {(A, H) : ∃B, G, (A, B) ∈ P, (G, H) ∈ Q and B ⊆ G} .

Proposition 2.1. If P and Q are pair-bases for a topological space X, then P ◦ Q

is a pair-base for X.

Proof. Let x ∈ X and O be an open neighborhood of x. Since P and Q are

pair-bases for X, there exists (G, H) ∈ Q with x ∈ G, H ⊆ O and then there
exists (A, B) ∈ P with x ∈ A and B ⊆ G. So, we have (A, H) ∈ P ◦ Q satisfying
x ∈ A and H ⊆ O. 

Proposition 2.2. Let P and Q be pair-collections in a topological space X. If

one of the pair-collections P or Q has one of the completeness properties in the
Definition 2, then P ◦ Q has the same completeness property.

Proof. We will prove for only one completeness property in the Definition 2.
Suppose P is a complete pair-collection in X and F is a filter base in P ◦ Q. Let
us see that the family
F1 = {(A, B) ∈ P : (A, H) ∈ F for a (G, H) ∈ Q with B ⊆ G} is a filter base in
P. Take (A1 , B1 ) , (A2 , B2 ) ∈ F1 . We have (G1 , H1 ) , (G2 , H2 ) ∈ Q such that
(A1 , H1 ) ∈ F and (A2 , H2 ) ∈ F. Since F is a filter base there is (A, H) ∈
F with (A, H)  (A1 , H1 ) and (A, H)  (A2 , H2 ). Definition of P ◦ Q gives us
nonempty subsets B, G of X such that (A, B) ∈ P, (G, H) ∈ Q and B ⊆ G. It
is easy to see that (A, B) is an element of F1 satisfying (A, B)  (A1 , B1 ) and
T
(A, B)  (A2 , B2 ) . Hence F1 is a filter base in P and so (A,B)∈F1 B 6= ∅ because
T T
P is complete. We also have (A,B)∈F1 B ⊆ (A,H)∈F H by the definition of
T
P ◦ Q. Then (A,H)∈F H 6= ∅.
If Q has one of the completeness properties, it can be proved by a similar
way. 

It is easy to see that if P is countably complete and L-complete, then P is com-

plete. So, Proposition 2.2 gives us the following Corollary.

Corollary 2.3. Let X be a topological space. If P is a countably complete pair-

base for X and Q is a L-complete pair-base for X, then P ◦ Q is a complete
pair-base for X.
936 SÜLEYMAN ÖNAL AND ÇETIN VURAL

3. Topological spaces having a complete pair-base

The following two theorems say us the property of having a complete pair-
base lies somewhere between subcompactness and domain representability(i.e.,
P -completeness).

Theorem 3.1. Subcompact spaces have a complete pair-base.

Proof. Let X be a subcompact space and B be a subcompact base for X. We

claim that P = {(U, V ) : U, V ∈ B and cl (U ) ⊆ V } be a complete pair-base for
X.
First, we will see that P is a pair-base for X. Let x ∈ X and O be an open
neighborhood of x. Regularity of the space X gives us U, V ∈ B such that x ∈
U ⊆ cl (U ) ⊆ V ⊆ cl (V ) ⊆ O. So, we have (U, V ) ∈ B satisfying x ∈ U and
V ⊆ O.
Now, let us see that P is complete. Take any filter base F in P. It is not difficult
to show that the family V = {V : (U, V ) ∈ F for an U ∈ B} is a regular filter
T T
base in B. Hence we have V ∈V V 6= ∅ and so (U,V )∈F V 6= ∅. Therefore P is a
complete pair-base for X. 

Theorem 3.2. Let X be a topological space and B be any base for X. If the space
X has a complete pair-base then there is a relation ≺ on B such that (B, ≺) is
generalised subcompact base for X.

Proof. Suppose P is a complete pair-base for X. Define the relation ≺ on B by

the rule:

U ≺ V if and only if U ⊆ A ⊆ B ⊆ V for a (A, B) ∈ P,

for each U, V ∈ B. It is obvious that the conditions (G1) , (G2) and (G3) are sat-
isfied. Let us see (G4) . Take any x ∈ X and suppose U, V ∈ B with x ∈ U ∩ V .
Since P is a pair-base for X we have (A, B) ∈ P satisfying x ∈ A and B ⊆ U ∩ V
and we also have W ∈ B with x ∈ W ⊆ A ⊆ B ⊆ U ∩ V. So, W ≺ U and W ≺ V.
Hence the family {W ∈ B : x ∈ W } is downward directed by ≺ .
Now, (G5) . Let F ⊆ B and (F, ≺) be downward directed. Since (F, ≺) is down-
ward directed, the family
S = {(A, B) ∈ P : U ⊆ A ⊆ B ⊆ V for some U, V ∈ F} is a filter base in P. Com-
T
pleteness of P leads us to the fact F 6= ∅. Hence the base B is a generalised
subcompact base for X. 
 COMPLETENESS PROPERTIES IN TOPOLOGICAL SPACES 937

In [13, Theorem 4.2], it was obtained that generalized subcompactness is equiva-

lent to domain representability. So we can give the following Corollary.

Corollary 3.1. If the topological space X has a complete pair-base then X is

P -complete, i.e., domain representable.

Now we recall that the internal characterization of p-spaces given by Burke in

[4, Theorem 1.3] or in [8, Theorem 3.21]. A space X is a p-space if and only if
there exists a sequence {On : n ∈ N} of open covers of X satisfying:
T
If On ∈ On for each n ∈ N and n∈N On 6= ∅, then
T
(i) n n∈N cl (On ) is compact,
o
T T
(ii) i≤n cl (Oi ) : n ∈ N is an outer network for the set n∈N cl (On ) .

Theorem 3.3. p-spaces have an L-complete pair-base.

Proof. Let X be p-space, and let {On : n ∈ N} be a sequence of open covers of

X satisfying the above conditions. Without loss of generality, we can assume that
X ∈ O1 and On+1 ⊆ On for each n ∈ N. For each open subset V of X, define the
number δ (V ) as follows: If for each n, there is an On ∈ On such that V ⊆ On ,
δ (V ) = ∞ and otherwise δ (V ) = max {n ∈ N : V ⊆ O for a O ∈ On }. We claim
that the family
P = {(U, V ) : U, V ⊆ X open, cl (U ) ⊆ V and δ (U ) > δ (V ) if δ (V ) < ∞} is a
L-complete pair base for X. First, let us see that P is a pair-base for X. Let
x ∈ X and V be an open subset of X such that x ∈ V. We have an open subset
W of X with x ∈ W ⊆ cl (W ) ⊆ V by the regularity of X. If δ (V ) = ∞, we
have (W, V ) ∈ P which we want it, otherwise, since the family Oδ(V )+1 is a cover
of X, we have an O ∈ Oδ(V )+1 with x ∈ O, and it is clear that (W ∩ O, V ) ∈ P
which we want it. Now, let us see that P is L-complete. Let F be a filter base
T
in P, and suppose (U,V )∈C V 6= ∅ for every countable subfamily C of F. We will
T
show that (U,V )∈F V 6= ∅. If F has a minimal element (U0 , V0 ) , it is obvious
T T
that (U,V )∈F V = V0 , and so (U,V )∈F V 6= ∅. Let F have no minimal element.
We have two cases.
Case 1: If there exists (U0 , V0 ) ∈ F such that δ (V0 ) = ∞; we have On ∈ On
T
such that V0 ⊆ On for each n ∈ N. Since V0 6= ∅, then n∈N On 6= ∅, and by
T
(i) in the above characterization of p-space, we have n∈N cl (On ) is compact.
Hence cl (U0 ) is compact. At the same time, since F is a filter base, we have
T
cl (U ) ∩ cl (U0 ) 6= ∅, for each (U, V ) ∈ F. These facts give us (U,V )∈F cl (U ) 6= ∅,
T
and so (U,V )∈F V 6= ∅.
938 SÜLEYMAN ÖNAL AND ÇETIN VURAL

Case 2: If δ (V ) < ∞ for each (U, V ) ∈ F; since F is a filter base and has no
minimal element, it is easy to see that the set {δ (V ) : (U, V ) ∈ F} is unbounded.
So, we can choose (Un , Vn ) ∈ F such that δ (Vn ) > n. By our hypotheses, we have
T
n∈N Vn 6= ∅. Now, let us choose an On ∈ Oδ(Vn ) with Vn ⊆ On for each n ∈ N.
Since δ (Vn ) > n and On+1 ⊆ On , we have On ∈ On for each n ∈ N. By (i) in the
T
above characterization of p-space, we have the set K = n∈N cl (On ) is compact.
Since F is a filter base, we have cl (U )∩K 6= ∅ by (ii) in the above characterization
T T
of p-space. Therefore (U,V )∈F clK (U ) 6= ∅, and so (U,V )∈F V 6= ∅. 

Corollary 3.2. Let X be a paracompact p-space. If X has a countable complete

pair-base then X is Čech-complete.

Proof. By Theorem 3.3, X has a L-complete pair-base, and so X has a complete

pair-base from Corollary 2.3. Hence X is domain representable by Corollary 3.1.
At the same time, we know that paracompact p-spaces are the perfect pre-images
of metric spaces by Corollaries 3.20 and 3.7 in [8]. So, there exist a metric space
Y and a perfect map f from X onto Y. We also have proved that perfect images
of domain representable spaces are domain representable in [12]. Since domain
representability is equivalent to Čech-completeness in metrizable spaces, we have
that the space Y is Čech-complete. Hence the space X is Čech-complete by
3.9.10.Theorem in [5]. 

4. Operations on topological spaces having a complete pair-base

In this section we will show that the property of having a complete pair-base
shows similarities with subcompactness and domain representability and they
have some common traits.
It is known that subcompactness and domain representability are hereditary
with respect to open subspaces. This heritability is also true for the property of
having a complete pair-base.

Proposition 4.1. If the topological space X has a complete pair-base then every
open subspace of X has such a base.

Proof. Let P be a complete pair-base for X, and let O be an open subspace

of X. Define the family Q = {(A, B) ∈ P : B ⊆ O} . It is clear that the family Q
is a pair-base for O. Since every filter base in Q is a filter base in P, we have
T
(A,B)∈F B 6= ∅ for every filter base F in Q. Hence Q is complete. 
 COMPLETENESS PROPERTIES IN TOPOLOGICAL SPACES 939

We observe that the other completeness properties in Definition 2 are also

hereditary with respect to open subspaces.

In [7, Theorem 2.5], it was showed that any finite union of subcompact spaces
is supcompact, and in [6, Proposition 7.1], it was showed that the analogous result
holds for domain representable spaces. This result is also true for spaces having
a complete pair-base. In the proof of the following proposition, we use similar
approach in the proof of Theorem 2.5 in [7].

Proposition 4.2. The union of two subspaces having complete pair-bases has a
complete pair-base.
Proof. Let X be a topological space and X = Y ∪ Z. Let Q and R be complete
pair-bases for the subspaces Y and Z, respectively. Using the similar way with the
proof of Theorem 2.5 in [7], it may be seen that the family P ={(U, V ) : U, V ⊆ X
open, cl (U ) ⊆ V, B ⊆ V and (U ∩Y ⊆ A or U ∩Z ⊆ A) for some (A, B) ∈ Q ∪ R}
is a complete pair-base for X. 

The authors, in [9, Theorem 2.1], showed that any union of open subcompact
subspaces is subcompact. In [6, Theorem 7.2], it was showed that any union of
open domain representable subspaces is domain representable. This property is
also true for any union of open subspaces which have a complete pair-base.
Theorem 4.1. If X is a topological space and O is a family of open subspaces
S
of X having a complete pair-base, then O has a complete pair-base.
Proof. Let κ be a cardinal number and O = {Oi : i < κ} . Without loss of gen-
S
erality, we may assume X = i<κ Oi . Let Pi be a complete pair-base for Oi ,
S
for each i < κ. Define the family Hi ={(U, B) : U ⊆ X open, B * j<i Oj
and (A, B) ∈ Pi for an A ⊆ Oi with cl (U ) ⊆ A} for each i < κ. Let us see
S
that the family A = i<κ Hi is a complete pair-base for X. Firstly, we will
show that A is a pair-base. Assume that x ∈ X and G is an open neighbor-
hood of x. Let i (x) = min {i < κ : x ∈ Oi } . Since Pi(x) is a pair-base for Oi(x) ,
S
there is (A, B) ∈ Pi(x) such that x ∈ A and B ⊆ G. Since x ∈ / j<i(x) Oj , we
S
have B * j<i(x) Oj . Since the space X is regular and the set A is open in X,
we have an open subset U of X such that x ∈ U ⊆ cl (U ) ⊆ A. So, we have
(U, B) ∈ Hi(x) , x ∈ U, B ⊆ G. Therefore the family A is a pair-base for X. Now,
we will show that the family A is complete. Let F ⊆ A be a filter base, and let
m = min {i < κ : F ∩ Hi 6= ∅} . We claim that the family
S = {(A, B) ∈ Pm : (U, B) ∈ F ∩ Hm , cl (U ) ⊆ A for an open set U in X}
940 SÜLEYMAN ÖNAL AND ÇETIN VURAL

is filter base in Pm . Take (A1 , B1 ) , (A2 , B2 ) ∈ S. We have (U1 , B1 ) , (U2 , B2 ) ∈

F ∩ Hm such that cl (U1 ) ⊆ A1 and cl (U2 ) ⊆ A2 . Since F is a filter base in A, we
have (U, B) ∈ A such that (U, B)  (U1 , B1 ) and (U, B)  (U2 , B2 ) . The fact
S
A = i<κ Hi and the minimality of m give us an α < κ with (U, B) ∈ Hα and
S
α ≥ m. Since B ⊆ U1 ⊆ B1 ⊆ Om , then (U, B) ∈ / i>m Hi . So, it must be α = m
and hence we have an A ⊆ Om such that (A, B) ∈ Pm and cl (U ) ⊆ A. Therefore
T
the family S is filter base in Pm . Since Pm is complete, we have (A,B)∈S B 6= ∅,
T T T
We also observe that (A,B)∈S B = (A,B)∈F B. Hence (A,B)∈F B 6= ∅ and so,
the family A is complete. 

In [1, Theorem 3.2], authors proved that Gδ -subspaces of domain representable

spaces are domain representable. This was also proved in [6, Proposition 9.1], by
using simplified definition of domain representability. In [6, Proposition 9.2], it
was also proved that dense Gδ -subspaces of subcompact spaces are domain rep-
resentable, but it is still not known whether dense Gδ -subspaces of subcompact
spaces are subcompact, in particular whether every Čech-complete space is sub-
compact. It is well known that dense Gδ -subspaces of Baire spaces are Baire
space. Another similar trait of the property having a complete pair-base and
domain representability is the following.
Theorem 4.2. If a topological space has a complete pair-base, then every dense
Gδ -subset of it has a complete pair-base.
Proof. Suppose that Y is a topological space having a complete pair-base and X
is a dense Gδ -subset of Y. Let P be a complete pair-base for Y and {Gn : n ∈ N}
T
be a decreasing family of open subsets of Y such that X = n∈N Gn . Without
loss of generality, we may assume G0 = Y. For each open subset V of X, define
the number

max {n ∈ N : cl (V ) ⊆ Gn } , cl (V ) * Gk for a k ∈ N;
δ (V ) =
∞ , otherwise.
For each x ∈ X and n ∈ N, choose an open subset W (x, n) of Y satisfying
x ∈ W (x, n) ⊆ cl (W (x, n)) ⊆ Gn . Define the family
PX ={(U, V ) : U and V are open subsets of X, clX (U ) ⊆ V, cl (U ) ⊆ A ⊆ B ⊆
cl (V ) for an (A, B) ∈ P and δ (U ) > δ (V ) if δ (V ) < ∞}.
We claim that the family PX is a complete pair-base for X. First, we will show
that the family PX is a pair-base. Assume that x ∈ X and H is an open subset
which contains x. Let O be an open subset of Y with H = O ∩ X. Since P is a
 COMPLETENESS PROPERTIES IN TOPOLOGICAL SPACES 941

pair-base for Y, we have (A, B) , (M, N ) ∈ P such that x ∈ A, B ⊆ O, x ∈ M,

N ⊆ A and cl(M ) ⊆ N. Define the set

M ∩X , δ (H) = ∞;
U=
M ∩ X ∩ W (x, δ (H) + 1) , otherwise.
The facts U ⊆ M ⊆ cl(M ) ⊆ N ⊆ A ⊆ B ⊆ O lead us to the fact cl(U ) ⊆
A ⊆ B ⊆ O, and so clX (U ) ⊆ H. Since X is dense and O is open in Y, we have
cl(U ) ⊆ A ⊆ B ⊆ cl(H). So (U, H) is the element of PX which we want. Now,
we will prove that PX is complete. Take any filter base F in PX . Without loss of
generality, we may assume that F has no minimal element. We claim that the set
{δ (V ) : ∃U, (U, V ) ∈ F} is not bounded. Assume contrary. Then there exist m ∈
N and (U0 , V0 ) ∈ F such that max {δ (V ) : ∃U, (U, V ) ∈ F} = m and δ (V0 ) = m.
Since (U0 , V0 ) ∈ F, we have δ (U0 ) > δ (V0 ) , and so δ (U0 ) > m. Since F is a filter
base and F has no minimal element, we have a pair (U, V ) ∈ F with (U, V ) 
(U0 , V0 ) . This leads us to the fact V ⊆ U0 , and hence δ (V ) ≥ δ (U0 ) > m. But
this contradicts the maximality of m. Hence the set {δ (V ) : ∃U, (U, V ) ∈ F} is
unbounded. So, we can choose (Un , Vn ) ∈ F such that δ (Vn ) > n for each n ∈ N.
T
Therefore we have cl(Vn ) ⊆ Gδ(Vn ) ⊆ Gn , and so n∈N cl(Vn ) ⊆ X. Since the
family F is a filter base in PX , it is easy to see that the family
S = {(A, B) ∈ P : ∃ (U, V ) ∈ F, cl (U ) ⊆ A ⊆ B ⊆ cl (V )}
T
is a filter base in P. Completeness of P give us (A,B)∈S B 6= ∅. At the same time
we have that the facts
T T T T
(A,B)∈S B ⊆ (U,V )∈F cl(V ) ⊆ n∈N cl(Vn ) ⊆ X. So, (U,V )∈F clX (V ) 6= ∅, and
T
hence (U,V )∈F V 6= ∅. 

Since compact spaces have a complete pair base, we can give the following Corol-
lary.
Corollary 4.3. Čech-complete spaces have a complete pair-base.

The authors, in [10, Theorem 2.2], proved that retracts of domain representable
spaces are domain representable. It is an open question, in [13, Question 6.6],
whether retracts of subcompact spaces are subcompact. Let us recall that a
retraction from the topological space Y onto a subspace X of Y is a continuous
map r : Y → X such that r (x) = x for all x ∈ X, and then X is called a
retract of Y. We can not obtain that the property of having a complete pair-base
is preserving under retractions but we obtain the following.
942 SÜLEYMAN ÖNAL AND ÇETIN VURAL

Theorem 4.3. Every retract of the space having a centered complete pair-base
has a complete pair-base.

Proof. Let the family P be a centered complete pair-base for a topological space
Y and let X be a retract of Y with the retraction r : Y → X.
We claim that the family PX = {(A ∩ X, r (B)) : (A, B) ∈ P} is a complete
pair-base for X. Since cl (A) ⊆ B, we have clX ((A ∩ X)) ⊆ r (B) , for each
(A, B) ∈ P. So, the family PX is a pair-collection. First, we will show that the
family PX is a pair-base. Assume that x ∈ X and H is an open neighborhood of
x in X. Since the retraction r is continuous and r (x) = x, we have an open subset
O of Y such that x ∈ O and r (O) ⊆ H, and then we have a pair (A, B) ∈ P with
x ∈ A, B ⊆ O. Therefore, x ∈ A ∩ X and r (B) ⊆ H.
Now, we will prove that PX is complete. Take any filter base F in PX . Since
every filter base is centered, the family
C = {(A, B) ∈ P : (A ∩ X, r (B)) ∈ F} is a centered family in P. Centered com-
T
pleteness of P leads us to (A,B)∈C B 6= ∅, and so we have
T
(A∩X,r(B))∈F r (B) 6= ∅. Hence the family PX is a complete pair-base for X. 

5. Relationships between bases and complete pair-bases

In [11, Corollary 3.5], authors show that for any domain representable space X
and for any base B of it, there exists a triple (P, , ϕ) representing X such that
ϕ (p) ∈ B for each p ∈ P. Similar argument is true for topological spaces having a
complete pair-base. That is, if a topological space X has a complete pair-base P,
the following proposition says us, for Y = X, we can assume that P is a subset
of B × B where B is any base of X.

Proposition 5.1. Let Y be a subspace of a topological space X and B be a base

for the space X. If PY is a complete pair-base for the subspace Y of X then the
family P = {(U, V ) ∈ B × B : cl (U ) ⊆ V and U ∩ Y ⊆ A ⊆ B ⊆ V for an
(A, B) ∈ PY ] is complete in X and it is an outer pair-base for Y in X.

Proof. First, we prove that P is complete in X. Let F be a filter base in P

and choose an (Aa , Ba ) ∈ PY such that U ∩ Y ⊆ Aa ⊆ Ba ⊆ V for each a =
(U, V ) ∈ F. It is easily seen that the family G = {(Aa , Ba ) ∈ PY : a ∈ F} is a
filter base in PY . Since PY is complete and Ba ⊆ V for each a = (U, V ) ∈ F we
T
have (U,V )∈F V 6= ∅.
Now, we will show that P is an outer pair-base for Y in X. Let y ∈ Y and G be
an open subset of X with y ∈ G. Since B is a base for the regular space X, there
 COMPLETENESS PROPERTIES IN TOPOLOGICAL SPACES 943

exist a V ∈ B and an open subset T of X such that y ∈ T ⊆ cl (T ) ⊆ V ⊆ G.

Since PY is a pair-base for the subspace Y, there exists (A, B) ∈ PY such that
y ∈ A, clY (A) ⊆ B ⊆ T ∩Y . Since the set A is open in the subspace Y, there is an
open subset W of X such that y ∈ W ∩ Y ⊆ A. Let H = W ∩ T. Since B is a base
for the space X, we have an U ∈ B with y ∈ U ⊆ H. It is clear that cl (U ) ⊆ V
and (U, V ) ∈ P. At the same time, we have y ∈ U and V ⊆ G. Therefore P is an
outer pair-base for Y in X. 

The proof of the following proposition is routine.

Proposition 5.2. Let P be a complete pair-base for a topological space X, and
Y be a subspace of X. Suppose that BY be a base of Y and O = {O ⊆ X :
O ∩ Y ∈ BY } . Then the family
A = {(G, H) ∈ O × O : G ⊆ A ⊆ B ⊆ H for some (A, B) ∈ P}
is complete and the family
AY = {(G ∩ Y, H ∩ Y ) : (G, H) ∈ A}
is a pair-base for Y.

References
[1] Harold Bennett and David Lutzer, Domain Representable spaces, Fundamenta Math. 189
(3) (2006), 255–268.
[2] Harold Bennett and David Lutzer, Domain-representability of certain complete spaces,
Houston J. Math. 34 (3) (2008), 753-772.
[3] Harold Bennett and David Lutzer, Subcompactness and domain representability in GO-
spaces on sets of real numbers, Topol. Appl. 156 (2009), 939-950.
[4] Dennis K. Burke, On p-spaces and ω∆-spaces, Pacific J. Math., 11 (1970), 105-126.
[5] Ryszard Engelking, General Topology, Heldermann Verlag, Berlin, 1989.
[6] William Fleissner and Lynne Yengulalp, From subcompact to domain representable, Topol.
Appl. 195 (2015), 174-195.
[7] William Fleissner, Viladimir Tkachuk and Lynne Yengulalp, Every scattered space is sub-
compact, Topol. Appl. 160 (2013), 1305-1312.
[8] Gary Gruenhage, Generalized Metric Spaces, in: Handbook of Set Theoretic Topology; K.
Kunen and J. E. Vaughan eds., Elsevier, North-Holland, Amsterdam, 1984, 423-501.
[9] Süleyman Önal and Çetin Vural, Every monotonically normal Čech-complete space is sub-
compact, Topol. Appl. 176 (2014), 35-42.
[10] Süleyman Önal and Çetin Vural, Domain representability of retracts, Topol. Appl. 194
(2015), 1-3.
[11] Süleyman Önal and Çetin Vural, There is no domain representable dense proper subsemi-
group of a topological group, Topol. Appl. 216 (2017), 79-84.
944 SÜLEYMAN ÖNAL AND ÇETIN VURAL

[12] Süleyman Önal and Çetin Vural, Certain Completeness Properties and Mappings, presub-
mit, (2020)
[13] Lynne Yengulalp, Coding strategies, the Choquet game, and domain representability, Topol.
Appl. 202 (2016), 384-396.
Received October 13, 2020
Revised version received March 22, 2021

(Süleyman Önal) Middle East Technical University, Department of Mathematics,

06531 Ankara, Turkey
Email address: osul@metu.edu.tr

(Çetin Vural) Gazi Üniversitesi, Fen Fakültesi, Matematik Bölümü, 06500 Tekniko-
kullar, Ankara, Turkey
Email address: cvural@gazi.edu.tr