NAME: Mrs.B.

JOTHILAKSMI
REG. NO.: 202017295
ASSISTANT PROFESSOR
DEPARTMENT OF MATHEMATICS
GOVERNMENT ARTS COLLEGE (AUTONOMOUS)
COIMBATORE – 641 018.
 On generalized closed sets

1. Introduction
Closed sets are fundamental objects in a topological space. For example, one can define
the topology on a set by using either the axioms for the closed sets or the Kuratowski
closure axioms. In 1970, Levine initiated the study of so-called generalized closed sets. By
definition, a subset S of a topological space X is called generalized closed if cl A ⊆ U
whenever A ⊆ U and U is open. This notion has been studied extensively in recent years
by many topologists because generalized closed sets are not only natural generalizations
of closed sets. More importantly, they also suggest several new properties of topological
spaces. Most of these new properties are separation axioms weaker than T 1, some of which
have been found to be useful in computer science and digital topology. For example, the
well-known digital line is T3/4 but not T1. Other new properties are defined by variations of
the property of submaximality. Furthermore, the study of generalized closed
sets also provides new characterizations of some known classes of spaces, for example, the
class of extremally disconnected spaces.
For the sake of convenience, we begin with some basic concepts, although most of
these concepts can be found from the references of this paper. A subset S of a topological
space X is called α-open (respectively semi-open, preopen, semi-preopen) if S ⊆ int(cl(int
S)) (respectively S ⊆ cl(int S), S ⊆ int(cl S), S ⊆ cl(int(cl S))). Moreover,
S is said to be α-closed (respectively semi-closed, preclosed, semi-preclosed) if X \ S is α-
open (respectively semi-open, preopen, semi-preopen) or, equivalently, if cl(int(cl S)) ⊆ S
(respectively int(cl S) ⊆ S, cl(int S) ⊆ S, int(cl(int S)) ⊆ S). The α-closure (respectively
semi-closure, preclosure, semi-preclosure) of S ⊆ X is the smallest α-closed (respectively
semi-closed, preclosed, semi-preclosed) set containing S. It is well known that α-cl S
= S ∪ cl(int(cl S)) and scl S = S ∪ int(cl S), pcl S = S ∪ cl(int S) and spcl S = S ∪ int(cl(int
S)). The α-interior of S ⊆ X is the largest α-open set contained in S, and we have α-int S =
S ∩ int(cl(int S)). It is worth mentioning that the collection α(X) of all α-
open subsets of X is a topology on X which is finer than the original one, and that a subset
S of X is α-open if and only if S is semi-open and preopen.

Definition 1: Let X be a topological space. A subset A of X is called:

(1) generalized closed (briefly, g-closed), if cl A ⊆ U , whenever A ⊆ U and U is open;
(2) semi-generalized closed (briefly, sg-closed), if scl A ⊆ U whenever A ⊆ U and
U is semi-open;
(3) generalized semiclosed (briefly, gs-closed), if scl A ⊆ U whenever A ⊆ U and U
is open;
(4) generalized α-closed (briefly, gα-closed) , if α-cl A ⊆ U whenever A ⊆ U and
U is α-open, or equivalently, if A is g-closed with respect to α(X);
(5) α-generalized closed (briefly, αg-closed), if α-cl A ⊆ U whenever A ⊆ U and
U is open;
(6) generalized semi-preclosed (briefly, gsp-closed), if spcl A ⊆ U whenever A ⊆ U
and U is open;
(7) regular generalized closed (briefly, r-g-closed), if cl A ⊆ U whenever A ⊆ U
and U is regular open.
 In addition to Definition 1 above, a subset A of X is g-open (sg-open ) if X \ A is
g-closed (sg-closed). Other classes of generalized open sets can be defined in a similar
manner. Recall that a space X is said to be submaximal if every dense subset
of X is open. As variations of submaximality, we obtain the notions of α-submaximal- ity, g-
submaximality and sg-submaximality. A space X is α-submaximal (respectively g-
submaximal, sg-submaximal) if every dense subset is α-open (respectively g-open, sg- open).
α-submaximal spaces have been studied by Ganster. Obviously, every submaximal space is g-
submaximal, that if (X, α(X)) is g-submaximal, then (X, α(X)) is also sg-submaximal. Note
that any indiscrete space with at least two points is g-submaximal but not submaximal, and an
sg-submaximal space which is not g-submaximal.
Characterize the spaces in which:
(i)Every semi-preclosed set is sg-closed.
(ii)Every preclosed set is gα-closed.
Proof:
Let S be a subset of a space X. A resolution of S is a pair (E1 , E2 ) of disjoint dense subsets of
S. The subset S is said to be resolvable if it possesses a resolution, otherwise S is called
irresolvable. In addition, S is called strongly irresolvable, if every open subspace of S is
irresolvable.
Observethat if (E1, E2⊕ is a resolution of S then E1 and E2 are codense in X, i.e., have empty
interior. We also note that every submaximal space is hereditarily irresolvable.

Lemma 1.2. Let X be a space. Then every singleton of X is either nowhere dense or
preopen.
Proof:
For a space X, we now define X1 = {x ∈ X: {x} is nowhere dense}, and
X2 = {x ∈X: {x} is preopen}. Then X = X1 ∪ X2 is a decomposition of X, which will be called
the
Janković–Reilly decomposition. Recall that a space X is said to be locally indiscrete if
every open subset is closed.

Theorem1.3 . For a space X with Hewitt decomposition X = F ∪ G the following are

equivalent:
(1) every semi-preclosed subset of X is sg-closed,
(2) X1 ∩ scl A ⊆ spcl A for each A ⊆ X,
(3) X1 ⊆ int(cl G),
(4) X ≈ Y ⊕ Z, where Y is locally indiscrete and Z is strongly irresolvable,
(5) every preclosed subset of X is gα-closed,
(6) X is g-submaximal with respect to α(X).

2. Lower separation axioms

A space X is called T1/2 if every g-closed subset of X is closed. The space X is T1/2 if and
only if singletons of X are either open or closed.
Theorem 2.1 Every submaximal space is T1/2.

It is obvious that in any topological space X, every sg-closed subset of X is gs-closed. In

[20], the class of Tgs-spaces was introduced where a space X is called Tgs if every gs-
closed subset of X is sg-closed. The following result exhibits the relationship between Tgs-
spaces and T1/2-spaces.

Theorem 2.2 . For a space X, the following statements are equivalent:

(1) X is a Tgs-space,
(2)every nowhere dense subset of X is a union of closed subsets of X (i.e., X is T1∗),
(3)(every generalized semi-preclosed subset of X is semi-preclosed (i.e., X is semi-pre-T1/2 ),
(4)every singleton of X is either preopen or closed.

Corollary 2.3. Every T1/2-space is Tgs.

A space X is called semi-T1 if each singleton is semi-closed, it is called semi- T1/2 if

every singleton is either semi-closed or semi-open. Let s(X) be the semi- regularization of
a space X. The closure of a subset A of X with respect to s(X) will be denoted by δ-cl
A. A subset A of X is called δ-generalized closed if δ-cl A ⊆ U when A ⊆ U and U is open
in X. Moreover, X is called a T3/4-space [8] if every δ- generalized closed subset of X is
closed in s(X). The well-known digital line, also called the Khalimsky line, is a T3/4-space
which fails to be T1.

Theorem 2.4. For any space X,

(1) T3/4 = Tgs + semi-T1.
(2) T1/2 = Tgs + semi-T1/2.

The results above clarify some connections between the Tgs property and other lower
separation axioms. Note, however, that not every Tgs-space is T0. For example, a three
point space X = {a, b, c} in which the only proper open subset is {a, b}, is a Tgs-space, but
not a T0-space. To obtain more characterizations of Tgs-spaces, we need the following
lemma.

Lemma 2.5. A subset A of a space X is gα-closed if and only if X1∩ α-cl A ⊆ A.

Proof. Suppose that A is gα-closed, and let x ∈ X1 ∩ α-cl A. If x ∈/ A, then X \ {x} is an

α-open set containing A and so α-cl A ⊆ X \ {x}, which is impossible.
Conversely, suppose that X1∩ α-cl A ⊆ A. Let U be an α-open set containing A, and let x
∈ α-cl A. If x ∈ X1 , then x ∈ A ⊆ U . Now let x ∈ X2 and assume that x ∈/ U . Then X \ U is an
α-closed set containing x, and thus, α-cl({x}) = {x}∪ cl(int(cl({x}))) ⊆ X \ U . Since {x} is
preopen, we have int(cl({x})) ∩ A /= ∅. Pick a point y ∈ int(cl({x})) ∩ A. Then y ∈ A ∩ (X \ U)
⊆ U ∩ (X \ U), which is a contradiction.

Theorem 2.6. A space X is Tgs if and only if every αg-closed subset of X is gα-closed.
 Proof. Suppose that X is Tgs. Let A be αg-closed and let x ∈ X1 ∩ α-cl A. Then {x} is
closed by Theorem 2.2. Assume that x ∈/ A, i.e., A ⊆ X \ {x}. Since A is αg-closed and X \
{x} is open we have x ∈ α-cl A ⊆ X \ {x}, which is a contradiction. Therefore, X1∩ α-cl A
⊆ A. By Lemma 3.5, A is gα-closed.

Conversely, assume that every αg-closed subset of X is gα-closed. Let x ∈ X1 and

suppose that {x} is not closed. Then X \ {x} is dense and αg-closed, thus gα-closed. It
follows from Lemma 3.5 that
. Σ
X1 ∩ α- cl X \ {x} = X1 ∩ X = X1 ⊆ X \ {x}.
So we obtain x ∈ X \ {x}, a contradiction.

Remark 2.7. A space whose αg-closed subsets are gα-closed may be called a Tαg -space.
The above Theorem shows, however, that the class of Tαg -spaces is precisely the class of
Tgs- spaces.

Proposition 2.8. A space X is Tgs and extremally disconnected if and only if every gs-
closed subset of X is preclosed.

Proof. Suppose that X is Tgs and extremally disconnected. Let A be a gs-closed subset.
Then A is sg-closed. A is preclosed.
Now assume that every gs-closed subset is preclosed. Let x ∈ X and suppose that {x} is
not closed. Then X \ {x} is not open, hence it is gs-closed. By assumption, X \ {x} is
preclosed, and so {x} is preopen. X is a Tgs-space. Moreover,
since every sg-closed subset is preclosed, X is extremally disconnected.

By Theorem 2.6, in a Tgs-space, every g-closed subset is gα-closed. This suggests a

natural question: Characterize those spaces whose gα-closed subsets are g-closed. Clearly,
if X is nodec, i.e., every nowhere dense subset of X is closed, then every gα-closed subset
of X is g-closed since in that case α(X) coincides with the given topology on X. Also
observe that X is nodec if and only if every nowhere dense subset is discrete as a subspace.
Theorem 2.9. For a space X the following are equivalent:
(1) every gα-closed set is g-closed,
(2) every nowhere dense subset is locally indiscrete as a subspace,
(3) every nowhere dense subset is g-closed,
(4) every α-closed set is g-closed.

Proof. (1) ⇒ (2): Let N ⊆ X be nowhere dense, let U be open and let N1 = U ∩ N . We
have to show that N1 is closed in N . Since N1 is nowhere dense it is α-closed, hence gα-
closed and so g-closed. Since N1 ⊆ U , we have cl N1 ⊆ U and hence cl N1 ∩ N = N1, i.e., N1
is closed in N .
⇒ (3): Let N ⊆ U where N is nowhere dense and U is open. Let x ∈ cl N . Then cl{x}⊆ cl N .
Since cl N is nowhere dense, by (2) we have that cl{x} is also open in cl N , i.e., there exists an
open set W such that cl{x} = W ∩ cl N . Suppose that x ∈/ U . Then cl{x}⊆ X \ U and so
W ∩ N ⊆ W ∩ cl N ∩ U = ∅, a contradiction. Therefore cl N ⊆ U.
 (2) ⇒ (4): Let F ⊆ X be α-closed. Then F = A ∪ N where A is closed and N is nowhere
dense. If F ⊆ U where U is open then, by our assumption, cl N ⊆ U and so cl F ⊆ U ,
i.e., F is g-closed.
(3) ⇒ (1): Let A be gα-closed with A ⊆ U where U is open. By assumption α- cl A
= A ∪ cl(int(cl A)) ⊆ U . It is easily checked that N = A \ cl(int(cl A)) is nowhere dense,
hence α-closed and so g-closed by assumption. Since N ⊆ U we have cl A ∩ (X \ cl(int(cl
A)) ⊆ cl N ⊆ U . It follows readily that cl A\ α-cl A ⊆ U and so cl A ⊆ U , i.e., A is g-closed.

In our next example we will show that there exist spaces whose nowhere dense subsets
are g-closed but which are not nodec.

Example 2.10. Let X be the real line and let X1 = {x ∈ X: x > 0} and X2 = {x ∈ X: x < 0}.
We now define a topology on X in the following way. Let {0} be open. If x ∈ X 1 , a basic
(minimal) open neighborhood of x is X1 ∪ {0}. If x ∈ X2, a basic (minimal) open
neighborhood of x is X2 ∪{0}. Clearly, if x ∈ X1 then {x} is nowhere dense but not closed, so
X fails to be nodec. Now let N ⊆ U where N is nowhere dense and U is open. Then 0 ∈/ N .
Let N1 = N ∩ X1 and N2 = N ∩ X2 . If x ∈ N1 then x ∈ U and so X1 ⊆ U . Hence cl N1 ⊆ X1 ⊆ U
. In the same manner, cl N2 ⊆ U and so cl N ⊆ U , i.e., N is g-closed.

3. More characterizations
We now return to the diagram in Section 1 to consider other possible converses of some
of the implications in that diagram. The following result about the class of extremally
disconnected spaces was proved .

Theorem 3.1 . For a space X, the following statements are equivalent:

(1) (X, τ) is extremally disconnected,
(2) scl(A ∪ B) = scl A ∪ scl B for all A, B ⊆ X,
(3) the union of two semi-closed subsets of X is semi-closed,
(4) the union of two sg-closed subsets of X is sg-closed,
(5) every semi-preclosed subset of X is preclosed,
(6) every sg-closed subset of X is preclosed,
(7) every semi-closed subset of X is preclosed,
(8) every semi-closed subset of X is α-closed,
(9) every semi-closed subset of X is gα-closed.

Next we present one more characterization of extremal disconnectedness using

generalized closed subsets.

Theorem 3.2. A space X is extremally disconnected if and only if every gs-closed subset of
X is αg-closed.
Proof. Suppose that X is extremally disconnected. Let A be gs-closed and let U be an open
set containing A. Then scl A ⊆ U , i.e., int(cl A) ⊆ U . Since int(cl A) is closed, we have α-
cl A = A ∪ cl(int(cl A)) ⊆ A ∪ int(cl A) ⊆ U . Hence A is αg-closed.
To prove the converse, let every gs-closed subset of X be αg-closed. Let A ⊆ X be
 regular open. Then A is gs-closed and so αg-closed. It follows that cl A = cl(int(cl A)) = α-
cl A ⊆ A. Therefore A is closed and X is extremally disconnected.
We now consider the property of sg-submaximality. First we give some elementary
characterizations of sg-submaximal spaces. Since the proof of the following result is
straightforward, we will omit it.

Theorem 3.3. For a space X, the following are equivalent:

(1) X is sg-submaximal,
(2) every subset of X is an intersection of a closed subset and an sg-open subset of X,
(3) every subset of X is a union of an open subset and an sg-closed subset of X,
(4) every codense subset A of X is sg-closed,
(5) cl A \A is sg-closed for every subset A of X.

A more advanced result about sg-submaximality was obtained.

= X F=G, the following are

Theorem 3.4 . For a space X with Hewitt decomposition
equivalent:
(1) X1 ⊆ cl G,
(2) every preclosed subset of X is sg-closed,
(3) X is sg-submaximal,
(4) X is sg-submaximal with respect to α(X).

We shall now improve the equivalence of (2) and (3) in Theorem 4.4 thereby providing
a new characterization of sg-submaximal spaces.

Theorem 3.5. A space X is sg-submaximal if and only if every preclosed subset of X is gs-
closed.

Proof. The necessity is trivial by Theorem 3.4(2). For the sufficiency, suppose that every
preclosed subset is gs-closed. Let X = F ∪ G be the Hewitt decomposition of X, and let
(E1, E2⊕ be a resolution of int F .
We first claim that every open set V ⊆ int F is regular open. In fact, V ∩ E1 is condense and
contained in V . Since codense sets are preclosed, by assumption, they are gs-closed.
Thus int(cl(V ∩ E1)) ⊆ V . On the other hand, E1 is dense in int F, hence we have
int(cl(V ∩ E1)) = int(cl V ). It follows that V = int(cl V ).
Now let x ∈ int F and let V = int F ∩ (X \ cl({x})). Suppose that {x} is nowhere dense.
Then X \ cl{x} is dense and int(cl V) = int(cl(int F)) = int F . By our claim, int F = V .
Hence int F ⊆ X \ {x}, a contradiction. Therefore {x} has to be preopen. We have thus proved
that int F ⊆ X2, i.e., X1 ⊆ cl G.
By Theorem 3.4, X is sg-submaximal.

Remark: One may define a space X to be gs-submaximal if each dense subset of X is gs-
open. Similar to the proof of Theorem 4.4, one checks easily that a space X is gs-submaximal
if and only if each preclosed subset of X is gs-closed. In the light of Theorem 4.5, the notion
of gs-submaximality coincides with that of sg-submaximality.