Abstract
We extend the concept of being countable dense homogeneous to include fuzzy topological spaces. Our extension is proved to be a good extension in the sense of Lowen. We study the relation between fuzzy CDH spaces and some ordinary topological spaces generated by these fuzzy spaces.
Introduction
Zadeh [34] introduced the notion of a fuzzy set in a set X as a function λ from X into the closed interval [0,1]. Chang [10] introduced the notion of fuzzy topology on a non-empty set X as a collection of fuzzy sets on X, closed under arbitrary suprema and finite infima and containing the constant fuzzy sets 0 and 1. The concept of induced fuzzy topological space was introduced by Weiss [32]. Since then many authors have investigated such spaces and much attention has been given to the extension of the separation notions to fuzzy topological spaces. Dealing with uncertain data is a very hot research topic [35, 36].
In 1920, Sierpinski [27] introduced the notion of a homogeneous space. Seven years earlier, Brouwer had shown that if A and B are two countable dense subsets of the n-dimensional Euclidean space , then there is an autohomeomorphism on that takes A to B. He needed this result in his development of dimension theory.
In 1972, Bennett [9] began the abstract study, who called such spaces countable dense homogeneous spaces. In 1954, Ford [17] introduced a new concept called strongly locally homogeneous and he denoted it by SLH. Fitzpatrick, Fletcher, McCoy, Ungar, Van Mill, and Lauer extended the study to new types of homogeneity.
Mathematicians generalized many concepts of ordinary topological spaces to include fuzzy topological spaces such as: separation axioms, connectedness, compactness, countability axioms. Several fuzzy homogeneity concepts were discussed in [1–6, 16]. Since countable dense homogeneity is of importance in general topology and still a hot area of research, as appears in [7, 30] and other papers, we found it necessary to extend it to include fuzzy topological spaces. One of the main goals of the present work is to show how the definition of countable dense homogeneous ordinary topological spaces can be modified in order to define a good extension of it in fuzzy topological spaces. We will see that the exact versions of some known results regarding countable dense homogeneous ordinary topological spaces are invalid in fuzzy topological spaces, and we will open the door for other researchers to study the validity or invalidity for many other related known results. The study will also deal with the relation between fuzzy CDH spaces and some ordinary topological spaces generated by these fuzzy spaces.
In this paper we shall follow [33] for the following notations: For λ, μ ∈ [0, 1] X , we write λ ⊆ μ iff λ (x) ≤ μ (x) for all x ∈ X. By λ = μ we mean that λ ⊆ μ and μ ⊆ λ, i.e., λ (x) = μ (x) for all x ∈X. If {λ i : i∈ I } is a collection of fuzzy sets in X, then (⋃ λ i ) (x) = sup{ λ i (x) : i ∈ I }, x ∈ X; and (⋂ λ i ) (x) = inf{ λ i (x) : i ∈ I }, x ∈ X. Let r denote the fuzzy set given by r (x) = r for all x ∈ X, where 0≤ r ≤1 ; i.e., r denotes the “constant” fuzzy set of level r. The complement λ c of a fuzzy set λ in X is given by λ c (x) = 1 - λ (x), x ∈ X. If A ⊆ X, then χA denotes the characteristic function of A. If (X, τ) is an ordinary topological space, then the class of all lower semi-continuous functions from (X, τ) to ([0, 1] , τ u ), where τ u is the usual Euclidean topology on [0, 1], is a fuzzy topology on X. This fuzzy topology is denotedby ω (τ).
In this paper we shall follow [33] for definitions of fuzzy point, the direct and the inverse images of a fuzzy set, and fuzzy continuous mapping. For instance, a fuzzy point x a in a set X is given by x a (y) = a for y = x (0 < a < 1) and x a (y) = 0 for y ≠ x, x ∈ X is called the support of x a and a the value (level) of x a . However, we shall agree that a fuzzy crisp point x a in X is a fuzzy set in X given by x a (y) = 1 for y = x and x a (y) = 0 for y ≠ x. We shall follow [28] for the definition of ’belonging to’. Namely: a fuzzy point x a in X is said to belong to a fuzzy set λ in X (notation: x a ∈ λ) iff a < λ (x). Two fuzzy points x a and y b in X are said to be distinct iff their supports are distinct, i.e., x≠ y. A bijective map h : (X, ℑ 1) → (Y, J2); where (X, ℑ1) and (Y, ℑ2) are both fuzzy topological spaces; will be called fuzzy homeomorphism iff h and h-1 are both fuzzy continuous.
The corresponding pair (M, ℑ M ) is called a fuzzy open (closed) subspace iff the fuzzy set χ M is fuzzy open (closed) in (X, ℑ).
If X is a set, then X will denote its cardinality. We write to denote the set of all rational numbers. On a non-empty set X, τ ind , τ disc will denote the indiscrete and the discrete topologies on X, respectively. The closure of a fuzzy set λ in a fuzzy topological space (X, ℑ) will be denoted by or Cl (λ).
The following result obtained in [15] will be used in proving that our main new definition in this paper is a good extension.
Separability in fuzzy topological spaces
We start this section with the following two definitions:
In 1989, Fora [14], proved that the above two definitions are independent of each other.
Depending on the above two different definitions of denseness, we define separability in fuzzy topological spaces in two ways as follows.
We will prove that Definitions 2.3 and 2.4 are equivalent. For this reason, we need the following notations and definitions.
and
The following result is important in proving our main result of this section.
If P is dense (I), then is dense (II). If P is dense (II), then is dense (I).
(ii) Suppose that P is dense (II) and assume on the contrary that is not dense (I). Then there is a non-zero fuzzy open set λ such that p ∉ λ for each . We are going to prove that ⋃{ p : p ∈ P } ⊆ λ c . If , then λ (x) = 0, so λ c (x) = 1 ≥ (⋃ { p : p ∈ P }) (x). If x∉ S (Q (S (P))), then . Therefore, ⋃{ p : p ∈ P } ⊆ λ c , and hence, λ c = 1. Thus λ = 0, which is a contradiction.
Following is the main result of this section:
The other direction is similar to the above one.
The above result makes it clear that the two concepts of separability in fuzzy spaces are equivalent. Then from now on, we shall say separable to represent each of them.
CDH fuzzy topological spaces
Frechet and Brouwer have observed that n-dimensional Euclidean space has the property that if A and B are two countable dense sets in , then there is a homeomorphism taking A to B. They used this theorem in their development of dimension theory.
In 1962, Fort [18], proved that the Hilbert cube has the same property. In 1972, Bennett [9], gave the following definition:
X is separable. If A and B are two countable dense subsets of X, then there is a homeomorphism h : X → X such that h (A) = B.
In this section, we extend the above definition to include fuzzy spaces. For this reason, we need the following definition:
X is separable. If P1and P2 are two countable dense (I) collections of fuzzy points of X, then there is a fuzzy homeomorphism h : (X, ℑ) → (X, ℑ) such that h (S (P1)) = S (P2).
Now we shall give the following lemma which will be useful in showing that our Definition 3.2 is a good extension of the well-known Definition 3.1, according to Lowen [24].
If A is dense in (X, τ), then is dense (I) in (X, ω (τ)). If P is dense (I) in (X, ω (τ)), then S (P) is dense in (X, τ).
(ii) Let U be any non-empty open subset of (X, τ). Then χU is a non-zero fuzzy open set of (X, ω (τ)). Choose x a ∈P such that x a ∈χU , so x ∈ U ∩ S (P), which completes the proof.
The proof of the other direction is similar to the above one.
In ordinary topological spaces, it is well-known that the only countable CDH space is the discrete space. However, this is invalid in fuzzy spaces since the fuzzy space in the following example will be a countable non-discrete CDH fuzzy space.
We are going to explore an alternative version of the above well-known fact about countable CDH fuzzy spaces. For this reason, we need the following definition and lemma:
If P is countable dense (I) in (X, ℑ), then X is countable. (X, ℑ) is separable iff X is countable.
(ii) If X is separable, then there is a countable dense (I) P of (X, ℑ), then by part (i) S (P) = X, but . Hence X is countable.
Conversely, if X is countable, then it is clear that is countable dense (I). Thus, (X, ℑ) isseparable.
The following result shows us that the only countable CDH fuzzy spaces on a countable set X are the semi discrete fuzzy spaces on X.
Conversely, if X is a semi-discrete fuzzy topological space, then by Lemma 3.8 (ii), (X, ℑ) is separable since X is countable. Let P1 and P2 be any two countable dense (I) in (X, ℑ). Then according to Lemma 3.8 (i), S (P1) = S (P2) = X. Therefore, the identity fuzzy homeomorphism completes the proof.
In ordinary topological spaces, we have the following definitions and results:
The above relation is an equivalence relation. So this relation induces a partition on X into equivalence classes, and it leads us to the following definition:
We are going to study the diversity of the above three results in fuzzy spaces. For this reason, we need the following definitions:
It is easy to see that the above relation is an equivalence relation. So this relation induces a partition on X into equivalence classes, and it leads us to the following definition.
The exact versions of Theorems 3.13, 3.14 and 3.15 are invalid in fuzzy spaces, as one can easily see from Example 3.6.
Recall that a property P of fuzzy topological spaces is called a fuzzy topological property if whenever (X, ℑ 1) possesses P and h : (X, ℑ 1) → (Y, ℑ 2) is a fuzzy homeomorphism, then (Y, ℑ 2)possesses P.
The following two lemmas will be used in proving that the property CDH fuzzy spaces is a fuzzy topological property.
Topologies induced by fuzzy topologies
In this section, we shall study an important class of topological spaces that are induced by fuzzy topological spaces. We will study the relation between a CDH fuzzy topological space and its induced topological spaces. This class is introduced in the following definition:
The following is an example of a non-fuzzy space (X, ℑ), for which (X, ℑ a ) is a topological space for all a ∈ [0, 1).
so ℑ a is a topology on X for all a ∈ [0, 1). On the other hand, since x b ∪ ∉ℑ, then ℑ is not a fuzzy topology on X.
The converse of the above result needs not be true in general as the following example shows.
According to Theorem 4.3, one can easily prove the following result.
In Theorem 4.5, the equality need not be true in general as can be deduced from Example 4.4. In fact , but . However, we have the following alternative result.
Conversely,
if t , then t ∈ .
The following lemma is needed in the following result.
If A is dense in (X, ℑ 0), then is dense in (X, ℑ). If P is dense in (X, ℑ), then S (P) is dense in (X, ℑ 0).
The following is our other main result:
In fact if a > 0, then (X, ℑ) being CDH does not imply, in general, that (X, ℑ a ) is CDH. This will be explained in the following counterexample.
Conclusion
In this paper, as an ordinary topological property countable dense homogeneity is extended to include fuzzy topological spaces. It is proved that this extension is a good extension in the sense of Lowen. A characterization of countable dense homogeneous fuzzy topological spaces on a countable set is obtained. The exact versions of four known results related to countable dense homogeneity in ordinary topological spaces are proved to be invalid in fuzzy topological spaces. It is proved that a-cut topological space (X, ℑ a ) of a CDH fuzzy space (X, ℑ) is CDH in general only for a = 0. In our future study concerning CDH fuzzy spaces, the following topics could be considered: 1) To define several modifications of CDH fuzzy spaces such as DH; 2) To define the degrees of CDH property of fuzzy sets in a similar way to that used in [23]; 3) To show how the definition of CDH spaces can be modified in order to define CDH generalized L-topological spaces [8].
Footnotes
Acknowledgments
The authors are very grateful to the referees for their valuable comments and suggestions.
