Abstract
In Wu Fuzzy Systems and Mathematics 3:94–99, 2012, the author introduced concepts of \(\alpha \)-remote neighborhood mapping and \(\alpha \)-quasi uniform, and obtained many good results in \(\alpha \)-quasi uniform spaces. This chapter will further investigate properties of \(\alpha \)-remote neighborhood mapping, and give some characterizations of \(\alpha \)-quasi uniforms. Based on this, this chapter also introduces concept of \(\alpha \)-P.Q. metric, and establishes the relations between \(\alpha \)-quasi uniforms and \(\alpha \)-P.Q. metrics.
Access provided by Autonomous University of Puebla. Download chapter PDF
Similar content being viewed by others
Keywords
- \(\alpha \)-Quasi uniform
- \(\alpha \)-Homeomorphism
- \(\alpha \)-P.Q. metric
- \(\alpha \)-Remote neighborhood mapping
1 Introduction
Theory of quasi-uniformity in completely distributive lattices was firstly introduced by Erceg [1] and Hutton [2]. Then it was developed into various forms and was extended into different topological spaces [3–9]. In [10], the author introduced the concept of \(\alpha \)-quasi uniform in \(\alpha \)-layer order-preserving operator spaces, and revealed the relations between \(\alpha \)-layer topological spaces and \(\alpha \)-quasi uniform spaces. In this chapter, firstly, we further study properties of \(\alpha \)-remote neighborhood mappings. Then we discuss some characterizations of \(\alpha \)-quasi uniformities. Secondly, we introduce the concept of \(\alpha \)-P.Q. metrics, and establish the relations between \(\alpha \)-quasi uniforms and \(\alpha \)-P.Q. metrics.
2 Preliminaries
In this chapter, \(X, Y\) will always denote nonempty crisp sets, A mapping \(A:X\rightarrow L\) is called an \(L\)-fuzzy set. \(L^X\) is the set of all \(L\)-fuzzy sets on \(X\). An element \(e\in L\) is called an irreducible element in \(L\), if \(p\vee q=e\) implies \(p=e\) or \(q=e\), where \(p, q\in L\). The set of all nonzero irreducible elements in \(L\) will be denoted by \(M(L)\). If \(x\in X\), \(\alpha \in M(L)\), then \(x_{\alpha }\) is called a molecule in \(L^X\). The set of all molecules in \(L^X\) is denoted by \(M^*(L^X)\). If \(A\in L^X\), \(\alpha \in M(L)\), take \(A_{[\alpha ]}=\{x\in X\mid A(x)\ge \alpha \}\) [3] and \(A^{\alpha }=\vee \{x_{\alpha }\mid x_{\alpha }\not \le A\}\) [11]. It is easy to check \((A_{[\alpha ]})^{\prime }= A^{\alpha }_{[\alpha ]}\).
Let \((L^X, \delta )\) be an \(L\)-fuzzy topological space, \(\alpha \in M(L)\). \(\forall A\in L^X\), \(D_{\alpha }(A)=\wedge \{G\in \delta ^{\prime }\mid G_{[\alpha ]}\supset A_{[\alpha ]}\}.\) Then the operator \(D_{\alpha }\) is a \(\alpha \)-closure operator of some co-topology on \(L^X\), denoted by \(D_{\alpha }(\delta )\). We called \(\alpha \)-layer topology. The pair \((L^X,D_{\alpha }(\delta ))\) is called \(\alpha \)-layer co-topological space [11]. An \(\alpha \)-layer topological space\((L^X,D_{\alpha }(\delta ))\) is called an \(\alpha \)-\(C_{II}\) space, if there is a countable base \(\fancyscript{B}_{\alpha }\) of \(D_{\alpha }(\delta )\).
A mapping \(F_{\alpha }: L^X\rightarrow L^Y\) is called an \(\alpha \)-mapping, if \(F_{\alpha }(A)_{[\alpha ]}=F_{\alpha }(B)_{[\alpha ]}\) whenever \(A_{[\alpha ]}=B_{[\alpha ]}\), and \(F_{\alpha }(A)=0_X\) whenever \(A_{[\alpha ]}=\emptyset \). The mapping \(F^{-1}_{\alpha }: L^Y\rightarrow L^X\) is called the reverse mapping of \(F_{\alpha }\), if for each \(B\in L^Y\), \(F^{-1}_{\alpha }(B)=\vee \{A\in L^X\mid F_{\alpha }(A)_{[\alpha ]}\subset B_{[\alpha ]}\}.\) Clearly, \(F^{-1}_{\alpha }\) is also an \(\alpha \)-mapping.
An \(\alpha \)-mapping \(F_{\alpha }: L^X\rightarrow L^Y\) is called an \(\alpha \)-order-preserving homomorphism (briefly \(\alpha \)-\(oph\)), iff both \(F_{\alpha }\) and \(F^{-1}_{\alpha }\) are \(\alpha \)-union preserving mappings.
An \(\alpha \)-mapping \(F_{\alpha }: L^X\rightarrow L^Y\) is called an \(\alpha \)-Symmetric mapping, if for every \(A,B\in L^X\), we have
An \(\alpha \)-mapping \(f_{\alpha }:L^X\rightarrow L^X\) is called an \(\alpha \)-remote neighborhood mapping, if for each \(A\in L^X\) with \(A_{[\alpha ]}\ne \emptyset \), we have \(A_{[\alpha ]}\not \subset f_{\alpha }(A)_{[\alpha ]}\). The set of all \(\alpha \)-remote neighborhood mappings is denoted by \(\fancyscript{F}_{\alpha }(L^X)\), (briefly by \(\fancyscript{F}_{\alpha }\)).
For \(f_{\alpha },g_{\alpha }\in \fancyscript{F}_{\alpha }\), let’s define
-
(1)
\(f_{\alpha }\le g_{\alpha } \Leftrightarrow \forall A\in L^X\), \(f_{\alpha }(A)_{[\alpha ]} \subset g_{\alpha }(A)_{[\alpha ]}\).
-
(2)
\((f_{\alpha }\vee g_{\alpha })(A)=f_{\alpha }(A)\vee g_{\alpha }(A)\).
-
(3)
\((f_{\alpha }\odot g_{\alpha })(A)= \wedge \{f_{\alpha }(B)\mid \exists B\in L^X\), \(B_{[\alpha ]}\not \subset g_{\alpha }(A)_{[\alpha ]}\}\).
An non-empty subfamily \(\fancyscript{D}_{\alpha }\subset \fancyscript{F}_{\alpha }\) is called an \(\alpha \)-quasi-uniform, if \(\fancyscript{D}_{\alpha }\) satisfies:
(\(\alpha \)-U1) \(\forall f_{\alpha }\in \fancyscript{D}_{\alpha },g_{\alpha }\in \fancyscript{F}_{\alpha }\) with \(f_{\alpha }\le g_{\alpha }\), then \(g_{\alpha }\in \fancyscript{D}_{\alpha }\).
(\(\alpha \)-U2) \(\forall f_{\alpha },g_{\alpha }\in \fancyscript{D}_{\alpha }\) implies \(f_{\alpha }\vee g_{\alpha }\in \fancyscript{D}_{\alpha }\).
(\(\alpha \)-U3) \(\forall f_{\alpha }\in \fancyscript{D}_{\alpha }\), then \(\exists g_{\alpha }\in \fancyscript{D}_{\alpha }\), such that \(g_{\alpha }\odot g_{\alpha }\ge f_{\alpha }\).
\((L^X,\fancyscript{D}_{\alpha })\) is called an \(\alpha \)-quasi-uniform space. A subset \(\fancyscript{B}_{\alpha }\subset \fancyscript{D}_{\alpha }\) is called a base of \(\fancyscript{D}_{\alpha }\), if \(\forall f_{\alpha }\in \fancyscript{D}_{\alpha }\), there is \(g_{\alpha }\in \fancyscript{B}_{\alpha }\), such that \(f_{\alpha }\le g_{\alpha }\). A subset \(\fancyscript{A}_{\alpha }\subset \fancyscript{D}_{\alpha }\) is called a subbase of \(\fancyscript{D}_{\alpha }\), if all of finite unions of the elements in \(\fancyscript{A}_{\alpha }\) consist a base of \(\fancyscript{D}_{\alpha }\). An \(\alpha \)-quasi-uniform \(\fancyscript{D}_{\alpha }\) is called an \(\alpha \)-uniform, if \(\fancyscript{D}_{\alpha }\) possesses a base whose elements are \(\alpha \)-symmetric. Usually, we call this base \(\alpha \)-symmetric base.
In [10], the author discussed the relation between an \(\alpha \)-quasi uniform space and an \(\alpha \)-layer co-topological space as following:
Let \(\fancyscript{D}_{\alpha }\) be an \(\alpha \)-quasi uniform on \(L^X\). \(\forall A\in L^X\), Let’s take \(c_{\alpha }(A)=\vee \{B\in L^X\mid \forall f_{\alpha }\in \fancyscript{D}_{\alpha }, A_{[\alpha ]}\not \subset f_{\alpha }(B)_{[\alpha ]}\}.\) Then \(c_{\alpha }\) is an \(\alpha \)-closure operator of some \(L\)-fuzzy co-topology, which is denoted by \(\eta _{\alpha }(\fancyscript{D}_{\alpha })\). Each \(\alpha \)-layer topological space \((L^X,D_{\alpha }(\delta ))\) can be \(\alpha \)-quasi uniformitale, i.e., there is an \(\alpha \)-quasi uniform \(\fancyscript{D}_{\alpha }\), such that \(D_{\alpha }(\delta )=\eta _{\alpha }(\fancyscript{D}_{\alpha })\).
Other definitions and notes not mentioned here can be seen in [12].
3 Properties of \(\alpha \)-Remote Neighborhood Mappings
Theorem 3.1
Let \(f_{\alpha },g_{\alpha }\in \fancyscript{F}_{\alpha }\). Then
-
(1)
\(f_{\alpha }\vee g_{\alpha }\in \fancyscript{F}_{\alpha }\), \(f_{\alpha }\odot g_{\alpha }\in \fancyscript{F}_{\alpha }\).
-
(2)
\(f_{\alpha }\odot g_{\alpha }\le f_{\alpha }, f_{\alpha }\odot g_{\alpha }\le g_{\alpha }\).
-
(3)
\((f_{\alpha }\odot g_{\alpha })\vee h_{\alpha }=(f_{\alpha }\vee h_{\alpha })\odot (g_{\alpha }\vee h_{\alpha })\),
\((f_{\alpha }\vee g_{\alpha })\odot h_{\alpha }=(f_{\alpha }\odot h_{\alpha })\vee (g_{\alpha }\odot h_{\alpha })\).
Theorem 3.2
Let \(f_{\alpha }\in \fancyscript{F}_{\alpha }\). If for each \(A\in L^X\),
and
Then
-
(1)
\(f^{\nabla }_{\alpha },f^{\diamond }_{\alpha }\in \fancyscript{F}_{\alpha }\).
-
(2)
\(f^{\diamond }_{\alpha }\le f^{\nabla }_{\alpha }\le f_{\alpha }\).
-
(3)
\(f^{\diamond }_{\alpha }(A)=\wedge \{f^{\nabla }_{\alpha }(B)\mid B_{[\alpha ]}\not \subset f_{\alpha }(A)_{[\alpha ]}\} =(f^{\nabla }_{\alpha }\odot f_{\alpha })(A)\).
-
(4)
\(f^{\nabla }_{\alpha }\) is \(\alpha \)-Symmetric.
Proof
(1) By \(A_{[\alpha ]}\subset \bigcup \limits _{G_{[\alpha ]}\not \subset A_{[\alpha ]}^{\alpha }}f_{\alpha }(G)_{[\alpha ]}^{\alpha }\), we have \(A_{[\alpha ]}\not \subset f^{\nabla }_{\alpha }(A)_{[\alpha ]}\). Thus \(f^{\nabla }_{\alpha }\in \fancyscript{F}_{\alpha }\). Again by \(A_{[\alpha ]}\not \subset f_{\alpha }(A)_{[\alpha ]}\), we get \(\bigcup \limits _{G_{[\alpha ]}\not \subset A_{[\alpha ]}^{\alpha }}f_{\alpha }(G)_{[\alpha ]}^{\alpha }\not \subset f_{\alpha }(A)_{[\alpha ]}\). Therefore, \(f^{\diamond }_{\alpha }\in \fancyscript{F}_{\alpha }\).
(2) The proof is obvious.
(3) For each \(B\in L^X\),
So
(4) \(\forall D,E\in L^X\). If there is \( A_{[\alpha ]}\not \subset D_{[\alpha ]}^{\alpha }\), such that
Then there is \(B\in L^X\), satisfying \(E_{[\alpha ]}\not \subset f_{\alpha }(B)_{[\alpha ]}\) and \(A_{[\alpha ]}\subset \bigcup \limits _{G_{[\alpha ]}\not \subset (B_{[\alpha ]}^{\alpha }}f_{\alpha }(G)_{[\alpha ]}^{\alpha }\). Clearly, \(\bigcup \limits _{G_{[\alpha ]}\not \subset B_{[\alpha ]}^{\alpha }}f_{\alpha }(G)_{[\alpha ]}^{\alpha }\not \subset D_{[\alpha ]}^{\alpha }\), i.e., \(D_{[\alpha ]}\not \subset \bigcap \limits _{G_{[\alpha ]}\not \subset B_{[\alpha ]}^{\alpha }}f_{\alpha }(G)_{[\alpha ]}\). Thus, there is \(C_{[\alpha ]}\not \subset B_{[\alpha ]}^{\alpha }\), such that \(D_{[\alpha ]}\not \subset f_{\alpha }(C)_{[\alpha ]}\). Conclusively, we have
and
Therefore, \(f^{\nabla }_{\alpha }\) is \(\alpha \)-Symmetric.
Theorem 3.3
Let \(f_{\alpha }\) be an \(\alpha \)-Symmetric remote neighborhood mapping, then
-
(1)
\(C_{[\alpha ]}\not \subset f_{\alpha }(A)_{[\alpha ]}\Rightarrow A_{[\alpha ]}\subset \bigcup \limits _{D_{[\alpha ]}\not \subset C_{[\alpha ]}^{\alpha }}f_{\alpha }(D)_{[\alpha ]}^{\alpha }\).
-
(2)
\(A_{[\alpha ]}\subset \bigcup \limits _{D_{[\alpha ]}\not \subset C_{[\alpha ]}^{\alpha }}f_{\alpha }(D)_{[\alpha ]}^{\alpha }\Rightarrow C_{[\alpha ]}\not \subset f_{\alpha }(A)_{[\alpha ]}\).
Proof
(1) Since \(f_{\alpha }\) is an \(\alpha \)-Symmetric mapping. \(\forall D_{[\alpha ]}\not \subset A_{[\alpha ]}^{\alpha }\), there is \(B_{[\alpha ]}\not \subset C_{[\alpha ]}^{\alpha }\), satisfying \(D_{[\alpha ]}\not \subset f_{\alpha }(B)_{[\alpha ]}\). Therefore,
(2) By \(A_{[\alpha ]}\subset \bigcup \limits _{D_{[\alpha ]}\not \subset C_{[\alpha ]}^{\alpha }}f_{\alpha }(D)_{[\alpha ]}^{\alpha }\), we have \(A_{[\alpha ]}^{\alpha }\supset \bigcap \limits _{D_{[\alpha ]}\not \subset C_{[\alpha ]}^{\alpha }}f_{\alpha }(D)_{[\alpha ]}\). So \(\forall x\not \in A_{[\alpha ]}^{\alpha }\), there is \(D_{[\alpha ]}\not \subset C_{[\alpha ]}^{\alpha }\), such that \(x\not \in f_{\alpha }(D)_{[\alpha ]}\). Since \(f_{\alpha }\) is \(\alpha \)-Symmetric. There is \(B^x_{[\alpha ]}\not \subset x_{[\alpha ]}^{\alpha }\), such that \(C_{[\alpha ]}\not \subset f_{\alpha }(B^x)_{[\alpha ]}\). Take \(E=\vee \{B^x\mid x\not \in A_{[\alpha ]}^{\alpha }\}\), then \(x\not \subset E_{[\alpha ]}^{\alpha }\). This implies \(E_{[\alpha ]}^{\alpha }\subset A_{[\alpha ]}^{\alpha }\), i.e., \(A_{[\alpha ]}\subset E_{[\alpha ]}\). Furthermore, we can conclude \(C_{[\alpha ]}\not \subset f_{\alpha }(A)_{[\alpha ]}\). Otherwise, if \(C_{[\alpha ]}\subset f_{\alpha }(A)_{[\alpha ]}\subset f_{\alpha }(E)_{[\alpha ]}\), then it contradicts with the statement: for each \(B^x\le E , C_{[\alpha ]}\not \subset f_{\alpha }(B^x)_{[\alpha ]}\).
Theorem 3.4
Let \(f_{\alpha }\in \fancyscript{F}_{\alpha }\). Then
Proof
\(\forall A\in L^X\),
By Theorem 2 (2), \(f_{\alpha }^{\nabla }\le f_{\alpha }\), we have
Therefore
Theorem 3.5
Let \(f_{\alpha }\in \fancyscript{F}_{\alpha }\). Then
-
(1)
\(f^{\nabla }_{\alpha }\le f_{\alpha }\odot f_{\alpha }\).
-
(2)
\(f_{\alpha }\odot f_{\alpha }\odot f_{\alpha }=f^{\diamond }_{\alpha }\).
Proof
(1) By Theorem 2, we have
(2) By Theorem 3 (2), we have
This shows \(f_{\alpha }\odot f_{\alpha }\odot f_{\alpha }\le f^{\diamond }_{\alpha }\). On the other hand, by Theorem 2 (2) and (1) above, it is easy to find \(f^{\diamond }_{\alpha }\le f^{\nabla }_{\alpha }\odot f_{\alpha }\le f_{\alpha }\odot f_{\alpha }\odot f_{\alpha }\). Therefore, (2) holds.
4 Characterizations of \(\alpha \)-Quasi Uniformities
Theorem 4.1
An non-empty subfamily \(\fancyscript{D}_{\alpha }\subset \fancyscript{F}_{\alpha }\) is an \(\alpha \)-uniform, iff \(\fancyscript{D}_{\alpha }\) satisfies:
-
(1)
\(f_{\alpha }\in \fancyscript{D}_{\alpha },g_{\alpha }\in \fancyscript{F}_{\alpha }\) with \(f_{\alpha }\le g_{\alpha }\), then \(g_{\alpha }\in \fancyscript{D}_{\alpha }\).
-
(2)
\(f_{\alpha },g_{\alpha }\in \fancyscript{D}_{\alpha }\) implies \(f_{\alpha }\vee g_{\alpha }\in \fancyscript{D}_{\alpha }\).
-
(3)
\(f_{\alpha }\in \fancyscript{D}_{\alpha }\), then \(\exists g_{\alpha }\in \fancyscript{D}_{\alpha }\), such that \(g^{\diamond }_{\alpha }\ge f_{\alpha }\).
Proof
Necessity. Since \(\fancyscript{D}_{\alpha }\) is an \(\alpha \)-uniform, (1) and (2) hold. Furthermore, if \(\fancyscript{B}_{\alpha }\subset \fancyscript{D}_{\alpha }\) is an \(\alpha \)-symmetric base. So for every \(f_{\alpha }\in \fancyscript{D}_{\alpha }\), there is \(g_{\alpha }\in \fancyscript{B}_{\alpha }\), such that \(g_{\alpha }\odot g_{\alpha }\odot g_{\alpha }\odot g_{\alpha }\ge f_{\alpha }\). By Theorem 5 (2), \(g^{\diamond }_{\alpha }=g_{\alpha }\odot g_{\alpha }\odot g_{\alpha } \ge g_{\alpha }\odot g_{\alpha }\odot g_{\alpha }\odot g_{\alpha }\ge f_{\alpha }\).
Sufficiency. If \(\fancyscript{B}_{\alpha }=\{g^{\nabla }_{\alpha }\mid g_{\alpha }\in \fancyscript{D}_{\alpha }\}\). For every \(f_{\alpha }\in \fancyscript{D}_{\alpha }\), there is \(g_{\alpha }\in \fancyscript{D}_{\alpha }\), such that \(g^{\diamond }_{\alpha }\ge f_{\alpha }\). By Theorem 4, \(g_{\alpha }\odot g_{\alpha }\ge g^{\diamond }_{\alpha }\ge f_{\alpha }\). Then \(\fancyscript{D}_{\alpha }\) is an \(\alpha \)-quasi-uniform. By Theorem 2 (2)and (4), we know \(g^{\nabla }_{\alpha }\ge g^{\diamond }_{\alpha }\ge f_{\alpha }\). This shows \(\fancyscript{B}_{\alpha }\) is an \(\alpha \)-symmetric base of \(\fancyscript{D}_{\alpha }\). Therefore \(\fancyscript{D}_{\alpha }\) is an \(\alpha \)-uniform.
Theorem 4.2
An non-empty subfamily \(\fancyscript{D}_{\alpha }\subset \fancyscript{F}_{\alpha }\) is an \(\alpha \)-uniform, iff \(\fancyscript{D}_{\alpha }\) satisfies:
-
(1)
\(f_{\alpha }\in \fancyscript{D}_{\alpha },g_{\alpha }\in \fancyscript{F}_{\alpha }\) with \(f_{\alpha }\le g_{\alpha }\), then \(g_{\alpha }\in \fancyscript{D}_{\alpha }\).
-
(2)
\(f_{\alpha },g_{\alpha }\in \fancyscript{D}_{\alpha }\) implies \(f_{\alpha }\vee g_{\alpha }\in \fancyscript{D}_{\alpha }\).
-
(3)
\(f_{\alpha }\in \fancyscript{D}_{\alpha }\), then \(\exists g_{\alpha }\in \fancyscript{D}_{\alpha }\), such that \(g^{\nabla }_{\alpha }\ge f_{\alpha }\).
Proof
Necessity. Since \(\fancyscript{D}_{\alpha }\) is an \(\alpha \)-uniform, (1) and (2) hold. By Theorem 6, For every \(f_{\alpha }\in \fancyscript{D}_{\alpha }\), there is \(g_{\alpha }\in \fancyscript{D}_{\alpha }\), such that \(g^{\diamond }_{\alpha }\ge f_{\alpha }\). So according to Theorem 2 (2), \(g^{\nabla }_{\alpha }\ge g^{\diamond }_{\alpha }\ge f_{\alpha }\). Thus (3) holds.
Sufficiency. If \(\fancyscript{B}_{\alpha }=\{g^{\nabla }_{\alpha }\mid g_{\alpha }\in \fancyscript{D}_{\alpha }\}\). By (3), For every \(f_{\alpha }\in \fancyscript{D}_{\alpha }\), there is \(g_{\alpha }\in \fancyscript{D}_{\alpha }\), such that \(g^{\nabla }_{\alpha }\ge f_{\alpha }\). Again, there is \(h_{\alpha }\in \fancyscript{D}_{\alpha }\), such that \(h^{\nabla }_{\alpha }\ge g_{\alpha }\). By Theorem 4, we get
So \(\fancyscript{B}_{\alpha }\) is an \(\alpha \)-symmetric base of \(\fancyscript{D}_{\alpha }\). Therefore, \(\fancyscript{D}_{\alpha }\) is an \(\alpha \)-uniform.
5 \(\alpha \)-P.Q. Metric and its Properties
A binary mapping \(d_{\alpha }: L^X\times L^X\rightarrow [0,+\infty )\) is called an \(\alpha \)-mapping, if \(\forall (A,B),(C,D)\in L^X\times L^X\) satisfying \(A_{[\alpha ]}=C_{[\alpha ]}\) and \(B_{[\alpha ]}=D_{[\alpha ]}\), then \(d_{\alpha }(A,B)=d_{\alpha }(A,B)\).
Definition 5.1
An \(\alpha \)-mapping \(d_{\alpha }: L^X\times L^X\rightarrow [0,+\infty )\) is called an \(\alpha \)-P.Q metric on \(L^X\), if
(\(\alpha \)-M1) \(d_{\alpha }(A,A)=0\).
(\(\alpha \)-M2) \(d_{\alpha }(A,C)\le d_{\alpha }(A,B)+d_{\alpha }(B,C)\).
(\(\alpha \)-M3) \(d_{\alpha }(A,B)=\bigwedge \limits _{C_{[\alpha ]} \subset B_{[\alpha ]}}d_{\alpha }(A,C)\).
Theorem 5.1
Let \(d_{\alpha }\) be an \(\alpha \)-P.Q. metrics on \(L^X\). \(\forall r\in (0,+\infty )\), a mapping \(P_r: L^X\rightarrow L^X\) is defined by \(\forall A\in L^X\),
Then
-
(1)
\(P_{\alpha }^r\) is \(\alpha \)-symmetric mapping.
-
(2)
\(\forall A,B\in L^X\), \(B_{[\alpha ]}\subset P_{\alpha }^r(A)_{[\alpha ]} \Leftrightarrow d_{\alpha }(A,B)\ge r\).
-
(3)
\(\forall A\in L^X, r>0\), \(A_{[\alpha ]}\not \subset P_{\alpha }^r(A)_{[\alpha ]}\).
-
(4)
\(\forall r,s\in (0,\infty )\), \(P^r_{\alpha }\odot P^s_{\alpha }\ge P^{r+s}_{\alpha }\).
-
(5)
\(\forall A\in L^X, r>0\), \(P_{\alpha }^r(A)_{[\alpha ]}=\bigcap \limits _{s<r}P_{\alpha }^s(A)_{[\alpha ]}\).
-
(6)
\(\forall A\in L^X\), \(\bigcap \limits _{r>0}P_{\alpha }^s(A)_{[\alpha ]}=\emptyset \).
Proof
Since \(d_{\alpha }\) is an \(\alpha \)-mapping, (1), (5) and (6) are easy.
(2) Clearly, \(d_{\alpha }(A,B)\ge r\Rightarrow B_{[\alpha ]}\subset P_{\alpha }^r(A)_{[\alpha ]}\). Conversely. \(\forall x\in B_{[\alpha ]}\), \(\exists x\in D_x\in L^X\), such that \(d_{\alpha }(A,D_x)\ge r\). So \(d_{\alpha }(A,\{x_{\alpha }\})\ge d_{\alpha }(A,D_x)\ge r\). Thereby \(d_{\alpha }(A,B)=\bigwedge \limits _{x\in B_{[\alpha ]}} d_{\alpha }(A,\{x_{\alpha }\})\ge r\).
(3) Suppose \(A_{[\alpha ]}\subset P_{\alpha }^r(A)_{[\alpha ]}\). By (2), we get \(d_{\alpha }(A,A)\ge r\). This is a contradiction with (\(\alpha \)-M1).
(4) \(\forall A,B\in L^X\), if \(B_{[\alpha ]}\not \subset P_{\alpha }^r\odot P_{\alpha }^s(A)_{[\alpha ]}\), then there is \(D\in L^X\), such that \(D_{[\alpha ]}\not \subset P^s_{\alpha }(A)_{[\alpha ]}\) and \(B_{[\alpha ]}\not \subset P^r_{\alpha }(D)_{[\alpha ]}\). By (2), we have \(d_{\alpha }(A,D)<s,d_{\alpha }(D,B)<r\). Then by (\(\alpha \)-M2), we gain \(d_{\alpha }(A,B)<r+s\). Therefore \(B_{[\alpha ]}\not \subset P_{\alpha }^{r+s}(A)_{[\alpha ]}\). This means \(P^r_{\alpha }\odot P^s_{\alpha }\ge P^{r+s}_{\alpha }\).
Theorem 5.2
If a family of \(\alpha \)-mappings \(\{P^r_{\alpha }\mid P_r: L^X\rightarrow L^X,r>0\}\) satisfies the conditions (2)–(5) in Theorem 8. For each \(A,B\in L^X\), let’s denote
Then
-
(1)
\(d_{\alpha }(A,B)<r\Leftrightarrow B_{[\alpha ]}\not \subset P_{\alpha }^r(A)_{[\alpha ]}\).
-
(2)
\(d_{\alpha }\) is \(\alpha \)-P.Q. metric on \(L^X\).
Proof
(1) By Theorem 8 (2),(5), we have
(2) \(\forall A,B\in L^X, r,s>0\), if \(d_{\alpha }(A,B)>r+s\), then
Thus \(\forall C\in L^X\), we have \(C_{[\alpha ]}\not \subset P^s_{\alpha }(A)_{[\alpha ]}\) and \(B_{[\alpha ]}\not \subset P^r_{\alpha }(C)_{[\alpha ]}\). this implies \(d_{\alpha }(A,C)>s\), and \(d_{\alpha }(A,B)>r\). Hence \(d_{\alpha }(A,B)+d_{\alpha }(A,C)>r+s\). Consequently, we obtain \(d_{\alpha }(A,B)+d_{\alpha }(A,C)\ge d_{\alpha }(A,B)\).
Theorem 5.3
An \(\alpha \)-mapping \(d_{\alpha }: L^X\times L^X\rightarrow [0,+\infty )\) satisfies (\(\alpha -M1\)),(\(\alpha -M2\)) and (\(\alpha -M3^{*}\)), then for each \(C_{[\alpha ]}\subset B_{[\alpha ]}\), \(d_{\alpha }(C,B)=0\).
(\(\alpha -M3^{*}) \forall A\in L^X , r>0, A_{[\alpha ]}\not \subset P_{\alpha }^r(A)_{[\alpha ]}\).
Proof
By (\(\alpha \)-M2), for each \(B,C\in L^X\), satisfying \(C_{[\alpha ]}\subset B_{[\alpha ]}\), we have \(d_{\alpha }(A,B)<d_{\alpha }(A,C)+d_{\alpha }(C,B)\). Here we can conclude \(d_{\alpha }(C,B)=0\). Otherwise, if \(d_{\alpha }(C,B)=s>0\), then \(B_{[\alpha ]}\subset P^s_{\alpha }(C)_{[\alpha ]}\). it contracts with \(C_{[\alpha ]}\not \subset P^s_{\alpha }(C)_{[\alpha ]}\) according to (\(\alpha \)-M3\(^{*}\)). Therefore \(d_{\alpha }(C,B)=0\).
Theorem 5.4
An \(\alpha \)-mapping \(d_{\alpha }: L^X\times L^X\rightarrow [0,+\infty )\) is an \(\alpha \)-P.Q metric on \(L^X\), iff \(d_{\alpha }\) satisfies (\(\alpha \)-M1),(\(\alpha \)-M2) and (\(\alpha \)-M3\(^*\)).
Proof
We only need to prove (\(\alpha \)-M3)\(\Leftrightarrow \)(\(\alpha \)-M3\(^*\)). By Theorem 8 (2), it is easy to check (\(\alpha \)-M3)\(\Rightarrow \)(\(\alpha \)-M3\(^*\)). If the converse result is not true, then there are \(r,s>0\), such that
so for each \(C_{[\alpha ]}\subset B_{[\alpha ]}\),
Thus, \(0<r-s<d_{\alpha }(B,C)\), which implies \(C_{[\alpha ]}\subset P^{r-s}_{\alpha }(B)_{[\alpha ]}\). As a result
However, it contradicts with (\(\alpha \)-M3\(^*\)). Therefore (\(\alpha \)-M3) holds.
Theorem 5.5
\(d_{\alpha }\) is \(\alpha \)-P.Q. metric on \(L^X\). Then \(\{P_{\alpha }^r\mid r>0\}\) satisfying (3)–(5) in Theorem 8 is an \(\alpha \)-base of some \(\alpha \)-uniform, which is called the \(\alpha \)-uniform induced by \(d_{\alpha }\).
Given an \(\alpha \)-quasi-uniform \(\fancyscript{D}_{\alpha }\), if we say \(\fancyscript{D}_{\alpha }\) is metricable, we mean there is an \(\alpha \)-P.Q. metric \(d_{\alpha }\), such that \(\fancyscript{D}_{\alpha }\) is induced by \(d_{\alpha }\).
Theorem 5.6
An \(\alpha \)-quasi-uniform spaces \((L^X,\fancyscript{D}_{\alpha })\) is \(\alpha \)-P.Q. metricable iff it has a countable \(\alpha \)-base.
Proof
By Theorem 12, the necessity is obvious. Let’s prove the sufficiency.
Let \(\fancyscript{D}_{\alpha }\) be an \(\alpha \)-uniformity on \(L^X\), which has a countable \(\alpha \)-base \(\fancyscript{B}_{\alpha }=\{P^n_{\alpha }\mid n\in N\}\). Let’s take \(g^1_{\alpha }=P^1_{\alpha }\), then there is \(g^2_{\alpha }\in \fancyscript{B}_{\alpha }\), such that \(g^2_{\alpha }\odot g^2_{\alpha }\odot g^2_{\alpha }\ge g^1_{\alpha }\vee P^2_{\alpha }\). In addition, there is \(g^3_{\alpha }\in \fancyscript{B}_{\alpha }\), such that \(g^3_{\alpha }\odot g^3_{\alpha }\odot g^3_{\alpha }\ge g^2_{\alpha }\vee P^3_{\alpha }\). The process can be repeated again and again, then \(\{g^n_{\alpha }\in n\in N\}\) is also an \(\alpha \)-base. Obviously, \(g^{n+1}_{\alpha }\odot g^{n+1}_{\alpha }\odot g^{n+1}_{\alpha } \ge g^n_{\alpha }\). Let’s take \(\varphi _{\alpha }:L^X\rightarrow L^X\), defined by: \(\forall A\in L^X\),
Clearly, \(\forall r>0\), \(A_{[\alpha ]}\not \subset \varphi ^r_{\alpha }(A)_{[\alpha ]}\). And \(\forall \frac{1}{2^n}<r\le \frac{1}{2^{n-1}}\), we have
Let’s define
Then \(\fancyscript{B}_{\alpha }=\{f^r_{\alpha }\mid r>0\}\subset \fancyscript{D}_{\alpha }\).
Finally, let’s prove \(\fancyscript{B}_{\alpha }\) is an \(\alpha \)-base of \(\fancyscript{D}_{\alpha }\) satisfying (3)-(6) in Theorem 8.
Step 1. For \(f_{\alpha }\in \fancyscript{D}\), there is \(n\in N\), such that \(g^n_{\alpha }\ge f_{\alpha }\). So if \(r\in (\frac{1}{2^{n+1}},\frac{1}{2^{n}}]\), then \(\varphi ^{2r}_{\alpha }=g^n_{\alpha }\). Besides, it is easy to check, \(\varphi ^{r_1}_{\alpha } \odot \varphi ^{r_2}_{\alpha }\odot \cdot \cdot \cdot \odot \varphi ^{r_k}_{\alpha }\ge \varphi ^{2r}_{\alpha }\) whenever \(\sum \limits _{i=1}^k r_i=r\). Thus \(f^r_{\alpha }\ge \varphi ^{2r}_{\alpha }=g^n_{\alpha }\ge f_{\alpha }\). This means \(\fancyscript{B}_{\alpha }\) be an \(\alpha \)-base of \(\fancyscript{D}_{\alpha }\).
Step 2. Obviously, \(f_{\alpha }\in \fancyscript{B}\) satisfies (3) and (6) in Theorem 8. Furthermore, for \(r,s>0, A\in L^X\), if there is \(B_{[\alpha ]}\not \subset (f^r_{\alpha }\odot f^s_{\alpha }(A))_{[\alpha ]}\). Then there is \(C\in L^X\), such that \(B_{[\alpha ]}\not \subset f^r_{\alpha }(C)_{[\alpha ]}\) and \(C_{[\alpha ]}\not \subset f^s_{\alpha }(A)_{[\alpha ]}\). So there are \(\sum \limits _{i=1}^k r_i=r\) and \(\sum \limits _{i=1}^m s_i=s\), such that
and
Thus
As a result, \(B_{[\alpha ]}\not \subset f^{r+s}_{\alpha }(A)_{[\alpha ]}\). Consequently, \(f^{r+s}_{\alpha }\le f^r_{\alpha }\odot f^s_{\alpha }\). This is the proof of (4) in Theorem 8.
Step 3. for \(r>s>0\), \(A\in L^X\),
Hence \(f^r_{\alpha }\le \bigwedge \limits _{s<r}f^s_{\alpha }\).
Conversely. Let’s prove the reverse result.
If \(B\in L^X\), \(B_{[\alpha ]}\not \subset f^r_{\alpha }(A)_{[\alpha ]}\), then there is \(\sum \limits _{i=1}^k r_i=r\), such that \(B_{[\alpha ]}\not \subset \varphi ^{r_1}_{\alpha } \odot \varphi ^{r_2}_{\alpha }\odot \cdot \cdot \cdot \odot \varphi ^{r_k}_{\alpha }(A)_{[\alpha ]}\). So there is \(C\in L^X\), such that \(C_{[\alpha ]}\not \subset (\varphi ^{r_2}_{\alpha } \odot \varphi ^{r_2}_{\alpha }\odot \cdot \cdot \cdot \odot \varphi ^{r_k}_{\alpha }(A)_{[\alpha ]}\) and \(B_{[\alpha ]}\not \subset \varphi ^{r_1}_{\alpha } (C)_{[\alpha ]}\). By \(\varphi ^{r_1}_{\alpha }=\bigwedge \limits _{t<r_1}\varphi ^t_{\alpha }\), there is \(t<r_1\), such that \(B_{[\alpha ]}\not \subset \varphi ^{t}_{\alpha } (C)_{[\alpha ]}\), i.e., \(B_{[\alpha ]}\not \subset \varphi ^{t}_{\alpha }\odot \varphi ^{r_2}_{\alpha }\odot \cdot \cdot \cdot \odot \varphi ^{r_k}_{\alpha }(A)_{[\alpha ]}\). Let’s take \(s=t+\sum \limits _{i=2}^k r_i\), we have \(s<r\) and \(B_{[\alpha ]}\not \subset f^{s}_{\alpha }(A)_{[\alpha ]}\). Therefore \(f^r_{\alpha }\ge \bigwedge \limits _{s<r}f^s_{\alpha }\). Therefore (5) in Theorem 8 holds.
Theorem 5.7
Each \(\alpha \)-\(C_{II}\) \(\alpha \)-layer topological space is P.Q.-metriclizable.
Proof
Let \((L^X,D_{\alpha }(\delta ))\) be an \(\alpha \)-\(C_{II}\) space, \(\{P_n\mid n\in N\}\) be an \(\alpha \)-base. \(\forall n\in N\), \(f^{P_n}_{\alpha }:L^X\rightarrow L^X\) is defined as: \(\forall A\in L^X\),
Let’s take \(\fancyscript{D}^*=\{f_{\alpha }^{P_n}\mid n\in N\}\) and
Then \(\fancyscript{B}_{\alpha }\) is an \(\alpha \)-base of some uniform denoted by \(\fancyscript{D}_{\alpha }\) and clearly, \(\eta _{\alpha }=\eta _{\alpha }(\fancyscript{D}_{\alpha })\). Furthermore, since \(\fancyscript{B}_{\alpha }\) is countable, we know \((L^X,D_{\alpha }(\delta ))\) is P.Q.-metriclizable.
Theorem 5.8
An \(\alpha \)-layer co-topology \(D_{\alpha }(\delta )\) on \(L^X\) can be \(\alpha \)-P.Q. metriclizable iff there is a Sequence of \(\alpha \)-remote neighborhood mappings \(\{f^n_{\alpha }\}_{n\in N}\) satisfying
-
(1)
\(\forall n\in N, f^n_{\alpha }\le f^{n+1}_{\alpha }\odot f^{n+1}_{\alpha }\odot f^{n+1}_{\alpha }\),
-
(2)
\(\forall a\in M^*(L^X)\), \(\{f^{n}_{\alpha }(a)\}_{n\in N}\) is the \(\alpha \)-remote neighborhood family of \(a\).
Proof
Necessary. If \(D_{\alpha }(\delta )\) can be \(\alpha \)-P.Q. metriclizable, there is an \(\alpha \)-P.Q. metric, say \(d_{\alpha }\). Let’s take \(f^n_{\alpha }=P^{\frac{1}{3^n}}_{\alpha }\). By Theorem 8, it is clear that (1) and (2) hold.
Sufficiency. If \(\{f^n_{\alpha }\}_{n\in N}\) satisfies (1) and (2). Clearly, \(\{f^n_{\alpha }\}_{n\in N}\) is countable. So by Theorem 12, it is an \(\alpha \)-base of some \(\alpha \)-quasi uniform \(\fancyscript{D}_{\alpha }\). Therefore \(D_{\alpha }(\delta )=\eta _{\alpha }(\fancyscript{D}_{\alpha })\). This means \(D_{\alpha }(\delta )\) is \(\alpha \)-P.Q. metriclizable.
Theorem 5.9
An \(\alpha \)-layer co-topology \(D_{\alpha }(\delta )\) on \(L^X\) can be \(\alpha \)-P.Q. metriclizable iff there is a Sequence of \(\alpha \)-symmetric remote neighborhood mappings \(\{f^n_{\alpha }\}_{n\in N}\) satisfying
-
(1)
\(\forall n\in N, f^n_{\alpha }\le f^{n+1}_{\alpha }\odot f^{n+1}_{\alpha }\odot f^{n+1}_{\alpha }\),
-
(2)
\(\forall a\in M^*(L^X)\), \(\{f^{n}_{\alpha }(a)\}_{n\in N}\) is the \(\alpha \)-remote neighborhood family of \(a\).
Theorem 5.10
An \(\alpha \)-layer co-topology \(D_{\alpha }(\delta )\) on \(L^X\) can be \(\alpha \)-P.Q. metriclizable iff there is a Sequence of \(\alpha \)-symmetric remote neighborhood mappings \(\{f^n_{\alpha }\}_{n\in N}\) satisfying
-
(1)
\(\forall n\in N, f^n_{\alpha }\le (f^{n+1}_{\alpha })^{\diamond }\le f^{n+1}_{\alpha }\),
-
(2)
\(\forall a\in M^*(L^X)\), \(\{f^{n}_{\alpha }(a)\}_{n\in N}\) is the \(\alpha \)-remote neighborhood family of \(a\).
Proof
Necessity. It is similar to that of Theorem 14.
Sufficiency. If \(\{f^n_{\alpha }\}_{n\in N}\) satisfies (1) and (2). By Theorem 2, \(\forall n\in N\), \(f^n_{\alpha }\le (f^{n+1}_{\alpha })^{\diamond }\le (f^{n+1}_{\alpha })^{\nabla }\le f^{n+1}_{\alpha }\). By Theorem 4, \(\forall n\in N\), \(f^n_{\alpha }\le (f^{n+1}_{\alpha })^{\diamond }\le f^{n+1}_{\alpha }\odot f^{n+1}_{\alpha }\).
Therefore \(\forall n\in N\),
Again, by Theorem 4,
\(\forall n\in N\), \(g_{\alpha }^n=(f^{4n-3}_{\alpha })^{\nabla }\). Then \(\{g_{\alpha }^n\}_{n\in N}\) is a family of \(\alpha \)-symmetric remote neighborhood mappings. It is clear that \(\{g^{n}_{\alpha }(a)\}_{n\in N}\) is the \(\alpha \)-remote neighborhood family of \(a\).
Theorem 5.11
An \(\alpha \)-layer co-topology \(D_{\alpha }(\delta )\) on \(L^X\) can be \(\alpha \)-P.Q. metriclizable iff there is a Sequence of \(\alpha \)-remote neighborhood mappings \(\{f^n_{\alpha }\}_{n\in N}\) satisfying
-
(1)
\(\forall n\in N, f^n_{\alpha }\le (f^{n+1}_{\alpha })^{\nabla }\le f^{n+1}_{\alpha }\),
-
(2)
\(\forall a\in M^*(L^X)\), \(\{f^{n}_{\alpha }(a)\}_{n\in N}\) is the \(\alpha \)-remote neighborhood family of \(a\).
References
Erceg, M.A.: Metric in fuzzy set theory. J. Math. Anal. Appl. 69, 205–230 (1979)
Hutton, B.: Uniformities on fuzzy topological spaces. J. Math. Anal. Appl. 58, 557–571 (1997)
Peng, Y.W.: Pointwise p.q. metric and the family of induced maps on completely distributive lattices. Ann. Math. 3, 353–359 (1992)
Shi, F.G.: Pointwise quasi-uniformities in fuzzy set theory. Fuzzy Sets Syst. 98, 141–146 (1998)
Shi, F.G.: Pointwise Pseudo-metrics in \(L\)-fuzzy set theory. Fuzzy Sets Syst. 121, 209–216 (2001)
Shi, F.G., Zhang, J., Zheng, C.Y.: \(L\)-proxinmities and totally bounded pointwise L-uniformities. Fuzzy Sets Syst. 133, 321–331 (2003)
Zhang, J.: Lattice valued smooth pointwise Quasi-uniformity on completely distributive lattice. Fuzzy Syst. Math. 17(2), 30–34 (2003)
Yue, Y., Fang, J.: Extension of Shi’s quaso-uniformities in a Kubiak-spstal sense. Fuzzy Sets Syst. 157, 1956–1969 (2006)
Yue, Y., Shi, F.G.: On (L, M)-fuzzy quasi-uniform spaces. Fuzzy Sets Syst. 158, 1472–1485 (2007)
Wu, X.Y.: \(L\)-fuzzy \(\alpha \)-quasi uniformtiy. Fuzzy Syst. Math. 3(3), 94–99 (2012)
Meng, G.W., Meng, H.: \(D\)-closed sets and their applications. Fuzzy Syst. Math. 17(1), 24–27 (2003)
Wang, G.J.: Theory of \(L\)-Fuzzy Topological Spaces. The Press of Shanxi normal University, Xi’an (1988)
Acknowledgments
Thanks to the support by: 1. National Science Foundation (No.10971125). 2. Science Foundation of Guangdong Province (No. 01000004) 3. The construct program of the key discipline in Hunan University of Science and Engineering.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2014 Springer-Verlag Berlin Heidelberg
About this chapter
Cite this chapter
Wu, XY., Xie, LL., Bai, SZ. (2014). Characterizations of \(\alpha \)-Quasi Uniformity and Theory of \(\alpha \)-P.Q. Metrics. In: Cao, BY., Nasseri, H. (eds) Fuzzy Information & Engineering and Operations Research & Management. Advances in Intelligent Systems and Computing, vol 211. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-38667-1_14
Download citation
DOI: https://doi.org/10.1007/978-3-642-38667-1_14
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-38666-4
Online ISBN: 978-3-642-38667-1
eBook Packages: EngineeringEngineering (R0)