Fuzzy Sets and Systems ( ) – www.elsevier.com/locate/fss
ଁ An approach to fuzzy frames via fuzzy posets Wei Yao∗
Department of Mathematics, Hebei University of Science and Technology, Shijiazhuang 050018, China
Received 10 October 2009; received in revised form 18 November 2010; accepted 19 November 2010
Abstract In this paper, based on a complete Heyting algebra L, we define a fuzzy version of frames, called L- frames. Then we construct an adjunction between the category of stratified L- topological spaces and that of L- locales (the opposite category of L- frames), which is a fuzzy counterpart of the Isbell-adjunction between topological spaces and locales. We also study the corresponding sobriety of stratified L- topologies and spatiality of L- frames and show the dual equivalence between them. © 2010 Published by Elsevier B.V.
Keywords: Topology; Adjunction; L- ordered set; (Stratified); L- topology; L- frame; L- locale; Sober space; Spatial L- frame
1. Introduction
A topological duality is a correspondence between two mathematical structures involving points and predicates such that isomorphic structures can be identified. Stone [29] firstly finds such a duality between topology and logic. In [16] Isbell gives an adjunction between the category of topological spaces with continuous functions and the opposite category of frames with frame homomorphism (which yields a duality between sober spaces and spatial frames). Stone duality in mathematical context is studied in a book of Johnstone [17]. Since topological ideas combined with those of fuzzy sets in 1968 [6], fuzzy topology has made a great progress now. It is natural to ask that whether or not it is possible to establish a category to play the same role with respect to a given notion of fuzzy topology as that locales play for the classical topological spaces. Rodabaugh is the first person to embark this question. In [21,22,25], he essentially captures the lattice theoretic behavior of the category L-Top of L-topology (not necessary stratified) by locales. The main result, as is stated in [34], is the fuzzification of the adjunction −→ pt via the introduction of L-fuzzy points of a crisp locale A, which are defined to be the frame morphisms from A to L. That is to say that he obtains an adjunction (L, Lpt):L-Top −→ Loc, where L takes every L-topological space to the locale of its open sets, and Lpt takes every locale A to the L-topological space of L-fuzzy points of A. Consequently, Rodabaugh establishes the Stone representation theorem for distributive lattices by means of this adjunction.
ଁ This paper is supported by the NNSF of China (10926055), the Foundation of Hebei Province (A2010000826, 09276158) and the Foundation of HEBUST (XL200821, QD200957). ∗ Tel.: +86 311 81668514. E-mail address: [email protected]
0165-0114/$- see front matter © 2010 Published by Elsevier B.V. doi:10.1016/j.fss.2010.11.010
Please cite this article as: W. Yao, An approach to fuzzy frames via fuzzy posets, Fuzzy Sets and Systems (2010), doi: 10.1016/j.fss.2010.11.010 2 W. Yao / Fuzzy Sets and Systems ( ) –
In [34], Zhang and Liu define a kind of an L-frame by a pair (A, i A), where A is a classical frame and i A : L −→ A is a frame morphism. For a stratified L-topological space (X, ), the pair (, iX ) is one of this kind of L-frames, where iX : L −→ is a map which sends a ∈ L to the constant map with the value a. Conversely, a point of an L-frame (A, i A) is frame morphism p :(A, i A) −→ (L, idL ) satisfying p ◦ i A = idL and Lpt(A) denotes the set of all points of (A, i A). Then {x : Lpt(A) −→ L|∀p ∈ Lpt(A), x (p) = p(x)} is a stratified L-topology on Lpt(A). By these two assignments, Zhang and Liu construct an adjunction between SL-Top and L-Loc and consequently establish the Stone representation theorem for distributive lattices by means of this adjunction. Zhang and Liu [34] point out that, from the viewpoint of lattice theory, Rodabaugh’s fuzzy version of Stone rep- resentation theory just is and has nothing differ from the classical one. While in our opinion, Zhang–Liu’s L-frames preserve many features of and also seem to have no strong difference from a crisp one. In [18,20] for L a complete chain, Pultr and Rodabaugh introduce a new approach to describing L-topological spaces using categorical constructs called lattice-valued frames or also L-frames. This approach not only gives new description of previously known type of sober spaces, but also leads naturally to a new type of sober spaces not previously documented in the literature. In [13], the kind of lattice-valued frames is extended from a complete chain to a completely distributive lattice. But for the case L = 2, their 2-Frm is categorical isomorphic to Frm2, not to Frm. In this paper, firstly we aim to define an L-frame by an L-ordered set equipped with some further conditions. For explicit, since a classical frame is a complete lattice satisfying the infinite distributive law of binary meets over arbitrary joins or the meet operation has a right adjoint, by means of fuzzy Galois connections in [32],anL-frame in this paper will be a complete L-ordered set with the meet operation having a right fuzzy adjoint. We then aim to establish an adjunction between the category of stratified L-topological spaces and the category of L-locales, the opposite category of this kind of L-frames. Corresponding sobriety of stratified L-topological spaces and spatiality of L-locales are defined and their duality is obtained.
2. Preliminaries
2.1. Category theory
For category theory, we refer to [1]. For two objects A, B in a category C,wewouldliketouse[A, B]C to denote the set of all C-morphism from A to B, and by |C| the class of C-objects and by Mor(C) the class of C-morphisms. Let F : A −→ B be a functor and B ∈|B|.Apair(u, A) with A ∈|A| and u : B −→ F(A) ∈ Mor(B) is called universal for B w.r.t. F provided that for each A ∈|A| and each B-morphism f : B −→ F(A) there exists a unique A-morphism f : A −→ A such that F( f )◦u = f . Dually, A pair (A, u) with A ∈|A| and u : F(A) −→ B ∈ Mor(B) is called co-universal for B w.r.t. F provided that for each A ∈|A| and each B-morphism f : F(A) −→ B there exists a unique A-morphism f : A −→ A such that u ◦ F( f ) = f . Let F : A −→ B and G : B −→ A be two functors. F is called a left adjoint of G (or G a right adjoint of F) or (F, G) is an adjunction between A, B, in symbols FٜG : AB, if for each A ∈|A|, there exists a universal pair (u A, F(A)) w.r.t. G (or equivalently, for each B ∈|B|, there exists a co-universal pair (G(B), u B) w.r.t. F).
2.2. Lattices
In this paper, if there is no further statement, L always denotes a complete Heyting algebra with , ⊥ as the top and bottom elements respectively. Thus there is an implication operation →: L × L −→ L induced by the binary meets ∧ on L, which is given by → = { ∈ L| ∧ ≤ }(∀, ∈ L). A complete Heyting algebra is equivalent to a frame, i.e., a complete lattice satisfying the distributive law of binary meets over arbitrary joins: ∧ = ∧ ∀ ∈ , ∀ ⊆ (IDL) a S s∈S(a s)( a L S L). Properties of complete Heyting algebras can be found in lots of literatures, e.g., [17]. Clearly, a complete lattice is a frame iff ∧ (−) has a right adjoint → (−)forany ∈ L. A frame morphism is a map preserves finite meets and arbitrary joins. Note that for a frame morphism f : A −→ B,wehave f () =
Please cite this article as: W. Yao, An approach to fuzzy frames via fuzzy posets, Fuzzy Sets and Systems (2010), doi: 10.1016/j.fss.2010.11.010 W. Yao / Fuzzy Sets and Systems ( ) – 3
(note that is the meet of empty set). The category of complete lattices with frame morphisms is called the category of semi-frames [23], denoted by SFrm, which firstly appeared in [22] under the symbol CSLF.AndFrm is its full subcategory of frames. Clearly, for objects a semi-frame has no difference from a complete lattice.
2.3. Fuzzy sets
For a set X, L X denotes the set of all L-fuzzy subsets of X.For ∈ L, the constant map with the value is X denoted by X . All operations on L can be extended onto L . For example, if L is a frame, define (A ⊗ B)(x) = A(x) ⊗ B(x)(∀A, B ∈ L X , x ∈ X)for⊗∈{∧, ∨, →},thenL X is also a frame. For f : X −→ Y an ordinary map, the usual forward and backward power set operators are respectively defined by f → :2X −→ 2Y , f →(A) ={f (x)|x ∈ A}(∀A ∈ 2X )and f ← :2Y −→ 2X , f ←(B) ={x ∈ X| f (x) ∈ B} (∀B ∈ Y → X −→ Y 2 ). Then their L-counterparts are fL : L L (called L-valued Zadeh function or L-forward powerset operator, ← Y −→ X → = cf. [24])and fL : L L (called L-backward powerset operator, cf. [24])by fL (A)(y) f (x)=y A(x)for ∈ X ∈ ← = ◦ ∈ Y A L and y Y ,and fL (B) B f for B L , respectively. −→ → X −→ Y ⊆ X → → ⊆ Foranordinarymap f : X Y , consider fL : L L as an ordinary map too, if L and ( fL ) ( ) ⊆ Y → −→ ← Y −→ X L , then we still use fL : to denote the restriction. Similarly, for the usual map fL : L L ,if ⊆ Y ← → ⊆ ⊆ X ← −→ L and ( fL ) ( ) L , then we still use fL : to denote the restriction.
2.4. (Stratified) L-topology
For materials related to fuzzy topology, we refer to [15]. X An L-topology on X is a family ⊆ L satisfying that (O1) X , ⊥X ∈ ; (O2) for any A, B ∈ , A ∧ B ∈ ; { | ∈ }⊆, ∈ ∈ (O3) for any Ai i I i Ai .AnL-topology on X is called stratified if it satisfies that (OS) X for any ∈ L. For an (resp., a stratified) L-topology on X, the pair (X, ) is called an (resp., a stratified) L-topological space. Amap f : X −→ Y is called continuous (resp., open) with respect to two given L-topological spaces (X, X )and , ← ∈ ∈ → ∈ ∈ (Y Y )iff fL (B) X for all B Y (resp., fL (A) Y for all A X ). The category of L-topological spaces with continuous maps is denoted by L-Top and by SL-Top its full subcategory of all stratified L-topological spaces. Amap f :(X, X ) −→ (Y, Y ) is called a homeomorphism between two L-topological spaces if f is a bijection −1 and both f :(X, X ) −→ (Y, Y ), f :(Y, Y ) −→ (X, X ) are continuous. Since for a bijection f : X −→ Y ,it → = −1 ← , −→ , is easy to show that fL ( f )L , we have that a bijection f :(X X ) (Y Y ) is a homeomorphism iff f is a categorical isomorphism in L-Top iff f is both open and continuous (cf. Theorem 5.1.2 in [23]).
2.5. L-ordered sets
Based on a (complete) Heyting algebra, Fan and Zhang [11,33] have studied quantitative domains under the frame- work of fuzzy set theory. Their approach firstly defines a fuzzy partial order on a non-empty set. Based on a (complete) residuated lattice L, in order to study fuzzy relational systems, Bˇelohlávek [2,3] has defined and studied an L-ordered set. In fact, a fuzzy partial order in the sense of Fan–Zhang (when being extended onto a residuated lattice) and an L-order in the sense of Bˇelohlávek are equivalent to each other (see Section 3 in [30]). Let P be a set and e : P × P −→ L be a map. The pair (P, e)iscalledanL-ordered set if for all x, y, z ∈ P, (E1) e(x, x) =; (E2) e(x, y) ∧ e(y, z) ≤ e(x, z); (E3) e(x, y) ∧ e(y, x) =implies x = y.
Let (P, ≤) be a classical poset. Then (P, ≤)isanL-ordered set, where ≤ is the characteristic function of ≤.Foran L-ordered set (P, e), ≤e={(x, y)| e(x, y) =}is a crisp partial order on X.If(P, e)isanL-ordered set and Q ⊆ P, then the restriction of e on Q × Q, still denoted by e (since there is no confusion will arise), is an L-order on Q.
Example 2.1. (1) Define eL : L × L −→ L by eL (x, y) = x → y for all x, y ∈ L.Then(L, eL )isanL-ordered set [2,33].
Please cite this article as: W. Yao, An approach to fuzzy frames via fuzzy posets, Fuzzy Sets and Systems (2010), doi: 10.1016/j.fss.2010.11.010 4 W. Yao / Fuzzy Sets and Systems ( ) – , ∈ X , = → (2) Let X be a set. For any S T L , the subsethood degree [12] of S in T is defined by subX (S T ) x∈X S(x) X , ≤ ≤ X T (x). Then (L subX )isanL-ordered set and subX is just the pointwise order on L [2].
P Definition 2.2. Let (P, e)beanL-ordered set and S ∈ L .Anelementx0 ∈ P is called a join (resp., meet) of S,in symbols x0 =S (resp., x0 =S), if for all x ∈ P, ≤ , ≤ , (J1) S(x) e(x x0) (resp., (M1) S(x) e(x0 x)); → , ≤ , → , ≤ , (J2) y∈X S(y) e(y x) e(x0 x) (resp., (M2) y∈X S(y) e(x y) e(x x0)).
Remark 2.3. (1) It is easy to verify by (E3) that if x1, x2 are two joins (or meets) of S,thenx1 = x2. That is, each S ∈ L P has at most one join (or meet). (2) In [4,5],forT a t-norm on the unit interval [0,1], Bodenhofer introduces and studies T .E-partial orders w.r.t. a T -equivalence E.In[8,9], Demirci extends Bodenhofer’s T–E-partial orders from [0,1] to an integral commutative complete quasi-monoidal-lattice [7] (L, ∗, ≤) and then studies vague lattices for L an integral, commutative cl-monoid. Proposition 3.2 in [31] shows that an L-ordered set is exactly an L–E-poset satisfying the condition that there exists at most one join (or meet) for any L-fuzzy subsets.
The following definitions and results can be found in [2–5,7–9,11,33]. Let (P, e)beanL-ordered set and S ∈ L P . Su ∈ L P is defined by ∀x ∈ P, Su(x) = S(y) → e(y, x). y∈P
Sl ∈ L P is defined by ∀x ∈ P, Sl (x) = S(y) → e(x, y). y∈P
For all S ∈ L P ,supS ∈ L P (resp., inf S ∈ L P )isdefinedby
∀x ∈ P, sup S = Su ∧ Sul (resp., inf S = Sl ∧ Slu).
The value Su(x) (resp., Sl (x)) can be interpreted as the degree of x being an upper (resp., a lower) bound of S.More general definitions of sups and infs of an L-fuzzy subset in L–E-ordered sets can be found in [8,9].
Proposition 2.4. Let (P, e) be an L-ordered set and S ∈ L P . Then = = ∈ , , = → , (1) x0 Siffsup S(x0) 1 iff for all x P e(x0 x) y∈P S(y) e(y x). = = ∈ , , = → , (2) x0 Siffinf S(x0) 1 iff for all x P e(x x0) y∈P S(y) e(x y).
u Proof. Some materials of the proof can be found in [2,3,11,33]. We here give a detailed proof for (1). Clearly, S (x0) = 1 ul u is equivalent to the condition (J1). And S (x0) = 1iffforanyx ∈ P, S (x) ≤ e(x0, x) iff (J2) holds. Hence x0 =S iff sup S(x0) = 1. P In the following, we only need to show that (J1) and (J2) hold for S ∈ L and x0 ∈ P,orsupS(x0) = 1ifffor u u u any x ∈ P, e(x0, x) = S (x). For the right direction, by (J2) we have e(x0, x) ≥ S (x)(∀x ∈ P). And S (x0) ≥ sup S(x0) = 1 and then for any x ∈ P, u u u S (x) = S (x0) → S (x) ≥ (S(y) → e(y, x0)) → (S(y) → e(y, x)) y∈P ≥ e(y, x0) → e(y, x) y∈P
≥ e(x0, x).
Please cite this article as: W. Yao, An approach to fuzzy frames via fuzzy posets, Fuzzy Sets and Systems (2010), doi: 10.1016/j.fss.2010.11.010 W. Yao / Fuzzy Sets and Systems ( ) – 5
u For the left direction, (J2) is equivalent to that S (x) ≤ e(x0, x)(∀x ∈ P). And for any x ∈ P, S(x) → e(x, x0) ≥ u S (x0) = e(x0, x0) = 1andthenS(x) ≤ e(x, x0), (J1) also holds. ç
Definition 2.5. An L-ordered set (P, e)iscalledcompleteifforallS ∈ L P , S and S exist. , = ∧ = For example, (L eL ) is a complete L-ordered set (cf. Example 3.4(1)), where S ∈L S( ) and S → ∈ L , , ≤ ∈L S( ) for all S L .If(P e) is a complete L-ordered set, then (P e) is a complete lattice, where = = ⊆ S S and S S for any S P.
Proposition 2.6 (Yao and Lu [32, Theorem 2.10 and Corollary 2.11]). For an L -ordered set (P, e), the following state- ments are equivalent: (1) (P, e) is complete. (2) For any S ∈ L P , Sexists. (3) For any S ∈ L P , Sexists.
The following proposition will be used later. −→ ∈ X → = ∧ Proposition 2.7. Let X be a set and f : X L be a map. Then for any S L , fL (S) x∈X f (x) S(x). → = → ∧ = ∧ = ∧ ç Proof. fL (S) a∈L fL (S)(a) a a∈L f (x)=a S(x) a x∈X f (x) S(x).
3. L-semi-frames and L-frames
In this section, we will introduce L-frames and L-semi-frames by means of L-ordered sets. Amap f :(P, eP ) −→ (Q, eQ) between two L-ordered sets is called monotone if for all x, y ∈ P, eP (x, y) ≤ eQ( f (x), f (y)).
Definition 3.1 (Yao [30], Yao and Lu [32]). Let (P, eP ), (Q, eQ)betwoL-ordered sets and f : P −→ Q, g : Q −→ P two monotone maps. The pair ( f, g) is called a fuzzy Galois connection between P and Q if
eQ( f (x), y) = eP (x, g(y)) for all x ∈ P, y ∈ Q,where f is called the fuzzy left adjoint of g and dually g the fuzzy right adjoint of f .
Proposition 3.2 (Yao [30], Yao and Lu [32]). Let f :(P, eP ) −→ (Q, eQ) and g :(Q, eQ) −→ (P, eP ) be two monotone maps. Then = → ∈ P (1) If P is complete, then f is monotone and has a fuzzy right adjoint if and only if f ( S) fL (S) for all S L . = → ∈ Q (2) If Q is complete, then g is monotone and has a fuzzy left adjoint if and only if g( T ) gL (T ) for all T L .
Definition 3.3. Let (P, e) be a complete L-ordered set and ∧ be the meet operation on (P, ≤e) such that for any a ∈ P, the map ∧a() = a ∧ (), b ∧a (b) = a ∧ b is monotone. We call (P, e)anL-frame if for any a ∈ P, ∧a has a fuzzy right adjoint, or equivalently, the following identity holds: ∧ =∧ → ∀ ∈ , ∀ ∈ P (FIDL) a( S) ( a)L (S)( a P S L ).
The condition (FIDL) could be called the fuzzy infinite distributive law of binary meets over arbitrary (fuzzy) joins, which is the fuzzy counterpart of (IDL). Clearly, for L = 2, (FIDL)=(IDL). For an arbitrary frame L,ifS is a crisp ∧ → { ∧ | ∈ } , subset of P,then( a)L (S) turns to the ordinary subset a s s S . It follows that if (P e)isanL-frame, then (P, ≤e) is a crisp one. But the converse need not be true as the following example shows.
Example 3.4. (1) The frame L itself is an L-frame with respect to eL . (2) Let be a stratified L-topology on X.Then(, subX )isanL-frame.
Please cite this article as: W. Yao, An approach to fuzzy frames via fuzzy posets, Fuzzy Sets and Systems (2010), doi: 10.1016/j.fss.2010.11.010 6 W. Yao / Fuzzy Sets and Systems ( ) –
(3) Suppose that L ={⊥, , , } is a diamond lattice, that is ⊥ < , < and , . Clearly, (L, ≤)isan ordinary frame, but (L, ≤) is neither an L-frame nor a complete L-ordered set. , ∈ L = ∧ ∈ Proof. (1) Step 1: (L eL ) is complete, where for any S L ,wehave S ∈L S( ) . In fact, for any L, ⎛ ⎞ ⎝ ⎠ eL S() ∧ , = S() → ( → ) = S() → eL (, ). ∈L ∈L ∈L
Step 2: For a ∈ L,define∧a : L −→ L by ∧a() = a ∧ . On one hand, for any , ∈ L,
eL (∧a(), ∧a()) = (a ∧ ) → (a ∧ ) ≥ → = eL (, ). ∧ , −→ , ∈ L ∧ = ∧ ∧ = Then a :(L eL ) (L eL ) is monotone. On the other hand, for any S L , a( S) a ∈L S( ) ∧ ∧ ∈ ∈L S( ) a .Andforany L, ∧ → → , = → → ( a)L (S)( ) eL ( ) S( ) ( ) ∈L ∈L a∧= = ((S() ∧ a ∧ ) → ) ∈L
=∧a(S) →
= eL (∧a(S), ).
Thus, the condition (FIDL) follows from Proposition 2.4(1). , A ∈ A = A ∧ A ∧ ∈ (2) Step 1: ( subX ) is complete, where for any L , U∈( (U))X U. Obviously, U∈( (U))X U since is stratified. And for any V ∈ , ⎛ ⎞ ⎛ ⎞ ⎝ ⎠ ⎝ ⎠ subX (A(U))X ∧ U, V = (A(U))X ∧ U (x) → V (x) U∈ x∈X U∈ = ((A(U) ∧ U(x)) → V (x)) x∈X U∈ = A(U) → U(x) → V (x) U∈ x∈X = A(U) → subX (U, V ). U∈ A = A ∧ Thus U∈( (U))X U. Step 2: For any U ∈ ,let∧U : −→ be defined by ∧U (V ) = U ∧ V (∀V ∈ ). On one hand, for any V1, V2 ∈ , subX (∧U (V1), ∧U (V2)) = (U(x) ∧ V1(x)) → (U(x) ∧ V2(x)) ≥ V1(x) → V2(x) = subX (V1, V2). x∈X x∈X
∧ , −→ , A ∈ ∧ A =∧ → A Then U :( subX ) ( subX ) is monotone. On the other hand, for any L ,wehave U ( ) ( U )L ( ), ∧ → −→ ∧ where ( U )L : L L is the L-forward operator of U . In fact, ⎛ ⎞ ⎝ ⎠ ∧U (A) = U ∧ (A(V ))X ∧ V = U ∧ (A(V ))X ∧ V V ∈ V ∈
Please cite this article as: W. Yao, An approach to fuzzy frames via fuzzy posets, Fuzzy Sets and Systems (2010), doi: 10.1016/j.fss.2010.11.010 W. Yao / Fuzzy Sets and Systems ( ) – 7 and for any W ∈ , ∧ → A → , = A → → ( U )L ( )(W1) subX (W1 W) (V ) (W1(x) W(x)) ∧ = ∈ W1∈ W1∈ U V W1 x X = ((A(V ) ∧ U(x) ∧ V (x)) → W(x)) x∈X V ∈ ⎛ ⎞ = ⎝ A(V ) ∧ U(x) ∧ V (x)⎠ → W(x) x∈X V ∈ = ∧U (A)(x) → W(x) x∈X
= subX (∧U (A), W).
(3) Put S = L , that is the constant map with the value .ThenS has no join in the L-ordered set (L, ≤). In fact, if = ∈ , = → , = → , = a S,thenwehaveforany b L, ≤(a b) c∈L S(c) ≤(c b) ( c∈L ≤(c b)). Put b a,wehave = , = → , ≤ , = = 1 ≤(a a) ( c∈L ≤(c a)) and c∈L ≤(c a), which implies that a ,thatis S. Now put b = ,wehave ⊥= ≤(, ) = → ≤(c, ) = →⊥=. c∈L It is impossible. ç
We call a map f :(A, eA) −→ (B, eB) between two complete L-ordered sets an L-frame homomorphism if f : , ≤ −→ , ≤ A ∈ (A eA ) (B eB ) is a frame morphism and f preserves joins of arbitrary L-fuzzy subset of A (i.e., for any A, A = → A L f ( ) fL ( )). The category of complete L-ordered sets (resp., L-frames) and L-frame homomorphisms is called the category of L-semi-frames (resp., L-frames), denoted by L-SFrm (resp., L-Frm). The opposite category is denoted by L-SLoc (resp., L-Loc). Obviously, L-Frm (resp., L-Loc) is a full subcategory of L-SFrm (resp., L-SLoc).
, −→ , ← , −→ , Proposition 3.5. Let f :(X ) (Y ) be a morphism in SL-Top. Then fL :( subY ) ( subX ) is a morphism in L-SFrm.
Proof. Obviously, f ← :(, ≤) −→ (, ≤) is a morphism in SFrm. We only need to show that f ← preserves arbitrary L L joins of L-fuzzy subsets of (, subY ). In fact, for A ∈ L ,foranyU ∈ ,wehave ← → A = A . ( fL )L ( )(U) (V ) ← = fL (V ) U Then ← → A = ← → A ∧ ( fL )L ( ) (( fL )L ( )(U))X U U∈ ⎛ ⎞ = ⎝ A(V )⎠ ∧ U ∈ f ←(V )=U U L X = (A(V ))X ∧ U ∈ ← = U fL (V ) U = A ∧ ← ( (V ))X fL (V ) V ∈
Please cite this article as: W. Yao, An approach to fuzzy frames via fuzzy posets, Fuzzy Sets and Systems (2010), doi: 10.1016/j.fss.2010.11.010 8 W. Yao / Fuzzy Sets and Systems ( ) – and ⎛ ⎞ ← A = ← ⎝ A ∧ ⎠ = ← A ∧ ← = A ∧ ← . fL ( ) fL ( (V ))Y V fL (( (V ))Y ) fL (V ) ( (V ))X fL (V ) V ∈ V ∈ V ∈ ← = ∈ ← A = ← → A ç Notice that fL ( Y ) X for all L. Hence fL ( ) ( fL )L ( ).
Let M be a distributive lattice. A map F : M −→ L is called an L-fuzzy filter of M if F(a ∧ b) = F(a) ∧ F(b)for a, b ∈ M. Denote the set of all L-fuzzy filters on M by FL (M). Obviously, for any ∈ L, M ∈ FL (M).
Lemma 3.6 (See in Zhang and Liu [34] for L-fuzzy ideals). Let :(M, ≤) −→ (L, ≤) be a monotone map. Then ∗ ∈ L M defined by ∗ (m) = (a)(∀m ∈ M) A∈[m] a∈A is the minimal L-fuzzy filter of M which is larger than or equal to (under pointwise order), where [m] ={A ⊆ M|A is finite and m ≥ A}.
∗ ∗ Proof. (1) is an L-fuzzy filter. Since it is easily shown that :(M, ≤) −→ (L, ≤) is monotone, we only need to ∗ ∗ ∗ show that for any m , m ∈ M, (m ) ∧ (m ) ≤ (m ∧ m ). In fact, 1 2 1 2 1 2 ∗ ∗ (m1) ∧ (m2) = (a) ∧ (b) A∈[m1],B∈[m2] a∈A,b∈B ≤ (c)
A∪B∈[m1∧m2] c∈A∪B ≤ (c)
C∈[m1∧m2] c∈C ∗ = (m1 ∧ m2). ∗ ∗ (2) Minimality of . Clearly, ≥ since {m}∈[m]foranym ∈ M. Suppose that F is an L-fuzzy filter of M such ∗ that F ≥ . Then for any m ∈ M, (m) = ∈ ∈ (a) ≤ ∈ ∈ F(a) = ∈ F( A) ≤ F(m) ∗ A [m] a A A [m] a A A [m] and hence ≤ F. ç
F F , A ∈ L (M) A = Proposition 3.7. ( L (M) subM ) is an L-frame, where for any L , we have ( F∈F (M) ∗ L (A(F))M ∧ F) .