On the embedding of convex spaces in stratified L-convex spaces

Consider L being a continuous lattice, two functors from the category of convex spaces (denoted by CS) to the category of stratified L-convex spaces (denoted by SL-CS) are defined. The first functor enables us to prove that the category CS can be embedded in the category SL-CS as a reflective subcategory. The second functor enables us to prove that the category CS can be embedded in the category SL-CS as a coreflective subcategory when L satisfying a multiplicative condition. By comparing the two functors and the well known Lowen functor (between topological spaces and stratified L-topological spaces), we exhibit the difference between (stratified L-)topological spaces and (stratified L-)convex spaces.

are some complete lattices. When L = 2, an (L, M)-convexity is precisely an M-fuzzifying convexity; when M = 2, an (L, M)-convexity is precisely an L-convexity; and when L = M = 2, an (L, M)-convexity is precisely a convexity. Similar to (lattice-valued) topology, the categorical relationships between convexity and latticed-valued convexity is an important direction of research. When L being a completely distributive complete lattices with some additional conditions, Pang and Shi (2016) proved that the category of convex spaces can be embedded in the category of stratified L-convex spaces as a coreflective subcategory.
In this paper, we shall continue to study the categorical relationships between convex spaces and stratified L-convex spaces. We shall investigate two embedding functors from the category of convex spaces (denoted by CS) to the category of stratified L-convex spaces (denoted by SL-CS). The first functor enables us to prove that the category CS can be embedded in the category SL-CS as a reflective subcategory when L being a continuous lattice. The second functor enables us to prove that the category CS can be embedded in the category SL-CS as a coreflective subcategory when the continuous lattice L satisfying a multiplicative condition. The second functor is an extension of Pang and Shi's functor (2016) from the lattice-context. Precisely, from completely distributive complete lattice to continuous lattice. And the second functor can be regarded as an analogizing of the well known (extended) Lowen functor between the category of topological spaces and the category of stratified L-topological spaces (Höle and Kubiak 2007;Lai and Zhang 2005;Li and Jin 2011;Lowen 1976;Warner 1990;Yue and Fang 2005). By comparing the two functors and Lowen functor, we exhibit the difference between (stratified L-)topological spaces and (stratified L-)convex spaces from the categorical sense.
The contents are arranged as follows. In "Preliminaries" section, we recall some basic notions as preliminary. In "CS reflectively embedding in SL-CS" section, we present the reflective embedding of the category CS in the category SL-CS. In "CS coreflectively embedding in SL-CS" section, we focus on the coreflective embedding of the category CS in the category SL-CS. Finally, we end this paper with a summary of conclusion.

Preliminaries
Let L = (L, ≤, ∨, ∧, 0, 1) be a complete lattice with 0 is the smallest element, 1 is the largest element. For a, b ∈ L, we say that a is way below (Gierz et al. 2003). For a directed subset D ⊆ L, we use ∨ ↑ D to denote its union.
Throughout this paper, L denote a continuous lattice, unless otherwise stated. The continuous lattice has a strong flavor of theoretical computer science (Gierz et al. 2003). The following lemmas collect some properties of way below relation on a continuous lattice.
Lemma 1 (Gierz et al. 2003 Lemma 2 (Gierz et al. 2003) Let L be a continuous lattice and let {a j,k | j ∈ J , k ∈ K (j)} be a nonempty family of element in L such that {a j,k | k ∈ K (j)} is directed for all j ∈ J. Then the following identity holds.
where N is the set of all choice functions h : Let X be a nonempty set, the functions X −→ L, denoted as L X , are called the L-subsets on X. The operators on L can be translated onto L X in a pointwise way. We make no difference between a constant function and its value since no confusion will arise. For a crisp subset A ⊆ X, we also make no difference between A and its characteristic function χ A . Clearly, χ A can be regarded as an L-subset on X. Let f : X −→ Y be a function.
For a nonempty set X, let 2 X denotes the powerset of X.
Definition 1 (Van De Vel 1993) A subset C of 2 X is called a convex structure on X if it satisfies: The pair (X, C) is called a convex space. A mapping f : (X, C X ) −→ (Y , C Y ) is called convexity-preserving (CP, in short) provided that B ∈ C Y implies f −1 (B) ∈ C X . The category whose objects are convex spaces and whose morphisms are CP mappings will be denoted by CS.
Definition 2 (Maruyama 2009; Pang and Shi 2016) A subset C of L X is called an L-convex structure on X if it satisfies: The category whose objects are stratified L-convex spaces and whose morphisms are L-CP mappings will be denoted by SL-CS.
Definition 3 (Adámek et al. 1990) Suppose that A and B are concrete categories;

CS reflectively embedding in SL-CS
In this section, we shall present a functor from the category CS to the category SL-CS, and then by using it to prove that the category CS can be embedded in the category SL-CS as a reflective subcategory.
At first, we fix some notations. For a ∈ L, x ∈ X, we denote x a as the L-subset values a at x and values 0 otherwise. For ∈ L X , let pt( ) = {x a | a ≪ (x)} and let Fin( ) denote the set of finite subset of pt( ). Obviously, = ∨pt( ) = ∨{∨F | F ∈ Fin( )}.

Definition 4 Let (X, C) be a convex space and let
Proof (LCS). It is obvious.
(LC2). It is easily seen that B is closed for the operator ∧.
. It follows by Lemma 1 (4) that a ≪ j xa (x) for some j x α ∈ J. Since { j } j∈J is directed then there exists a j ∈ J, denote as j F , such that j xa ≤ j F for all j x a . By Lemma 1 (2) we get a ≪ j F (x). This shows that F ∈ Fin( j F ). By a similar discussion on j F we have that F ∈ Fin(µ j F ,k F ) for some k F ∈ K (j F ). It follows that ∨F ≤ µ j F ,k F .
We have proved that for any F ∈ Fin( ), there exists a µ j,k such that ∨F ≤ µ j,k . Because B is closed for ∧, we get that σ is definable.

Lemma 3 Let
Proof The sufficiency is obvious. We check the necessity. Let χ U ∈ ω 1 L (C). Then with ∀j ∈ J , a j ∈ L, U j ∈ C and {a j ∧ U j } j∈J is directed. Without loss of generality, we assume that a j � = 0 for all j ∈ J. It is easily seen that {U j } j∈J is directed. In the following we check that On one hand, it is obvious that U ⊇ U j for any j ∈ J and so U ⊇ ↑ j∈J U j . On the other hand, for any x ∈ U we have which means x ∈ U j for some j ∈ J. Thus U ⊆ ↑ j∈J U j as desired.
It follows that f : (X, ω 1 Conversely, let f : It is easily seen that the correspondence (X, C) � → (X, ω 1 L (C)) defines an embedding functor Proposition 2 (Pang and Shi 2016) Let (X, C) be a stratified L-convex space. Then the set ρ L (C) = {U ∈ 2 X | U ∈ C} forms a convex structure on X and the correspondence (X, C) � → (X, ρ L (C)) defines a concrete functor Theorem 1 The pair (ρ L , ω 1 L ) is a Galois correspondence and ρ L is a left inverse of ω 1 L .
Proof It is sufficient to show that ρ L • ω 1 L (C) = C for any (X, C) ∈ CS and ω 1 L • ρ L (C) ⊆ C for any (X, C) ∈ L-CS.
with ∀j ∈ J , a j ∈ L, U j ∈ C and {a j ∧ U j } j∈J is directed. It follows by the definition of stratified L-convex space that ∈ C.

CS coreflectively embedding in SL-CS
In this section, we shall give a functor from the category CS to the category SL-CS, and then by using it to prove that the category CS can be embedded in the category SL-CS as a coreflective subcategory. This functor extends Pang and Shi's functor (2016) from the lattice-context. Precisely, from completely distributive complete lattice to continuous lattice.
Firstly, we fix some notations used in this section. Let ∈ L X and a ∈ L. Then the set [a] := {x ∈ X| a ≤ (x)} and the set (a) := {x ∈ X| a ≪ (x)} are called the a-cut and strong a-cut of , respectively. Let a, b ∈ L, we say that a is wedge below b (in symbol, a ⊳ b) if for all subsets D ⊆ L, y ≤ ∨D always implies that x ≤ d for some d ∈ D. For each a ∈ L, denote β(a) = {b ∈ L| b ⊳ a}.
The following lemma generalizes Huang and Shi's result from lattice-context. Huang and Shi (2008) defined (a) := {x ∈ X| a ⊳ (x)} and assumed that L being completely distributive complete lattice.
The way below relation ≪ on L is called multiplicative (Gierz et al. 2003

Lemma 6 Assume that the way below relation ≪ on L is multiplicative. Then for any
∈ L X and any a ∈ L, the set In addition, by the multiplicative condition we have a ≪ b ∧ c. This proves that Pang and Shi (2016) proved a similar result when L being a completely distributive complete lattice with the condition β(a ∧ b) = β(a) ∩ β(b) for any a, b ∈ L. It is easily seen that this condition is equivalent to that the wedge below relation on L is multiplicative. Definition 5 Let (X, C) be a convex space and the way below relation ≪ on L be multiplicative. Then the set ω 2 L (C) defined below is a stratified L-convex structure on X, Proof The proofs of (LCS) and (LC2) are obvious. We only check (LC3) below.
Let { j } j∈J ⊆ ω 2 L (C) be directed and a ∈ L. Then It follows immediately that j∈J ↑ j ∈ ω 2 L (C) by ( j ) [c] ∈ C for any j ∈ J , c ∈ L and C being closed for intersection and directed union.
Lemma 4(1) Similar to Pang and Shi (2016), we can prove that the correspondence (X, C) � → (X, ω 2 L (C)) defines an embedding functor Let (X, C) be a stratified L-convex space. Then Pang and Shi (2016) defined ι L (C) as the finest convex structure on X which contains all [a] for all ∈ C, a ∈ L. They proved that the correspondence (X, C) � → (X, ι L (C)) defined a concrete functor Similar to Pang and Shi (2016), when the way below relation ≪ on L being multiplicative, we get the following results.

Theorem 2
The pair (ω 2 L , ι L ) is a Galois correspondence and ι L is a left inverse of ω 2 L .

Corollary 2
The category CS can be embedded in the category SL-CS as a coreflective subcategory.
Remark 1 Let us replace the convex space (X, C) in Definition 4 and Definition 5 with a topological space (X, T ). Then ω 2 L defines an embedding functor from the category of topological spaces to the category of stratified L-topological spaces. This functor was first proposed by Lowen (1976) for L = [0, 1] and then extended by many researchers (Höle and Kubiak 2007;Lai and Zhang 2005;Liu and Luo 1997;Wang 1988;Warner 1990). If we further remove the directed condition in ω 1 L then we also get an embedding functor from the category of topological spaces to the category of stratified L-topological spaces. By the definition of stratified L-topology, it is easily seen that ω 1 L (T ) ⊆ ω 2 L (T ). Conversely, if ∈ ω 2 L (T ) then = a∈L (a ∧ [a] ) ∈ ω 1 L (T ). Thus ω 1 L = ω 2 L and it follows the following well known result. That is, the category of topological spaces can be embedded in the category of stratified L-topological spaces as a both reflective and coreflective subcategory.
Remark 2 Does CS can be embedded in L-CS as a both reflective and coreflective subcategory? Now, we can not answer it. For a convex space (X, C), the inclusion ω 1 L (C) ⊆ ω 2 L (C) holds obviously. But the reverse inclusion seems do not hold. The reason is that for an L-subset ∈ L X , the set {a ∧ [a] | a ∈ L} is generally not directed.
At last, we give two interesting examples to distinguish (L-)convex space from (L-) topological spaces. Example 1 An upper set U on L is called Scott open if for each directed set D ⊆ L, ∨ ↑ D ∈ U implies that d ∈ U for some d ∈ D. It is known that the Scott open sets on L form a topology L, called the Scott topology (Gierz et al. 2003). It is not difficult to check that the Scott open sets on L do not form a convex structure on L since they are not closed for intersection.
Example 2 An L-filter (Höhle and Rodabaugh 1999) on a set X is a function F : L X −→ L such that for all , µ ∈ L X , (F1) The set of L-filters on X is denoted by F L (X). Since F L (X) is a subset of L (L X ) , hence, there is a natural partial order on F L (X) inherited from L (L X ) . Precisely, for F, G ∈ F L (X) , It is known that F L (X) is closed for intersection, but is not closed for union (Fang 2010;Jäger 2001). In the following, we check that F L (X) is closed for directed union.
Let {F j } j∈J ⊆ F L (X) be directed. Then it is readily seen that ↑ j∈J F j satisfies the conditions (F1) and (F2). Taking , µ ∈ L X , then Thus ↑ j∈J F j satisfies the condition (F3). We have proved that F L (X) is closed for directed union.
1. Let Y = L X and C = {0, 1} ∪ F L (X). Then it is easily seen that C is an L-convex structure on Y but not an L-topology on Y. 2. If we call a function F : L X −→ L satisfying (F2) and (F3) as a nearly L-filter on X.
Let F N L (X) denote the set of nearly L-filters on X. Then it is easily seen that F N L (X) is a stratified L-convex structure on Y but not a stratified L-topology on Y.
Note that L X forms a continuous lattice. If replacing L X with a continuous lattice M, similar to (1)-(2), we can define (stratified) L-convex structure on M.

Conclusions
When L being a continuous lattice, an embedding functor from the category CS to SL-CS is introduced, then it is used to prove that the category CS can be embedded in the category SL-CS as a reflective subcategory. When L being a continuous lattice with a multiplicative condition, Pang and Shi's functor (2016) is generalized from the lattice context, then it is used to prove that the category CS can be embedded in the category SL-CS as a coreflective subcategory. It is well known that the category of topological spaces can be embedded in the category of stratified L-topological spaces as a both reflective and coreflective subcategory. But, we find that the category of convex spaces seem not be embedded in the category of stratified L-convex spaces as a both reflective and coreflective subcategory. This shows the difference between (stratified L-)topological spaces and (stratified L-) convex spaces from categorical sense.