betaFeedback

WorksMathematics

On the structural aspects of Convex Subhyperlattices under different orders

Abstract

Nayak et al. (2025) introduced the concept of convex subhyperlattices of a strong join hyperlattice L\mathfrak{L} and demonstrated that the collection of all convex subhyperlattices, CS⁡(L)\operatorname{CS}(\mathfrak{L}), together with empty set, forms a lattice. In particular, for PP-hyperlattices and Nakano hyperlattices, they obtained a unique representation of convex subhyperlattices in terms of hyperideals and hyperfilters. In the present work, we extend these studies by proving that CS⁡(L)∪{∅}\operatorname{CS}(\mathfrak{L}) \cup \{\emptyset\}, under set inclusion, forms an algebraic lattice. Moreover, we introduce two new orders on CS⁡(L)\operatorname{CS}(\mathfrak{L}) and examine their properties, including conditions for equivalence of these orders. In the class of PP-hyperlattices, we characterize convex subhyperlattices in terms of initial and final segments, providing a structural insight into their formation. When convexity is extended to arbitrary join hyperlattices, many results known for PP-hyperlattices and Nakano hyperlattices fail to hold. To address these limitations of convexity in arbitrary hyperlattices, we introduce the notion of strongly convex sets, denoted SC⁡(L)\operatorname{SC}(\mathfrak{L}), which naturally generalizes the concept while preserving its fundamental properties and extends the applicability of results to a wider setting. We establish several results for SC⁡(L)\operatorname{SC}(\mathfrak{L}) that do not hold for CS⁡(L)\operatorname{CS}(\mathfrak{L}).

Introduction and Preliminaries

Hyperoperations serve as a broad generalization of binary operations. Building on this idea, Konstantinidou and Mittas [9] extended the classical notion of lattices by replacing the binary join and meet operations with corresponding hyperoperations. In [12], the authors introduced and investigated various generalizations of prime hyperideals, such as 2-absorbing and primary 2-absorbing hyperideals, within the framework of meet hyperlattices. Bideshki et al. [4] introduced the ideas of strong meet hyperlattices, hyperideals, and hyperfilters in strong meet hyperlattices. They also studied distributivity and dual distributivity in meet hyperlattices. Ameri et al. [1] examined residual mappings in meet hyperlattices through the framework of principal downsets and established specific conditions under which a mapping between two hyperlattices corresponds to a residual mapping. They further explored the structure and properties of convex subsets in meet hyperlattices. Later, Nayak et al. [11] introduced two kinds of hyperfilters in join hyperlattices and explored their relationships within PP-hyperlattices and Nakano hyperlattices. They also examined convex subhyperlattices and provided a unique representation for them in terms of hyperideals and hyperfilters within PP-hyperlattices and Nakano hyperlattices. Lavanya and Parameshwara Bhatta [10] investigated the collection of all convex sublattices of a lattice LL under a partial order distinct from inclusion. They showed that the lattice of convex sublattices under this new order and the lattice under consideration belong to the same equational class. Extending their ideas, Bhargava and Hegde [3] introduced the concept of strongly convex subtrellises in trellises, a generalization of lattices, and generalized several corresponding results. Since hyperlattices are also natural generalizations of lattices, it is meaningful to ask whether these developments can be extended to the framework of hyperlattices. Nayak et al. [11] established that the set CS⁡(L)\operatorname{CS}(\mathcal{L}) of all convex subhyperlattices of a join hyperlattice L\mathcal{L}, including the empty set, forms a lattice under inclusion. In the present study, we demonstrate that CS⁡(L)∪{∅}\operatorname{CS}(\mathcal{L}) \cup\{\emptyset\} is an algebraic lattice. Furthermore, we introduce a new order on CS⁡(L)\operatorname{CS}(\mathcal{L}), defined for X,Y∈CS⁡(L)X,Y \in\operatorname{CS}(\mathcal{L}) such that X≤bYX \le_b Y if and only if, for every element x∈Xx \in X, there exists y∈Yy \in Y such that x⪯yx \preceq y, and for every y∈Yy \in Y, there exists x∈Xx \in X such that x⪯yx \preceq y. Under this relation, CS⁡(L)\operatorname{CS}(\mathcal{L}) forms a partially ordered set. We also define another order on CS⁡(L)\operatorname{CS}(\mathcal{L}) using hyperideals and hyperfilters, and show that these two orders are equivalent in PP-hyperlattices, though not necessarily in general, as illustrated through examples. Although convexity is well defined in strong hyperlattices, its direct extension to arbitrary hyperlattices is inadequate for preserving key results. The absence of unique representation of convex subhyperlattices motivates the need for the introduction of strongly convex sets. In the last section, we introduce strongly convex sets and discuss some of their properties.

Throughout this paper we use the following notations:

  • ∨\vee is the join of two elements in a lattice LL.

  • ∨⌣\mathbin{\mathop{\vee}\limits^{\scriptscriptstyle\smile}} is the hyperoperation join.

  • ⋁\bigvee is the classical join of a subset of a lattice LL.

  • ⪯\preceq is the partial order on the hyperlattice L\mathcal{L}.

  • ≺\prec is the covering relation in a classical lattice.

Definition 1.1 ([5]). Let HH be a non-empty set and let P∗(H)\mathcal{P}^{*}(H) be the set of all non-empty subsets of HH. A hyperoperation on HH is a map ∘:H×H→P∗(H)\circ: H \times H \to\mathcal{P}^{*}(H) and (H,∘)(H,\circ) is called a hypergroupoid.

If AA and BB are non-empty subsets of HH and ⋎\curlyvee is a hyperoperation, then

A⋎B=⋃a∈A,b∈Ba⋎b,A⋎x=A⋎{x}andx⋎B={x}⋎B.A \curlyvee B = \bigcup_{\substack{a \in A,b \in B}} a \curlyvee b,\qquad A \curlyvee x = A \curlyvee\{x\}\quad\text{and}\quad x \curlyvee B = \{x\} \curlyvee B.

Definition 1.2 ([15]). Let LL be a non-empty set, ⋎:L×L→P∗(L)\curlyvee: L \times L \to\mathcal{P}^{*}(L) be a hyperoperation, and ∧:L×L→L\wedge: L \times L \to L be a binary operation. Then (L,⋎,∧)(L,\curlyvee,\wedge) is a join hyperlattice if for all x,y,z∈Lx,y,z \in L the following conditions hold:

i. x∈x⋎xx \in x \curlyvee x and x=x∧xx = x \wedge x.

ii. x⋎(y⋎z)=(x⋎y)⋎zx \curlyvee(y \curlyvee z) = (x \curlyvee y) \curlyvee z and x∧(y∧z)=(x∧y)∧zx \wedge(y \wedge z) = (x \wedge y) \wedge z.

iii. x⋎y=y⋎xx \curlyvee y = y \curlyvee x and x∧y=y∧xx \wedge y = y \wedge x.

iv. x∈x∧(x⋎y)∩x⋎(x∧y)x \in x \wedge(x \curlyvee y) \cap x \curlyvee(x \wedge y).

We say LL is a strong join hyperlattice if whenever y∈x⋎yy \in x \curlyvee y then x=x∧yx = x \wedge y.

Throughout, L\mathcal{L} denotes a strong join hyperlattice unless otherwise specified, and ⋎\curlyvee and ∨\vee denote the join hyperoperation and the classical join operation respectively.

Definition 1.3 ([13]). Define the relation ⪯\preceq on L\mathcal{L} as follows:

l1⪯l2 if and only if l1∧l2=l1 if and only if l2∈l1⋎l2.l_{1} \preceq l_{2}\ \text{if and only if}\ l_{1} \wedge l_{2} = l_{1}\ \text{if and only if}\ l_{2} \in l_{1} \curlyvee l_{2}.

Then (L,⪯)(\mathcal{L},\preceq) is poset.

Henceforth, ⪯\preceq will denote a partial order on the strong hyperlattice L\mathcal{L}.

A non-empty subset HH of L\mathcal{L} is a subhyperlattice if HH is closed under the hyperoperation ⋎\curlyvee and the operation ∧\wedge in L\mathcal{L} [14]. A non-empty subset JJ of L\mathcal{L} which is closed under ⋎\curlyvee is a hyperideal if a∈La \in\mathcal{L}, b∈Jb \in J with a⪯ba \preceq b implies that a∈Ja \in J [15]. A non-empty subset FF of L\mathcal{L} which is closed under ∧\wedge is a hyperfilter

i. type-I, if whenever a∈Fa \in F, x∈Lx \in\mathcal{L} with a⪯xa \preceq x implies x∈Fx \in F [15].

ii. type-II if a1∈Fa_{1} \in F, x∈Lx \in\mathcal{L} with a1⪯xa_{1} \preceq x implies a1⋎x⊆Fa_{1} \curlyvee x \subseteq F [11].

iii. type-III if a∈Fa \in F, x∈Lx \in\mathcal{L} implies that a⋎x⊆Fa \curlyvee x \subseteq F [11].

Definition 1.4 ([13]). Let A⊆LA \subseteq\mathcal{L}. The intersection of all hyperideals of L\mathcal{L} containing AA is called the hyperideal generated by AA and it is denoted by ⟨A]\langle A]. If A={a}A = \{a\}, then we call the hyperideal generated by AA a principal hyperideal and denote it by ⟨a]\langle a].

Definition 1.5. The intersection of all type-III hyperfilters of L\mathcal{L} containing AA is called the type-III hyperfilter generated by AA (or simply, the hyperfilter generated by AA).

In the sequel, we use the notation ⟨A]\langle A] to denote the hyperideal generated by AA and [A⟩[A\rangle to denote the (type-III) hyperfilter generated by AA. The set of all hyperideals of L\mathcal{L} forms a lattice I(L)I(\mathcal{L}) under inclusion, where the meet and join operations are given as: for J1,J2∈I(L)J_1,J_2 \in I(\mathcal{L}), J1∧J2=J1∩J2J_1 \wedge J_2 = J_1 \cap J_2 and J1∨J2=⟨J1∪J2]J_1 \vee J_2 = \langle J_1 \cup J_2]. Similarly, the set of all type-III hyperfilters forms a lattice F(L)F(\mathcal{L}) under reverse inclusion where the meet and join operations are given as: for J1,J2∈F(L)J_1,J_2 \in F(\mathcal{L}), J1∧J2=[J1∪J2⟩J_1 \wedge J_2 = [J_1 \cup J_2\rangle and J1∨J2=J1∩J2J_1 \vee J_2 = J_1 \cap J_2.

Theorem 1.6 [8]. Let (L,∧,∨)(L,\wedge,\vee) be lattice and PP be a non-empty subset of LL. We define a hyperoperation ∨P\mathop{\vee}\limits^{P} by

l1∨Pl2=l1∨l2∨P={l1∨l2∨p∣p∈P},l_1 \mathop{\vee}\limits^{P} l_2 = l_1 \vee l_2 \vee P = \{l_1 \vee l_2 \vee p \mid p \in P\},

for all l1,l2∈Ll_1,l_2 \in L. Then (L,∨P,∧)(L,\mathop{\vee}\limits^{P},\wedge) is a join hyperlattice if and only if for each l1∈Ll_1 \in L there exists a p∈Pp \in P such that p⪯l1p \preceq l_1. In this case we call LL a PP-hyperlattice. Also, LL is a strong hyperlattice.

Definition 1.7 [14]. Let (L,∧,∨)(L,\wedge,\vee) be a modular lattice. For any a,b∈La,b \in L, define the Nakano hyperoperation a⊔b={z∈L:z∨a=z∨b=a∨b}a \sqcup b = \{z \in L : z \vee a = z \vee b = a \vee b\}. Then (L,⊔,∧)(L,\sqcup,\wedge) is a strong join hyperlattice.

We call the hyperlattice (L,⊔,∧)(L,\sqcup,\wedge) a Nakano join hyperlattice.

Lemma 1.8 [11]. Intersection of hyperfilters of one type is again a hyperfilter of the same type, provided it is non-empty.

Proposition 1.9 [11].

  • ii In a join PP-hyperlattice, all the three types of hyperfilters are equivalent.

  • iiii In a Nakano join hyperlattice, the type-II and type-III hyperfilters are equivalent.

Definition 1.10 [11]. A subhyperlattice KK of L\mathcal{L} is a convex subhyperlattice if for any p1,p2∈Kp_1,p_2 \in K, p3∈Lp_3 \in\mathcal{L} with p1⪯p3⪯p2p_1 \preceq p_3 \preceq p_2 implies that p3∈Kp_3 \in K.

Proposition 1.11 [11]. For any L\mathcal{L},

  • ii All three types of hyperfilters of L\mathcal{L} are convex subsets of L\mathcal{L}.

  • iiii Every hyperideal of L\mathcal{L} is a convex subhyperlattice.

  • iiiiii Every type-III hyperfilter of L\mathcal{L} is a convex subhyperlattice.

Proposition 1.12 [11]. Let {Ci}i∈τ\{C_i\}_{i \in\tau} be a collection of convex subhyperlattices of L\mathcal{L}. Then Co=⋂i∈τCiC_o = \bigcap_{i \in\tau} C_i is a convex subhyperlattice provided CoC_o is non-empty.

Lemma 1.13 [11]. Let CS(L)CS(\mathcal{L}) be the set of all convex subhyperlattices of L\mathcal{L}. Then (CS(L)∪{∅},⊆)(CS(\mathcal{L}) \cup\{\emptyset\},\subseteq) forms a lattice under inclusion order.

Theorem 1.14 [11]. Let II be a hyperideal and FF be a type-III hyperfilter of L\mathcal{L}. Then I∩FI \cap F is a convex subhyperlattice provided it is non-empty.

Theorem 1.15 [11]. In a join PP-hyperlattice, any convex subhyperlattice can be uniquely written as the intersection of a hyperideal and a hyperfilter.

On the lattice (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L}) \cup\{\varnothing\}, \subseteq)

The authors in [11], have established that (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L}) \cup\{\varnothing\}, \subseteq) is a lattice. In this section, we explore various properties of the lattice (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L}) \cup\{\varnothing\}, \subseteq) in detail. We derive the expression for the convex subhyperlattice generated by a non-empty subset of a join hyperlattice, and characterize the compact elements of (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L}) \cup\{\varnothing\}, \subseteq). Using these results, we show that the lattice (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L}) \cup\{\varnothing\}, \subseteq) is algebraic.

Definition 2.1. For a non-empty subset AA of a join hyperlattice L\mathcal{L}, the convex subhyperlattice generated by AA is denoted by ⟨A⟩\langle A\rangle and is defined as the intersection of all convex subhyperlattices containing AA.

For a non-empty subset AA of a PP-hyperlattice L\mathcal{L}, the convex subhyperlattice generated by AA is given by ⟨A⟩=⟨A]∩[A⟩\langle A\rangle= \langle A] \cap[A\rangle, where ⟨A]\langle A] denotes the hyperideal generated by AA and [A⟩[A\rangle denotes the hyperfilter generated by AA.

If A={a}A = \{a\}, we write ⟨a⟩\langle a\rangle instead of ⟨{a}⟩\langle\{a\}\rangle.

Remark 2.2. For non-empty subsets H1H_1 and H2H_2 of a PP-hyperlattice L\mathcal{L},

i. We observe that ⟨H1]∩[H1⟩\langle H_1] \cap[H_1\rangle is the smallest convex subhyperlattice containing H1H_1. Indeed, ⟨H1]∩[H1⟩\langle H_1] \cap[H_1\rangle is itself a convex subhyperlattice containing H1H_1. If G1G_1 is any convex subhyperlattice containing H1H_1, then ⟨H1]⊆⟨G1]\langle H_1] \subseteq\langle G_1] and [H1⟩⊆[G1⟩[H_1\rangle\subseteq[G_1\rangle. Hence, ⟨H1]∩[H1⟩⊆⟨G1]∩[G1⟩=G1\langle H_1] \cap[H_1\rangle\subseteq\langle G_1] \cap[G_1\rangle= G_1.

ii. H1H_1 is a convex subhyperlattice if and only if ⟨H1⟩=H1\langle H_1\rangle= H_1.

iii. If H1⊆H2H_1 \subseteq H_2, then ⟨H1⟩⊆⟨H2⟩\langle H_1\rangle\subseteq\langle H_2\rangle.

Remark 2.3. Let (L,∨,∧)(\mathcal{L}, \vee, \wedge) be any join hyperlattice and a∈La \in\mathcal{L}. Then {a}\{a\} is a convex subhyperlattice if and only if a∨a={a}a \vee a = \{a\}.

Proposition 2.4. For any L\mathcal{L}, (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L}) \cup\{\varnothing\}, \subseteq) forms a complete lattice where for X,Y∈CS⁡(L)X,Y \in\operatorname{CS}(\mathcal{L}),

X∧Y=X∩YX \wedge Y = X \cap Y
X∨Y=⟨X∪Y⟩X \vee Y = \langle X \cup Y\rangle

Proof. This follows from the fact that the intersection of any collection of convex subhyperlattices of L\mathcal{L} is either empty or a convex subhyperlattice.

Lemma 2.5. Let (L,∨,∧)(\mathcal{L}, \vee, \wedge) be a strong join hyperlattice. If {Xα}\{X_\alpha\} is a chain in (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L}) \cup\{\varnothing\}, \subseteq), then X=⋃αXαX = \bigcup_\alpha X_\alpha is a convex subhyperlattice of L\mathcal{L}.

Proof. Let x,y∈Xx,y \in X and a∈La \in\mathcal{L} such that x⪯a⪯yx \preceq a \preceq y. Since {Xα}\{X_\alpha\} is a chain, we get x,y∈Xβx,y \in X_\beta for some β\beta. By the convexity of XβX_\beta we then have a∈Xβ⊆Xa \in X_\beta\subseteq X. Hence XX is a convex set. Next, let a,b∈Xa,b \in X. Then a,b∈Xαa,b \in X_\alpha for some α\alpha. Now, XαX_\alpha is a subhyperlattice implies that a∧b∈Xα⊆Xa \wedge b \in X_\alpha\subseteq X and a∨b⊆Xα⊆Xa \vee b \subseteq X_\alpha\subseteq X. Therefore XX is a subhyperlattice.

Proposition 2.6. For a non-empty subset AA of L\mathcal{L},

⟨A⟩=⋃{⟨F⟩∣F⊆A,F is finite}.\langle A\rangle= \bigcup\{\langle F\rangle\mid F \subseteq A, F\text{ is finite}\}.

Proof. The proof is by transfinite induction on ∣A∣|A|. If AA is finite, then the result holds trivially. Now, suppose that AA is infinite and the result holds for all sets of cardinality smaller than ∣A∣|A|. We well-order the set AA in the order type of its cardinal. Then for each a∈Aa \in A, the initial segment S(a)={x∈A∣x<a}S(a)=\{x \in A \mid x<a\} determined by aa has cardinality strictly less than ∣A∣|A|. Furthermore, as AA is infinite, its cardinal is a limit ordinal and hence it follows that

A=⋃a∈AS(a)(1)A=\bigcup_{a\in A}S(a) \tag*{(1)}

Let C={⟨S(a)⟩∣a∈A}C=\{\langle S(a)\rangle\mid a\in A\}. Let a,b∈Aa,b\in A. Since AA is a chain with respect to ≤\leq, we get a≤ba\leq b or b≤ab\leq a. Without loss of generality, assume that a≤ba\leq b. Then S(a)⊆S(b)S(a)\subseteq S(b) and hence, ⟨S(a)⟩⊆⟨S(b)⟩\langle S(a)\rangle\subseteq\langle S(b)\rangle. This shows that CC is a chain with respect to ⊆\subseteq. Let C=⋃C=⋃a∈A⟨S(a)⟩C=\bigcup C=\bigcup_{a\in A}\langle S(a)\rangle. Then by Lemma 2.5, CC is a convex subhyperlattice. Now, for every a∈Aa\in A, since S(a)⊆⟨S(a)⟩⊆CS(a)\subseteq\langle S(a)\rangle\subseteq C, it follows from equation (1) that A⊆CA\subseteq C. Hence ⟨A⟩⊆⟨C⟩=C\langle A\rangle\subseteq\langle C\rangle=C because C∈CS⁡(L)C\in\operatorname{CS}(\mathcal{L}). Also for any a∈Aa\in A, since S(a)⊆AS(a)\subseteq A, we get ⟨S(a)⟩⊆⟨A⟩\langle S(a)\rangle\subseteq\langle A\rangle which implies C⊆⟨A⟩C\subseteq\langle A\rangle. Thus we get ⟨A⟩=C\langle A\rangle=C. By induction hypothesis, for each a∈Aa\in A, we have

⟨S(a)⟩=⋃{⟨F⟩∣F⊆S(a), F is finite}\langle S(a)\rangle=\bigcup\{\langle F\rangle\mid F\subseteq S(a),\ F\text{ is finite}\}

Thus, ⟨A⟩=⋃a∈A⟨S(a)⟩=⋃{⟨F⟩∣F⊆A, F is finite}\langle A\rangle=\bigcup_{a\in A}\langle S(a)\rangle=\bigcup\{\langle F\rangle\mid F\subseteq A,\ F\text{ is finite}\}, as any finite subset FF of AA is contained in S(a)S(a) for some a∈Aa\in A.

□\square

Definition 2.7 ([6]). An element pp of a lattice LL is compact if whenever p≤⋁Sp\leq\bigvee S for a subset SS of LL, then p≤⋁Fp\leq\bigvee F for a finite subset FF of SS.

Definition 2.8 ([7]). A complete lattice LL is said to be algebraic (compactly generated) if every element of LL is a join of some compact elements of LL.

Lemma 2.9. Let S={x1,x2,…,xn}S=\{x_1,x_2,\ldots,x_n\} be a non-empty finite subset of L\mathcal{L}. Then

⟨S⟩=⋁i=1n⟨xi⟩.\langle S\rangle=\bigvee_{i=1}^{n}\langle x_i\rangle.

Proof. We know that xi∈⟨xi⟩x_i\in\langle x_i\rangle for each ii, 1≤i≤n1\leq i\leq n. This gives S⊆⋃i=1n⟨xi⟩S\subseteq\bigcup_{i=1}^{n}\langle x_i\rangle and hence ⟨S⟩⊆⟨⋃i=1n⟨xi⟩⟩=⋁i=1n⟨xi⟩\langle S\rangle\subseteq\left\langle\bigcup_{i=1}^{n}\langle x_i\rangle\right\rangle=\bigvee_{i=1}^{n}\langle x_i\rangle. Since xi∈Sx_i\in S, we have ⟨xi⟩⊆⟨S⟩\langle x_i\rangle\subseteq\langle S\rangle for each ii. This implies ⋃i=1n⟨xi⟩⊆⟨S⟩\bigcup_{i=1}^{n}\langle x_i\rangle\subseteq\langle S\rangle. Thus ⋁i=1n⟨xi⟩=⟨⋃i=1n⟨xi⟩⟩⊆⟨S⟩\bigvee_{i=1}^{n}\langle x_i\rangle=\left\langle\bigcup_{i=1}^{n}\langle x_i\rangle\right\rangle\subseteq\langle S\rangle. Hence

⟨S⟩=⋁i=1n⟨xi⟩.\langle S\rangle=\bigvee_{i=1}^{n}\langle x_i\rangle.

□\square

Proposition 2.10. For any (L,∨,∧)(\mathcal{L},\vee,\wedge), the complete lattice (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L})\cup\{\emptyset\},\subseteq) is algebraic. The set of all compact elements of this lattice is

{⟨F⟩∣F is a finite subset of L}\{\langle F\rangle\mid F\text{ is a finite subset of }\mathcal{L}\}

Proof. Let F={f1,f2,…,fn}F=\{f_1,f_2,\ldots,f_n\} be a finite subset of L\mathcal{L}, for some n∈Nn\in\mathbb{N}. We show that ⟨F⟩\langle F\rangle is compact in (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L})\cup\{\varnothing\},\subseteq). Suppose ⟨F⟩⊆⋁α∈IXα=⟨⋃α∈IXα⟩\langle F\rangle\subseteq\bigvee_{\alpha\in\mathcal{I}}X_\alpha=\left\langle\bigcup_{\alpha\in\mathcal{I}}X_\alpha\right\rangle for some sub collection {Xα∣α∈I}\{X_\alpha\mid\alpha\in\mathcal{I}\} in CS⁡(L)\operatorname{CS}(\mathcal{L}). By Proposition 2.6, we have

⟨⋃α∈IXα⟩=⋃{⟨S⟩∣S is finite subset of ⋃α∈IXα}.\left\langle\bigcup_{\alpha\in\mathcal{I}}X_\alpha\right\rangle = \bigcup\{\langle S\rangle\mid S\text{ is finite subset of }\bigcup_{\alpha\in\mathcal{I}}X_\alpha\}.

Since fi∈F⊆⟨F⟩f_i\in F\subseteq\langle F\rangle for each ii, there exists a finite subset SiS_i of ⋃α∈IXα\bigcup_{\alpha\in\mathcal{I}}X_\alpha such that fi∈⟨Si⟩f_i\in\langle S_i\rangle. Each SiS_i is a finite subset of ⋃α∈IXα\bigcup_{\alpha\in\mathcal{I}}X_\alpha. So, there exists a finite subset Ii⊆I\mathcal{I}_i\subseteq\mathcal{I} such that Si⊆⋃α∈IiXαS_i\subseteq\bigcup_{\alpha\in\mathcal{I}_i}X_\alpha. Put ΛF=⋃j=1nIj\Lambda_F=\bigcup_{j=1}^{n}\mathcal{I}_j. Then ΛF\Lambda_F is a finite subset of I\mathcal{I}. We have fj∈⟨Sj⟩f_j\in\langle S_j\rangle, which implies

F⊆⋃j=1n⟨Sj⟩⊆⟨⋃j=1nSj⟩F\subseteq\bigcup_{j=1}^{n}\langle S_j\rangle \subseteq \left\langle\bigcup_{j=1}^{n}S_j\right\rangle
⊆⟨⋃j=1n⋃α∈IjXα⟩=⟨⋃α∈ΛFXα⟩.\begin{aligned} \subseteq \left\langle\bigcup_{j=1}^{n}\bigcup_{\alpha\in\mathcal{I}_j}X_\alpha\right\rangle\\ &= \left\langle\bigcup_{\alpha\in\Lambda_F}X_\alpha\right\rangle. \end{aligned}

Hence, ⟨F⟩⊆⟨⋃α∈ΛFXα⟩=⋁α∈ΛFXα\langle F\rangle\subseteq\left\langle\bigcup_{\alpha\in\Lambda_F}X_\alpha\right\rangle=\bigvee_{\alpha\in\Lambda_F}X_\alpha. Therefore, we get ⟨F⟩\langle F\rangle is a compact element in (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L})\cup\{\varnothing\},\subseteq). Now, let X∈CS⁡(L)X\in\operatorname{CS}(\mathcal{L}) be any compact element. Clearly, X⊆⋃x∈X⟨x⟩⊆⟨⋃x∈X⟨x⟩⟩=⋁x∈X⟨x⟩X\subseteq\bigcup_{x\in X}\langle x\rangle\subseteq\left\langle\bigcup_{x\in X}\langle x\rangle\right\rangle=\bigvee_{x\in X}\langle x\rangle. Since XX is compact, there exist finitely many elements x1,x2,…,xn∈Xx_1,x_2,\ldots,x_n\in X such that

X⊆⋁i=1n⟨xi⟩.(2)X\subseteq\bigvee_{i=1}^{n}\langle x_i\rangle. \tag*{(2)}

Let F={x1,x2,…,xn}F=\{x_1,x_2,\ldots,x_n\}. By Lemma 2.9, ⟨F⟩=⋁i=1n⟨xi⟩\langle F\rangle=\bigvee_{i=1}^{n}\langle x_i\rangle. From equation (2), we get X⊆⟨F⟩X\subseteq\langle F\rangle. Since F⊆XF\subseteq X, we have ⟨F⟩⊆⟨X⟩=X\langle F\rangle\subseteq\langle X\rangle=X. Hence X=⟨F⟩X=\langle F\rangle where FF is a finite subset of L\mathcal{L}. Thus the set of all compact elements of the complete lattice (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L})\cup\{\varnothing\},\subseteq) is {⟨F⟩∣F finite subset of L}\{\langle F\rangle\mid F\text{ finite subset of }\mathcal{L}\}

Finally, we show that (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L})\cup\{\varnothing\},\subseteq) is algebraic. Let X∈CS⁡(L)X\in\operatorname{CS}(\mathcal{L}). By Proposition 2.6, we have

X=⋃{⟨F⟩∣F⊆X, F is finite}X=\bigcup\{\langle F\rangle\mid F\subseteq X,\ F\text{ is finite}\}

Then,

⋁{⟨F⟩∣F⊆X, F is finite}=⟨⋃{⟨F⟩∣F⊆X, F is finite}⟩=⟨X⟩=X(because X is convex subhyperlattice)\begin{aligned} \bigvee\{\langle F\rangle\mid F \subseteq X,\ F\text{ is finite}\} = \left\langle\bigcup\{\langle F\rangle\mid F \subseteq X,\ F\text{ is finite}\}\right\rangle\\ &= \langle X\rangle= X\quad\text{(because $X$ is convex subhyperlattice)} \end{aligned}

This shows that XX is a join of some compact elements in (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L}) \cup\{\varnothing\}, \subseteq). □\square

Example 2.11. Consider the lattice L={0,a,b,c,d,e,f,h,1}\mathcal{L} = \{0, a, b, c, d, e, f, h, 1\} shown in Figure 1. The PP-hyperlattice obtained by taking P={0,c}P = \{0,c\} is given in Figure 2. In the diagrammatic representation, the meet of two elements is the lattice element occurring in the same position as their meet in the lattice. The hyperoperation join of two elements is the set occurring in the same position as their join in the lattice. For instance, 0∨0={0,c}0 \vee0 = \{0,c\}, a∨e={1}a \vee e = \{1\}, c∨f={f}c \vee f = \{f\}.

Figure 1. Lattice L\mathcal{L} from Example 2.11

Figure 2. Hyperlattice L\mathcal{L} from Example 2.11

The set CS⁡(L)\operatorname{CS}(\mathcal{L}) with respect to orders ≤b\leq_b and ⊑\sqsubseteq

In this section, we investigate the set CS⁡(L)\operatorname{CS}(\mathcal{L}) with respect to two new orders, ≤b\leq_b and ⊑\sqsubseteq. We determine classes of hyperlattices where the two orders are equivalent and give examples to show that they do not coincide in general. We explore the relationship between CS⁡(L)\operatorname{CS}(\mathcal{L}) and I(L)×F(L)I(\mathcal{L}) \times F(\mathcal{L}) in PP-hyperlattices.

Proposition 3.1. Let ≤b\leq_b be a relation defined on CS⁡(L)\operatorname{CS}(\mathcal{L}) as follows: for C1,C2∈CS⁡(L)C_1, C_2 \in\operatorname{CS}(\mathcal{L}), C1≤bC2C_1 \leq_b C_2 if and only if for every a∈C1a \in C_1, there exists b∈C2b \in C_2 such that a⪯ba \preceq b and for every b∈C2b \in C_2 there exists a∈C1a \in C_1 such that a⪯ba \preceq b. Then (CS⁡(L),≤b)(\operatorname{CS}(\mathcal{L}), \leq_b) forms a poset.

Lattice diagram for the hyperlattice considered in Example 2.11

Figure 3. Lattice (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L}) \cup\{\varnothing\}, \subseteq) for the hyperlattice considered in Example 2.11

Proof. Clearly ≤b\le_b is reflexive and transitive. Now suppose C1≤bC2C_1 \le_b C_2 and C2≤bC1C_2 \le_b C_1. Let a∈C1a \in C_1. Then there exists b∈C2b \in C_2 such that a⪯ba \preceq b. Since C2≤bC1C_2 \le_b C_1, there exists b′∈C2b' \in C_2 such that b′⪯ab' \preceq a. Hence b′⪯a⪯bb' \preceq a \preceq b where b,b′∈C2b,b' \in C_2. By the convexity of C2C_2, it follows that a∈C2a \in C_2. Thus C1⊆C2C_1 \subseteq C_2. A similar argument gives C2⊆C1C_2 \subseteq C_1, and hence C1=C2C_1 = C_2. Therefore ≤b\le_b is anti-symmetric. Hence, (CS⁡(L),≤b)(\operatorname{CS}(\mathcal{L}), \le_b) is a poset.

Proposition 3.2. Let L\mathcal{L} be a PP-hyperlattice. The relation ⊑\sqsubseteq defined on CS⁡(L)\operatorname{CS}(\mathcal{L}) by: for C1,C2∈CS⁡(L)C_1, C_2 \in\operatorname{CS}(\mathcal{L}), C1⊑C2C_1 \sqsubseteq C_2 if and only if ⟨C1]⊆⟨C2]\langle C_1] \subseteq\langle C_2] and [C1⟩⊇[C2⟩[C_1\rangle\supseteq[C_2\rangle, is a partial order on CS⁡(L)\operatorname{CS}(\mathcal{L}).

Proof. By definition, ⊑\sqsubseteq is reflexive and transitive. Now suppose C1⊑C2C_1 \sqsubseteq C_2 and C2⊑C1C_2 \sqsubseteq C_1. Then ⟨C1]⊆⟨C2]\langle C_1] \subseteq\langle C_2], [C1⟩⊇[C2⟩[C_1\rangle\supseteq[C_2\rangle and ⟨C2]⊆⟨C1]\langle C_2] \subseteq\langle C_1], [C2⟩⊇[C1⟩[C_2\rangle\supseteq[C_1\rangle. Thus ⟨C1]=⟨C2]\langle C_1] = \langle C_2] and [C1⟩=[C2⟩[C_1\rangle= [C_2\rangle. Since L\mathcal{L} is a PP-hyperlattice, using Theorem 1.15, we get C1=⟨C1]∩[C1⟩C_1 = \langle C_1] \cap[C_1\rangle and C2=⟨C2]∩[C2⟩C_2 = \langle C_2] \cap[C_2\rangle from which it follows that C1=C2C_1 = C_2. Therefore ⊑\sqsubseteq is anti-symmetric.

Proposition 3.3. Let A,B∈CS⁡(L)A, B \in\operatorname{CS}(\mathcal{L}). If A≤bBA \le_b B then A⊑BA \sqsubseteq B.

Proof. Let A,B∈CS⁡(L)A, B \in\operatorname{CS}(\mathcal{L}). Suppose A≤bBA \le_b B. Let x∈Ax \in A. Then, there exists z∈Bz \in B such that x⪯zx \preceq z. Since z∈B⊆⟨B]z \in B \subseteq\langle B], and x⪯zx \preceq z, we get x∈⟨B]x \in\langle B]. Therefore, A⊆⟨B]A \subseteq\langle B] which implies, ⟨A]⊆⟨B]\langle A] \subseteq\langle B]. Now, let x∈Bx \in B. Then, there exists y∈Ay \in A such that y⪯xy \preceq x. Since y∈A⊆[A⟩y \in A \subseteq[A\rangle and y⪯xy \preceq x, we get that x∈[A⟩x \in[A\rangle. Hence B⊆[A⟩B \subseteq[A\rangle which implies [B⟩⊆[A⟩[B\rangle\subseteq[A\rangle. Thus, A⊑BA \sqsubseteq B.

As illustrated in Example 3.4, the relation ⊑\sqsubseteq is not a partial order on an arbitrary strong join hyperlattice. In addition, Example 3.4 provides a counterexample to the equivalence of ≤b\leq_b and ⊑\sqsubseteq.

Example 3.4. Consider the hyperlattice (L,∨,∧)(\mathcal{L},\vee,\wedge) where ∨\vee and ∧\wedge are given in Table 1. The hyperideals, type-III hyperfilters and convex subhyperlattices are listed in Table 2. The poset (CS⁡(L),≤b)(\operatorname{CS}(\mathcal{L}),\leq_b) is given in Figure 4. The relation ⊑\sqsubseteq is not a partial order on CS⁡(L)\operatorname{CS}(\mathcal{L}) because the sets {x1}\{x_1\} and {x1,x2,1}\{x_1,x_2,1\} violate antisymmetry. Further, {x1,x2,1}⊑{x1}\{x_1,x_2,1\}\sqsubseteq\{x_1\}, we have {x1,x2,1}≰b{x1}\{x_1,x_2,1\}\nleq_b\{x_1\} demonstrating that ⊑\sqsubseteq does not imply ≤b\leq_b.

∨\vee00x1x_1x2x_211
00{0}\{0\}{x1,x2,1}\{x_1,x_2,1\}{x2,1}\{x_2,1\}{1}\{1\}
x1x_1{x1,x2,1}\{x_1,x_2,1\}{x1}\{x_1\}{x2,1}\{x_2,1\}{1}\{1\}
x2x_2{x2,1}\{x_2,1\}{x2,1}\{x_2,1\}{x2}\{x_2\}{1}\{1\}
11{1}\{1\}{1}\{1\}{1}\{1\}{1}\{1\}
∧\wedge00x1x_1x2x_211
0000000000
x1x_100x1x_1x1x_1x1x_1
x2x_200x1x_1x2x_2x2x_2
1100x1x_1x2x_211

Table 1. ∨\vee and ∧\wedge on L\mathcal{L} considered in Example 3.4

HyperidealsType-III HyperfiltersConvex subhyperlattices
I1={0}I_1=\{0\}F1={1}F_1=\{1\}C1={0}C_1=\{0\} C6={x1,x2,1}C_6=\{x_1,x_2,1\}
I2={0,x1,x2,1}I_2=\{0,x_1,x_2,1\}F2={x2,1}F_2=\{x_2,1\}C2={x1}C_2=\{x_1\} C7={0,x1,x2,1}C_7=\{0,x_1,x_2,1\}
I4=LI_4=LF3={x1,x2,1}F_3=\{x_1,x_2,1\}C3={x2}C_3=\{x_2\}
F={0,x1,x2,1}F=\{0,x_1,x_2,1\}C4={1}C_4=\{1\}
C5={x2,1}C_5=\{x_2,1\}

Table 2. Hyperideals, Type-III Hyperfilters and Convex subhyperlattices of (L,∨,∧)(\mathcal{L},\vee,\wedge) considered in Example 3.4

Figure 4. Poset (CS⁡(L),≤b)(\operatorname{CS}(\mathcal{L}),\leq_b) for the hyperlattice considered in Example 3.4

Proposition 3.5 [11]. Let H⊆LH\subseteq L be non-empty, where LL is a PP-hyperlattice. Then

ii ⟨H]={x∈L:x⪯y\langle H] = \{x \in\mathcal{L}: x \preceq y for some y∈h1∨Ph2∨P…∨Phn,y \in h_1 \overset{P}{\vee} h_2 \overset{P}{\vee} \ldots\overset{P}{\vee} h_n, hi∈H,n∈N,1≤i≤n}h_i \in H, n \in\mathbb{N}, 1 \leq i \leq n\}.

iiii [H⟩={x∈L:h1∧h2∧⋯∧hn⪯x,hi∈H,n∈N,1≤i≤n}[H\rangle= \left\{x \in\mathcal{L}: h_1 \wedge h_2 \wedge\cdots\wedge h_n \preceq x, h_i \in H, n \in\mathbb{N}, 1 \leq i \leq n\right\}.

The following proposition shows that orders ≤b\leq_b and ⊑\sqsubseteq are equivalent in a PP-hyperlattice.

Proposition 3.6. Let L\mathcal{L} be a PP-hyperlattice. Then the orders ≤b\leq_b and ⊑\sqsubseteq are equivalent.

Proof. Assume that A⊑BA \sqsubseteq B where A,B∈CS⁡(L)A, B \in\operatorname{CS}(\mathcal{L}). Let x∈Ax \in A. Then since ⟨A]⊆⟨B]\langle A] \subseteq\langle B], we get x∈⟨B]x \in\langle B]. This implies x⪯yx \preceq y for some y∈h1∨Ph2∨P…∨Phny \in h_1 \overset{P}{\vee} h_2 \overset{P}{\vee} \ldots\overset{P}{\vee} h_n where hi∈Bh_i \in B (1≤i≤n1 \leq i \leq n) and n∈Nn \in\mathbb{N}. Since BB is a subhyperlattice, we get y∈By \in B. For the verification of the second condition, let z∈Bz \in B. Then since B⊆[B⟩⊆[A⟩B \subseteq[B\rangle\subseteq[A\rangle, we have g1∧g2∧⋯∧gk⪯zg_1 \wedge g_2 \wedge\cdots\wedge g_k \preceq z where gi∈Ag_i \in A (1≤i≤k1 \leq i \leq k) and k∈Nk \in\mathbb{N}. Now the fact that AA is a subhyperlattice gives that w=g1∧g2∧⋯∧gk∈Aw = g_1 \wedge g_2 \wedge\cdots\wedge g_k \in A. Thus, we have a w∈Aw \in A such that w⪯zw \preceq z. Therefore, A≤bBA \leq_b B. Hence, the orders ≤b\leq_b and ⊑\sqsubseteq are equivalent by Proposition 3.3.

Example 3.7. Let L={a,b,c,d,e}\mathcal{L} = \{a, b, c, d, e\}. The hyperoperation ∨\vee and classical operation ∧\wedge are given in Table 3. Then (L,∨,∧)(\mathcal{L}, \vee, \wedge) forms a join hyperlattice but it is not a strong hyperlattice. In this hyperlattice, the orders ≤b\leq_b and ⊑\sqsubseteq on CS⁡(L)\operatorname{CS}(\mathcal{L}) are not equivalent. Here, (CS⁡(L),≤b)(\operatorname{CS}(\mathcal{L}), \leq_b) is a poset (infact a lattice) which is shown in Figure 5. But ⊑\sqsubseteq is not a partial order on CS⁡(L)\operatorname{CS}(\mathcal{L}). If we take A={b}A = \{b\} and B={a,b}B = \{a, b\}, then we get A⊑BA \sqsubseteq B and B⊑AB \sqsubseteq A. But A≠BA \ne B.

∨\veeaabbccddee
aa{a}\{a\}{a,b}\{a,b\}{a,c}\{a,c\}{a,d}\{a,d\}{a,e}\{a,e\}
bb{a,b}\{a,b\}{b}\{b\}{b,c,e}\{b,c,e\}{b,d,e}\{b,d,e\}{b,e}\{b,e\}
cc{a,c}\{a,c\}{b,c,e}\{b,c,e\}{c}\{c\}{c,d}\{c,d\}{c,e}\{c,e\}
dd{a,d}\{a,d\}{b,d,e}\{b,d,e\}{c,d}\{c,d\}{d}\{d\}{d,e}\{d,e\}
ee{a,e}\{a,e\}{b,e}\{b,e\}{c,e}\{c,e\}{d,e}\{d,e\}{e}\{e\}
∧\wedgeaabbccddee
aaaaaaaaaaaa
bbaabbaaaabb
ccaaaacccccc
ddaaaaccdddd
eeaabbccddee

Table 3. ∨\vee and ∧\wedge on L\mathcal{L} considered in Example 3.7

Remark 3.8. Let L\mathcal{L} be a PP-hyperlattice. Then,

ii for a non-empty subset AA of L\mathcal{L}, we have D⟨A⟩=⟨A]D\langle A\rangle= \langle A] and h⟨A⟩=[A⟩h\langle A\rangle= [A\rangle.

iiii for two non-empty subsets AA and BB of L\mathcal{L}, suppose that for every a∈Aa \in A, there is a b∈Bb \in B such that a⪯ba \preceq b and for every b∈Bb \in B, there is a a∈Aa \in A such that a⪯ba \preceq b. Then A⊆⟨B]A \subseteq\langle B] and B⊆[A⟩B \subseteq[A\rangle. Then by Proposition 3.6, we get ⟨A⟩≤b⟨B⟩\langle A\rangle\leq_b \langle B\rangle.

Figure 5. Lattice (CS⁡(L),≤b)(\operatorname{CS}(\mathcal{L}), \le_b) for the hyperlattice considered in Example 3.7

  • iii for a,b∈La,b \in\mathcal{L} such that a⪯ba \preceq b, we have ⟨a⟩≤b⟨b⟩\langle a\rangle\le_b \langle b\rangle.

  • iv The map f:L→(CS⁡(L),≤b)f:\mathcal{L}\to(\operatorname{CS}(\mathcal{L}),\le_b) defined by f(a)=⟨a⟩f(a)=\langle a\rangle is order preserving.

Proposition 3.9. Let L\mathcal{L} be a PP-hyperlattice. Then the poset (CS⁡(L),⊑)(\operatorname{CS}(\mathcal{L}),\sqsubseteq) forms a lattice with the following meet and join operations: For C1,C2∈CS⁡(L)C_1,C_2\in\operatorname{CS}(\mathcal{L}),

C1∧C2=(⟨C1]∧⟨C2])∩([C1⟩∧[C2⟩)C_1\wedge C_2=(\langle C_1]\wedge\langle C_2])\cap([C_1\rangle\wedge[C_2\rangle)
C1∨C2=(⟨C1]∨⟨C2])∩([C1⟩∨[C2⟩)C_1\vee C_2=(\langle C_1]\vee\langle C_2])\cap([C_1\rangle\vee[C_2\rangle)

where ⟨C1]∧⟨C2]=⟨C1]∩⟨C2]\langle C_1]\wedge\langle C_2]=\langle C_1]\cap\langle C_2], ⟨C1]∨⟨C2]=⟨⟨C1]∪⟨C2]⟩\langle C_1]\vee\langle C_2]=\langle\langle C_1]\cup\langle C_2]\rangle are the meet and join operations in I(L)I(\mathcal{L}), and [C1⟩∧[C2⟩=[C1⟩∪[C2⟩[C_1\rangle\wedge[C_2\rangle=[C_1\rangle\cup[C_2\rangle, [C1⟩∨[C2⟩=[C1⟩∩[C2⟩[C_1\rangle\vee[C_2\rangle=[C_1\rangle\cap[C_2\rangle are the meet and join operations in F(L)F(\mathcal{L}).

Proof. Let M=(⟨C1]∧⟨C2])∩([C1⟩∧[C2⟩)M=(\langle C_1]\wedge\langle C_2])\cap([C_1\rangle\wedge[C_2\rangle) and N=(⟨C1]∨⟨C2])∩([C1⟩∨[C2⟩)N=(\langle C_1]\vee\langle C_2])\cap([C_1\rangle\vee[C_2\rangle). First we show that MM and NN are non-empty. Choose x∈C1x\in C_1 and y∈C2y\in C_2. Then x∧y∈⟨C1],⟨C2]x\wedge y\in\langle C_1],\langle C_2]. Also, x∈C1⊆[C1⟩⊆⟨[C1⟩∪[C2⟩x\in C_1\subseteq[C_1\rangle\subseteq\langle[C_1\rangle\cup[C_2\rangle. Similarly, y∈⟨[C1⟩∪[C2⟩y\in\langle[C_1\rangle\cup[C_2\rangle. By the definition of hyperfilter, we get that x∧y∈⟨[C1⟩∪[C2⟩x\wedge y\in\langle[C_1\rangle\cup[C_2\rangle. Thus MM is non-empty. To show NN is non-empty, let a∈C1⊆[C1⟩a\in C_1\subseteq[C_1\rangle and b∈C2⊆[C2⟩b\in C_2\subseteq[C_2\rangle. This implies a∨b⊆[C1⟩∩[C2⟩a\vee b\subseteq[C_1\rangle\cap[C_2\rangle by the definition of hyperfilter. We also have a,b∈⟨C1]∪⟨C2]⊆⟨⟨C1]∪⟨C2]⟩a,b\in\langle C_1]\cup\langle C_2]\subseteq\langle\langle C_1]\cup\langle C_2]\rangle. Since $\langle\langle C_1]\cup\langle C_2]\rangle is a hyperideal, we get a∨Pb⊆⟨C1⟩∪⟨C2⟩a \mathbin{\vee^P} b \subseteq\langle C_1\rangle\cup\langle C_2\rangle. This implies that a∨b⊆Na \vee b \subseteq N, and hence, NN is non-empty. Thus by Theorem 1.14, it follows that MM and NN are convex subhyperlattices. Since L\mathcal{L} is a PP-hyperlattice, MM attains a unique representation in terms of ⟨M⟩\langle M\rangle and [M⟩[M\rangle. By the uniqueness, we get that ⟨M⟩=⟨C1⟩∩⟨C2⟩\langle M\rangle= \langle C_1\rangle\cap\langle C_2\rangle and [M⟩=[C1⟩∪[C2⟩[M\rangle= [C_1\rangle\cup[C_2\rangle. By a similar argument, we can show that ⟨N⟩=⟨C1⟩∪⟨C2⟩\langle N\rangle= \langle C_1\rangle\cup\langle C_2\rangle and [N⟩=[C1⟩∩[C2⟩[N\rangle= [C_1\rangle\cap[C_2\rangle. Using these facts, we conclude that MM and NN are greatest lower bound and least upper bound of {C1,C2}\{C_1,C_2\} respectively, in (CS(L),⊑)(\mathrm{CS}(\mathcal{L}),\sqsubseteq).

Corollary 3.10. (CS(L),≤b)(\mathrm{CS}(\mathcal{L}),\le_b) is a lattice for a PP-hyperlattice L\mathcal{L}.

Proof. Follows by Proposition 3.6 and Proposition 3.9.

The following example shows that (CS(L),≤b)(\mathrm{CS}(\mathcal{L}),\le_b) need not be a lattice for an arbitrary join hyperlattice.

Example 3.11. For L={a1,b1,c1}\mathcal{L}=\{a_1,b_1,c_1\}, let the operation ∧\wedge and hyperoperation ∨\vee be as given in Table 4. Then the poset (CS(L),≤b)(\mathrm{CS}(\mathcal{L}),\le_b) is given in Figure 6. Here (CS(L),≤b)(\mathrm{CS}(\mathcal{L}),\le_b) is not a lattice.

∨\veea1a_1b1b_1c1c_1
a1a_1{a1}\{a_1\}{b1}\{b_1\}{c1}\{c_1\}
b1b_1{b1}\{b_1\}{b1}\{b_1\}{b1,c1}\{b_1,c_1\}
c1c_1{c1}\{c_1\}{b1,c1}\{b_1,c_1\}{c1}\{c_1\}
∧\wedgea1a_1b1b_1c1c_1
a1a_1a1a_1a1a_1a1a_1
b1b_1a1a_1b1b_1a1a_1
c1c_1a1a_1a1a_1c1c_1

Table 4. ∨\vee and ∧\wedge on L\mathcal{L} considered in Example 3.11

Figure 6. Poset (CS(L),≤b)(\mathrm{CS}(\mathcal{L}),\le_b) for the hyperlattice considered in Example 3.11

Definition 3.12. Let XX be a non-empty subset of L\mathcal{L}.

i The initial segment of XX with respect to ⪯\preceq is defined by

↓⪯(X)={y∈L∣y⪯x for some x∈X}\mathord{\downarrow}_{\preceq}(X)=\{y\in\mathcal{L}\mid y\preceq x\text{ for some }x\in X\}

If X={x}X=\{x\}, we use the symbol ↓⪯(x)\mathord{\downarrow}_{\preceq}(x) in place of ↓⪯({x})\mathord{\downarrow}_{\preceq}(\{x\}). ii The final segment of XX with respect to ⪯\preceq is defined by

↑⪯(X)={y∈L∣x⪯y for some x∈X}\uparrow_{\preceq}(X)=\{y\in\mathcal{L}\mid x\preceq y\text{ for some }x\in X\}

If X={x}X=\{x\}, we use the symbol ↑⪯(x)\uparrow_{\preceq}(x) in place of ↑⪯({x})\uparrow_{\preceq}(\{x\}).

Note that X⊆↓⪯(X)⊆⟨X]X\subseteq\downarrow_{\preceq}(X)\subseteq\langle X] and X⊆↑⪯(X)⊆[X⟩X\subseteq\uparrow_{\preceq}(X)\subseteq[X\rangle for any non-empty subset XX of L\mathcal{L}.

Example 3.13. Consider the hyperlattice shown in Figure 7. Suppose X={e,f}X=\{e,f\}. Then we get ↓⪯(X)={0,a,b,c,e,f}\downarrow_{\preceq}(X)=\{0,a,b,c,e,f\}, ↑⪯(X)={e,f,1}\uparrow_{\preceq}(X)=\{e,f,1\}, [X⟩={c,e,f,1}[X\rangle=\{c,e,f,1\} and ⟨X]=L\langle X]=\mathcal{L}. This shows that ↓⪯(X)⊂⟨X]\downarrow_{\preceq}(X)\subset\langle X] and ↑⪯(X)⊂[X⟩\uparrow_{\preceq}(X)\subset[X\rangle. The initial segment of XX is shown by dotted lines and the final segment of XX is shown by dashed lines in Figure 7.

Figure 7. Hyperlattice considered in Example 3.13

Proposition 3.14 gives a characterization of those non-empty subsets of a PP-hyperlattice that are convex subhyperlattices.

Proposition 3.14. A non-empty set CC of a PP-hyperlattice L\mathcal{L} is a convex subhyperlattice if and only if ↓⪯(C)=⟨C]\downarrow_{\preceq}(C)=\langle C], ↑⪯(C)=[C⟩\uparrow_{\preceq}(C)=[C\rangle and CC is a convex set.

Proof. Let ∅≠C⊆L\varnothing\ne C\subseteq\mathcal{L} be a convex subhyperlattice. Clearly, CC is a convex set. Already, ↓⪯(C)⊆⟨C]\downarrow_{\preceq}(C)\subseteq\langle C]. Let a∈⟨C]a\in\langle C]. Then a⪯ya\preceq y for some y∈h1∨h2∨…∨hny\in h_{1}\vee h_{2}\vee\ldots\vee h_{n} where hi∈Ch_{i}\in C, n∈Nn\in\mathbb{N} and 1≤i≤n1\le i\le n. Note that h1∨h2∨…∨hn⊆Ch_{1}\vee h_{2}\vee\ldots\vee h_{n}\subseteq C since CC is a subhyperlattice. Thus we have a⪯ya\preceq y where y∈Cy\in C. This implies a∈↓⪯(C)a\in\downarrow_{\preceq}(C) and hence ⟨C]=↓⪯(C)\langle C]=\downarrow_{\preceq}(C). Now, let b∈[C⟩b\in[C\rangle. Then g1∧g2∧⋯∧gs⪯bg_{1}\wedge g_{2}\wedge\cdots\wedge g_{s}\preceq b for some gi∈Cg_{i}\in C, s∈Ns\in\mathbb{N}. Since CC is a subhyperlattice, g1∧g2∧⋯∧gs∈Cg_{1}\wedge g_{2}\wedge\cdots\wedge g_{s}\in C, so we get b∈↑⪯(C)b\in\uparrow_{\preceq}(C). This shows that [C⟩⊆↑⪯(C)[C\rangle\subseteq\uparrow_{\preceq}(C) and hence, [C⟩=↑⪯(C)[C\rangle=\uparrow_{\preceq}(C). Conversely, assume that ↓⪯(C)=⟨C]\downarrow_{\preceq}(C)=\langle C], ↑⪯(C)=[C⟩\uparrow_{\preceq}(C)=[C\rangle and CC is a convex set. To show CC is a convex subhyperlattice, it remains to show that CC is a subhyperlattice. Clearly C⊆⟨C]∩[C⟩C\subseteq\langle C]\cap[C\rangle. Let b∈⟨C]∩[C⟩=↓⪯(C)∩↑⪯(C)b\in\langle C]\cap[C\rangle=\downarrow_{\preceq}(C)\cap\uparrow_{\preceq}(C). Then x⪯b⪯yx\preceq b\preceq y for some x,y∈Cx,y\in C. Since CC is convex, we get b∈Cb\in C. Thus C=⟨C]∩[C⟩C=\langle C]\cap[C\rangle which implies CC is a subhyperlattice.

Remark 3.15. The convexity of CC is necessary in the converse part of Proposition 3.14. In the PP-hyperlattice shown in Figure 7, consider the subhyperlattice C={0,c,f}C=\{0,c,f\}. Then ↑⪯(C)=L=[C⟩\uparrow_{\preceq}(C)=\mathcal{L}=[C\rangle and ↓⪯(C)={0,b,c,f}=⟨C]\downarrow_{\preceq}(C)=\{0,b,c,f\}=\langle C]. Note that CC is not a convex set.

Proposition 3.16. ([15]) Let (L,∨,∧)(\mathcal{L},\vee,\wedge) be a strong join hyperlattice. For every a,b∈La,b\in\mathcal{L}, there exist a1,a2∈a∨ba_{1},a_{2}\in a\vee b such that a⪯a1a\preceq a_{1} and b⪯a2b\preceq a_{2}.

Proposition 3.17. Let XX and YY be two non-empty subsets of (L,∨,∧)(\mathcal{L},\vee,\wedge), and A=X∧Y={x∧y∣x∈X,y∈Y}A=X\wedge Y=\{x\wedge y\mid x\in X,y\in Y\}, B=X∨Y=⋃a∈X,b∈Ya∨bB=X\vee Y=\bigcup_{a\in X,b\in Y}a\vee b, then

ii ⟨A]⊆⟨X]∩⟨Y]\langle A]\subseteq\langle X]\cap\langle Y] and [B⟩⊆[X⟩∩[Y⟩[B\rangle\subseteq[X\rangle\cap[Y\rangle,

iiii [A⟩=[X⟩∪[Y⟩[A\rangle=[X\rangle\cup[Y\rangle and ⟨B]=⟨X]∪⟨Y]\langle B]=\langle X]\cup\langle Y],

Proof. ii Let p∈Ap\in A. Then p=x∧yp=x\wedge y for some x∈X,y∈Yx\in X,y\in Y. Since x∧y⪯x,yx\wedge y\preceq x,y, we get p∈⟨X]∩⟨Y]p\in\langle X]\cap\langle Y]. This gives A⊆⟨X]∩⟨Y]A\subseteq\langle X]\cap\langle Y], which implies ⟨A]⊆⟨X]∩⟨Y]\langle A]\subseteq\langle X]\cap\langle Y]. Now suppose q∈Bq\in B. Then q∈a∨bq\in a\vee b for some $a\in X,b\in YY. Note that a∈X⊆[X⟩a \in X \subseteq[X\rangle and b∈Y⊆[Y⟩b \in Y \subseteq[Y\rangle. By the definition of type-III hyperfilter, we get a∨b⊆[Y⟩a \vee b \subseteq[Y\rangle (because a∈La \in L, b∈[Y⟩b \in[Y\rangle). Similarly, we get a∨b⊆[X⟩a \vee b \subseteq[X\rangle. This gives q∈[X⟩∩[Y⟩q \in[X\rangle\cap[Y\rangle. Thus B⊆[X⟩∩[Y⟩B \subseteq[X\rangle\cap[Y\rangle, and so, [B⟩⊆[X⟩∩[Y⟩[B\rangle\subseteq[X\rangle\cap[Y\rangle.

ii We first show that A⊆[X⟩∪[Y⟩A \subseteq[X\rangle\cup[Y\rangle. Let a∈Aa \in A. Then a=x∧ya = x \wedge y for some x∈Xx \in X and y∈Yy \in Y. Note that x,y∈[X⟩∪[Y⟩⊆⟨[X⟩∪[Y⟩x,y \in[X\rangle\cup[Y\rangle\subseteq\langle[X\rangle\cup[Y\rangle. So we get a∈⟨[X⟩∪[Y⟩a \in\langle[X\rangle\cup[Y\rangle. Thus A⊆⟨[X⟩∪[Y⟩A \subseteq\langle[X\rangle\cup[Y\rangle, which implies [A⟩⊆⟨[X⟩∪[Y⟩[A\rangle\subseteq\langle[X\rangle\cup[Y\rangle.

To show ⟨[X⟩∪[Y⟩⊆[A⟩\langle[X\rangle\cup[Y\rangle\subseteq[A\rangle, it is enough to show that X,Y⊆[A⟩X,Y \subseteq[A\rangle. Let x∈Xx \in X. Then, if y∈Yy \in Y is arbitrary, a=x∧y∈A⊆[A⟩a=x\wedge y \in A \subseteq[A\rangle and since x∧y⪯xx\wedge y \preceq x, we get x∈[A⟩x \in[A\rangle. This implies X⊆[A⟩X \subseteq[A\rangle which shows that [X⟩⊆[A⟩[X\rangle\subseteq[A\rangle. By a similar argument, we get that [Y⟩⊆[A⟩[Y\rangle\subseteq[A\rangle. Therefore [A⟩=[X⟩∪[Y⟩[A\rangle= [X\rangle\cup[Y\rangle.

For the second part, let b∈Bb \in B. Then b∈s∨tb \in s \vee t for some s∈Xs \in X and t∈Yt \in Y. Note that s,t∈⟨X]∪⟨Y]⊆⟨[X⟩∪[Y⟩s,t \in\langle X] \cup\langle Y] \subseteq\langle[X\rangle\cup[Y\rangle. Then by the definition of hyperideal, we get b∈s∨t⊆⟨[X⟩∪[Y⟩b \in s \vee t \subseteq\langle[X\rangle\cup[Y\rangle. Thus, we get ⟨B]⊆⟨[X⟩∪[Y⟩\langle B] \subseteq\langle[X\rangle\cup[Y\rangle.

Now to show the reverse inclusion, we show that X,Y⊆⟨B]X,Y \subseteq\langle B]. Let x∈Xx \in X. Then for any y∈Yy \in Y, we have x∨y⊆B⊆⟨B]x \vee y \subseteq B \subseteq\langle B]. By Proposition 3.16, there exists u∈x∨yu \in x \vee y such that x⪯ux \preceq u. Now, u∈⟨B]u \in\langle B] and x⪯ux \preceq u implies that x∈⟨B]x \in\langle B]. Thus X⊆⟨B]X \subseteq\langle B]. By a similar argument, we get Y⊆⟨B]Y \subseteq\langle B]. Hence, ⟨X]∪⟨Y]⊆⟨B]\langle X] \cup\langle Y] \subseteq\langle B], which implies ⟨X]∪⟨Y]⊆⟨B]\langle X] \cup\langle Y] \subseteq\langle B]. Therefore, ⟨B]=⟨X]∪⟨Y]\langle B] = \langle X] \cup\langle Y].

Proposition 3.18. Let XX and YY be subhyperlattices of a PP-hyperlattice (L,∨P,∧)(\mathcal{L},\overset{P}{\vee},\wedge), and A=X∧^Y={x∧y∣x∈X,y∈Y}A=X\mathbin{\widehat{\wedge}}Y=\{x\wedge y\mid x\in X,y\in Y\}, B=X∨^PY=⋃a∈X,b∈Ya∨PbB=X\overset{P}{\mathbin{\widehat{\vee}}}Y=\bigcup_{a\in X,b\in Y}a\overset{P}{\vee}b. Then ⟨A]=⟨X]∩⟨Y]\langle A]=\langle X]\cap\langle Y] and [B⟩=[X⟩∩[Y⟩[B\rangle=[X\rangle\cap[Y\rangle.

Proof. From Proposition 3.17, we have ⟨A]⊆⟨X]∩⟨Y]\langle A]\subseteq\langle X]\cap\langle Y] and [B⟩⊆[X⟩∩[Y⟩[B\rangle\subseteq[X\rangle\cap[Y\rangle. Let p∈⟨X]∩⟨Y]p\in\langle X]\cap\langle Y]. Then p⪯sp\preceq s for some s∈h1∨Ph2∨P⋯∨Phns\in h_1\overset{P}{\vee}h_2\overset{P}{\vee}\cdots\overset{P}{\vee}h_n where hi∈X(1≤i≤n)h_i\in X(1\leq i\leq n) and n∈Nn\in\mathbb{N}, and p⪯tp\preceq t for some t∈g1∨Pg2∨P⋯∨Pgmt\in g_1\overset{P}{\vee}g_2\overset{P}{\vee}\cdots\overset{P}{\vee}g_m where gj∈Yg_j\in Y, m∈Nm\in\mathbb{N} and 1≤j≤m1\leq j\leq m. Since XX and YY are subhyperlattices, we get s∈Xs\in X and t∈Yt\in Y. We have p⪯s∧tp\preceq s\wedge t where s∧t∈A⊆⟨A]s\wedge t\in A\subseteq\langle A]. This implies p∈⟨A]p\in\langle A]. Thus ⟨X]∩⟨Y]⊆⟨A]\langle X]\cap\langle Y]\subseteq\langle A].

Let b∈[X⟩∩[Y⟩b\in[X\rangle\cap[Y\rangle. Then s=g1∧g2∧⋯∧gk⪯bs=g_1\wedge g_2\wedge\cdots\wedge g_k\preceq b for some gi∈Xg_i\in X and k∈Nk\in\mathbb{N} where 1≤i≤k1\leq i\leq k and t=f1∧f2∧⋯∧fl⪯bt=f_1\wedge f_2\wedge\cdots\wedge f_l\preceq b for some fj∈Yf_j\in Y and l∈Nl\in\mathbb{N} where 1≤j≤l1\leq j\leq l. Since XX and YY are subhyperlattices, we get s∈Xs\in X and t∈Yt\in Y. Hence, we have s∨Pt⊆B⊆[B⟩s\overset{P}{\vee}t\subseteq B\subseteq[B\rangle. Let k1∈b∨Pbk_1\in b\overset{P}{\vee}b. Then k1=b∨Pb∨Pp1k_1=b\overset{P}{\vee}b\overset{P}{\vee}p_1 for some p1∈Pp_1\in P. Note that s∨Pt∨Pp1⪯b∨Pb∨Pp1s\overset{P}{\vee}t\overset{P}{\vee}p_1\preceq b\overset{P}{\vee}b\overset{P}{\vee}p_1 where k2=s∨Pt∨Pp1∈s∨Ptk_2=s\overset{P}{\vee}t\overset{P}{\vee}p_1\in s\overset{P}{\vee}t. Thus for any k∈b∨Pbk\in b\overset{P}{\vee}b, there exists l∈s∨Ptl\in s\overset{P}{\vee}t such that l⪯kl\preceq k. In particular, for b∈b∨Pbb\in b\overset{P}{\vee}b, there exists w∈s∨Ptw\in s\overset{P}{\vee}t such that w⪯bw\preceq b. We have w∈s∨Pt⊆[B⟩w\in s\overset{P}{\vee}t\subseteq[B\rangle and w⪯bw\preceq b. This gives b∈[B⟩b\in[B\rangle. Hence [B⟩=[X⟩∩[Y⟩[B\rangle=[X\rangle\cap[Y\rangle.

Remark 3.19. If XX and YY are not subhyperlattices in Proposition 3.18, then the equality may not hold. Consider the hyperlattice given in Figure 8. Let X={0,b}X=\{0,b\}, Y={c}Y=\{c\}. Clearly XX and YY are not subhyperlattices. We have A=X∧^Y={0}A=X\mathbin{\widehat{\wedge}}Y=\{0\} and B=X∨^PY={c,1}B=X\overset{P}{\mathbin{\widehat{\vee}}}Y=\{c,1\}. Note that ⟨X]={0,a,b}\langle X]=\{0,a,b\}, ⟨Y]={0,a,b,c,1}\langle Y]=\{0,a,b,c,1\} and ⟨A]={0,a}\langle A]=\{0,a\}. Here ⟨A]={0,a}⊂{0,a,b}=⟨X]∩⟨Y]\langle A]=\{0,a\}\subset\{0,a,b\}=\langle X]\cap\langle Y].

Figure 8. Hyperlattice considered in Remark 3.19

The expressions for meet and join in (CS⁡(L),≤b)(\operatorname{CS}(\mathcal{L}),\leq_b) given by Lavanya and Parameshwara Bhatta [10] motivated Bhargava to develop analogous results in trellises [2]. The result below on join hyperlattices is derived from the ideas in these works.

Theorem 3.20. Let XX and YY be any two convex subhyperlattices of a PP-hyperlattice L\mathcal{L}, and A=X∧Y={x∧y∣x∈X,y∈Y}A=X\wedge Y=\{x\wedge y\mid x\in X,y\in Y\}, B=X∨Y=⋃a∈X,b∈Ya∨bB=X\vee Y=\bigcup_{a\in X,b\in Y}a\vee b. Then X∧Y=⟨A⟩X\wedge Y=\langle A\rangle and X∨Y=⟨B⟩X\vee Y=\langle B\rangle where X∧YX\wedge Y and X∨YX\vee Y are respectively the meet and join of XX and YY in (CS⁡(L),⊑)(\operatorname{CS}(\mathcal{L}),\sqsubseteq).

Proof. This follows from Proposition 3.9, Proposition 3.17 and Proposition 3.18.

Definition 3.21 ([7]). A subposet QQ of a poset PP is cover preserving if whenever x≺yx\prec y in QQ for any x,y∈Qx,y\in Q, then x≺yx\prec y in PP.

Theorem 3.22. For a PP-hyperlattice L\mathcal{L}, the map f:CS⁡(L)→I(L)×F(L)f:\operatorname{CS}(\mathcal{L})\to I(\mathcal{L})\times F(\mathcal{L}) defined by, for C∈CS⁡(L)C\in\operatorname{CS}(\mathcal{L}), f(C)=(⟨C],[C⟩)f(C)=(\langle C], [C\rangle), is an injective homomorphism. Moreover, the sublattice K={(I,D):I∈I(L),D∈F(L),I∩D≠∅}K=\{(I,D):I\in I(\mathcal{L}),D\in F(\mathcal{L}),I\cap D\ne\varnothing\} of I(L)×F(L)I(\mathcal{L})\times F(\mathcal{L}) is cover preserving and is isomorphic to (CS⁡(L),⊑)(\operatorname{CS}(\mathcal{L}),\sqsubseteq).

Proof. The map ff is well defined. Also, ff is a one-to-one map. Note that for any X1,Y1∈CS⁡(L)X_1,Y_1\in\operatorname{CS}(\mathcal{L}), by Remark 3.8, we have ⟨X1∧Y1]=⟨X1]∧⟨Y1]\langle X_1\wedge Y_1]=\langle X_1]\wedge\langle Y_1] and [X1∧Y1⟩=[X1⟩∪[Y1⟩=[X1⟩∧[Y1⟩[X_1\wedge Y_1\rangle=[X_1\rangle\cup[Y_1\rangle=[X_1\rangle\wedge[Y_1\rangle. Now,

f(X1∧Y1)=(⟨X1∧Y1],[X1∧Y1⟩)=(⟨X1]∧⟨Y1],[X1⟩∧[Y1⟩)=(⟨X1],[X1⟩)∧(⟨Y1],[Y1⟩)=f(X1)∧f(Y1)\begin{aligned} f(X_1\wedge Y_1)&=(\langle X_1\wedge Y_1],[X_1\wedge Y_1\rangle)\\ &=(\langle X_1]\wedge\langle Y_1],[X_1\rangle\wedge[Y_1\rangle)\\ &=(\langle X_1],[X_1\rangle)\wedge(\langle Y_1],[Y_1\rangle)\\ &=f(X_1)\wedge f(Y_1) \end{aligned}

Therefore, ff is a meet homomorphism of lattices. Similarly, ff is a join homomorphism of lattices. Thus, ff is a homomorphism. Clearly, f(CS⁡(L))=Kf(\operatorname{CS}(\mathcal{L}))=K and hence CS⁡(L)≅K\operatorname{CS}(\mathcal{L})\cong K. Now, we show that KK is cover preserving. Let (I1,D1),(I2,D2)∈K(I_1,D_1),(I_2,D_2)\in K such that (I1,D1)≺(I2,D2)(I_1,D_1)\prec(I_2,D_2) in KK. It is enough to show that I1=I2I_1=I_2 and D1≺D2D_1\prec D_2 in F(L)F(\mathcal{L}) or I1≺I2I_1\prec I_2 in I(L)I(\mathcal{L}) and D1=D2D_1=D_2. Suppose I1≠I2I_1\ne I_2 and D1≠D2D_1\ne D_2. Then I1⊂I2I_1\subset I_2 and D1⊃D2D_1\supset D_2. Let p∈I2∖I1p\in I_2\setminus I_1. Take I=I1∨⟨p]I=I_1\vee\langle p] and D=D1D=D_1. Then (I,D)∈K(I,D)\in K and (I1,D1)<(I,D)<(I2,D2)(I_1,D_1)<(I,D)<(I_2,D_2) in KK. This is a contradiction to the fact that (I2,D2)(I_2,D_2) covers (I1,D1)(I_1,D_1) in KK. Thus I1=I2I_1=I_2 or D1=D2D_1=D_2. Suppose I1=I2I_1=I_2. If D1<D<D2D_1<D<D_2 in F(L)F(\mathcal{L}) for some D∈F(L)D\in F(\mathcal{L}), then, (I2,D)∈K(I_2,D)\in K and (I1,D1)<(I2,D)<(I2,D2)(I_1,D_1)<(I_2,D)<(I_2,D_2) in KK, which is a contradiction. Hence, D1≺D2D_1\prec D_2 in F(L)F(\mathcal{L}). Similarly, if D1=D2D_1=D_2, then I1≺I2I_1\prec I_2 in I(L)I(\mathcal{L}).

Remark 3.23. The sublattice KK of I(L)×F(L)I(\mathcal{L})\times F(\mathcal{L}) in Theorem 3.22 may not be convex. For example, consider the hyperlattice L\mathcal{L} given in Example 2.11. We note that ({0,c},{c,f,h,1})≤({0,c},{h,1})≤({0,c,d,e,h},{h,1})(\{0,c\},\{c,f,h,1\})\le(\{0,c\},\{h,1\})\le(\{0,c,d,e,h\},\{h,1\}) and ({0,c},{h,1})∈I(L)×F(L)(\{0,c\},\{h,1\})\in I(\mathcal{L})\times F(\mathcal{L}) but ({0,c},{h,1})∉K(\{0,c\},\{h,1\})\notin K.

Proposition 3.24. For a PP-hyperlattice L\mathcal{L}, I(L)I(\mathcal{L}) is the principal ideal of (CS⁡(L),⊑)(\operatorname{CS}(\mathcal{L}),\sqsubseteq) generated by L\mathcal{L} and F(L)F(\mathcal{L}) is the principal filter of (CS⁡(L),⊑)(\operatorname{CS}(\mathcal{L}),\sqsubseteq) generated by L\mathcal{L}.

Proof. Let X∈CS⁡(L)X\in\operatorname{CS}(\mathcal{L}) such that X⊑LX\sqsubseteq\mathcal{L}. Then [X⟩⊇[L⟩=L[X\rangle\supseteq[\mathcal{L}\rangle=\mathcal{L}. Thus [X⟩=L[X\rangle=\mathcal{L}. This shows that X=⟨X]∩[X⟩=⟨X]∩L=⟨X]∈I(L)X=\langle X]\cap[X\rangle=\langle X]\cap\mathcal{L}=\langle X]\in I(\mathcal{L}). Suppose X∈I(L)X\in I(\mathcal{L}), then X=⟨X]∩LX=\langle X]\cap\mathcal{L}. Being a hyperideal, XX is a convex subhyperlattice. By the uniqueness of the representation for XX, we get [X⟩=L[X\rangle= \mathcal{L}. Hence we get that X⊆LX \subseteq\mathcal{L}. By a similar argument, we can show that F(L)F(\mathcal{L}) is the principal filter generated by L\mathcal{L}.

Definition 3.25 ([6]). Let pi=qip_i=q_i be identities for i∈Ii \in I. The class κ\kappa of all lattices satisfying the identities pi=qip_i=q_i, for i∈Ii \in I, is called a equational class of lattices.

Corollary 3.26. Let L\mathcal{L} be a PP-hyperlattice and γ\gamma be an equational class of lattices. If I(L)I(\mathcal{L}) and F(L)F(\mathcal{L}) lie in γ\gamma, then (CS⁡(L),⊑)(\operatorname{CS}(\mathcal{L}),\sqsubseteq) lies in γ\gamma. Conversely, if (CS⁡(L),⊑)(\operatorname{CS}(\mathcal{L}),\sqsubseteq) lies in γ\gamma, then both I(L)I(\mathcal{L}) and F(L)F(\mathcal{L}) lie in γ\gamma.

Proof. This follows from Theorem 3.22 and Proposition 3.24.

In general, I(L)I(\mathcal{L}) and F(L)F(\mathcal{L}) need not be in a same equational class. For instance, in Example 2.11, I(L)I(\mathcal{L}) is modular, whereas F(L)F(\mathcal{L}) is non-modular. Secondly, even when I(L)I(\mathcal{L}) and F(L)F(\mathcal{L}) are in the same equational class, (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L})\cup\{\varnothing\},\subseteq) may fail to lie in the same class. In the PP-hyperlattice shown in Figure 9, I(L)I(\mathcal{L}), F(L)F(\mathcal{L}) and (CS⁡(L),⊑)(\operatorname{CS}(\mathcal{L}),\sqsubseteq) are distributive whereas (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L})\cup\{\varnothing\},\subseteq) is not distributive. Moreover, the result stated in Corollary 3.26 does not hold if we consider (CS⁡(L)∪{∅},⊆)(\operatorname{CS}(\mathcal{L})\cup\{\varnothing\},\subseteq) instead of (CS⁡(L),⊑)(\operatorname{CS}(\mathcal{L}),\sqsubseteq).

Figure 9. PP-hyperlattice considered in Example 3.27

Example 3.27. Let L={0,p,q,r,s,1}\mathcal{L}=\{0,p,q,r,s,1\}. The PP-hyperlattice with P={0,q}P=\{0,q\} is given in Figure 9.

Let P∗(L)\mathcal{P}^{*}(\mathcal{L}) denote the collection of all non-empty subsets of a PP-hyperlattice L\mathcal{L}. For P1,P2∈P∗(L)P_1,P_2\in\mathcal{P}^{*}(\mathcal{L}), define P1≤P2P_1\leq P_2 if and only if ⟨P1]⊆⟨P2]\langle P_1]\subseteq\langle P_2] and [P1⟩⊇[P2⟩[P_1\rangle\supseteq[P_2\rangle. Then clearly ≤\leq is reflexive and transitive. Now, define a new relation θ\theta on P∗(L)\mathcal{P}^{*}(\mathcal{L}) as follows: For P1,P2∈P∗(L)P_1,P_2\in\mathcal{P}^{*}(\mathcal{L}), define P1≡P2(θ)P_1\equiv P_2(\theta) if and only if P1≤P2P_1\leq P_2 and P2≤P1P_2\leq P_1. We note that θ\theta is an equivalence relation on P∗(L)\mathcal{P}^{*}(\mathcal{L}). Note that P1≡⟨P1⟩(θ)P_1\equiv\langle P_1\rangle(\theta), and P1≡P2(θ)P_1\equiv P_2(\theta) if and only if ⟨P1⟩=⟨P2⟩\langle P_1\rangle=\langle P_2\rangle. Hence the relation ≤\leq defined on P∗(L)/θ\mathcal{P}^{*}(\mathcal{L})/\theta by, for [P1],[P2]∈P∗(L)/θ[P_1],[P_2]\in\mathcal{P}^{*}(\mathcal{L})/\theta, [P1]≤[P2][P_1]\leq[P_2] if and only if ⟨P1⟩≤⟨P2⟩\langle P_1\rangle\leq\langle P_2\rangle is a partial order on P∗(L)/θ\mathcal{P}^{*}(\mathcal{L})/\theta. The map [P1]↦⟨P1⟩[P_1]\mapsto\langle P_1\rangle defines an order isomorphism between the two posets (P∗(L)/θ,≤)(\mathcal{P}^{*}(\mathcal{L})/\theta,\leq) and (CS⁡(L),⊑)(\operatorname{CS}(\mathcal{L}),\sqsubseteq). Since (CS⁡(L),⊑)(\operatorname{CS}(\mathcal{L}),\sqsubseteq) is a lattice, we note that (P∗(L)/θ,≤)(\mathcal{P}^{*}(\mathcal{L})/\theta,\leq) is also a lattice. Thus we have the following proposition.

Proposition 3.28. For a PP-hyperlattice L\mathcal{L},

(CS⁡(L),⊑)≅(P∗(L)/θ,≤)(\operatorname{CS}(\mathcal{L}),\sqsubseteq) \cong(\mathcal{P}^{*}(\mathcal{L})/\theta,\leq)

The notion of convexity is well understood in lattices and extends naturally to some classes of strong hyperlattices, but it fails to exhibit the same structural properties in an arbitrary join hyperlattice. In particular, a convex subhyperlattice need not admit a unique representation in terms of hyperideals and hyperfilters. In addition to this, several results valid in PP-hyperlattices and Nakano hyperlattices fail to extend directly. To overcome this, we introduce the concept of strongly convex sets, which refines the existing notion of convexity while preserving its properties. This concept serves as a natural generalization, ensuring that results established for PP-hyperlattices remain valid in the broader setting of arbitrary hyperlattices.

Strongly convex sets in L\mathcal{L}

In this section, we introduce the notion of strongly convex sets in an arbitrary join hyperlattice and study their basic properties. We show that every strongly convex set XX admits a unique representation in terms of ⟨X]\langle X] and [X⟩[X\rangle. This characterization allows us to extend several structural results known for PP-hyperlattices to a more general setting.

Definition 4.1. A non-empty set SS of L\mathcal{L} is said to be strongly convex if S=I∩FS=I\cap F for some hyperideal II and type-III hyperfilter FF.

Remark 4.2. By Theorem 1.14, every strongly convex subset of L\mathcal{L} is a convex subhyperlattice, but the converse does not hold in general. This is illustrated in Example 3.4, where {x1}\{x_1\} is a convex subhyperlattice which is not strongly convex. The example also illustrates that {1}\{1\} is a convex subhyperlattice which is a strongly convex set. Furthermore, in a PP-hyperlattice or a Nakano hyperlattice, every convex subhyperlattice is strongly convex.

We denote the set of all strongly convex sets of L\mathcal{L} by SC⁡(L)\operatorname{SC}(\mathcal{L}).

Proposition 4.3. Let XX be a strongly convex set of (L,∨,∧)(\mathcal{L},\vee,\wedge). Then X=⟨X]∩[X⟩X=\langle X]\cap[X\rangle.

Proof. Let X=I∩FX=I\cap F for some hyperideal II and type-III hyperfilter FF. Then, clearly, ⟨X]⊆I\langle X]\subseteq I and [X⟩⊆F[X\rangle\subseteq F. Let I1I_1 be a hyperideal of L\mathcal{L} containing XX. Suppose that a∈Ia\in I and let x∈Xx\in X be arbitrary. Then, a∨x⊆Ia\vee x\subseteq I. Also, x∈F,a∈Lx\in F,a\in\mathcal{L} implies that a∨x⊆Fa\vee x\subseteq F since FF is a type-III hyperfilter. This implies a∨x⊆I∩F=X⊆I1a\vee x\subseteq I\cap F=X\subseteq I_1. Note that there exists u∈a∨xu\in a\vee x such that a⪯ua\preceq u. Thus, u∈I1u\in I_1 and a⪯ua\preceq u implies that a∈I1a\in I_1. Hence, I⊆I1I\subseteq I_1. Now ⟨X]\langle X] is a hyperideal containing XX. By the above argument, we get that I⊆⟨X]I\subseteq\langle X], and hence I=⟨X]I=\langle X]. Now, let F1F_1 be any type-III hyperfilter of L\mathcal{L} containing XX. Let p∈Fp\in F. Then, for any q∈Xq\in X, we have p∧q∈Fp\wedge q\in F. Also, since q∈Iq\in I, we get that p∧q∈Ip\wedge q\in I. Therefore p∧q∈X⊆F1p\wedge q\in X\subseteq F_1. Now, p∧q∈F1p\wedge q\in F_1 and p∧q⪯pp\wedge q\preceq p implies that p∈F1p\in F_1. This shows that F⊆F1F\subseteq F_1. Therefore F=[X⟩F=[X\rangle. □

The following theorems are generalizations of Proposition 3.2, Proposition 3.9 and Theorem 3.22 to arbitrary join hyperlattices. Their proofs are omitted, as they follow directly from the arguments used in the corresponding results. In the sequel, let L\mathcal{L} be such that SC⁡(L)=CS⁡(L)\operatorname{SC}(\mathcal{L})=\operatorname{CS}(\mathcal{L}).

Theorem 4.4. For a join hyperlattice L\mathcal{L}, (SC⁡(L),⊑)(\operatorname{SC}(\mathcal{L}),\sqsubseteq) is a poset.

Theorem 4.5. The poset (SC⁡(L),⊑)(\operatorname{SC}(\mathcal{L}),\sqsubseteq) is a lattice with meet and join as given in Proposition 3.9.

Theorem 4.6. For a hyperlattice L\mathcal{L}, the map f:SC⁡(L)⟶I(L)×F(L)f:\operatorname{SC}(\mathcal{L})\longrightarrow I(\mathcal{L})\times F(\mathcal{L}) defined by, for C∈SC⁡(L)C\in\operatorname{SC}(\mathcal{L}), f(C)=(⟨C],[C⟩)f(C)=(\langle C], [C\rangle), is an injective homomorphism. Moreover, the sublattice K={(I,D):I∈I(L),D∈F(L),I∩D≠∅}\mathcal{K}=\{(I,D):I\in I(\mathcal{L}),D\in F(\mathcal{L}),I\cap D\ne\varnothing\} of I(L)×F(L)I(\mathcal{L})\times F(\mathcal{L}) is cover preserving and is isomorphic to (SC⁡(L),⊑)(\operatorname{SC}(\mathcal{L}),\sqsubseteq).

Conclusion

It is well known that the family CS⁡(L)∪{∅}\operatorname{CS}(\mathcal{L})\cup\{\varnothing\}, ordered by inclusion, forms a lattice. In this work, we have extended the study of this structure. First, we have shown that CS⁡(L)∪{∅}\operatorname{CS}(\mathcal{L})\cup\{\varnothing\} is not only a lattice but an algebraic lattice, and we have characterized its compact elements. Then, we have introduced two new orders on CS⁡(L)\operatorname{CS}(\mathcal{L}), namely ≤b\leq_b and ⊑\sqsubseteq, and have examined their properties. We have proved that for A,B∈CS⁡(L)A,B\in\operatorname{CS}(\mathcal{L}), A≤bBA\leq_b B implies A⊑BA\sqsubseteq B. However, these two orders are not generally equivalent, though they coincide for PP-hyperlattices. We have shown, by example, that in an arbitrary join hyperlattice, the relation ⊑\sqsubseteq need not be a partial order. Next, we have characterized convex subhyperlattices of PP-hyperlattices using initial and final segments, and have embedded the lattice (CS⁡(L),⊑)(\operatorname{CS}(\mathcal{L}),\sqsubseteq) into the product lattice I(L)×F(L)I(\mathcal{L})\times F(\mathcal{L}). We have introduced the notion of strongly convex sets and have shown that every such set is a convex subhyperlattice, but not vice versa. Although many results are given for PP-hyperlattices, the key arguments do not rely on the PP-hyperlattice condition. Consequently, these results also hold for hyperlattices satisfying SC⁡(L)=CS⁡(L)\operatorname{SC}(\mathcal{L})=\operatorname{CS}(\mathcal{L}). Finally, we have justified the use of type-III hyperfilters in defining strongly convex sets, since type-III hyperfilters, unlike other types, form subhyperlattices. We have proved that when L\mathcal{L} is a PP-hyperlattice, (CS⁡(L),≤b)(\operatorname{CS}(\mathcal{L}),\leq_b) is a lattice, and we have given an example of a non-strong hyperlattice where this does not hold. Based on our observations, we have conjectured that (CS⁡(L),≤b)(\operatorname{CS}(\mathcal{L}),\leq_b) is a lattice for every strong hyperlattice. In conclusion, it is natural to investigate whether the operator SC⁡\operatorname{SC} preserves finite products of hyperlattices and to identify the structural conditions under which such preservation holds. Additionally, examining how homomorphisms and congruence relations on a hyperlattice L\mathcal{L} carry over to the induced structure on SC⁡(L)\operatorname{SC}(\mathcal{L}) offers scope for further research.

Acknowledgement(s)

The first author acknowledges the Department of Science and Technology (DST), Government of India, for the DST INSPIRE Fellowship to carry-out this research work. The first, fourth, and sixth authors acknowledge the Manipal Institute of Technology, Manipal Academy of Higher Education, Manipal for its kind encouragement. The second author acknowledges the Department of Mathematics, Mangalore university, Mangalagangotri for its support. The third and fifth authors thank the Manipal Institute of Technology Bengaluru, Manipal Academy of Higher Education, Manipal for its encouragement and institutional support.

References

References

  1. [1]Reza Ameri, M Amiri-Bideshki, and A Borumand Saeid. Some properties of residual mapping and convexity in Λ-hyperlattices. Ratio Math., 27(1):27–36, 2014.DOI
  2. [2]K Bhargava, S Parameshwara Bhatta, and Chandru Hegde. On strongly-convex subtrellises of a trellis. Acta Math. Univ. Comen., 95(2):79–94, 2026.
  3. [3]K Bhargava and Chandru Hegde. Some properties of strongly-convex subtrellises of a trellis. Palest. J. Math., 14(2), 2025.
  4. [4]M Amiri Bideshki, Reza Ameri, and A Borumand Saeid. On prime hyperfilters (hyperideals) in Λ-hyperlattices. Eur. J. Pure Appl. Math., 11(1):169–188, 2018.
  5. [6]George Grätzer. Lattice Theory: Foundation. Springer, 2011.DOI
  6. [7]George Grätzer. General lattice theory, volume 52. Birkhäuser, 2012.
  7. [8]M Konstantinidou. On p-hyperlattices and their distributivity. Rend. Circ. Mat. Palermo, 42:391–403, 1993.
  8. [9]M Konstantinidou and J Mittas. An introduction to the theory of hyperlattices. Math. Balkanica, 7:187–193, 1977.
  9. [10]S Lavanya and S Parameshwara Bhatta. A new approach to the lattice of convex sublattices of a lattice. Algebra Universalis, 35(1):63–71, 1996.DOI
  10. [11]N. Aishwarya Nayak, Pallavi Panjarike, Syam Prasad Kuncham, Tapatee Sahoo, and Harikrishnan Panackal. Hyperfilters and convex subhyperlattices in a join hyperlattice. Indian Journal of Pure and Applied Mathematics, 56(3):1036–1047, 2025.
  11. [12]Panjarike Pallavi, Syam Prasad Kuncham, Gopadi Ramachandra Bhatta Vadiraja, and Panackal Harikrishnan. Computation of prime hyperideals in meet hyperlattices. Bull. Comput. Appl. Math., 10(1):33–58, 2022.
  12. [14]S Rasouli and B Davvaz. Lattices derived from hyperlattices. Commun. Algebra., 38(8):2720–2737, 2010.DOI
  13. [15]A Soltani Lashkenari and B Davvaz. Complete join hyperlattices. Indian J. Pure Appl. Math., 46:633–645, 2015.DOI

Paper details

vlnc:W_25mc4cgu
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents