betaFeedback

WorksMathematics

Sequential convergence in the lower topology and co-sobriety of Scott spaces

Abstract

We investigate how order structure controls sequential convergence in the lower topology of a poset and how this, in turn, governs topological duality and co-sobriety of Scott spaces. We first give a purely order-theoretic characterization of lower convergence. For complete lattices this yields a lower-limsup formula, and on complete chains the lower limsup coincides with the usual lattice limsup. We also characterize first countability of the lower topology at a point by the existence of a countable coinitial subset outside its principal ideal. We prove that every compact saturated subset of a Scott space is sequentially closed in the lower topology and deduce that, for every poset, the lower topology and the co-compact (de Groot dual) topology of its Scott space have exactly the same convergent sequences and limits. For coherent Scott spaces we identify the lower closure of a kk-irreducible compact saturated set with its double-cone closure and obtain an infimum criterion for co-sobriety. Consequently, every bounded-complete dcpo with sequential lower topology has a co-sober Scott space. In particular, every countable bounded-complete dcpo, and hence every countable complete lattice, has a co-sober Scott space.

Keywords: Lower topology; sequential convergence; Scott topology; bounded-complete dcpo; complete lattice; co-sobriety.

2020 Mathematics Subject Classification: 06F30, 06B35, 54A20, 54D30.

Funding: This research was supported by the National Natural Science Foundation of China (Nos. 12471070, 12071199, 12561083).

Introduction

Co-sobriety was introduced by Escardó, Lawson and Simpson in their study of a dual form of the Hofmann–Mislove theorem [2]. A T0T_0-space is co-sober if every kk-irreducible compact saturated subset is the saturation of a point. Co-sobriety and sobriety are distinct: sober spaces need not be co-sober, and even a sober Scott space can fail to be co-sober [8, 9].

Let PP be a poset. We write ω(P)\omega(P) for its lower topology, σ(P)\sigma(P) for its Scott topology, and ΣP=(P,σ(P))\Sigma P=(P,\sigma(P)) for its Scott space. The co-compact, or de Groot dual, topology of ΣP\Sigma P will be denoted by σ(P)d\sigma(P)^d; it is generated by the complements of the compact saturated subsets of ΣP\Sigma P.

Topology-level duality between one-sided order topologies and de Groot duals has been studied under stable compactness and related structural hypotheses [5, 11, 12]. In particular, systematic criteria for when the de Groot dual of a Scott topology is the lower topology are developed in [12 Chapters 9–10]. Lawson and Xu developed a systematic account of the lower topology on general T0T_0-spaces and its interaction with compactness, well-filteredness and coherence [6]. Zhao, He and Wang proved, by a different method, that every locally compact coherent T0T_0-space is co-sober [14]. The present paper takes a different viewpoint, centered on the underlying order. Rather than assuming global topological hypotheses that force topology-level duality, we begin with the order relation itself and use ordinary sequences as the organizing principle. This leads from an order-theoretic description of lower convergence to topological consequences for de Groot duality and co-sobriety.

Our first result makes the order-theoretic nature of lower convergence explicit. For a poset PP and a sequence (xn)n≥1(x_n)_{n\ge1} in PP, (xn)n≥1(x_n)_{n\ge1} converges to xx in ω(P)\omega(P) precisely when every element not below xx is eventually not below xnx_n. Equivalently,

⋂m≥1↓{xn:n≥m}⊆↓x.\bigcap_{m\ge1}\mathord{\downarrow}\{x_n:n\ge m\}\subseteq\mathord{\downarrow}x.

For a complete lattice this gives a natural lower limsup. If

lim sup⁡ℓxn=⋁⋂m≥1↓{xn:n≥m},\limsup_{\ell} x_n=\bigvee\bigcap_{m\ge1}\mathord{\downarrow}\{x_n:n\ge m\},

then

xn⟶x in ω(L)⟺lim sup⁡ℓxn≤x.x_n\longrightarrow x\text{ in }\omega(L)\quad\Longleftrightarrow\quad\limsup_{\ell}x_n\le x.

On a complete chain, lim sup⁡ℓxn\limsup_{\ell}x_n agrees with the usual lattice limsup

⋀m≥1⋁n≥mxn.\bigwedge_{m\ge1}\bigvee_{n\ge m}x_n.

Thus, for the extended real line with its usual order, lower-topological convergence is exactly the one-sided condition lim sup⁡ℓxn≤x\limsup_{\ell}x_n\le x.

The same order-theoretic viewpoint also identifies the precise local countability condition behind this convergence. The lower topology is first-countable at xx if and only if P∖↓xP \setminus\mathord{\downarrow}x has a countable coinitial subset. This supplies a purely order-theoretic source of first countability, and hence of the Fréchet–Urysohn and sequential properties, even for uncountable posets.

The order-theoretic convergence description then interacts directly with the Scott topology. Our main sequential duality theorem states that, for an arbitrary poset PP, the topologies ω(P)\omega(P) and σ(P)d\sigma(P)^{d} have the same convergent sequences, with the same limits. The topologies themselves can differ [11], but they coincide whenever the lower topology is sequential. In particular, topology-level equality follows if the lower topology is Fréchet–Urysohn or first-countable, and it holds for every countable poset. Thus the global duality studied in the literature has an unconditional sequential analogue: ordinary sequences cannot distinguish ω(P)\omega(P) from the de Groot dual of the Scott topology.

For coherent Scott spaces we then determine the lower closure of a kk-irreducible compact saturated set KK:

cl⁡ω(P)K=Kℓu.\operatorname{cl}_{\omega(P)} K = K^{\ell u}.

where KℓK^{\ell} is the set of lower bounds of KK and KℓuK^{\ell u} is the set of upper bounds of KℓK^{\ell}. If ⋀K\bigwedge K exists, the formula reduces to

cl⁡ω(P)K=↑⋀K.\operatorname{cl}_{\omega(P)} K = \mathord{\uparrow}\bigwedge K.

Together with lower sequential closedness of compact saturated sets, this yields an infimum criterion for co-sobriety. We apply the criterion to bounded-complete dcpos. If the lower topology is sequential—in particular, if it is Fréchet–Urysohn or first-countable—then the Scott space is co-sober. Hence the first-countability criterion gives a purely order-theoretic sufficient condition without requiring the dcpo itself to be countable.

On the other hand, Scott co-sobriety can fail without a countability assumption even for complete Boolean algebras. Ji and Xu constructed a complete Boolean algebra whose Scott space is not co-sober [4]. Their example is necessarily uncountable, since every countable complete Boolean algebra is finite [4 Proposition 5.4]. Thus the uncountable and countable settings exhibit sharply different behavior. One of the main consequences below is therefore a structural theorem for countable ordered sets with sufficient completeness: every countable bounded-complete dcpo has a co-sober Scott space, and hence so does every countable complete lattice.

Miao, Xi, Li and Zhao constructed a countable complete distributive lattice whose Scott space is not sober [7]. Our result shows that their example is nevertheless co-sober. Thus, even among countable complete distributive lattices, Scott sobriety and Scott co-sobriety separate.

The paper is organized as follows. Section 2 develops the order theory of sequential convergence in the lower topology, including the lower-limsup and coinitiality criteria for first countability, and then connects this order convergence with compact saturated subsets of Scott spaces. Section 3 establishes sequential duality between the lower topology and the co-compact topology. Section 4 derives the lower-closure and infimum criteria for coherent Scott spaces. Section 5 applies these results to bounded-complete dcpos and complete lattices.

Sequential convergence in the lower topology

For A⊆PA \subseteq P, put

↑A={x∈P:(∃a∈A) a≤x},↓A={x∈P:(∃a∈A) x≤a}.\uparrow A = \{x \in P : (\exists a \in A)\ a \leq x\}, \qquad\downarrow A = \{x \in P : (\exists a \in A)\ x \leq a\}.

For a singleton, we write ↑x\uparrow x and ↓x\downarrow x.

The lower topology ω(P)\omega(P) is generated by

{P∖↑x:x∈P}\{P \setminus\uparrow x : x \in P\}

as a subbase. Thus the finite intersections

P∖↑F:=P∖⋃a∈F↑a,F⊆P finite,P \setminus\uparrow F := P \setminus\bigcup_{a \in F} \uparrow a,\qquad F \subseteq P\ \text{finite},

form a base.

A set U⊆PU \subseteq P is Scott open if it is an upper set and, whenever a directed set DD has a supremum with ⋁D∈U\bigvee D \in U, one has D∩U≠∅D \cap U \neq\varnothing. The Scott open sets form the topology σ(P)\sigma(P), called the Scott topology of PP.

For a T0T_0-space XX, let Q(X)\mathcal{Q}(X) be the family of its nonempty compact saturated subsets. No separation axiom is included in the meaning of compactness. A member K∈Q(X)K \in\mathcal{Q}(X) is kk-irreducible if

K=K1∪K2,K1,K2∈Q(X),K = K_1 \cup K_2,\quad K_1,K_2 \in\mathcal{Q}(X),

implies K=K1K = K_1 or K=K2K = K_2. The space XX is co-sober if every kk-irreducible member of Q(X)\mathcal{Q}(X) is the saturation of a point. A T0T_0-space is coherent if the intersection of any two compact saturated subsets is compact.

A subset AA of a topological space XX is said to be sequentially closed if for any sequence in AA, AA contains all its limits. The space is sequential if every sequentially closed subset is closed. It is known that every first-countable space is Fréchet–Urysohn and every Fréchet–Urysohn space is sequential (see, e.g., [1 Theorem 1.6.14]).

For A⊆PA \subseteq P, define

Aℓ={x∈P:x≤a for every a∈A},A^\ell= \{x \in P : x \leq a\ \text{for every }a \in A\},

and

Aℓu={y∈P:x≤y for every x∈Aℓ}.A^{\ell u} = \{y \in P : x \leq y\ \text{for every }x \in A^\ell\}.

Proposition 2.1 (Order characterization of lower convergence). Let (xn)n≥1(x_n)_{n\ge1} be a sequence in a poset PP and let x∈Px\in P. The following are equivalent.

  1. xn→xx_n\to x in ω(P)\omega(P).

  1. For every a∈Pa\in P with a≰xa\nleq x, one has a≰xna\nleq x_n eventually.

3.

⋂m≥1↓{xn:n≥m}⊆↓x.\bigcap_{m\ge1}\downarrow\{x_n:n\ge m\}\subseteq\downarrow x.

Proof. A subbasic lower-open set P∖↑aP\setminus\uparrow a contains xx exactly when a≰xa\nleq x. Convergence to all subbasic neighbourhoods is equivalent to convergence in the topology, because eventual membership in finitely many subbasic neighbourhoods implies eventual membership in their intersection. This proves (1) ⇔\Leftrightarrow (2).

Assume (2), and let zz belong to the intersection in (3). If z≰xz\nleq x, then (2) gives z≰xnz\nleq x_n eventually. On the other hand, membership of zz in every tail-downset means that for every mm there is n≥mn\ge m with z≤xnz\le x_n, a contradiction. Thus z≤xz\le x.

Conversely, if (2) fails, there is a a≰xa\nleq x such that a≤xna\le x_n for infinitely many nn. Hence for every ll some n≥ln\ge l satisfies a≤xna\le x_n, so

a∈⋂l≥1↓{xn:n≥l},a\in\bigcap_{l\ge1}\downarrow\{x_n:n\ge l\},

contradicting (3). Therefore (3) ⇒\Rightarrow (2). □

For a sequence (xn)(x_n) in a complete lattice LL, define its lower limsup by

limsup⁡ℓxn:=⋁⋂m≥1↓{xn:n≥m}.\mathop{\operatorname{limsup}}\limits_{\ell}x_n:=\bigvee\bigcap_{m\ge1}\downarrow\{x_n:n\ge m\}.

We also write

limsup⁡oxn:=⋀m≥1⋁n≥mxn\mathop{\operatorname{limsup}}\limits_{o}x_n:=\bigwedge_{m\ge1}\bigvee_{n\ge m}x_n

for the usual lattice limsup.

Corollary 2.2. Let LL be a complete lattice and (xn)n≥1(x_n)_{n\ge1} a sequence in LL. Then, for every x∈Lx\in L,

xn→x in ω(L)⟺limsup⁡ℓxn≤x.x_n\to x\text{ in }\omega(L)\Longleftrightarrow\mathop{\operatorname{limsup}}\limits_{\ell}x_n\le x.

Moreover,

limsup⁡ℓxn≤limsup⁡oxn.\mathop{\operatorname{limsup}}\limits_{\ell}x_n\le\mathop{\operatorname{limsup}}\limits_{o}x_n.

Proof. By Proposition 2.1, lower convergence to xx is equivalent to

⋂m≥1↓{xn:n≥m}⊆↓x,\bigcap_{m\ge1}\downarrow\{x_n:n\ge m\}\subseteq\downarrow x,

which, in a complete lattice, is equivalent to the join of the left-hand side being at most xx.

Let

A=⋂m≥1↓{xn:n≥m}.A=\bigcap_{m\ge1}\downarrow\{x_n:n\ge m\}.

For a∈Aa\in A and every mm, some n≥mn\ge m satisfies a≤xna\le x_n, hence

a≤⋁n≥mxn.a\le\bigvee_{n\ge m}x_n.

Therefore a≤lim sup⁡oxna\le\limsup_o x_n for every a∈Aa\in A. Taking joins gives lim sup⁡ℓxn≤lim sup⁡oxn\limsup_\ell x_n\le\limsup_o x_n.

Proposition 2.3. If LL is a complete chain, then for every sequence (xn)n≥1(x_n)_{n\ge1} in LL,

lim sup⁡ℓxn=lim sup⁡oxn.\limsup_\ell x_n=\limsup_o x_n.

Consequently,

xn→x in ω(L)⟺lim sup⁡oxn≤x.x_n\to x\text{ in }\omega(L)\Longleftrightarrow\limsup_o x_n\le x.

Proof. Put

A=⋂m≥1↓{xn:n≥m},a=⋁A,r=⋀m≥1⋁n≥mxn.A=\bigcap_{m\ge1}\downarrow\{x_n:n\ge m\},\qquad a=\bigvee A,\qquad r=\bigwedge_{m\ge1}\bigvee_{n\ge m}x_n.

By Corollary 2.2, a≤ra\le r. Suppose that a<ra<r.

If there is b∈Lb\in L with a<b<ra<b<r, then for every mm we have

b<r≤⋁n≥mxn.b<r\le\bigvee_{n\ge m}x_n.

Since LL is a chain, some n≥mn\ge m must satisfy b≤xnb\le x_n; otherwise bb would be an upper bound of the tail. Hence b∈Ab\in A, contradicting b>a=⋁Ab>a=\bigvee A.

It remains to consider the case where no element lies strictly between aa and rr. For every mm, since a<r≤⋁n≥mxna<r\le\bigvee_{n\ge m}x_n, some n≥mn\ge m satisfies a<xna<x_n. By the absence of an element strictly between aa and rr, this implies r≤xnr\le x_n. Thus r∈Ar\in A, again contradicting a=⋁A<ra=\bigvee A<r. Hence a=ra=r.

Remark 2.4. For the extended real line R‾\overline{\mathbb{R}} with its usual order, Proposition 2.3 says

xn→x in the lower topology⟺lim sup⁡n→∞xn≤x.x_n\to x\text{ in the lower topology}\Longleftrightarrow\limsup_{n\to\infty}x_n\le x.

Thus lower-topological convergence is a direct order-topological extension of the familiar one-sided limsup condition from analysis.

We next isolate the exact local countability condition for the lower topology. A subset CC of a poset SS is called coinitial if for every s∈Ss \in S there is c∈Cc \in C with c≤sc \le s, or equivalently, S=↑CS = \uparrow C.

Theorem 2.5 (First-countability criterion). Let PP be a poset and x∈Px \in P. The lower topology ω(P)\omega(P) is first-countable at xx if and only if P∖↓xP \setminus\downarrow x has a countable coinitial subset.

Proof. Suppose first that ω(P)\omega(P) is first-countable at xx. Let (Bn)n≥1(B_n)_{n \ge1} be a countable local base at xx. For every nn, choose a basic lower-open neighbourhood Vn=P∖↑FnV_n = P \setminus\uparrow F_n, where Fn⊆P∖↓xF_n \subseteq P \setminus\downarrow x is finite, such that

x∈Vn⊆Bn.x \in V_n \subseteq B_n.

Put C=⋃n≥1FnC = \bigcup_{n \ge1} F_n. Then CC is countable and C⊆P∖↓xC \subseteq P \setminus\downarrow x. Let a∈P∖↓xa \in P \setminus\downarrow x. Since P∖↑aP \setminus\uparrow a is a neighbourhood of xx, there is nn with

Bn⊆P∖↑a.B_n \subseteq P \setminus\uparrow a.

Hence

P∖↑Fn=Vn⊆P∖↑a,P \setminus\uparrow F_n = V_n \subseteq P \setminus\uparrow a,

so ↑a⊆↑Fn\uparrow a \subseteq\uparrow F_n. In particular, a∈↑Fna \in\uparrow F_n, and therefore some c∈Fn⊆Cc \in F_n \subseteq C satisfies c≤ac \le a. Thus CC is coinitial in P∖↓xP \setminus\downarrow x.

Conversely, let C⊆P∖↓xC \subseteq P \setminus\downarrow x be countable and coinitial. If C=∅C = \varnothing, then P∖↓x=∅P \setminus\downarrow x = \varnothing, so xx is a greatest element and {P}\{P\} is a local base at xx. Assume therefore that C≠∅C \ne\varnothing, and enumerate it as C={c1,c2,…}C = \{c_1,c_2,\ldots\}, repeating elements if CC is finite. Put

Un=P∖⋃k=1n↑ck.U_n = P \setminus\bigcup_{k=1}^{n} \uparrow c_k.

Then each UnU_n is a neighbourhood of xx. We show that (Un)n≥1(U_n)_{n \ge1} is a local base at xx. Let

V=P∖⋃i=1m↑aiV = P \setminus\bigcup_{i=1}^{m} \uparrow a_i

be a basic neighbourhood of xx, so each ai≰xa_i \nleq x. By coinitiality, choose cki≤aic_{k_i} \le a_i for i=1,…,mi = 1,\ldots,m. If n≥max⁡ikin \ge\max_i k_i, then

⋃i=1m↑ai⊆⋃k=1n↑ck,\bigcup_{i=1}^{m} \uparrow a_i \subseteq\bigcup_{k=1}^{n} \uparrow c_k,

and hence Un⊆VU_n \subseteq V. Thus ω(P)\omega(P) is first-countable at xx.

Corollary 2.6. For a poset PP, the following are equivalent.

(1) ω(P)\omega(P) is first-countable.

(2) For every x∈Px \in P, the poset P∖↓xP \setminus\mathord{\downarrow}x has a countable coinitial subset.

In either case ω(P)\omega(P) is Fréchet–Urysohn and hence sequential.

We now turn to compact saturated subsets of Scott spaces. The following compact-down lemma is standard; see, for example, [3 Section II-1] and [13 Lemma 3.14]. The proof is included because the topology on PP is otherwise arbitrary.

Lemma 2.7. Let PP be equipped with a topology in which every principal upper set is closed. If A⊆PA \subseteq P is compact in that topology, then ↓A\mathord{\downarrow}A is Scott closed.

Proof. The set ↓A\mathord{\downarrow}A is a lower set. Let D⊆↓AD \subseteq\mathord{\downarrow}A be directed, suppose that d=⋁Dd = \bigvee D exists, and assume that d∉↓Ad \notin\mathord{\downarrow}A. Then d≰ad \nleq a for every a∈Aa \in A. For each a∈Aa \in A, choose da∈Dd_a \in D such that da≰ad_a \nleq a. Thus

{P∖↑e:e∈D}\{P \setminus\mathord{\uparrow}e : e \in D\}

is an open cover of AA. Compactness gives e1,…,en∈De_1,\ldots,e_n \in D with

A⊆⋃i=1n(P∖↑ei).A \subseteq\bigcup_{i=1}^{n}(P \setminus\mathord{\uparrow}e_i).

Choose e∈De \in D above all eie_i. Since P∖↑ei⊆P∖↑eP \setminus\mathord{\uparrow}e_i \subseteq P \setminus\mathord{\uparrow}e for every ii, we have A⊆P∖↑eA \subseteq P \setminus\mathord{\uparrow}e. On the other hand, e∈↓Ae \in\mathord{\downarrow}A, so e≤ae \leq a for some a∈Aa \in A, a contradiction. Hence d∈↓Ad \in\mathord{\downarrow}A.

Lemma 2.8. Suppose that xn→xx_n \to x in ω(P)\omega(P). For n≥1n \geq1, let

An={x}∪{xm:m≥n},Fn=↓An.A_n = \{x\} \cup\{x_m : m \geq n\},\quad F_n = \mathord{\downarrow}A_n.

Then every FnF_n is Scott closed and

⋂n≥1Fn=↓x.\bigcap_{n \geq1} F_n = \mathord{\downarrow}x.

Proof. The set AnA_n is a convergent sequence tail together with its limit, hence it is compact in ω(P)\omega(P). Principal upper sets are closed in the lower topology, so Lemma 2.7 shows that FnF_n is Scott closed.

The inclusion ↓x⊆⋂nFn\mathord{\downarrow}x \subseteq\bigcap_n F_n is immediate. Conversely, let z∈⋂nFnz \in\bigcap_n F_n and suppose that z≰xz \nleq x. Then x∈P∖↑z∈ω(P)x \in P \setminus\mathord{\uparrow}z \in\omega(P). As xn→xx_n \to x in ω(P)\omega(P), there is mm such that xn∈P∖↑zx_n \in P \setminus\mathord{\uparrow}z for all n≥mn \geq m, and hence z∉Fmz \notin F_m, which contradicts z∈⋂nFn⊆Fmz \in\bigcap_n F_n \subseteq F_m. Thus $\bigcap_{n\geq 1} F_n = \mathord{\downarrow}x.

Proposition 2.9. Every compact saturated subset of ΣP\Sigma P is sequentially closed in ω(P)\omega(P).

Proof. Let K∈Q(ΣP)K \in\mathcal{Q}(\Sigma P), and suppose that xn∈Kx_n \in K and xn→xx_n \to x in ω(P)\omega(P). Let FnF_n be as in Lemma 2.8. The sets K∩FnK \cap F_n form a decreasing sequence of nonempty closed subsets of the compact subspace KK: indeed, xn∈K∩Fnx_n \in K \cap F_n. Hence

⋂n≥1(K∩Fn)≠∅.\bigcap_{n \ge1}(K \cap F_n) \ne\varnothing.

No separation axiom is used in this compactness argument. By Lemma 2.8, the intersection is K∩↓xK \cap\mathord{\downarrow}x. Choose z∈K∩↓xz \in K \cap\mathord{\downarrow}x. Since KK is an upper set and z≤xz \le x, it follows that x∈Kx \in K.

Sequential duality with the co-compact topology

Let σ(P)d\sigma(P)^d be the topology generated by

{P∖K:K∈Q(ΣP)}\{P \setminus K : K \in\mathcal{Q}(\Sigma P)\}

as a subbase.

Theorem 3.1 (Sequential duality theorem). For every poset PP, the topologies ω(P)\omega(P) and σ(P)d\sigma(P)^d have the same convergent sequences and the same limits.

Proof. For every p∈Pp \in P, the principal upper set ↑p\mathord{\uparrow}p is compact and saturated in ΣP\Sigma P: a Scott open set containing pp contains all of ↑p\mathord{\uparrow}p. Therefore P∖↑pP \setminus\mathord{\uparrow}p is open in σ(P)d\sigma(P)^d, and

ω(P)⊆σ(P)d.\omega(P) \subseteq\sigma(P)^d.

Thus convergence in σ(P)d\sigma(P)^d implies convergence, to the same limit, in ω(P)\omega(P).

Conversely, suppose that xn→xx_n \to x in ω(P)\omega(P), and let K∈Q(ΣP)K \in\mathcal{Q}(\Sigma P) with x∉Kx \notin K. If infinitely many xnx_n belonged to KK, a subsequence contained in KK would still converge to xx in ω(P)\omega(P). Proposition 2.9 would then give x∈Kx \in K, a contradiction. Hence xn∉Kx_n \notin K eventually. The sequence is eventually in every subbasic σ(P)d\sigma(P)^d-neighbourhood of xx, and therefore in every finite intersection of such neighbourhoods. Thus xn→xx_n \to x in σ(P)d\sigma(P)^d.

Corollary 3.2. If ω(P)\omega(P) is sequential, then

ω(P)=σ(P)d.\omega(P) = \sigma(P)^d.

In particular, the equality holds whenever ω(P)\omega(P) is Fréchet–Urysohn or first-countable, and it holds for every countable poset PP.

Proof. By Theorem 3.1, every σ(P)d\sigma(P)^d-open set is sequentially open with respect to ω(P)\omega(P). If ω(P)\omega(P) is sequential, it is therefore ω(P)\omega(P)-open. The reverse inclusion always holds. The Fréchet–Urysohn and first-countable cases follow from the standard implications. If PP is countable, the defining subbase of ω(P)\omega(P) is countable, so ω(P)\omega(P) is second-countable and hence sequential.

Combining Theorem 2.5 with Corollary 3.2 gives a purely order-theoretic version.

Corollary 3.3. If, for every x∈Px \in P, the set P∖↓xP \setminus\downarrow x has a countable coinitial subset, then

ω(P)=σ(P)d.\omega(P) = \sigma(P)^{d}.

Remark 3.4. Corollary 3.2 concerns the co-compact topology of the Scott space. It does not identify this topology, or the lower topology, with the Scott topology of the order dual PopP^{\mathrm{op}}.

The distinction between Theorem 3.1 and topology-level de Groot duality is also worth emphasizing. Global criteria for σ(P)d=ω(P)\sigma(P)^{d} = \omega(P) involve additional structural hypotheses and are part of the theory of stable compactness and ordered patch spaces; see [5, 11, 12]. Theorem 3.1 requires no such hypothesis: it asserts an unconditional equality of the sequential convergence structures. Sequentiality of ω(P)\omega(P) is exactly what is needed here to promote this sequential duality to equality of the topologies.

Coherent Scott spaces

Proposition 4.1. Suppose that ΣP\Sigma P is coherent. If K∈Q(ΣP)K \in Q(\Sigma P) is kk-irreducible, then

cl⁡ω(P)K=Kℓu.\operatorname{cl}_{\omega(P)} K = K^{\ell u}.

Proof. For every a∈Kℓa \in K^{\ell}, one has K⊆↑aK \subseteq\uparrow a. Since ↑a\uparrow a is closed in ω(P)\omega(P),

cl⁡ω(P)K⊆⋂a∈Kℓ↑a=Kℓu.\operatorname{cl}_{\omega(P)} K \subseteq\bigcap_{a \in K^{\ell}} \uparrow a = K^{\ell u}.

For the reverse inclusion, let y∉cl⁡ω(P)Ky \notin\operatorname{cl}_{\omega(P)} K. There are a1,…,an∈Pa_1,\ldots,a_n \in P such that

y∈P∖(↑a1∪⋯∪↑an)y \in P \setminus(\uparrow a_1 \cup\cdots\cup\uparrow a_n)

and

K⊆↑a1∪⋯∪↑an.K \subseteq\uparrow a_1 \cup\cdots\cup\uparrow a_n.

Hence

K=⋃i=1n(K∩↑ai).K = \bigcup_{i=1}^{n}(K \cap\uparrow a_i).

Discard the empty terms. Each principal upper set is compact saturated, and coherence implies that every remaining K∩↑aiK \cap\uparrow a_i is compact saturated. Finite unions of compact saturated sets are compact saturated, so binary kk-irreducibility applies inductively to this finite union. Thus K=K∩↑aiK = K \cap\uparrow a_i for some ii. Hence ai∈Kℓa_i \in K^{\ell}. Since ai≰ya_i \not\leq y, we have y∉Kℓuy \notin K^{\ell u}. □\square

Corollary 4.2. Under the hypotheses of Proposition 4.1, if ⋀K\bigwedge K exists, then

cl⁡ω(P)K=↑⋀K.\operatorname{cl}_{\omega(P)} K = \uparrow\bigwedge K.

Proof. If c=⋀Kc = \bigwedge K, then Kℓ=↓cK^{\ell} = \mathord{\downarrow}c and Kℓu=↑cK^{\ell u} = \mathord{\uparrow}c.

Theorem 4.3. Suppose that ΣP\Sigma P is coherent and ω(P)\omega(P) is sequential. Then ΣP\Sigma P is co-sober if and only if every kk-irreducible member of Q(ΣP)Q(\Sigma P) has an infimum in PP.

Proof. Suppose first that ΣP\Sigma P is co-sober. Its specialization order is the original order on PP, so the saturation of xx is ↑x\mathord{\uparrow}x. Hence every kk-irreducible K∈Q(ΣP)K \in Q(\Sigma P) has the form K=↑xK = \mathord{\uparrow}x, and ⋀K=x\bigwedge K = x.

Conversely, let K∈Q(ΣP)K \in Q(\Sigma P) be kk-irreducible and suppose that ⋀K\bigwedge K exists. By Proposition 2.9, KK is sequentially closed in ω(P)\omega(P). Since ω(P)\omega(P) is sequential, KK is closed. Corollary 4.2 now gives

K=cl⁡ω(P)K=↑⋀K.K = \operatorname{cl}_{\omega(P)} K = \mathord{\uparrow}\bigwedge K.

Thus every kk-irreducible compact saturated set is the saturation of a point.

Bounded-complete dcpos

We use the convention that a bounded-complete dcpo is a dcpo in which every subset having an upper bound has a supremum, including the empty subset. In particular, ⊥=⋁∅\bot= \bigvee\varnothing exists. This convention is stated explicitly because a weaker usage of the term also occurs in the literature.

Xi and Lawson proved that the Scott space of every complete lattice or bounded-complete dcpo is well-filtered and coherent [10 Corollary 4.3].

Theorem 5.1. Let PP be a bounded-complete dcpo. If ω(P)\omega(P) is sequential, then ΣP\Sigma P is co-sober.

Proof. We verify the infimum condition in Theorem 4.3. Let A⊆PA \subseteq P be nonempty. Its set of lower bounds

Aℓ={x∈P:x≤a for every a∈A}A^{\ell} = \{x \in P : x \leq a \text{ for every } a \in A\}

is nonempty because ⊥∈Aℓ\bot\in A^{\ell}. It is bounded above by any fixed a0∈Aa_0 \in A, so

c=⋁Aℓc = \bigvee A^{\ell}

exists. Every a∈Aa \in A is an upper bound of AℓA^{\ell}, and therefore c≤ac \leq a. Thus cc is the greatest lower bound of AA. In particular, every kk-irreducible compact saturated subset has an infimum. Coherence follows from the result of Xi and Lawson, so Theorem 4.3 applies.

Corollary 5.2. Let PP be a bounded-complete dcpo. If ω(P)\omega(P) is Fréchet–Urysohn, then ΣP\Sigma P is co-sober. In particular, this holds if ω(P)\omega(P) is first-countable.

Proof. Every Fréchet–Urysohn space is sequential, and every first-countable space is Fréchet–Urysohn. Apply Theorem 5.1.

Corollary 5.3. Let PP be a bounded-complete dcpo. If, for every x∈Px \in P, the set P∖↓xP \setminus\downarrow x has a countable coinitial subset, then ΣP\Sigma P is co-sober.

Proof. By Theorem 2.5, ω(P)\omega(P) is first-countable. Apply Corollary 5.2.

Corollary 5.4. Every countable bounded-complete dcpo has a co-sober Scott space.

Proof. The lower topology of a countable poset is second-countable and hence sequential. Apply Theorem 5.1.

Corollary 5.5. Every countable complete lattice, and hence every countable complete Heyting algebra, has a co-sober Scott space.

The countability hypothesis in Corollary 5.5 is essential for such a general statement. Indeed, Ji and Xu constructed a complete Boolean algebra BB whose Scott space ΣB\Sigma B is not co-sober [4]. Since every countable complete Boolean algebra is finite, this counterexample is necessarily uncountable. Consequently, even within complete Boolean algebras, Scott co-sobriety is automatic in the countable case but may fail once countability is dropped.

Corollary 5.6. There exists a countable complete distributive lattice whose Scott space is co-sober but not sober.

Proof. Miao, Xi, Li and Zhao constructed a countable complete distributive lattice whose Scott space is not sober [7]. Corollary 5.4 gives co-sobriety.

Concluding remarks

The results above show how order structure, sequential convergence and Scott topology interact in a direct way. Proposition 2.1 shows that lower convergence is controlled by eventual order avoidance; on complete chains it is exactly a limsup condition. Theorem 3.1 shows that ordinary sequences cannot distinguish the lower topology from the co-compact topology of the Scott space, even though the two topologies may be different. First countability and the Fréchet–Urysohn property provide natural bridges from this sequential duality to topology-level duality.

For coherent Scott spaces, Proposition 4.1 makes the order-theoretic content of lower closure explicit. When the relevant infima exist, co-sobriety is reduced to the passage from lower sequential closedness to lower closedness. This explains why sequential properties of the lower topology lead directly to co-sobriety results for bounded-complete dcpos.

Several questions suggested by this viewpoint remain open. First, it would be useful to obtain intrinsic order-theoretic characterizations of when ω(P)\omega(P) is Fréchet–Urysohn or sequential, extending the first-countability criterion in Theorem 2.5. Second, for complete lattices one may ask for structural conditions under which the lower limsup lim sup⁡ℓ\limsup_{\ell} agrees with the lattice limsup lim sup⁡o\limsup_{o} beyond the case of complete chains. Finally, the sequentiality hypothesis in Theorem 5.1 is sufficient but is not claimed to be necessary; weaker closure assumptions ensuring that kk-irreducible compact saturated subsets are lower closed would lead to further co-sobriety criteria.

Statements and Declarations

Funding. This research was supported by the National Natural Science Foundation of China (Nos. 12471070, 12071199, 12561083).

Competing interests. The authors have no relevant financial or nonfinancial interests to disclose.

Use of generative AI. During the preparation of this manuscript, the authors used ChatGPT (OpenAI) for language editing, presentation suggestions, and manuscript checking. The authors independently verified all mathematical arguments and references, revised the output as necessary, and take full responsibility for the final content.

Data availability. No data were used for the research described in this article.

References

  1. [1]R. Engelking, General Topology, Polish Scientific Publishers, Warsaw, 1989.
  2. [2]M. Escardó, J.D. Lawson, A. Simpson, Comparing Cartesian closed categories of (core) compactly generated spaces, Topology Appl. 143 (1–3) (2004) 105–145. https://doi.org/10.1016/j.topol.2004.02.011.
  3. [3]G. Gierz, K.H. Hofmann, K. Keimel, J.D. Lawson, M. Mislove, D.S. Scott, Continuous Lattices and Domains, Encyclopedia of Mathematics and its Applications, Vol. 93, Cambridge University Press, Cambridge, 2003. https://doi.org/10.1017/CBO9780511542725.
  4. [4]W. Ji, X. Xu, Scott spaces of complete Boolean algebras need not be co-sober, arXiv:2608.08484 (2026).
  5. [5]J.D. Lawson, Stably compact spaces, Math. Struct. Comput. Sci. 21 (1) (2011) 125–169. https://doi.org/10.1017/S0960129510000319.
  6. [6]J.D. Lawson, X. Xu, T₀-spaces and the lower topology, Math. Struct. Comput. Sci. 34 (6) (2024) 467–490. https://doi.org/10.1017/S0960129524000240.
  7. [7]H. Miao, X. Xi, Q. Li, D. Zhao, Not every countable complete distributive lattice is sober, Math. Struct. Comput. Sci. 33 (9) (2023) 809–831. https://doi.org/10.1017/S0960129523000269.
  8. [8]C. Shen, G. Wu, X. Xi, D. Zhao, Sober Scott spaces are not always co-sober, Topology Appl. 282 (2020) 107316. https://doi.org/10.1016/j.topol.2020.107316.
  9. [9]X. Wen, X. Xu, Sober is not always co-sober, Topology Appl. 250 (2018) 48–52. https://doi.org/10.1016/j.topol.2018.10.003.
  10. [10]X. Xi, J.D. Lawson, On well-filtered spaces and ordered sets, Topology Appl. 228 (2017) 139–144. https://doi.org/10.1016/j.topol.2017.06.002.
  11. [11]X. Xu, A note on duals of topologies, Topology Appl. 199 (2016) 95–101. https://doi.org/10.1016/j.topol.2015.12.006.
  12. [12]X. Xu, Order and Topology (in Chinese), 2nd ed., Science Press, Beijing, 2022.
  13. [13]X. Xu, D. Zhao, On topological Rudin’s lemma, well-filtered spaces and sober spaces, Topology Appl. 272 (2020) 107080. https://doi.org/10.1016/j.topol.2020.107080.
  14. [14]B. Zhao, Z. He, K. Wang, On co-sober spaces, Houston J. Math. 49 (4) (2023) 971–988.DOI

Paper details

vlnc:W_cw8f2f5z
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

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