WorksMathematics
Wilker property, function spaces and hyperspaces
Abstract
The Wilker property requires that a compact set covered by two open sets admit compact covering pieces inside those open sets. We study this property through products, compact-open function spaces and hyperspaces. We construct countable compact Fréchet–Urysohn hereditarily Wilker spaces with non-Wilker products and squares. For spaces whose compact subsets are closed, Wilker-ness is equivalent to every compact subspace being Hausdorff; this yields a positive product theorem when one Wilker factor has closed compact subsets. For an irreducible subset system , we characterize preservation of -sobriety by compact-open function spaces in terms of -supercompactness of compact subsets of the domain, and show that Wilker spaces satisfy this criterion. The Hoare hyperspace of a space is Wilker exactly when binary union on the Scott space of its open-set lattice is jointly continuous for the ordinary product topology. Consequently, the Hoare construction preserves and reflects Wilker-ness on consonant bases, although preservation fails in general. We also identify core-compactness of the base with local compactness of its Hoare hyperspace and prove that the Smyth construction reflects Wilker-ness.
Acknowledgements. This research was supported by the National Natural Science Foundation of China (Nos. 12471070 and 12071199).
Introduction
The compact-cover splitting condition studied in this paper has two classical sources. It appears already in Choquet’s theory of capacities, where a family of sets satisfying the corresponding binary refinement property is called rich [3]. It was later isolated by Wilker as condition (D) in his study of adjoint product and hom functors [17]. We follow the established terminology and call a space Wilker if, whenever a compact set is contained in the union of two open sets , one can cover by compact sets . Wilker proved, in particular, that every Hausdorff space and every locally compact space has this property.
The condition also occurs in measure-extension theory [10] and in compact-open constructions for multifunction spaces [14]. More recently, every quasi-Polish space was shown to be Wilker [2], every LCS-complete space is Wilker [1], and co-consonance implies Wilker-ness through strong compactness properties [23].
We develop the basic theory of the Wilker property and examine its connections with compact-open function spaces and hyperspaces. The hyperspaces considered here are the Hoare and Smyth powerspaces, equipped with the lower and upper Vietoris topologies, respectively. These constructions connect structural questions in general topology with the order-theoretic study of open-set lattices.
We first study the topology of the class itself. Wilker spaces are hereditary with respect to open, closed and locally closed subspaces, are preserved by retracts, and are preserved and reflected by arbitrary topological sums. Arbitrary heredity, however, fails in a particularly strong way: every -space is homeomorphic to a subspace of a locally hypercompact Wilker space. A new separation-theoretic feature emerges when Wilker-ness is combined with the KC property: we prove that
This characterization yields a mixed product theorem. If is Wilker and every compact subspace of is Hausdorff, then is Wilker. Consequently,
and in particular the product of a Wilker space with a Hausdorff space is Wilker. The KC hypothesis cannot be omitted. We construct two countable compact Fréchet–Urysohn anti-Hausdorff hereditarily Wilker spaces whose product is not Wilker (Theorem 3.16). A finite sum gives a hereditarily Wilker space with a non-Wilker square; a co-finite-core construction gives an anti-Hausdorff square counterexample as well. Consequently, the full category of Wilker spaces is neither reflective nor epireflective in .
The counterexample uses the co-finite modifications studied by Xu and Ji in their preprint Products of co-sober spaces need not be co-sober (2026, arXiv:2610.14434v1) for co-sobriety. Its additional ingredient is that every proper closed subspace of the particular Kovár–van Douwen seed is compact Hausdorff (Lemma 3.14). This makes the singleton modifications hereditarily Wilker. An explicit binary open cover of a compact diagonal then has no compact refinement: the projections of any proposed pieces would express the original anti-Hausdorff KC-space as a union of two proper closed sets. The modifications themselves are credited to the earlier work; the Wilker analysis and the compact-cover obstruction are established here. Together with the mixed product theorem, this locates failure of productivity precisely outside the guarantee supplied by a KC factor.
We next compare Wilker-ness with core-compactness. The rational line is Wilker but not core-compact. In the opposite direction, we show that every subspace of is Wilker whenever is Hausdorff and carries the Scott topology of the reverse usual order. Applied to the classical Hofmann–Lawson construction, this yields a second-countable core-compact Wilker -space that is not locally compact. Hence
is a strict implication, while the question whether every core-compact -space is Wilker remains open.
The Scott-topological consequences make the connection with domain theory more explicit. The Scott space of every quasicontinuous poset is locally compact [13]; hence it is Wilker. For complete lattices one can go further: since their Scott spaces are well-filtered, core-compactness already forces local compactness and therefore Wilker-ness [21, 23]. On the other hand, complete lattices have coherent Scott spaces, and Zhao, He and Wang showed that coherent Wilker spaces are co-sober. Combining this with the recent construction of Ji and Xu of a complete Boolean algebra with a non-co-sober Scott space, given in their preprint Scott spaces of complete Boolean algebras need not be co-sober (2026, arXiv:2608.08484v1), yields a complete Boolean algebra whose Scott space is not Wilker. A parallel phenomenon occurs for upper topologies: the dual of Isbell’s lattice, together with the lower-topology example of Wen and Xu and the coherence theorem of Xu, Yang and Chen, yields a complete lattice whose upper topology is not Wilker. Thus strong order-completeness alone does not force Wilker-ness for either of these standard one-sided order topologies. The unrestricted Boolean, Heyting and meet-continuous Scott-space versions all fail, while the corresponding countable questions for complete Heyting algebras and complete meet-continuous lattices remain open.
The second part of the paper turns from the basic theory to compact-open function spaces. For a -space , let
where is generated by the sets
with compact in . Given an irreducible subset system , we call a subset -supercompact if every -family in whose union covers already contains one member covering . For a fixed , we prove the equivalence of the following four conditions:
is -sober;
every compact subset of is -supercompact;
for every -space the compact-open topology on has property with respect to ;
is -sober for every -sober space .
This reorganizes several compact-open function-space questions from [20] into a single structural statement.
Wilker’s condition is the main bridge to this theory. We prove that every compact subset of a Wilker space is -supercompact for every irreducible subset system . Consequently, if is Wilker, then is -sober for every , the compact-open topology on has property with respect to every , and is -sober whenever is. In particular, is sober for every Wilker space . If is also consonant, then , and the sobriety of the Scott space follows immediately.
The third part concerns the Hoare and Smyth hyperspace constructions. For the nonempty Hoare powerspace we prove
We further show that every is coherent. Since coherent Wilker spaces are co-sober, a known countable zero-dimensional metrizable space with non-co-sober gives
Thus the Hoare construction does not preserve Wilker-ness in general. A further characterization identifies exactly what the Wilker property of measures:
The domain here is the ordinary product of Scott spaces, not the Scott topology of the order product. The proof combines the compact-family correspondence of de Brecht and Kawai [2 Theorem 6.10] with a coherent-space refinement lemma. Wilker-ness of itself is characterized by continuity of union on . Hence, for every consonant base,
This gives a positive solution of the Hoare-reflection problem under consonance; the unrestricted implication remains open. The relation between consonance, Wilker-ness and Scott continuity of union is already recorded in [7 Proposition J]; the additional point here is its connection with the Hoare criterion.
For the Smyth powerspace , the known local-compactness/core-compactness characterization is complemented by the reflection theorem
Thus the two power constructions interact with the Wilker property and with core-compactness in markedly different ways.
The paper is organized as follows. Section 2 fixes notation and recalls the function-space preliminaries. Section 3 develops the basic theory, the KC–Wilker characterization and the mixed product theorem, followed by the product and square counterexamples and the non-reflectivity consequence. Section 4 studies the relation with core-compactness and applications to order topologies. Sections 5 and 6 introduce -supercompactness and apply Wilker’s condition to compact-open function spaces. Section 7 treats the Hoare construction, including the Scott-union criterion and consonant reflection; Section 8 treats the Smyth construction. Section 9 singles out two central open problems, Smyth preservation and the core-compactness implication, with a brief discussion of their related directions. The final section summarizes the resulting structural picture.
Preliminaries
Throughout, spaces are unless explicitly stated otherwise. For a space , denotes its lattice of open subsets, ordered by inclusion. A subset is saturated if it is an intersection of open sets. We write for the set of all nonempty compact saturated subsets of .
Definition 2.1 (Choquet–Wilker [3, 17]) A topological space is a Wilker space if, whenever is compact and satisfy , there are compact subsets such that
The compact sets are allowed to be empty and need not be subsets of . The original condition is binary, but the finite form is immediate.
Lemma 2.2 If is Wilker, is compact, and
with each open, then there are compact such that
Proof. Induct on , using the binary Wilker property for .
We shall repeatedly use the following elementary fact.
Lemma 2.3 The saturation of a compact subset of a topological space is compact.
Proof. Let be compact and let be an open cover of . It covers , so finitely many members have union containing . Since is open and contains , it contains . Hence those finitely many cover .
Thus the Wilker condition may equivalently be stated using compact saturated sets only.
For later use it is convenient to adjoin the empty compact saturated set and write
ordered by the Smyth order
Binary meets in are unions.
For a compact subset , put
The topology on generated by the family is denoted by , and
As observed in [20],
and has inclusion as its specialization order. The latter can also be seen directly, since singletons are compact.
Let be an irreducible subset system in the sense of [19]: for every -space ,
and continuous maps preserve -subsets under direct image. A space is -sober if every has a unique generic point, i.e., there is a unique such that
The systems , , and yield, respectively, -spaces, well-filtered spaces (for both and ), and sober spaces; see [19, 21].
For spaces , let be the set of continuous maps from to . The compact-open topology on has subbasic open sets
where is compact and . The resulting function space is denoted by .
We use property from [20]. An order-compatible topology on has property with respect to if, for every and every satisfying
one has .
We shall use the following result proved in [20 Theorem 36].
Proposition 2.4. Let be an irreducible subset system, let be -spaces, and let be a topology on finer than the pointwise convergence topology. If is -sober, then is -sober if and only if has property with respect to .
We shall also use the standard homeomorphism
where is the Sierpiński space with open sets ; see [20].
Basic theory of Wilker spaces
This section develops the structural properties of the Wilker condition used throughout the paper. After elementary permanence properties and the KC characterization, we prove the mixed product theorem and then give counterexamples to unrestricted productivity, already for squares. The same examples yield non-reflectivity.
Heredity, retracts and sums
Proposition 3.1 (Open and closed heredity) Let be Wilker.
(1) Every open subspace of is Wilker.
(2) Every closed subspace of is Wilker.
(3) Consequently, every locally closed subspace of is Wilker.
Proof. Let .
If is open, every compact is compact in , and every open subset of is open in . The Wilker decomposition in therefore already lies in .
If is closed, write with open in . If is compact and , apply Wilker-ness of to , obtaining compact covering . Then is compact, , and .
The locally closed case follows by composition of the first two cases. □
The preceding result is essentially the strongest general hereditary statement one can expect. In fact, arbitrary heredity fails in a very strong form.
Theorem 3.2 (Failure of arbitrary heredity) Every -space is homeomorphic to a subspace of a locally hypercompact Wilker space. Consequently, the class of Wilker spaces is not hereditary with respect to arbitrary subspaces.
Proof. Let be the Sierpiński space and let be any set. Identify with by sending to
Under this identification, a basis for the product topology consists of the sets
Each is supercompact. Indeed, if an open cover of is given, one member contains the point ; since every open subset of is an upper set for inclusion, that member contains all of . Thus has a basis of supercompact open sets. In particular, it is locally hypercompact, hence locally compact and therefore Wilker by Wilker’s classical theorem.
Now let be a -space. The canonical map
is a topological embedding. Indeed, it is continuous because the inverse image under the -coordinate of the open singleton is exactly ; it is injective because is ; and these same coordinate inverse images recover the topology of . Hence every -space embeds into a locally hypercompact Wilker space.
Finally, non-Wilker -spaces exist. For example, Wilker observed that the one-point compactification of the rational line is a compact KC non-Hausdorff space and does not satisfy condition (D) [17]. Therefore embeds into a Wilker space while itself being non-Wilker, so arbitrary heredity fails.
Remark 3.3 The proof shows more than non-heredity. The Sierpiński powers , which are naturally identified with the Scott spaces of the algebraic lattices , form a universal source of locally hypercompact Wilker spaces for embeddings. This is one place where the basic theory of Wilker spaces meets a standard domain-theoretic representation.
Proposition 3.4 (Retracts) Every retract of a Wilker space is Wilker.
Proof. Let and be continuous with , and suppose is Wilker. Let be compact and with open. Then is compact and
Choose compact covering . Then is compact, lies in , and the cover .
Theorem 3.5 (Topological sums) For a family of spaces, the topological sum is Wilker if and only if every is Wilker.
Proof. Each summand is clopen, so the forward implication follows from Proposition 3.1.
Conversely, let with every Wilker, and let be compact. Since the summands form an open cover of , only finitely many of them meet . On each of those finitely many components apply Wilker-ness to and the restrictions of the two given open sets. Finite unions of the resulting compact pieces give the required decomposition in .
Let denote the full subcategory of consisting of Wilker spaces. Proposition 3.4 and Theorem 3.5 show that is closed under retracts and arbitrary coproducts.
Separation properties
Recall that a space is if every compact subset is closed. Following Wilker [17, p. 279], a space is called a space if every compact subspace is normal: for any disjoint subsets that are closed in , there are such that
Here normality does not include the axiom. Wilker proved that every space satisfies condition (D), and in particular every Hausdorff space is Wilker [17 Theorem 8]. The following elementary characterization shows that, after imposing the condition, Wilker-ness exactly forces compact subspaces to be Hausdorff.
Theorem 3.6 (KC–Wilker characterization). For a space , the following conditions are equivalent.
(1) is and Wilker.
(2) Every compact subspace of is Hausdorff.
Proof. (1) (2): Assume first that is and Wilker, and let be compact. Since is , is closed in ; hence is Wilker by Proposition 3.1. Let be distinct. Singletons are compact, hence closed in a space, so
is an open cover of the compact space . By Wilker-ness there are compact subsets
with . The sets and are compact as subsets of , hence closed by the property. Therefore
are disjoint open neighborhoods in of and , respectively. Thus is Hausdorff.
(2) (1): Suppose that every compact subspace of is Hausdorff. We first show that is . Let be compact. If , then is compact and hence Hausdorff by assumption. Since is compact, it is closed in the Hausdorff subspace , contradicting . Thus every compact subset of is closed.
It remains to prove that is Wilker. Let be compact, with open in . The subspace is compact Hausdorff, hence normal. Applying the finite shrinking lemma in to the open cover
we obtain closed subsets of such that
Each is compact, so these sets give the required Wilker decomposition.
Corollary 3.7 If is compact and , then is Wilker if and only if is Hausdorff.
Proof. The forward implication follows from Theorem 3.6, since itself is a compact subspace. The reverse implication is Wilker’s classical Hausdorff result.
Remark 3.8 Thus Wilker-ness by itself is far from being a separation axiom, but in the presence of it has a rigid separation-theoretic meaning: it is equivalent to Hausdorffness on every compact subspace. In particular, every compact non-Hausdorff space is automatically non-Wilker.
Mixed products
The preceding characterization gives a positive product theorem under a separation hypothesis on one factor. The counterexamples in Subsection 3.4 will show that this hypothesis cannot simply be omitted.
Theorem 3.9 (Mixed product theorem) Let be a Wilker space, and suppose that every compact subspace of is Hausdorff. Then is Wilker.
Proof. Let be compact and let be open in with
Put . Then is compact and hence Hausdorff by hypothesis. Moreover, Theorem 3.6 shows that is .
Fix and write
Since is , is closed, so is compact. Also
By the Wilker property of , choose compact sets
such that . For each , compactness of and openness of give an open set and an open neighborhood of in such that
If , take and . Put
Then . Hence
is compact and . Since is KC, the compact set is closed. Therefore
is an open neighborhood of , and
The set
is an open neighborhood of in the compact Hausdorff space . Choose an open neighborhood of in such that
Then is compact; since is KC, it is also closed in . Consequently
is compact, and so is its projection
By (3.1) and ,
Apply the Wilker property of once more to obtain compact sets
with . Since ,
and the two compact rectangles cover . The family covers the compact space . Choose a finite subcover
For , set
Each is compact, , and . Hence is Wilker.
Combining Theorems 3.6 and 3.9 gives a convenient intrinsic form.
Corollary 3.10 If and are Wilker spaces and at least one of them is KC, then is Wilker.
Proof. Suppose, for example, that is KC. By Theorem 3.6, every compact subspace of is Hausdorff, so Theorem 3.9 applies. The other case follows by symmetry.
Corollary 3.11 (Hausdorff factor) If is Wilker and is Hausdorff, then is Wilker.
Proof. Every compact subspace of a Hausdorff space is Hausdorff, so this is an immediate consequence of Theorem 3.9.
Remark 3.12. Consequently, if the product of two Wilker spaces fails to be Wilker, then both factors must be non-KC, and in particular both must be non-Hausdorff. The next subsection constructs counterexamples of exactly this type.
Remark 3.13. A naive Wallace–Tychonoff argument does not establish productivity in the non-Hausdorff case. Given compact covered by finitely many open rectangles, compactness of the projections and Tychonoff’s theorem do not by themselves give a cover of by compact rectangles subordinate to those open rectangles. The proof of Theorem 3.9 succeeds because the second factor makes compact projections closed and supplies finite shrinking inside the compact Hausdorff space . The counterexample below shows that an unrestricted compact-cover refinement cannot hold.
Failure of product and square stability
We now prove that the separation hypothesis in the mixed product theorem cannot be omitted. The construction uses the co-finite modifications of Xu and Ji (arXiv: 2610.14434v1, Sections 3 and 6), originally used to study products of co-sober spaces. We retain those constructions and analyze their Wilker property; the obstruction to splitting a compact cover in the product is proved directly.
A space is hereditarily Wilker if all of its subspaces are Wilker. A nonempty space is anti-Hausdorff if any two nonempty open sets intersect. We use the particular countably infinite compact Fréchet–Urysohn anti-Hausdorff KC-space obtained from van Douwen’s construction [4] in the realization described by Kovár [11 Example 2.2 and the following paragraph, p. 6]; see also the product preprint (arXiv:2610.14434v1, Lemma 5.1). In this realization there are a family of infinite subsets of and a map , with , such that a closed subbasis for is
Here we use the specific realization with the stated topological properties, not an arbitrary choice of an almost-disjoint family and of the map . The following additional consequence of the closed subbasis is what is needed for Wilker-ness.
Lemma 3.14. Every proper closed subspace of is compact Hausdorff. Consequently, for each , every compact subspace of is Hausdorff.
Proof. Let be a member of (3). For , the set is closed in , so is open in . Thus every point of except possibly is isolated. Since is , an isolated point and any other point are separated by and . Hence is Hausdorff. It is also compact because it is closed in the compact space .
Let be closed and nonempty. Choose a point of and a basic open neighborhood of it disjoint from . The complement of this neighborhood is a finite union of members of (3) containing . A finite union of closed Hausdorff subspaces is Hausdorff: each diagonal is closed in , hence in , and is closed. Therefore is Hausdorff. It is compact by closedness in ; the empty case is immediate.
Finally, a compact subset of is compact in , and hence closed in by the KC property. Since it omits , it is proper, so the first assertion applies.
Let more generally be a -space and . Following the product preprint (arXiv:2610.14434v1, Section 3), write for the co-finite topology, including the empty set, and put
This is a compact topology contained in . Every subset containing is compact, because every neighborhood of is co-finite. The restrictions of and to coincide: for , the set is -open and belongs to . In particular, a subset avoiding is compact in if and only if it is compact in . These are the singleton cases recorded in the product preprint (arXiv:2610.14434v1, Lemma 3.1).
Lemma 3.15. The space is Wilker if and only if the original subspace is Wilker. If every compact subspace of is Hausdorff, then is hereditarily Wilker.
Proof. Necessity of the first assertion follows from open heredity. Conversely, suppose that is Wilker, and let , where is compact in and the $U_i are open in . If , put
The set is compact because it contains , and is finite because is co-finite. They lie in , respectively, and cover . The case is symmetric. If neither open set contains , apply Wilker-ness in the original open subspace . Notice that the compact pieces need not be subsets of , as in Definition 2.1.
For the hereditary assertion, take . If , every compact subset of is a compact Hausdorff subset of . Theorem 3.6 makes Wilker. If , every relative neighborhood of is co-finite in , and every subset of containing is compact. The argument above applies inside , using the already established Wilker property of .
Theorem 3.16 (Failure of productivity) There exist two countably infinite compact Fréchet–Urysohn anti-Hausdorff hereditarily Wilker spaces whose product is not Wilker.
Proof. Take the space used in Lemma 3.14 and distinct . By Lemmas 3.14 and 3.15, the modifications are hereditarily Wilker. They are compact and , and they remain anti-Hausdorff because their topologies are coarser than . They are also Fréchet–Urysohn, as proved in the product preprint (arXiv:2610.14434v1, Lemma 3.3). Indeed, at the modified point sequence of distinct points of any infinite set converges to that point; away from it, closure can be tested in the original open complement, where the Fréchet–Urysohn property applies.
The diagonal map from to is continuous. Thus its image
is compact. The two open sets
cover , since .
Suppose that compact sets covered . The projection is compact in and avoids , so it is compact in and therefore is a proper -closed set. Similarly, is a proper -closed set avoiding . Every belongs to or , so . This contradicts anti-Hausdorffness of . Hence the cover (3.4) has no Wilker refinement.
The compact sets in the proof are allowed to contain points outside . No intersection is assumed compact. Also, the proof does not infer failure of Wilker-ness merely from the known failure of co-sobriety of the product: it excludes a compact-cover splitting directly by taking projections.
Corollary 3.17 (A non-Wilker square) There is a countable compact Fréchet–Urysohn T hereditarily Wilker space whose square is not Wilker.
Proof. Let for the two spaces in Theorem 3.16. A finite topological sum preserves compactness, the Fréchet–Urysohn property and . Every subspace of is a sum of subspaces of the two factors, so Theorem 3.5 gives hereditary Wilker-ness. The product is a clopen subspace of . If were Wilker, Proposition 3.1 would contradict Theorem 3.16. □
The square can also be made anti-Hausdorff. We use the co-finite-core construction in the product preprint (arXiv:2610.14434v1, § 6) and record its additional Wilker permanence property.
Proposition 3.18 Let be disjoint Wilker spaces and an infinite set disjoint from them. On put
Then is Wilker. Consequently, a countable compact Fréchet–Urysohn anti-Hausdorff Wilker space can have a non-Wilker square.
Proof. The basic properties of (3.5) are established in the product preprint (arXiv:2610.14434v1, Lemma 6.1): are closed subspaces with their original topologies, and is compact if and only if and are compact in their respective spaces. In particular, every subset of is compact.
Let be a compact set covered by two open sets of . Split using compact covering pieces , and similarly split using compact . Put
Then are compact subsets of covering .
For the last assertion, take disjoint copies of from Theorem 3.16 and countable . The construction preserves compactness and the Fréchet–Urysohn property and gives a anti-Hausdorff space (arXiv:2610.14434v1, Lemma 6.1). Its square contains the closed subspace , which is not Wilker. Closed heredity completes the proof. □
Categorical consequence
Theorem 3.19 The full subcategory of is neither reflective nor epireflective. More precisely, the square in Corollary 3.17 has no reflection into Wilker spaces.
Proof. Let be as in Corollary 3.17, and suppose that were a reflection into Wilker spaces. Since is Wilker, its coordinate projections factor uniquely through : there are continuous with . Then satisfies . The uniqueness clause of the reflection, applied to , gives , since both sides have composite with . Thus is a homeomorphism, contradicting that is Wilker and is not. Hence is not reflective, and therefore cannot be epireflective.
This is the product obstruction to reflectivity, expressed directly by the universal property. Together with Theorem 3.2, it settles the unrestricted heredity and reflection questions negatively. The behavior under quotients and quotient-like maps is not settled by these arguments and remains a separate direction.
Sources of Wilker spaces and the core-compactness problem
Wilker proved that every space satisfies condition (D), hence every Hausdorff space is Wilker; he also proved that every locally compact space is Wilker [17]. Later work produced large non-Hausdorff classes. Every quasi-Polish space is Wilker [2], every LCS-complete space is Wilker [1], and every co-consonant -space is Wilker by the strong-compactness argument of [23].
Recall that a space is Alexandroff if arbitrary intersections of open subsets of are open. Equivalently, every point has a least open neighborhood, namely
where is taken in the specialization order. Thus the open sets of an Alexandroff -space are exactly the upper sets of its specialization order.
For completeness we record two elementary sources; under our standing convention the zero-dimensional case is already covered by Hausdorffness, while the Alexandroff case is genuinely non-Hausdorff.
Proposition 4.1 Every zero-dimensional space is Wilker. Every Alexandroff space is Wilker.
Proof. Let be zero-dimensional, compact. For each choose a clopen neighborhood contained in one of the . Compactness gives finitely many such . Let be the union of those assigned to and put . Each is closed in , hence compact, lies in , and the two sets cover .
Now let be Alexandroff. By Lemma 2.3 we may assume that is compact saturated. Since is an upper set and the principal upper sets form an open cover of , compactness yields a finite such that . Assign each point of to one of the containing it, and let be the corresponding finite subsets. Then are compact, lie in , and cover .
Core-compactness
Recall that is core-compact if is a continuous lattice. Every locally compact space is core-compact, and for sober spaces core-compactness is equivalent to local compactness [5]. Hence local compactness gives
Neither condition alone characterizes local compactness; the examples below also distinguish their conjunction from local compactness.
The Wilker property does not imply core-compactness: the rational line is Hausdorff, hence Wilker, but it is not locally compact; being Hausdorff, it is sober, so it cannot be core-compact.
The opposite implication is one of the central open problems of this paper.
Question 4.2 (Core-compact versus Wilker) Is every core-compact -space Wilker?
The classical Hofmann–Lawson example does not provide a negative answer. In fact it yields a rather strong positive test case. We first prove a general lemma.
Let equipped with the Scott topology of the reverse usual order. Thus the open sets of are , .
Theorem 4.3 Let be a Hausdorff space. Every subspace of is Wilker.
Proof. Let be compact and let be open in with . Write with open in . Let be projection and put . Then is compact Hausdorff.
For , the fiber
is nonempty and compact: indeed, is homeomorphic to the closed subspace of , because is Hausdorff. Every nonempty compact subset of has a largest element in the usual order. Indeed, if were not attained (or were equal to ), the increasing family of open initial intervals with (respectively ) would cover without a finite subcover. Write
The map is upper semicontinuous. For ,
Here is closed in , since its complement is Scott open. Hence the set inside the projection is compact; its image is compact in the Hausdorff space , and therefore closed. Thus is upper semicontinuous in the ordinary real-valued sense. For and , let
where denotes the open neighborhoods of in . Each is lower semicontinuous: if , one may choose and a witnessing neighborhood ; then .
Put
Since is upper semicontinuous and is lower semicontinuous, is open in . Moreover . Indeed, for the point belongs to , hence to one of the . By openness, there are and an open neighborhood of such that , so .
Since is compact Hausdorff, the finite shrinking lemma gives closed sets with . Define
Then each is compact and . We claim . If , then . Choose with . By the definition of the supremum, some is witnessed by a neighborhood of with . Since , we obtain .
Hofmann and Lawson constructed a second-countable core-compact -space which is not locally compact; in fact every compact subset of has empty interior [5, 9]. Their space is a subspace of .
Corollary 4.4 There exists a second-countable core-compact Wilker -space that is not locally compact.
Proof. Apply Theorem 4.3 to the Hofmann–Lawson space.
Thus
In particular, even the conjunction of core-compactness and Wilker-ness does not force local compactness.
Scott spaces and domain-theoretic consequences
We now record several consequences for Scott spaces. They provide a direct link between the basic theory developed above and standard classes from domain theory.
Recall that Mao and Xu call a poset quasicontinuous if, for every , there is a nonempty finite set such that
It is clear that the Scott space of every quasicontinuous poset is locally compact (cf. [13 Proposition 2.6]). Wilker’s classical theorem therefore gives the following immediate consequence.
Corollary 4.5. If is a quasicontinuous poset, then its Scott space is Wilker.
In particular, the conclusion applies to quasicontinuous domains, continuous posets and continuous domains. Notice that directed completeness is not required in Corollary 4.5.
The core-compact Scott-space problem has a complete answer for complete lattices.
Proposition 4.6. Let be a complete lattice. If is core-compact, then is locally compact, sober and Wilker.
Proof. The Scott space of every complete lattice is well-filtered; see, for example, [18]. Every core-compact well-filtered -space is locally compact (and sober) by [21]. Since every locally compact space is Wilker, the conclusion follows. □
Thus the specialization of Question 4.2 to complete lattices is positive. For general Scott spaces, however, a natural domain-theoretic version of the question remains.
Question 4.7 (Core-compact Scott spaces). Let be a poset, or even a dcpo. If its Scott space is core-compact, must be Wilker?
We next record a useful obstruction. Recall that a -space is co-sober if every -irreducible compact saturated subset is the saturation of a point. Zhao, He and Wang proved that every Wilker -space is weakly co-sober and that, in a coherent space, weak co-sobriety implies co-sobriety [23]. Hence:
Proposition 4.8. Every coherent Wilker -space is co-sober.
Since the Scott space of every complete lattice is coherent, we immediately obtain the following complete-lattice form.
Corollary 4.9. Let be a complete lattice. If is Wilker, then is co-sober.
This turns recent non-co-sobriety examples into Wilker counterexamples without any additional analysis of compact covers. Ji and Xu constructed a complete Boolean algebra such that is not co-sober; in fact their example is also non-sober and well-filtered (arXiv:2608.08484v1). Corollary 4.9 therefore yields:
Corollary 4.10. There exists a complete Boolean algebra whose Scott space is not Wilker. Consequently, there exist a complete Heyting algebra and a complete meet-continuous lattice whose Scott spaces are not Wilker.
Remark 4.11. The counterexample in Corollary 4.10 is much stronger than a merely distributive complete-lattice example: every complete Boolean algebra is a frame, hence a complete Heyting algebra, and in particular a complete meet-continuous lattice. Thus none of these three unrestricted classes has Wilker Scott spaces automatically.
Countability changes the picture at the Boolean level. Ji and Xu proved that every countable complete Boolean algebra is finite (arXiv:2608.08484v1, Proposition 5.4). Hence its Scott topology is finite Alexandroff and therefore Wilker by Proposition 4.1. The next two cases are substantially less clear.
Question 4.12 (Countable Scott–Wilker problems)
Is the Scott space of every countable complete Heyting algebra (equivalently, every countable frame) Wilker?
Is the Scott space of every countable complete meet-continuous lattice Wilker?
The second problem admits an even broader variant in which “complete meet-continuous lattice” is replaced by “countable meet-continuous dcpo”. These countable questions are not consequences of Corollary 4.10, since the Boolean case collapses to the finite case. They therefore isolate a genuinely countable boundary between the positive quasicontinuous theory and the unrestricted Boolean counterexample.
Upper topologies of complete lattices
There is a parallel negative phenomenon for the upper topology. For a poset , let denote the upper topology generated by the subbasic sets
Equivalently, , where denotes the lower topology.
Wen and Xu proved that Isbell’s complete lattice equipped with its lower topology is sober but not co-sober [16]. Let . Then is again a complete lattice and
as topological spaces. On the other hand, Xu, Yang and Chen proved that the upper topology of every sup-complete poset is coherent (indeed, it satisfies considerably stronger properties) [22]. Hence is coherent. Proposition 4.8 now gives the following.
Corollary 4.13 There exists a complete lattice whose upper topology is not Wilker.
Proof. For the lattice above, is not co-sober but is coherent. If it were Wilker, Proposition 4.8 would make it co-sober, a contradiction.
Remark 4.14 Corollaries 4.10 and 4.13 give two complementary complete-lattice obstructions:
Scott topology | even complete Boolean algebras may fail to be Wilker,
upper topology | even complete lattices may fail to be Wilker.
Thus Wilker-ness is not forced merely by completeness, even for standard order-compatible topologies.
H-supercompactness and compact-open function spaces
We now introduce the graded compactness notion that controls the compact-open topology.
Definition 5.1 Let be an irreducible subset system and let be a -space. A subset is called -supercompact if, for every ,
When , we also say that is irreducibly supercompact.
Remark 5.2 If , then every -supercompact subset is an -supercompact subset. Hence, for the standard systems,
Ordinary supercompactness implies -supercompactness.
The directed level is exactly ordinary compactness.
Proposition 5.3 For a subset of a -space , the following are equivalent:
(1) is compact;
(2) is -supercompact.
Proof. Suppose first that is compact, and let with . Then a finite subfamily covers . Since is directed under inclusion, there exists with . Hence .
Conversely, assume that is -supercompact and let be an open cover of . Let be the directed family of all finite unions of members of . Then , so for some . Thus finitely many members of cover .
The next observation is the technical link between compact-open function spaces and .
Lemma 5.4 Let be -spaces and let . Then the map
is continuous.
Proof. For every compact ,
which is open in the compact-open topology.
We now obtain the central equivalence.
Theorem 5.5 Let be an irreducible subset system and let be a -space. Then the following conditions are equivalent:
is -sober;
every compact subset of is -supercompact;
for every -space , the compact-open topology on has property with respect to ;
for every -sober space , the function space is -sober.
Proof. (1) (2). Let be compact and let satisfy . By the -sobriety of , there is a unique such that
Since the specialization order of is inclusion, every satisfies , whence .
For the reverse inclusion, let . Then . Since , the open set meets . Hence for some . Thus
Now , and again yields . Therefore for some .
(2) (1). Let and put . Since every is contained in ,
We show that . Let be a basic neighbourhood of , where is compact. Then
By (2), for some , so . Hence every basic neighbourhood of meets , and therefore
The generic point is unique because is .
(2) (3). Let be a -space, , and satisfy
Let be a subbasic compact-open neighbourhood of . By Lemma 5.4 and functoriality of ,
Moreover,
By (2), there is such that , equivalently . Thus every subbasic neighbourhood of meets . Since is irreducible, every finite intersection of such neighbourhoods also meets . Hence , proving property .
(3) (4) follows from Proposition 2.4, since the compact-open topology is finer than the pointwise convergence topology.
(4) (1). The Sierpiński space is sober, hence -sober. Therefore is -sober by (4). By the homeomorphism (2.1), is -sober.
The theorem immediately reorganizes several of the questions from [20].
Corollary 5.6 For a fixed -space the following hold.
(1) The following are equivalent:
(a) is sober;
(b) every compact subset of is -supercompact;
(c) for every -space , the compact-open topology on has property with respect to ;
(d) for every sober space , is sober.
(2) The following are equivalent:
(a) is well-filtered;
(b) every compact subset of is WD-supercompact;
(c) every compact subset of is RD-supercompact;
(d) for every -space , the compact-open topology on has property with respect to WD (equivalently, with respect to RD);
(e) for every well-filtered space , is well-filtered.
Consequently, the universal forms of Questions 49, 55, 58 and 60 in [20] are equivalent, and the universal forms of Questions 48, 56, 57 and 59 are equivalent (for either WD or RD).
Proof. Apply Theorem 5.5 with , and , and use the characterizations of sober and well-filtered spaces by -, - and -sobriety from [19, 21].
Corollary 5.7 For every -space , the following are equivalent:
every compact subset of is WD-supercompact;
every compact subset of is RD-supercompact.
Remark 5.8 Corollary 5.7 concerns the assertion that all compact subsets have the corresponding property. It does not assert that the WD and RD notions of supercompactness coincide for an individual fixed subset.
There is a close relation with the SI-compactness introduced by Zhao and Ho [24]. They call SI-compact if every open cover of which is irreducible in the Scott space contains itself. Since , every subset irreducible in the Scott topology is irreducible in . Hence R-supercompactness in the sense of Definition 5.1 is at least as strong as the corresponding Scott-irreducible covering condition. If is consonant, then , and the two irreducibility tests agree.
Wilker property and H-supercompactness
We now show that Wilker’s condition is a strong bridge from ordinary compactness to the whole -supercompactness hierarchy. The Wilker definition and its finite form were given in § 2; we therefore pass directly to the main implication.
Theorem 6.1 Let be a Wilker space and let be an irreducible subset system. Then every compact subset of is -supercompact.
Proof. Let be compact and let satisfy . Since is compact, there are such that
By Lemma 2.2, choose compact such that
For each , the basic open set of meets , because . Since , irreducibility gives
Choose in this intersection. Then for every , and therefore . Thus is -supercompact.
Combining Theorems 5.5 and 6.1 yields the main function-space consequence.
Corollary 6.2 Let be a Wilker space and let be an irreducible subset system. Then we have the following conclusions:
is -sober.
For every -space , the compact-open topology on has property with respect to .
For every -sober space , is -sober.
Corollary 6.3 Let be a Wilker space. Then we have the following conclusions:
is sober (hence well-filtered).
If is sober, then is sober.
If is well-filtered, then is well-filtered.
The compact-open topology on has property with respect to , WD and RD, for every -space .
In particular, Questions 48, 49, 55, 56, 59 and 60 of [20] have affirmative answers whenever the domain space is Wilker.
Wilker showed that condition (D) (i.e., the condition of Definition 2.1) holds for some important classes of spaces, including Hausdorff spaces and locally compact spaces [17]. Thus Corollary 6.2 applies to all such domain spaces.
We also obtain a short proof of a recent result on Scott-open set lattices.
Corollary 6.4 If is consonant and Wilker, then the Scott space is sober.
Proof. By Corollary 6.2, is sober. Consonance is equivalent to
see, for example, [6]. Hence is sober. □
Remark 6.5 Corollary 6.4 recovers the consonant-Wilker part of Z. He’s preprint When do weakly first-countable spaces and the Scott topology of open set lattice become sober? (2025, arXiv:2503.12766v2). The present proof separates the two mechanisms: Wilker’s condition supplies the irreducible compact-cover refinement, while consonance identifies the compact-generated topology with the Scott topology.
The Hoare power construction
Coherence, local compactness and non-preservation
Let denote the set of all nonempty closed subsets of equipped with the lower Vietoris topology, generated by
Schalk proved that the Hoare hyperspace (with the empty closed set included) is always sober [15]; the nonempty part is the open subspace , hence it is sober as well.
Proposition 7.1 (Schalk) For every -space , is sober.
The lower Vietoris semilattice also carries a useful compactness property that is independent of the base space.
Proposition 7.2 For every -space , the Hoare powerspace is coherent.
Proof. Put and consider the union map
For every ,
so is continuous. Let be compact saturated subsets of . Since the specialization order of is inclusion, both families are upward closed under inclusion. Therefore
Indeed, belongs to both families whenever and , while every equals . The intersection is thus a continuous image of a compact space, hence compact; it is saturated as an intersection of saturated sets. Thus is coherent.
Combining Proposition 7.2 with Proposition 4.8 yields a useful obstruction.
Corollary 7.3 For every -space ,
For later use we include a direct compact-neighborhood argument in the locally compact case. Schalk proved the corresponding lower-Vietoris local-compactness result [15 Proposition 6.11]; stronger formulations for core-compact bases are also standard [8 Proposition 2.9].
Lemma 7.4 If is locally compact, then is locally compact.
Proof. Let and let
be a basic neighborhood of . Choose . By local compactness, choose open and compact saturated such that
Then
It remains to show that is compact. Consider
For open ,
because an open set meets if and only if it contains at least one . Hence is continuous. Its image is compact. Since the specialization order of is inclusion, the saturation of this image consists exactly of the closed sets containing for some ; equivalently, it is . By Lemma 2.3, is compact.
We now relate the Hoare construction to core-compactness. Let be the sobrification of . Closed subsets of and correspond naturally: for closed , put
Lemma 7.5 The map is a homeomorphism
Proof. The standard sobrification induces an order isomorphism and hence, by complementation, a bijection between the closed subsets of and those of ; under this bijection is sent to . It remains only to compare the lower Vietoris topologies. If corresponds to
then
Indeed, the forward implication is immediate. Conversely, if , then is irreducible closed and meets . Thus lower Vietoris subbasic opens correspond exactly.
The implication from core-compactness of to local compactness of the full lower Vietoris hyperspace is contained in the classical theory recalled above. The following formulation for the nonempty Hoare powerspace also records the converse in a form useful below.
Theorem 7.6 For every -space , the following are equivalent:
(1) is core-compact;
(2) is locally compact;
(3) is core-compact.
Proof. (1) (2). The sobrification of a core-compact space is locally compact [5]. By Lemma 7.4, is locally compact, and by Lemma 7.5 it is homeomorphic to .
(2) (1). Let . The canonical map
is continuous and satisfies . By local compactness of , choose an open and a compact set such that
Put . We show in . Let be directed with . Then
Compactness and directedness yield with . Hence
Thus every has a neighborhood , so is core-compact.
(2) (3). Local compactness always implies core-compactness. Conversely, is sober by Proposition 7.1, and a sober core-compact space is locally compact [5].
Corollary 7.7 The Hoare powerspace construction preserves and reflects core-compactness.
Corollary 7.8 The Hoare powerspace construction preserves local compactness but does not reflect local compactness.
Proof. Preservation is Lemma 7.4. For non-reflection, take the Hofmann–Lawson space . It is not locally compact, but it is core-compact, so is locally compact by Theorem 7.6.
The reason for this asymmetry is transparent: carries sobriety automatically. The Hoare construction therefore turns the core-compactness of into core-compactness plus sobriety of , which is exactly local compactness. Combining Theorem 7.6 with Wilker’s classical theorem gives the following large positive region.
Corollary 7.9. If is core-compact, then is locally compact, sober and Wilker.
This places Question 4.2 in a new light: for every core-compact , the Hoare powerspace is already Wilker; the missing issue is whether the Wilker property can be recovered on the base space.
The preservation problem, however, has a strong negative answer. In their preprint A countable metric space whose Hoare power space is not co-sober (2026, arXiv:2610.05868v1), Xu and Ji constructed a countable zero-dimensional metrizable space such that is not co-sober. Since is metrizable, it is Hausdorff and hence Wilker, whereas Corollary 7.3 excludes Wilker-ness of its Hoare powerspace.
Corollary 7.10 (Failure of Hoare preservation) There exists a countable zero-dimensional metrizable Wilker space such that is not Wilker. Consequently, the Hoare powerspace construction does not preserve Wilker-ness, even within countable zero-dimensional metrizable spaces.
Proof. Take as above. If were Wilker, then it would be co-sober by Corollary 7.3, contrary to the metric-space preprint (arXiv:2610.05868v1).
The unrestricted reflection direction remains open. The results below give an operation-theoretic reformulation and settle it on consonant bases.
Question 7.11 (Hoare reflection) If is Wilker, must be Wilker?
A negative answer to Question 4.2 would automatically give a negative answer to Question 7.11: if were core-compact but non-Wilker, then would be locally compact and hence Wilker by Corollary 7.9.
Although preservation fails in general, it holds on important subclasses. Core-compact bases are covered by Corollary 7.9; moreover, lower powerspaces of quasi-Polish spaces are quasi-Polish, and quasi-Polish spaces are Wilker [2].
Union on the compact-open lattice
We first record an elementary operation-theoretic form of the Wilker condition. In the proof, empty compact saturated sets are allowed.
Proposition 7.12 A space is Wilker if and only if
is continuous, where the domain has the ordinary product topology.
Proof. Suppose is Wilker and for a compact saturated . Choose compact saturated and with . Then is a neighborhood of mapped into .
Conversely, let be compact saturated and suppose union is continuous. The sets form a base for , since finite intersections correspond to finite unions of compact sets. There are therefore compact saturated and such that
If , saturation provides open sets containing and omitting . Their union must contain , a contradiction. Thus , proving the Wilker condition by Lemma 2.3.
A Scott-lattice criterion for Hoare powerspaces
We next show that the corresponding operation on the Scott space characterizes Wilker-ness of the Hoare powerspace. One technical step is needed to pass from lower Vietoris subbasic covers to arbitrary open covers.
Lemma 7.13 (A coherent-space refinement lemma) Let be coherent, and let be an open subbasis closed under finite unions and containing . Suppose that whenever a compact saturated satisfies with , there are compact saturated sets , covering . Then is Wilker.
Proof. By Lemma 2.3, it suffices to split compact saturated sets. We may replace the compact pieces supplied by the hypothesis with their intersections with , using coherence.
Let with compact saturated and open. Choose a finite cover of by basic opens, each contained in one , and group them into two finite unions. Distributivity and finite-union closure of express these unions as
where . An empty group means that is contained in the other open set and is immediate; a whole-space group is equally immediate. We may therefore take .
For each , use to obtain compact saturated covering , with and . Put
Finite unions and coherence make both sets compact saturated, and , . If , then some satisfies for every , so for every and hence . Thus .
For the next correspondence we temporarily include the empty closed set. Let be the space of all closed subsets of with the lower Vietoris topology, and use for when working in . Then is its open subspace of nonempty closed sets.
Lemma 7.14. The space is Wilker if and only if is Wilker. Moreover, is coherent.
Proof. The forward implication follows from open heredity. An open set in containing the point is the whole space. If a two-set cover contains such an open set, take the entire compact set as one piece and the empty set as the other. Otherwise the open sets and the compact set being covered lie in , so Wilker-ness there proves the converse.
For coherence, the proof of Proposition 7.2 applies unchanged with the empty closed set included: union is continuous, and its image on the product of two compact saturated families is their intersection.
Let be a compact saturated subset of , possibly empty, and let . Define
We use the following established order correspondence, without any countable-basis or consonance hypothesis.
Proposition 7.15 (de Brecht–Kawai [2 Definition 6.7 and Theorem 6.10]) The maps in (7.1) are mutually inverse, inclusion-reversing bijections between the compact saturated subsets of and the Scott-open subsets of . In particular,
In the Smyth order on compact families these maps are order-preserving; “inclusion-reversing” refers to ordinary set inclusion on both sides. The empty cases are included: and . Only the order correspondence is used below, not the stronger topological assertion in [2 Theorem 6.11], which requires a countable base.
Theorem 7.16 (Hoare–Scott union criterion) For every -space , the following are equivalent:
(1) is Wilker;
(2) the union operation
is jointly continuous, where the domain is the ordinary product of the two Scott spaces.
Proof. Assume (1), and let . By Proposition 7.15, the compact saturated family satisfies
Use Lemma 7.14 and Wilker-ness to choose compact saturated families and covering . Put . These are Scott-open neighborhoods of , respectively, and (7.3) gives
For and , upperness implies . This proves joint continuity.
Conversely, assume (2). Suppose that a compact saturated family is covered by and . Then . Continuity gives Scott-open neighborhoods of such that whenever and . Taking shows
Apply (7.4) to obtain
These are compact saturated pieces. The diamonds, together with , form an open subbasis closed under finite unions; a cover using itself splits trivially. Lemma 7.14 gives coherence, and Lemma 7.13 upgrades this subbasic splitting property to Wilker-ness of . Another application of Lemma 7.14 proves (1).
Remark 7.17 Union always preserves directed suprema as a map from the order product . This does not imply the joint continuity in Theorem 7.16: its domain has topology , not the possibly finer topology . Thus the theorem is not a consequence of order-theoretic Scott continuity alone.
Consonant bases and the reflection question
Recall that consonance is the equality . The preceding two operation-theoretic characterizations therefore agree on consonant base spaces.
Corollary 7.18 (Hoare preservation and reflection on consonant bases) For every consonant -space ,
In particular, Question 7.11 has an affirmative answer for consonant , and every counterexample in the general case must be non-consonant.
Proof. Consonance identifies with . Apply Proposition 7.12 and Theorem 7.16.
For consonant spaces, the equivalence between Wilker-ness and joint Scott continuity of union was already obtained by de Brecht and is recorded in [7 Proposition J], where it is attributed to a conversation with Keimel. We use this established connection; the Hoare consequence here comes from combining it with Theorem 7.16.
Remark 7.19 (Compact transversals). There is also a direct proof of the reflection direction that isolates the use of consonance. Let be compact in , and put . If is Wilker, split into compact families . Saturate the families, which keeps them compact and inside . Consonance gives compact saturated such that
Every meets : otherwise belongs to , but cannot belong to . If , then , and saturation of implies . Thus . This is the compact transversal mechanism of [2 Lemma 6.12 and Theorem 6.13].
For arbitrary , Question 7.11 is now equivalent to whether the following implication always holds:
Both domains here have the ordinary product topology. The inclusion alone does not justify this transfer of joint continuity. We have proved it when is consonant, but do not claim it for arbitrary . In particular, the general Hoare-reflection problem and the core-compactness question 4.2 remain unresolved here.
The Smyth power construction
Let be equipped with the upper Vietoris topology, whose basic open sets are
The specialization order is reverse inclusion.
The local-compactness behavior of the Smyth construction is very different from that of the Hoare construction.
Theorem 8.1 (Lyu–Chen–Jia [12]) For every topological space , the following are equivalent:
(1) is locally compact;
(2) is locally compact;
(3) is core-compact.
Hence local compactness gives a first positive answer to the Wilker-preservation problem for Smyth powerspaces.
Corollary 8.2 If is locally compact, then is Wilker.
We now prove a reflection theorem. We use the following standard compact-union lemma.
Proof. The union is saturated because every member of is saturated. Let be an open cover of . For each , compactness of gives a finite such that . The sets form an open cover of . Choose finitely many whose corresponding boxes cover . The finitely many appearing in the cover every member of , hence cover .
Theorem 8.4 (Smyth reflection of Wilker-ness) If is Wilker, then is Wilker.
Proof. Let be compact and with open. Consider the canonical map
The set is compact saturated by Lemma 2.3, applied to the compact singleton . The map is continuous, since . Therefore is compact in , and
Since is Wilker, there are compact subsets such that
Set . By Lemma 8.3, is compact saturated, and . If , then belongs to one of the , hence . Thus .
So the difficult direction is preservation.
Question 8.5 (Smyth preservation) If is Wilker, must be Wilker?
Again there are large positive subclasses. Besides the locally compact case from Corollary 8.2, the Smyth powerspace of a quasi-Polish space is quasi-Polish, hence Wilker [2].
The Hofmann–Lawson space provides a particularly natural test case: it is core-compact, Wilker and not locally compact. By Theorem 8.1, its Smyth powerspace is not core-compact and not locally compact. Thus Question 8.5 is genuinely outside the automatic locally compact regime already for this classical example.
Two central open problems and further directions
Two central open problems on Wilker spaces
Theorem 3.16 settles the binary product problem negatively, even for countable compact hereditarily Wilker factors. Corollary 3.17 gives a square counterexample, and Corollary 3.19 settles categorical reflectivity negatively. We therefore single out two remaining structural problems.
(W1) Smyth preservation. If is Wilker, must be Wilker?
This is Question 8.5. The reverse implication is Theorem 8.4, and preservation holds for locally compact and quasi-Polish spaces. The Hofmann–Lawson space remains a natural test case: it is core-compact and Wilker but not locally compact, and its Smyth powerspace lies outside the automatic locally compact regime.
(W2) Core-compactness versus Wilker-ness. Is every core-compact -space Wilker?
This is Question 4.2. The complete-lattice Scott-space case is positive by Proposition 4.6. Corollary 4.4 shows that core-compactness together with Wilker-ness does not force local compactness, so an affirmative answer cannot be reduced to a general local-compactness assertion.
The product counterexamples do not answer either question. They give failure of a binary permanence property, whereas (W1) concerns a specific powerspace construction and (W2) concerns the existence of compact-cover refinements in all core-compact spaces.
The Hoare connection and related directions
Question 7.11 is now equivalent to the transfer of joint continuity in (11). It has a positive answer on consonant bases by Corollary 7.18, but its general form remains open. This question is closely tied to (W2): a negative answer to (W2) would give a core-compact non-Wilker space whose Hoare powerspace is locally compact and Wilker by Corollary 7.9. Conversely, a positive answer to general Hoare reflection would imply a positive answer to (W2). The new criterion does not by itself supply this transfer outside consonance.
The order-topological variants remain useful test problems rather than additional central questions. Question 4.7 asks for (W2) on Scott spaces of arbitrary posets or dcpos; Question 4.12 concerns countable frames and countable complete meet-continuous lattices. The unrestricted Boolean and upper-topology counterexamples do not decide those countable cases. The quotient behavior of Wilker spaces also remains separate from the product and reflection results proved here.
On the function-space side, it remains natural to ask whether sobriety of , equivalently -supercompactness of every compact subset of , forces Wilker-ness. A negative answer would isolate a weakening of compact-cover splitting. Other directions concern strictness and possible collapses in the supercompactness hierarchy of Remark 5.2, and detection of Wilker-type properties by property or -sobriety of for a fixed finite , such as or a small Alexanderoff space. These connect naturally with consonance, co-consonance and Hofmann–Mislove correspondences; (W1) and (W2) remain the two principal open problems emphasized here.
Conclusions and outlook
Wilker’s compact-cover condition admits descriptions in terms of both topological structure and continuity of union. The KC characterization supplies a sufficient separation hypothesis for the mixed product theorem. The co-finite modifications, by contrast, give hereditarily Wilker factors with a non-Wilker product and square. Thus the class is stable under retracts and sums, but not under products, and it does not form a reflective full subcategory of .
The compact-open results use -supercompactness to describe the role of the Wilker property. A compact refinement on the domain yields -sobriety of , property for compact-open function spaces, and preservation of -sobriety by . The order-topological examples delimit this mechanism: quasicontinuous Scott spaces are Wilker, whereas complete Boolean algebras need not have Wilker Scott spaces and complete lattices need not have Wilker upper topologies.
The Hoare–Scott union criterion supplies another link between the topology of a space and the topology of its open-set lattice. Wilker-ness of is continuity of union on , whereas Wilker-ness of is continuity of union on . Consonance identifies these topologies and gives preservation and reflection on that class of bases. Outside consonance, preservation fails and the general reflection implication is not settled here. Smyth powerspaces always reflect Wilker-ness, but their preservation problem remains open.
The resulting picture leaves (W1), Smyth preservation, and (W2), the core-compactness implication, as the two main remaining questions. The counterexamples remove productivity from that list, while the Hoare criterion gives a precise additional route toward (W2) without claiming an answer to it.
Statements and Declarations
Funding. This research was supported by the National Natural Science Foundation of China (Nos. 12471070 and 12071199).
Competing interests. The author has no relevant financial or non-financial interests to disclose.
Author contributions. Xiaoquan Xu is the sole author and is responsible for the research and the preparation of the manuscript.
Data availability. Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.
Use of AI-assisted tools. During the preparation of this manuscript, the author used ChatGPT (OpenAI) to assist with language editing, organization of the exposition, bibliographic checks, and the development and checking of mathematical arguments. The author is responsible for verifying all mathematical statements, proofs and references and takes full responsibility for the final text.
References
- [1]de Brecht, M., Goubault-Larrecq, J., Jia, X., Lyu, Z.: Domain-complete and LCS-complete spaces. Electron. Notes Theor. Comput. Sci. 345, 3–35 (2019). https://doi.org/10.1016/j.entcs.2019.07.014
- [2]de Brecht, M., Kawai, T.: On the commutativity of the powerspace constructions. Log. Methods Comput. Sci. 15(3), 13:1–13:25 (2019). https://doi.org/10.23638/LMCS-15(3:13)2019
- [3]Choquet, G.: Theory of capacities. Ann. Inst. Fourier (Grenoble) 5, 131–295 (1953–1954). https://doi.org/10.5802/aif.53
- [4]van Douwen, E.K.: An anti-Hausdorff Fréchet space in which convergent sequences have unique limits. Topology Appl. 51, 147–158 (1993). https://doi.org/10.1016/0166-8641(93)90147-6
- [5]Gierz, G., Hofmann, K.H., Keimel, K., Lawson, J.D., Mislove, M., Scott, D.S.: Continuous Lattices and Domains. Encyclopedia of Mathematics and its Applications, vol. 93. Cambridge University Press, Cambridge (2003). https://doi.org/10.1017/CBO9780511542725
- [6]Goubault-Larrecq, J.: Non-Hausdorff Topology and Domain Theory: Selected Topics in Point-Set Topology. New Mathematical Monographs, vol. 22. Cambridge University Press, Cambridge (2013). https://doi.org/10.1017/CBO9781139524438
- [7]Goubault-Larrecq, J.: Choquet–Wilker spaces. Electronic supplement to Non-Hausdorff Topology and Domain Theory, 20 April 2026, Proposition J. https: //topology.lmf.cnrs.fr/choquet-wilker-spaces/. Accessed 14 October 2026
- [8]Hofmann, D., Nora, P.: Dualities for modal algebras from the point of view of triples. Algebra Universalis 73, 297–320 (2015). https://doi.org/10.1007/ s00012-015-0324-5
- [9]Hofmann, K.H., Lawson, J.D.: The spectral theory of distributive continuous lattices. Trans. Amer. Math. Soc. 246, 285–310 (1978). https://doi.org/10.1090/ S0002-9947-1978-0515540-7
- [10]Keimel, K., Lawson, J.D.: Measure extension theorems for T₀-spaces. Topology Appl. 149, 57–83 (2005). https://doi.org/10.1016/j.topol.2004.02.019
- [11]Kovár, M.M.: The Hofmann–Mislove theorem for general topological structures. In: Spatial Representation: Discrete vs. Continuous Computational Models. Dagstuhl Seminar Proceedings 04351, pp. 1–9. Schloss Dagstuhl (2005). https://doi.org/10. 4230/DagSemProc.04351.21
- [12]Lyu, Z., Chen, Y., Jia, X.: Core-compactness, consonance and the Smyth power-spaces. Topology Appl. 312, 108066 (2022). https://doi.org/10.1016/j.topol. 2022.108066
- [13]Mao, X., Xu, L.: Quasicontinuity of posets via Scott topology and sobrification. Order 23(4), 359–369 (2006). https://doi.org/10.1007/s11083-007-9054-4
- [14]Papadopoulos, B.K.: A topological lattice on the set of multifunctions. Int. J. Math. Math. Sci. 12, 665–668 (1989). https://doi.org/10.1155/S0161171289000815
- [15]Schalk, A.: Algebras for Generalized Power Constructions. Ph.D. thesis, Technische Hochschule Darmstadt (1993)
- [16]Wen, X., Xu, X.: Sober is not always co-sober. Topology Appl. 250, 48–52 (2018). https://doi.org/10.1016/j.topol.2018.10.003
- [17]Wilker, P.: Adjoint product and hom functors in general topology. Pacific J. Math. 34, 269–283 (1970). https://doi.org/10.2140/pjm.1970.34.269
- [18]Xi, X., Lawson, J.: On well-filtered spaces and ordered sets. Topology Appl. 228, 139–144 (2017). https://doi.org/10.1016/j.topol.2017.06.002
- [19]Xu, X.: On H-sober spaces and H-sobrifications of T₀ spaces. Topology Appl. 289, 107548 (2021). https://doi.org/10.1016/j.topol.2020.107548
- [20]Xu, X., Bao, M., Zhang, X.: On function spaces equipped with Isbell topology and Scott topology. Math. Structures Comput. Sci. 32, 1099–1116 (2022). https://doi.org/10.1017/S0960129523000014
- [21]Xu, X., Shen, C., Xi, X., Zhao, D.: On T₀ spaces determined by well-filtered spaces. Topology Appl. 282, 107323 (2020). https://doi.org/10.1016/j.topol.2020.107323
- [22]Xu, X., Yang, Y., Chen, L.: Strong well-filteredness of upper topology on sup-complete posets. Math. Structures Comput. Sci. 36, e19 (2026). https://doi.org/10.1017/S0960129526100589
- [23]Zhao, B., He, Z., Wang, K.: On co-sober spaces. Houston J. Math. 49(4), 971–988 (2023)DOI
- [24]Zhao, D., Ho, W.: On topologies defined by irreducible sets. J. Log. Algebra Methods Program. 84, 185–195 (2015). https://doi.org/10.1016/j.jlamp.2014.10.003