Priestley Duality

Origin. Hilary A. Priestley, "Representation of distributive lattices by means of ordered Stone spaces" (Bulletin of the London Mathematical Society 2, 1970), with the full categorical duality in "Ordered topological spaces and the representation of distributive lattices" (Proceedings of the London Mathematical Society 24, 1972). Marshall Stone had already represented bounded distributive lattices on their prime spectra in "Topological representations of distributive lattices and Brouwerian logics" (Časopis pro pěstování matematiky a fysiky 67, 1938), but his spaces are not Hausdorff; Priestley kept the same points — the prime filters — and added the inclusion order to a compact Hausdorff topology. William Cornish, "On H. Priestley's dual of the category of bounded distributive lattices" (Matematički Vesnik 12, 1975), proved the two presentations interchangeable. Davey and Priestley's Introduction to Lattices and Order (Cambridge, 1990; second edition 2002) is the standard textbook treatment.

Models. A bounded distributive lattice and a Priestley space are two presentations of one object: the points of the space are the prime filters of the lattice, the order is inclusion, and the lattice reappears as the clopen upsets. Dropping complementation from a Boolean algebra is what makes the order visible — ultrafilters are pairwise incomparable, so in the Boolean case the order is equality and the space is a Stone space, while prime filters of a merely distributive lattice nest, and that nesting is exactly the structure the dual must carry. This is the base layer for the dualities of non-classical logic: an Esakia space is a Priestley space satisfying one further condition, so the duals of Heyting algebras, De Morgan algebras, and distributive lattices with operators generally are Priestley spaces with added structure, as Stone spaces are Priestley spaces with trivial order.

Formalism.

The representation: For a bounded distributive lattice L, let X_L = {prime filters of L}, ordered by ⊆. Topologize by the subbasis {â : a ∈ L} ∪ {X_L ∖ â : a ∈ L}, where â = {F ∈ X_L : a ∈ F}. Then a ↦ â is an isomorphism of L onto the clopen upsets of X_L.

Priestley space: A compact ordered topological space (X, τ, ≤) that is totally order-disconnected: whenever x ≰ y, some clopen upset contains x and omits y. Compactness comes from the prime ideal theorem; total order-disconnectedness is the Hausdorff separation plus the order.

The duality: DLat ≃ Pries^op. Homomorphisms L → M correspond contravariantly to continuous order-preserving maps X_M → X_L. Congruences ↔ closed subsets; quotient lattices ↔ closed subspaces; sublattices ↔ Priestley quotient spaces.

The Boolean case: L is Boolean iff the order of X_L is equality. Priestley spaces with discrete order are the Stone spaces, and the duality restricts to Stone's.

Spectral spaces: Cornish (1975): Pries and Spec are isomorphic categories. The open upsets of a Priestley space form a spectral topology; the Priestley topology is its patch topology, and the specialization order of the spectral topology returns ≤. Hochster, "Prime ideal structure in commutative rings" (1969): the spectral spaces are exactly the Zariski spectra of commutative rings. So Stone's 1938 duality and Priestley's are one duality carried in two topologies.

Added operations: Heyting implication ↔ ↓U clopen for every clopen U — the Esakia spaces (Esakia, "Topological Kripke models", 1974). Operators in the Jónsson–Tarski sense ↔ additional relations on the space, with canonicity: Gehrke and Jónsson, "Bounded distributive lattices with operators" (Mathematica Japonica 40, 1994).

The finite case: Finite distributive lattices ↔ finite posets (Garrett Birkhoff, 1937). Priestley duality is this correspondence with the topology doing the work of finiteness.

Symbols.

Metatheory. Priestley (1972) proved the full dual equivalence, and Cornish (1975) proved that the category of Priestley spaces is isomorphic to Hochster's category of spectral spaces, so Priestley duality and Stone's 1938 spectral duality carry the same information and differ in which of two interdefinable topologies is written down. Hochster (1969) identified the spectral spaces as the prime spectra of commutative rings, so the duals of bounded distributive lattices are the same spaces algebraic geometry works over. The existence of enough prime filters is the prime ideal theorem for distributive lattices, equivalent over ZF to BPI, so the duality sits at the same point of the choice hierarchy as Stone's and fails constructively for the same reason. Congruences correspond to closed subsets, so the congruence lattice of L is isomorphic to the frame of open subsets of X_L, and homomorphisms, sublattices, and quotients all have named duals. Gehrke and Jónsson (1994) extended the duality and its canonicity theory to distributive lattices with operators, the distributive counterpart of Jónsson–Tarski, which is the form in which the duality reaches logics whose algebras carry more than the lattice operations.

Applies to. Esakia duality and through it the intermediate logics, whose lattice is dual to the subvarieties of Heyting algebras. Priestley-style dualities for the distributive-lattice-based algebras of non-classical logic — De Morgan algebras, distributive residuated lattices, and the other algebras of substructural and many-valued systems — via distributive lattices with operators. Spectral spaces in commutative algebra, through Cornish and Hochster. Canonical extensions of distributive lattices. Natural duality theory, where the two-element lattice generating the duality is the founding example.

Limitations. Distributivity is required, and not as a convenience: the points of the dual are prime filters, and a non-distributive lattice can lack them entirely — the five-element modular lattice M₃ has no prime filters at all. Non-distributive lattices need different machinery: Urquhart, "A topological representation theory for lattices" (Algebra Universalis 8, 1978), replaces prime filters by maximal filter–ideal pairs on a doubly ordered space, reducing to Priestley duality in the distributive case but covering only surjective homomorphisms in the Urquhart–Hartung line. Hartonas and Dunn, "Stone duality for lattices" (Algebra Universalis 37, 1997), use polarities between two spaces instead, and the non-distributive theory remains a family of competing dualities. The duality is a theorem about algebras, so a logic reaches it only through its algebraic semantics: intermediate logics inherit everything through Esakia duality, while a substructural logic without distribution — full Lambek and its relatives — is outside the duality's scope. Each operation beyond the lattice ones must be dualized separately, with its own condition on the space, as Heyting implication needs the Esakia condition. And the representation requires BPI, so in a constructive metatheory the spaces have too few points and locale-theoretic substitutes must be used.

© 2026 Lingenic LLC