SLIDE 1
Choice-free Stone duality Wesley H. Holliday University of - - PowerPoint PPT Presentation
Choice-free Stone duality Wesley H. Holliday University of - - PowerPoint PPT Presentation
Choice-free Stone duality Wesley H. Holliday University of California, Berkeley Joint work with Nick Bezhanishvili University of Amsterdam BLAST 2018 August 7, 2018 What does choice-free mean? We give a choice-free topological duality
SLIDE 2
SLIDE 3
What does “choice-free” mean?
We give a choice-free topological duality for Boolean algebras. We work in what Schechter (Handbook of Analysis and Its Foundations) calls quasiconstructive mathematics: “mathematics that permits conventional rules of reasoning plus ZF + DC, but no stronger forms of Choice” (p. 404).
SLIDE 4
What does “choice-free” mean?
We give a choice-free topological duality for Boolean algebras. We work in what Schechter (Handbook of Analysis and Its Foundations) calls quasiconstructive mathematics: “mathematics that permits conventional rules of reasoning plus ZF + DC, but no stronger forms of Choice” (p. 404). Note: only our applications, not the duality itself, uses DC.
SLIDE 5
What does “choice-free” mean?
We give a choice-free topological duality for Boolean algebras. We work in what Schechter (Handbook of Analysis and Its Foundations) calls quasiconstructive mathematics: “mathematics that permits conventional rules of reasoning plus ZF + DC, but no stronger forms of Choice” (p. 404). Note: only our applications, not the duality itself, uses DC. Note: of course we won’t prove that every BA is isomorphic to a field of sets, since this implies the Boolean Prime Filter Theorem.
SLIDE 6
Slogans
Two slogans describing our duality:
SLIDE 7
Slogans
Two slogans describing our duality:
1
“a mix of Stone and Tarski, connected by Vietoris”;
SLIDE 8
Slogans
Two slogans describing our duality:
1
“a mix of Stone and Tarski, connected by Vietoris”;
2
“a hyperspace approach, in contrast to a pointfree approach.”
SLIDE 9
Stone Representation of BAs
Theorem (Stone 1936). Every Boolean algebra is isomorphic to the BA of clopen sets of some topological (Stone) space. Marshall Stone (1903 – 1989)
SLIDE 10
Stone Representation of BAs
Stone representation uses the Prime Filter Theorem.
SLIDE 11
Stone Representation of BAs
Stone representation uses the Prime Filter Theorem. Let A be a Boolean algebra.
SLIDE 12
Stone Representation of BAs
Stone representation uses the Prime Filter Theorem. Let A be a Boolean algebra. Let XA be the space of all prime filters.
SLIDE 13
Stone Representation of BAs
Stone representation uses the Prime Filter Theorem. Let A be a Boolean algebra. Let XA be the space of all prime filters. The topology is generated by sets a = {x ∈ XA | a ∈ x} for a ∈ A.
SLIDE 14
Stone Representation of BAs
Stone representation uses the Prime Filter Theorem. Let A be a Boolean algebra. Let XA be the space of all prime filters. The topology is generated by sets a = {x ∈ XA | a ∈ x} for a ∈ A. Then XA is a compact Hausdorff space with a clopen basis.
SLIDE 15
Stone Representation of BAs
Stone representation uses the Prime Filter Theorem. Let A be a Boolean algebra. Let XA be the space of all prime filters. The topology is generated by sets a = {x ∈ XA | a ∈ x} for a ∈ A. Then XA is a compact Hausdorff space with a clopen basis. A is isomorphic to the Boolean algebra Clop(XA) of clopen sets.
SLIDE 16
Stone Representation of BAs
Stone representation uses the Prime Filter Theorem. Let A be a Boolean algebra. Let XA be the space of all prime filters. The topology is generated by sets a = {x ∈ XA | a ∈ x} for a ∈ A. Then XA is a compact Hausdorff space with a clopen basis. A is isomorphic to the Boolean algebra Clop(XA) of clopen sets. The isomorphism ϕ : A → Clop(XA) is given by ϕ(a) = a.
SLIDE 17
Prime Filter Theorem
The Prime Filter Theorem is used for showing that ϕ is injective.
SLIDE 18
Prime Filter Theorem
The Prime Filter Theorem is used for showing that ϕ is injective. Suppose a b. Then ↑a ∩ ↓b = ∅.
SLIDE 19
Prime Filter Theorem
The Prime Filter Theorem is used for showing that ϕ is injective. Suppose a b. Then ↑a ∩ ↓b = ∅. By the Prime Filter Theorem, there is a prime filter F such that ↑a ⊆ F and F ∩ ↓b = ∅.
SLIDE 20
Prime Filter Theorem
The Prime Filter Theorem is used for showing that ϕ is injective. Suppose a b. Then ↑a ∩ ↓b = ∅. By the Prime Filter Theorem, there is a prime filter F such that ↑a ⊆ F and F ∩ ↓b = ∅. So F ∈ ϕ(a) and F / ∈ ϕ(b), yielding ϕ(a) ⊆ ϕ(b).
SLIDE 21
Prime Filter Theorem
The Prime Filter Theorem is used for showing that ϕ is injective. Suppose a b. Then ↑a ∩ ↓b = ∅. By the Prime Filter Theorem, there is a prime filter F such that ↑a ⊆ F and F ∩ ↓b = ∅. So F ∈ ϕ(a) and F / ∈ ϕ(b), yielding ϕ(a) ⊆ ϕ(b). Our aim is to obtain Stone-like representation of Boolean algebras choice free.
SLIDE 22
Prime Filter Theorem
The Prime Filter Theorem is used for showing that ϕ is injective. Suppose a b. Then ↑a ∩ ↓b = ∅. By the Prime Filter Theorem, there is a prime filter F such that ↑a ⊆ F and F ∩ ↓b = ∅. So F ∈ ϕ(a) and F / ∈ ϕ(b), yielding ϕ(a) ⊆ ϕ(b). Our aim is to obtain Stone-like representation of Boolean algebras choice free. This will resemble Stone’s representation of distributive lattices.
SLIDE 23
Stone Representation of DLs
Theorem (Stone 1937). Every distributive lattice is isomorphic to the distributive lattice of compact open sets of some topological (spectral) space. Marshall Stone (1903 – 1989)
SLIDE 24
Boolean algebra of regular open sets
Theorem (Tarski 1937). For every topological space X, the set RO(X) of regular open subsets of X forms a (complete) BA. Alfred Tarski (1901 – 1983)
SLIDE 25
Regular open sets
A set U is regular open if Int(Cl(U)) = U.
SLIDE 26
Regular open sets
A set U is regular open if Int(Cl(U)) = U. For U, V ∈ RO(X) we put U ∧ V = U ∩ V, U ∨ V = Int(Cl(U ∪ V)), ¬U = Int(X \ U).
SLIDE 27
Choice-free Stone duality
Let A be a Boolean algebra.
SLIDE 28
Choice-free Stone duality
Let A be a Boolean algebra. Let XA be the space of all proper filters of A.
SLIDE 29
Choice-free Stone duality
Let A be a Boolean algebra. Let XA be the space of all proper filters of A. We generate a topology by sets a = {x ∈ XA | a ∈ x} for a ∈ A.
SLIDE 30
Choice-free Stone duality
Let A be a Boolean algebra. Let XA be the space of all proper filters of A. We generate a topology by sets a = {x ∈ XA | a ∈ x} for a ∈ A. Then XA is a spectral space, i.e., compact, T0, coherent (compact
- pen sets are closed under intersection and form a basis), sober.
SLIDE 31
Choice-free Stone duality
Let A be a Boolean algebra. Let XA be the space of all proper filters of A. We generate a topology by sets a = {x ∈ XA | a ∈ x} for a ∈ A. Then XA is a spectral space, i.e., compact, T0, coherent (compact
- pen sets are closed under intersection and form a basis), sober.
Good formulation of sober for us: every completely prime filter in Ω(X) is Ω(x) = {U ∈ Ω(X) | x ∈ U} for some x ∈ X.
SLIDE 32
Choice-free Stone duality
Let A be a Boolean algebra. Let XA be the space of all proper filters of A. We generate a topology by sets a = {x ∈ XA | a ∈ x} for a ∈ A. Then XA is a spectral space, i.e., compact, T0, coherent (compact
- pen sets are closed under intersection and form a basis), sober.
SLIDE 33
Choice-free Stone duality
Let A be a Boolean algebra. Let XA be the space of all proper filters of A. We generate a topology by sets a = {x ∈ XA | a ∈ x} for a ∈ A. Then XA is a spectral space, i.e., compact, T0, coherent (compact
- pen sets are closed under intersection and form a basis), sober.
The specialization order is the inclusion ⊆ on proper filters.
SLIDE 34
Choice-free Stone duality
Let A be a Boolean algebra. Let XA be the space of all proper filters of A. We generate a topology by sets a = {x ∈ XA | a ∈ x} for a ∈ A. Then XA is a spectral space, i.e., compact, T0, coherent (compact
- pen sets are closed under intersection and form a basis), sober.
The specialization order is the inclusion ⊆ on proper filters. Second topology to consider: Up(XA), the -upset topology.
SLIDE 35
Choice-free Stone duality
Let A be a Boolean algebra. Let XA be the space of all proper filters of A. We generate a topology by sets a = {x ∈ XA | a ∈ x} for a ∈ A. Then XA is a spectral space, i.e., compact, T0, coherent (compact
- pen sets are closed under intersection and form a basis), sober.
The specialization order is the inclusion ⊆ on proper filters. Second topology to consider: Up(XA), the -upset topology. Let Int be the interior operation associated with Up(XA): Int(U) = {x ∈ X | ∀x′ x x′ ∈ U}.
SLIDE 36
Choice-free Stone duality
Let A be a Boolean algebra. Let XA be the space of all proper filters of A. We generate a topology by sets a = {x ∈ XA | a ∈ x} for a ∈ A. Then XA is a spectral space, i.e., compact, T0, coherent (compact
- pen sets are closed under intersection and form a basis), sober.
The specialization order is the inclusion ⊆ on proper filters. Second topology to consider: Up(XA), the -upset topology. Let Int be the interior operation associated with Up(XA): Int(U) = {x ∈ X | ∀x′ x x′ ∈ U}. U ⊆ X is -regular open if it is regular open in Up(XA)
SLIDE 37
Choice-free Stone duality
Let A be a Boolean algebra. Let XA be the space of all proper filters of A. We generate a topology by sets a = {x ∈ XA | a ∈ x} for a ∈ A. Then XA is a spectral space, i.e., compact, T0, coherent (compact
- pen sets are closed under intersection and form a basis), sober.
The specialization order is the inclusion ⊆ on proper filters. Second topology to consider: Up(XA), the -upset topology. Let Int be the interior operation associated with Up(XA): Int(U) = {x ∈ X | ∀x′ x x′ ∈ U}. U ⊆ X is -regular open if it is regular open in Up(XA): an upset s.th. if y ∈ U, then ∃z y: z ∈ Int(XA \ U).
SLIDE 38
Choice-free Stone duality
Let A be a Boolean algebra. Let XA be the space of all proper filters of A. We generate a topology by sets a = {x ∈ XA | a ∈ x} for a ∈ A. Then XA is a spectral space, i.e., compact, T0, coherent (compact
- pen sets are closed under intersection and form a basis), sober.
The specialization order is the inclusion ⊆ on proper filters. Second topology to consider: Up(XA), the -upset topology. Let Int be the interior operation associated with Up(XA): Int(U) = {x ∈ X | ∀x′ x x′ ∈ U}. U ⊆ X is -regular open if it is regular open in Up(XA): an upset s.th. if y ∈ U, then ∃z y: z ∈ Int(XA \ U). (XA, ) is a separative poset, i.e., every principal upset is regular open
SLIDE 39
Choice-free Stone duality
Let A be a Boolean algebra. Let XA be the space of all proper filters of A. We generate a topology by sets a = {x ∈ XA | a ∈ x} for a ∈ A. Then XA is a spectral space, i.e., compact, T0, coherent (compact
- pen sets are closed under intersection and form a basis), sober.
The specialization order is the inclusion ⊆ on proper filters. Second topology to consider: Up(XA), the -upset topology. Let Int be the interior operation associated with Up(XA): Int(U) = {x ∈ X | ∀x′ x x′ ∈ U}. U ⊆ X is -regular open if it is regular open in Up(XA): an upset s.th. if y ∈ U, then ∃z y: z ∈ Int(XA \ U). (XA, ) is a separative poset, i.e., every principal upset is regular open: if y x, then ∃z y: z ∈ Int(XA \ ↑x).
SLIDE 40
Choice-free Stone duality
CO(X) = {compact open subsets of X}.
SLIDE 41
Choice-free Stone duality
CO(X) = {compact open subsets of X}. Let CORO(XA) be the set of compact open -regular open sets.
SLIDE 42
Choice-free Stone duality
CO(X) = {compact open subsets of X}. Let CORO(XA) be the set of compact open -regular open sets. If U ∈ CORO(XA), then Int(X \ U) ∈ CORO(XA).
SLIDE 43
Choice-free Stone duality
CO(X) = {compact open subsets of X}. Let CORO(XA) be the set of compact open -regular open sets. If U ∈ CORO(XA), then Int(X \ U) ∈ CORO(XA). Then CORO(XA) is a Boolean algebra, where U ∧ V = U ∩ V, ¬U = Int(X \ U).
SLIDE 44
Choice-free Stone duality
Theorem (Choice-free representation of BAs). Each Boolean algebra A is isomorphic to the Boolean algebra CORO(XA).
SLIDE 45
Choice-free Stone duality
Theorem (Choice-free representation of BAs). Each Boolean algebra A is isomorphic to the Boolean algebra CORO(XA). The isomorphism ϕ : A → CORO(XA) is given by ϕ(a) = a.
SLIDE 46
Choice-free Stone duality
Theorem (Choice-free representation of BAs). Each Boolean algebra A is isomorphic to the Boolean algebra CORO(XA). The isomorphism ϕ : A → CORO(XA) is given by ϕ(a) = a. To show that ϕ is injective we do not need the PFT.
SLIDE 47
Choice-free Stone duality
Theorem (Choice-free representation of BAs). Each Boolean algebra A is isomorphic to the Boolean algebra CORO(XA). The isomorphism ϕ : A → CORO(XA) is given by ϕ(a) = a. To show that ϕ is injective we do not need the PFT. What kind of space is XA?
SLIDE 48
UV-spaces
A UV-space is a T0 space X such that:
SLIDE 49
UV-spaces
A UV-space is a T0 space X such that:
1
CORO(X) is closed under ∩ and Int(X \ ·);
SLIDE 50
UV-spaces
A UV-space is a T0 space X such that:
1
CORO(X) is closed under ∩ and Int(X \ ·);
2
x y ⇒ there is a U ∈ CORO(X) s.t. x ∈ U and y / ∈ U;
SLIDE 51
UV-spaces
A UV-space is a T0 space X such that:
1
CORO(X) is closed under ∩ and Int(X \ ·);
2
x y ⇒ there is a U ∈ CORO(X) s.t. x ∈ U and y / ∈ U;
3
every proper filter in CORO(X) is CORO(x) for some x ∈ X.
SLIDE 52
UV-spaces
A UV-space is a T0 space X such that:
1
CORO(X) is closed under ∩ and Int(X \ ·);
2
x y ⇒ there is a U ∈ CORO(X) s.t. x ∈ U and y / ∈ U;
3
every proper filter in CORO(X) is CORO(x) for some x ∈ X.
- Proposition. Every UV-space is a spectral space.
SLIDE 53
Choice-free representation of BAs
Theorem (Choice-free representation of BAs). For each BA A there is a UV-space X such that A is isomorphic to CORO(X).
SLIDE 54
Choice-free representation of BAs
Theorem (Choice-free representation of BAs). For each BA A there is a UV-space X such that A is isomorphic to CORO(X). We extend this correspondence to a full duality of the corresponding categories.
SLIDE 55
Vietoris space of a Stone space
Theorem (Vietoris 1922, Stone version). For every Stone space X its Vietoris space, i.e., the space of nonempty closed sets equipped with the hit-and-miss topology, is again a Stone space. Leopold Vietoris (1891 – 2002)
SLIDE 56
Vietoris space of a Stone space
Let X be a Stone space.
SLIDE 57
Vietoris space of a Stone space
Let X be a Stone space. Let F(X) be the set of all nonempty closed subsets of X.
SLIDE 58
Vietoris space of a Stone space
Let X be a Stone space. Let F(X) be the set of all nonempty closed subsets of X. The upper Vietoris topology has the basis U = {F ∈ F(X) | F ⊆ U}, U ∈ Ω(X). The lower Vietoris topology has the subbasis ♦V = {F ∈ F(X) | F ∩ V = ∅}, V ∈ Ω(X).
SLIDE 59
Vietoris space of a Stone space
Let X be a Stone space. Let F(X) be the set of all nonempty closed subsets of X. The upper Vietoris topology has the basis U = {F ∈ F(X) | F ⊆ U}, U ∈ Ω(X). The lower Vietoris topology has the subbasis ♦V = {F ∈ F(X) | F ∩ V = ∅}, V ∈ Ω(X). The Vietoris topology is the join of the upper and lower Vietoris topologies.
SLIDE 60
Examples of UV-spaces
Let X be a Stone space and UV(X) its upper Vietoris space.
SLIDE 61
Examples of UV-spaces
Let X be a Stone space and UV(X) its upper Vietoris space. Then UV(X) is a UV-space.
SLIDE 62
Examples of UV-spaces
Let X be a Stone space and UV(X) its upper Vietoris space. Then UV(X) is a UV-space. Assuming the PFT, every UV-space is homeomorphic to UV(X) for a Stone space X.
SLIDE 63
Examples of UV-spaces
Let X be a Stone space and UV(X) its upper Vietoris space. Then UV(X) is a UV-space. Assuming the PFT, every UV-space is homeomorphic to UV(X) for a Stone space X. We can prove a localic version of this result choice free.
SLIDE 64
UV-maps
Let X and X′ be spectral spaces.
SLIDE 65
UV-maps
Let X and X′ be spectral spaces. A map f : X → Y is called spectral if f −1[U] is compact open for each compact open U.
SLIDE 66
UV-maps
Let X and X′ be spectral spaces. A map f : X → Y is called spectral if f −1[U] is compact open for each compact open U. A UV-map between UV-spaces X and X′ is a spectral map f : X → X′
SLIDE 67
UV-maps
Let X and X′ be spectral spaces. A map f : X → Y is called spectral if f −1[U] is compact open for each compact open U. A UV-map between UV-spaces X and X′ is a spectral map f : X → X′ that is also a p-morphism: if f(x) ′ y′, then ∃y : x y and f(y) = y′. x f(x) y′ x ∃y f(x) y′ ⇒
SLIDE 68
Duality
Theorem (Choice-free UV-duality for BAs). The category of UV-spaces with UV-maps is dually equivalent to the category of Boolean algebras with Boolean homomorphisms.
SLIDE 69
Comparison
Theorem (Choice-free UV-duality for BAs). The category of UV-spaces with UV-maps is dually equivalent to the category of Boolean algebras with Boolean homomorphisms. Theorem (Stone duality). Assuming the PFT, the category of Stone spaces with continuous maps is dually equivalent to the category of Boolean algebras with Boolean homomorphisms.
SLIDE 70
Comparison
Theorem (Choice-free UV-duality for BAs). The category of UV-spaces with UV-maps is dually equivalent to the category of Boolean algebras with Boolean homomorphisms. Theorem (Stone duality). Assuming the PFT, the category of Stone spaces with continuous maps is dually equivalent to the category of Boolean algebras with Boolean homomorphisms. Let’s further compare these with a localic Stone duality.
SLIDE 71
Stone locales
A locale is a complete lattice L satisfying the join-infinite distributive law for each a ∈ L and S ⊆ L: a ∧
- S =
- {a ∧ s | s ∈ S}.
For any space X, Ω(X) ordered by ⊆ is a locale.
SLIDE 72
Stone locales
A locale is a complete lattice L satisfying the join-infinite distributive law for each a ∈ L and S ⊆ L: a ∧
- S =
- {a ∧ s | s ∈ S}.
For any space X, Ω(X) ordered by ⊆ is a locale. A Stone locale is a locale that is: compact – S = 1 implies T = 1 for a finite T ⊆ S. zero-dimensional – every element is a join of complemented elements.
SLIDE 73
Stone locales
A locale is a complete lattice L satisfying the join-infinite distributive law for each a ∈ L and S ⊆ L: a ∧
- S =
- {a ∧ s | s ∈ S}.
For any space X, Ω(X) ordered by ⊆ is a locale. A Stone locale is a locale that is: compact – S = 1 implies T = 1 for a finite T ⊆ S. zero-dimensional – every element is a join of complemented elements. Theorem (Choice-free localic Stone duality). The category of Stone locales with localic maps is dually equivalent to the category of Boolean algebras with Boolean homomorphisms.
SLIDE 74
Comparison
Theorem (Choice-free UV-duality for BAs). The category of UV-spaces with UV-maps is dually equivalent to the category of Boolean algebras with Boolean homomorphisms. Theorem (Stone duality). Assuming the PFT, the category of Stone spaces with continuous maps is dually equivalent to the category of Boolean algebras with Boolean homomorphisms. Theorem (Choice-free localic Stone duality). The category of Stone locales with localic maps is dually equivalent to the category of Boolean algebras with Boolean homomorphisms.
SLIDE 75
Comparison
Theorem (Choice-free UV-duality for BAs). The category of UV-spaces with UV-maps is dually equivalent to the category of Boolean algebras with Boolean homomorphisms. Theorem (Stone duality). Assuming the PFT, the category of Stone spaces with continuous maps is dually equivalent to the category of Boolean algebras with Boolean homomorphisms. Theorem (Choice-free localic Stone duality). The category of Stone locales with localic maps is dually equivalent to the category of Boolean algebras with Boolean homomorphisms. Let’s now relate the first and third approaches.
SLIDE 76
Vietoris space of a Stone locale
The Vietoris space of X may be defined using the open sets instead of the closed sets, which led Johnstone to define for any Stone locale L the Vietoris space of L.
SLIDE 77
Vietoris space of a Stone locale
The Vietoris space of X may be defined using the open sets instead of the closed sets, which led Johnstone to define for any Stone locale L the Vietoris space of L. Similarly, we can define the upper Vietoris space of L.
SLIDE 78
Upper Vietoris space of a Stone locale
The upper Vietoris space of L has as its set of points L− = {a ∈ L | a = 1} with the topology generated by the sets a = {b ∈ L− | a ∨ b = 1}, a ∈ L.
SLIDE 79
Upper Vietoris space of a Stone locale
The upper Vietoris space of L has as its set of points L− = {a ∈ L | a = 1} with the topology generated by the sets a = {b ∈ L− | a ∨ b = 1}, a ∈ L. Idea: switch from nonempty closed sets to their open complements.
SLIDE 80
Upper Vietoris space of a Stone locale
The upper Vietoris space of L has as its set of points L− = {a ∈ L | a = 1} with the topology generated by the sets a = {b ∈ L− | a ∨ b = 1}, a ∈ L. Idea: switch from nonempty closed sets to their open
- complements. Since the basic opens of upper Vietoris are
U = {F ∈ F(X) | F ⊆ U} for U ∈ Ω(X), when we switch to open complements we look at U = {V ∈ Ω(X)\{X} | Vc ⊆ U} = {V ∈ Ω(X)\{X} | U ∪V = X}.
SLIDE 81
Upper Vietoris space of a Stone locale
The upper Vietoris space of L has as its set of points L− = {a ∈ L | a = 1} with the topology generated by the sets a = {b ∈ L− | a ∨ b = 1}, a ∈ L. Idea: switch from nonempty closed sets to their open
- complements. Since the basic opens of upper Vietoris are
U = {F ∈ F(X) | F ⊆ U} for U ∈ Ω(X), when we switch to open complements we look at U = {V ∈ Ω(X)\{X} | Vc ⊆ U} = {V ∈ Ω(X)\{X} | U ∪V = X}. Theorem (Representation of UV-spaces). X is a UV-space iff X is homeomorphic to the upper Vietoris space of a Stone locale.
SLIDE 82
Upper Vietoris locale of a Stone locale
Johnstone also defined the notion of the Vietoris locale of a Stone locale L such that its space of points is homeomorphic to the Vietoris space of L.
SLIDE 83
Upper Vietoris locale of a Stone locale
Johnstone also defined the notion of the Vietoris locale of a Stone locale L such that its space of points is homeomorphic to the Vietoris space of L. Similarly, one can define the upper Vietoris locale of a Stone locale L such that its space of points is homeomorphic to the upper Vietoris space of L.
SLIDE 84
Upper Vietoris locale of a Stone locale
Johnstone also defined the notion of the Vietoris locale of a Stone locale L such that its space of points is homeomorphic to the Vietoris space of L. Similarly, one can define the upper Vietoris locale of a Stone locale L such that its space of points is homeomorphic to the upper Vietoris space of L.
- Theorem. X is a UV-space iff X is homeomorphic to the space of
points of the upper Vietoris locale of a Stone locale.
SLIDE 85
Hyperspace & pointfree approaches related
upper Vietoris locale of L upper Vietoris space of X pt pt locale L of filters Stone space X upper Vietoris space of L BA A UV(A) locales: spaces:
SLIDE 86
Duality dictionary
BA UV Stone BA UV-space Stone space homomorphism UV-map continuous map filter ↑x, x ∈ X closed set ideal U ∈ ORO(X)
- pen set
principal filter U ∈ CORO(X) clopen set principal ideal U ∈ CORO(X) clopen set maximal filter {x}, x ∈ Max(X) {x}, x ∈ X maximal ideal X \ ↓x, x ∈ Max(X) X \ {x}, x ∈ X relativization subspace U ∈ CORO(X) subspace U ∈ Clop(X) complete algebra complete UV-space ED Stone space atom isolated point isolated point atomic algebra Cl(Xiso) = X Cl(Xiso) = X atomless algebra Xiso = ∅ Xiso = ∅ homomorphic image subspace induced by ↑x, x ∈ X closed set subalgebra image under UV-map image under continuous map direct product UV-sum disjoint union canonical completion RO(X) ℘(X) MacNeille completion RO({x ∈ X | ↑x ∈ CORO(X)}) RO(X)
Table: Dictionary for BA, UV, and Stone.
SLIDE 87
Example applications: antichains of BAs
SLIDE 88
Example applications: antichains of BAs
By an antichain in a BA, we mean a collection C of elements such that for all x, y ∈ C with x = y, we have x ∧ y = 0.
SLIDE 89
Example applications: antichains of BAs
By an antichain in a BA, we mean a collection C of elements such that for all x, y ∈ C with x = y, we have x ∧ y = 0.
- Proposition. Every infinite BA contains infinite chains and
infinite antichains.
SLIDE 90
Example applications: antichains of BAs
By an antichain in a BA, we mean a collection C of elements such that for all x, y ∈ C with x = y, we have x ∧ y = 0.
- Proposition. Every infinite BA contains infinite chains and
infinite antichains. The standard Stone duality proof uses the fact that if X is an infinite set and U ⊆ X, then either U is infinite or X \U is infinite.
SLIDE 91
Example applications: antichains of BAs
By an antichain in a BA, we mean a collection C of elements such that for all x, y ∈ C with x = y, we have x ∧ y = 0.
- Proposition. Every infinite BA contains infinite chains and
infinite antichains. The standard Stone duality proof uses the fact that if X is an infinite set and U ⊆ X, then either U is infinite or X \U is infinite. Our proof is very similar, but we use the fact that if X is an infinite separative poset and U ∈ RO(X), then either U is infinite
- r ¬U = Int(X \ U) = {x ∈ X | ∀x′ x x′ ∈ U} is infinite.
SLIDE 92
Example applications: antichains of BAs
- Proposition. Every infinite BA contains infinite chains and
infinite antichains.
Proof.
SLIDE 93
Example applications: antichains of BAs
- Proposition. Every infinite BA contains infinite chains and
infinite antichains.
- Proof. By duality, it suffices to show that in any infinite UV-space X,
there is an infinite descending chain U0 U1 . . . of sets from CORO(X), as well as an infinite family of pairwise disjoint sets from CORO(X).
SLIDE 94
Example applications: antichains of BAs
- Proposition. Every infinite BA contains infinite chains and
infinite antichains.
- Proof. By duality, it suffices to show that in any infinite UV-space X,
there is an infinite descending chain U0 U1 . . . of sets from CORO(X), as well as an infinite family of pairwise disjoint sets from CORO(X). For this it suffices to show that (⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n.
SLIDE 95
Example applications: antichains of BAs
- Proposition. Every infinite BA contains infinite chains and
infinite antichains.
- Proof. By duality, it suffices to show that in any infinite UV-space X,
there is an infinite descending chain U0 U1 . . . of sets from CORO(X), as well as an infinite family of pairwise disjoint sets from CORO(X). For this it suffices to show that (⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n. For then by DC, there is an infinite descending chain U0 ⊇ U1 ⊇ . . . of sets from CORO(X) with Ui ∩ ¬Ui+1 = ∅ for each i ∈ N, in which case {U0 ∩ ¬U1, U1 ∩ ¬U2, . . . } is our antichain.
SLIDE 96
Example applications: antichains of BAs
(⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n. We prove (⋆) by induction.
SLIDE 97
Example applications: antichains of BAs
(⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n. We prove (⋆) by induction. Let U0 = X.
SLIDE 98
Example applications: antichains of BAs
(⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n. We prove (⋆) by induction. Let U0 = X. For the inductive step: Since Un is infinite and X is T0, there are x, y ∈ Un such that x y.
SLIDE 99
Example applications: antichains of BAs
(⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n. We prove (⋆) by induction. Let U0 = X. For the inductive step: Since Un is infinite and X is T0, there are x, y ∈ Un such that x y. Then by the separation property of UV-spaces, there is a V ∈ CORO(X) such that x ∈ V and y ∈ V,
SLIDE 100
Example applications: antichains of BAs
(⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n. We prove (⋆) by induction. Let U0 = X. For the inductive step: Since Un is infinite and X is T0, there are x, y ∈ Un such that x y. Then by the separation property of UV-spaces, there is a V ∈ CORO(X) such that x ∈ V and y ∈ V, which with y ∈ Un and Un, V ∈ RO(X) implies that there is a z y such that z ∈ Un ∩ ¬V.
SLIDE 101
Example applications: antichains of BAs
(⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n. We prove (⋆) by induction. Let U0 = X. For the inductive step: Since Un is infinite and X is T0, there are x, y ∈ Un such that x y. Then by the separation property of UV-spaces, there is a V ∈ CORO(X) such that x ∈ V and y ∈ V, which with y ∈ Un and Un, V ∈ RO(X) implies that there is a z y such that z ∈ Un ∩ ¬V. Since Un, V ∈ CORO(X), we have Un ∩ V, Un ∩ ¬V ∈ CORO(X) by the closure conditions on UV-spaces;
SLIDE 102
Example applications: antichains of BAs
(⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n. We prove (⋆) by induction. Let U0 = X. For the inductive step: Since Un is infinite and X is T0, there are x, y ∈ Un such that x y. Then by the separation property of UV-spaces, there is a V ∈ CORO(X) such that x ∈ V and y ∈ V, which with y ∈ Un and Un, V ∈ RO(X) implies that there is a z y such that z ∈ Un ∩ ¬V. Since Un, V ∈ CORO(X), we have Un ∩ V, Un ∩ ¬V ∈ CORO(X) by the closure conditions on UV-spaces; and since z ∈ Un ∩ ¬V and x ∈ Un ∩ V, we have z ∈ Un ∩ ¬(Un ∩ V) = ∅ and x ∈ Un ∩ ¬(Un ∩ ¬V) = ∅.
SLIDE 103
Example applications: antichains of BAs
(⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n. We prove (⋆) by induction. Let U0 = X. For the inductive step: Since Un is infinite and X is T0, there are x, y ∈ Un such that x y. Then by the separation property of UV-spaces, there is a V ∈ CORO(X) such that x ∈ V and y ∈ V, which with y ∈ Un and Un, V ∈ RO(X) implies that there is a z y such that z ∈ Un ∩ ¬V. Since Un, V ∈ CORO(X), we have Un ∩ V, Un ∩ ¬V ∈ CORO(X) by the closure conditions on UV-spaces; and since z ∈ Un ∩ ¬V and x ∈ Un ∩ V, we have z ∈ Un ∩ ¬(Un ∩ V) = ∅ and x ∈ Un ∩ ¬(Un ∩ ¬V) = ∅. Thus, if Un ∩ V is infinite, then we can set Un+1 := Un ∩ V,
SLIDE 104
Example applications: antichains of BAs
(⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n. We prove (⋆) by induction. Let U0 = X. For the inductive step: Since Un is infinite and X is T0, there are x, y ∈ Un such that x y. Then by the separation property of UV-spaces, there is a V ∈ CORO(X) such that x ∈ V and y ∈ V, which with y ∈ Un and Un, V ∈ RO(X) implies that there is a z y such that z ∈ Un ∩ ¬V. Since Un, V ∈ CORO(X), we have Un ∩ V, Un ∩ ¬V ∈ CORO(X) by the closure conditions on UV-spaces; and since z ∈ Un ∩ ¬V and x ∈ Un ∩ V, we have z ∈ Un ∩ ¬(Un ∩ V) = ∅ and x ∈ Un ∩ ¬(Un ∩ ¬V) = ∅. Thus, if Un ∩ V is infinite, then we can set Un+1 := Un ∩ V, and otherwise we claim that Un ∩ ¬V is infinite, in which case we can set Un+1 := Un ∩ ¬V.
SLIDE 105
Example applications: antichains of BAs
(⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n. We prove (⋆) by induction. Let U0 = X. For the inductive step: Since Un is infinite and X is T0, there are x, y ∈ Un such that x y. Then by the separation property of UV-spaces, there is a V ∈ CORO(X) such that x ∈ V and y ∈ V, which with y ∈ Un and Un, V ∈ RO(X) implies that there is a z y such that z ∈ Un ∩ ¬V. Since Un, V ∈ CORO(X), we have Un ∩ V, Un ∩ ¬V ∈ CORO(X) by the closure conditions on UV-spaces; and since z ∈ Un ∩ ¬V and x ∈ Un ∩ V, we have z ∈ Un ∩ ¬(Un ∩ V) = ∅ and x ∈ Un ∩ ¬(Un ∩ ¬V) = ∅. Thus, if Un ∩ V is infinite, then we can set Un+1 := Un ∩ V, and otherwise we claim that Un ∩ ¬V is infinite, in which case we can set Un+1 := Un ∩ ¬V. Since Un ∈ RO(X), we may regard Un as a separative poset.
SLIDE 106
Example applications: antichains of BAs
(⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n. We prove (⋆) by induction. Let U0 = X. For the inductive step: Since Un is infinite and X is T0, there are x, y ∈ Un such that x y. Then by the separation property of UV-spaces, there is a V ∈ CORO(X) such that x ∈ V and y ∈ V, which with y ∈ Un and Un, V ∈ RO(X) implies that there is a z y such that z ∈ Un ∩ ¬V. Since Un, V ∈ CORO(X), we have Un ∩ V, Un ∩ ¬V ∈ CORO(X) by the closure conditions on UV-spaces; and since z ∈ Un ∩ ¬V and x ∈ Un ∩ V, we have z ∈ Un ∩ ¬(Un ∩ V) = ∅ and x ∈ Un ∩ ¬(Un ∩ ¬V) = ∅. Thus, if Un ∩ V is infinite, then we can set Un+1 := Un ∩ V, and otherwise we claim that Un ∩ ¬V is infinite, in which case we can set Un+1 := Un ∩ ¬V. Since Un ∈ RO(X), we may regard Un as a separative poset. Given V ∈ RO(X), we have Un ∩ V, Un ∩ ¬V ∈ RO(Un)
SLIDE 107
Example applications: antichains of BAs
(⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n. We prove (⋆) by induction. Let U0 = X. For the inductive step: Since Un is infinite and X is T0, there are x, y ∈ Un such that x y. Then by the separation property of UV-spaces, there is a V ∈ CORO(X) such that x ∈ V and y ∈ V, which with y ∈ Un and Un, V ∈ RO(X) implies that there is a z y such that z ∈ Un ∩ ¬V. Since Un, V ∈ CORO(X), we have Un ∩ V, Un ∩ ¬V ∈ CORO(X) by the closure conditions on UV-spaces; and since z ∈ Un ∩ ¬V and x ∈ Un ∩ V, we have z ∈ Un ∩ ¬(Un ∩ V) = ∅ and x ∈ Un ∩ ¬(Un ∩ ¬V) = ∅. Thus, if Un ∩ V is infinite, then we can set Un+1 := Un ∩ V, and otherwise we claim that Un ∩ ¬V is infinite, in which case we can set Un+1 := Un ∩ ¬V. Since Un ∈ RO(X), we may regard Un as a separative poset. Given V ∈ RO(X), we have Un ∩ V, Un ∩ ¬V ∈ RO(Un) and Un ∩ ¬V = ¬n(Un ∩ V), where ¬n is the negation in RO(Un).
SLIDE 108
Example applications: antichains of BAs
(⋆) for any n ∈ N, there is a descending chain U0 ⊇ U1 ⊇ · · · ⊇ Un of infinite sets from CORO(X) such that Ui ∩ ¬Ui+1 = ∅ for i ∈ n. We prove (⋆) by induction. Let U0 = X. For the inductive step: Since Un is infinite and X is T0, there are x, y ∈ Un such that x y. Then by the separation property of UV-spaces, there is a V ∈ CORO(X) such that x ∈ V and y ∈ V, which with y ∈ Un and Un, V ∈ RO(X) implies that there is a z y such that z ∈ Un ∩ ¬V. Since Un, V ∈ CORO(X), we have Un ∩ V, Un ∩ ¬V ∈ CORO(X) by the closure conditions on UV-spaces; and since z ∈ Un ∩ ¬V and x ∈ Un ∩ V, we have z ∈ Un ∩ ¬(Un ∩ V) = ∅ and x ∈ Un ∩ ¬(Un ∩ ¬V) = ∅. Thus, if Un ∩ V is infinite, then we can set Un+1 := Un ∩ V, and otherwise we claim that Un ∩ ¬V is infinite, in which case we can set Un+1 := Un ∩ ¬V. Since Un ∈ RO(X), we may regard Un as a separative poset. Given V ∈ RO(X), we have Un ∩ V, Un ∩ ¬V ∈ RO(Un) and Un ∩ ¬V = ¬n(Un ∩ V), where ¬n is the negation in RO(Un). Then since Un is infinite, either Un ∩ V or ¬n(Un ∩ V) is infinite.
SLIDE 109
Priestley-like duality
The spectral duality of distributive lattices can be reformulated in terms of Priestley spaces.
SLIDE 110
Priestley-like duality
The spectral duality of distributive lattices can be reformulated in terms of Priestley spaces. Assuming AC, we can do the same for the duality via UV-spaces.
SLIDE 111
Priestley-like duality
The spectral duality of distributive lattices can be reformulated in terms of Priestley spaces. Assuming AC, we can do the same for the duality via UV-spaces. Let A be a BA.
SLIDE 112
Priestley-like duality
The spectral duality of distributive lattices can be reformulated in terms of Priestley spaces. Assuming AC, we can do the same for the duality via UV-spaces. Let A be a BA. Let XA be the space of all proper filters, with topology generated by a subbasis of sets of the form { a : a ∈ A}, {XA \ a : a ∈ A}.
SLIDE 113
Priestley-like duality
The spectral duality of distributive lattices can be reformulated in terms of Priestley spaces. Assuming AC, we can do the same for the duality via UV-spaces. Let A be a BA. Let XA be the space of all proper filters, with topology generated by a subbasis of sets of the form { a : a ∈ A}, {XA \ a : a ∈ A}. Let ≤ be the inclusion of proper filters.
SLIDE 114
Priestley-like duality
The spectral duality of distributive lattices can be reformulated in terms of Priestley spaces. Assuming AC, we can do the same for the duality via UV-spaces. Let A be a BA. Let XA be the space of all proper filters, with topology generated by a subbasis of sets of the form { a : a ∈ A}, {XA \ a : a ∈ A}. Let ≤ be the inclusion of proper filters. Then (XA, ≤) is a Priestley space.
SLIDE 115
Priestley-like duality
In addition, if ClopRO(X, ≤) = {clopen ≤-regular open sets}:
1
if U ∈ ClopRO(X, ≤), then Int≤(X \ U) ∈ ClopRO(X, ≤);
SLIDE 116
Priestley-like duality
In addition, if ClopRO(X, ≤) = {clopen ≤-regular open sets}:
1
if U ∈ ClopRO(X, ≤), then Int≤(X \ U) ∈ ClopRO(X, ≤);
2
x ≤ y ⇒ there is a U ∈ ClopRO(X, ≤) s.t. x ∈ U and y / ∈ U;
SLIDE 117
Priestley-like duality
In addition, if ClopRO(X, ≤) = {clopen ≤-regular open sets}:
1
if U ∈ ClopRO(X, ≤), then Int≤(X \ U) ∈ ClopRO(X, ≤);
2
x ≤ y ⇒ there is a U ∈ ClopRO(X, ≤) s.t. x ∈ U and y / ∈ U;
3
every proper filter in ClopRO(X, ≤) is ClopRO(x) for some x ∈ X.
SLIDE 118
Priestley-like duality
Theorem (Priestley-like representation of BAs). Every Boolean algebra A is isomorphic to ClopRO(XA, ≤).
SLIDE 119
Priestley-like duality
Theorem (Priestley-like representation of BAs). Every Boolean algebra A is isomorphic to ClopRO(XA, ≤). Such ordered spaces are order-homeomorphic to the Vietoris space of a Stone space ordered by ⊆.
SLIDE 120
Conclusions and further directions
SLIDE 121
Conclusions and further directions
We developed choice-free topological duality for Boolean algebras.
SLIDE 122
Conclusions and further directions
We developed choice-free topological duality for Boolean algebras. With choice this can be converted into a Priestley-like
- rder-topological duality.
SLIDE 123
Conclusions and further directions
We developed choice-free topological duality for Boolean algebras. With choice this can be converted into a Priestley-like
- rder-topological duality.
We also have extensions of this duality to modal algebras (see our presentation at Advances in Modal Logic 2018).
SLIDE 124
Conclusions and further directions
We developed choice-free topological duality for Boolean algebras. With choice this can be converted into a Priestley-like
- rder-topological duality.
We also have extensions of this duality to modal algebras (see our presentation at Advances in Modal Logic 2018). It should also be possible to give choice-free dualities for Heyting algebras and distributive lattices.
SLIDE 125