Categorical Structures of Basic Pairs and
Concrete Spaces
Tatsuji Kawai
Japan Advanced Institute of Science and Technology
February 24, 2012
Introduction
Several competing notions of general topology exist in constructive mathematics:
◮ Neighbourhood space — Topology with base
◮ Formal topology — Point-free
◮ Basic pair and concrete space — Similar to neighbourhood space but with difference in maps
Theorem (Ishihara and K, 2012)
The category of basic pairs and concrete spaces are complete and cocomplete.
Constructive Zermelo-Frankel Set Theory (CZF)
Set theory for constructive mathematics based on ZF.
◮ Uses intuitionistic logic — no excluded middle ϕ∨ ¬ϕ.
◮ No Powerset — Pow(X) is not a set.
◮ Restricted separation —{x ∈ X | ϕ(x)} is a set only for restricted formula ϕ; quantifications must be of the forms
∀x ∈ a or ∃x ∈ a.
◮ Has an interpretation in Martin-Löf’s Intuitionistic Type Theory (ITT) [Aczel 1977,1982,1986].
Basic Pairs
Definition (Basic Pair)
Abasic pairis a triple(X, , S) where X and S are sets and ⊆ X × S is a relation.
Maps between basic pairs
Definition (Relation Pair)
A mapping between basic pairs(X1, 1, S1) and (X2, 2, S2) is a pair(r, s) of relations r ⊆ X1× X2and s⊆ S1× S2 such that
2 ◦ r = s ◦ 1
X1 r
1
//S
1 s
X2 2
//S2
where◦ is the composition of relations:
r2◦ r1= {(x, z) ∈ X × Z | (∃y ∈ Y) x r1y∧ y r2z} for r1⊆ X × Y, and r2⊆ Y × Z.
The pair(r, s) is called arelation pair.
Equivalent relation pairs
Definition
Two relation pairs(r1, s1), (r2, s2) : (X, , S) → (X′, ′, S′) are said to beequivalent,(r1, s1) ∼ (r2, s2), if
′ ◦r1 = ′ ◦r2
s1◦ = s2◦
X
r1
r2
//S
s1
s2
X′
′
//S′
∼ is an equivalence relation on the relation pairs from X to X′.
Category
Definition (Category) AcategoryC consists of:
◮ Objects — A, B, C, . . .
◮ Arrows — f , g, h, . . .
◮ Each arrow f is assigned objects dom(f ), cod(f ). f : A → B ⇐⇒ dom(f ) = A & cod(f ) = Bdef
◮ For any f : A → B and g : B → C, there is an arrow g◦ f : A → C.
◮ For each object A, there is an arrow 1A: A → A. Those are required to satisfy
(Associativity) h◦ (g ◦ f ) = (h ◦ g) ◦ f , (Identity) f◦ 1dom(f )= f = 1cod(f )◦ f .
Category of Basic Pair (BP)
Definition (The category BP)
The category of basic pair BP consists of
◮ Objects — basic pairs(X, , S), (Y, ′, T), · · ·
◮ Arrows — equivalence classes of relation pairs (r, s), (u, v), · · ·
◮ Composition is defined by
(u, v)◦(r, s) = (u ◦ r, v ◦ s)
X1 1
//
r
u◦r
S1 s
v◦s
X2 2 //
u
S2
v
X3 3 //S3
◮ The identity on(X, , S) is (idX, idS) where idX = {(x, x) | x ∈ X} .
Completeness of categories
Definition (Completeness)
A category iscompleteif it has all limits. Proposition
A category is complete if and only if it has
◮ a product for any family of objects(Ai)i∈I,
◮ an equaliser for any pair of arrows A f //
g //B .
Products
Definition (Product)
Aproductfor a family of objects(Ai)i∈I is an objectQi∈IAi together with a family of arrows
πi :Qi∈IAi → Aii∈I such that for any family of arrows(fi: T → Ai)i∈I, there is unique arrow h: T →Qi∈IAisuch that
T
fi
%%
KK KK KK KK KK KK
h
Q
i∈IAi πi
//A
i
fi = πi◦ h
for all i∈ I.
Equalisers
Definition (Equaliser)
Anequaliserfor any pair of arrows A
f //
g //B is an arrow e: E → A such that
◮ E e //A
f //
g //B , i.e. f ◦ e = g ◦ e,
◮ for any h: T → A such that T h //A
f //
g //B ,
there is a unique arrow u: T → E such that T
h
##
FF FF F FF FF
u
E e //A
f //
g //B
h= e ◦ u.
Cocompleteness of categories
Definition (Cocompleteness)
A category iscocompleteif it has all colimits. Proposition
A category is cocomplete if and only if it has
◮ a coproduct for any family of objects(Ai)i∈I,
◮ a coequaliser for any pair of arrows A f //
g //B .
Coproduct
Definition (Coproducts)
Acoproductfor a family of objects(Ai)i∈I is an object`i∈IAi
together with a family of arrows
σi : Ai→`i∈IAi
i∈I
such that for any family of arrows(fi: Ai → T)i∈I, there is unique arrow h:`i∈IAi→ T such that
Ai σi //`i∈IAi
h
T
fi
%%
KK KK KK KK KK KK
fi = h ◦ σi
for all i∈ I.
Coequaliser
Definition (Coequalisers)
Acoequaliserfor any pair of arrows A
f //
g //B is an arrow q: B → Q such that
◮ A f //
g //B
q //Q, i.e. q◦ f = q ◦ g,
◮ for any h: B → T such that A f //
g //B h
//T ,
there is a unique arrow u: Q → T such that
A
f //
g //B
q //Q
u
T
h III$$ II II
II h= u ◦ q.
Coproducts in BP
Proposition
BP has coproducts.
Proof.
Given a family of basic pairs(Xi)i∈I, define a basic pair
`
i∈IXi= (
P
i∈IXi, Σ,
P
i∈ISi) by
Σ=((i, x), (j, a)) ∈ Pi∈IXi×Pi∈ISi | i = j & x i a and a family of relation pairs (ri, si) : Xi→`i∈IXii∈I by
ri =(x, (j, x′)) ∈ Xi×Pi∈IXi | i = j & x = x′ , si =(a, (j, a′)) ∈ Si×Pi∈ISi | i = j & a = a′ .
Set-generated Class
Definition
A class C of subsets of a set S isset-generatedif there is a subset G of C such that
(∀U ∈ C) (∀σ ∈ Fin (U)) (∃V ∈ G) σ ⊆ V ⊆ U, where
◮ Fin(X) is the set of finitely enumerable subsets X A set A isfinitely enumerable
⇐⇒ there is a surjection f : n = {0, · · · , n − 1} → Adef
◮ Gis called agenerating subsetof C
Generalized Geometric Theory
Definition
Ageneralized geometric implication(simply implication) and ageneralized geometric theory(simply theory) over a set S, and their ranks, are defined simultaneously by
1. sis an implication of rank 0 for each s∈ S;
2. if σ∈ Fin(S) and Γ is a set of theories of rank n, then V σ → WU∈ΓV U
is an implication of rank n+ 1;
3. a set T of implications of rank≤ n is a theory of rank n.
Model of a theory
Definition
The relation|= between a set α ∈ Pow(S), and implications s (of rank 0),V σ → WU∈ΓV U (of positive rank) and a theory T over Sis defined by
1. α|= s if s ∈ α;
2. α|=V σ → WU∈ΓV U if σ ⊆ α implies α |= U for some U∈ Γ;
3. α|= T if α |= θ for all θ ∈ T.
A set α∈ Pow(S) is amodelof a theory T if α|= T.
Fundamental theorem
Theorem (Aczel, Ishihara, Nemoto, Sangu 2012)
Assuming RRS2-uREA or RDC, the class M(T) of models of a theory T of rank n is set-generated.
Coequalisers in BP
Proposition
BP has coequalisers. Proof.
Given a pair of relation pairs (X1, 1, S1) = X1
(r1,s1)
//
(r2,s2)
//X2= (X2, 2, S2) ,
define a class Q by
Q=U ∈ Pow(S2) | (s1◦ 1)−U = (s2◦ 1)−U . Qis the class of models of the theory T over the set S2.
T =nV{a} → Vx∈(s1◦ 1)−{a}Wb∈(s2◦ ){x}V{b}a∈ S2
o
∪nV{a} → Vx∈(s2◦ 1)−{a}Wb∈(s1◦ ){x}V{b}a∈ S2o.
Coequalisers in BP Proof continued.
Q=U ∈ Pow(S2) | (s1◦ 1)−U = (s2◦ 1)−U . (s1◦ 1)−U = (s2◦ 1)−U
⇐⇒ (s1◦ 1)−U⊆ (s2◦ 1)−U& (s1◦ 1)−U ⊇ (s2◦ 1)−U. (s1◦ 1)−U ⊆ (s2◦ 1)−U
⇐⇒ (∀a ∈ U) (s1◦ 1)−{a} ⊆ (s2◦ 1)−U
⇐⇒ (∀a ∈ U) ∀x (s1◦ 1)−{a} (∃b ∈ (s2◦ 1) {x}) b ∈ U T =nV{a} → Vx∈(s1◦ 1)−{a}Wb∈(s2◦ ){x}V{b}a∈ S2o
∪nV{a} → Vx∈(s2◦ 1)−{a}
W
b∈(s1◦ ){x}V{b}
a∈ S2
o .
Coequalisers in BP
Proof continued.
Therefore, the class Q has a generating subset G. Then
X2
(idX2,∈)
//G = (X
2, G, G)
X2 idX2
2
//S2
∈
X2 G
//G
where
G = {(x, U) ∈ X2× G | (∃a ∈ U) x 2a}
is a coequaliser for X1
(r1,s1)
//
(r2,s2)
//X2.
Completeness and Cocompleteness of BP
Corollary
BP is cocomplete.
Proposition
BP is isomorphic to its own dual, i.e. BP ∼= BPop.
Corollary BP is complete.
Concrete Space
Definition (Concrete Space)
Aconcrete spaceis a basic pair(X, , S) which satisfies (B1) ext(S) = X,
(B2) ext(a) ∩ ext(b) = ext(a ↓ b), where
ext(a) = {x ∈ X | x a} , ext(U) =Sa∈Uext(a),
a↓ b = {c ∈ S | ext(c) ⊆ ext(a) ∩ ext(b)} . for a, b∈ S and U ⊆ S.
Convergent Relation Pair
Definition (Convergent Relation Pair)
A mapping between concrete spacesX1 = (X1, 1, S1) and X2= (X2, 2, S2) is a relation pair (r, s) : X1→ X2such that
(C1) r−ext(a ↓ b) = ext (s−a↓ s−b), (C2) r−ext S2= ext S1
for a, b∈ S2, where
r−(U) = {x ∈ X | (∃y ∈ U) x r y} for a relation r⊆ X × Y.
A relation pair(r, s) which satisfies (C1) and (C2) is called a convergent relation pair.
Category of Concrete Space (CSpa)
Definition (The category CSpa)
The category of concrete spaces CSpa consists of
◮ Objects — concrete spaces(X, , S), (Y, ′, T), · · ·
◮ Arrows — convergent relation pairs(r, s), (u, v), · · ·
◮ Composition of arrows and identity arrows
— defined as in BP:
(u, v) ◦ (r, s) = (u ◦ r, v ◦ s), 1X = (idX, idS) for a concrete spaceX = (X, , S).
Completeness and Cocompleteness of CSpa
Proposition
The category of concrete spaces (CSpa) has (co)products and (co)equalisers.
Theorem (Ishihara and K, 2012) CSpa is complete and cocomplete.
RRS2-uREA
A set A isregularif
◮ it is transitive, i.e. a⊆ A for each a ∈ A,
◮ for each a∈ A and R ⊆ a × A such that
(∀x ∈ a) (∃y ∈ A) (x, y) ∈ R, there exists b ∈ A such that (∀x ∈ a) (∃y ∈ b) (x, y) ∈ R & (∀y ∈ b) (∃x ∈ a) (x, y) ∈ R. A set A isunion-closedifS a ∈ A for each a ∈ A.
A regular set A isRRS2-regularif for each A′ ∈ Pow(A) and R⊆ (A′× A′) × A′such that∀(x, y) ∈ A′× A′∃z ∈ A′((x, y), z) ∈ R, for each a0∈ A′, there exists A0∈ A such that a0∈ A0⊆ A′ and
∀(x, y) ∈ A0× A0∃z ∈ A0((x, y), z) ∈ R. Definition (RRS2-uREA)
Every set is a subset of a union-closed RRS2-regular set.
Relativised Dependent Choice
Definition (RDC)
For any formulae ϕ and ψ, if∀x [ϕ(x) → ∃y [ϕ(y) ∧ ψ(x, y)]] and ϕ(a0), then there exists a function f with domain N such that f(0) = a0 and∀n ∈ N [ϕ(f (n)) ∧ ψ(f (n), f (n + 1))].