TheoremBase

Each predicativity clause is proved by writing down a formula that quantifies over sets only, the Borel condition being read off the membership criterion of the generated σ-algebra; the two families are then the maps given by the resulting defined set symbols.

Proof

Throughout, formulas are written with abbreviations; each formula below quantifies over set variables only and is therefore predicative, its bound variables being renamed apart from the other variables in play where needed (Predicativity under Renaming of Set Variables by Variables, of Formulas with Ordered Pairs, and of Formulas for Relations, Functions, Maps, Injectivity, Surjectivity and Bijectivity (i)). For a function FF and a set uu in its domain, (u,v)∈F(u,v)\in F holds exactly for v=F(u)v=F(u) (Functions, Values of a Function, and Functions from One Class to Another (ii)), and μ(F,A,B)\mu(F,A,B) is the formula of Predicativity under Renaming of Set Variables by Variables, of Formulas with Ordered Pairs, and of Formulas for Relations, Functions, Maps, Injectivity, Surjectivity and Bijectivity (vii), which holds exactly when FF is a map F:A→BF:A\to B.

Proof of (i). Let a,b,c,e,f,p,s,x,y,za,b,c,e,f,p,s,x,y,z be further set variables. Let δ(S,d)\delta(S,d) be

∃c (∀p (p∈c⇔∃x ∃y (x∈S∧y∈S∧p=(x,y)))∧μ(d,c,R)).\exists c\,\Bigl(\forall p\,\bigl(p\in c\Leftrightarrow\exists x\,\exists y\,(x\in S\wedge y\in S\wedge p=(x,y))\bigr)\wedge\mu(d,c,\mathbb{R})\Bigr).

The only cc satisfying its first conjunct is the Cartesian product S×SS\times S, so δ(S,d)\delta(S,d) holds exactly when dd is a map S×S→RS\times S\to\mathbb{R}. Let κ(S,d)\kappa(S,d) be

∀x ∀y ∀z ∀a ∀b ∀e ∀f ∀s ((x∈S∧y∈S∧z∈S∧((x,y),a)∈d∧((y,x),b)∈d∧((x,z),e)∈d∧((y,z),f)∈d∧((a,f),s)∈+)⇒((0,a)∈≤∧(a=0⇔x=y)∧a=b∧(e,s)∈≤)).\forall x\,\forall y\,\forall z\,\forall a\,\forall b\,\forall e\,\forall f\,\forall s\,\Bigl(\bigl(x\in S\wedge y\in S\wedge z\in S\wedge((x,y),a)\in d\wedge((y,x),b)\in d\wedge((x,z),e)\in d\wedge((y,z),f)\in d\wedge((a,f),s)\in{+}\bigr)\Rightarrow\bigl((0,a)\in{\le}\wedge(a=0\Leftrightarrow x=y)\wedge a=b\wedge(e,s)\in{\le}\bigr)\Bigr).

When δ(S,d)\delta(S,d) holds and x,y,z∈Sx,y,z\in S, its premises single out a=d(x,y)a=d(x,y), b=d(y,x)b=d(y,x), e=d(x,z)e=d(x,z), f=d(y,z)f=d(y,z) and s=d(x,y)+d(y,z)s=d(x,y)+d(y,z) (Predicativity of Expressions with Abbreviations, and Symbol-Free Forms of the Basic Defined Set Symbols (xi) for the sum), so κ(S,d)\kappa(S,d) states the four conditions of Metric Space (i). Hence

χ(S,d)=δ(S,d)∧κ(S,d)\chi(S,d)=\delta(S,d)\wedge\kappa(S,d)

expresses that (S,d)(S,d) is a metric space (Metric Space (vi)); its parameters are R\mathbb{R}, ≤\le, ++ and 00.

Proof of (ii). Let zz be a further set variable. By Open Ball in a Metric Space (ii), y∈Bd(x,r)y\in B_{d}(x,r) means y∈Sy\in S and d(x,y)<rd(x,y)<r, so the condition is expressed by

χ(S,d)∧x∈S∧r∈R∧(0,r)∈<∧y∈S∧∃z (((x,y),z)∈d∧(z,r)∈<),\chi(S,d)\wedge x\in S\wedge r\in\mathbb{R}\wedge(0,r)\in{<}\wedge y\in S\wedge\exists z\,\bigl(((x,y),z)\in d\wedge(z,r)\in{<}\bigr),

whose parameters are R\mathbb{R}, ≤\le, <<, ++ and 00.

Proof of (iii). Let ω(U,S,d)\omega(U,S,d) be the formula ω(U)\omega(U) of The Open Sets of a Metric Space Form a Set, Given by a Predicative Formula (i) with SS in place of XX. That clause holds for every metric space (X,d)(X,d), a statement about parameters holding for all their values satisfying its hypotheses (Conditions Expressed by a Predicative Formula, and Guards (i)). So under every assignment for which (S,d)(S,d) is a metric space, ω(U,S,d)\omega(U,S,d) holds if and only if U⊆SU\subseteq S and UU is open in (S,d)(S,d), that is, U∈TdU\in\mathcal{T}_{d} (The Open Sets of a Metric Space Form a Set, Given by a Predicative Formula (ii)). Hence χ(S,d)∧ω(U,S,d)\chi(S,d)\wedge\omega(U,S,d) expresses the condition, with parameters R\mathbb{R}, ≤\le, <<, ++ and 00.

Proof of (iv). Let A,c,f,n,wA,c,f,n,w be further set variables, and let ς(G,Y)\varsigma(G,Y) be the conjunction of the four formulas

∀A (A∈G⇒∀w (w∈A⇒w∈Y)),Y∈G,\forall A\,\bigl(A\in G\Rightarrow\forall w\,(w\in A\Rightarrow w\in Y)\bigr),\qquad Y\in G, ∀A (A∈G⇒∃c (∀w (w∈c⇔(w∈Y∧w∉A))∧c∈G)),\forall A\,\Bigl(A\in G\Rightarrow\exists c\,\bigl(\forall w\,(w\in c\Leftrightarrow(w\in Y\wedge w\notin A))\wedge c\in G\bigr)\Bigr), ∀f (μ(f,N,G)⇒∃c (∀w (w∈c⇔∃n ∃A (n∈N∧(n,A)∈f∧w∈A))∧c∈G)).\forall f\,\Bigl(\mu(f,\mathbb{N},G)\Rightarrow\exists c\,\bigl(\forall w\,(w\in c\Leftrightarrow\exists n\,\exists A\,(n\in\mathbb{N}\wedge(n,A)\in f\wedge w\in A))\wedge c\in G\bigr)\Bigr).

The first says G⊆P(Y)G\subseteq\mathcal{P}(Y) (The Union Set and the Power Set of a Set (ii)). In the third, the only cc with the stated elements is the difference Y∖AY\setminus A. In the fourth, the sets ff with μ(f,N,G)\mu(f,\mathbb{N},G) are the sequences (An)(A_{n}) in GG, and the only cc with the stated elements is the union ⋃n∈NAn\bigcup_{n\in\mathbb{N}}A_{n}, a set by The Union and Product of a Family of Sets Indexed by a Set, and the Intersection of a Family with an Inhabited Index Class, Are Sets (i). So ς(G,Y)\varsigma(G,Y), whose only parameter is N\mathbb{N}, holds if and only if GG is a σ\sigma-algebra on YY (σ-Algebra and Measurable Space (i)).

Proof of (v). Let G,U,wG,U,w be further set variables and let β(B,S,d)\beta(B,S,d) be

χ(S,d)∧∀w (w∈B⇒w∈S)∧∀G ((ς(G,S)∧∀U (ω(U,S,d)⇒U∈G))⇒B∈G).\chi(S,d)\wedge\forall w\,(w\in B\Rightarrow w\in S)\wedge\forall G\,\Bigl(\bigl(\varsigma(G,S)\wedge\forall U\,(\omega(U,S,d)\Rightarrow U\in G)\bigr)\Rightarrow B\in G\Bigr).

If (S,d)(S,d) is not a metric space, β(B,S,d)\beta(B,S,d) and the condition both fail. Let (S,d)(S,d) be a metric space. Then B(S,d)=σ(Td)\mathcal{B}(S,d)=\sigma(\mathcal{T}_{d}) (Borel Sigma-Algebra of a Metric Space (i)), and by The σ-Algebra Generated by a Set of Subsets (ii), B∈σ(Td)B\in\sigma(\mathcal{T}_{d}) if and only if B⊆SB\subseteq S and B∈GB\in G for every σ\sigma-algebra GG on SS with Td⊆G\mathcal{T}_{d}\subseteq G. By (iv), as proved above, ς(G,S)\varsigma(G,S) says that GG is a σ\sigma-algebra on SS; and by the property of ω\omega shown in the proof of (iii), ∀U (ω(U,S,d)⇒U∈G)\forall U\,(\omega(U,S,d)\Rightarrow U\in G) says Td⊆G\mathcal{T}_{d}\subseteq G. So β(B,S,d)\beta(B,S,d) expresses the condition, with parameters R\mathbb{R}, ≤\le, <<, ++, 00 and N\mathbb{N}.

Proof of (vi). Let Ω\Omega, and Σ\Sigma, DD and Ξ\Xi, be parameters whose values are Ω\Omega and the functions giving the families (Sj)j∈J(S_{j})_{j\in J}, (dj)j∈J(d_{j})_{j\in J} and (Xj)j∈J(X_{j})_{j\in J}, and let j,z,s,e,x,A,G,B,c,w,yj,z,s,e,x,A,G,B,c,w,y be set variables. Let π(c,B)\pi(c,B) be ∀w (w∈c⇔∃y (y∈B∧(w,y)∈x))\forall w\,(w\in c\Leftrightarrow\exists y\,(y\in B\wedge(w,y)\in x)), the formula of Preimages under a Map: Membership, the Preimage Map, Sequences of Preimages, and the Set of Sets with Preimage in a Given Set (i) with cc, BB and xx for its zz, VV and uu, and let θ(z,j)\theta(z,j) be

∃s ∃e ∃x ((j,s)∈Σ∧(j,e)∈D∧(j,x)∈Ξ∧∀A (A∈z⇔(∀w (w∈A⇒w∈Ω)∧∀G ((ς(G,Ω)∧∀B (β(B,s,e)⇒∃c (π(c,B)∧c∈G)))⇒A∈G)))).\exists s\,\exists e\,\exists x\,\Bigl((j,s)\in\Sigma\wedge(j,e)\in D\wedge(j,x)\in\Xi\wedge\forall A\,\Bigl(A\in z\Leftrightarrow\bigl(\forall w\,(w\in A\Rightarrow w\in\Omega)\wedge\forall G\,\bigl((\varsigma(G,\Omega)\wedge\forall B\,(\beta(B,s,e)\Rightarrow\exists c\,(\pi(c,B)\wedge c\in G)))\Rightarrow A\in G\bigr)\bigr)\Bigr)\Bigr).

Let j∈Jj\in J. The only sets ss, ee and xx satisfying the first three conjuncts are SjS_{j}, djd_{j} and XjX_{j}, and XjX_{j} is a map Ω→Sj\Omega\to S_{j} (Probability Spaces, Events and Random Elements (iii)). For every set BB, the only cc with π(c,B)\pi(c,B) is Xj−1(B)X_{j}^{-1}(B) (Preimages under a Map: Membership, the Preimage Map, Sequences of Preimages, and the Set of Sets with Preimage in a Given Set (i)), and β(B,Sj,dj)\beta(B,S_{j},d_{j}) holds if and only if B∈B(Sj)B\in\mathcal{B}(S_{j}), by (v) as proved above. So, with Cj={Xj−1(B):B∈B(Sj)}\mathcal{C}_{j}=\{X_{j}^{-1}(B):B\in\mathcal{B}(S_{j})\}, the subset of P(Ω)\mathcal{P}(\Omega) of σ-Algebras Generated by Random Elements (i), the inner formula says that A⊆ΩA\subseteq\Omega and A∈GA\in G for every σ\sigma-algebra GG on Ω\Omega with Cj⊆G\mathcal{C}_{j}\subseteq G, which by The σ-Algebra Generated by a Set of Subsets (ii) means A∈σ(Cj)=σ(Xj)A\in\sigma(\mathcal{C}_{j})=\sigma(X_{j}). Thus θ(z,j)\theta(z,j) holds exactly when zz and σ(Xj)\sigma(X_{j}) have the same elements, that is, when z=σ(Xj)z=\sigma(X_{j}). So σ(Xj)\sigma(X_{j}) is a defined set symbol with argument jj, given by θ\theta and used properly at every j∈Jj\in J. As σ(Xj)\sigma(X_{j}) is a σ\sigma-algebra on Ω\Omega (σ-Algebras Generated by Random Elements (i)), σ(Xj)⊆P(Ω)\sigma(X_{j})\subseteq\mathcal{P}(\Omega), that is, σ(Xj)∈P(P(Ω))\sigma(X_{j})\in\mathcal{P}(\mathcal{P}(\Omega)) (σ-Algebra and Measurable Space (i), The Union Set and the Power Set of a Set (ii)). By Maps and Relations Given by Formulas (i), with JJ for aa, P(P(Ω))\mathcal{P}(\mathcal{P}(\Omega)) for bb and σ(Xj)\sigma(X_{j}) for t(j)t(j), exactly one map G:J→P(P(Ω))\mathcal{G}:J\to\mathcal{P}(\mathcal{P}(\Omega)) has G(j)=σ(Xj)\mathcal{G}(j)=\sigma(X_{j}) for every j∈Jj\in J; it is a family indexed by JJ. Any family (Gj′)j∈J(\mathcal{G}'_{j})_{j\in J} with Gj′=σ(Xj)\mathcal{G}'_{j}=\sigma(X_{j}) for every j∈Jj\in J is a function with domain JJ and range contained in P(P(Ω))\mathcal{P}(\mathcal{P}(\Omega)), hence a map J→P(P(Ω))J\to\mathcal{P}(\mathcal{P}(\Omega)) (Functions, Values of a Function, and Functions from One Class to Another (iii)), and so it is G\mathcal{G}.

Proof of (vii). Let Ξ\Xi and Γ\Gamma be parameters whose values are the functions giving (Xj)j∈J(X_{j})_{j\in J} and (gj)j∈J(g_{j})_{j\in J}, let j,z,x,h,p,u,v,wj,z,x,h,p,u,v,w be set variables, and let θ′(z,j)\theta'(z,j) be

∃x ∃h ((j,x)∈Ξ∧(j,h)∈Γ∧∀p (p∈z⇔∃u ∃v ∃w ((u,v)∈x∧(v,w)∈h∧p=(u,w)))).\exists x\,\exists h\,\Bigl((j,x)\in\Xi\wedge(j,h)\in\Gamma\wedge\forall p\,\bigl(p\in z\Leftrightarrow\exists u\,\exists v\,\exists w\,((u,v)\in x\wedge(v,w)\in h\wedge p=(u,w))\bigr)\Bigr).

Let j∈Jj\in J. The only xx and hh satisfying the first two conjuncts are XjX_{j} and gjg_{j}, so by Relations, Domain, Range, Inverse and Composition (v) the only set zz with θ′(z,j)\theta'(z,j) is the composite gj∘Xj=gj(Xj)g_{j}\circ X_{j}=g_{j}(X_{j}), which is a map Ω→Tj\Omega\to T_{j} by Basic Properties of Functions: Equality, Composition, Identity, Inverse and Restriction (ii). Thus gj(Xj)g_{j}(X_{j}) is a defined set symbol with argument jj, given by θ′\theta' and used properly at every j∈Jj\in J. Let U=⋃j∈JTj\mathcal{U}=\bigcup_{j\in J}T_{j}, a set by The Union and Product of a Family of Sets Indexed by a Set, and the Intersection of a Family with an Inhabited Index Class, Are Sets (i). Every element of gj(Xj)g_{j}(X_{j}) is an ordered pair (u,w)(u,w) with u∈Ωu\in\Omega and w∈Tj⊆Uw\in T_{j}\subseteq\mathcal{U} (Functions, Values of a Function, and Functions from One Class to Another (iii), Indexed Families of Sets and Their Union, Intersection and Product (ii)), so gj(Xj)∈P(Ω×U)g_{j}(X_{j})\in\mathcal{P}(\Omega\times\mathcal{U}). As in the proof of (vi), Maps and Relations Given by Formulas (i) with JJ for aa, P(Ω×U)\mathcal{P}(\Omega\times\mathcal{U}) for bb and gj(Xj)g_{j}(X_{j}) for t(j)t(j) gives exactly one map J→P(Ω×U)J\to\mathcal{P}(\Omega\times\mathcal{U}) with value gj(Xj)g_{j}(X_{j}) at every j∈Jj\in J, and every family with these values is such a map, hence equal to it.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…