• 検索結果がありません。

ファイル置き場 Sendai Logic Homepage Arai20121102

N/A
N/A
Protected

Academic year: 2018

シェア "ファイル置き場 Sendai Logic Homepage Arai20121102"

Copied!
25
0
0

読み込み中.... (全文を見る)

全文

(1)

Proof theory of indescribable cardinals

Toshiyasu Arai(Chiba)

(2)

I will explain how to lift up the ordinal analyses to a Π1n+1- indescribable cardinals. This yields a proof-theoretic reduction of the existence of a Π1n+1-indescribable cardinal to iterations of Π1n-indescribabilities over ZF+V=L.

(3)

PLAN of the talk 1. Π1n-indescribability, pp. 4-9

2. Reduction of Π1N+1-indescribability, pp. 10-24

(4)

1 Π1n-indescribability

Ordinal analyses in [Rathjen94], [A09] yield a proof-theoretic re- duction of Π3-reflecting ordinals to iterations of recursively Mahlo operations.

Recall that a Π3-reflecting ordinal is a recursive analogue of the weakly compact cardinal, i.e., Π11-indescribable cardinal, and a ΠN+2-reflecting ordinal is a recursive analogue of Π1N-indescribable cardinal for N ≥ 1.

For a universal ΠN-formula ΠN(a)

P ∈ RMN(X ) :⇔ ∀b ∈ P [P |= ΠN(b) → ∃Q ∈ X ∩ P (Q |= ΠN(b))] (read:P is Π -reflecting on X .)

(5)

Let <ε be a ∆-predicate such that for any transitive and well- founded model V of KPω, <ε is a canonical well ordering of type εK+1 for the order type K of the class Ord of ordinals in V , and P ∈ RMN(a; ≺) iff

a ∈ P ∈ !{RMN(RMN(b; ≺)) : b ∈ P |= b ≺ a}

for the ΠN-recursively Mahlo operation RMN and definable rela- tions ≺.

Theorem 1.1 ([A∞2]) For each N ≥ 2, KPΠN+1 is ΠN+1- conservative over the theory

(6)

The easier half, KPΠN+1 , V ∈ RMN(*ωn(K + 1)+; <ε) for each n < ω, follows from metainduction on n using the fact that for each n, the transfinite induction up to *ωn(K + 1)+ along <ε is provable in KPω, and a fortiori in KPΠN+1, and RMN(a; <ε) is

a ΠN+1-definable class: Assume ∀b <ε a(V ∈ RMN(b; <ε)) and a ΠN-sentence ϕ holds in V . Then for each b <ε a, the ΠN+1- sentence θ :⇔ (V ∈ RMN(b; <ε))∧ϕ holds in the ΠN+1-reflecting universe V . Pick a transitive P ∈ V so that P contains parameters occurring in θ, and θ holds in P . Since P |= V ∈ RMN(b; <ε) iff P ∈ RMN(b; <ε), this means that V ∈ RMN(a; <ε).

For the other half I need an ordinal analysis based on cut- elimination with operator controlled derivations.

(7)

In lifting up Theorem 1.1 to the indescribable cardinals, a re- cent characterization of Π1n+1-indescribability due to J. Bagaria, M. Magidor and H. Sakai, is helpful.

Definition 1.2 Let κ be a regular uncountable cardinal, and n ∈ ω. S ⊂ κ is said to be Π1n-indescribable in κ if for any A ⊂ Vκ and any Π1n-formula ϕ with Vκ |= ϕ[A] there exists a µ ∈ S such that Vµ |= ϕ[A∩Vµ]. κ is said to be Π1n-indescribable iff κ itself is Π1n-indescribable in κ.

Π1N-Mahlo operation MN:

κ ∈ M (S) iff S ∩ κ is Π1 -indescribable in κ

(8)

Definition 1.3 n-stationary subsets, and (n + 1)-club (closed and unbounded) subsets of a regular uncountable cardinal κ are defined by recursion on n ≥ −1 as follows.

1. S ⊂ κ is (−1)-stationary in κ if S is unbounded in κ.

2. For n ≥ 0, S ⊂ κ is n-stationary in κ if S meets every n-club subset of κ.

3. C ⊂ κ is (n + 1)-club in κ if C is n-stationary in κ, and λ ∈ C for all Π1n-indescribable λ <κ such that C ∩ λ is n-stationary in λ, where λ is defined to be Π1−1-indescribable iff λ is a limit ordinal.

(9)

‘S ⊂ κ is n-stationary’ is Π1n+1-definable uniformly on regular uncountable cardinals κ, while ‘C ⊂ κ is n-club’ is Π1n-definable.

The case n = 0 of the following theorem is due to [Jensen72]. Theorem 1.4 ([Bagaria-Magidor-Sakai∞], cf. [Hellsten03])

Assume V = L. Let n ∈ ω. Then the following are equivalent for every Π1n-indescribable cardinal K:

1. K is Π1n+1-indescribable.

2. For any n-stationary subset S of K, there exists a Π1n-indescribable cardinal λ < K such that S ∩ λ is n-stationary in λ.

(10)

2 Reduction of Π1N+1-indescribability

The Π1N-Mahlo operation MN is Π1N+1-definable uniformly on any regular uncountable cardinals. Hence the operation MN can be iterated along any definable relations in Π1N+1-indescribable car- dinals K as in ΠN+1-reflecting ordinals. By ‘definable’ relation, I mean a definable relation on the universe V , in which a Π1N+1- indescribable cardinal exists.

(11)

Let <ε be a ∆-predicate such that for any transitive and well- founded model V of KPω, <ε is a canonical well ordering of type εI+1 for the order type I of the class Ord of ordinals in V .

I will show that the assumption of the Π1N+1-indescribability is proof-theoretically reducible to iterations of an operation along initial segments of <ε over ZF+V=L. The operation is a mixture of Π1N-Mahlo operation MN and Mostowski collapsings.

Let K be a Π1N+1-indescribable cardinal. For α <ε εK+1 and finite sets Θ ⊂f in (K + 1), Πn+1-classes M hαn[Θ] are defined so that the following holds.

(12)

Theorem 2.1 ([A∞3], [A∞4]) 1. For each n < ω,

ZF+(V = L)+(K is Π1N+1-indescribable) , K ∈ M hωnn(I+1)[∅]. 2. For any Σ1N+2-sentences ϕ, if

ZF + (V = L) + (K is Π1n+1-indescribable) , ϕLK, then we can find an n < ω such that

ZF + (V = L) + (K ∈ Mhωnn(I+1)[∅]) , ϕLK.

(13)

Let us recall ordinals for ZF+V=L in [A∞1]. Let I be a weakly inaccessible cardinal.

Definition 2.2 For X ⊂ LI, HullΣn(X) denotes the Σn-Skolem hull of X in LI. a ∈ HullΣn(X) ⇔ {a} ∈ ΣLnI(X) (a ∈ LI).

Definition 2.3 (Mostowski collapsing function F )

By the Condensation Lemma we have an isomorphism (Mostowski collapsing function)

F : HullΣn(X) ↔ Lγ

for an ordinal γ ≤ I such that F ! Y = id ! Y for any transitive

(14)

Let us denote, though I 4∈ dom(F ) = HullΣn(X) F (I) := γ.

Let us denote the isomorphism F on HullΣn(X) ↔ Lγ by FXΣn. In what follows K denotes a Π1N+1-indescribable cardinal, and I the least weakly inaccessible cardinal above K.

(15)

Definition 2.4 Hα,n(X) is a Skolem hull of {0, K, I}∪X under the functions +, α 6→ ωα,

ΨI,n !α, Ψκ,n!α (regular κ <I ), the Σn-definability: X 6→ HullΣn(X ∩ I)

and the Mostowski collapsing functions

(x = Ψκ,nγ, δ) 6→ Fx∪{κ}Σ1 (δ) (κ ∈ R) and

(x = ΨI,nγ, δ) 6→ FxΣn(δ). For κ ≤ I

(16)

Γ denotes a sequent, a finite set of sentences. Its intended mean- ing is the disjunction "{A : A ∈ Γ}.

The idea of the controlled derivations due to [Buchholz92] is to consider a relation

H ,a Γ

where the set k(A) of L-ranks rkL(c) of sets c occurring in sentences A ∈ Γ together with the depth a of derivations are controlled by Skolem hulls H in such a way that

{a} ∪ k(A) ⊂ H

and simultaneously the L-rank of witnessing sets is to be bounded by the depths.

(17)

Let us see how the class M hαn[Θ] looks like through cut-elimination. Over ZF + (V = L) with K ∈ MN, the Π1N+1-indescribability of K is codified using the L-least counter example S ∈ HullΣ1({K, K+}) to the Π1N+1-indescribability of K.

H ,a! Γ, ¬τN(S, K) H ,ar Γ, ∀ρ ∈ MN ∩ K[τN(S, ρ)]

H ,a Γ (RefK)

where τN(S, ρ) says that S is N -thin(non-stationary)

τN(S, ρ) :⇔ ∃C ⊂ ρ[(C is N -club)ρ ∧ (S ∩ C = ∅)] (Σ1N+1 on ρ)

(18)

Let us try to show the following lemma by induction on ξ.

Lemma 2.5 (?) For Π1N+1-sentences A and a ξ-times iterations M hξ of MN

H ,ξ AK ⇒ ∀π ∈ M hξ(Aπ is true).

(19)

H ,ξ AK ⇒ ∀π ∈ M hξ(Aπ is true).

H ,ξ! AK, ¬τN(S, K) H ,ξr AK, ∀ρ ∈ MN ∩ K[τN(S, ρ)]

H ,ξ AK (RefK)

Let π ∈ M hξ. IH yields ∀ρ ∈ MN ∩ π[Aπ ∨ τN(S, ρ)] and

∀ρ ∈ M hξ! ∩ π[Aρ ∨ ¬τN(S, ρ)]. Supposing

M hξ! ⊂ MN (1)

we have ∀ρ ∈ M hξ! ∩ π[Aρ ∨ Aπ].

(20)

H ,ξ AK ⇒ ∀π ∈ M hξ(Aπ is true). On the other hand we have π ∈ M hξ ∧ ξ/ ∈ H ∩ ξ.

Moreover suppose

π ∈ M hξ ∧ ξ/ ∈ H ∩ ξ ⇒ π ∈ MN(M hξ!) (2) Then Proposition 2.6 with Aπ ∨ ∀ρ ∈ M hξ! ∩ π Aρ yields Aπ. Proposition 2.6 Let A be a Π1N+1-sentence, and π ∈ MN(X).

If ∀ρ ∈ X ∩ π[Lλ |= A], then Lπ |= A.

(21)

H ,ξ AK ⇒ ∀π ∈ M hξ(Aπ is true).

For a true literal (d ∈ S), where S ⊂ K such that S ∈ HullIΣ1({K, K+}) and d ∈ K. (d ∈ S)π ≡ (d ∈ (S ∩ π)).

H ,ξ d ∈ S (

#)

We have d ∈ H ∩ K, and hence d < π if

π ∈ M hξ ⇒ H ∩ K ⊂ π (3)

(22)

H ,ξ AK ⇒ ∀π ∈ M hξ(Aπ is true).

{H[{rkL(C)}] ,ξ(C) (S ∩ C 4= ∅) : C is N -club in K} H ,ξ ¬τN(S, K) (S ∈ HullΣ1({K, K+})) (

#)

Let π ∈ M hξ, and suppose τN(S, π), i.e., S ∩ π is N -thin, and let C0 := µC ∈ Lπ+[(C ⊂ π is N -club) ∧ (S ∩ C = ∅)]

It turns out that the multiplication C of C0 is also N -club in K: C = {γ ∈ K : ∃x, y < K(γ = π · x + y ∧ y ∈ C0 ∪ {0})}. Since C ∈ HullΣ1({K, K+, π, π+}) ⊂ H[{π}], we have

H[{π}] ,ξ(C) (S ∩ C 4= ∅)

(23)

H ,ξ AK ⇒ ∀π ∈ M hξ(Aπ is true).

H[{π}] ,ξ(C) (S ∩ C 4= ∅) IH yields ∀ρ ∈ M hξ(C)(S ∩ C ∩ ρ 4= ∅), and

∅ = S ∩ C0 = S ∩ C ∩ π 4= ∅ by Proposition 2.6 if

π ∈ M hξ ∧ ξ(C) ∈ H[{π}] ∩ ξ ⇒ π ∈ MN(M hξ(C)) (4) Finally let us define classes M hξ.

(24)

π ∈ M hξ ⇒ H ∩ K ⊂ π (3)

M hξ! ⊂ MN (1)

π ∈ M hξ ∧ ξ(C) ∈ H[{π}] ∩ ξ ⇒ π ∈ MN(M hξ(C)) (4) Definition 2.7 Let Θ ⊂f in (K + 1) and K ≥ π be regular uncountable. Then π ∈ M hαn[Θ] iff

Hα,n(π) ∩ K ⊂ π & α ∈ Hα,n[Θ](π)

& ∀ξ ∈ Hξ,n[Θ ∪ {π}](π) ∩ α[π ∈ MN(M hξn[Θ ∪ {π}])]

(25)

References

[A97] T. Arai, A sneak preview of proof theory of ordinals, an invited talk at Kobe seminar on Logic and Computer Science, 5-6 Dec. 1997. appeared in Ann. Japan Asso. Phil. Sci. 20(2012), 29-47. [A09] T. Arai, Iterating the recursively Mahlo operations, 13th LMPS(2009), pp. 21-35.

[A∞1] T. Arai, Lifting up the proof theory to the countables: Zermelo-Fraenkel’s set theory, submitted. arXiv: 1101.5660.

[A∞2] T. Arai, Conservations of first-order reflections, submitted. arXiv 1204.0205. [A∞3] T. Arai, Proof theory of weak compactness, submitted. arXiv:1111.0462. [A∞4] T. Arai, Proof theory of Π1n-indescribability, in preparation.

[Bagaria-Magidor-Sakai∞] J. Bagaria, M. Magidor and H. Sakai, private communication.

[Buchholz92] W. Buchholz, A simplified version of local predicativity, P. H. G. Aczel, H. Simmons and S. S. Wainer(eds.), Proof Theory, Cambridge UP, 1992, pp. 115-147.

[Hellsten03] A. Hellsten, Diamonds on large cardinals, Ann. Acad. Sci. Fenn., Ser. A I, Diss. 133, 2003. [Jensen72] R. Jensen, The fine structure of the constructible hierarchy, AML 4(1972), 229-308.

参照

関連したドキュメント

Specifically, we consider the glueing of (i) symmetric monoidal closed cat- egories (models of Multiplicative Intuitionistic Linear Logic), (ii) symmetric monoidal adjunctions

3.1, together with the result in (Barber and Plotkin 1997) (completeness via the term model construction), is that the term model of DCLL forms a model of DILL, i.e., a

Light linear logic ( LLL , Girard 1998): subsystem of linear logic corresponding to polynomial time complexity.. Proofs of LLL precisely captures the polynomial time functions via

Polarity, Girard’s test from Linear Logic Hypersequent calculus from Fuzzy Logic DM completion from Substructural Logic. to establish uniform cut-elimination for extensions of

In [9], it was shown that under diffusive scaling, the random set of coalescing random walk paths with one walker starting from every point on the space-time lattice Z × Z converges

It turned out that the propositional part of our D- translation uses the same construction as de Paiva’s dialectica category GC and we show how our D-translation extends GC to

John Baez, University of California, Riverside: [email protected] Michael Barr, McGill University: [email protected] Lawrence Breen, Universit´ e de Paris

In this chapter, we shall introduce light affine phase semantics, which is meant to be a sound and complete semantics for ILAL, and show the finite model property for ILAL.. As