Suppose κ is a regular uncountable cardinal, λ is a cardinal > 1, and is a κ-complete uniform ideal on κ. This paper deals with a saturation property Sat(κ, λ, ) of , which is a weakening of usual λ-saturatedness. Roughly speaking, Sat(κ, λ, ) means that can be densely extended to λ-saturated ideals on small fields of subsets of κ. We will show that some consequences of the existence of a λ-saturated ideal on κ follow from weaker ∃: Sat(κ, λ, ), and that ∃: Sat(κ, λ, ) is connected with weak compactness and complete ineffability of κ in much the same way as the existence of a saturated ideal on κ is connected with measurability of κ.

In §2, we define Sat(κ, λ, ), mention a few results that can be proved by straightforward adaptation of known methods, and discuss generic ultrapowers of ZFC−, which will be used repeatedly in the subsequent sections as a main technical tool. A related concept Sat(κ, λ) is also defined and shown to be equivalent to ∃: Sat(κ, λ, ) under a certain condition.

In §3, we show that ∃: Sat(κ, κ, ) implies that κ is highly Mahlo, improving results in [KT] and [So].

Qλ is the set of nonprincipal filters on ω which are generated by fewer than λ sets, for λ a fixed uncountable, regular cardinal ≤ c. We analyze forcing with Qλ, where Qλ is partially ordered in such a way that a filter F1 is more informative than F2 iff F1 includes F2. Qλ-forcing adjoins an ultrafilter on ω but adds no new reals. We analyze Qλ-forcing from a forcing-theoretic viewpoint. We also analyze the properties of Qλ-generic ultrafilters. These properties are independent of ZFC and depend very much on the ground model. In particular, we study Qλ-forcing over ground models which are Cohen real extensions, random real extensions, and models which satisfy Martin's Axiom.

In §2 we give notations and definitions, and review some of the basic facts about forcing and ultrafilters which we will use. In §3 we introduce Qλ-forcing and prove some basic lemmas about it. §4 studies Qc-forcing. §§5, 6, and 7 analyze Qλ-forcing over ground models of Martin's Axiom, ground models which are generated by Cohen reals, and ground models which are generated by random reals, respectively. Qλ-forcing over Cohen real and random real models is isomorphic to the notion of forcing which adjoins a Cohen generic subset of λ; this is proved in §8.

Definition. A cardinal κ is strong iff for every x there is an elementary embedding j:V → M with critical point κ such that x ∈ M.

κ is superstrong iff ∃j:V → M with critical point κ such that Vj(κ) ∈ M.

These definitions are natural weakenings of supercompactness and hugeness respectively and display some of the same relations. For example, if κ is superstrong then Vκ ⊨ “∃ proper class of strong cardinals”, but the smallest superstrong cardinal is less than the smallest strong cardinal (if both types exist). (See [SRK] and [Mo] for the arguments involving supercompact and huge, which translate routinely to strong and superstrong.)

Given any two types of large cardinals, a typical vague question which is often asked is “How large is the gap in consistency strength?” In one sense the gap might be considered relatively small, since the “higher degree” strong cardinals described below (a standard trick that is nearly always available) and the Shelah and Woodin hierarchies of cardinals (see [St] for a definition of these) seem to be (at least at this point in time) the only “natural” large cardinal properties lying between strong cardinals and superstrong cardinals in consistency strength.

Let be any proper ideal of subsets of the real line R which contains all finite subsets of R. We define an ideal * ∣ as follows: X ∈ * ∣ if there exists a Borel set B ⊂ R × R such that X ⊂ B and for any x ∈ R we have {y ∈ R: 〈x, y〉 ∈ B} ∈ . We show that there exists a family ⊂ * ∣ of power ω1 such that ⋃ ∉ * ∣ .

In the last section we investigate properties of ideals of Lebesgue measure zero sets and meager sets in Cohen extensions of models of set theory.

Quoi qu'en pensent certains, la déviation de Saharon Shelah n'est pas qu'une construction technique servant à un résultat plus technique encore, le théorème de classification; c'est la notion de dépendance la plus générale qui soit apparue dans un contexte mathématique, si bien que toute propriété qu'on pourra attribuer à cette déviation sera, dans le fond, une propriété de la notion même de corrélation.

L'auteur de ces lignes a trouvé la vie beaucoup plus plaisante quand il a pu lire la déviation, associée à une théorie stable T, dans un ordre qu'il était facile de définir sur les types complets au-dessus des modèles de T; s'il a qualifié cet ordre de “fondamental”, ce n'est pas pour donner une emphase hors de propos à sa modeste contribution au démélage de l'échevau shelahien, mais parce que le rang U de Lascar n'était autre que le rang de fondation de l'ordre fondamental.

L'ordre fondamental condense l'essentiel de la déviation, sous une forme pour ainsi dire graphique; certes, tout ne peut s'y retrouver, et en particulier des théories peuvent avoir des ordres fondamentaux semblables si on se limite aux 1-types, et fort différents si on regarde aussi les bitypes; mais il est clair que dès qu'apparait l'ordre fondamental se pose de façon urgente le problème de décrire tous les ordres qui peuvent être ordre fondamental d'une théorie stable: tant que ce programme n'est pas rempli, il reste des propriétés cachées de la déviation.

The aim of the paper is to prove the completeness theorem for biprobability models. This also solves Keisler's Problem 5.4 (see [4]).

Let be a countable admissible set and ω ∈ . The logic is similar to the standard probability logic . The only difference is that two types of probability quantifiers and are allowed.

A biprobability model is a structure (, μ1, μ2) where is a classical structure without operations and μ1, μ2 are two types of probability measures such that μ1 is absolutely continuous with respect to μ2, i.e. μ1 ≪ μ2.

The quantifiers are interpreted in the natural way, i.e.

for i = 1, 2. (The measure is the restriction of the completion of to the σ-algebra generated by the measurable rectangles and the diagonal sets

Axioms and rules of inference are those of , as listed in [2] with the axiom B4 from [4], with the remark that both P1 and P2 can play the role of P, together with the following axioms:

Axioms of continuity.

• 1) .

• 2) .

Axiom of absolute continuity:

where and Φn = {φ ∈ Φ: φ has n free variables}.

The theory of large cardinals in the absence of the axiom of choice (AC) has been examined extensively by set theorists. A particular motivation has been the study of large cardinals and their interrelationships with the axiom of determinacy (AD). Many important and beautiful theorems have been proven in this area, especially by Woodin, who has shown how to obtain, from hypermeasurability, models for the theories “ZF + DC + ∀α < ℵ1(ℵ1 → (ℵ1)α)” and . Thus, consequences of AD whose consistency strength appeared to be beyond that of the more standard large cardinal hypotheses were shown to have suprisingly weak consistency strength.

In this paper, we continue the study of large cardinals in the absence of AC and their interrelationships with AD by examining what large cardinal structures are possible on cardinals below ℵω in the absence of AC. Specifically, we prove the following theorems.

Theorem 1. Con(ZFC + κ1 < κ2are supercompact cardinals) ⇒ Con(ZF + DC + The club filter on ℵ1is a normal measure + ℵ1and ℵ2are supercompact cardinals).

Theorem 2. Con(ZF + AD) ⇒ Con(ZF + ℵ1, ℵ2and ℵ3are measurable cardinals which carry normal measures + μωis not a measure on any of these cardinals).

We characterize some large cardinal properties, such as μ-measurability and P2(κ)-measurability, in terms of ultrafilters, and then explore the Rudin-Keisler (RK) relations between these ultrafilters and supercompact measures on Pκ(2κ). This leads to the characterization of 2κ-supercompactness in terms of a measure on measure sequences, and also to the study of a certain natural subset, Fullκ, of Pκ(2κ) whose elements code measures on cardinals less than κ. The hypothesis that Fullκ is stationary (a weaker assumption than 2κ-supercompactness) is equivalent to a higher order Löwenheim-Skolem property, and settles a question about directed versus chain-type unions on Pκλ.

It is a theorem of Rowbottom [12] that if κ is measurable and I is a normal prime ideal on κ, then for each λ < κ,

In this paper a natural structural property of ideals, distributivity, is considered and shown to be related to this and other ideal theoretic partition relations.

The set theoretical terminology is standard (see [7]) and background results on the theory of ideals may be found in [5] and [8]. Throughout κ will denote an uncountable regular cardinal, and I a proper, nonprincipal, κ-complete ideal on κ. NSκ is the ideal of nonstationary subsets of κ, and Iκ = {X ⊆ κ∣∣X∣<κ}. If A ∈ I+ (= P(κ) − I), then an I-partition of A is a maximal collection W ⊆, P(A) ∩ I+ so that X ∩ Y ∈ I whenever X, Y ∈ W, X ≠ Y. The I-partition W is said to be disjoint if distinct members of W are disjoint, and in this case, for denotes the unique member of W containing ξ. A sequence 〈Wα ∣ α < η} of I-partitions of A is said to be decreasing if whenever α < β < η and X ∈ Wβ there is a Y ∈ Wα such that X ⊆ Y. (i.e., Wβ refines Wα).

A natural deduction formulation is given for the intermediate logic called MH by Gabbay in [4]. Proof-theoretic methods are used to show that every deduction can be normalized, that MH is the weakest intermediate logic for which the Glivenko theorem holds, and that the Craig-Lyndon interpolation theorem holds for it.

Namba [N] proved that the nonstationary ideal over a measurable (NSκ) cannot be κ+-saturated. Baumgartner, Taylor and Wagon [BTW] asked if it is possible for NSκ to be precipitous over a measurable κ. A model with this property was constructed by the author, and shortly after Foreman, Magidor and Shelah [FMS] proved a general theorem that after collapsing of a supercompact or even a superstrong to the successor of κ, NSκ became precipitous. This theorem implies that it is possible to have the nonstationary ideal precipitous over even a supercompact cardinal. Just start with a supercompact κ and a superstrong λ > κ. Make supercompactness of κ indistractible as in [L] and then collapse λ to be κ+.

The aim of our paper is to show that the existence of a supercompact cardinal alone already implies the consistency of the nonstationary ideal precipitous over a supercompact. The proof gives also the following: if κ is a λ-supercompact for λ ≥ (2κ)+, then there exists a generic extension in which κ is λ-supercompact and NSκ is precipitous. Thus, for a model with NSκ precipitous over a measurable we need a (2κ)+-supercompact cardinal κ. Jech [J] proved that the precipitous of NSκ over a measurable κ implies the existence of an inner model with o(κ) = κ+ + 1. In §3 we improve this result a little by showing that the above assumption implies an inner model with a repeat point.

The paper is organized as follows. In §1 some preliminary facts are proved. The model with NSκ precipitous over a supercompact is constructed in §2.

The complexity of the set of truths of arithmetic is determined for various theories of truth deriving from Kripke and from Gupta and Herzberger.

Let PA be Peano arithmetic with function symbols′, + and ·. The length of a proof P, denoted by lh(P), is the maximum length of threads of P (for the term ‘thread’, see [T, p. 14]). For a formula A and a natural number m, PA ⊢mA denotes the fact that there is a proof in PA of A whose length is less than or equal to m. PA ⊢ A denotes the fact that there is a proof in PA of A.

G. Kreisel conjectured that the following proposition holds.

“Let m be a natural number and A(a) be a formula. If for each natural number n, then PA ⊢ ∀xA(x)”.

Let PA1 be the corresponding system with + and · replaced by ternary predicates and axioms saying that these predicates represent functions. Parikh [P] proved the following proposition which is obtained from Kreisel's conjecture by replacing PA by PA1.

Proposition. Let A(a) be a formula in PA1and m be a natural number. Assume thatfor each natural number n. Then PA1 ⊢ ∀xA(x).

The reason why Parikh's method succeeds is the fact that the only function symbol ′ in PA1 is unary. So his method fails for PA.

To solve this conjecture for PA, we must make syntactical investigation into proofs in PA of formulas of the form A() with length ≤ m. Even if lengths of proofs are less than or equal to m, depths of occurrences of bound variables in induction axiom schemata or equality axiom schemata in proofs are not always bounded.

Suppose κ is a supercompact cardinal. It is known that for every λ ≥ κ, many normal ultrafilters on Pκ(λ) have the partition property. It is also known that certain large cardinal assumptions imply the existence of normal ultrafilters without the partition property. In [1], we introduced the tree T of normal ultrafilters associated with κ. We investigate the distribution throughout T of normal ultrafilters with and normal ultrafilters without the partition property.

Let M be an O-minimal structure. We use our understanding, acquired in [KPS], of the structure of definable sets of n-tuples in M, to study definable (in M) equivalence relations on Mn. In particular, we show that if E is an A-definable equivalence relation on Mn (A ⊂ M) then E has only finitely many classes with nonempty interior in Mn, each such class being moreover also A-definable. As a consequence, we are able to give some conditions under which an O-minimal theory T eliminates imaginaries (in the sense of Poizat [P]).

If L is a first order language and M an L-structure, then by a definable set in M, we mean something of the form X ⊂ Mn, n ≥ 1, where X = {(a1…,an) ∈ Mn: M ⊨ϕ(ā)} for some formula ∈ L(M). (Here L(M) means L together with names for the elements of M.) If the parameters from come from a subset A of M, we say that X is A-definable.

M is said to be O-minimal if M = (M, <,…), where < is a dense linear order with no first or last element, and every definable set X ⊂ M is a finite union of points, and intervals (a, b) (where a, b ∈ M ∪ {± ∞}). (This notion is as in [PS] except here we demand the underlying order be dense.) The complete theory T is said to be O-minimal if every model of T is O-minimal. (Note that in [KPS] it is proved that if M is O-minimal, then T = Th(M) is O-minimal.) In the remainder of this section and in §2, M will denote a fixed but arbitrary O-minimal structure. A,B,C,… will denote subsets of M.

We give an alternative proof of the stability of separably closed fields of fixed Éršov invariant to the one given in [W]. We show that in case the Éršov invariant is finite, the theory is in fact equational. We also characterize the independence relation in those theories.

Fred Richman conjectured that the following principle is not constructive:

(*) If A is a decidable subset of the set N of natural numbers and if, for every decidable subset B of N, either A ⊆ B or A ⊆ N − B, then, for some n ∈ N, A ⊆ {n}.

Richman's conjecture was motivated by the problem of uniqueness of divisible hulls of abelian groups in constructive algebra. Richman showed that a countable discrete abelian p-group G has a unique (up to isomorphism over G) divisible hull if the subgroup pG is decidable. He also showed that the converse implies.

We confirm the nonconstructive nature of by showing (in §1) that it is not provable in intuitionistic set theory, IZF. Thus, in the models we construct, there are countable discrete abelian p-groups G whose divisible hulls are unique but whose subgroups pG are not decidable.

Our models do not satisfy further conditions imposed by Richman, namely Church's Thesis and Markov's Principle, so the full conjecture remains an open problem. We do, however, show (in §2) how to embellish our first model so that the fan theorem (i.e., compactness of 2N) fails. (Church's Thesis implies the stronger statement that the negation of the fan theorem holds.)

Our models will be constructed by the method of sheaf semantics [1], [3]. That is, we shall construct Grothendieck topoi in whose internal logic fails.

By a “partly numerical structure” (p.n.s.) we shall here mean a quadruple , where M is a set, ω = the natural numbers, ω ⊆ M, and are disjoint sets, is a set of relations (of various positive integral arities) on M, and is a set of functions (of various positive integral arities) with arguments and values in M. Thus, in calculated disharmony with common practice, we do not (except as noted below, in connection with naming the elements of ω) fix a similarity type as part of our notion of a “structure”. Suppose a finitary first-order language (with identity) has been specified, with constant symbols , n ∈ ω, and with exactly enough relation and function symbols of each arity to enable us to interpret in . We wish to consider the variation in the degree (relative to a fixed Gödel-numbering of ) of the complete -theory of as we vary the way in which elements of ∪ are assigned as interpretations to the relation and function symbols of . We shall in fact, therefore, be concerned exclusively with p.n.s.'s for which ∪ is countable. More: we assume to be such that we can effectively tell, uniformly in n > 0, exactly how many n-ary relations has and exactly how many n-ary functions has.

Myhill [12] extended the ideas of Shapiro [15], and proposed a system of epistemic set theory IST (based on modal S4 logic) in which the meaning of the necessity operator is taken to be the intuitive provability, as formalized in the system itself. In this setting one works in classical logic, and yet it is possible to make distinctions usually associated with intuitionism, e.g. a constructive existential quantifier can be expressed as (∃x) □ …. This was first confirmed when Goodman [7] proved that Shapiro's epistemic first order arithmetic is conservative over intuitionistic first order arithmetic via an extension of Gödel's modal interpretation [6] of intuitionistic logic.

Myhill showed that whenever a sentence □A ∨ □B is provable in IST, then A is provable in IST or B is provable in IST (the disjunction property), and that whenever a sentence ∃x.□A(x) is provable in IST, then so is A(t) for some closed term t (the existence property). He adapted the Friedman slash [4] to epistemic systems.

Goodman [8] used Epistemic Replacement to formulate a ZF-like strengthening of IST, and proved that it was a conservative extension of ZF and that it had the disjunction and existence properties. It was then shown in [13] that a slight extension of Goodman's system with the Epistemic Foundation (ZFER, cf. §1) suffices to interpret intuitionistic ZF set theory with Replacement (ZFIR, [10]). This is obtained by extending Gödel's modal interpretation [6] of intuitionistic logic. ZFER still had the properties of Goodman's system mentioned above.

In this paper I show how to develop stability theory within the context of the topological logic first introduced by McKee [Mc 76], Garavaglia [G 78] and Ziegler [Z 76]. I then study some specific applications to topological modules; in particular I prove two quantifier élimination theorems, one a generalization of a result of Garavaglia.

In the first section I present a summary of basic results on topological model theory, mostly taken from the book of Flum and Ziegler [FZ 80]. This is done primarily to fix notation, but I also introduce the notion of an Lt-elementary substructure. The important point with this concept, as with many others, appears to be to allow only individuals to appear as parameters, not open sets.

In the second section I begin the study of stability theory for Lt. I first develop a translation of the topological language Lt into an ordinary first-order language L*. The first main theorem is (2.3), which shows that the translation is faithful to the model-theoretic content of Lt, and provides the necessary tools for studying Lt theories in the context of ordinary first-order logic. The translation allows me to consider individual stability theory for Lt: the stability-theoretic study of those types of Lt in which only individual variables occur freely and in which only individuals occur as parameters. I originally developed this stability theory entirely within Lt; the fact that the theorems and their proofs were virtually identical to those in ordinary first order logic suggested the reduction from Lt to L*.