RVM, RVC revisited: Clubs and Lusin sets

Similar documents
ADDING A LOT OF COHEN REALS BY ADDING A FEW II. 1. Introduction

Tall, Strong, and Strongly Compact Cardinals

Extender based forcings, fresh sets and Aronszajn trees

LARGE CARDINALS AND L-LIKE UNIVERSES

Interpolation of κ-compactness and PCF

Sy D. Friedman. August 28, 2001

DEPTH OF BOOLEAN ALGEBRAS SHIMON GARTI AND SAHARON SHELAH

Generalization by Collapse

Silver type theorems for collapses.

A precipitous club guessing ideal on ω 1

COMBINATORICS AT ℵ ω

The Semi-Weak Square Principle

Level by Level Inequivalence, Strong Compactness, and GCH

SHORT EXTENDER FORCING

HEIKE MILDENBERGER AND SAHARON SHELAH

STRONGLY UNFOLDABLE CARDINALS MADE INDESTRUCTIBLE

GUESSING MODELS IMPLY THE SINGULAR CARDINAL HYPOTHESIS arxiv: v1 [math.lo] 25 Mar 2019

Strongly compact Magidor forcing.

Annals of Pure and Applied Logic

SHIMON GARTI AND SAHARON SHELAH

Axiomatization of generic extensions by homogeneous partial orderings

THE NUMBER OF UNARY CLONES CONTAINING THE PERMUTATIONS ON AN INFINITE SET

MITCHELL S THEOREM REVISITED. Contents

arxiv:math/ v1 [math.lo] 15 Jan 1991

Global singularization and the failure of SCH

being saturated Lemma 0.2 Suppose V = L[E]. Every Woodin cardinal is Woodin with.

Chromatic number of infinite graphs

A Laver-like indestructibility for hypermeasurable cardinals

arxiv: v2 [math.lo] 13 Feb 2014

Währinger Strasse 25, 1090 Vienna Austria

Characterizing large cardinals in terms of layered partial orders

January 28, 2013 EASTON S THEOREM FOR RAMSEY AND STRONGLY RAMSEY CARDINALS

Easton s theorem and large cardinals from the optimal hypothesis

ARONSZAJN TREES AND THE SUCCESSORS OF A SINGULAR CARDINAL. 1. Introduction

Large Cardinals with Few Measures

CONSECUTIVE SINGULAR CARDINALS AND THE CONTINUUM FUNCTION

Chain conditions, layered partial orders and weak compactness

A relative of the approachability ideal, diamond and non-saturation

PARTITIONS OF 2 ω AND COMPLETELY ULTRAMETRIZABLE SPACES

ON THE SINGULAR CARDINALS. A combinatorial principle of great importance in set theory is the Global principle of Jensen [6]:

The (λ, κ)-fn and the order theory of bases in boolean algebras

arxiv:math/ v1 [math.lo] 9 Dec 2006

On the strengths and weaknesses of weak squares

SUCCESSIVE FAILURES OF APPROACHABILITY

PERFECT TREE FORCINGS FOR SINGULAR CARDINALS

Notes to The Resurrection Axioms

The first author was supported by FWF Project P23316-N13.

Cardinal arithmetic: The Silver and Galvin-Hajnal Theorems

MODIFIED EXTENDER BASED FORCING

Two Stationary Sets with Different Gaps of the Power Function

COLLAPSING SUCCESSORS OF SINGULARS

On the Splitting Number at Regular Cardinals

On almost precipitous ideals.

COMBINATORICS OF REDUCTIONS BETWEEN EQUIVALENCE RELATIONS

A HIERARCHY OF RAMSEY-LIKE CARDINALS

Strongly Unfoldable Cardinals Made Indestructible

LECTURE NOTES - ADVANCED TOPICS IN MATHEMATICAL LOGIC

On Singular Stationarity II (tight stationarity and extenders-based methods)

arxiv: v1 [math.lo] 9 Mar 2015

INDESTRUCTIBLE STRONG UNFOLDABILITY

2. The ultrapower construction

NORMAL MEASURES ON A TALL CARDINAL. 1. Introduction We start by recalling the definitions of some large cardinal properties.

Generalising the weak compactness of ω

Large cardinals and their effect on the continuum function on regular cardinals

arxiv: v1 [math.lo] 24 May 2009

Cardinal characteristics at κ in a small u(κ) model

Notes on getting presaturation from collapsing a Woodin cardinal

More on the Pressing Down Game

arxiv: v2 [math.lo] 21 Mar 2016

THE TREE PROPERTY UP TO ℵ ω+1

FORCING AND THE HALPERN-LÄUCHLI THEOREM. 1. Introduction This document is a continuation of [1]. It is intended to be part of a larger paper.

EASTON FUNCTIONS AND SUPERCOMPACTNESS

Continuous images of closed sets in generalized Baire spaces ESI Workshop: Forcing and Large Cardinals

UPWARD STABILITY TRANSFER FOR TAME ABSTRACT ELEMENTARY CLASSES

Determinacy models and good scales at singular cardinals

arxiv: v1 [math.lo] 26 Mar 2014

Generic embeddings associated to an indestructibly weakly compact cardinal

SOME CONSEQUENCES OF REFLECTION ON THE APPROACHABILITY IDEAL

arxiv: v1 [math.lo] 8 Oct 2015

CARDINALITIES OF RESIDUE FIELDS OF NOETHERIAN INTEGRAL DOMAINS

Chapter 4. Cardinal Arithmetic.

arxiv: v3 [math.lo] 23 Jul 2018

Short Extenders Forcings II

Large cardinals and the Continuum Hypothesis

Philipp Moritz Lücke

Bounds on coloring numbers

On Singular Stationarity I (mutual stationarity and ideal-based methods)

Fat subsets of P kappa (lambda)

ANNALES ACADEMIÆ SCIENTIARUM FENNICÆ DIAMONDS ON LARGE CARDINALS

DIAGONAL PRIKRY EXTENSIONS

Combinatorics, Cardinal Characteristics of the Continuum, and the Colouring Calculus

On almost precipitous ideals.

LOCAL CLUB CONDENSATION AND L-LIKENESS

arxiv:math/ v2 [math.lo] 17 Feb 2007

BLOWING UP POWER OF A SINGULAR CARDINAL WIDER GAPS

arxiv: v1 [math.lo] 27 Mar 2009

Covering properties of derived models

The Outer Model Programme

ON SCH AND THE APPROACHABILITY PROPERTY

THE TREE PROPERTY AT ALL REGULAR EVEN CARDINALS

Transcription:

RVM, RVC revisited: Clubs and Lusin sets Ashutosh Kumar, Saharon Shelah Abstract A cardinal κ is Cohen measurable (RVC) if for some κ-additive ideal I over κ, P(κ)/I is forcing isomorphic to adding λ Cohen reals for some λ. Such cardinals can be obtained by starting with a measurable cardinal κ and adding at least κ Cohen reals. We construct various models of RVC having different properties than this model. Our main results are: (1) κ = 2 ℵ 0 is RVC does not decide S for various stationary S κ. (2) κ is RVC and κ < λ = cf(λ) < 2 ℵ 0 does not decide S for various stationary S λ. (3) κ = 2 ℵ 0 is RVC does not decide the existence of a Lusin set of size κ. We also prove analogues of (1), (2) for real valued measurable cardinals. 1 Introduction A cardinal κ is (atomlessly) real valued measurable (RVM) if for some κ-additive ideal I over κ, P(κ)/I is forcing isomorphic to adding λ random reals for some λ. In [13], Solovay showed that if we start with a measurable cardinal κ and add κ random reals then in the resulting model κ is RVM. The category analogue of RVM was introduced in [5]: A cardinal κ is Cohen measurable (RVC) if for some κ-additive ideal I over κ, P(κ)/I is forcing isomorphic to adding λ Cohen reals for some λ. If we start with a measurable cardinal κ and add κ Cohen reals, then in the resulting model κ is RVC. In [5] it was shown that 2 ℵ 0 is RVC does not decide the statement: For every function f : R R, there is a non meager set X R such that f X i s continuous. We continue this line of investigation by constructing models of RVC with different properties than Prikry s model. In Sections 2 and 3, we mainly deal with the existence of club guessing sequences. In Section 4, we consider the existence of Lusin sets. 2010 Mathematics Subject Classification: 03E35, 03E55. Key words: Forcing Einstein Institute of Mathematics, The Hebrew University of Jerusalem, Edmond J Safra Campus, Givat Ram, Jerusalem 91904, Israel; email: akumar@math.huji.ac.il; Supported by a Postdoctoral Fellowship at the Einstein Insititute of Mathematics funded by European Research Council grant 338821 Einstein Institute of Mathematics, The Hebrew University of Jerusalem, Edmond J Safra Campus, Givat Ram, Jerusalem 91904, Israel and Department of Mathematics, Rutgers, The State University of New Jersey, Hill Center-Busch Campus, 110 Frelinghuysen Road, Piscataway, NJ 08854-8019, USA; email: shelah@math.huji.ac.il; Partially supported by European Research Council grant 338821. Paper no. 1046. 1

1.1 On notation For λ < κ = cf(κ), Sλ κ = {δ < κ : cf(δ) = cf(λ)}. In forcing we use the convention that a larger condition carries more information - So p q means q extends p. For a forcing notion P, G P is the P name for the generic added by P. For a regular infinite cardinal κ, H κ is the family of sets whose transitive closure has size < κ. An uncountable regular cardinal κ is called quasi measurable if there exists a κ-additive ℵ 1 -saturated ideal over κ. For a regular uncountable cardinal κ, N S κ denotes the non-stationary ideal over κ. If S κ is stationary, the non-stationary ideal restricted to S is defined by N S κ S = {X κ : X S N S κ }. For a set X, Leb = Leb X denotes the Lebesgue measure on 2 X ; we may drop X when cle ar from the context. 1.2 Review of diamonds and clubs For a regular uncountable cardinal κ and a stationary S κ of limit ordinals, (a) S is the statement: There exists a sequence A α : α S such that each A α α and for every A κ, the set {α S : A α = A α} is stationary. (b) S is the statement: There exists a sequence A α : α S such that each A α [P(α)] α and for every A κ, the set {α S : A α / A α } is non-stationary. (c) S is the statement: There exists a sequence A α : α S such that each A α is an unbounded subset of α and for every unbounded X κ, the set {α S : A α X} is non empty (equivalently stationary). We need the following facts about diamonds and clubs. Fact 1.1. For κ, S as above, S is equivalent to S + κ <κ = κ. Fact 1.2. Let κ be a successor cardinal and S κ be stationary. Assume S. Then for every stationary T S, T holds. Fact 1.3. Let κ, S be as above. Assume S, P is a κ-cc poset and P κ. Then V P S. Furthermore, if V = κ, so does V P. Proof: Using P κ, nice P-names for subsets of κ (union of sets of form W {α} where α < κ and W is an antichain in P) can be thought of as subsets of κ by identifying P κ with κ. If A κ codes such a nice name, then let n(a) denote the corresponding P-name. Fix a S witnessing sequence A α : α S in V. In V [G], define B α : α S by B α = eval G (n(a α )) α if A α codes a nice name for a subset of κ, otherwise B α = 0. Fix a code A for a nice P-name n(a) for a subset of κ. Then for each α < κ, A α also codes a nice name. Moreover, using κ-ccness of P, the set {α < κ : eval G (n(a α)) α = eval G (n(a)) α} contain a club. Since A α = A α on a stationary subset of S, B α = eval G (n(a)) α on a stationary subset as well. Hence B α : α S witnesses S in V [G]. The preservation of κ has a similar proof. Fact 1.4. Suppose κ is regular uncountable and for every stationary S κ, S holds. Let P add κ Cohen reals. Then V P = S for every stationary S κ. 2

Proof: Let p S κ is stationary. Put S 1 = {δ < κ : ( p δ p )(p δ δ S)}. So S 1 is stationary. For δ S 1, choose p δ p such that p δ δ S.. By Fodor s lemma, we can find a stationary S 2 S 1 such that {p δ : δ S 2 } forms a delta system with root p p. Note that for every unbounded X S 2, p X S φ. So using S2, we can proceed as in the proof of Fact 1.3. Fact 1.5 (Shelah). Suppose κ = κ ℵ 0 is regular uncountable and S κ is stationary. Suppose that for every S S, stationary in κ, S holds. (1) Let P add κ random reals. Then, in V P, for every stationary S S, S holds. (2) Let Q add κ Cohen reals. Then, in V Q, for every stationary S S, S holds. (3) Suppose ℵ 0 < = cf(), κ < = κ. Let R be a -c.c. forcing of size κ. Then, in V R, for every stationary S S, S holds. Proof: (1) Suppose p S S is stationary in κ. Let S 1 = {δ < κ : δ limit ordinal and ( p δ p )(p δ δ S)}. So S 1 is stationary. For δ S 1, choose p δ p such that p δ δ S. Claim 1.6. There exists p p such that for every p p, {δ S 1 : p, p δ are compatible} is stationary. Proof: If not, let X be a maximal antichain above p such that for every q X, S 1,q = {δ S 1 : q, p δ are compatible} is non-stationary. Since X is countable we can choose δ S 1 \ q X S 1,q. But now p δ is compatible with some q X which is impossible. Claim 1.7. For every stationary S S, there are κ + many stationary subsets of S whose pairwise intersections are bounded below κ. Hence the non-stationary ideal over κ restricted to any stationary subset of S is not κ + -saturated. Proof: Fix a S sequence C δ : δ S. Let {A i : i < κ} be a partition of κ into sets of size κ. Inductively construct {A i : κ < i < κ + } such that each A i [κ] κ and for every i < j < κ +, A i A j < κ. For each i < κ +, let S i = {δ S : C δ A i }. Then {S i : i < κ + } is as required. Claim 1.8. Let Y = {q P : q p }. For q Y, let S 2,q = {δ S 1 : q, p δ are compatible}. There exists a family S 3,q : q Y such that each S 3,q S 2,q is stationary and if q 1 q 2 Y then S 3,q1 S 3,q2 = φ. Proof: Since P = κ ℵ 0 = κ, Y κ. By previous claim, N S κ S is nowhere κ + - saturated. Hence by Theorem 1.2 in [1], we can find a disjoint refinement for any family of size κ. Let Λ be the family of all triplets (p, α, p) where p P, α = α k : k < ω, α k < κ and p = p k : k < ω is a maximal antichain of P above p. Since κ ℵ 0 = κ, we can get a list (p i, α i, p i ) : i < κ in which each member of Λ occurs κ times. For each q Y, let C q δ : δ S 3,q be a S3,q witnessing sequence. We can assume that C δ,q is an unbounded subset of δ of order type cf(δ). Put S 3 = {S 3,q : q Y } and C = C δ : δ S 3 = q Y Cq δ : δ S 3,q. For q Y and δ S 3,q, let D δ = {α i,k : i C δ and ( k < ω)(p i,k G P )}. Let S 4 = {δ S 3 S : D δ is an unbounded subset of δ of order type cf(δ)}. It is enough to show the following. 3

Claim 1.9. Suppose q p and q Å [κ]κ. Then for some q q, δ < κ, q (δ S 4 and D δ Å). Proof: Note that since q p, for every δ S 3,q, q, p δ are compatible. Let β j : j < κ be defined by β j = min(å\j). For each j < κ, choose (β j, q j ) such that q j = q j,k : k < ω is a maximal antichain above q, β j = β j,k : k < ω and q j,k β j = β j,k. Let f(j) be the least i [j, κ) such that (q, β j, q j ) = (p i, α i, p i ). Let E κ be a club such that for each δ E, j < δ and k < ω, f(j) < δ and α j,k < δ. Let A = {f(j) : j < κ}. Since C δ : δ S 3,q is a S3,q witnessing sequence, we can choose δ E S 3,q such that C δ A. As p δ, q are compatible, we can choose q p δ, q. Let G be P-generic over V with q G. So δ S = S[G]. Since δ E, D δ δ. As C δ is unbounded in δ and β j j, D δ is also unbounded in δ. Hence δ S 4. Finally note that since C δ A, each member of D δ is of the form α f(j),k where p f(j),k G. As q f(j),k β f(j),k = β j and p f(j),k = q f(j),k, α f(j),k = β f(j),k, we get α f(j),k A. (2) Follows from (3). (3) This has essentially the same proof as (1). In the construction of D δ, we use a modified version of Λ where we consider sequences of length <. The assumptions κ < = κ, R satisfies -c.c. ensure that Λ κ. Fact 1.10. Let ω σ = cf(σ) < κ = cf(κ). Suppose P satisfies σ + -c.c., S κ is stationary and V P S. Then V S. Proof: Let p Åα : α S is a S witnessing sequence. For each α S, choose B α [P(α)] σ such that the following holds: If for some q p and B α V, q Åα = B α, then B α B α. It is easy to see that, for every B κ, for stationary many α S, B α B α. Let f : κ σ κ be a bijection. Let C κ be a club such that for every α C, f α : α σ α is a bijection. For α C S, let D α = {f[b] : B B α }. Note that for every D σ κ, there are stationary many α C S such that D (σ α) D α. Enumerate D α = {D α,i : i < σ}. Fo r j < σ, let D α,i,j = {β : (j, β) D α,i }. We claim that for some i < σ, D α,i,i : α C S witnesses S. Suppose not and for each i < σ, let D i κ be such that {α C S : D i α D α,i,i } is non-stationary. Put D = {(i, α) : α D i }. But then {α C S : D (σ α) D α } is non-stationary: A contradiction. Fact 1.11 (Kunen). Suppose κ 2 ℵ 0 is quasi measurable with I a witnessing normal ideal. Then, for every I-positive S κ, S holds. Note that the assumption κ 2 ℵ 0 cannot be dropped even for getting κ. For example, if we start with κ measurable and add κ + Cohen reals then in the resulting model κ is quasi measurable (in fact, κ is RVC) but κ fails. Question 1.12. Suppose κ < 2 ℵ 0 is quasi measurable. Can κ hold? We intend to deal with this in a forthcoming work. 4

2 Clubs: When continuum is RVC/RVM Lemma 2.1. It is consistent that κ = 2 ℵ 0 is RVC and for every stationary S κ, S holds. It is also consistent that κ = 2 ℵ 0 is RVM and for every stationary S κ, S holds. Proof: Start with κ measurable and V = L[I] where I is a witnessing normal prime ideal over κ. Then for every stationary S κ, S holds. Now add κ Cohen/random reals and use Fact 1.5. We next show that κ = 2 ℵ 0 is RVC/RVM and S (equivalently, S ) for various stationary S κ is consistent. Suppose λ is inaccessible and GCH holds. Let ω = cf() < λ and S S λ be a stationary set of singular cardinals in λ. Let Q = Q S be the forcing whose conditions are p 2 <λ where p 1 [{1}] S is nowhere reflecting, ordered by end extension. Let S 1 S be the generic added by Q 1. In V [S 1 ], let (A, ρ) = A i, ρ i : i S 1 satisfy: A i is an unbounded subset of i of order type, ρ i : A i 2. Let R = R (A,ρ) V [S 1 ] be the forcing to uniformize (A, ρ). So p R iff p = (β, E, g) where β < λ, E β + 1 is a club and g : β 2 such that if i S 1 E, then ρ i g where means inclusion modulo ρ i restricted to a bounded subset of A i. The orderin g is (β, E, g) (β, E, g ) iff β β, E is an end extension of E and g g. Claim 2.2. Q R is λ-strategically closed. In fact, it has a dense λ-closed subset. Proof: The completeness player always plays conditions of the form (p, q) where q = (β, E, g) and β dom(p). Definition 2.3. For λ,, S as above, let P = P i, Q j : i λ +, j < λ + be an iteration with < λ support such that the following hold. If j is odd, in V P j, Q j = Q S for some stationary S S λ, ω = cf() < λ. Let S j be a P j+1 name for the generic subset of S added by Q j If j is even, in V P j, Q j = R (A,ρ) where for some j < j, (A, ρ) is a uniformizing candidate on S j Through usual bookkeeping, we ensure that every stationary set of singular cardinals and every uniformizing candidate is taken care of at some stage before κ + Claim 2.4. Let P be as in Definition 2.3. Then P i satisfies λ + -c.c. and is λ-strategically closed for every i λ +. In V P λ +, every stationary set S λ consisting of singular cardinals has a stationary subsets S such that S does not hold. Proof: The main point is that the stationarity of S j V P j+1 is preserved during the later stages of the iteration. For a proof of this, see [8] and more in [9]. Theorem 2.5. Assume GCH. Suppose κ is κ + -supercompact. Then there is a forcing P such that in V P the following hold. 5

1. κ is measurable, 2. If S κ is stationary consisting of singular cardinals, then for some stationary S 1 S, S1 does not hold. Proof: Let P α, Q β : α κ + 1, β κ be an Easton support iteration where If α is not inaccessible, Q α is trivial If α is inaccessible, Q α V Pα is the iteration of Definition 2.3, with λ = α. It is easily checked that all regular cardinals are preserved in V P κ+1. GCH holds as well. By Claim 2.4, in V P κ+1, every stationary S κ consisting of singular cardinals has a stationary subset S 1 such that S1 does not hold. To finish, we show that κ remains measurable in V P κ+1. Let j : V M be an elementary embedding witnessing the κ + - supercompactness of κ. So crit(j) = κ, κ ++ < j(κ) < κ +++ and M κ+ M. In M, we can write j(p κ ) = P κ Q κ R (as Q κ M). Let G be P κ -generic over V and let H be Q κ -generic over V [G]. Let R = R[G][H]. Working in V [G][H], we would like to extend j to V [G][H]. We do this in two steps. Claim 2.6. In V [G][H], there exists an R-generic filter K over M[G][H] and an elementary extension j 1 : V [G] M[G][H][K] of j. Proof: As P κ+1 has κ + -c.c., every nice name for a sequence of ordinals of length κ + is in M. Hence (M[G][H]) κ+ V [G][H] M[G][H]. It follows that, in V [G][H], R is κ ++ - strategically closed since M[G][H] = R is κ +++ -strategically closed. So in V [G][H] the completeness player can play for κ ++ moves using this strategy. This works because every initial segment of a play is in M[G][H]. Now in M[G][H], R has j(κ) maximal antichains. In V [G][H], we can list them in order type κ ++ and the use the κ ++ -strategic closure of R to construct a filter K R meeting all of them. Finally since j[g] G H K, we can lift j to j 1 : V [G] M[G][H][K]. Now we would like to extend j 1 to V [G][H]. In M, let j(p κ+1 ) = P κ Q κ R S. Claim 2.7. In V [G][H], there exists an S-generic filter L over M[G][H][K] and an elementary extension j 2 : V [G][H] M[G][H][K][L] of j 1. Proof: Since V [G] = Q κ is κ-strategically closed and has κ + maximal antichains, by elementarity, M[G][H][K] = S is j(κ)-strategically closed and has j(κ + ) maximal antichains. Hence as before, in V [G][H], S is κ ++ -strategically closed so that we can build an S-generic filter over M[G][H][K] above a master condition which we now define as follows. In V [G][H], consider the pointwise image j 1 [H]. Since H can be thought of as a κ + -sequence of subsets of κ, there is a condition q S such that every member of j[h] is below q. So we construct an S-generic filter L over M[G][H][K] with q L. It is now easy to lift j 1 to j 2 : V [G][H] M[G][H][K][L]. Corollary 2.8. It is consistent that κ = 2 ℵ 0 is RVC (resp. RVM) and for every stationary S 1 κ consisting of singular cardinals, there exists stationary S S 1 such that S. Proof: Start with κ measurable, P as above and let Q add κ Cohen reals (resp. random reals). Use Fact 1.10. 6

3 Clubs: When 2 ℵ 0 > κ and κ is RVC/RVM Lemma 3.1. Suppose κ is RVC or RVM and ω 1 = cf() < κ. Then. Proof: Let I be an RVC (resp. RVM) witnessing ideal over κ. Force with I and form the generic ultrapower M V [G]. Let j : V M be the generic elementary embedding. By [4], G is generic for adding > κ Cohen reals (resp. random reals) over V so fails in V [G] hence also in M as M κ V [G] M. By elementarity, this also holds in V. Lemma 3.2. Suppose κ is measurable and λ = λ ℵ 0 > κ. Let P be the forcing for adding λ Cohen reals (resp. random reals). Then, in V P, κ is RVC (resp. RVM), 2 ℵ 0 = λ and for every uncountable regular < λ, we have. Proof: Use the Cohen/random sequence to outguess every potential club guessing sequence. So we can ask the following Question 3.3. Can we force S for various stationary S where κ < 2 ℵ 0 with κ is RVC (resp. RVM)? together This is addressed in the next two sections. 3.1 Some club preserving forcings On the following forcing, see [2], and for a more general treatment [11]. Definition 3.4. For ω σ κ = cf(κ) < λ, we define Q = Q λ,κ,σ as follows: (A) p Q iff p = (u, f) = (u p, f p ) where u [λ] <κ and f : u 2 <σ (B) For p, q Q, p q iff for some r Q, p pr r (read r is a pure extension of p ) and r apr q (read r is a apure extension of p ), where (C) p pr q iff u p u q and f p = f q u p (D) p apr q iff u p = u q = u and for every α u, f p (α) f q (α) and {α u : f p (α) f q (α)} < σ (E) For p Q, we write Q[p] for the forcing whose conditions are apr -extensions of p ordered under Q (here it is same as apr ) Lemma 3.5. For σ, κ, λ and Q = Q λ,κ,σ as in Definition 3.4, the following hold. (1) If κ <κ = κ, then Q satisfies κ + -c.c. (even κ + -Knaster) (2) Any pr -increasing sequence of length < κ has a least upper bound (3) Q is < σ-directed complete (4) If σ <σ = σ and p Q, then Q[p] satisfies σ + -c.c. and is σ-closed 7

(5) Assume σ = σ <σ, σ + < κ. If p Q τ V, then for some q Q, p pr q the set {r Q : q apr r r forces a value to τ} is predense above q (6) Assume σ = σ <σ, σ + < κ. If < κ and p Q τ : V, then for some q Q, p pr q and for every i <, the set {r Q : q apr r r forces a value to τ(i)} is predense above q (7) If σ <σ = σ and κ is inaccessible, then Q preserves all regular cardinals. Proof: Clauses (1)-(4) are easy to check. For (5), suppose p Q τ V. We try to construct (p i, q i ) : i < σ + such that p 0 = p and p i s are pr -increasing continuous for every i < σ +, p i+1 apr q i q i s are pairwise incompatible q i forces a value to τ Suppose we can complete the construction. Let p σ + be the union of p i s for i < σ +. Let r i be the least upper bound of p σ +, q i. Then r i s are pairwise incompatible apr extensions of p σ + which contradicts (4). So the construction cannot be carried out at some stage i < σ +. If i is limit, let q = p i be the union of p j s for j < i. Let r q be any condition deciding τ. If possible suppose r is incompatible with every r j for some j < i (r j is the least upper bound of q, q j ). Then we can set q i = r u q, p i+1 = q r (u r \u q ) contradicting that the construction could not be carried out at stage i. The case when i is a successor is similar. Clause (6) has a similar proof. For clause (7), note that if σ = κ, Q is < κ-closed and has κ + -c.c. so this is clear. So assume σ < κ. Note that all regular car dinal κ + or σ are preserved. Suppose σ < δ κ be regular cardinals and p Q τ : δ. Using clause (6), get q pr p as there. For each i <, choose a maximal antichain in Q[q] deciding τ(i) and let α < δ be the greater than all these values. Then q Q range(τ) α. Lemma 3.6. Let σ = σ <σ κ = κ <κ < λ and Q = Q λ,κ,σ. Let < κ be an infinite regular cardinal. (1) If S S κ+ is stationary in κ + and S holds as witnessed by B = B i : i S, with B i, then V Q = S with the same witness. (2) If λ 1 = cf(λ 1 ) > κ and α < λ 1 = α <κ < λ 1, S S λ 1 is stationary in λ 1 and S holds as witnessed by B = B i : i S, with B i, then V Q = S with the same witness. (3) Assume 2 κ = κ + and let S S κ+ <κ be stationary in κ +. Then S holds in both V and V Q. (4) Suppose 2 κ = κ +, S Sκ κ+ is stationary and S holds. Then in V Q, S holds iff λ = κ +. Moreover, if S holds for every stationary S Sκ κ+ and λ = κ +, then in V Q, S holds for every stationary S Sκ κ+ 8

Proof: (1) Follows from (2). (2) Suppose p Q Å is an unbounded subset of λ 1. For each α < λ 1, let p α, γ α be such that (a) p p α (b) γ α < λ 1 and α < β = γ α < γ β (c) p α Q γ α Å By the assumption on λ 1, using delta system lemma, we can find an unbounded W λ 1 such that {u pα : α W } forms a delta system with root u and p α u does not depend on α W. Since {γ α : α W } is an unbounded subset of λ 1, we can find some δ S such that B δ {γ α : α W }. Since B δ, letting q to be the lub of {p α : α B δ }, q Q B δ Å. (3) For some < κ, S = S κ+ S is stationary in κ +. By [10], S holds in V. Now use (1). (4) If λ > κ +, we can diagonalize against any potential S sequence. When λ = κ +, Q satisfies κ + -c.c. and Q κ +. Since 2 κ = κ +, S holds in V so we can use Fact 1.3. Next suppose λ = κ + and S holds for every stationary S Sκ κ+ and let p S Sκ κ+ is stationary. Let S 1 = {α Sκ κ+ : ( p α p)(p α α S)}. By Fodor s lemma, we can find stationary S 2 S 1, such that {p α : α S 2 } forms a delta system with root p p. Note that every unbounded subset of S 2 is forced by p to meet S. Since 2 κ = κ +, S2 holds in V. So we can proceed as in the proof of Fact 1.3. Lemma 3.7. Suppose κ is inaccessible, λ > κ σ = σ <σ, S κ is stationary and S holds as witnessed by B = B i : i S. Let Q = Q λ,κ,σ. Then V Q = S holds with witness B. Proof: This is clear if σ = κ since Q is < κ-closed. So assume σ < κ. Suppose p Å [κ]κ. Construct (p i, u i, q i, α i, F i ) : i < κ such that the following hold 1. For each i < j < κ, α i < α j 2. p 0 = p and p i s are pr -increasing continuous 3. p i apr q i, u i = dom(p i ) 4. every common extension of p i+1 and q i forces α i Å 5. {β u i : p i (β) q i (β)} = F i For i S κ σ, let h(i) be the least j < i such that F i u j. Then h is constant on some some stationary set W. By thinning W further, we can assume that for some F and q : F 2 <σ, we have F i = F and q i F = q for every i W. Now choose δ S such that B δ {α i : i W }. Let j W be larger than sup{i < κ : α i B δ } and q = {q i : i W, i j}. Then q B δ Å. 9

3.2 Getting clubs Theorem 3.8. Suppose κ < µ are measurable, µ remains measurable in any < µ-directed closed forcing extension 1 and 2 µ = µ + < λ = λ µ. Let S 0 = S µ µ+. Then for some forcing P, in V P we have (a) No cardinal > µ is collapsed (b) cf(µ) = ℵ 0 (c) 2 ℵ 0 = λ (d) κ is RVC (e) For every regular < µ +, S µ + holds. (f) Moreover, for every stationary S µ + \S 0, S holds. Proof: Remark 3.9. Can we get S for all stationary subsets of µ +? We intend to do this in a subsequent work both for RVC and RVM (see Corollary 3.24 below). Claim 3.10. For some < µ + -directed closed, µ ++ -c.c. forcing P 1 V, letting V 1 = V P 1, we have V 1 = µ +. Proof: See [6] Chapter 7, H(20). Claim 3.11. Let P 2 = Q λ,µ,µ V 1. Put V 2 = V P 2 1. Then, in V 2, 2 µ = λ and for every < µ, holds. S µ + Proof: Note that P 2 is < µ-directed closed so κ, µ remain measurable. Also, 2 µ = λ is easily verified. For S µ +, we use Lemma 3.6. Claim 3.12. In V 2, let P 3 be Prikry forcing w.r.t. some normal ultrafilter D over µ. Then in V 3 = V P 3 2, κ is measurable, µ is a strong limit of countable cofinality, 2 µ = λ and for every regular < µ +, S µ + holds. Proof: We only check the club principle. If ℵ 0 < < µ, then S = (S µ+ ) V 2 = (S µ+ ) V 3. If = ℵ 0, note that S = (S µ+ ) V 2 (S µ+ ) V 3. Hence it is enough to show V 3 = S. In V 2, fix B = B i : i S witnessing S there. Let p P4 Å µ + is unbounded. Construct (p i, γ i ) : i < µ + such that for i < j, γ i < γ j, p p i and p i P4 γ i Å. We can assume that each p i = (s, X i ), s [µ] <ℵ 0, X i D. Let W = {γ i : i < µ + }. Since W is unbounded in µ +, we can choose δ S such that B δ W. Let p = (s, X) where X = {X i : γ i B δ }. Then p P4 B δ Å. Claim 3.13. In V 3, let P 4 add µ Cohen reals. Then in V 4 = V P 4 3, we have 1 See [3] for examples. 10

(1) κ < µ < λ = 2 ℵ 0 (2) κ is RVC (3) For every regular < µ +, S µ + holds Proof: In V 4, since µ is a strong limit of countable cofinality, µ ℵ 0 = 2 µ = λ. Hence V 4 = 2 ℵ 0 = λ. It is clear that κ is RVC in V 4. Also S µ + is preserved by the same witness because P 4 = µ. We next show clause (f). Lemma 3.14. If (A), then (B) where (A) (a) µ is strongly inaccessible (b) 2 µ = µ + < λ = λ µ and µ + (c) S 0 = S µ+ µ (d) P = Q 0 Q 1 Q 2 where holds (i) Q 0 = Q λ,µ,µ (ii) D is a Q 0 -name such that V Q 0 = D is a normal ultrafilter over µ (iii) Q 1 V Q 0 is Prikry forcing w.r.t. D (iv) Q 2 is the forcing for adding µ Cohen reals (B) V P = S holds for every stationary S µ + \S 0. Proof: Suppose p S µ + \S 0 is stationary. Note that the set of conditions p = (p 0, p 1, p 2 ) P where p 2 is an actual finite partial function from µ to 2 and p 1 = (w, Å) where w is an actual finite subset of µ is dense in P. Let us call such conditions good and assume, WLOG, that p is good. Since P satisfies µ + -c.c., for each α < µ +, we can find a maximal antichain of good conditions p α,i : i u α µ above p deciding α S. Let p α,i = (p 0 α,i, (w α,i, Åα,i), f α,i ). Let u [λ] µ+ be such that for every α < µ +, i u α, p 0 α,i Q u,µ,µ and Åα,i is a Q u,µ,µ-name. Let G be Q u,µ,µ-generic over V with p 0 G. By Fact??, µ still holds in V [G]. Now (Q + λ,µ,µ ) V [G] = Q 0 /G, so we might as well assume that p 0 = φ and for each α < µ +, i u α, p α,i = (φ, (w α,i, A α,i ), f α,i ). Claim 3.15. There exist j, w, f,, E such that (a) j < µ, w [µ] <ℵ 0, f is finite partial function from µ to 2, < µ is regular infinite, E µ + is a club (b) Letting S 1 = {δ < µ + : cf(δ) =, j u δ, w δ,j = w, f δ,j = f and p δ,j δ S}, and S 2 = {δ S 1 : p δ,j G P }, we have, for every δ E S 1, p δ,j S 2 is stationary. 11

Proof: We can easily choose j, w, f, such that S 1 is stationary. Next we can find a club E µ + such that for every δ E S 1, w [A δ,j ] <ℵ 0, the set {α S 1 : w A α,j } is stationary. Let δ S 1 E and suppose p δ,j S 2 is stationary. Hence there is a good q = (q 0, (w, Å), q2 ) p δ,j and a club E µ + (since P satisfies µ + -c.c.) such that q S 2 E = φ. Let δ S 1 E E be such that w A δ,j w. But now q and p δ,j are compatible and p δ,j δ S 2 E - A contradiction. For every w w [µ] <ℵ 0, let S 1,w = {δ S 1 : w\w A δ,j }. Let W = {w : w w [µ] <ℵ 0 and S 1,w is stationary in µ + }. Let E 1 E be a club in µ + such that for every w w [µ] <ℵ 0, if S 1,w is not stationary, then E 1 S 1,w = φ. Claim 3.16. There exists S 3,w : w W such that (1) S 3,w S 1,w is stationary (2) w w W implies S 3,w S 3,w = φ Proof: Using Theorem 1.2 in [1]. For each w W, choose a w = a δ : δ S 3,w witnessing S3,w. Let a = a δ : δ E 1 S 1. It suffices to show Claim 3.17. For each δ S 1 E 1, p δ,j a S witnesses S. Proof: Suppose not and let δ 1 S 1 E 1, q p δ1,j, Å be such that q is good and q Å µ+ is unbounded and for every δ S 2, a δ Å. Let (q α, γ α ) : α < µ + be such that for every α < µ +, q q α = (qα, 0 (v α, B α ), qα) 2 is good and q α γ α is the α-th member of Å. Let E 3 = {δ < µ + : δ is limit and α < δ = γ α < δ}. By Fodor s lemma, for some stationary S 3 S µ µ+ E 3, qα 0 : α S forms a delta system with root q 0 and v α = v, qα 2 = q 2 do not depend on α S 3. As q α q p δ1,j and δ 1 S 1 E 1, the set S 1,v is stationary. Hence S 3,v is also stationary so we can find δ 2 E 3 S 3,v such that a δ2 {γ α : α S 3 }. But now recalling the definition of S 1,v, the set {q α : γ α a δ2 } {p δ2,j } has an upper bound. This completes the proof of Theorem 3.8. What happens when we add µ random reals to V 3? In this case, we get a weak version of S which we now describe. Definition 3.18. For ℵ 0 < κ = cf(κ) and S κ stationary, the principle inf S says the following: There exists A δ : δ S such that each A δ is an unbounded subset of δ of order type cf(δ) and for every A [κ] κ, for every ε > 0, for stationary many (equivalently, some) δ S, for every α A δ, lim inf n { Aδ A [α, β] n } : α < β < δ n = A δ [α, β] 1 ε Remark 3.19. Suppose κ < λ = λ ℵ 0, κ is measurable and P adds λ random reals. Then, in V P, for every < λ = 2 ℵ 0, inf does not hold. 12

Definition 3.20. For ℵ 0 < κ = cf(κ), S κ stationary, ℵ 0, σ < κ, the principle sp(,σ) S says the following: There exists A δ : δ S such that each A δ is an unbounded subset of δ of order type cf(δ) and for every A [κ] κ, f α : α A, where each f α : u α σ with u α [κ] <, there exists stationary many (equivalently, some) δ S such that A δ A and f α : α A δ forms a delta system - i.e., for some u, for every α β A δ, u α u β = u and f α u = f β u. Claim 3.21. Let κ, σ,, S be as in Definition 3.20. Suppose for every α < κ, α < < κ. Then, sp(,σ) S is equivalent to S via the same witness. Proof: Easily follows from: Whenever f α : α A is as in Definition 3.20, for some B [A] κ, f α : α A forms a delta system. Claim 3.22. Suppose µ is measurable with a normal ultrafilter D and S S µ+ is a stationary subset of µ + where = cf() < µ. Assume S. Let P be Prikry forcing w.r.t. D. Then, for every, σ < µ, V P = sp(,σ) S. Proof: Note that V P = µ ℵ 0 µ + so it is not enough to just check S. Using Claim 3.21, fix a sp(,σ) S witnessing sequence A δ : δ S in V. We ll show that sp(,σ) S holds in V P with the same witness. So suppose p forces Å [µ+ ] µ+, f α : α Å, ů α [µ + ] <, f α : ů α σ. Let A 1 = {α < µ + : ( p α p)(p α α Å)}. For each α A 1, let p α = (w α, B α ), ξ α < be such that p α p, w α [µ] <ℵ 0, B α D and p α forces α Å and order type of u α is ξ α. Choose A 2 [A 1 ] µ+, w and ξ such that for each α A 2, w α = w, ξ α = ξ. For α A 2, let γ α,i : i < ξ list ů α in increasing order. Since P does not add bounded subsets of µ, we can also assume that each p α forces a value ζ α,i to f α ( γ α,i ). Let X = A 2 ξ. Let E be the equivalence relation on X define by ((α 1, i 1 ), (α 2, i 2 )) E iff for some B D, (w, B) γ α1,i 1 = γ α2,i 2. Let g : X µ + be such that for all x, y X, (x, y) E iff g(x) = g(y). Define f α : α A 2 by dom(f α) = {g(α, i) : i < ξ }, f α(g(α, i)) = ζ α,i. Choose δ S such that A δ A 2 and f α : α A δ forms a delta system. Put B = {B α : α A δ }, q = (w, B). Then q A δ Å. Using the Prikry property, it is easily verified that q f α : α A δ forms a delta system. Claim 3.23. Suppose 2 ℵ 0 < µ, = cf() < µ, S S µ+ is a stationary subset of µ +, {S i : i < µ} is a partition of S into stationary sets and A = A δ : δ S is such that for each i < µ, A S i witnesses sp(ℵ 1,ℵ 0 ) S i with A as a witness.. Let P add µ random reals. Then V P = inf S holds Proof: Suppose p Å [µ+ ] µ+ and 0 < ε < 1. Let A 1 = {α < µ + : ( p α p)(p α α Å)}. For α A 1, choose a compact p α such that p p α α Å. By thinning down A 1, we can also assume that for some clopen set q, Leb(q p α ) < (1 ε/8)leb(p α ). Let γ α,i : i < ξ α where γ α,i < µ are increasing with i and ω ξ α < ω 1 be a list containing supports of p α, p, q. Using 2 ℵ 0 < µ, choose A 2 [A 1 ] µ+, ξ, K 2 ξ compact such that for every α A 2, ξ α = ξ and p α = {x 2 µ : x α K where x α 2 ξ satisfies x α (i) = x(γ α,i )}. Let i = {i n : n < ω}. For α A 2, define f α = {(γ α,in, n) : n < ω}. Choose δ S such that A δ A 2 and f α : α A δ forms a delta system so for some v, for every α, β A δ, α β, {γ α,i : i < ξ } {γ β,i : i < ξ } = v and f α v = f β v. Let q be the constant projection 13

of p α to v. Since both p, q are supported in v, p q, q q and for all α A δ, q p α hence Leb(p α ) > (1 ε/8)(leb(q)). Now the sequence p α : α A δ is independent over q hence the conclusion follows from the law of large numbers. Corollary 3.24. Let V 3 be as in Claim 3.12. In V 3, let P 4 be the measure algebra on 2 µ. Then in V 4 = V P 4 3, we have (1) κ < µ < λ = 2 ℵ 0 (2) κ is RVM (3) For every regular < µ +, inf S µ+ holds. Proof: Follows from Claims 3.22 and 3.23. 4 RVC and Lusin sets Recall that an uncountable set of reals X is Lusin if every meager subset of X is countable. Fact 4.1 (Gitik-Shelah). Suppose κ is RVC. Then for each uncountable < κ, there exists a Lusin set of size. Proof: Let I be a witnessing ideal over κ. Let G be P(κ)/I-generic over V. In V [G], let M be the transitive collapse of the well founded G-ultrapower of V and j : V M V [G], the generic elementary embedding. By [4], G is generic for adding > κ Cohen reals over V so V [G] and therefore M contains a κ-sequence of Cohen reals over V. Hence for each < κ, M and therefore V (by elementarity) contains a Lusin set of size. It is natural to ask: Suppose κ is RVC. Must there exist a Lusin set of size κ? The usual way of getting a model of κ is RVC is to start with a measurable cardinal κ and add κ Cohen reals. The Cohen reals added constitute a Lusin set of size κ. So this is consistent. On the other hand, we ll show the following. Theorem 4.2. It is consistent that 2 ℵ 0 is RV C and there is no Luzin set of size continuum. The proof uses iterations with restricted memory as defined in [5]. The main difference here is that instead of adding a continuous function, we add a meager set by finite approximation. 4.1 Cohen forcing We describe another way of looking at adding ω 1 Cohen reals. Compare with [7]. Definition 4.3. Let σ k : k < ω be an enumeration of 2 <ω. For X = x α : α < ω 1 2 ω, let Q X be the forcing whose conditions are p = (F, g, N) where (1) N < ω, F = F i : i < n, n < ω and each F i is a finite subset of ω 1. 14

(2) g = g i : i < n and each g i : {σ k : k < N} 2 <ω satisfies: For each k < N, g i (σ k ) σ k and [g i (σ k )] {x α : α F i } = φ. Here, for a finite partial function σ from ω to 2, [σ] = {x 2 ω : σ x}. (3) Ordering is defined naturally: F i s and g i s increase, F, g lengthen, N increases. Fact 4.4. Let X, Q X be as above. (a) Q X satisfies ω 1 -Knaster condition: For every uncountable A Q X, there exists B [A] ℵ 1 such that the elements of B are pairwise compatible. (b) In V Q X, X is meager. Let C X = C(X) = {C : i X} (finite support product) denote Cohen forcing for adding X many Cohen reals. So C is 2 <ω ordered by end extension. We also write C(X, Y ) = C(X Y ) = C(X) C(Y ) when X, Y are disjoint. Lemma 4.5. Let P = C ω1 add ω 1 Cohen reals c α : α < ω 1. Let τ be a C-name for a real in 2 ω. Let X = X τ = τ[c α ] : α < ω 1. Let Q X be as above. Then, P Q X is forcing equivalent to P. Moreover, if ω < ω 1, then (P Q X )/C is also forcing equivalent to P. Proof: For ω λ ω 1, consider the forcing A λ of all pairs (p, q) where p C λ, p = ρ α : α u p, ρ α 2 <ω, u p [λ] <ℵ 0 {ρ α : α u p } = {ρ n : n u p ω} q = (F, g, N), F = F i : i < n, g = g i : i < n, N < ω For every i < n, F i u p and g i : {σ k : k < N} 2 <ω For every k < N, σ k g i (σ k ) and p Cλ [g i (σ k )] {τ[c α ] : α F i } = φ A λ is ordered in the natural way. Observe that for any (p, q) C λ Q X λ, there exists (p, q ) A λ such that p p, p P q QX λ q. Hence A λ is dense in C λ Q X λ. For ω < λ ω 1, (p, q) A λ, define (p, q) = (p, q ) A in the natural way. Note that if (p, q) A λ, (p, q ) A and (p, q ) (p, q), then (p, q) and (p, q ) are compatible. It follows that A A λ. Hence A λ : λ [ω, ω 1 ] is -increasing and at limit stages λ, A λ is the finite support limit of A β : β < λ. Moreover, since each quotient A λ+1 /A λ is Cohen (being countable atomless), it follows that A ω1 and A ω1 /A are forcing equivalent to C ω1 for every ω < ω 1. Hence (P Q X )/C is equivalent to A ω1 /A A /C so it is also forcing equivalent to C ω1. 15

4.2 Iterations with restricted memory For proofs and details on this section, see Section 3 in [5]. In the definition of the class K δ (S) below, we assume that each forcing in K δ (S) is a finite support iteration (P α, Q α ) : α < δ satisfying: There are posets R α H ω2 of size ω 1 such that each Q α is a P α name for an upward closed subordering (not necessarily complete) of R α. Members of P α are sequences of length α with finite support ordered coordinatewise - So if p, q P α, then p q iff for all β < α, p β Pβ p(β) Rβ q(β). For u α, P u is the suborder of all conditions in P α whose supports are in u. We now introduce the class K δ (S). (1) A memory template of length δ is a sequence u = u α : α < δ such that (i) u α α (ii) β u α = u β u α (2) Let u be a template of length δ. A subset u of δ is u-closed if for every α u, u α u. (3) Let u be a template of length δ. A finite support iteration (P α, Q α ) : α < δ is a u-iteration if for all α < δ, Q α is a P uα -name for a subset of R α. (4) Let S S δ ω 1. A memory template u of length δ is an S-template if (i) For all α δ\s, u α = φ (ii) For all α, u α ω 1 (iii) For all α S, if β S α is a limit point of u α, then β u α (5) Let u be an S-template of length δ where S S δ ω 1. (i) For u v δ, both u-closed, we say that v is a u-straight extension of u if for every β v S, if u β is unbounded in β, then β u. (ii) A u-closed set u is u-straight if every u-closed set v u is a u-straight extension of u. (6) Let K δ (S) consist of finite support u-iterations P = (P α, Q α ) : α < δ where u is an S-template and the following hold (i) If α δ\s, Q α = C adds a Cohen real (ii) For all α < δ, R α ω 1 and R α H ω2 (iii) For all β < α < δ, β / S, α S, P uα {α}/p uα β is isomorphic to Cohen forcing Fact 4.6. Suppose u is a template of length δ and u v δ are u-closed. Let P = (P α, Q α ) : α < δ be a u-iteration. Then, P u P v witnessed by the function p p u. Fact 4.7. Let P K δ (S) with witnessing template u. Let ρ be a P δ -name for a real. Then there is a u-straight set u of size ω 1 such that δ is a P u name. 16

Fact 4.8. Let P K δ (S) with witnessing template u. Let v be a u-straight extension of u and suppose S does not reflect at any limit point of v\u. Then P v /P u is isomorphic to Cohen forcing. In particular, if v is a u-straight extension of u and v\u ω 1, then P v /P u is isomorphic to Cohen forcing. 4.3 No Lusin set of size κ The aim of this section is to prove the following. Theorem 4.9. Let κ be inaccessible and S Sω κ 1 be stationary. Assume S. Then there exists some P K κ (S) such that, in V Pκ, every set of reals of size κ has a meager subset of size ω 1. P along with its witnessing template u will be constructed inductively using a S sequence. To see how this construction should go let us suppose P has already been constructed and let x α : α κ\s be a κ-sized set of reals in V P δ. We d like to have added a meager set containing uncountably many x α s by some stage of the iteration. Let v α be a u-straight set of size ω 1 such that x α is a P vα -name. We have P vα = P vα α (P vα /P vα α) Using Fact 4.8, we can identify P vα α = C(X α ), for some X α with X α ω 1. Also, in V Pvα α, we can identify P vα /P vα α = C(Y α ) for some Y α with Y α ω 1. So P vα = C(Xα, Y α ) Since κ is inaccessible, there are a stationary S 1 κ\s and f α : α S 1, φ, v, X, Y, τ, such that (a) α S 1 = v α α = v (c) X, Y are disjoint sets and for each α S 1, f α : P vα C(X, Y ), φ : P v C(X) are dense embeddings satisfying f α P v = φ (b) τ is a C(X, Y ) name for a real such that for each α S 1, f α ( x α ) = τ For α S 1, let γ α > α be a successor ordinal such that v α γ α. Let E = {α < κ : α = sup(s 1 α) and α S 1 α = γ α < α}. Then E is a club in κ. Let δ E S. Construct α i : i < ω 1 increasing such that for every i < ω 1, α i S 1, i < j = γ i < α j and sup{α i : i < ω 1 } = δ. Put A = {α i : i < ω 1 }. The next lemma says that we can make { x α : α A} meager in V P δ+1. Lemma 4.10. Let S S κ ω 1, δ S and P K δ (S δ) with u as its witnessing template. Suppose that α i, v i, x i : i < ω 1, X, Y, τ satisfy the following. (1) α i : i < ω 1 is an increasing unbounded subset of δ disjoint with S (2) For each i < ω 1, v i is a u-straight subset of δ of size ω 1, v i α i+1, v i α i = v does not depend on i 17

(3) X, Y are disjoint and there are dense embeddings f i : P vi C(X, Y ), φ : P v C(X) such that f i P v = φ (4) τ is a C(X, Y )-name for a real, x i is a P vi -name for a real and f i ( x i ) = τ Let v = {v i : i < ω 1 }. Then v {δ} is u-straight and there is a P v -name Q for a forcing such that P Q K δ+1 (S δ + 1) witnessed by u v and in V P δ+1, the set { xi : i < ω 1 } is meager. Proof: v {δ} is u-straight because every limit point of v S, other than δ, is a limit point of some v i, i < ω 1. The forcing P v is isomorphic to C(X) {C(Y i ) : i < ω 1 } where Y i s are disjoint copies of Y as witnessed by the function that takes p P v to (q, r i : i < ω 1 ) where φ(p v ) = q C(X) and f i (p v i ) = (q, r i ) C(X, Y i ). Let τ = τ/g C(X) so that τ[g C(X,Y ) ] = τ [G C(X) ][G C(Y ) ]. Let Y Y be infinite countable such that τ is a C(Y )-name (so essentially C-name). Let P = C(X) {C(Y i \Y i ) : i < ω 1 }. Then we can write P v as P {C(Y i ) : i < ω 1 } Let c i to be the Cohen real added by C(Y i ). In V Pv, let X = {τ [c i ] : i < ω 1 }. Put Q δ = Q X. Clearly, { x i : i < ω 1 } = X is meager in V P δ+1. To show that Pδ+1 K δ+1 (S) with witnessing template u v, it is enough to show that for each β δ\s, (P v Q X ) /P v β is isomorphic to Cohen forcing. Pick ω k < ω 1 large enough so that α k > β and write P v αk as C(X) {C(Y i ) : i < k} Hence (P v Q X ) /P v αk is forcing isomorphic to [( ) {C(Yi \Y i ) : k i < ω 1 } {C(Y i ) : i < ω 1 } Q X / ] {C(Y i ) : i < k} which is isomorphic to Cohen forcing by Lemma 4.5. By Fact 4.8, P v αk /P v β is isomorphic to Cohen forcing. It follows that (P v Q X ) /P v β is also isomorphic to Cohen forcing. Proof of Theorem 4.9: We can now define P, u. Using S, get f δ : δ S such that for each δ S, f δ is a function, dom(f δ ) is an unbounded subset of δ, range(f δ ) V δ and for every unbounded T κ, F : T V κ, there are stationary many δ S such that F dom(f δ ) = f δ. Suppose P K δ (S δ), u δ have already been defined. If δ / S, put u δ = φ, Q δ = C. If δ S, dom(f δ ) = {α i : i < ω 1 } unbounded in δ, for every i, f δ (α i ) = (v i, X, Y, τ) where α i, v i : i < ω 1, X, Y, τ satisfy the hypothesis of the Lemma 4.10, then define u δ, Q δ as given in the conclusion of this lemma. We need the following facts from [5]. Fact 4.11. Suppose κ is measurable and 2 κ = κ +. Then there is a forcing extension W of V such that, in W, κ is measurable and there is a stationary S Sω κ 1 such that S holds and j(s) reflects nowhere in the interval (κ, j(κ)]. 18

Fact 4.12. Let P K δ (S). If α < δ, α / S, S reflects nowhere in the interval (α, δ], then P δ /P α is isomorphic to Cohen forcing. Proof of Theorem 4.2: Let W, κ, S Sω κ 1 4.9. Then, W Pκ is as required. be as above. Let P K κ (S) be as in Theorem References [1] B. Balcar, P. Simon, Disjoint refinement, Handbook of Boolean algebras Vol. 2 (J. D. Monk and R. Bonnett, Eds.), North-Holland Amsterdam (1989), pp. 333-388 [2] S. Fuchino, S. Shelah and L. Soukup, Sticks and clubs, Annals Pure and Applied Logic 90 (1997), 57-77 [3] M. Gitik and S. Shelah, On certain indestructibility of strong cardinals and a question of Hajnal, Archive for Math. Logic 28 (1989), 35-42 [4] M. Gitik and S. Shelah, Forcings with ideals and simple forcing notions, Israel J. Math. 68 (1989), 129-160 [5] N. Greenberg, S. Shelah, Models of Cohen measurability, Annals Pure and Applied Logic 165 (2014), 1557-76 [6] K. Kunen, Set Theory - An introduction to independence proofs, North-Holland 1980 [7] S. Shelah, Possibly every real function is continuous on a non-meagre set, Publications de L Institute Mathematique - Beograd, Nouvelle Series 57(71) (1995), 47-60 [8] S. Shelah, Not collapsing cardinals κ in (< κ)-support iterations, Israel J. Math 136 (2003), 29-115 [9] S. Shelah, Many forcing axioms for all regular uncountable cardinals, Publication 832 on Shelah s archive [10] S. Shelah, Diamonds, Proc. Amer. Math. Soc. 138 (2010), 2151-2161 [11] S. Shelah, Many partition relations below density, Israel J. Math 192 (2012), 507-543 [12] S. Shelah, Proper and improper forcing, Perspectives in mathematical logic, Springer- Verlag Berlin, Second edition, 1998 [13] R. Solovay, Real valued measurable cardinals, Axiomatic set theory Proc. Sympos. Pure Math. Vol. 13 Part I Amer. Math. Soc. 1971, 397-428 19