Silver type theorems for collapses.

Similar documents
Strongly compact Magidor forcing.

Extender based forcings, fresh sets and Aronszajn trees

The Semi-Weak Square Principle

Two Stationary Sets with Different Gaps of the Power Function

Sy D. Friedman. August 28, 2001

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

On the Splitting Number at Regular Cardinals

LARGE CARDINALS AND L-LIKE UNIVERSES

Annals of Pure and Applied Logic

Short Extenders Forcings II

Global singularization and the failure of SCH

DEPTH OF BOOLEAN ALGEBRAS SHIMON GARTI AND SAHARON SHELAH

Determinacy models and good scales at singular cardinals

On almost precipitous ideals.

Level by Level Inequivalence, Strong Compactness, and GCH

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

SHORT EXTENDER FORCING

Tall, Strong, and Strongly Compact Cardinals

MITCHELL S THEOREM REVISITED. Contents

Generalization by Collapse

A precipitous club guessing ideal on ω 1

Notes on getting presaturation from collapsing a Woodin cardinal

MODIFIED EXTENDER BASED FORCING

CONSECUTIVE SINGULAR CARDINALS AND THE CONTINUUM FUNCTION

On the strengths and weaknesses of weak squares

On almost precipitous ideals.

Easton s theorem and large cardinals from the optimal hypothesis

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

Philipp Moritz Lücke

COMBINATORICS AT ℵ ω

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

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

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

2. The ultrapower construction

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

Chain conditions, layered partial orders and weak compactness

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

Characterizing large cardinals in terms of layered partial orders

BLOWING UP POWER OF A SINGULAR CARDINAL WIDER GAPS

The Outer Model Programme

Cardinal arithmetic: The Silver and Galvin-Hajnal Theorems

LECTURE NOTES - ADVANCED TOPICS IN MATHEMATICAL LOGIC

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.

SOME CONSEQUENCES OF REFLECTION ON THE APPROACHABILITY IDEAL

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

ON NORMAL PRECIPITOUS IDEALS

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

STRONGLY UNFOLDABLE CARDINALS MADE INDESTRUCTIBLE

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

DIAGONAL PRIKRY EXTENSIONS

Interpolation of κ-compactness and PCF

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

arxiv: v1 [math.lo] 26 Mar 2014

A HIERARCHY OF RAMSEY-LIKE CARDINALS

Generalising the weak compactness of ω

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

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

LOCAL CLUB CONDENSATION AND L-LIKENESS

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

Large Cardinals with Few Measures

THE TREE PROPERTY UP TO ℵ ω+1

COLLAPSING SUCCESSORS OF SINGULARS

Währinger Strasse 25, 1090 Vienna Austria

Chapter 4. Cardinal Arithmetic.

THE SHORT EXTENDERS GAP THREE FORCING USING A MORASS

COMBINATORICS OF REDUCTIONS BETWEEN EQUIVALENCE RELATIONS

Generic embeddings associated to an indestructibly weakly compact cardinal

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

PERFECT TREE FORCINGS FOR SINGULAR CARDINALS

SUCCESSIVE FAILURES OF APPROACHABILITY

Bounds on coloring numbers

Axiomatization of generic extensions by homogeneous partial orderings

ON SCH AND THE APPROACHABILITY PROPERTY

Strongly Unfoldable Cardinals Made Indestructible

Hierarchies of (virtual) resurrection axioms

Fat subsets of P kappa (lambda)

arxiv: v2 [math.lo] 13 Feb 2014

Large cardinals and the Continuum Hypothesis

RVM, RVC revisited: Clubs and Lusin sets

ANNALES ACADEMIÆ SCIENTIARUM FENNICÆ DIAMONDS ON LARGE CARDINALS

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

THE FIRST MEASURABLE CARDINAL CAN BE THE FIRST UNCOUNTABLE REGULAR CARDINAL AT ANY SUCCESSOR HEIGHT

THE TREE PROPERTY AT ALL REGULAR EVEN CARDINALS

A Laver-like indestructibility for hypermeasurable cardinals

Hod up to AD R + Θ is measurable

Notes to The Resurrection Axioms

Chromatic number of infinite graphs

INDESTRUCTIBLE STRONG UNFOLDABILITY

Covering properties of derived models

arxiv: v3 [math.lo] 23 Jul 2018

UPWARD STABILITY TRANSFER FOR TAME ABSTRACT ELEMENTARY CLASSES

PARTITIONS OF 2 ω AND COMPLETELY ULTRAMETRIZABLE SPACES

Dropping cofinalities and gaps

The Resurrection Axioms

EASTON FUNCTIONS AND SUPERCOMPACTNESS

Satisfaction in outer models

FORCING AXIOMS, SUPERCOMPACT CARDINALS, SINGULAR CARDINAL COMBINATORICS MATTEO VIALE

DIAGONAL SUPERCOMPACT RADIN FORCING

Preservation theorems for Namba forcing

arxiv: v1 [math.lo] 8 Oct 2015

Transcription:

Silver type theorems for collapses. Moti Gitik May 19, 2014 The classical theorem of Silver states that GCH cannot break for the first time over a singular cardinal of uncountable cofinality. On the other hand it is easy to obtain a situation where GCH breaks on a club below a singular cardinal κ of an uncountable cofinality but 2 κ = κ +. We would like here to investigate the situation once blowing up power of singular cardinals is replaced by collapses of their successors. 1 ZFC results. The following basic result should be well known and goes back to Silver: Theorem 1.1 Suppose that V W are transitive models of ZFC with the same ordinals such that: 1. κ is a cardinal in W, 2. κ changes its cofinality to ω 1 in V witnessed by a club κ α α < ω 1, 3. for every α < ω 1, (κ + α ) W < κ + α (or only for stationary many α s), 4. κ is a strong limit in V or just it is a limit cardinal and κ ω 1 α < κ, for every α < ω 1. Then (κ + ) W < κ +. Proof. Suppose that (κ + ) W = κ +. Fix in W a sequence f i i < κ + of κ + first canonical functions in ν<κ ν+, < J bd κ just any sequence of κ + many functions in ν<κ ν+ increasing mod Jκ bd. Set in V g i = f i {κ α α < ω 1 }, for every i < κ +. Then g i i < κ + is an increasing sequence 1 or

of functions in α<ω 1 (κ + α ) W, < J bd ω 1. By the assumption (3) we have that for every α < ω 1, (κ + α ) W < κ + α. Now, as in the Baumgartner-Prikry proof of the Silver Theorem (see K. Kunen [2] p.296 (H5)), it is impossible to have κ + many such functions. Hence (κ + ) W < κ +. Let us deal now with double successors. Theorem 1.2 Suppose that V W are transitive models of ZFC with the same ordinals such that: 1. κ is a cardinal in W, 2. 2 κ κ ++, and moreover there is a sequence of κ ++ many functions in ν<κ ν++ increasing mod J bd κ, 3. κ changes its cofinality to ω 1 in V witnessed by a club κ α α < ω 1, 4. for every α < ω 1, (κ ++ α ) W < κ + α (or only for stationary many α s), 5. κ is a strong limit in V or just it is a limit cardinal and κ ω 1 α < κ, for every α < ω 1. Then (κ ++ ) W < κ +. The condition (2) allows to repeat the proof of 1.1. Let state the following relevant result of Shelah ([3](4.9,p.304)), which says that once (κ + ) W changes its cofinality, then we must have (κ ++ ) W < κ + unless cof((κ + ) W ) = cof( (κ + ) W ) = cof(κ). Proposition 1.3 Let F be the κ complete filter of co-bounded subsets of P κ (κ + ), i.e. the filter generated by the sets {P P κ (κ + ) α P }, α < κ +. Then there is a sequence f i i < κ ++ of functions such that 1. f i : P κ (κ + ) κ, 2. f i (P ) < P +, for all P P κ (κ + ), 3. f i > F f j, whenever i > j. Proof. We define a sequence f i i < κ ++ by induction. Suppose that f j j < i is defined. Define f i. 2

Case 1. i = i + 1. Set f i (P ) = f i (P ) + 1. Case 2. i is a limit ordinal of cofinality δ < κ. Pick a cofinal in i sequence i τ τ < δ. Set f i (P ) = τ<δ f i τ (P ) + 1. Case 3. i is a limit ordinal of cofinality δ κ, i.e. δ = κ or δ = κ +. Pick a cofinal in i sequence i τ τ < δ. Set f i (P ) = τ P f i τ (P ) + 1. Theorem 1.4 Suppose that V W are transitive models of ZFC with the same ordinals such that: 1. κ is an inaccessible in W, 2. κ > (cof(κ)) V = δ for some uncountable (in V ) cardinal δ. 3. κ is a strong limit in V or just it is a limit cardinal and for every ξ < κ, ξ δ < κ. 4. There exist a club κ α α < δ in κ (or just a stationary set) 1 and a sequence P α α < δ such that (a) P α (P κ (κ + )) W, for each α < δ, (b) ( P α + ) W < κ + α, for each α < δ, (c) (κ + ) W = α<δ P α, (d) for every Q (P κ (κ + )) W, there is α < δ such that for every β, α β < δ, Q P β. Then (κ ++ ) W < κ +. Proof. Suppose otherwise. Then (κ ++ ) W = κ +, by the assumption (b),(c) above. Let f i i < κ ++ be a sequence of functions in W given by Proposition 1.3. We can repeat the argument of 1.1 with slight adaptations. Thus, set in V g i (α) = f i (P α ), for every α < ω 1 and i < (κ ++ ) W = κ +. Let ν α := ( P α + ) W. By the assumption, ν α < κ + α. Then g i i < κ + is an increasing sequence of functions in α<δ ν α, < J bd, δ since for every A F we have {P α α α 0 } A, for some α 0 < δ. This is impossible, since ν α < κ + α, for every α < δ. Contradiction. 1 Note that if δ = ω 1, then we can just force a club into it without effecting things above. 3

Theorem 1.5 Suppose that V W are transitive models of ZFC with the same ordinals such that for some inaccessible in W cardinal κ both κ and its successor in W change their cofinality to some uncountable (in V ) cardinal δ and κ remains a cardinal in V. Then the following conditions are equivalent: 1. (*) There are a clubs κ α α < δ in κ and η α α < δ in (κ + ) W such that for every limit α < δ (or just for stationary many α s) 2 the set {η β β < α} can be covered by a set a α W with ( a α + ) W < κ + α. 2. (**) There are a clubs κ α α < δ in κ and η α α < δ in (κ + ) W such that for every limit α < δ (or just for stationary many α s) the set {η β β < α} has an unbounded intersection with a set b α W with ( b α + ) W < κ + α. 3. There exist a club κ α α < δ in κ and a sequence P α α < δ such that (a) P α (P κ (κ + )) W, for each α < δ, (b) P α κ = κ α, for each α < δ, (c) ( P α + ) W < κ + α, for each α < δ, (d) (κ + ) W = α<ω 1 P α, (e) for every Q (P κ (κ + )) W, there is α < ω 1 such that for every β, α β < ω 1, Q P β. 4. There exists an increasing sequence P α α < δ which satisfies all the requirements of the previous item. Proof. Split the proof into lemmas. Lemma 1.6 (*) iff (**). Proof. Clearly, (*) implies (**). Let us show the opposite direction. We fix a bijection π ξ : κ ξ in W, for every ξ < (κ + ) W. Fix in V a function π : κ onto (κ + ) W. Set now for every α < δ, η α = sup(π κ α ). Then, clearly, {η α α < δ} is a club in (κ + ) W. Now given a sequence which witnesses (**). Without loss of generality we can assume that it is the sequence η α η < δ defined above. Otherwise 2 If δ = ω 1, then it is basically the same, since once we have only stationary many such α s, then force a club into it. Everything is a the level of ω 1, so this will have no effect on the cardinal arithmetic above. 4

just intersect two clubs. Define an increasing continuous sequence N α α < δ of elementary submodels of some H χ, with χ big enough such that 1. δ, κ, κ α α < δ, π ξ ξ < (κ + ) W, π N 0, 2. N α < δ, 3. N α δ is an ordinal, 4. N β β α N α+1. Denote N α δ by δ α. Then sup(n α κ) = κ δα and sup(n α (κ + ) W ) = η δα. Clearly, δ α = α for a club many α s. Suppose now that for some α < δ we have δ α = α and there is a set X W such that ( X + ) W < κ + α, X {η β β < α} is unbounded in η α. Note that η β N α, for every β < α and then, also, π ηβ N α. By elementarity, then π ηβ (N α κ α ) : N α κ α N α η β. In particular, π ηβ κ α {η γ γ < β}. Set Y := {π ζ κ α ζ X η α }. Then, Y W, Y W κ α + X W, and so ( Y + ) W < κ + α. But, in addition, Y {η γ γ < α}, since for unboundedly many β < α, we have η β X and so, π ηβ κ α {η γ γ < β}. of the lemma. Lemma 1.7 (1) implies (3) Proof. Fix clubs κ α α < δ and η α α < δ witnessing (1). Let us build first a sequence R α α < δ which satisfies all the requirements of (3), but probably is not increasing. Set R 0 = κ 0 ((π η0 κ 0 ) \ κ). Let α, 0 < α < δ be an ordinal. Pick a α W, a α η α to be a cover of {η β β < α} with ( a α + ) W < κ + α. Set R α = {π ξ κ α ξ b α {η α }}. Let R α = κ α (R α \ κ). 5

The constructed sequence satisfies trivially the requirements (a),(b) and (c). Let us check (e). (d) clearly follows from (e). Let Q (P κ (κ + )) W. There is β < ω 1 such that Q η β. Consider π 1 η β Q. It is a bounded subset of κ. Hence there is γ < ω 1 such that κ γ π 1 η β Q. So π ηβ κ β Q. Let α < ω 1 be an ordinal above β, γ. Then R δ Q, for every δ α. of the lemma. Lemma 1.8 (3) iff (4). Proof. Clearly (4) implies(3). Let us show the opposite direction. Let a club κ α α < δ in κ and a sequence R α α < δ witness (3). Define an increasing subsequence P α α < δ Set P 0 = R 0. By (e) there is α 0 such that for every β, α 0 β < δ, P 0 R β. Set P 1 = R α1. Continue by induction. Suppose that ν < δ and for every ν < ν, increasing sequences α ν ν < ν and P ν ν < ν are defined and satisfy the following: 1. P ν = R α ν, 2. for every β, α ν β < δ, P ν R β. If ν is a successor ordinal, then let ν = µ + 1, for some µ. Set P ν = R αµ be such that for every β, α ν β < δ, P ν R β. and let α ν < δ If ν is a limit ordinal, then let P ν = R ν <ν α and define α ν ν as in the successor case. Finally let us define an increasing subsequence of P α α < δ which satisfies the properties (a)-(e) of (3). Let C := {ν < δ ν = ν <ν α ν }. Clearly it is a club. Set P ν = P ν, for every ν C. Then κ α α C and P α α C are as desired. of the lemma. Lemma 1.9 (3) implies (1). Proof. Let a club κ α α < δ in κ and a sequence P α α < δ witness (3). Let η α α < δ be a club in (κ + ) W. We claim that there is a club C δ such that for every α C, P α {η β β < α}. Suppose otherwise. Then there is a stationary S δ such that for every α S there is α < α with η α P α. Then there are a stationary set S S and α < δ such that for every α S, η α P α. This is impossible by (d). 6

of the lemma. Theorem 1.10 Suppose that V W are transitive models of ZFC with the same ordinals such that: 1. κ is an inaccessible in W, 2. κ > (cof(κ)) V = δ for some uncountable (in V ) cardinal δ > ω 1. Let κ α α < δ be a witnessing club. 3. For every α < δ, (κ ++ α ) W < κ + α (or only for stationary many α s), 4. κ is a strong limit in V or just it is a limit cardinal and κ ω 1 α < κ, for every α < δ. 5. There is a regular cardinal δ, ω < δ < δ such that for every regular cardinal ρ < κ of W which became a singular of cofinality δ in V, there is a club a club sequence ρ i i < δ in ρ such that for every club c δ the set {(cof(ρ i )) W i c} is unbounded in ρ. Or 6. Like the previous item but only for ρ s of the form (cof(η α )) W with α < δ of cofinality δ, where η α α < δ is a club in (κ + ) W. Then (κ ++ ) W < κ +. Proof. Let us argue that (**) of 1.4 holds. Assume for simplicity that δ = ω 1. Let N α α < δ and η α α < δ be as in 1.6. Pick α < δ of cofinality ω 1 with δ α = α. Consider η α. Then cof(η α ) = ω 1. If (cof(η α )) W < κ + α, then we pick in W a club X in η α of the order type (cof(η α )) W. Then X {η β β < α} is a club, and so, unbounded in η α. Suppose now that (cof(η α )) W κ + α. Denote (cof(η α )) W by ρ. Then ρ κ, since η α < (κ + ) W. It is impossible to have ρ = κ, since cof(κ) > ω 1 = cof(α) = cof(η α ) = cof(ρ). Hence κ + α ρ < κ. In particular, ρ κ + α. By the assumption (5) of the theorem, there is a club a club sequence ρ i i < ω 1 such that for every club c ω 1 the set {(cof(ρ i )) W i c} is unbounded in ρ. Let e = {e ξ ξ < ρ} W be a club in η α. Consider d := {η β β < α} e. It is a club in η α. So there are some 7

γ < α and j < ω 1 such that η γ = e ρj and (cof(ρ j )) W > κ α. But this is impossible, since η γ N α, and hence, (cof(η γ )) W = (cof(ρ j )) W N α κ κ α. Hence, always (cof(η α )) W < κ + α. So, the set {η α α < δ and cof(α) = ω 1 } witnesses (**) and we are done. Lemma 1.11 For every β < δ, {(cof(η γ )) W γ < β} κ β. Proof. Otherwise there is γ < β such that (cof(η γ )) W κ β. Recall that κ < η γ < (κ + ) W. Hence, (cof(η γ )) W κ. It is impossible to have (cof(η γ )) W = κ, since cof(κ) = δ > N γ cof(η γ ) = cof((cof(η γ )) W ). So, (cof(η γ )) W < κ. But (cof(η γ )) W N β and sup(n β κ) = κ β. Lemma 1.12 Suppose that for every β < δ, κ + β is is successor cardinal in W and ν β is its immediate predecessor, then, for a club many β < δ of uncountable cofinality (cof(η β )) W ν β. Proof. Otherwise there will be stationary many β s of uncountable cofinality with (cof(η β )) W < ν β. Then (**) holds on this stationary set. Lemma 1.13 Suppose that for every β < δ, κ + β many β < δ of uncountable cofinality (cof(η β )) W > κ + β. is a limit cardinal of W, then, for a club Proof. Otherwise there will be stationary many β s of uncountable cofinality with (cof(η β )) W < κ + β. Then (**) holds on this stationary set. Theorem 1.14 Suppose that V W are transitive models of ZFC with the same ordinals such that: 1. κ is an inaccessible in W, 2. κ > (cof(κ)) V = δ for some uncountable (in V ) cardinal δ > ω 1. Let κ α α < δ be a witnessing club. 8

3. For every α < δ, (κ ++ α ) W < κ + α (or only for stationary many α s), 4. κ is a strong limit in V or just it is a limit cardinal and κ ω 1 α < κ, for every α < δ. Assume that (κ ++ ) W κ +. Then there is an increasing unbounded in κ sequence ρ α α < δ such that ρ α is a regular cardinal in W, for every limit α, cof(ρ α ) = cof(α), for every limit α of uncountable cofinality, ρ α ρ α > κ α sup({ρ β β < α}), for every limit α of uncountable cofinality, there is a club c α in ρ α such that for every τ c α we have (cof(τ)) W {ρ β β < α}. Proof. Just take ρ α = (cof(η α )) W. Suppose that α has an uncountable cofinality. Then, by 1.13, ρ α ρ α κ + α, and by 1.11, {ρ β β < α} κ α. Fix some increasing continuous function φ α : ρ α η α in W with ran(φ α ) unbounded in η α. Set c α := {φ 1 α (η β ) β < α limit and η β is a limit point of ran(φ α )}. Let τ c α. Then τ = φ 1 α (η β ) for a limit β < α and η β is a limit point of ran(φ α ). Now the continuity of φ α implies that (cof(τ)) W = (cof(η β )) W which is ρ β. 2 A forcing construction. We would like to show the following: Theorem 2.1 Suppose that κ is a κ +3 supercompact cardinal. Let S be subset of ω 1. Then there are generic extensions V V such that 1. κ changes its cofinality to ω 1 in V, 2. there is a closed unbounded in κ sequence κ α α < ω 1 of cardinals in V such that S = {α < ω 1 (κ + α ) V = (κ + α ) V } and ω 1 \ S = {α < ω 1 (κ + α ) V < (κ + α ) V }. 9

Let us describe the construction. Assume GCH, κ is a κ +3 supercompact cardinal and S is a subset of ω 1. 3 Fix a coherent sequence W = W (α, β) α dom( W ), β < o W (α) such that 1. κ = max(dom( W ), 2. o W (κ) = ω 1, 3. for every α dom( W ), β < o W (α), W (α, β) is a normal ultrafilter over P α (α ++ ), 4. W (α, β) = jw (α,β) (f)(α), for some f : α V. Consider the Levy collapse Col(κ, κ + ). Let p Col(κ, κ + ). Set F p = {D Col(κ, κ + ) D is a dense open above p}. Then F p is a κ complete filter over a set of cardinality κ +, for every p Col(κ, κ + ). It is also fine in a sense that for every η < κ +, {q Col(κ, κ + ) η ran(q)} F p. Let j : V M be an elementary embedding with κ a critical point and κ++ M M. For every p Col(κ, κ + ), pick p j F p. 4 So, p (Col(j(κ), j(κ + ))) M. Set F p = {X Col(κ, κ + ) p j(x)}. Then F p is a κ complete ultrafilter which extends F p. Note that F p is a filter on P κ (κ κ + ), hence F p is an ultrafilter there. Now find, in M, some (least) η < j(κ + ) which codes p p Col(κ, κ + ). Define a κ complete ultrafilter W over P κ (κ + ) κ + as follows: X W iff j κ +, η j(x). For every p Col(κ, κ + ), fix a projection π p : P κ (κ + ) κ + Col(κ, κ + ) of W onto Fp. 3 The interesting case is when S and its compliment are both stationary. 4 In some fixed in advance well ordering. 10

Now use the coherent sequence W to define in the obvious fashion a new coherent sequence W where each W (α, β) is an α complete ultrafilter over P α (α + ) α + defined from W (α, β) as above. Note that W (α, β) will belong already to the ultrapower by W (α, β) P α (α + ) = W (α, β) P α (α + ). Thus, W (α, β) belongs to the ultrapower by W (α, β), by coherency. By the condition (4) above it will be in the ultrapower by W (α, β) P α (α + ), since this ultrapower is closed under κ + sequences. Force the supercompact Magidor forcing with W. 5 Denote by V a resulting generic extension. Let P ν, η ν ν < ω 1 be the generic sequence. Then P ν ν < ω 1 be the supercompact Magidor sequence. Denote P ν κ by κ ν. If ν < ν < ω 1, then P ν, η ν P ν, η ν. In particular, η ν P ν. Also, η ν codes elements of Col(κ ν, P ν ). 6 For every ν S fix a cofinal sequence ν n n < ω. Let ν S. Consider η νn Col(κ νn+1, P νn+1 ) codded by η νn. Let tr ν : P ν κ + ν be the transitive collapse of P ν. Consider a set n < ω. Denote by t i ν,n i < κ + ν n the sequence of members of Z ν := {tr ν t i ν,n n < ω, i < κ + ν n }. It is a subset of Col(κ ν, κ + ν ). Define a partial order ν on Z ν as follows: tr ν t i ν,n ν tr ν t j ν,m iff n m and tr ν t i ν,n Col(κν,κ + ν ) tr ν t j ν,m. Set G ν to be the set of all unions of all < ν increasing ω sequences of elements of Z ν. Lemma 2.2 There is g G ν which is generic for Col(κ ν, κ + ν ) over V. Proof. Work in V. Define a function g as follows. Start with tr ν t 0 ν,0. Pick i 1 < κ + ν 1 such that t i 1 ν,1 comes from the ultrafilter F t 0 ν,0 over Col(κ, κ + ). Continue by induction. Suppose that t n ν,i n is defined. Pick i n+1 < κ + ν n such that t i n+1 ν,n+1 comes from the ultrafilter F t in ν,n over Col(κ, κ + ). Finally set g = n<ω tr ν t i n ν,n. We claim that g is as desired. 5 Set here Q, ξ P, η iff Q {ξ} P and Q < P κ. 6 Note that η ν need not code only members of Col(κ ν, P ν ), or even of Col(κ ν, P ν ). 11

Work in V above a condition which already decides κ ν. Suppose for simplicity that none of κ νn, n < ω is decided yet. Let D be a dense open subset of Col(κ ν, κ + ν ). Intersect the measure one set of F 0 with D. The resulting condition will force g extends a member of Ď. The next lemma follows from the definition of G ν. Lemma 2.3 For every n 0 < ω, G ν V [ tr ν P νn n 0 < n < ω ]. Set V = V [ G ν ν S ]. Let now ρ ω 1 \ S. We need to argue that (κ + ρ ) V = (κ + ρ ) V. By Lemma 2.3, it follows that V [ G ν ν S \ ρ ] V [ P τ, η τ ρ < τ < ω 1 ], i.e. the extension of V by the same forcing but which only starts above κ ρ. Such extension does not add new bounded subsets to κ + ρ and below. Hence, it is enough to deal with the forcing up to κ ρ. Let us split the argument into two cases. Case 1. ρ is a limit point of ρ ω 1 \ S. Let then ρ k k < ω be a cofinal sequence consisting of elements of ω 1 \ S. Assume for simplicity that ρ 0 = 0. For every ν S ρ find the least k(ν) such that ν < ρ k(ν). Let n ν be the least n < ω such that nu n > ρ k(ν) 1, if k(ν) 1 and 0 otherwise. Consider V ρ := V [ κ τ τ < ρ, tr ν P νn, tr ν η νn n ν n < ω ν S ρ ]. Then V [ G ν ν S ρ ] V ρ. Lemma 2.4 V ρ is a generic extension of V by a Prikry type forcing which satisfies κ + ρ c.c. Case 2. ρ is not a limit point of ρ ω 1 \ S. The treatment of this case is similar and even a bit simpler than the previous one. 12

References [1]. Gitik, Prikry type forcings, Handbook of Set Theory [2] K. Kunen, Set Theory. [3] S. Shelah, Cardinal arithmetic 13