A Laver-Type Indestructability for Accessible Cardinals

A Laver-Type Indestructability for Accessible Cardinals

Logic Colloquium '86 F.R.Drake and J.K. Truss (Editors) 9 0 Elsevier Science Publishers B.V. (North-Holland), 1988 A LAVER-TYPE INDESTRUCTABILITY ...

442KB Sizes 2 Downloads 35 Views

Logic Colloquium '86

F.R.Drake and J.K. Truss (Editors)

9

0 Elsevier Science Publishers B.V. (North-Holland), 1988

A LAVER-TYPE INDESTRUCTABILITY FOR ACCESSIBLE CARDINALS Shai Ben David, Department of Mathematics, Hebrew University of Jerusalem. INTRODUCTION In [L78] Laver introduced a forcing notion super-compactness of a super compact x

PH

that makes the

indestructible under x-directed-

closed forcing extensions. to an accessible cardinal and

We wish to collapse a large cardinal H

still have indestructibility for some of its compactness properties. If x

A

is a super compact and

stationary subset S of

A

is a cardinal above it then every

reflects, provided all its elements have co-

x.

finalities less than

As an application

We get an indestructibility theorem for this property. we get a model where for every long as [a E S:cof(a.)'

every stationary subset S C A

# A} is stationary in

super compact cardinal, p notion P

A

regular ( H

A.

reflects as

x is a

We prove that if

x 5 A then there is a forcing

and

so that: p

(i) P is p-closed (so in particular

and cardinals below it are not

collapsed); (ii) In

(iii) P (iv) if

V[PI x = p+;

does not collapse cardinals or change cofinalities above x;

P

is generic for

order then in V[Pl[k] For

A

regular

P

P

then V[P]

1

"If R

is a x-closed partial

every stationary subset of

S
reflects".

x to A+.

is just the usual Levy collapse of

For

is the Levy collapse folowed by a A+-strategically closed

singular

A,P forcing of size A+. 1.

A VERY WEAK BOX PRINCIPLE Assume

(A+)<'

=

Let (A :a. < )'1

A+.

+ a

sequence of subsets of ]Aa] 5 A

IJ

and

P
Aa = PcA(i+).

c

),+:]XI

For

be an increasing and continuous such that for each a,

< A}

6 < A+

a limit ordinal we say that

a
Ai.

Let

S*(A+) =

i<6 the definition of

and every initial segment of

I6 < A+:&

* +

S (1 )

* +)

s (A

uniquely defined.

is a member of

is not approachable}.

depends upon the sequence

such sequences agree on a closed unbounded subset of class of

c6

(Aa:CI

A+,

Note that although

< A'),

as any two

the equivalence

modulo the ideal of non-stationary subsets of A +

is

S.Ben David

10

Let us denote by 0;

the property

'I

S

* (A+

".

A+

is not stationary in

(The superscript s stands for Shelah who introduced this property in [ 5 7 8 ] . ) Another way of formulating 0; (b) the sequence

3 (Aa:Q <)'A

is:

such that:

A,

(a) for each a , IAaI 2

A>

(Aa:a <

(c) for each limit 6 <

A+

is 5-increasing and continuous, there is a closed

c6,

6,

cofinal in

otp(c ) = cof(6) and for every 6 E c6, c6n6 E A6 (by continuity A 6 * 6 =a
Oh

above and (c'): For each limit

6 , otp(c ) = cof 6 and for every 6

6 < A+ there is a closed c6, cofinal in 6 Ec6, c6r16 E A 6'

OBSERVATION ( 1 ) If

(K,,s0)

+

(A+,A)

One way to see it is:

holds.

then 0;

the transfer property implies the existence of a

A+-special Aronszajn tree, which is equivalent to the principle 0;

and

0;

implies 0;. ( 2 ) p<'

'

implies 0'.

=

directly.

THEOREM 1 . O Shelah

This follows from ( 1

A,

For a singular

( [ 5781 ) :

supercompact cardinal H

implies that there is no

0;

A.

cof(h) < H. <

such that

COROLLARY The failure of 0;

but can also be easily proven

for a singular

is consistent with

ZFC

if

the existence of a supercompact cardinal is. REMARK

In

[ S 7 8 ] a model of

*

model f o r 7 0

ZFC for - 0

is constructed, and in [BS86] a YJJ

Iyw

is constructed.

Both models are forcing extensions of a

ground model with a supercompact cardinal. The relevance of 0;

to our topic is through Shelah's theorem 20 of [ S 7 8 ] .

This theorem implies that: SA+ and <' V[+I 14

+

P

is a '-closed

If VkO;,

then if

forcing notion,

s is stationary in A+

h

S

is a stationary subset of

V-generic for P, then

$1.

The next theorem shows that any model can be extended by a 'mild' forcing to where 0; THEOREM 1 . 1

holds (for any given For every cardinal h

A). such that '2 2 h

for all

< 1,

is a A-strongly-strategically-closed forcing not'ion PA such that and a generic set for PA introduces a 0; sequence. For a definition of paragraph of section 2 .

'' A-strongly-strategically-closed

"

there

lPAl = 2'

see the first

A Laver-Type Indestructability for Accessible Cardinals REMARK

11

The cardinal arithmetic assumption can be weakened at the cost of

complicating the definition of 0; along the lines of Definitions 1 , 2 in [5781. proof. -

A condition p

A3

p = (A :a < 1.1 <

;

in

is just an initial segment of a 0:

PA

is a set of 5 A

each A

is 5-increasing, continuous, and for each limit 6 5 p exists a closed cofinal subset c6 proper initial segment of p

q

c6

if Dom(p) 5 Dom(q)

and

q

(=

Dom(p))

of order type cof(6)

belongs to

sequence,

subsets of a , the sequence

A6.

there

such that every

The order is end-extensions:

(Dom(p)) = p.

CLAIM For every p

Proof. c6

E PA and

p

p < A+

there is an extension of

p,q

such that

E Dom(q). Let p = (A :a < 11)

and

a

6

closed cofinal in &:p 5

11 5 p .

For each limit p 5 6 5 p

of order-type cof(6).

6 5 p}.

Let q

be

Now, for

6

11

(Aa:a < p)^ (A6:p 5

6

pick some 5 p

let

5 p).

CLAIM PA is A-strongly-strategically closed.

Proof.

Having a chain of conditions of cardinality
condition extending all of them should extend

(pi:i < p ) , any pi; on the other hand, if

u

iEp U pi is a condition in PA. so to i


extends

U pi then

guarantee that the union of this chain is a condition. obstacle to

U p. being a member of PA

i


The only possible

is that maybe for 6

such that otp(c6) = cof(6)

Let us defineplayer 1's strategy H, by

H (pi:i 5

a)

-

=

=

iLlllDom(pi),

and for each

a!i[

"i] AADom(pa)

A is {a:a c {Dompi:i a and pi was picked by player I}) -Dom(pa) where a is the closure of a in the ordinal topology. Note that by the where

assumption p <

A

a,

I A ~ ~ ~ ( ~ L~ A.) I

Now at any limit stage p , (Dom(pi):pi was picked by player I) is cofinal in 6 = U Dom(p. ) , so i


3

21-I 5

Every initial segment of such

a

is a member of some

for some stage i < P that was played by player ADom ( p i shown that H is a winning strategy for player I.

I.

We have thus

S. Ben David

12

The definition of

PA and the previous claim imply that a generic set for sequence: this completes the proof of Theorem 1.1.

PA introduces a 0;

2. ON !.I-STRATEGICALLY-CLOSEDFORCING NOTIONS

A partial order

DEFINITION

is p-strongly-strategically-closed if player

has a winning strategy in the following game.

I

Each player in his turn

chooses a condition p E p extending all the conditions that have been chosen at previous stages, thus constructing an increasing chain in p (Pi:i < a ). Player I1 gets to play all even and limit stages and Player I the odd stages. Player I wins if for any a 5 p

.

(P. :i < a)

there is condition in

Clearly, any p-closed p

a p-strongly-strategically-closed p so

it is

extending

is a !.I-strategically-closedpartial order

p

distributive (forcing via such

(p,m)

p

is !.I-strongly-strategically-closedand adds no p-sequences to the

ground model). In Theorem 2.5 we indicate why being p-strongly-strategically-closed is strictly weaker than being p-closed. We will now generalize some theorems due to Magidor and Shelah (respectively) concerning closed partial orders, to strongly-strategically-closed partial orders. THEOREM 2.1

(Magidor [M82], Lemma 3 )

notion

regular) and

(H

p

Let p

be a X-closed separative forcing

a cardinal above the cardinalities of p

and the

set of its dense subsets. (a) In V[col(X,
(b) For every 6, of

V[6]

p

there is a

generic filter over

is a x-closed forcing extension of g-generic over

V,

V,6,

such that

v[6].

there is a x-closed forcing extension

making it a model of the form V[col(x,
and such that

6 is

the g-generic object defined in this model by part (a). OBSERVATION 2.2 "

P

The conclusions of Theorem 2.1 still hold when the assumption

is x-closed"

is replaced by

"

p is x-strongly-strategically-closed" .

This was independently noticed by Matt Foreman. Proof.

One just repeats Magidor's proof but in the construction of the

sequence

(pa:CY < 8)

V[col(H,
uses Player's I strategy.

define a V-generic subset-for (pas < x)

an increasing sequence one-to-one mapping of

x

onto

generic set of thecollapse. (pa:"

< 0)

,

if

F(P)

{D

p

More specifically, in by constructing inductively

of elements of p.

c P:D

Let p

is dense in

Let

F

be the

introduced by the

be the minimal upper bound of

is a dense subset of

p

we define pD

to be some

A Laver-Type Indestructability for Accessible Cardinals canonical extension of

in

p

F(p)

is the strategy for Player I; if

pp = H( (Pa:@ < p ) * @ d )

and

to be any common extension of p and q x-closed any sequence of length < x

of

pick

if such exists, and p

p

of elements of

H

P

P

otherwise,

Note that as the collapse to

x

is

is a member of the

ground mode1,so it belongs to the domain of the strategy H. x-strongly-strategically-closed and

where H

of p

F(p) is some member q

and proceed as in the previous case.

13

As

p

is

is used in successor stages (the choice

can be regarded as being made by Player 11) an upper bound to

pp

ba:a < p) Let

exists for all

p <

x.

be the p-generic filter generated by

G

check that V[col(x,
+,:a < M).

It remains to

is a x-closed extension of V[G].

This is done by

repeating the argument of Magidor's proof.

p

REMARK We do not know if the theorem still holds when the assumption on is replaced by:

Shelah IS781 (Theorems 20, 21, 22) Supposeforaregular p

THEOREM 2 . 3 above it,

V[61,

S

p is x-strategically-closed.

S

5 gcf(X+).

Then there is a p-closed forcing

is not stationary if and only if

REMARK The assumption

S

THEOREM 2 . 4

X for all

is not stationary.


For p , ~ , s 5 'A

S.

It follows (for every

S

g A+)

from

p < p.

A+

as above if

(S\S*(A+))~S<~is stationary

p,

then for any p-strategically-closed forcing

S

is stationary in V[i].

Note that we don't assume strong-strategical-closure. demand on

and

such that in

5 gcf(X+) relates to the cardinal arithmetics of

the cofinalities of the elements of the assumption 2 ' 5

[s\S*(X')lnSx+

p

This relaxes the

p by allowing games in which Player I has to play the limit

stages in order to guarantee a winning strategy. Proof.

The proof is the proof of Shelah's theorem.

submodel N < H

of size , Isuch that its A+

Once forms an elementary

is some 6 E

(A++) constructs ih this model a cofinal sequence of conditions of of Player I is used (instead of the p-closedness of

p

S

and

p. The strategy

in Shelah's proof) to

guarantee that there is some p E p above all the members of the sequence. is stationary allows to pick 8 outside The assumption " (S\S*(A')) 'I

S*(X')

so there is some cofinal sequence in

segments are members of THEOREM 2 . 5

6

such that all its initial

N.

If there exists a super compact cardinal, then the assertion

S. Ben David

14

P

'I

is A-strongly-strategically-closed

A-closed

1.1 <

More precisely, there are cardinals

(i)

is strictly weaker than

"

p

p

is p-closed.

(iii) The cardinality of

K

is

Let

A and a partial order p

so that:

x).

Let

one of them extends the other.

PA

+W

V[D]

and

A

(the omegath

indestructible under

K

be the partial order of Theorem 1 . 1 .

PA,

if

p

V[6.J.

If

is a cardinal such that

D 5 PA

If

and q

It follows that if

is a super compact cardinal in

holds in

= K

is A-strongly-strategically-closed.

PA

then it is X-directed closed as in then H

A

Apply Laver's method to make

x-directed closed forcings. [PA[= A+;

A+.

be a super-compact cardina1;set

K

cardinal above

0:

is

is A-strongly-strategically-closed.

(ii) No dense subset of

Proof.

P

'I

'I.

is H-closed

are compatible, then

PA D

is V-generic for

is dense in

cof(A) < K < A.

PA

then

This

contradicts Theorem 1 . 0 . This theorem aswers a question raised by Forman.

3. THE MAIN RESULT The next theorem is the key to the results of this paper.

It is a

generalization to K-strongly-strategically-closed partial orders, of a claim that appears implicitly in the proof of Magidor's Theorem 2 o'f [M82]. Let

THEOREM 3 . 1

below n.

n be a supercompact cardinal and

p

If

V[col(p,
p

some regular cardinal

is a x-directed-strongly-strategically-closed

V[cOl(p,
satisfies:

A s 5 S C K , if s\S*(A) stationary in

'I.

COROLLARY 3.2

For

"

For every

is stationary in

K,p,

p

A,

h 1 K s.t. cof(A) 1 K then for some a <

as above, V[cOl(P,
(a) For any cardinal A > p, 0;

forcing in

P,

then the model obtained by further forcing via

and every

A,

Sna

satisfies:

implies that every stationary subset of

sA+ reflects. CP

(b) For any

p > p

if 1Jcp = p

then every stationary subset of

S"

CP

reflects.

LEMMA 3.3

Proof. As

The theorem holds for K-directed-closed P's.

Pick a cardinal A * K

above K

and

2"'.

is a supercompact cardinal (in the ground model V)

elementary embedding

j:V+ M

is

to some transitive model M

there is an

of

ZFC s.t.

A Laver-Type Indestmctability for Accessible Cardinals N

j(x) > A * and ~ A 5 M,

is its critical point,

follows that

P, col(p,
M.

Extend

conclude that for some P-generic extension of

j to an embedding of

-G

M[j(col(p,
v[col(p,
po

is in

j[P]

be a

T

P

j(x)-directed-closed.

= {j(a):a <

A}

p

and

=

Sup(B)

of

p

M[j(colp,
pol/-j(p,

'j(T)ny = j"E'

)J

'.

As

in this model

A

of

p

disjoint from

is a closed unbounded subset of X

M[j(colp,
E

is not stationary in A .

M[col( p,
E

was stationary in

is x-closed,

P

E

A.

j"E.

there is a

It is not hard to check

disjoint from

E,

therefore

On the other hand, in Now we get the desired contra-

is a x-closed extension of

diction by recallingthat j(colp,
G

has cofinality p

therefore it follows that already in M[j(col(p,
closed cofinal subset H

In

I f this is not the case, let

I .

is not stationary in

col(p,
Therefore,

E G and

'j"E

j-'(€i)

is

then in M[col(p,
and

M[j(COl(p,
is j(x)-closed in

po

p

such that

in

E 6)

of size < A * < j(x)

name for a stationary

pok j(p) 'j"E is stationary in

(as j(H)>A)

j[P]

-

is

j(P)

that

-

so G ={j(p):p

M

be a generic set for

j[P]

is a x-closed forcing

E j[P] above all members of G (in V[j(col(p,
let

,B

M, jlP

is realised as a stationary subset E

CLAIM

is a

= jot), V= M[co~(~,
I.(!

i , M[j(col(p,
is a directed subset of

is x-directed-closed

there is some

T

M[COl(p,
so

M[COl(p,
By the closure properties of

and as

It

M.

M[j(col(p,
Applying Observation 2 . 2 to our case

in

E

to M[j(COl(p,
j(col(p,
implies A

and all their subsets and the names in their

forcing languages are members of V[COl(p,
5 A*

15

is forced to be a stationary subset of (s?,\s;)

its

stationariness cannot be destroyed by such forcing extensions (Thm. 2 . 3 ) . M [ j ( c o l ( p , c ~ ) ) l * [ j ( P ) ] I ~ " c o f j2( ~l(X) )

of this model and has cardinality p and

jot) >

A,

so

j"(h)

is bounded in

j"E is a stationary initial segment of implies therefore that in

"

(as cof(X) 2 K ) j"(A)

there, as in this model

j(A)). j(E).

V[col(p,
is a member

j(x)

is

The elementariness of E

p+

It follows that in this model j

has a stationary initial

segment.

Proof of the theorem forcing in

V'

model, where

Let

Q

be a x-directed-strongly-strategically-closed

Let p be COl(H,<6) in the sense of this 6 is some cardinal above 2IQ1. P is x-directed closed, so =

V[col(p,
S. Ben David

16

by the previous lemma in

By Lemma 3 . 3 x-closed.

x

S 5 S,,

and

Sx

and


V'[hl.

S\S*(A)

As

-

for some P

such that in

is stationary, then

is H-closed and in

It follows that in

The same a

and

S

S

is

is still stationary in

for some a < A Sna

V'[$]

V'[Q]

is a stationary

cof(A) 1 H

V"Q1

is

S\S*(A)

a.

is stationary in

It follows from Theorem 2 . 3 that if in V"Q1

subset of

a.

A, Sna

is V'[Q][P]

V'[h]

V'[Ql[Fl = V'[h]. in

cof X 2 x

if

then for some a <

A,

stationary in

V'[fi],

is therefore a reflection point for

S

the same holds

is stationary in

in

V"Q1.

The following theorem is a straightforward corollary to Theorem 3 . 1 . THEOREM 3 . 4

If

ZFC

cardinal then so is

is consistent with the existence of a supercompact ZFC with "For every H-directed-strongly-strategically-

closed forcing notion cof(X) 2 H

g

after forcing via

and every stationary subset

is stationary then

S

g

for every ordinal

S of

S
n

reflects", this for every

A

s.t.

if {aEs:cof(a)+ # cofll successor of a regular

cardinal.

4 . APPLICATIONS

If

THEOREM 4 . 1

V

V[&l

" ZFC + a-many supercompact cardinals ", 6 such that cof(6) = a there is a forcing extension

is a model of

then for every ordinal

adding no a-sequences to

V,

in which every stationary subset of

K6+ 1

reflects.

Proof.

Let

(H.:i < a )

be an increasing sequence of supercompact cardinals.

lim N ~ ,and let (xi:i < a ) a i
Let

H

=

partial orders Pi; Po in

V[hil

and

ei:i < p) for g.

)I.

is colt h,,

Pi+, = Pi*Qi+l.

followed by

col( Xp

0'

'"0.

be an increasing sequence of regular By induction on Qi+,

i < a

define

is col(Xi,
In limit stages P p is the inverse limit of

,
P

).

Let

g

= Pa

and let b be generic

Standard arguments about the Levy collapse show that in

V[i]

is a cardinal, lim xi = 8, and X z is a cardinal, so it is i

each

b+,-

Hi+,-directed-closed forcing notion (for any that any

A -sequence of V[hl -

Pi, P,-generic).

is an element of

V

is a

It follows

and in particular

p adds

no &sequences. Let

R

be the forcing notion described in.section1 for adding a 0'

#6 sequence to V[h]. -

Recall that

R

preserves cardinalities and cofinalities.

A Laver-Type Indestmctability for Accessible Cardinals

v[~I[1 ~ I"Every stationary subset of K , + ~ reflects".

CLAIM 4 . 2

Proof of the claim S . = {p

E

S 5

Let

K,+,

S:cof(p) < A , }

The function f:S + a

be stationary.

X.1

f ( p ) = min{i:cof(p) <

defined by Let

17

is pressing down so for some

is stationary.

Let u s show that

i < a

Si reflects.

Qi denote the V[il

Levy collapse colOti, <%+I). i V[g]*[Qi] = VIPil*[P1l*[Qi], and P *Qi isa K.-closed forcing notion (as an By Theorem 2 . 4 , as R

iteration of two x.-closed notions). strategically-closed in V[Pi], closed

Q.

is x.-strongly-

can be regarded as R*Q*

for some

K.-

Q*.

In V[P]*[R],S*(K6+1)= p'mod the non stationary ideal, so therefore without

xs+l1.

loss of generality Si 5 K,+,\S*( that

Si

Applying Theorem 2 . 4 , we conclude

is still a stationary subset of

(k,+ljJ[pR1 is

an ordinal of cofinality xl.

[pR1 &+' , 1

reflects".

some p <

sinp i s stationary in p .

is stationary in

p

REMARK 4 . 3 in V[p]

as V[pR]

(Xi:i < a)

and

=

such that for some Qi:

(a) Qi

Sup 1 i
every stationary subset of If

ZFC

s:ii

A+

is a forcing notion

.

%Xi '

reflects.

is consistent with the existence of arbitrarily many

X,

existence of a non-reflecting stationary subset of

Oi

Oi

a forcing notion BX

without collapsing cardinals or changing cofinalities and

such that for each regular the demands (a) and (b) for

Assume

does not imply the

X.

In [B,S861 we constructed for every singular X

that introduces

THEOREM 4 . 5

V[pRlk"Sinp

p is a Levy collapse of superthen if R

supercompact cardinals then, for every singular

Proof.

reflects

In particular, for

is X.-closed ;

then in V[pR] COROLLARY 4 . 4

(though

V[pQi].

does not destroy stationarity of subsets of

(b) R*Qi

cof(h,+l tv[pRl

For the same p ,

is a submodel of

What we actually proved was that if

compact cardinals

V[_p][Qi]

Col(Ki,<&+l)

every stationary subset of

and so "every stationary subset of

'I

in

makes the old K,+,
no longer a cardinal).

corollary (b) in V[p][Qi]

(%+l) VIP1[R1 -

ZFC

1-(

<

X there is some forcing notion Qp satisfying

p,X

of Remark 4 . 3 .

is consistent with the existence of a proper class

of supercompact cardinals.

Then

ZFC

is consistent with "For every cardinal

every stationary subset of { a < X:cof(a)+ # A}

reflects".

S. Ben David

18

REMARK

(a) The result is the strongest that can be expected in the sense

4.6

X

that if

= p',

p

a regular cardinal, then

{a < X:cof(a)

= p} is a non-

A.

reflecting stationary subset of

( b ) The assumption is a 'natural' one in the sense that reflection of

Xt

stationary subsets of

implies strong failure of KJx

(even 0

Xp - a - fails for all p < Xi.

sequence defined for ordinals of cofinalities 2 p

In [B,S86] it is shown that 0 is consistent with the existence of superXlJ compact cardinals K s.t. cof(X) < K < p ) . It follows that a model of ZFC with full reflection as above has an inner model with proper class of large cardinals.

Proof.

Let

be a model of

V

(xCY:aE On,a successor).

ZFC with unboundedly-many supercompactcardinals

Without loss of generality every limit of super-

compact cardinals is a singular cardinal. regular limit and replace V

( ~ ~ E: Ona) notion (1)

is

H(6)

by

of

(If not, let

is increasing and continuous.

p so that: For each CY p

= P *Pa

where

is

Pa

6 be the first such

Let us also assume that

V.)

We define a proper class forcing

K

~

-+ C.C. ~

(and a set) and

Pa

K -directed-strongly-strategically-closed.

( 2 ) If

p

is a successor of a regular cardinal in

V[i],

then in

V

p

is one of the supercompacts in our sequence. 1 3 ) If in

V"1

p

is p'

is regular, then for some a E On

p

and

Pa+, = Pa*co~(~t<~CYtl) *re c~l(p.
tion.

p is by induction on a.

The definition of

where

= Pa *Col(K

pa+ 1 sense of

v[pa].

For a limit a

? ~ = v ' ~ = s KU p is the PA p

P

a+1

Col(Xt,
(P :B < a). It is quite B p does satisfy our list of demands. (One

is the inverse limit of

straightforward to check that this has to use the inaccessibility of the closure property of

Y

= P *R *Col(X+,
forcing notion defined in section 1 for intro-

VIPal [RCYl.

At limit a's, P

K+

For successor a

Col(K , < H ~ + ~ isthe ) Levy collapse in the

Pa

are still cardinals in

REMARK

K

~

for a < y

+to~ get the

for P "a+1 - c.c* to verify that for limit y K

Y

and and

V"].)

4.7 'Actually we get a somewhat stronger result in

v"].

Not only

do we have the full reflection but for every p-directed-stronqly-strategicallyclosed

R

in

V[i][fi]

we still have relfection for the stationary subsets of

ordinals of cofinality less than

p.

A Laver-Type Indestructability for Accessible Cardinals Added in Proof. questions:

The proofs of Theorems 4 . 1 and 4.5 raise the following

Could we get the models of reflection by just Levy collapsing the

supercompact cardinals of the ground model? 0:

19

Do we really have to force the

sequences to get f u l l reflection of stationary sets? Shela has given a partial answer to these questions by proving that one can

force over the ground model

V

the supercompact cardinals of V

v*

to get an intermediate extension V' are still supercompact in

V'

so

that

and the model

obtained by iteratively Levy collapsing the supercompact cardinals of V'

(in the manner described in the proofs of Theorems 4 . 1 and 4 . 5 ) has the desired reflection properties.

S. Ben-David, S. Shelah, Non-special Aronszajn trees on Journal of Mathematics 53, ( 1 9 8 6 ) 93-96.

,fi

Israel

R. Laver, Making the supercompactness of x indestructible under x-directed closed forcings, Israel Journal of Mathematics 29, ( 1 9 7 8 ) 385-388.

M. Magidor, Reflecting stationary sets, Journal of Symbolic Logic 47, 4,

( 1 9 8 2 ) 755-771.

S. Shelah, On successors of singular cardinals, in M. Boffa, D. Van Dalen, K. McAloon (eds) Logic Colloquium '78, North Holland ( 1 9 7 9 )

357-380.