Está en la página 1de 8

7

5
8


r
e
v
i
s
i
o
n
:
2
0
0
1
-
1
2
-
1
3







m
o
d
i
f
i
e
d
:
2
0
0
1
-
1
2
-
1
4


NOWHERE PRECIPITOUSNESS OF THE
NON-STATIONARY IDEAL OVER T

Yo Matsubara
1
and Saharon Shelah
2
Abstract. We prove that if is a strong limit singular cardinal and a regular
uncountable cardinal < , then NS

, the non-stationary ideal over P, is nowhere


precipitous. We also show that under the same hypothesis every stationary subset
of P can be partitioned into
<
disjoint stationary sets.
1. Introduction
Throughout this paper we let denote an uncountable regular cardinal and
a cardinal . Let NS

denote the non-stationary ideal over T

. NS

is the
minimal -complete normal ideal over T

. If X is a stationary subset of T

, then
NS

[X denotes the -complete normal ideal generated by the members of NS

and T

X. We refer the reader to Kanamori [6, Section 25] for basic facts about
the combinatorics of T

.
Large cardinal properties of ideals have been investigated by various authors.
One of the problems studied by these set theorists was to determine which large
cardinal properties can NS

or NS

[X bear for various , and X T

. In
the course of this investigation, special interest has been paid to two large cardinal
properties, namely precipitousness and saturation.
If NS

[X is not precipitous for every stationary X T

, then we say that


NS

is nowhere precipitous. In [8] Matsubara and Shioya proved that if is a


strong limit singular cardinal and cf < , then NS

is nowhere precipitous. In
2 we extend this result by showing that NS

is nowhere precipitous if is a
strong limit singular cardinal.
In [10] Menas conjectured the following:
Menas Conjecture. Every stationary subset of T

can be partitioned into


<
disjoint stationary sets.
This conjecture implies that NS

[X cannot be
<
-saturated for every station-
ary X T

. By the work of several set theorists we know that Menas Conjecture


is independent of ZFC. One of the most striking results concerning this conjecture
is the following theorem of Gitik [4].
1
The rst author was partially supported by Grant-in-Aid for Scientic Research (No.11640112),
Ministry of Education, Science and Culture of Japan.
2
The second author was partially supported
by The Israel Science Foundation funded by the Israel Academy of Sciences and Humanities. Pub-
lication 758
Typeset by A
M
S-T
E
X
1
7
5
8


r
e
v
i
s
i
o
n
:
2
0
0
1
-
1
2
-
1
3







m
o
d
i
f
i
e
d
:
2
0
0
1
-
1
2
-
1
4


2 YO MATSUBARA
1
AND SAHARON SHELAH
2
Gitiks Theorem. Suppose that is a supercompact cardinal and > . Then
there is a p.o. P that preserves cardinals such that
P
X (X is a stationary
subset of T

X cannot be partitioned into


+
disjoint stationary sets).
In 2 we also show that if is a strong limit singular cardinal, then every
stationary subset of T

can be partitioned into


<
disjoint stationary sets. Gitik
[4] mentions that GCH fails in his model of a non-splittable stationary subset
of T

. Our result shows that GCH must fail in such a model of a non-splittable
stationary subset of T

if is singular.
We often consider the poset P
I
of I-positive subsets of T

i.e. subsets of T

not belonging to I, ordered by


X
PI
Y X Y.
We say that an ideal I is proper if P
I
is a proper poset. In [9] Matsubara proved
the following result:
Proposition. Let be a cardinal 2
2
2

. If there is a proper
+
-complete
normal ideal over T

+ then NS
1
is precipitous.
It is not known whether NS

can be precipitous for singular . In [1] it is con-


jectured that NS

cannot be precipitous if is singular. Therefore it is interesting


to ask the following question:
Question. Can T

bear a proper -complete normal ideal where is the suc-


cessor cardinal of a singular cardinal?
In 3 we give a negative answer to this question.
2. On NS

for strong limit singular


We rst state our main results.
Theorem 1. If is a strong limit singular cardinal, then NS

is nowhere pre-
cipitous.
Theorem 2. If is a strong limit singular cardinal, then every stationary subset
of T

can be partitioned into


<
disjoint stationary sets.
One of the key ingredients of our proof of the main results is Lemma 3. Part
(ii) of Lemma 3 was proved in Matsubara [7]. Part (i) appeared in Matsubara-
Shioya [8]. For the proof of Part (ii) we refer the reader to Kanamori [6, page 345].
However we will present the proof of (i) because the idea of this proof will be used
later.
Lemma 3. If 2
<
<
<
= 2

, then
(i) NS

is nowhere precipitous
(ii) every stationary subset of T

can be partitioned into


<
disjoint stationary
sets.
Before we present the proof of part (i), we make some comments concerning this
lemma. First note that the hypothesis of our lemma is satised if is a strong limit
cardinal with cf < . Secondly under this hypothesis every unbounded subset of
T

must have a size of 2

. We also note that Lemma 3 can be generalized in the


following manner:
7
5
8


r
e
v
i
s
i
o
n
:
2
0
0
1
-
1
2
-
1
3







m
o
d
i
f
i
e
d
:
2
0
0
1
-
1
2
-
1
4


NOWHERE PRECIPITOUSNESS OF THE NON-STATIONARY IDEAL OVER P 3
For an ideal I over some set A, we let non(I) = min[X[ [ X A, X / I and
cof(I) = min[J[ [ J I, X I, Y J (X Y ). The proof of Lemma 3
actually shows that if non(I) = cof(I) then I is nowhere precipitous (i.e. for every
I-positive X, I[X is not precipitous) and every I-positive subset X of A can be
partitioned into non(I) many disjoint I-positive sets.
Proof of Lemma 3 (i). For I an ideal over T

, let G(I) denote the following game


between two players, Nonempty and Empty: Nonempty and Empty alternately
choose I-positive sets X
n
, Y
n
T

respectively so that X
n
Y
n
X
n+1
for
n = 1, 2, . . . . After moves, Empty wins G(I) if

n{0}
X
n
= . See [3] for a
proof of the following characterization.
Proposition. I is nowhere precipitous if and only if Empty has a winning strategy
in G(I).
Let f

[ < 2

) enumerate functions from


<
into T

. For a function
f :
<
T

, we let C(f) = s T

[

f

s
<
s. For X T

, X is
stationary if and only if C(f

) X ,= for every < 2

.
We now describe Emptys strategy in G(NS

) using the hypothesis 2


<
<

<
= 2

. Suppose that X
1
is Nonemptys rst move. Choose s
1

[ < 2

), a
sequence of elements of X
1
by induction on in the following manner: Let s
1
0
be
any element of X
1
C(f
0
). Suppose we have s
1

[ < ) for some < 2

. Since
s
1

[ < is a non-stationary, in fact bounded, subset of T

, X
1
s
1

[ <
is stationary. Pick an element from (X
1
s
1

[ < ) C(f

) and call it s
1

.
Let Empty play Y
1
= s
1

[ < 2

. It is easy to see that Y


1
is a stationary subset
of T

. Inductively suppose Nonempty plays his n + 1-st move X


n+1
immediately
following Emptys n-th move Y
n
= s
n

[ < 2

. Choose s
n+1

[ < 2

), a
sequence from X
n+1
in the following manner: Let s
n+1
0
be any element of (X
n+1

s
n
0
) C(f
0
). Suppose we have s
n+1

[ < ), for some < 2

. Pick an element
of the stationary set (X
n+1
C(f

)) (s
n+1

[ < s
n

[ ) and call it
s
n+1

. Let Empty play Y


n+1
= s
n+1

[ < 2

. This denes a strategy for Empty.


Claim. The strategy described above is a winning strategy for Empty.
Proof of Claim. Suppose X
1
, Y
1
, X
2
, Y
2
, . . . is a run of the game G(NS

) where
Empty followed the above strategy. We want to show that

n{0}
Y
n
= .
Suppose otherwise. Let t be an element of

n{0}
Y
n
. Then for each m 0,
there is a unique ordinal
m
< 2

such that s
m
m
= t. But by the way the s
n

s are
chosen, s
0
0
= s
1
1
= s
2
2
= implies
0
>
1
>
2
> . This is impossible.
Thus we must have

n{0}
Y
n
= .
End of proof of Lemma 3 (i).
We now prove Theorem 2 using Lemma 3 and Theorem 1.
Proof of Theorem 2. Let be a strong limit singular cardinal. If cf < then by
Lemma 3 (ii), we are done. So assume cf . In this case we have
<
= .
Therefore it is enough to show that NS

[X is not -saturated for every stationary


X T

. But this is a consequence of NS

being nowhere precipitous. In fact


we know that NS

[X cannot be
+
-saturated for every stationary X T

.
We need some preparation to present the proof of Theorem 1. Let be a strong
limit singular cardinal and be a regular uncountable cardinal < . If cf <
then by Lemma 3 we conclude that NS

is nowhere precipitous.
7
5
8


r
e
v
i
s
i
o
n
:
2
0
0
1
-
1
2
-
1
3







m
o
d
i
f
i
e
d
:
2
0
0
1
-
1
2
-
1
4


4 YO MATSUBARA
1
AND SAHARON SHELAH
2
From now on let us assume that is a strong limit cardinal with cf < .
Let

[ < cf ) be a continuous increasing sequence of strong limit singular


cardinals converging to with
0
> cf . The following lemma is another key
ingredient of our proof.
Lemma 4. For every X T

, if for each < cf with cf < , [t X [


sup(t) =

[ < 2

, then X is non-stationary.
Proof of Lemma 4. Since t X [ sup(t) / t is a club subset of T

, without loss
of generality we may assume that sup(t) / t for every t in X. For each < cf
with cf < , we let X

= t X [ sup(t) =

. We need the following fact from


pcf theory by S. Shelah.
Fact. There is a club subset C cf such that pp(

) = 2

for every C.
The proof of the above fact can be obtained from 5.15 of [12] or by combining
Conclusion XI 5.13 [11, page 414], Corollary VIII 1.6(2) [11, page 321], and Conclu-
sion II 5.7 [11, page 94]. [12] contains updates and corrections to [11]. The reader
can look at Holz-Steens-Weitz [5] for the pcf theory, particularly Theorem 9.1.3
[5, page 271].
For each C with cf < , let a

be a set of regular cardinals conal in

such that
(a) every member of a

is above cf
(b) [a

[ = cf

, and
(c)

> [X

[ [

pcf (a

)]
Let a =

a

[ C cf < . Let f

[ < ) enumerate all of the


members of f [ f is a function, domain(f) is a bounded subset of , and f is
regressive i.e. f() < for every domain(f).
For each t T

we dene g
t


a by letting g
t
() = supf

() + 1 [
t dom(f

), if

t
domain(f

), and g
t
() = 0 otherwise. Note that
[t[ < cf < min(a) guarantees g
t

a. Now by (c) in the denition of a

s
and the fact that g
t
a

[ t X

is a subset of

a

of cardinality [X

[ <

pcf(a

), there is some h

such that t X

[g
t
a

<
J
<
(a)
h

].
(For the denition of J
<
(a

), we refer the reader to section 3.4 of [5].) Therefore


(1) t X

[g
t
() < h

()]
holds. As min(a) > cf and a =

[ C cf < , there is h

a such
that h

< h a

for every C with cf < .


Let W = t T

[ (i) for some C sup(t) =

with cf < , and (ii) if


t then for some t, h (a ) = f

. Note that W is a club subset of T

.
Claim. X W = .
Proof of Claim. Suppose otherwise, say t X W. By (i) in the denition of W,
t X

for some C with cf < . By (1) we have


(2) a

[g
t
() < h

()].
Since sup(t) =

, there must be some t such that > . Now by (ii) in the


denition of W, h (a ) = f

for some t. Since a , h() = f

().
By the denition of g
t
we have f

() < g
t
(). From h

< h a

, we know
h

() < h(). Therefore we have h

() < g
t
() contradicting (2).
7
5
8


r
e
v
i
s
i
o
n
:
2
0
0
1
-
1
2
-
1
3







m
o
d
i
f
i
e
d
:
2
0
0
1
-
1
2
-
1
4


NOWHERE PRECIPITOUSNESS OF THE NON-STATIONARY IDEAL OVER P 5
End of proof of Lemma 4.
For each < cf with cf < , let us x a sequence f

[ < 2

) that
enumerates members of f [ f is a function such that domain(f)
<

and
range(f)

. Furthermore for each function f with domain(f)


<

and
range(f)

, we let C

[f] = t T

[ t
<
domain(f), sup(t) =

, and t is
closed under f. We need the following lemma to present the proof of Theorem 1.
Lemma 5. Suppose X is a stationary subset of T

. For every Y s T

[
s , if for each < cf with cf < the following condition () holds, then
Y is stationary.
() < 2

([C

[f

] X[ = 2

[f

] Y ,= )
Proof of Lemma 5. Since s for every s Y , to show that Y is stationary
it is enough to show that Y C[g] ,= for every function g :
<
where C[g]
denotes the set t T

[ g

t
<
t. For the proof of this fact, we refer the reader
to Foreman-Magidor-Shelah [2, Lemma 0]. Let us x a function g :
<
. Now
we let E = < cf [ cf < and for each E we let W

= s T

[
sup(s) =

/ s. Note that

E
W

is a club subset of T

. For each
E, we let g

denote g (
<

). Now partition E into two sets E


+
and E

where
E
+
= E [ [C

[g

] X[ = 2

and
E

= E [ [C

[g

] X[ < 2

.
We need the following:
Claim. X

[ E

is non-stationary.
Proof. It is enough to show that Z = C[g] X

[ E

is non-stationary.
Note that for each E
+
, ZW

= and for each E

, ZW

[g

] X.
Therefore [Z W

[ < 2

for every E. Hence, by Lemma 4, we conclude that


Z is non-stationary.
From Claim we know that X

[ E
+
is stationary. Pick an element

from E
+
. Consider the partial function g

(= g (
<

)). Let

< 2

be such that f

= g

. Since

E
+
, we have [C

[g

] X[ = 2

. Since
f

= g

and Y satises condition (), we know that C

[g

] Y ,= . Therefore
C[g] Y ,= showing that Y is stationary.
End of proof of Lemma 5.
Finally we are ready to complete the proof of Theorem 1. To present a winning
strategy for Empty in the game G(NS

), we introduce some new types of games.


For each E = < cf [ cf < , we dene the game G

between Nonempty
and Empty as follows: Nonempty and Empty alternately choose sets X
n
, Y
n

W

= s T

[ sup(s) =

/ s respectively so that X
n
Y
n
X
n+1
and
< 2

([C

[f

] X
n
[ = 2

[f

] Y
n
,= ) for n = 1, 2, . . . . Empty wins
G

n{0}
Y
n
= .
By the same argument as the proof of Lemma 3 (i), we know that Empty has
a winning strategy, say

, in the game G

for each E. Now we show how to


7
5
8


r
e
v
i
s
i
o
n
:
2
0
0
1
-
1
2
-
1
3







m
o
d
i
f
i
e
d
:
2
0
0
1
-
1
2
-
1
4


6 YO MATSUBARA
1
AND SAHARON SHELAH
2
combine the strategies

s to produce a winning strategy for Empty in G(NS

).
Suppose X
1
is Nonemptys rst move in G(NS

). We let X

1
= X
1
s T

[
s

[ E. Since s T

[ s

[ E is a club
subset of T

, X

1
is stationary in T

. For each E, we simulate a run of the


game G

as follows: Let us pretend that Nonemptys rst move in G

is X

1
W

.
Let Empty play her strategy

, so Emptys rst move is

(X

1
W

)). Now in
the game G(NS

), let Empty play Y


1
=

(X

1
W

)) [ E. Lemma 5
guarantees that Y
1
is stationary in T

. In general if X

1
, Y
1
, X
2
, Y
2
, . . . , X
n
) is
a run of G(NS

) up to Nonemptys n-th move, then we let Empty play Y


n
=

(X

1
W

, X
2
W

, . . . , X
n
W

)) [ E. Once again we know Y


n
is a
stationary subset of X
n
. For each E, since

is a winning strategy in G

we
have

n{0}

(X

1
W

, X
2
W

, . . . , X
n
W

)) = .
Because the W

s are pairwise disjoint, we conclude that



n{0}
Y
n
= . There-
fore we have a winning strategy for Empty in the game G(NS

). This proves that


NS

is nowhere precipitous for every strong limit singular .


End of proof of Theorem 1.
3. On proper ideals over T

First we dene that we mean by a proper ideal.


Denition. An ideal I over a set A is a proper ideal if the corresponding p.o. P
I
is proper (in the sense of proper forcing).
We refer the reader to Shelah [13] for the background of properness.
As we mentioned in 1, we are interested in the question of whether it is possible
to have a -complete normal proper ideal over T

where is the successor of


some singular cardinal. We give a negative answer to this question. Here we present
a more general result.
Theorem 6. (i) Suppose I is a -complete normal ideal over . If < [ cf =
/ I for some cardinal satisfying
+
< , then I is not proper.
(ii) Suppose I is a -complete normal ideal over T

. If s T

[ cf(s ) =
/ I for some cardinal satisfying
+
< , then I is not proper.
Note that if is the successor cardinal of a singular cardinal, then every -
complete normal ideal over T

satises the hypothesis of (ii).


Proof of Theorem 6. Since the proof of (ii) is identical to that of (i), we only present
the proof of (i).
Let I and be as in the hypothesis of (i). First note that if =
0
then the set
< [ cf = forces cf =
0
showing P
I
cannot be proper. Therefore we
may assume that is uncountable.
We need the following claim:
Claim 1. There are a stationary subset E of < [ cf =
0
and an I-positive
subset X of < [ cf = such that E is non-stationary for every in X.
Proof. Let E

[ <
+
be a family of pairwise disjoint stationary subsets of
< [ cf =
0
. For each < with cf = , there must be a club subset of
7
5
8


r
e
v
i
s
i
o
n
:
2
0
0
1
-
1
2
-
1
3







m
o
d
i
f
i
e
d
:
2
0
0
1
-
1
2
-
1
4


NOWHERE PRECIPITOUSNESS OF THE NON-STATIONARY IDEAL OVER P 7
with cardinality . Therefore for such an ordinal , there is some

<
+
such
that E

is non-stationary. By the -completeness of I, there is some

<
+
such that X = < [ cf =

/ I. If we let E = E

, then E is
non-stationary for every in X.
For each from X, let c

be a club subset of with c

E = . Let

C denote
c

[ X). Let be a large enough regular cardinal. Assume that N is a


countable elementary substructure of H(), ) satisfying I, E, X,

C N and
sup(N ) E.
We are ready to show that I is not proper.
Claim 2. If Y is a subset of X such that Y / I (therefore Y P
I
and Y X),
then Y is not (N, P
I
)-generic.
Claim 2 implies that P
I
is not proper.
Proof of Claim 2. Suppose otherwise. Assume that there exists Y X such that
Y is (N, P
I
)-generic.
For each < we dene a function f

: X by f

() = Min(c

) if
> , and f

() = 0 otherwise. It is clear that f

N for each N .
For each < , we let T

= X [ f

() = . For each xed < ,


using the normality of I, we see that T

[ < , T

/ I is a maximal
antichain below X in P
I
. Let

T

= T

[ < , T

/ I). It is clear that

N for N .
Since Y is (N, P
I
)-generic, for N T

[ N T

/ I is
predense below Y in P
I
. So we must have Y

[ N T

/
I I for each N . Let Y

= Y

[ N T

/ I. We
have

N
Y

I. This implies Y

N
Y

/ I. Let

be an element of
Y

N
Y

with

> sup(N ). Note that

Y Y

for each N .
Hence if N , then there exists

N such that

. Thus
f

) =

N for each N . This means that Min(c

) N
for each N , showing c

N is unbounded in sup(N ).
Since sup(N ) <

, we must have sup(N ) c

. But this implies


sup(N) c

E which contradicts c

E = for each X and

Y X.
This contradiction shows that Y cannot be (N, P
I
)-generic.
End of proof of Theorem 6.
References
1. D. Burke and Y. Matsubara, The extent of strength of the club lters, Israel Journal of
Mathematics 114 (1999), 253263.
2. M. Foreman, M. Magidor and S. Shelah, Martins Maximum, saturated ideals, and non-regular
ultralters. Part I, Annals of Mathematics 127 (1988), 147.
3. F. Galvin, T. Jech and M. Magidor, An ideal game, Journal of Symbolic Logic 43 (1978),
284292.
4. M. Gitik, Nonsplitting subset of P(
+
), Journal of Symbolic Logic 50 (1985), 881894.
5. M. Holz, K. Steens and E. Weitz, Introduction to Cardinal Arithmetic, Birkhauser, 1999.
6. A. Kanamori, The Higner Innite, Springer-Verlag, 1994.
7. Y. Matsubara, Consistency of Menas conjecture, Journal of the Mathematical Society of
Japan 42 (1990), 259263.
8. Y. Matsubara and M. Shioya, Nowhere precipitousness of some ideals, Journal of Symbolic
Logic 63 (1998), 10031006.
7
5
8


r
e
v
i
s
i
o
n
:
2
0
0
1
-
1
2
-
1
3







m
o
d
i
f
i
e
d
:
2
0
0
1
-
1
2
-
1
4


8 YO MATSUBARA
1
AND SAHARON SHELAH
2
9. Y. Matsubara, Proper ideals over P.
10. T. Menas, On strong compactness and supercompactness, Annals of Mathematical Logic 7
(1974), 327359.
11. S. Shelah, Cardinal Arithmetic, Oxford Science Publications, 1994.
12. S. Shelah, Analytical guide and updates for cardinal arithmetic E-12.
13. S. Shelah, Proper and Improper Forcing, Springer-Verlag, 1998.