7

5
5


r
e
v
i
s
i
o
n
:
2
0
0
2
-
0
4
-
0
3







m
o
d
i
f
i
e
d
:
2
0
0
2
-
0
4
-
0
3


WEAK DIAMOND
SAHARON SHELAH
Abstract. Under some cardinal arithmetic assumptions, we prove that
every stationary subset of of a right conality has weak diamond.
This is a strong negation of uniformization. We then deal with a weaker
version of the weak diamond that involves restricting the domain of the
colourings. We then deal with semi- saturated (normal) lters.
Key words and phrases. Set theory, Normal ideals, Weak diamond, precipituous lters,
semi saturated lters .
Research supported by the United States-Israel Binational Science Foundation Publi-
cation 755. corections after proof reading to the journal.
1
7
5
5


r
e
v
i
s
i
o
n
:
2
0
0
2
-
0
4
-
0
3







m
o
d
i
f
i
e
d
:
2
0
0
2
-
0
4
-
0
3


2 SAHARON SHELAH
Annotated Content
1. Weak Diamond: sucient condition
[We prove that if = 2

=
<
is weakly inaccessible,
= : = cf() < and < [[
tr,
< and S < : cf()
is stationary then it has weak diamond. We can omit or weaken the demand
=
<
if we restrict the colouring F (in the denition of the weak diamond)
such that for

, F() depends only on C

where C

, =
|C

|
].
2. On versions of precipitousness
[We show that for successor >

, the club lter on is not semi-

saturated (even every normal lter concentrating on any S I[] of co-
nality from a large family). Woodin had proved D

2
+ S
2
0
consistently is
semi-saturated].
7
5
5


r
e
v
i
s
i
o
n
:
2
0
0
2
-
0
4
-
0
3







m
o
d
i
f
i
e
d
:
2
0
0
2
-
0
4
-
0
3


WEAK DIAMOND 3
1. Weak Diamond: sufficient condition
On the weak diamond see [DvSh 65], [Sh:f, Appendix 1], [Sh 208], [Sh 638];
there will be subsequent work on the middle diamond.
Denition 1.1. For regular uncountable ,
(1) We say S is small if it is F-small for some function F from
>

to 0, 1, which means
()
F,S
for every c
S
2 there is

such that S : F(
) = c

is not stationary.
(2) Let D
wd

= A : A is small , it is a normal ideal (the weak

diamond ideal).
Claim 1.2. Assume
(a) =
<
= 2

(b) = : = cf() and for every < , we have [[

<>
< or just
[[
<tr,>
< (see below; so if >

every large enough regular

<

is in )
(c) S < : cf() , and

divides is stationary.
Then S is not in the ideal D
wd

of small subsets of .
Denition 1.3. (1) Let

= Min[T[ : T []

and every A []

is included in the union of < members of T.

(2)

tr
= sup[lim

(t)[ : t is a tree with nodes and levels

Remark 1.4. (1) On

tr
see [Sh 589], on

see there and in [Sh 460]

but no real knowledge is assumed here.
(2) The interesting case of 1.2 is (weakly) inaccessible; for successor
we know more; but in later results even if 2

is successor we say on
it new things.
(3) Actually only F (

mark. ??
Proof. Let F be a function from

to 0, 1, i.e., F is a colouring, and

we shall nd f
S
as required for it.
Let
i
: i < list

<

such that
< lg(
i
)
i

j
: j < i.
For S let T

=

: ( < )(
i
: i < ).
Clearly S [T

[ [[
<tr,>
< by assumption (c). For each T

we dene h

2 by: h

() = F(g
,
) where for < , we let g
,


2
be dened by g
,
() = ( + ) for < , recalling that

divides as
S. So h

: T

is a subset of

2 of cardinality [T

[ < = 2

hence we can choose g

2 g

: T

. For < let f

S
2 be
f

() = 1g

(). If for some < the function f

serve as a weak diamond

7
5
5


r
e
v
i
s
i
o
n
:
2
0
0
2
-
0
4
-
0
3







m
o
d
i
f
i
e
d
:
2
0
0
2
-
0
4
-
0
3


4 SAHARON SHELAH
sequence for F, we are done so assume that (for each < ) there are

and E

such that:
(a) E

is a club of .
(b)

.
(c) if E

S then F(

) = 1 f

() and

.
Now dene

2 by ( + ) =

() for < , < .

Let E = < : is divisible by

and < E

and ( < )[

i
: i < ]. Clearly E is a club of hence we can nd E S. So
by the denition of T

we have T

and for < we have g

,

is equal to

(Why? note that = as E and see the denition of

g
,
and of , so : < g
,
() = (+) =

()). Hence h

2
is well dened and by the choice of we have < g
,
=

so by
its denition, h

for each < satises h

() = F(g
,
) = F = (

).
Now by clause (c) and the choice of f

we have F(

) = 1f

() = g

()
so h

= g

, but h

whereas we have chosen g

such that g

/ T

, a
contradiction.
We may consider a generalization.
Denition 1.5. (1) We say

C is a Wd-parameter if:
(a) is a regular uncountable,
(b) S a stationary subsets of ,
(c)

C = C

: S), C

(1A) We say

C is a (, , ) -Wd-parameter if in addition (
S)[cf() = [C

[ < ]. We may also say that

C is (S, , )-
parameter.
(2) We say that F is a

C -colouring if:

C is a -Wd-parameter and F is
a function from
>
to 2 such that :
if S,
0
,
1


and
0
C

=
1
C

then F(
0
) = F(
1
).
(2A) If

C = : S) we may omit it writing S - colouring
(2B) In part (2) we can replace F by F

: S) where F

:
(C

)
2
such that

S F() = F

( C

). So abusing
notation we may write F( C

)
(3) Assume F is a

C-clouring,

C a -Wd-parameter.
We say c
S
2 (or c

2) is an F-wd-sequence if :
(*) for every

, the set S : F( ) = c

is a stationary
subset of .
We also may say c is an (F, S)-Wd-sequence.
(3A) We say c
S
2 is a D F-Wd-sequence if D is a lter on to
which S belongs and
(*)for every

we have
S : F( ) = c

, = modD
(4) We say

C is a good -Wd-parameter, if for every < we have
> [C

: S and C

[.
7
5
5


r
e
v
i
s
i
o
n
:
2
0
0
2
-
0
4
-
0
3







m
o
d
i
f
i
e
d
:
2
0
0
2
-
0
4
-
0
3


WEAK DIAMOND 5
Similarly to 1.2 we have
Claim 1.6. Assume
(a)

C is a good (, , )-Wd-parameter.
(b) [[
tr,
< for every < .
(c) = 2

and =
<
(d) F is a

C- colouring.
Then there is a F-Wd-sequence.
Proof. Let cd be a 1-to-1 function from

onto , for simplicity, and without
loss of generality
= cd(

: < )) sup

: <
and let the function cd
i
: for i < be such that cd
i
(cd(

: <
))) =
i
.
Let T = : for some C of cardinality < , we have
C
, so
by assumption (c) clearly [T[ = , so let us list T as

: < with no
repetitions, and let T
<
=

: < . For S let T

= : a function
from C

to such that for every C

we have (C

) T
<
.
By

C being good and clause (b) of the assumption necessarily T

has
cardinality < . For each T

and < we dene

,

C

by

,
() = cd

(()) for C

. Now for T

, clearly

=: F(
,
) :
< ) belongs to

2. Clearly

: T

is a subset of

2 of cardinality
[T

[ which as said above is < . But [

2[ = 2

= by clause (c) of the

assumption, so we can nd

: T

.
For each < we can consider the sequence c

= 1

() : S) as
a candidate for being an F -Wd-sequence. If one of then is, we are done.
So assume toward contradiction that for each < there is

which
exemplify its failure, so there is a club E

of such that

1
S E

F(

) ,= c

and without loss of generality

2
< E

() < .
But c

= 1

() and so z 0, 1 &z ,= c

z =

() hence
we have got

3
S E F(

) =

()
Dene

by

() = cd(

() : < )), now as is regular

uncountable clearly E =: < : for every < we have

() < and
if

S, C

= C

& C

then

T
<
is a club of (see the
choice of T, T
<
, recall that by assumption (a) the sequence

C is good, see
Denition 1.5(4)).
Clearly E

= E

: < E is a club of . Now for each E

S,
clearly

; just check the denitions of T

and E, E

. Now recall

,
is the function from C

to dened by

,
() = cd

()).
7
5
5


r
e
v
i
s
i
o
n
:
2
0
0
2
-
0
4
-
0
3







m
o
d
i
f
i
e
d
:
2
0
0
2
-
0
4
-
0
3


6 SAHARON SHELAH
But by our choice of

clearly cd

()) =

(), so
C

,
() =

() so

,
=

,
Hence F(

C
,
) = F(

), however as E

clearly F(

) =

(), together F(

,
) =

().
As

clearly

2, moreover for each < we know
that

(), see its denition above, is equal to F(

,
) which by the
previous sentence is equal to

(). As this holds for every < and

are members of

2, clearly they are equal. But

so

: T

whereas

has been chosen outside this set,

contradiction.
Well, are there good (, , )-parameters? (on I[] see [Sh 420, 1]).
Claim 1.7. (1) If S is a stationary subset of the regular cardinal and
S I[] and ( S)cf() = then for some club E of , there is
a good (S E, , )-parameter.
(2) If = cf(),
+
< = cf() then there is a stationary S I[] with
( S)[cf() = ].
Proof. (1) By the denition of I[]
(2) By [Sh 420, 1].

We can note
Claim 1.8. (1) Assume the assumption of 1.6 or 1.2 with C

= and
D is a
+
- complete lter on , S D, and D include the club lter.
Then we can get that there is a D F-Wd-sequence.
(2) In 1.6, we can weaken the demand = 2

to = cf(2

) that is,
assume
(a)

C is a good (, , )-Wd-parameter.
(b) [[
tr,
< 2

for every < .

(c) = cf(2

) and 2

= (2

)
<
(d) F is a

C-colouring
(e) D is a
+
-complete lter on extending the club lter to which
Dom(

C) belongs.
Then
1
there is a D F-Wd-sequence.
(3) In 1.6+1.8(2) we can omit regular.
Proof. (1) The same proof.
(2) Let H

: 2

be increasing continuous with unbounded range

and let S I[] be stationary, such that ( S)cf() = , and

C = C

: S) is a good (cf(), , )-Wd- parameter, let

S

= h

() : S, C

()
= h

() : C

,

C

= C

: S

)
1
in fact if = cf(2

) < 2

then the demand

C is good is not necessary; see more in
[Sh 775]
7
5
5


r
e
v
i
s
i
o
n
:
2
0
0
2
-
0
4
-
0
3







m
o
d
i
f
i
e
d
:
2
0
0
2
-
0
4
-
0
3


WEAK DIAMOND 7
and repeat the proof using

= 2

,

C

= C

: S

) instead ,

C.
Except that in the choice of the club E we should use E

= < :
for every Rang (h

) we have

() < and is a limit

ordinal and

= C

T
<
.
(3) Similarly.

This lead to considering the natural related ideal.

Denition 1.9. Let

C be a (, , )- parameter.
(1) For a family T of

C-colouring and T

2, let id
C,F,P
be
W : for some F T for every c T for some

the set
W S : F( C

) = c

is not stationary.
(2) If T is the family of all

C- colouring we may omit it. If we write Def
instead T this mean as in [Sh 576, 1].
We can strengthen 1.6 as follows.
Denition 1.10. We say the -colouring F is (S, )- good if:
(a) S < : cf() < is stationary
(b) we can nd E and C

: S E) such that
() E a club of .
() C

is an unbounded subset of , [C

[ < .
() if ,

, S E, and

= f C

then F(

) = F()
() for every < we have
> [C

: S E and C

[
() S [
cf ()
tr
] :
Claim 1.11. Assume
(a) = cf(2

)
(b) F is an (S, )- good -colouring.
Then there is a (F, S)-Wd-sequence, see Denition 1.5(3).
Remark 1.12. So if = cf(2

) and we let

=: = cf() and ( <

)([[
tr,
< ) then
(a)

large (e.g. contains every large enough Reg

if

< )
and
(b) if = cf()
+
< then there is a stationary S I[] such that
S cf() = .
(c) if , S are as above then there is a good C

: S)
(d) for , S,

C as above, if F = F

: S) and F

() depend just on
C

and D is a normal ultralter on (or less), and lastly S D

then there is an D F-Wd-sequence; see Denition 1.5(3A).
7
5
5


r
e
v
i
s
i
o
n
:
2
0
0
2
-
0
4
-
0
3







m
o
d
i
f
i
e
d
:
2
0
0
2
-
0
4
-
0
3


8 SAHARON SHELAH
2. On versions of precipitousness
Denition 2.1. (1) We say the D is (P, D

) -precipituous if
(a) D is a normal lter on , a regular uncountable cardinal.
(b) P is forcing notion with
P
minimal.
(c) D

a P-name of an ultralter of the Boolean Algebra T()

(d) letting for p P
D
p,D

=: A : p A D

we have:
() D

P
,D

= D and
() D
p,D

is normal lter on
(e)
P
V

/D

is well founded.
(2) For regular uncountable and D a normal lter on let NOR
D
=
D

: D

a normal lter on extending D ordered by inclusion and

D

= D

: D

NOR
D

Woodin [W99] dened and was interested in semi-saturation for =

2
,
where!.
(1A) If D

is clear from the context (as in part (2)) we may omit D

.
Denition 2.2. For regular uncountable cardinal, a normal lter D on
is called semi-saturated when for every forcing notion P and P-name D

of
a normal (for regressive f V) ultralter on T()
V
, we have: D is (P, D

)-
precipitous.
Woodin proved Con(D

2
S
2
0
is semi saturated), he proved that the
existence of such lter has large consistency strength by proving 2.3 below.
This is related to [Sh:g, V].
Claim 2.3. If =
+
, D a semi-saturated lter or , then every f

is
<
D
- than the -th function for some <
+
(on the -th function see e.g
[Sh:g, XVII, 3])
In fact
Claim 2.4. If =
+
and D is NOR

-precipitous then every f

is
<
D
- smaller than the -th function for some <
+
Proof. The point is that
(a) if D is a normal lter on , f

: <
+
) is
<
D -increasing in and
f

, <
+
(f
D
f

) then there is a normal lter D

1
on
extending D such that <
+
f

<
D
1
f
(b) if f

:
+
) is <
D
- increasing f

, and =
+
and
X = < : cf(f

+()) = , = mod D then there are functions

g
i


for i < such that g
i
< f

+mod(D + X), and ( <

+
)(i < )(g
i
<
D
f

).
[In detials let = (D
1
, f, ) : D
1
NOR

, f ,

, D
1

NOR

f/D

is
the -the ordinal in V

/D

and f f

mod D
1
for <
+
, for some
7
5
5


r
e
v
i
s
i
o
n
:
2
0
0
2
-
0
4
-
0
3







m
o
d
i
f
i
e
d
:
2
0
0
2
-
0
4
-
0
3


WEAK DIAMOND 9
f

:< D
1
- increasing with . If the conclusion fails then ,= 0,
choose (D
1
, f, ) with minimal and by clause (a) without loss of
generality <
+
f

< fmodD
1
. By (b) there is g < fmodD
1
such
that <
+
(g < f

modD
1
), without loss of generality <
+
f

<
gmod D
1
and for some < and D
2
NOR

extending D
1
, D
2

NOR

g/D

is the the ordinal of V

/D

, contradiction to the minimality of ]

Claim 2.5. (1) If =

+

then the club lter on is not semi-

saturated.
(2) If =
+

then for every large enough regular <

, there
is no semi-saturated normal lter D

on to which S

= < :
cf() = belongs.
(3) If 2
2

< =
+
> = cf() >
0
and for every f

we have
rk
J
bd

(f) < then there is no semi-saturated normal lter D

on
to which < : cf() = belongs.
(4) In 1), 2), 3), if D is Nor
D
-semi-saturated then the conclusion
holds for D.
Remark: We can replace

by any strong limit uncountable cardinal.

Proof. (1) Follows by (2)
(2) By [Sh 460] for some
0
<

, for every regular (

0
,

) we
have:

= , see 1.3. Let D = A : sup( A) < .

By part (3) it is enough to prove
if f

then rk
D
(f) <
proof of If not then for every < there is
f

such that f

<
D
f and rk
D
(f) =
and dene
D

=: A : A D or A / D, and rk
D+(\A)
(f

) < .
This is a -complete lter on see [Sh 589]. So for some D

the set
A = : D

= D

is unbounded in . By [Sh 589, 4] (alterna-

tively use [Sh:g, V] on normal lters)
(*) for < from A, f

<
D
f

and D

is a -complete lter on
.
But as =

letting

= sup(Rang(f)) + 1 which is < , so

[

[ , there is a family T [

such that for every a [

,
for some i() < and a
i
T for i < i() we have a

i<i()
a
i
hence
for every A, for some a

T we have
i < : f

(i) a

, = mod D

.
So for some a

and unbounded B A we have B a

= a

and
moreover for some b

we have B b

= i < : f

(i) a

and moreover B f

= f

. But this contradict (*).

7
5
5


r
e
v
i
s
i
o
n
:
2
0
0
2
-
0
4
-
0
3







m
o
d
i
f
i
e
d
:
2
0
0
2
-
0
4
-
0
3


10 SAHARON SHELAH
(3) We can nd u
,
: < , <
+
) such that:
(a) u

, : < ) is -increasing continuous such that [u

,
[ < ,
and u
,
: < = .
(b) if < <
+
and u
,
then u
,
= u
,
.
Let f

be f

() = otp(u
,
), so it is well known that f

/D

is the -th function, in particular < f

<
D

where D

is the club lter on ; in fact < <

+
f

<
J
bd

. Choose
2

C = C

: S

), C

a club of of order type , and let g

enumerate C

, i.e. g

(i) is the i-th member of C

For < let g

be constantly , and let g

be dened
by g

() = rk
J
bd

(g

)
()
0
g

and g

()
[why? by an assumption]
For <
+
we dene f

by:
f

() =

rk
J
bd

(f

) if S

0 if S

Note that f

is a function from to .
Now
()
1
f

for <
+
[Why? as f

, so by a hypothesis rk
J
bd

(f

) < ]
()
2
for <
+
()
2

= < : if < then f

() <
is a club of
[Why? Obvious]
()
3
for <
+
we have
E

() < g

(), so f

<
D

[Why? the rst statement by the denition of E

, of f

and of
g

(). The second by the rst ()

0
.]
()
4
if < <
+
then f

<
J
bd

hence f

<
D

[Why? the rst as f

<
J
bd

hence for some < , we have

< < f

() < f

() hence S

( + 1)
f

<
J
bd
C

<
J
bd

rk
J
bd

(f

) < rk
J
bd

(f

)
f

() < f

()
Let f

+
=: g

, so
()
+
f

and <
+
f

<
D

This of course suces by ??.

2
recall S

= { < : cf() = }
7
5
5


r
e
v
i
s
i
o
n
:
2
0
0
2
-
0
4
-
0
3







m
o
d
i
f
i
e
d
:
2
0
0
2
-
0
4
-
0
3


WEAK DIAMOND 11
(4) The same proof.

Remark: In the proof of 2.5(2) it is enough that U

J
bd

() = (see [Sh 589]).

[References of the form math.XX/ refer to arXiv.org ]
References
[DvSh 65] Keith J. Devlin and Saharon Shelah. A weak version of which follows from
2

0
< 2

1
. Israel Journal of Mathematics, 29:239247, 1978.
[Sh 638] Saharon Shelah. More on Weak Diamond. The East-West Journal of Mathe-
matics, accepted. math.LO/9807180.
[Sh 208] Saharon Shelah. More on the weak diamond. Annals of Pure and Applied Logic,
28:315318, 1985.
[Sh 420] Saharon Shelah. Advances in Cardinal Arithmetic. In Finite and Innite Com-
binatorics in Sets and Logic, pages 355383. Kluwer Academic Publishers, 1993.
N.W. Sauer et al (eds.). 0708.1979.
[Sh:g] Saharon Shelah. Cardinal Arithmetic, volume 29 of Oxford Logic Guides. Ox-
ford University Press, 1994.
[Sh:f] Saharon Shelah. Proper and improper forcing. Perspectives in Mathematical
Logic. Springer, 1998.
[Sh 589] Saharon Shelah. Applications of PCF theory. Journal of Symbolic Logic,
65:16241674, 2000.
[Sh 460] Saharon Shelah. The Generalized Continuum Hypothesis revisited. Israel Jour-
nal of Mathematics, 116:285321, 2000. math.LO/9809200.
[Sh 576] Saharon Shelah. Categoricity of an abstract elementary class in two successive
cardinals. Israel Journal of Mathematics, 126:29128, 2001. math.LO/9805146.
[Sh 775] Saharon Shelah. Middle Diamond. Archive for Mathematical Logic, 44:527560,
2005. math.LO/0212249.
Institute of Mathematics The Hebrew University of Jerusalem Jerusalem
91904, Israel and Department of Mathematics Rutgers University New Brunswick,
NJ 08854, USA
E-mail address: shelah@math.huji.ac.il
URL: http://www.math.rutgers.edu/~shelah