Andrzej Roslanowski and Saharon Shelah - Lords of The Iteration

8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

LORDS OF THE ITERATION
ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
Abstract. We introduce several properties of forcing notions which imply
that their support iterations are proper. Our methods and techniques
rene those studied in [RS01], [RS07], [RS05] and [RS], covering some new
forcing notions (though the exact relation of the new properties to the old
ones remains undecided).
0. Introduction
Since the beginning of 1980s it has been known that the theory of proper forcing
does not admit naive generalization to the context of larger cardinals and iterations
with larger supports. The evidence of that was given already in Shelah [She82] (see
[She98, Appendix 3.6(2)]). It seems that the rst steps towards developing the
theory of forcing iterated with uncountable supports were done in Shelah [She03a],
[She03b], but the properties introduced there were aimed at situations when we
do not want to add new subsets of (corresponding to the case of no new reals
in CS iterations of proper forcing notions). Later Roslanowski and Shelah [RS01]
introduced an iterable property called properness over semi-diamonds and then
Eisworth [Eis03] proposed an iterable relative of it. These properties work nicely
for support iterations (where =
<
is essentially arbitrary) and forcings adding
new subsets of , but the price to pay is that many natural examples are not covered.
If we restrict ourselves to inaccessible , then the properties given by Roslanowski
and Shelah [RS07, RS05, RS] may occur useful. Those papers give both iteration
theorems and new examples of forcing notions for which the theorems apply.
In the present paper we further advance the theory and we give results applicable
to both the case of inaccessible as well as those working for successor cardinals.
The tools developed here may be treated as yet another step towards comparing and
contrasting the structure of

with that of

. That line of research already has
received some attention in the literature (see e.g., Cummings and Shelah [CS95],
Shelah and Spasojevic [SS02] or Zapletal [Zap97]). Also with better iteration the-
orems one may hope for further generalizations of Roslanowski and Shelah [RS99]
to the context of uncountable cardinals. (Initial steps in the latter direction were
presented in Roslanowski and Shelah [RS07].) However, while we do give some
examples of forcing notions to which our properties apply, we concentrate on the
Date: July 2010.
1991 Mathematics Subject Classication. Primary 03E40; Secondary:03E35.
Key words and phrases. iterated forcing, support, iteration theorems,
The rst author would like to thank the Hebrew University of Jerusalem and the Lady Davis
Fellowship Trust for awarding him with Schonbrunn Visiting Professorship under which this
research was carried out.
Both authors acknowledge support from the United States-Israel Binational Science Foundation
(Grant no. 2002323). This is publication 888 of the second author.
1
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

2 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
development of the theory of forcing leaving the real applications for further in-
vestigations. The need for the development of such general theory was indirectly
stated by Hytinnen and Rautila in [HR01], where they commented:
Our proof is longer than the one in [MS93] partly because we are
not able to utilize the general theory of proper forcing, especially the
iteration lemma, but we have to prove everything from scratch.
We believe that the present paper brings us substantially closer to the right general
iteration theorems for iterations with uncountable supports.
In the rst section we introduce Tparameters (which will play an important
role in our denitions) and a slight generalization of the Bbounding property from
[RS05]. We also dene a canonical example for testing usefulness of our iteration
theorems: the forcing Q
E
E
in which conditions are complete trees in which along
each -branch the set of splittings forms a set from a lter E (and the splitting at
is into a set from a lter E
on ). The main result of the rst section (Theorem

1.10) says that we may iterate with supports forcing notions Q
E
E
, provided is
inaccessible and E is always the same and has some additional properties.
If we want to iterate forcing notions like Q
E
E
but with dierent E on each coor-
dinate (when the result of the rst section is not applicable), we may decide to use
very orthogonal lters. Section 2 presents an iteration theorem 2.7 which is tailored
for such situation. Also here we need the assumption that is inaccessible.
The following section introduces Bnoble forcing notions and the iteration the-
orem 3.3 for them. The main gain here is that it allows us to iterate (with
supports) forcing notions like Q
E
E
even if is not inaccessible. The fourth section
gives more examples of forcing notions and shows a possible application. In Corol-
lary 4.5 we substantially improve a result from [RS05] showing that dominating
numbers associated with dierent lters may be distinct even if is a successor.
The fth section shows that some of closely related forcing notions may have
dierent properties. Section 6 presents yet another property that is useful in
support iterations (for inaccessible ): reasonably merry forcing notions. This
property has the avour of putting together being Bbounding (of [RS05]) with
being fuzzy proper (of [RS07]). We also give an example of a forcing notion which
is reasonably merry but which was not covered by earlier properties. We conclude
the paper with a section listing open problems.
This research is a natural continuation of papers mentioned earlier ([She03a],
[She03b], [RS01], [RS07], [RS05] and [RS]). All our iteration proofs are based on
trees of conditions and the arguments are similar to those from the earlier works.
While we tried to make this presentation self-contained, the reader familiar with
the previous papers will denitely nd the proofs presented here easier to follow
(as several technical aspects do re-occur).
0.1. Notation. Our notation is rather standard and compatible with that of clas-
sical textbooks (like Jech [Jec03]). In forcing we keep the older convention that a
stronger condition is the larger one.
(1) Ordinal numbers will be denoted be the lower case initial letters of the Greek
alphabet (, , , . . .) and also by i, j (with possible sub- and superscripts).
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

LORDS OF THE ITERATION 3
Cardinal numbers will be called , , ; will be always assumed to
be a regular uncountable cardinal such that
<
= (we may forget
to mention this).
Also, will denote a suciently large regular cardinal; H() is the family
of all sets hereditarily of size less than . Moreover, we x a well ordering
<
of H().
(2) We will consider several games of two players. One player will be called
Generic or Complete or just COM, and we will refer to this player as she.
Her opponent will be called Antigeneric or Incomplete or just INC and will
be referred to as he.
(3) For a forcing notion P, almost all Pnames for objects in the extension
via P will be denoted with a tilde below (e.g.,
, X
). There will be some

exceptions to this rule, however.
P
will stand for the canonical Pname
for the generic lter in P. Also some (names for) normal lters generated
in the extension from objects in the ground model will be denoted by D,
D
P
or D[P].
The weakest element of P will be denoted by
P
(and we will always
assume that there is one, and that there is no other condition equivalent to
it). All forcing notions under considerations are assumed to be atomless.
By support iterations we mean iterations in which domains of con-
ditions are of size . However, on some occasions we will pretend that
conditions in a support iteration

Q = P
, Q
: <
) are total functions

on
and for p lim(
Q) and
dom(p) we will let p() =
.
(4) By a sequence we mean a function dened on a set of ordinals (so the
domain of a sequence does not have to be an ordinal). For two sequences
, we write whenever is a proper initial segment of , and
when either or = . The length of a sequence is the order type of
its domain and it is denoted by lh().
(5) A tree is a downward closed set of sequences. A complete tree is a
tree T
<
such that every -chain of size less than has an -bound
in T and for each T there is T such that .
Let T be a tree. For T we let
succ
T
() = < :
) T and (T)
= T : or .
We also let root(T) be the shortest T such that [succ
T
()[ > 1 and
lim
(T) =

: ( < )( T).
0.2. Background on trees of conditions.
Denition 0.1. Let P be a forcing notion.
(1) For a condition r P, let
0
(P, r) be the following game of two players,
Complete and Incomplete:
the game lasts at most moves and during a play the
players construct a sequence (p
i
, q
i
) : i < ) of pairs of
conditions from P in such a way that (j < i < )(r
p
j
q
j
p
i
) and at the stage i < of the game, rst
Incomplete chooses p
i
and then Complete chooses q
i
.
Complete wins if and only if for every i < there are legal moves for both
players.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

(2) We say that the forcing notion P is strategically (<)complete if Complete
has a winning strategy in the game
0
(P, r) for each condition r P.
(3) Let N (H(), , <
) be a model such that

<
N N, [N[ = and
P N. We say that a condition p P is (N, P)generic in the standard
sense (or just: (N, P)generic) if for every Pname
N for an ordinal
we have p
N .
(4) P is proper in the standard sense (or just: proper) if there is x H()
such that for every model N (H(), , <
) satisfying
<
N N, [N[ = and P, x N,
and every condition q N P there is an (N, P)generic condition p P
stronger than q.
Remark 0.2. Let us recall that if P is either strategically (<
+
)complete or
+
cc,
then P is proper. Also, if P is proper then

+
is not collapsed in forcing by P, moreover
for every set of ordinals A V
P
of size there is a set A
+
V of size
such that A A
+
.
Denition 0.3 (Compare [RS07, Def. A.1.7], see also [RS05, Def. 2.2]).
(1) Let be an ordinal, ,= w . A (w, 1)
tree is a pair T = (T, rk) such

that
rk : T w ,
if t T and rk(t) = , then t is a sequence (t)
: w ),
(T, ) is a tree with root ) and
if t T, then there is t
T such that t t
and rk(t
) = .
(2) If, additionally, T = (T, rk) is such that every chain in T has a upper
bound in T, we will call it a standard (w, 1)
tree
We will keep the convention that T
x
y
is (T
x
y
, rk
x
y
).
(3) Let

Q = P
i
, Q
i
: i < ) be a support iteration. A tree of conditions in
Q is a system p = p
t
: t T) such that
(T, rk) is a (w, 1)
tree for some w ,

p
t
P
rk(t)
for t T, and
if s, t T, st, then p
s
= p
t
rk(s).
If, additionally, (T, rk) is a standard tree, then p is called a standard tree
of conditions.
(4) Let p
0
, p
1
be trees of conditions in

Q, p
i
= p
i
t
: t T). We write p
0
p
1
whenever for each t T we have p
0
t
p
1
t
.
Note that our standard trees and trees of conditions are a special case of that
[RS07, Def. A.1.7] when = 1.
Proposition 0.4. Assume that

Q = P
i
, Q
i
: i < ) is a support iteration such
that for all i < we have
Pi
Q
i
is strategically (<)complete .
Suppose that p = p
t
: t T) is a tree of conditions in

Q, [T[ < , and 1 P
is
open dense. Then there is a tree of conditions q = q
t
: t T) such that p q and
(t T)(rk(t) = q
t
1).
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

Proof. This is essentially [RS07, Proposition A.1.9] and the proof there applies here
without changes.
1. Tparameters
In this section we introduce Tparameters and we use them to get a possible
slight improvement of [RS05, Theorem 3.1] (in Theorem 1.10). We also dene our
canonical testing forcing Q
E
E
to which this result can be applied.
Denition 1.1. (1) A pre-Tparameter on is a triple p = (

P, S, D) =
(

P
p
, S
p
, D
p
) such that
D is a proper uniform normal lter on , S D,

P = P
: S) and P
]
<
for each S.
(2) For a function f

and a pre-Tparameter p = (

P, S, D) we let
set
p
(f) = S : f P
.
(3) We say that a pre-Tparameter p = (

P, S, D) is a Tparameter on if
set
p
(f) D for every f
.
Example 1.2. (1) If is strongly inaccessible, D is the lter generated by
club subsets of and P
=

,

P = P
: < ), then (

P, , D) is a
Tparameter on .
(2)
+
is a statement asserting existence of a Tparameter with the lter

generated by clubs of .
(3)
implies the existence of a Tparameter (

P, S, D) such that [P
[ = [[.
(4) For more instances of the existence of Tparameters we refer the reader
to Shelah [She00, 3].
Denition 1.3. Let p be a pre-Tparameter on and Q be a forcing notion not
collapsing . In V
Q
we dene
D
p
[Q] = D
p[Q]
is the normal lter generated by D
p
set
p
(f) : f
,
p[Q] = (

P
p
, S
p
, D
p[Q]
).
Remark 1.4. If Q is a strategically (<)complete forcing notion and D is a (proper)
normal lter on , then in V
Q
the normal lter on generated by D V is also a
proper lter. Abusing notation, we will denote this lter by D (or D
Q
). The lter
D
p
[Q] can be larger, but it is still a proper lter, provided p is a Tparameter.
Lemma 1.5. Assume that p = (

P, S, D) is a Tparameter on and Q is a
strategically (<)complete forcing notion. Then
Q
/ D
p
[Q]. Consequently,
Q
p[Q] is a Tparameter on .
Proof. Assume that p Q and A

is a Qname for an element of D V and f
is
a Qname for an element of

(for < ). Using the strategic completeness of Q
build a sequence p
, A
, f
: < ) such that for each < :

(i) p
Q, p p
0
p
for < ,
(ii) A
D V, f

and
(iii) p

Q
A
= A
and f
= f
for all .
Since p is a Tparameter, we know that B =
<
A

<
set
p
(f
) D. Let
B. Then
p

Q

<
A

and f
= f
for all < ,

8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

so p

Q

<
A

<
set
p
(f
) .
Lemma 1.6. Assume that
<
= , p = (

P, S, D) is a Tparameter on , Q is
a strategically (<)complete forcing notion and N (H(), , <
) is such that
p N, [N[ = and
<
N N. Let N
: < ) be an increasing continuous

sequence of elementary submodels of N such that p N
0
, N
, P
N
+1
,
N
: ) N
+1
and [N
[ < (for < ). Then
Q

_
A N
___
< : A N
N
+1
_
D[Q]
_
.
Proof. We may nd an increasing continuous sequence
: < ) and a
bijection f : N such that f[N
] =
and fN
N
+1
(for < ). For
A N let
A
: 2 be such that
A
() = 1 if and only if f
1
() A. Plainly,
if =
and
A
P
, then A N
N
+1
.
Let C
0
be a forcing notion consisting of all pairs (, f) such that < and
f
<
( +1) ordered by the extension (so (, f) (
, f
) if and only if f f
).
Thus it is a (<)complete forcing notion which is an incarnation of the Cohen
forcing notion.
Proposition 1.7. Assume is strongly inaccessible. If p = (

P, S, D) is a T
parameter on such that ( S)([P
[ [[), then
C
0
D
C
0
,= D[C
0
] .
Proof. Let f
be the canonical C
0
name for the generic function in
<
( + 1),
so (, f)
C
0
f f
. Plainly,
C
0
set
p
(f
) D[C
0
] and we are going to argue
that
C
0
set
p
(f
)
_
D
C
0
_
+
. To this end, suppose that p C
0
and A
is a
C
0
name for an element of D V (for < ). By induction on < choose
, B
, p
: < ) so that
()
: < ) is an increasing continuous sequence of ordinals below ,

() B
D, p
= p
:

<
(
+ 1)) C
0
, p
0
= p = (
0
, f
0
),
() if

<
(
+ 1), then p
= (
, f
) and f
) = () for < ,
() if <
<
+ 1) and =
, then p
,
() if < is limit,

<
(
+ 1), then (
= sup(
: < ) and)
f
<
f
, and
() p
+1

for every

+ 1).
(Remember that is inaccessible, so
<
(
+ 1)
< for each .) Next, consider

the set B =
<
B
D. Let BS be a limit ordinal. Since [P
[ [[ <
<
[
[,
we may pick

<
(
+1) such that f
/ P
. Then p

<
A

set
p
(f
).
Denition 1.8. Let p = (

P, S, D) be a Tparameter on , Q be a strategically
(<)complete forcing notion.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

(1) For a condition p Q we dene a game
rbB
p
(p, Q) between two players,
Generic and Antigeneric, as follows. A play of
rbB
p
(p, Q) lasts steps and
during a play a sequence
_
I
, p
t
, q
t
: t I
) : <
_
is constructed. Suppose that the players have arrived to a stage < of
the game. Now,
()
rst Generic chooses a set I
of cardinality < and a system p
t
: t
I
) of conditions from Q,
1
()
then Antigeneric answers by picking a system q
t
: t I
) of condi-
tions from Q such that (t I
)(p
t
q
t
).
At the end, Generic wins the play
, p
t
, q
t
: t I
) : <
_
of
rbB
p
(p, Q) if and only if
()
p
rbB
there is a condition p
Q stronger than p and such that

p
Q

_
< :
_
t I
__
q
t

Q
__
D[Q] .
(2) A forcing notion Q is reasonably Bbounding over p if for any p Q,
Generic has a winning strategy in the game
rbB
p
(p, Q).
Remark 1.9. The notion introduced in 1.8 is almost the same as the one of [RS05,
Denition 3.1(2),(5)]. The dierence is that in ()
p
rbB
we use the lter D[Q] and
not D
Q
= D, so potentially we have a weaker property here. We do not know,
however, if there exists a forcing notion which is reasonably Bbounding over p
and not reasonably Bbounding over D. (See Problem 7.1.)
In a similar fashion we may also modify the property of being nicely double
bbounding (see [RS, Denition 2.9(2),(4)]) and get the parallel iteration theorem.
Theorem 1.10. Assume that
(1) is a strongly inaccessible cardinal and p is a Tparameter on ,
(2)

Q = P
, Q
: < ) is a support iteration,

(3) for every < ,
P
Q
is reasonably Bbounding over p[P
] .
Then
(a) P
= lim(
Q) is proper,
(b) if
is a P
name for a function from to V, p P
, then there are q p

and A
: < ) such that ( < )([A
[ < ) and
q < :
() A
D
p
[P
] .
Proof. The proof is essentially the same as that of [RS05, Theorem 3.1] with a small
modication at the end (in Claim 3.1 there); compare with the proof of Theorem
2.7 here and specically with 2.7.1.
Denition 1.11. Let

E = E
:
<
) be a system of (<)complete non-
principal lters on and let E be a normal lter on . We dene a forcing notion
Q
E
E
as follows.
A condition p in Q
E
E
is a complete tree p
<
such that
for every p, either [succ
p
()[ = 1 or succ
p
() E
, and
for every lim
(p) the set < : succ

p
() E
belongs to E.
1
Note that no relation between p
t
and p
s
for < is required to hold.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

The order =
Q
E
E
is the reverse inclusion: p q if and only if (p, q Q
E
E
and )
q p.

E, E are as in 1.11. Let p = (

P, S, D) be a
Tparameter on such that S E.
(1) Q
E
E
is a (<)complete forcing notion of size 2
.
(2) Q
E
E
is reasonably Bbounding over p.
(3) If is strongly inaccessible and ( S)([P
[ [[), then
Q
E
E
D
Q
E
E
,=
D[Q
E
E
].
Proof. (1) Should be clear.
(2) Let p Q
E
E
. We are going to describe a strategy st for Generic in
rbB
p
(p, Q
E
E
).
In the course of the play, Generic constructs aside a sequence T
: < ) so that
if
, p
t
, q
t
: t I
) : <
_
is the sequence formed by the innings of the two
players, then the following conditions are satised.
(a) T
E
E
and if < < then p = T
0
T
and T
= T
.
(b) If < is limit, then T
<
T
.
(c) If S then
I
= P
and p
t
= (T
)
t
for t I
,
T
+1
=
t
: t I

_
(T
:

T
_
.
(d) If / S, then I
= and T
+1
= T
.
Conditions (a)(d) fully describe the strategy st. Let us argue that it is a winning
strategy and to this end suppose that
, p
t
, q
t
: t I
) : <
_
is a play of
rbB
p
(p, Q
E
E
) in which Generic uses st and constructs aside the sequence T
: < )
so that (a)(d) are satised. Put p
<
T

<
. It follows from (a)+(b) that
p
is a complete tree and for each p
either [succ
p
()[ = 1 or succ
p
() E
.
Suppose now that lim
(p
) and for < let B
def
= < : succ
T
()
E
. Since lim
(T
) for each < , we know that B
E. Let
B =
<
B
< : is limit and / S.

It follows from our assumptions that B E. For each B we know that
succ
T
() E
for < and T
<
T
, so succ
T
() E
. More-
over, T

+1
= T

+1
for all > (remember (a)+(d)) and consequently
succ
p
() = succ
T
() E
. Thus we have shown that p
E
E
.
Let W
be a Q
E
E
name given by
Q
E
E
W
root(p) : p
Q
E
E
. It should be
clear that
Q
E
E
W

and thus
Q
E
E
set
p
(W
) D[Q
E
E
]. Plainly, if S and
t
, then (p
)
t
q
t
and hence
p
E
E
if set
p
(W
), then W
and q
E
E
,
so Generic won the play.
(3) We are going to show that
Q
E
E
set
p
(W
) / D
Q
E
E
. To this end suppose that
p Q
E
E
and A

(for < ) are Q
E
E
names for elements of D. Let st be the winning
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

strategy of Generic in
rbB
p
(p, Q
E
E
) described in part (2) above. Consider a play
, p
t
, q
t
: t I
) : <
_
of
rbB
p
(p, Q
E
E
) in which
()
1
Generic follows st and constructs aside a sequence T
: < ),
()
2
Antigeneric plays so that at a stage S he picks a set B
D and
conditions q
t
p
t
(for t I
) such that
_
t I
__
q
t
B

_
.
Let p
<
T
be the condition determined at the end of part (2) and let B =
<
B
. Choose an increasing continuous sequence
: < ) and a complete

tree T p
such that for every < we have

()
3
if T
, then [ T
+1
: [ = [
[ and
()
4
if T
+1
, then P
for some (
,
+1
) S.
(The choice can be done by induction on ; remember that p is a Tparameter
and is assumed to be inaccessible.) Pick a limit ordinal B S such that
=
. Since [T
[ > [
[, we may choose T
. Put q = (p
. Then
q p
p and q
Q
E
E

<
A

set
p
(W
) (remember ()
2
+ ()
4
).
2. Iterations with lords
Theorem 1.10 can be used for -support iteration of forcing notions Q
E
E
when
on each coordinate we have the same lter E. But if we want to use dierent lters
on various coordinates we have serious problems. However, if we move to the other
extreme: having very orthogonal lters we may use a dierent approach to argue
that the limit of the iteration is proper.
Denition 2.1. (1) A forcing notion with complete (, )purity is a triple
(Q, ,
pr
) such that ,
pr
are transitive reexive (binary) relations on Q
such that
(a)
pr
,
(b) both (Q, ) and (Q,
pr
) are strategically (<)complete,
(c) for every p Q and a (Q, )name
for an ordinal below , there are

a set A of size less than and a condition q Q such that p
pr
q
and q forces (in (Q, )) that
A.
(2) If (Q, ,
pr
) is a forcing notion with complete (, )purity for every ,
then we say that it has complete (, )purity.
(3) If (Q, ,
pr
) is a forcing notion with complete (, )purity, then all
our forcing terms (like forces, name etc) refer to (Q, ). The relation
pr
has an auxiliary character only and if we want to refer to it we add
purely (so q is stronger than p means p q, and q is purely stronger
than p means that p
pr
q).
Denition 2.2. Let Q = (Q, ,
pr
) be a forcing notion with complete (,
+
)
purity, p = (

P, S, D) be a Tparameter on , | be a normal lter on and
=
: < ) be a sequence of cardinals below .

pr
U,p,
pr
U,p,
(p, Q) lasts steps
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

and during a play a sequence
_
, p
t
, q
t
: t
) : <
_
is constructed. So suppose that the players have arrived to a stage <
of the game. Now,
()
pr
rst Antigeneric pics
0, 1.
()
pr
After this, Generic chooses a system p
t
: t
) of paiwise incom-
patible conditions from Q, and
()
pr
Antigeneric answers with a system of conditions q
t
Q (for t
)
such that for each t
:
p
t
q
t
, and
if
= 1, then p
t

pr
q
t
.
_
, p
t
, q
t
: t
) : <
_
if and only if either < :
= 1 / |, or
()
p
pr

p
Q

_
< :
_
t
__
q
t

Q
_
D[Q] .
(2) We say that the forcing notion Q (with complete (,
+
)purity) is purely
B
bounding over |, p, if for any p Q, Generic has a winning strategy

in the game
pr
U,p,
(p, Q).
Remark 2.3. Note that in the denition of the game
pr
U,p,
(p, Q) the size of the
index set used at stage is declared to be
(while in the related game

rbB
p
(p, Q)
we required just [I
[ < ). The reason for this is that otherwise in the proof of the
iteration theorem for the current case we could have problems with deciding the
size of the set I
; compare clause ()
4
of the proof of Theorem 2.7.
Observation 2.4. Assume

E, E are as in 1.11. For p, q Q
E
E
let p
pr
q mean
that p q and root(p) = root(q). Then
(1) (Q
E
E
, ,
pr
) is a forcing notion with complete (,
+
)purity,
(2) if, additionally, each E
(for
<
) is an ultralter on , then (Q
E
E
,
,
pr
) has (, 2)purity for every < .

E, E are as in 1.11, p = (

P, S, D) is a T
parameter on and =
: < ) is a sequence of non-zero cardinals below

such that ( S)([P
). Then (Q
E
E
, ,
pr
) is purely B
bounding over
E, p, .
Proof. Let p Q
E
E
and let st be the strategy described in the proof of 1.12(2) with
a small modication that we start the construction with
0
= lh(root(p)) + 1 (so
T
0
= p and the rst
0
steps of the play are not relevant). Then we also replace
clauses (c)+(d) there by
(cd) If
0
then
I
is of size
, and p
t
= (T
)
t
for t I
, and
if S then P
, and
T
+1
=
t
: t I

_
(T
:

T
_
.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

(So, in particular, Antigenerics choice of
has no inuence on the answers by

Generic.) We are going to show that st is a winning strategy for Generic in
pr
E,p,
(p, Q
E
E
). To this end suppose that
, p
t
, q
t
: t I
) : <
_
is a play
of
pr
E,p,
(p, Q
E
E
) in which Generic follows st (we identify I
with [I
[ =
) and
T
: < ) is the sequence of side objects constructed in the course of the play.
Assume A = < :
= 1 E (otherwise Generic wins by default). Like in

1.12(2), put p
<
T
. To argue that p
E
E
we note that if lim
(p
) and

<
_
< : succ
T
() E
_
A < : >
0
is limit ,
then
succ
p
() =
_
succ
T
() if / I
succ
q
() if I
.
Exactly as in 1.12(2) we justify that p
witnesses ()
p
pr
.
Lemma 2.6. Assume that
(1) is strongly inaccessible,
(2)

Q = P
, Q
: < ) is a support iteration, w , [w[ < ,

0
w,
(3) for every < ,
P
Q
= (Q
, ,
pr
) is a forcing notion with complete (,
+
)purity ,
(4) P
0
is proper,
(5) T = (T, rk) is a standard (w, 1)
tree, [T[ < ,

(6) p = p
t
: t T) is a standard tree of conditions in P
, and
(7)
is a P
name for an ordinal.

Then there are a set A of size and a standard tree of conditions q = q
t
: t
T) P
such that
(a)
_
t T
__
rk(t) = q
t

A
_
, and
(b) p q and if t T, rk(t) >
0
then q
t0

P
0
p
t
(
0
)
pr
q
t
(
0
).
Proof. Let us start with the following observation.
Claim 2.6.1. If p P
then there are a set A

0
of size and a condition q p
such that q
P

A
0
and q
0

P
0
p(
0
)
pr
q(
0
).
Proof of the Claim. Let us look at P
as the result of 3 stage composition P

0

Q
0
P
(0+1),
, where P
(0+1),
is a P
0+1
name for the following forcing notion.
The set of conditions in P
(0+1),
is r(
0
, ) : r P
(so it belongs to V); the

order of P
(0+1),
is such that if G
0+1
P
0+1
is generic over V, then
V[G
0+1
] [= r
P
(
0
+1),
[G
0
]
s if and only if
there is q G
0+1
such that q
r
P
q
s .
Now, pick a P
0+1
name (r
) such that
p(
0
+ 1)
P
0
+1
p(
0
, ) r
and r
and then choose a P

0
name A
for a subset of P
(0+1),
ON and a P
0
name
q(
0
) for a condition in Q
0
such that
p
0

P
0
p(
0
)
pr
q(
0
) and [A
[ = and
q(
0
)
Q
0
_
(s, ) A
__
r
= s &
=
_
.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

Since P
0
is proper, we may choose a set A
+
P
(0+1),
ON of size and a
condition q
0
p
0
such that q
0
A
A
+
. Then
q(
0
+ 1)
P
0
+1
_
(s, ) A
+
__
r
= s &
=
_
.
Put A = : (s)((s, ) A
+
). Now we may easily dene q(
0
, ) so that
dom
_
q(
0
, )
_
=
dom(s) : ()((s, ) A
+
) and
q(
0
+ 1)
P
0
+1
r
(
0
+1),
q(
0
, ) and
A .
Fix an enumeration t
) of t T : rk(t) = (so
< ). For each

0
x a P
name st
for a winning strategy of Complete in the game
0
_
(Q
, ),
_
such that as long as Incomplete plays
, Complete answers
with
as well. Let st
pr
be the <
rst P
0
name for a winning strategy of
Complete in
0
_
(Q
0
,
pr
), p
t
(
0
)
_
(for
). Note that if ,
and
t
(
0
+ 1) = t
(
0
+ 1), then st
pr
= st
pr
.
By induction on
we choose a sequence p
, q
, A
) so that the
following demands are satised.
(i) p
= p
t
: t T), q
= q
t
: t T) are standard trees of conditions, A
is a
set of ordinals of size .
(ii) If <
, then p p
and A
.
(iii) p
.
(iv) If
0
, ,
, then
q

P
p
(), q
() : ) is a result of a play of
0
_
(Q
, ),
_
in which Complete uses st
.
(v) If ,
, then
q
P
0
p
0
(
0
), q
0
(
0
) : ) is a result of a play
of
0
_
(Q
0
,
pr
), p
t
(
0
)
_
in which Complete uses st
pr
.
Suppose that we have determined p
, q
, A
for <
. First we choose
p
= p
t
: t T) P
. If = 0 then we set p
= p. Otherwise we choose p
so that
for t T we have:
(vi) dom(p
t
) =
<
dom(q
t
), and
(vii) if dom(p
t
)
0
, then p
() is the <
rst P
name for a condition

in Q
such that p
t

P
( < )(q
t
() p
t
()), and
(viii) p
(
0
) is the <
rst P
0
name for a condition in Q
0
such that
p
0

P
0
( < )(q
t
(
0
)
pr
p
t
(
0
)).
The choice is possible by (iv)+(v), and since we pick the <
rst names we easily

see that p
is a standard tree of conditions. Now we use 2.6.1 to nd a set A
of
size and a condition p
such that
_
<
A
, p
, p
0

P
0
p
(
0
)
pr
p
(
0
) and p
.
Next, for each t T we let p
t
P
rk(t)
be such that
(ix) if s = t t
, then
p
t
rk(s) = p
rk(s) and p
t
[rk(s), rk(t)) = p
t
[rk(s), rk(t)).
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

Clearly, p
= p
t
: t T) is a standard tree of conditions satisfying the relevant
parts of the demands in (ii)(v). Now we choose a tree of conditions q
= q
t
: t
T) so that the requirements of (iv)+(v) hold (for this we proceed like in (vi)(viii)
above).
After the construction is carried out we note that q
and A
are as required
in the assertion of the lemma.
(1) is strongly inaccessible, =
: < ) is a sequence of cardinals below

, p = (

P, S, D) is a Tparameter on , and
(2)

Q = P
, Q
: < ) is a support iteration,

(3) |
is a P
name for a normal lter on (for < ),

(4) A
,
is such that
P
A
,
|
and
P
A
,
|
(for < <

), and
(5) for every < ,
P
Q
is purely B
bounding over |
, p[P
], .
Then P
= lim(
Q) is proper.
Proof. The arguments follow closely the lines of the arguments for [RS05, Thm.
3.1, 3.2] and [RS, Thm. 2.12]. The proof is by induction on , so assume that we
know also that each P
is proper for < .

Let N (H(), , <
) be such that
<
N N, [N[ = and

Q, A
,
: < <
), p, . . . N. Let p N P
and
: < ) list all P
names for ordinals from

N. For each N x a P
name st
N for a winning strategy of Complete

in
0
(Q
) such that it instructs Complete to play
as long as her opponent

plays
.
By induction on < we will choose
()
a
, p
, q
, r
, r
, w
, Z
, and
()
b

,
, p
,
, q
,
and st

for N ,
so that the following demands are satised.
()
0
All objects listed in ()
a
+ ()
b
belong to N. After stage < of the

construction, the objects in ()
a
are known as well as those in ()

b
for
w
.
()
1
r
, r
, r
0
(0) = r
0
(0) = p(0), w
, [w
[ = [[ + 1,
<
dom(r
) =
<
w
= N , w
0
= 0, w
w
+1
and if is limit then w
<
w
.
()
2
For each < < we have ( w
+1
)(r
() = r
()) and p r
.
()
3
If ( w
) N, then
r
the sequence r
(), r
() : ) is a legal partial play of
0
_
Q
_
in which Complete follows st

and if w
+1
w
, then st

N is a P
name for a winning strategy

of Generic in
pr
U
,p,
(r
(), Q
). (And st
0
N is a winning strategy of
Generic in
pr
U
0,p,
(p(0), Q
0
).)
()
4
T
= (T
, rk
) is a standard (w
, 1)
tree, T
.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

()
5
p
= p
,t
: t T
) and q
= q
,t
: t T
) are standard trees of conditions,

p
.
()
6
For t T
we have that dom(p
,t
) =
_
dom(p)

<
dom(r
) w
_
rk
(t)
and for each dom(p
,t
) w
:
p
,t

P
if the set r
() : < p() has an upper bound in Q
,
then p
,t
() is such an upper bound .
()
7
For N ,
,
0, 1 and p
,
, q
,
are P
names for sequences of

conditions in Q
of length
.
()
8
If either = 0 = or w
+1
w
, < , then

,
, p
,
, q
,
: < ) is a play of
pr
U
,p,
(r
(), Q
)
in which Generic uses st

.
()
9

will be called the lord of stage ) and

_
w
__
/
_
A
,
: w
_
A
,
: w
( + 1)
__
.
()
10

,
= 0 for N
and
,
= 1.
()
11
If t T
, rk
(t) = < , then for each <
,t

P
,
() = p
,t
() and q
,
() = q
,t
() .
()
12
If t
0
, t
1
T
, rk
(t
0
) = rk
(t
1
) and w
rk
(t
0
), t
0
= t
1
but
_
t
0
_
,=
_
t
1
_
, then
p
,t0

P
the conditions p
,t0
(), p
,t1
() are incompatible .
()
13
Z
is a set of ordinals, [Z
[ = and for each t T
with rk
(t) = we have
q
,t

P
_

__
_
.
()
14
dom(r
) = dom(r
) =
tT
dom(q
,t
) dom(p) and if t T
, dom(r
)
rk
(t) w
, and q
,t
q P
, r
q, then
q
P
if the set r
() : < q
,t
(), p() has an upper bound in Q
,
then r

First we x an increasing continuous sequence w
: < ) of subsets of N
such that the relevant demands in ()
1
are satised. Now, suppose that we have
arrived to a stage < of the construction and all objects listed in ()
a
and
relevant cases of ()
b
(see ()
0
) have been determined for < .
To ensure ()
0
, all choices below are made in N (e.g., each time we choose an
object with some properties, we pick the <
rst such object).

If is a successor ordinal and w
w
1
, then we let st

N be a P
name
for a winning strategy of Generic in
pr
U
,p,
(r
1
(), Q
). We also put
,
= 0 for
all < and we pick p
,
, q
,
(for < ) so that the suitable parts of ()
7
+()
8
at are satised.
Clause ()
4
fully describes T
. Now we choose the lord of stage . If for some

w
we have

_
A
,
: w
_
A
,
: w
( + 1)
_
,
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

then
is equal to this (note that there is at most one w
with the required

property). Otherwise we let
= 0. Then we put
,
= 1 and
,
= 0 for all
w
.
Next, for each w
we choose a P
name p
,
such that
,
= p
,
() : <
) is given to Generic by st

as an answer to
,
, p
,
, q
,
: < )
,
) .
(Note that
P
conditions p
,
(
0
), p
,
(
1
) are incompatible whenever
0
<
1
<
and w
.) After this we may choose a tree of conditions p
= p
,t
: t T
)
such that for each t T
:
dom(p
,t
) =
_
dom(p)

<
dom(r
) w
_
rk
(t) and
for dom(p
,t
) w
, p
,t
() is the <
rst P
name for a condition in

Q
such that
p
,t

P
if the set r
,
then p
,t
() is such an upper bound ,
p
,t
() = p
,
_
(t)
_
for dom(p
,t
) w
.
Using Lemma 2.6 we may pick a tree of conditions q
= q
,t
: t T
) and a set Z
of ordinals such that

p
, [Z
[ = ,
if t T
, rk
(t) = then q
,t

P
( )(
),
if t T
, rk
(t) >
then q
,t
,t
(
)
pr
q
,t
(
).
Note that if w
and
0
<
1
<
, t T
, rk
(t) = , then
q
,t

P
the conditions q
,t
0
(), q
,t
Hence we have no problems with nding P
names q
,
(for w
) such that

P
,
= q
,
() : <
) is a sequence of conditions in Q
,

P
( <
)( p
,
() q
,
()) and
P
( <
)( p
()
pr
q
()) ,
if t T
, rk
(t) > , then q
,t

P
,t
() = q
,
_
(t)
_
.
Now we dene r
, r
so that
dom(r
) = dom(r
) =
_
tT
dom(q
,t
) dom(p)
and
if w
+1
, < , then r
() = r
() = r
(),
if dom(r
) w
, then r
() is the <
rst P
name for an element of

Q
such that
r
() is an upper bound of r
() : < p() and

if t T
, rk
(t) > , and q
,t

P
and the set

r
() : < q
,t
,
then r
() is such an upper bound ,

and r
() is the <
rst P
name for an element of Q
such that
r
() is given to Complete by st
as the answer to
r
(), r
() : < )
())
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

Note that by a straightforward induction on dom(r
) one easily applies ()

3
from previous stages to show that r
, r
are well dened and r
, p for
< . If = 0 we also stipulate r
0
(0) = r
0
(0) = p(0).
This completes the description of stage of our construction. One easily veries
that the demands ()
0
()
14
are satised.
After completing all stages of the construction, for each N we look at
the sequence
,
, p
,
, q
,
: < ). For < such that w
let
B
_
A
,
: w
_
A
,
: w
( + 1)
_
,
and for < such that / w
put B
= . It follows from our assumptions that
( < )(B
) and thus also

P

<
B
. Note that if is a limit

ordinal, w
and
<
B
, then also B
and hence =
(remember
()
9
) and
,
= 1 (by ()
10
). Consequently for each N
< :
,
= 1 |
.
Therefore, for every N we may pick a P
name q() for a condition in Q
such that
if w
+1
w
, < (or = 0 = ), then
q() r
() and q()
Q
_
<:
_
<
__
q
,
()
Q
__
D[P
+1
] .
This determines a condition q P
(with dom(q) = N) and easily ( < )(p

r
q) (remember ()
2
). For each N x P
+1
names C
i
, f
i
(for i < )
such that
q( + 1)
P
+1

_
i <
__
C
i
D V & f
_
and
_

i<
C
i

i<
set
p
(f
i
)
__
<
__
q
,
()
Q
_
.
Claim 2.7.1. For each limit ordinal < , the condition q forces (in P
) that
_
w
__

i<
C
i

i<
set
p
(f
i
)
_

_
t T
__
rk
(t) = & q
,t

P
_
.
Proof of the Claim. The proof is essentially the same as that for [RS05, Claim 3.1],
however for the sake of completeness we will present it fully. Suppose that r q
and a limit ordinal < are such that
()
a
r
P
_
w
__

i<
C
i

i<
set
p
(f
i
)
_
.
For each < x a P
name st
for a winning strategy of Complete in
0
(Q
)
such that as long as Incomplete plays
, Complete answers with
as well.
We are going to show that there is t T
such that rk
(t) = and the condi-

tions q
,t
and r are compatible. Let
) = w
be the increasing
enumeration. By induction on
we will choose conditions r
, r
and
t = (t)
: <
) T
such that letting t
= (t)
: < ) T
we have
()
b
q
,t
and r
,
()
c
for every < and <
,
r
(), r
() :
< ) is a legal partial play of
0
(Q
)
in which Complete uses her winning strategy st
.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

Suppose that
is a limit ordinal and we have already dened t
= (t)
:
< ) and r
, r
: < ). Let = sup(
: < ). It follows from ()

c
that
we may nd a condition r
such that r
is stronger than all r
(for
< ) and also r
[,
) = r[,
). Clearly r
and also q
,t
(remember ()
b
for < ). Now by induction on [,
) we argue that
q
,t
. So suppose that <
and we know q
,t
. It
follows from ()
3
+()
5
+()
6
that r
(i < )(r
i
() p
,t
() q
,t
()) and
therefore we may use ()
14
to conclude that
r
,t
() r
() q() r() = r
().
Finally we let r
be a condition such that for each <
() is given to Generic by st
as the answer to
r
(), r
() : < )
()) .
Now suppose that = + 1
and we have already dened r
, r
and
t
. It follows from the choice of q and ()

a
+ ()
b
+ ()
11
that
r
r(
)
Q
_
<
__
q
,t
)
Q
_
.
Therefore we may choose = (t)
<
(thus dening t
) and a condition r
such that
r
and
r
) r(
) & r
) q
,t
) ,
r
) = r(
).
Exactly like in the limit case we argue that r
has the desired properties and then

in the same manner as there we dene r
.
We nish the proof of the claim noting that t = t
and the condition r
are such that r
r and r
,t
.
Let us use 2.7.1 to argue that q is (N, P
)generic. To this end suppose
N is a
P
name for an ordinal, say
, < , and let q
q. Since P
is strategically
(<)complete we may build an increasing sequence q
i
: i < ) of conditions above
q
and a sequence C
i
, f
i
: N , i < ) such that C
i
D V, f
i

and
for each w
i
q
i

_
j i
__
C
j
= C
j
& f
j
i = f
j
i
_
.
The set
_
< :
_
w
__
j <
__
C
j
set
p
(f
j
)
__
is in D, so we may choose
a limit ordinal > such that for each w
we have

i<
C
i

i<
set
p
(f
i
).
Then q

_
w
__

i<
C
i

i<
set
p
(f
i
)
_
and therefore, by 2.7.1,
q

_
t T
__
rk
(t) = & q
,t

P
_
.
Using ()
13
we conclude q
and hence q
N.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

Remark 2.8. Naturally, we want to apply Theorem 2.7 to =
++
in a model
where 2
=
+
, so one may ask if the assumptions (3) + (4) of 2.7 can be satised
in such a universe. But they are not so unusual: suppose that we start with
V [=
<
= & 2
=
+
. Consider the following forcing notion C
+
.
A condition p C
+
is a function p : dom(p) 2 such that dom(p)
+

and [dom(p)[ < .
The order is the inclusion.
Plainly, C
+
is a (<)complete
+
cc forcing notion of size
+
. Suppose now
that G C
+
is generic over V and let us work in V[G]. Put f =
G (so
f :
+
2) and for <
+
and i < 2 let A
i
= < : f(, ) = i.
For a function h :
+
2 let U
h
be the normal lter generated by the family
_
A
h()
: <
+
_
. One easily veries that each U
h
is a proper (normal) lter and
plainly if h, h
:
+
2 are distinct, say h() = 0, h
() = 1, then A
0
U
h
and
A
0
= A
1
U
h
.
3. Noble iterations
The iteration theorems 1.10 and 2.7 have one common drawback: they assume
that is strongly inaccessible. In this section we introduce a property slightly
stronger than being Bbounding over p and we show the corresponding iteration
theorem. The main gain is that the only assumption on is =
<
.
Denition 3.1. Let Q = (Q, ) be a forcing notion and p = (

P, , D) be a T
parameter on .
B+
p
B+
p
(p, Q) lasts steps
during which the players construct a sequence
, A
, p
, q
: <
_
such
that
(a) f
: Q and f
for < ,
(b) A
and for every A
the sequence f
_
()
_
: < ) Q
has an upper bound in Q and if
0
,
1
A
are distinct, then for some

< the conditions f
0
()
_
, f
1
()
_
are incompatible,
(c) p
= p
: A
) Q is a system of conditions in Q such that

_
<
__
f
_
()
_
p
_
for A
,
(d) q
= q
: A
) Q is a system of conditions in Q such that

( A
)(p
)
The choices of the objects listed above are done so that at stage < of
the play:
()
rst Generic picks a function f
: Q with the property described

in (a) above (so if is limit, then f
<
f
). She also chooses A
,
p
satisfying the demands of (b)+(c) (note that A
could be empty).
()
Then Antigeneric decides a system q
as in (d).
, A
, p
, q
: <
_
if and only if
()
B+
p

p
Q

_
< :
_
A
__
q

Q
__
D[Q] .
(2) A forcing notion Q is Bnoble over p if it is strategically (<)complete
and Generic has a winning strategy in the game
B+
p
(p, Q) for every p Q.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

Note that in the above denition we assumed that

P = P
: < ). This was

caused only to simplify the description of the game if the domain of

P is S D,
then we may extend it to in some trivial way without changing the resulting
properties.
Observation 3.2. If p is a Tparameter and a forcing notion Q is Bnoble over
p, then Q is reasonably Bbounding over p.
(1) =
<
and p = (

P, , D) is a Tparameter on , and
(2)

Q = P
, Q
: < ) is a support iteration such that for every < ,
is Bnoble over p[P
] .
Then
(a) P
= lim(
Q) is proper,
(b) for each P
name
for a function from to V and a condition p P
there are q P
and A
: < ) such that [A
[ < (for < ) and

q p and
q
P
< :
() A
D[P
] .
Proof. (a) Assume that N (H(), , <
) is such that
<
N N, [N[ = and
Q, p, . . . N. Let p N P
. Choose

N = N
: < ) and =
: < ) such
that

N is an increasing continuous sequence of elementary submodels of N,
is an increasing continuous sequence of ordinals below ,
N =
<
N
,

Q, p, p, . . . N
0
, N
, P
N
+1
,

N( + 1) N
+1
,
[N
[ < and

+ otp(N
) + 888 <
+1
, ( + 2) N
+1
.
Put w
= N
and for each < let st
be the <
rst P
name for a winning

strategy of Complete in
0
(Q
) such that it instructs Complete to play
as
long as her opponent plays
. We also assume that whenever possible,
is the
<
rst name for the answer by st
to a particular sequence of names. Note that

st
: < ) N
0
.
By induction on < we will construct
()
, p
, q
, r
, r
and f
,
, A
,
, p
,
, q
,
, st

and p
,
for w
so that the following conditions ()

0
()
16
are satised.
()
0
Objects listed in ()
form the <
rst tuple with the properties described

in ()
1
()
16
below. Consequently, the sequence
objects listed in ()
: < )
is denable from

N
, ,

Q, p, p (in the language L(, <
)), so if =
is limit, then this sequence belongs to N

+1
. Also, objects listed in ()
are known after stage (and they all belong to N).

()
1
r
, r
, w
dom(r
) = dom(r
) and r
0
() = r
0
() = p() for w
0
.
()
2
For each < < we have ( w
+1
)(r
() = r
()) and p r
.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

()
3
If dom(r
) w
, then
r
the sequence r
(), r
0
_
Q
_

and if w
+1
w
, then st

is the <
rst P
name for a winning strategy

of Generic in
B+
p
(r
(), Q
). (And for w
0
, st

is the <
rst P
name
for a winning strategy of Generic in
B+
p
(p(), Q
). Note that st

N
0
for
w
0
and st
+1
for w
+1
.)
()
4
T
= (T
, rk
) is a (w
, 1)
tree, T
_
P
_
. (Note that we
do not require here that T
is standard, so some chains in T
may have no
bounds.)
()
5
p
= p
t
: t T
) and q
= q
t
: t T
) are trees of conditions, p
.
()
6
For t T
we have that dom(p
t
)
_
dom(p)

<
dom(r
) w
_
rk
(t)
and for each dom(p
t
) w
:
p
t

P
if the set r
,
then p
t
()
7
If w
+1
w
, < , then
,
, A
,
, p
,
, q
,
: < ) is a partial play of
B+
p
(r
(), Q
) in which Generic uses st

.
()
8
dom(r
) = dom(r
) =
tT
dom(q
t
) and if t T
, dom(r
)rk
(t)w
,
and q
t
q P
, r
q, then
q
P
if the set r
() : < q
t
,
then r

()
9
p
,
= p
,
t
: t T
& rk
(t) ) P
is a tree of conditions (for

w
), p
,
= p
.
()
10
If , w
, < and t T
, rk
(t) = , then p
,
t
p
,
t
.
The demands ()
11
()
16
formulated below are required only if =
is a limit
ordinal.
()
11
If t T
, rk
(t) = w
and X
t
= (s)
: ts T
, then
either ,= X
t
P
and p
,
t

P
,
= X
t
,
or X
t
= and p
,
t

P
,
= .
()
12
If s T
, rk
(s) = , w
and (s)
,= , then
p
,
s

P
,
((s)
) p
,
s
() .
()
13
If w
, otp(w
) = then
t T
: rk
(t) N
++1
, t T
: rk
(t) , p
,
:
(w
) ( + 1)) N
++2
, and
if is limit, then p
,
: w
) N
++1
, and if

t = t
i
: i <
i
) N
++1
is a chain in t T
: rk
(t) < with sup(rk
(t
i
) :
i < i
) = sup(w
), then

t has a bound in T
.
()
14
If t T
, w
rk
(t) and (t)
,= , then
q
t

P
,
((t)
) = q
t
() .
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

()
15
If t
0
, t
1
T
, rk
(t
0
) = rk
(t
1
) and w
rk
(t
0
), t
0
= t
1
but
_
t
0
_
,=
_
t
1
_
, then
p
t0

P
the conditions p
t0
(), p
t1
()
16
If
is a P
name for an ordinal and t T
satises rk
(t) = , then
the condition q
t
forces a value to
.
The rule ()
0
(and conditions ()
1
()
16
) actually fully determines our objects, but
we should argue that at each stage there exist objects with properties listed in
()
1
()
16
.
Suppose we have arrived to a stage < of the construction and all objects
listed in ()
for < have been determined so that all relevant demands are
satised, in particular the sequence objects listed in ()
: < ) is denable
from

N
, ,

Q, p, p.
If is a successor ordinal and w
w
1
, then we let st

be the <
rst
P
name for a winning strategy of Generic in

B+
p
(r
1
(), Q
). We also pick
the <
rst sequence f
,
, A
,
, p
,
, q
,
: < ) so that ()
7
is satised. Then
assuming that is not limit or ,=
we may nd objects listed in ()
so that
the demands in ()
1
()
10
are satised and [t T
: rk
(t) = [ = 1.
So suppose now that =
is a limit ordinal. For each w
we let f
,
be the
<
rst P
name such that

P
,
=
<
f
,
, and A
,
, p
,
be the <
rst
P
names such that
,
, A
,
, p
,
are given to Generic by st

as the answer to f
,
, A
,
, p
,
, q
,
: < ) .
Note that
objects listed in ()
: < )
,
, A
,
, p
,
: w
) N
+1
.
Now by induction on w
we will choose t T
: rk
(t) and p
,
and
auxiliary objects p
,
so that, in addition to demands ()
9
()
13
we also have:
()
17
p
,
= p
,
t
: t T
& rk
(t) ) P
is a tree of conditions, p
,
p
,
and dom(p
,
t
)
_
dom(p)

<
dom(r
) w
_
rk
(t) whenever t T
,
rk
(t) , and
()
18
if
0
<
1
are from w
, t T
, rk
(t) =
0
, then p
,0
t
p
,1
t
, and
()
19
if t T
, rk
(t) = , then dom(p

,
t
) = dom(p
,
t
) and for dom(p
,
t
) we
have
p
,
t

P
the sequence p
,
t
(), p
,
t
() : (w
) ( + 1)) is
a legal partial play of
0
(Q
) in which Complete uses

the winning strategy st
.
To take care of clause ()
13
, each time we pick an object, we choose the <
rst
one with the respective property.
Case 1: otp(w
) = + 1 is a successor ordinal.
Let
0
= max(w
) and suppose that we have dened T
= t T
: rk
(t)
0
and p
,0
, p
,0
satisfying the relevant demands of ()
9
()
19
. Let t T
be such
that rk
(t) =
0
. It follows from ()
11
that either p
,0
t
A
,0
= or p
,0
t

A
,0
= X
t
for some non-empty set X
t
P
. In the former case stipulate

8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

X
t
= . Note that necessarily X
t
N
+1
and X
t
N
++2
(remember ()
13
).
We declare that
t T
: rk
(t) = T
t (
0
, a) : t T
& rk
(t) =
0
& a X
t
.
Plainly, [t T
: rk
(t) [ < and even t T
: rk
(t) N
++1
and t T
: rk
(t) N
++2
(again, by ()
13
). Choose a tree of conditions
p
+
= p
+
t
: t T
& rk
(t) ) P
so that
dom(p
+
t
)
_
dom(p)

<
dom(r
) w
_
rk
(t) for t T
, rk
(t) ,
if t T
, rk
(t) < then p

+
t
= p
,0
t
,
if t T
, rk
(t) = and (t)

0
,= , then p
+
t
(
0
) is a P
0
name such that
p
,0
t

P
0
p
,0
_
(t)
0
_
p
+
t
(
0
) ,
if t T
, rk
(t) = and dom(p

+
t
) (
0
+ 1), then
p
+
t

P
if the set r
,
then p
+
t
(Note: p
+
N
++2
.) Next we may use Proposition 0.4 to pick a tree of conditions
p
,
= p
,
t
: t T
& rk
(t) ) such that p

+
p
,
and
if < , t T
, rk
(t) = , then either p

,
t
A
,
= or for some
non-empty set X
t
P
we have p
,
t
A
,
= X
t
.
(Again, by our rule of picking the <
rst, p
,
N
++2
.) Then we choose a
tree of conditions p
,
= p
,
t
: t T
& rk
(t) ) so that p
,
p
,
and for
every t T
with rk
(t) = we have dom(p

,
t
) = dom(p
,
t
) and for dom(p
,
t
),
p
,
t
() is the <
rst P
name for a condition in Q
such that
p
,
t

P
p
,
t
() is given to Generic by st
as the answer to
p
,
t
(), p
,
t
() : w
p
,
t
()) .
Note that, by the rule of picking the <
rst, p
,
N
++2
. It should be also
clear that p
,
, p
,
satisfy all the relevant demands stated in ()
9
()
19
.
Case 2: otp(w
) = is a limit ordinal.
Suppose we have dened t T
: rk
(t) and p
,
, p
,
for w
.
By our rule of choosing the <
rst objects, we know that the sequence
t
T
: rk
(t) , p
,
, p
,
: w

_
belongs to N
++1
. We also know that
t T
: rk
(t) < N
+
. Let T
+
be the set of all limit branches in
_
t
T
: rk
(t) < ,
_
, so elements of T
+
are sequences s = (s)
: w
) such
that s = (s)
) t T
: rk
(t) for w
. (Of course,
T
+
N
++1
.) We put
t T
: rk
(t) = t T
: rk
(t) <
_
T
+
N
++1
_
N
++2
.
Due to ()
19
at stages w
, we may choose a tree of conditions p

+
= p
+
t
:
t T
& rk
(t) ) P
such that
dom(p
+
t
)
_
dom(p)

<
dom(r
) w
_
rk
(t) for t T
, rk
(t) , and
if t T
, rk
(t) =
0
< then dom(p
+
t
) dom(p
,0
t
) and for each
dom(p
+
t
)
0
we have
p
+
t

P
p
,0
t
() p
+
t
() , and
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

if t T
, rk
(t) = , sup(w
) < , dom(p
+
t
), then
p
+
t

P
if the set r
,
then p
+
t
Then, like in the successor case, we may nd a tree of conditions p
,
= p
,
t
: t
T
& rk
(t) ) such that p

+
p
,
and
if < , t T
, rk
(t) = , then either p

,
t
A
,
= or for some
non-empty set X
t
P
we have p
,
t
A
,
= X
t
.
Also like in that case we choose p
,
= p
,
t
: t T
& rk
(t) ). Clearly, all

relevant demands in ()
9
()
19
are satised.
The last stage of the above construction gives us a tree T
= t T
: rk
(t)
and a tree of conditions p
,
= p
= p
t
: t T
). Since T
N
+otp(w
)+1
, we
know that [T
[ < so we may apply Proposition 0.4 to get a tree of conditions

q
= q
t
: t T
) p
such that ()
16
is satised. Remembering ()
15
+ ()
12
, we
easily nd P
names q
,
(for w
) such that

P
,
= q
,
() : A
,
) is a system of conditions in Q
,

P
( A
,
)( p
,
() q
,
()) , and
if t T
, rk
(t) > , then q
t

P
,
,= q
t
() = q
,
_
(t)
_
.
So then ()
14
is satised. Now we dene r
, r
essentially by ()
1
()
3
and
()
8
.
After completing all stages of the construction, for each N we look at
the sequence f
,
, A
,
, p
,
, q
,
: < ). By ()
7
, it is a P
name for a play of
B+
p
(r
(), Q
) (where w
+1
w
) in which Generic uses her winning strategy

st

. Therefore, for every N we may pick a P
name q() for a condition in

Q
such that
if w
+1
w
, < (or w
0
, = 0), then
q() r
() and q()
Q
_
<:
_
A
,
__
q
,
()
Q
__
D
p
[P
+1
].
This determines a condition q P
(with dom(q) = N) and easily ( < )(p

r
q) (remember ()
2
). For each N x P
+1
names C
i
, g
i
(for i < )
such that
q( + 1)
P
+1

_
i <
__
C
i
D V & g
_
and
_

i<
C
i

i<
set
p
(g
i
)
__
A
,
__
q
,
()
Q
_
.
Let B
be a P
name for the set < :

P
N
N
+1
. It follows from Lemma
1.6 that
P
B
D
p
[P
].
Claim 3.3.1. If
= is limit, then
q
P
if B
and
_
w
__

i<
C
i

i<
set
p
(g
i
)
_
then
_
t T
__
rk
(t) = & q
t

P
_
.
Proof of the Claim. Suppose that =
is a limit ordinal and a condition r q

forces (in P
) that
()
a
20
B
, and
()
b
20
_
w
__

i<
C
i

i<
set
p
(g
i
)
_
.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

Passing to a stronger condition if necessary, we may also assume that
()
a
21
if r
, then either r
r or r
, r are incompatible in P
.
Let H
= r
: r
r. It follows from ()
a
20
+ ()
a
21
that
()
b
21
r
P
N
= H
and H
N
+1
.
By 3.1(1)(a)+ ()
7
+ ()
0
we may choose a sequence
(, ) : w
& <
) N
+1
such that

(, ) is a P
name for an element of Q
(, ) N
,

P
(, ) = f
,
().
Next we choose a sequence t
= (t
: w
)

w
_
P

_
so that for each
w
, (t
is the <
rst member of P
satisfying:
()
22
if t = t
= (t
: w
) T
and
(i) for some non-empty set X P
, p
,
t

P
X = A
,
(remember ()
11
)
and there is X such that
(ii)
_
<
__
(, ()) r
() : r
_
,
then (t
X and
_
<
__
(, (t
()) r
() : r
_
.
Note that for every w
the sequence t
is denable (in L(, <
))
from p,
, H
, w
, and p
,
: w
). Consequently, if w
and
= otp(w
), then t
N
++1
. Now, by induction on w
we are
going to show that t
and choose conditions r
, r
such that
()
a
23
q
, r r
and if w
then r
, and
()
b
23
dom(r
) = dom(r
) and r
and for every dom(r
)
r
(), r
() : w
( + 1)) is a partial play of
0
(Q
)
.
Suppose that otp(w
) is a limit ordinal and for w
we know that
t
and we have dened r
, r
13
that t
. Let
= sup(w
) . It follows from ()
b
23
that we may nd a condition r
such that r
is stronger than all r
(for w
) and r
[, ) = r[, ).
Clearly r r
and also q
(remember ()
a
23
for w
). Now by
induction on [, ) we argue that q
. So suppose that <

and we know already that q
3
+ ()
5
+ ()
6
that
r

P
_
i <
__
r
i
() p
() q
()
_
and therefore we may use ()
8
to
conclude that
r

P
q
() r
() q() r() = r
(),
as desired. Finally we dene r
essentially by ()
b
23
.
Now suppose that otp(w
) is a successor ordinal and let

0
= max(w
).
Assume we know that t
0
T
and that we have already dened r
0
, r
0
P
0
.
It follows from the choice of q and from ()
b
20
that
r
0

P
0
r(
0
)
Q
0
_
A
,0
__
q
,0
()
Q
0
_
.
Thus we may choose r
P
0+1
and P
such that r
0
r
0
, r
0
r(
0
)
r
(
0
) and r
0

P
0
A
,0
& q
,0
() r
(
0
) . Then r(
0
+ 1) r
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

and (by ()
7
, 3.1(1)(c,d)) r
0

P
0
_
<
__
(
0
, ()) r
(
0
)
_
and hence (by
()
a
21
) r
0

(
0
, ()) r(
0
) for all < . Therefore
()
a
24
for all < ,
(
0
, ()) r
(
0
) : r
.
Since p
,0
t
0
r
0
(remember ()
a
23
for
0
and ()
5
) we may use ()
11
to conclude
that for some non-empty set X P
we got p
,0
t
0
X = A
,0
and X
satises (ii) of ()
22
. Hence (t
)
0
X is such that
()
b
24
for all < ,
(
0
, (t
)
0
()) r
(
0
) : r
,
and in particular t
(remember ()
11
). We claim that (t
)
0
= . If not,
then by 3.1(1)(b) we have
r
0

P
0
_
<
__
(
0
, (t
)
0
()),
(
0
, ()) are incompatible in Q
0
_
,
so we may pick < and a condition r
+
P
0
such that r
0
r
+
and
r
+
(
0
, (t
)
0
()),
(
0
, ()) are incompatible in Q
0
.
However, r
+
(
0
, (t
)
0
()) r(
0
) &
(
0
, ()) r(
0
) (by ()
a
24
+
()
b
24
), a contradiction.
Now we dene r
so that r
(
0
+ 1) = r
and r
(
0
, ) = r(
0
, ). By
the above considerations and ()
14
we know that q
(
0
+ 1) r
= r
(
0
+ 1).
Exactly like in the case of limit otp(w
) we argue that q
. Finally, we
choose r
by ()
b
23
.
The last stage of the inductive process described above shows that t
and
q
t
r
, r r
. Now the claim readily follows.

We nish the proof of part (a) of the theorem exactly like in the proof of 2.7.
(b) Included in the proof of the rst part.
4. Examples and counterexamples
Let us note that our canonical test forcing Q
E
E
is Bnoble:

E, E are as in 1.11 and p = (

P, S, D) is a T
parameter on such that S E. Then the forcing Q
E
E
is Bnoble over p.
Proof. The proof is a small modication of that of 1.12(2). First we x an enumer-
ation
: < ) =
<
(remember
<
= ), and for < let f() Q
E
E
be a
condition such that root
_
f()
_
=
and
_
f()
__
succ
f()
() =
_
.
Let p Q
E
E
. Consider the following strategy st of Generic in
B+
p
(p, Q
E
E
). In the
course of the play Generic is instructed to build aside a sequence T
: < ) so
that if f
, A
, p
, q
: < ) is the sequence of the innings of the two players, then

the following conditions (a)(d) are satised.
(a) T
E
E
and if < < then p = T
0
T
and T
= T
,
(b) if < is limit, then T
<
T
,
(c) if S, then
f
= f and A
is a maximal set (possibly empty) such that

() for each A
the family
_
f
_
()
_
: <
_
T
has an upper
bound in Q
E
E
and lh
_
()
: <
_
= ,
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

() if
0
,
1
A
are distinct, then for some < the conditions f

_
0
()
_
,
f
_
1
()
_
are incompatible,
for A
the condition p
is an upper bound to
_
f
_
()
_
: <
_
T
,
T
+1
=
: A

_
(T
and / p
for A
_
,
(d) if / S, then A
= , f
= f and T
+1
= T
.
After the play is over, Generic puts p
<
T

<
. Almost exactly as in
the proof of 1.12(2), one checks that p
E
E
is a condition witnessing ()
B+
p
of
3.1(1).
Denition 4.2. Let

E, E be as in 1.11. We dene a forcing notion P
E
E
as follows.
A condition p in P
E
E
<
such that
for every p, either [succ
p
()[ = 1 or succ
p
() E
, and
for some set A E we have
_
p
__
lh() A succ
p
() E
_
.
The order =
P
E
E
is the reverse inclusion: p q if and only if (p, q P
E
E
and )
q p.

E, E are as in 1.11 and p = (

P, S, D) is a T
parameter on such that S E. Then P
E
E
is a (<)-complete forcing notion
of size 2
which is B-noble over p.

Proof. The arguments of 4.1 can be repeated here with almost no changes (a slight
modication is needed for the justication that p
E
E
).
We may use the forcing P
E
E
to substantially improve [RS05, Corollary 5.1]. First,
let us recall the following denition.
Denition 4.4. Let T be a lter on including all co-bounded subsets of , / T.
(1) We say that a family F

is Tdominating whenever
_
g
__
f F
__
< : g() < f() T
_
.
The Tdominating number d
F
is the minimal size of an Tdominating
family in

.
(2) We say that a family F

is Tunbounded whenever
_
g
__
f F
__
< : g() < f() (T)
+
_
.
The Tunbounded number b
F
is the minimal size of an Tunbounded
family in

.
(3) If T is the lter of co-bounded subsets of , then the corresponding dom-
inating/unbounded numbers are also denoted by d
, b
. If T is the lter
generated by club subsets of , then the corresponding numbers are called
d
cl
, b
cl
.
Corollary 4.5. Assume =
<
, 2
=
+
. Suppose that p = (

P, S, D) is a
Tparameter on , and E is a normal lter on such that S E. Then there
is a
++
cc proper forcing notion P such that
P
2
=
++
= b
E
P = d
E
P = d
& b
= b
D[P]
= d
D[P]
=
+
.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

Proof. For
<
let E
be the lter generated by clubs of and let

E = E
:

<
). Let

Q = P
, Q
: <
++
) be a support iteration such that for every
<
++
,
P
= P
E
E
. (Remember, we use the convention that in V
P
the
normal lter generated by E is also denoted by E etc.) Let P = P
++ = lim(
Q).
It follows from 3.3(a)+4.3 that P is proper. Using [RS, Theorem 2.2] (see also
Eisworth [Eis03, 3]) we see that P satises the
++
cc,
P
2
=
++
and P is
(<)complete. Thus, the forcing with P does not collapse cardinals and it also
follows from 3.3(b) that
P

V is D[P]dominating in

.
It is also easy to check, that for each <
++
P

V
P
is not Eunbounded in

and hence we may easily conclude that
P
b
E
P = 2
.
Denition 4.6. Assume that
is weakly inaccessible,
<
= ,
H : is such that [[
+
[H()[ for each < ,
F is a normal lter on ,

F = F
<
<
H()) where F
is a
_
<[[
+
_
complete lter on H() whenever

<
H(), < .
We dene forcing notions Q
H
F,F
and P
H
F,F
as follows.
(1) A condition p in Q
H
F,F

<
<
H() such that
(a) for every p, either [succ
p
()[ = 1 or succ
p
() F
, and
(b) for every lim
(p) the set < : succ

p
() F
belongs to
F.
The order of Q
H
F,F
is the reverse inclusion.
(2) A condition p in P
H
F,F

<
<
H() satisfying
(a) above and
(b)
+
for some set A F we have
_
p
__
lh() A succ
p
() F
_
.
The order of P
H
F,F
is the reverse inclusion.
Proposition 4.7. Assume that , H,

F, F are as in 4.6. Let p = (

P, S, D) be a
Tparameter such that S F. Then both Q
H
F,F
and P
H
F,F
are strategically
(<)complete forcing notions of size 2
which are also Bnoble over p.

Proof. Like 1.12(2), 4.1, 4.3.
The property of being Bnoble seems to be a relative of properness for Dsemi
diamonds introduced in [RS01] and even more so of properness over Ddiamonds
studied in Eisworth [Eis03]. However, technical dierences make it dicult to
see what are possible dependencies between these notions (see Problem 7.2). In
this context, let us note that there are forcing notions which are proper over semi
diamonds, but are not Bnoble over any Tparameter p. Let us consider, for
example, a forcing notion P
dened as follows:
a condition in P
is a function p such that

8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

(a) dom(p)
+
, rng(p)
+
, [dom(p)[ < , and
(b) if
1
<
2
are both from dom(p), then p(
1
) <
2
;
the order of P
is the inclusion .
Proposition 4.8 (See [RS01, Prop. 4.1, 4.2]). P
is (<)-complete forcing notion

which is proper over all semi diamonds.
Proposition 4.9. P
is not Bnoble over any Tparameter p on .

Proof. Let q P
be such that that dom(q) and let W
0
be a P
name such
that
P
W
0
=
P
. Clearly
q
P
W
0
is a function with dom(W
0
) and rng(W
0
) .
Let W
be a P
name for a member of

such that
q
P
W
0
W
and
_
dom(W
0
)
__
W
() =
_
.
Now suppose that p = (

P, S, D) is a Tparameter and A
: < ) is a sequence
of subsets of such that [A
[ < for < . The following claim implies that P
(above the condition q) is not Bnoble over p (remember 3.3(b)).

Claim 4.9.1.
P

_
< : W
() A
_
/ D[P
] .
Proof of the Claim. Suppose that p q and B
i
, f
i
(for i < ) are P
names for
members of D V and members of

, respectively, such that
p
P

i<
B
i

i<
set
p
(f
i
) < : W
() A
.
Build inductively a sequence p
i
, B
i
, f
i
: i < ) such that for each i < :
(i) p
i
P
, p p
0
p
j
p
i
for j < i,
(ii) B
i
D V, f
i

and
(iii) p
i

P
B
i
= B
i
and f
j
i = f
j
i for all j i, and
(iv) i < sup
_
dom(p
i
)
_
.
Since B =
i<
B
i

i<
set
p
(f
i
) D, we may pick a limit ordinal B such
that
_
i <
__
sup
_
(dom(p
i
) rng(p
i
))
_
<
_
. Put = sup(A
) + 888 and
p
+
=
i<
p
i
(, ). Then p
+
P
is a condition stronger than all p

i
for i <
and p
+
P

i<
B
i

i<
set
p
(f
i
) and W
() = / A
, a contradiction.
A similar construction can be carried out above any condition q such that for
some dom(q) we have cf() = (the set of such conditions is dense in P
).
5. Q
E
E
vs P
E
E
and Cohen reals
The forcing notions Q
E
E
and P
E
E
(introduced in 1.11 and 4.2, respectively) may
appear to be almost the same. However, at least under some reasonable assumptions
on

E, E they do have dierent properties.
Suppose that V V
are transitive universes of ZFC (with the same ordinals)

such that
<
V =
<
V
. We say that a function c
2 V
is a Cohen
over V if for every open dense set U
<
2 (where
<
2 is equipped with the partial
order of the extension of sequences), U V, there is < such that c U.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

(a) is a strongly inaccessible cardinal,
(b) S is the set of all strong limit cardinals < of countable conality,
(c) E is a normal lter on such that S E,
(d)

E = E
:
<
) is a system of (<)complete non-principal lters on
.
Then the forcing notion P
E
E
adds a Cohen over V.
Proof. Let S. We will say that a tree T
<
is interesting if
()
0
[ lim
(T)[ = 2
and for some increasing conal in sequence

n
: n < )
we have
_
n <
__
T
__
lh()
n
sup
_
rng()
_
<
n+1
_
.
Note that there are only 2
many interesting trees (for S). Therefore we

may x a well ordering of the family of interesting trees of length 2
and choose
by induction a function f
:

<
2 ) such that
()
1
if T
<
is a interesting tree, then
_

<
2 )
__
lim
(T)
__
f
() =
_
.
Let W
be a P
E
E
name such that
P
E
E
W
=
_
root(p) : p
P
E
E
_
. Plainly, W
. Next, let C
be a P
E
E
name such that
P
E
E
C
= S : W

and let
be a P
E
E
name such that
E
E

is the concatenation of all elements of the sequence

f
(W
) : C
), i.e.,
= . . .
(W
. . .)
C
.
Plainly,

2 (remember, ) / rng(f
) for S).
We are going to argue that
E
E

is -Cohen over V .
To this end suppose that U
<
2 is an open dense set and p P
E
E
. Let
B =
_
S :
_
p
__
succ
p
() E
__
(so B E). By induction on n < choose
n
, T
n
so that
()
2

n
B,
n
<
n+1
, T
n

n
is a complete tree (thus every chain in T
n
has a bound in T
n
),
()
3
T
n
T
n+1
p, T
n+1
n
= T
n
n
, T
0

0
0
, [T
0
0
[ = 1,
()
4
if T
n
<n
, then succ
Tn
() ,= and
2 [succ
Tn
()[ lh()
0
, . . . ,
n1
,
()
5
if T
n+1
n
, then [succ
Tn+1
()[ =
n
and
sup
_
rng()
_
<
n+1
< min
_
succ
Tn+1
()
_
.
Next put = sup(
n
: n < ) and T =
n<
T
n
. Clearly S and T
<
is a
interesting tree such that
_
lim
(T)
__
= min S :
0
< &

_
.
Let T
0

0
0
=
0
and C(
0
) = S (
0
+ 1) :
0

, and let
0
be the concatenation of all elements of the sequence f
(
0
) : C(
0
)), i.e.,
0
= . . .
(
0
)
. . .)
C(0)

<
2. Pick
<
2 such that
0
U. It
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

follows from ()
1
that we may nd lim
(T) such that =

0
(). Now note

that (p)
E
E
.
(a) is a measurable cardinal,
(b)

E = E
:
<
) is a system of normal ultralters on ,
(c) E is a normal lter on ,
(d) p Q
E
E
and
is a Q
E
E
name such that p
2,
(e)

=
: < ) is an increasing continuous sequence of non-successor

ordinals below such that
0
= 0 and 2
2
||
< [
+1
[ for all < .
Then there are a condition q Q
E
E
and a sequence A
: < ) such that

(i) q p and A

[,+1)
2, [A
[ < [
+1
[ for < , and
(ii) q
Q
E
E
_
<
__
,
+1
) A
_
.
In particular, the forcing notion Q
E
E
does not add any Cohen over V.
Proof. Let
: < ) be an enumeration of
<
such that
implies < .
By induction on < we will construct a sequence A
, p
, X
: < ) so that
for each < we have:
()
1
A

[,+1)
2, [A
[ < [
+1
[, p
E
E
, X
, [X
,
()
2
if < < , then p
and X
,
()
3
p
0
= p, X
0
= root(p
0
), and if is limit then p
<
p
and X
<
X
,
()
4
if X
+1
, then succ
p
() E
,
()
5
if is limit, X
and

<
succ
p
(), then for some X

+1
we
have
) ,
()
6
if
, then there is X
+1
such that
and if (additionally)
succ
p
(
) E
, then =
,
()
7
if is limit and

p
and succ
p
() E
, then X
+1
,
()
8
p
+1

,
+1
) A
.
Suppose that we have determined p
, X
for < and A
for + 1 < so that

the relevant instances of ()
1
()
8
are satised. If is limit or 0, then p
, X
are
dened by ()
3
(and A
will be chosen at the next step). One easily veries that

p
, X
satisfy the requirements in ()

1
()
4
.
So suppose now that = + 1 (and we have dened p
, X
and A
for < ).
We may easily choose a set X
such that
()
9
X
, [X
[ <
and X
satises ()
4
()
7
(with there correspond-
ing to here), and
()
10
if
0
,
1
X
, =
0

1
, then X
, and
()
11
if
: < ) X
is increasing, then there is X
such that
( < )(
).
Next, for each X
choose a function
: [
,
+1
) 2 and a condition
q
E
E
so that
()
12
root(q
) = , (p
,
_
X
__
(lh()) / succ
q
()
_
and
q
,
+1
) =
.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

(Possible by assumption (b) and 2.4.) Put p
X
q
and A
: X
.
Plainly, [A
[ [X
[ <
+1
and p
E
E
(to verify that p
is a complete -tree use

()
10
+ ()
11
; the other requirements easily follow from the fact that [X
[ < ).
One also easily checks that p
,
+1
) A
.
After the inductive construction is carried out, we put q =
<
p
. It follows
from ()
2
+()
5
+()
6
that q is a complete tree, [succ
q
()[ = 1 or succ
q
() E
for each q, and

_
<
X
= q : succ
q
() E
.
Suppose now that lim
(q). Then for each < we have lim
(p
) and
hence B
def
= < : succ
p
() E
E. Let
C = < : is limit and

(it is a club of ). Since E is a normal lter, C

<
B
E. Suppose
C
<
B
. Then by ()
3
+()
7
we have X
+1
and thus succ
q
() E
.
Consequently, q Q
E
E
.
Finally, it follows from ()
8
that q
_
<
__
,
+1
) A
_
.
Let us note that forcing notions of the form Q
E
E
may add Cohens if the lters
E
are far from being ultralters.

(a) E is a normal lter on =
<
,
(b)

E = E
:
<
) is a system of (<)complete lters on ,
(c) for every
<
there is a family A
: < of pairwise disjoint sets

from (E
)
+
.
Then both the forcing notions Q
E
E
and P
E
E
add Cohens over V.
Proof. We will sketch the argument for Q
E
E
only (no changes are needed for the
case of P
E
E
).
For each
<
choose a function h
:
<
2 ) such that
_

<
2
__
<
__
A
__
h
() =
_
.
Let W
be a Q
E
E
name such that
Q
E
E
W
=
_
root(p) : p
Q
E
E
_
and let
be a
Q
E
E
name such that
E
E

is the concatenation of all elements of the sequence
h
W
_
W
()
_
: <
_
, i.e.,
_
W
(0)
_
h
W
(0)
_
W
(1)
_
. . .
h
W
_
W
()
_
. . .)
<
.
One easily veries that
2 is a Cohen over V .
The result in 5.2 would be specially interesting if we only knew that it is preserved
in support iterations. Unfortunately, at the moment we do not know if this is true
(see Problem 7.3(1)). However, we may consider properties stronger than adding
Cohens and then our earlier results give some input.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

Denition 5.4. Suppose that V V
are transitive universes of ZFC (with the

same ordinals) such that
<
V =
<
V
. We say that a function c
2 V
is a
(1) strongly
Cohen over V if it is a Cohen (i.e., for every open dense set

U
<
2 from V there is < such that c U) and
() if
: < ) V is such that <
< and

[,)
2 for
< , then
V
[= < :
c is stationary;
(2) strongly
Cohen over V if it is a Cohen and

() if
< and

[,)
2 for
< , then
V
[= < :
c is stationary.
(3) More generally, if D is a normal lter on , D V then we say that c
2 V
is Dstrongly
Cohen over V if in () we replace stationary

by (D
V
)
+
(where D
V
is the normal lter generated by D in V
).
Similarly for strongly
.
Remark 5.5. (1) To explain our motivation for 5.4, let us recall that if c

2
is Cohen over V and
< and

[,)
2 for < , then
V
[= both < :
c and < :
c are unbounded in .
(2) Let
: < ) V be such that <
< and

[,)
2 for
< . Let C = (
<
2, ) (so this is the Cohen forcing notion) and let c
be the canonical Cname for the generic real (i.e.,

C
c
C
). Let Q
be a Cname for a forcing notion in which conditions are closed bounded

sets d such that ( d)(
) ordered by end extension. Then

C Q
is essentially the -Cohen forcing and
CQ
is not strongly
Cohen over V .
Hence, if we add a Cohen then we also add a non-strong
Cohen.
(3) Note that strongly
Cohen implies strongly
Cohen. (Simply, for a

sequence
: < ) consider 1
: < ).)
Proposition 5.6. Assume that is a strongly inaccessible cardinal and p =
(

P, , D
) is a Tparameter such that P
=

and D
is the lter generated

by club subsets of .
(1) If a forcing notion Q is reasonably Bbounding over p, then
Q
there is no strongly
Cohen over V .
(2) If

Q = P
, Q
: < ) is a support iteration such that for every < ,
P
Q
is reasonably Bbounding over p[P
] ,
then
P
there is no strongly
Cohen over V .
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

Proof. (1) Note that, in V
Q
, D
[Q] is the lter generated by clubs of .

Let p Q and let
be a Qname such that p

2. Let st be a winning
strategy of Generic in the game
B+
p
(p, Q).
Let us consider a play of
B+
p
(p, Q) in which Generic follows the instructions
of st while Antigeneric plays as follows. In the course of the play, in addition to
his innings q
t
, Antigeneric constructs aside a sequence
t
: t I
) : <
_
such that if
, p
t
, q
t
: t I
) : <
_
is the sequence of the innings of the two
players then the following two demands are satised.
()
1

is a cardinal such that 2

|I|++0
<
and
t

2 (for t I
),
()
2
q
t

Q

t
for each t I
.
Since the play is won by Generic, there is a condition q p such that
q
Q
< : (t I
)(q
t

Q
) contains a club of .
It follows from ()
1
that for each < we may choose
0
<
1
from the in-

terval (,
) such that
_
t I
__
t
(
0
) =
t
(
1
)
_
. For each < choose
: [,
) 2 so that
(
0
) = 0 and
(
1
) = 1. Then
q
Q
< :
contains a club of .
(2) Similar, but we have to work with trees of conditions as in the proof of 1.10.
6. Marrying Bbounding with fuzzy proper
In this section we introduce a property of forcing notions which, in a sense,
marries the Bbounding forcing notions of [RS05, Denition 3.1(5)] with the fuzzy
proper forcings introduced in [RS07, A.3]. This property, dened in the language
of games, is based on two games: the servant game
servant
S,D
which is the part
coming from the fuzzy properness and the master game
master
S,D
which is related to
the reasonable boundedness property. Later in this section we will even formulate
a true preservation theorem for a slightly modied game.
In this section we assume the following:
Context 6.1. (1) is a strongly inaccessible cardinal,
(2) D is a normal lter on ,
(3) S D, 0 / S, all successor ordinals below belong to S, S is unbounded.
Denition 6.2. Let Q be a forcing notion.
(1) A Qservant over S is a sequence q = q
t
: S & t I
) such that
[I
[ < (for S) and q
t
Q (for S, t I
).
(2) Let q be a Qservant over S and q Q. We dene a game
servant
S,D
( q, q, Q)
as follows. A play of
servant
S,D
( q, q, Q) lasts at most steps during which the
players, COM and INC, attempt to construct a sequence r
, A
: < )
such that
r
Q, q r
, A
D and < < r
.
The terms r
, A
are chosen successively by the two players so that

if / S, then INC picks r
, A
, and
if S, then COM chooses r
, A
.
If at some moment of the play one of the players has no legal move, then INC
wins; otherwise, if both players always had legal moves and the sequence
r
, A
: < ) has been constructed, then COM wins if and only if

8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

()
_
S
__
[

<
A
& is limit ] (t I
)(q
t
r
)
_
.
(3) If COM has a winning strategy in the game
servant
S,D
( q, q, Q), then we will
say that q is an (S, D)knighting condition for the servant q.
Denition 6.3. Let Q be a strategically (<)complete forcing notion.
master
S,D
(q, Q) between two players,
Generic and Antigeneric. The game is a small modication of
rbB
p
(p, Q)
(see 1.8) the main dierence is in the winning condition. A play of
master
S,D
(p, Q) lasts steps and during a play a sequence
_
I
, p
t
, q
t
: t I
) : <
_
is constructed. Suppose that the players have arrived to a stage < of
the game. Now,
()
rst Generic chooses a non-empty set I
of cardinality < and a

system p
t
: t I
) of conditions from Q,
()
then Antigeneric answers by picking a system q
t
: t I
) of condi-
tions from Q such that (t I
)(p
t
q
t
).
, p
t
, q
t
: t I
) : <
_
of
master
S,D
(p, Q) if and only if letting q = q
t
: S & t I
) (it is a
Qservant over S) we have
()
D,S
master
there exists an (S, D)knighting condition q p for the servant q.
(2) A forcing notion Q is reasonably merry over (S, D) if (it is strategically
(<)complete and) Generic has a winning strategy in the game
master
S,D
(p, Q)
for any p Q.
Theorem 6.4. Assume that , S, D are as in 6.1. Let

Q = P
, Q
: < ) be a
support iteration such that for each < :
P
Q
is reasonably merry over (S, D).

Then
(a) P
= lim(
Q) is proper, and
(b) for every P
name
for a function from to V and a condition p P
,
there are q p and A
: < ) such that ( < )([A
[ < ) and
q < :
() A

_
D
P
_
+
.
Proof. (a) The proof starts with arguments very much like those in [RS05, Theo-
rems 3.1, 3.2], so we will state only what should be done (without actually describ-
ing how the construction can be carried out). The major dierence comes later, in
arguments that the chosen condition is suitably generic.
Suppose that N (H(), , <
) is such that
<
N N, [N[ = and

Q, S, D, . . . N.
Let p N P
and
: < ) list all P
names for ordinals from N. For each

N x a P
name st
N for a winning strategy of Complete in
0
(Q
)
such that it instructs Complete to play
as long as her opponent plays
.
Let us pick an increasing continuous sequence w
: < ) of subsets of such

that
<
w
= N , w
0
= 0 and [w
[ < .
By induction on < choose
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

()
, p
, q
, r
, r
,
, p
,
, q
,
: w
), and st

for w
+1
w
so that if the following conditions ()

0
()
11
are satised (for each < ).
()
0
All objects listed in ()
belong to N and they are known after stage of

the construction.
()
1
r
, r
, r
0
(0) = r
0
(0) = p(0), and for each < < we have
( w
+1
)(r
() = r
() = r
()) and p r
.
()
2
If dom(r
) w
, then
r
the sequence r
(), r
0
_
Q
_

and if w
+1
w
, then st

is a P
name for a winning strategy of Generic

in
master
S,D
(r
(), Q
) such that if p
t
: t I
) is given by that strategy to

Generic at stage , then I
is an ordinal below . Also st

0
is a suitable
winning strategy of Generic in
master
S,D
(p(0), Q
0
).
()
3
T
= (T
, rk
) is a standard (w
, 1)
tree, [T
[ < .
()
4
p
= p
t
: t T
) and q
= q
t
: t T
) are standard trees of conditions in
Q, p
.
()
5
If t T
, rk
(t) = , then the condition q
t
decides the values of all names
: ).
()
6
For t T
we have
_
dom(p)

<
dom(r
) w
_
rk
(t) dom(p
t
) and
for each dom(p
t
) w
:
p
t

P
if the set r
,
then p
t
()
7
If w
, then
,
is a P
name for an ordinal below , p
,
, q
,
are
P
names for
,
sequences of conditions in Q
.
()
8
If w
+1
w
, then
,
, p
,
, q
,
: < ) is a play of
master
S,D
(r
(), Q
)
in which Generic uses st

.
()
9
If t T
, rk
(t) = < , then the condition p
t
decides the value of
,
,
say p
,
=
t
,
, and (s)
: ts T
=
t
,
and
q
t

P
,
() p
() and q
,
() = q
() for <
t
,
.
()
10
If t
0
, t
1
T
, rk
(t
0
) = rk
(t
1
) and w
rk
(t
0
), t
0
= t
1
but
_
t
0
_
,=
_
t
1
_
, then
p
t0

P
the conditions p
t0
(), p
t1
()
11
dom(r
) = dom(r
) =
tT
dom(q
t
) dom(p) and if t T
, dom(r
)
rk
(t) w
, and q
t
q P
, r
q, then
q
P
if the set r
() : < q
t
,
then r

After the construction is carried out we dene a condition r P
as follows. We
let dom(r) = N and for dom(r) we let r() be a P
name for a condition

8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

in Q
such that if w
+1
w
, < (or = 0 = ), then
r() r
() is an (S, D)knighting condition for the servant

q
def
= q
,
() : S & <
,
) .
Clearly r is well dened (remember ()
8
). Note also that r
r for all < and

p r. We will argue that r is an (N, P
)generic condition. To this end suppose

towards contradiction that r
r,
< and r
/ N.
For each N x a P
name st
for a winning strategy of COM in the

game
servant
S,D
( q
, r(), Q
). Moreover, for each < x a P
name st
for a
winning strategy of Complete in
0
(Q
) such that it instructs Complete to

play
.
By induction on < we will build a sequence
, r
+
, A
,i
, A
,i
: i < & N ) : <
_
such that the following demands ()
12
()
16
are satised:
()
12
r
, r
r
+
for < < ,

()
13
A
,i
is a P
name for an element of D V (for N , i < ),

()
14
if S and w
, then r
( < )(i < )(A
,i
= A
,i
),
()
15
if < < and w
+1
w
, then for some P
names s
: ) we
have
r
the sequence s
,
i<
A
,i
: )
(),
i<
A
,i
: < )
is a legal partial play of
servant
S,D
( q
, r(), Q
)
in which Generic follows st
,
()
16
dom(r
+
) = dom(r
), r
+
= r
and for each dom(r

+
) w
we
have
r
+
the sequence r
(), r
+
0
_
Q
_
.
So suppose that we have arrived to a stage < of the construction and
r
, r
+
for < ,
A
,i
for i < , < and

<
w
,
A
,i
for , i < sup( S) and

<sup(\S)
w
have been determined.

Case 1: / S.
Note that by our assumption on S (in 6.1), is not a successor ordinal, so w
<
w
(or = 0 and w
0
= 0). By ()
16
+ ()
15
we may choose a condition r
stronger than all r

+
(for < ) and stronger than r
and such that for each w
if < and i < , then r
forces a value to A
,i
, say
r
,i
= A
,i
.
For w
and i < we also let A
,i
be a P
name for the interval (, ). The

condition r
+
is fully determined by ()
16
.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

Case 2: S is a successor ordinal, say = + 1.
First, for each w
we pick P
names s
and A
,i
(for < , i < ) such
that r
+
r
+
() = s
0
and
r
the sequence s
,
i<
A
,i
: ) is a legal partial play of
servant
S,D
( q
, r(), Q
) in which Generic follows st
.
Next, we let dom(r
) = dom(r
+
) and for each w
we choose P
names r
()
and A
,i
(for i < ) such that
r
+
(),
i<
A
,i
is the answer to the partial game as in ()
15
given by st
.
For dom(r
) w
we put r
() = r
+
(). Then we dene condition r

+
by ()
16
.
Case 3: S is a limit ordinal.
We let dom(r
) =
<
dom(r
+
) and by induction on dom(r
) we dene r
()
so that
if / w
, then r
( < )(r
+
() r
()) (exists by ()
16
),
if w
then for some P
names A
,i
for members of D V
r
(),
i<
A
,i
is the answer to the partial game as in ()
15
given by st
.
The condition r
+
is given by ()
16
.
After the above construction is carried out we note that
_
< : ( w
)(, i < )( A
,i
)
_
D,
so we may choose an ordinal S (
+ 1) which is a limit of points from S

and such that

<
i<
A
,i
for all w
. The following claim provides the

desired contradiction (remember ()
0
+ ()
5
).
Claim 6.4.1. For some t T
such that rk
(t) = the conditions q
t
and r
are
compatible
Proof of the Claim. The proof is very much like that of Claim 2.7.1. Let
) = w
be the increasing enumeration. For each < x a P
name st
for a winning strategy of Complete in
0
(Q
) such that it instructs Complete

to play
.
By induction on
we will choose conditions s
, s
and t = (t)
:
<
) T
such that letting t
= (t)
< ) T
we have
()
a
q
and r
,
()
b
dom(s
) = dom(s
) and for every <
,
s
(), s
() :
< ) is a partial legal play of
0
(Q
)
.
Suppose that
is a limit ordinal and we have already dened t
= (t)
< ) and s
, s
< ). Let = sup(
< ). It follows from ()

b
that we may nd a condition s
such that s
is stronger than all s
(for
< ) and s
[,
) = r
[,
). Clearly r
and also q
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

(remember ()
a
). Now by induction on [,
] we argue that q
.
Suppose that <
and we know q
. By ()
2
+()
4
+()
6
we know
that s
(i < )(r
i
() p
() q
()) and therefore we may use ()

11
to
conclude that
s
() r
() r() r
() = s
().
Then the condition s
is determined ()
b
.
Now suppose that =
+ 1
and we have already dened s
, s
and t
. It follows from the choice of and ()

14
that r

<
i<
A
,i
and hence, by the choice of r and ()
15
we have (remember () of 6.2(2))
s

_
<
__
q
() r
)
_
.
Therefore we may use ()
9
to choose = (t)
and a condition s
such that
t
def
= t
, ) T
, s
and
s
) r
) = s
) ,
r
) = s
).
We nish exactly like in the limit case.
After the inductive construction is completed, look at t = t
and s
.
(b) Should be clear at the moment.
Denition 6.5 (See [RS05, Def. 3.1]). Let Q be a strategically (<)complete
forcing notion.
(1) Let p Q. A game
rcB
D
(p, Q) is dened similarly to
rbB
p
(p, Q) (see 1.8)
except that the winning criterion ()
p
rbB
is weakened to
()
rc
B

p
Q

_
< :
_
t I
__
q
t

Q
__
D
Q
.
(2) A forcing notion Q is reasonably Bbounding over D if for any p Q,
Generic has a winning strategy in the game
rcB
D
(p, Q).
Observation 6.6. If Q is reasonably Bbounding over D, then it is reasonably
merry over (S, D).
It is not clear though, if forcing notions which are reasonably Bbounding over a
Tparameter p are also reasonably merry (see Problem 7.4). Also, we do not know
if fuzzy properness introduced in [RS07, A.3] implies that the considered forcing
notion is reasonably merry (see Problem 7.5), even though the former property
seems to be almost built into the latter one.
One may ask if being reasonably merry implies being Bbounding. There are
examples that this is not the case. The forcing notion Q
2
D
(see 6.8 below) was
introduced in [RS05, Section 6] and by [RS05, Proposition 6.4] we know that it is
not reasonably Bbounding over D. However we will see in 6.12 that it is reasonably
merry over (S, D).
Denition 6.7 (See [RS05, Def. 5.1]). (1) Let < < . An (, )extending
function is a mapping c : T() T() T() such that c(u) = u for
all u T().
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

(2) Let C be an unbounded subset of . A Cextending sequence is a sequence
c = c
: C) such that each c
is an (, min(C ( + 1)))extending
function.
(3) Let C , [C[ = , C, w and let c = c
: C) be a C
extending sequence. We dene pos
+
(w, c, ) as the family of all subsets u
of such that
(i) if
0
= min
_
C : ( w)( < )
_
, then u
0
= w (so if
0
= , then u = w), and
(ii) if
0
,
1
C, w
0
<
1
= min(C (
0
+ 1)) , then either
c
0
(u
0
) = u
1
or u
0
= u
1
,
(iii) if sup(w) <
0
= sup(C
0
) / C,
1
= min
_
C (
0
+1)
_
, then
u
1
= u
0
.
For
0
C such that w
0
, the family pos(w, c,
0
, ) consists of all
elements u of pos
+
(w, c, ) which satisfy also the following condition:
(iv) if
1
= min
_
C (
0
+ 1)
_
, then u
1
= c
0
(w).
Denition 6.8 (See [RS05, Def. 6.2]). We dene a forcing notion Q
2
D
as follows.
A condition in Q
2
D
is a triple p = (w
p
, C
p
, c
p
) such that
(i) C
p
D, w
p
min(C
p
),
(ii) c
p
= c
p
: C
p
) is a C
p
extending sequence.
The order
Q
2
D
= of Q
2
D
is given by
p
Q
2
D
q if and only if
(a) C
q
C
p
and w
q
pos
+
(w
p
, c
p
, min(C
q
)) and
(b) if
0
,
1
C
q
,
0
<
1
= min(C
q
(
0
+1)) and u pos
+
(w
q
, c
q
,
0
), then
c
q
0
(u) pos(u, c
p
,
0
,
1
).
For p Q
2
D
, C
p
and u pos
+
(w
p
, c
p
, ) we let p
u
def
= (u, C
p
, c
p
(C
p
)).
In [RS05, Problem 6.1] we asked if support iterations of forcing notions Q
2
D
are proper. Now we may answer this question positively (assuming that is
strongly inaccessible). First, let us state some auxiliary denitions and facts.
Proposition 6.9. (1) Q
2
D
is a (<)complete forcing notion of cardinality 2
.
(2) If p Q
2
D
and C
p
, then
for each u pos
+
(w
p
, c
p
, ), p
u Q
2
D
is a condition stronger than
p, and
the family p
u : u pos
+
(w
p
, c
p
, ) is pre-dense above p.
(3) Let p Q
2
D
and < be two successive members of C
p
. Suppose that
for each u pos
+
(w
p
, c
p
, ) we are given a condition q
u
Q
2
D
such that
p
c
p
(u) q
u
. Then there is a condition q Q
2
D
such that letting
=
min(C
q
) we have
(a) p q, w
q
= w
p
, C
q
= C
p
and c
q
= c
p
for C
q
, and
(b)
_
w
qu
: u pos
+
(w
p
, c
p
, )
_

, and
(c) q
u
q
c
q
(u) for every u pos

+
(w
p
, c
p
, ).
(4) Assume that p Q
2
D
, C
p
and
is a Q
2
D
name such that p
V.
Then there is a condition q Q
2
D
stronger than p and such that
(a) w
q
= w
p
, C
q
and C
q
= C
p
, and
(b) if u pos
+
(w
q
, c
q
, ) and = min(C
q
( + 1)), then the condition
q
c
q
(u) forces a value to
.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

Proof. Fully parallel to [RS05, Proposition 5.1].
Denition 6.10. The natural limit of an
Q
2
D
increasing sequence p = p
: <
) Q
2
D
(where < is a limit ordinal) is the condition q = (w
q
, C
q
, c
q
) dened
as follows:
w
q
=
<
w
p
, C
q
=
<
C
p
and
c
q
= c
q
: C
q
) is such that for C
q
and u we have c
q
(u) =
<
c
p
(u).
Proposition 6.11. (1) Suppose p = p
: < ) is a
Q
2
D
increasing sequence
of conditions from Q
2
D
such that
(a) w
p
= w
p0
for all < , and
(b) if < is limit, then p
is the natural limit of p, and

(c) for each < , if C
p
, otp(C
p
) = , then C
p
+1
( + 1) =
C
p
( + 1) and for every C

p
+1
we have c
p
+1
= c
p
.
Then the sequence p has an upper bound in Q
2
D
.
(2) Suppose that p Q
2
D
and h
is a Q
2
D
name such that p h
: V.
Then there is a condition q Q
2
D
stronger than p and such that
() if <
are two successive points of C

q
, u pos(w
q
, c
q
, ), then the
condition q
c
q
(u) decides the value of h
( + 1).
Proof. Fully parallel to [RS05, Proposition 5.2].
Proposition 6.12. Assume that , S, D are as in 6.1. The forcing notion Q
2
D
is
reasonably merry over (S, D).
Proof. Let p Q
2
D
. We will describe a strategy st for Generic in the game
master
S,D
(p, Q
2
D
) - this strategy is essentially the same as the one in the proof of
[RS05, Proposition 5.4], only the argument that it is a winning strategy is dierent.
In the course of a play the strategy st instructs Generic to build aside an in-
creasing sequence of conditions p
= p
: < ) Q
2
D
such that for each < :
(a) p
0
= p and w
p
= w
p
, and
(b) if < is limit, then p
is the natural limit of p
, and
(c) if C
p
, otp(C
p
) = , then C
p
+1
( +1) = C
p
( + 1) and for
every C
p
+1
we have c
p
+1
= c
p
, and
(d) after stage of the play of
master
S,D
(p, Q
2
D
), the condition p
+1
is determined.
After arriving to the stage , Generic is instructed to pick C
p
such that
otp(C
p
) = , put = min(C
p
( +1)) and play as her innings of this stage:

I
= pos
+
(w
p
, c
p
, ) and p
u
= p
c
p
(u) for u I
.
Then Antigeneric answers with q
u
: u I
) Q
2
D
. Since p
c
p
(u) q
u
for
each u pos
+
(w
p
, c
p
, ), Generic may use 6.9(3) (with , , p
, q
u
here standing
for , , p, q
u
there) to pick a condition p
+1
such that, letting
= min(C
p
+1
),
we have
(e) p
+1
, w
p
+1
= w
p
, C
p
+1
= C
p
and c
p
+1
= c
p
for
C
p
+1
, and
(f)
_
w
q
u
: u I
, and
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

(g) q
u
p
+1
c
p
+1
(u) for every u I
.
This completes the description of st. Suppose that
, p
u
, q
u
: u I
) : <
_
is
the result of a play of
master
S,D
(p, Q
2
D
) in which Generic followed st and constructed
aside p
= p
: < ). By 6.11, there is a condition p
Q
2
D
stronger than all
p
(for < ). We claim that p
is an (S, D)knighting condition for the servant

q = q
u
: S & u I
). To this end consider the following strategy st
of COM
in
servant
S,D
( q, p
, Q
2
D
). After arriving to a stage S of a play of
servant
S,D
( q, p
, Q
2
D
),
when r
, A
: < ) has been already constructed, COM plays as follows.

If is a successor or /

<
A
, then she just puts r
, A
such that:
(h) r
for all < and if C

p
is such that otp( C

p
) = , then
w
r
( + 1) ,= , and
(i) A
<
A

<
C
r
(sup(w
r
) + 1).
If

<
A
is a limit ordinal, then COM rst lets u =
<
w
r
. It follows from
(h)+(i) from earlier stages that u , C
p
and otp( C
p
) = . Note that

<
C
r
, u

<
pos
+
(w
r
, c
r
, ) and u I
. Let
= min
_
C
p
( + 1)
_
and

<
C
r
( + 1). It follows from (c)+(g) that for each < we have

q
u
p
+1
c
p
+1
(u) p
c
p
(u) r
c
r
(u).
Hence COM may choose a condition r
u
stronger than all r
(for < ) and

satisfying (h). Then A
is given by (i).
It follows directly from the description of st
that it is a winning strategy of

COM in
servant
S,D
( q, p
, Q
2
D
).
The master game
master
S,D
used to dene the property of being reasonably merry
is essentially a variant of the Areasonable boundedness game
rcA
of [RS05, Def.
3.1]. The related bounding property was weakened in [RS, Def. 2.9] by introducing
double areasonably completeness game
rc2a
. We may use these ideas to introduce
a property much weaker than reasonably merry, though the description of the
resulting notions becomes somewhat more complicated. As an award for additional
complication we get a true preservation theorem, however. In the rest of this
section, in addition to 6.1 we assume also
Context 6.13. =
: < ) is a sequence of cardinals below such that

( < )(
0

=
||
).
Denition 6.14. Let Q be a forcing notion.
(1) A double Qservant over S, is a sequence
q =
, q
: S & <
)
such that for S,
0 <
< and q
Q (for <
),
_
i, i
<
__
j <
__
i
< i q
+j
q
i+j
_
.
(Here
is treated as an ordinal and
is the ordinal product of
and
.)
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

(2) Let q be a double Qservant over S, and let q Q. We dene a game
2ser
S,D,
( q, q, Q) as follows. A play of
2ser
S,D,
( q, q, Q) lasts at most steps
during which the players, COM and INC, attempt to construct a sequence
r
, A
: < ) such that

r
Q, q r
, A
D and < < r
.
The terms r
, A
are chosen successively by the two players so that

if / S, then INC picks r
, A
, and
if S, then COM chooses r
, A
.
If at some moment of the play one of the players has no legal move, then INC
wins; otherwise, if both players always had legal moves and the sequence
r
, A
: < ) has been constructed, then COM wins if and only if

()
_
S
__
[
<
A
& is limit ] (j <
)(i <
)(q
i+j
r
)
_
.
(3) If COM has a winning strategy in the game
2ser
S,D,
( q, q, Q), then we will
say that q is an knighting condition for the double servant q.
Denition 6.15. Let Q be a strategically (<)complete forcing notion.
2mas
S,D,
Generic and Antigeneric. A play of
2mas
S,D,
(p, Q) lasts steps and during a
play a sequence
, p
, q
: <
) : <
_
.
is constructed. (Again,
is the ordinal product of
and
.) Sup-
pose that the players have arrived to a stage < of the game. First,
Antigeneric picks a non-zero ordinal
< . Then the two players start a

subgame of length
alternately choosing the terms of the sequence

p
, q
: <
). At a stage =
i + j (where i <
, j <
) of
the subgame, rst Generic picks a condition p
Q stronger than p and

stronger than all conditions q
for < of the form =
+j (where
i
< i), and then Antigeneric answers with a condition q
stronger than p
.
At the end, Generic wins the play if
(a) there were always legal moves for both players (so a sequence
, p
, q
: <
) : <
_
has been constructed) and
(b) for each S, the conditions in p
j
: j <
are pairwise incom-

patible, and
(c) letting
q =
, q
: S & <
)
(it is a double Qservant over S) we may nd a knighting condition
q p for the double servant q.
(2) A forcing notion Q is reasonably double merry over (S, D, ) if (it is strate-
gically (<)complete and) Generic has a winning strategy in the game
2mas
S,D,
(p, Q) for any p Q.
Theorem 6.16. Assume that , S, D, are as in 6.1+6.13. Let

Q = P
, Q
: <
) be a support iteration such that for each < :
P
Q
is reasonably double merry over (S, D, ).

8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

Then P
= lim(
Q) is reasonably double merry over (S, D, ) (so also proper).

Proof. Combine the proof of [RS, Thm 2.12] (the description of the strategy here
is the same as the one there) with the end of the proof of 6.4(a).
7. Open problems
Problem 7.1. Let p = (

P, S, D) be a Tparameter. Does reasonably B-bounding
over p (see 1.8) imply reasonably B-bounding over D (of [RS05, Def. 3.1])? Does
reasonably B-bounding over p imply Bnoble over p ? (Note 6.6.)
Problem 7.2. Are there any relations among the notions of properness over
Dsemi diamonds (of [RS01]), properness over Ddiamonds (of [Eis03]) and B
nobleness (of 3.1)?
Problem 7.3. Does support iterations of forcing notions of the form Q
E
E
add
Cohens? Here we may look at iterations as in 3.3 or 2.7.
Problem 7.4. Does reasonably Bbounding over p (for a Tparameter p)
imply reasonably merry over (S, D) (for some S, D as in 6.1)?
Problem 7.5. Does fuzzy proper over quasi Ddiamonds for W (see [RS07,
Def. A.3.6]) imply reasonably merry? (Any result in this direction may require
additional assumptions on W,

Y in [RS07, A.3.1, A.3.3].)
References
[CS95] James Cummings and Saharon Shelah, Cardinal invariants above the continuum, Annals
of Pure and Applied Logic 75 (1995), 251268, math.LO/9509228.
[Eis03] Todd Eisworth, On iterated forcing for successors of regular cardinals, Fundamenta
Mathematicae 179 (2003), 249266.
[HR01] Tapani Hyttinen and Mika Rautila, The canary tree revisited, The Journal of Symbolic
Logic 66 (2001), 16771694.
[Jec03] Thomas Jech, Set theory, Springer Monographs in Mathematics, Springer-Verlag, Berlin,
2003, The third millennium edition, revised and expanded.
[MS93] Alan H. Mekler and Saharon Shelah, The canary tree, Canadian Journal of Mathematics.
Journal Canadien de Mathematiques 36 (1993), 209215, math.LO/9308210.
[RS] Andrzej Roslanowski and Saharon Shelah, Reasonable ultralters, again, Notre Dame
Journal of Formal Logic submitted, math.LO/0605067.
[RS99] , Norms on possibilities I: forcing with trees and creatures, Memoirs of the Amer-
ican Mathematical Society 141 (1999), no. 671, xii + 167, math.LO/9807172.
[RS01] , Iteration of -complete forcing notions not collapsing
+
., International Journal
of Mathematics and Mathematical Sciences 28 (2001), 6382, math.LO/9906024.
[RS05] , Reasonably complete forcing notions, Quaderni di Matematica 17 (2005),
math.LO/0508272.
[RS07] , Sheva-Sheva-Sheva: Large Creatures, Israel Journal of Mathematics 159 (2007),
109174, math.LO/0210205.
[She82] Saharon Shelah, Proper forcing, Lecture Notes in Mathematics, vol. 940, Springer-
Verlag, Berlin-New York, xxix+496 pp, 1982.
[She98] , Proper and improper forcing, Perspectives in Mathematical Logic, Springer,
1998.
[She00] , The Generalized Continuum Hypothesis revisited, Israel Journal of Mathematics
116 (2000), 285321, math.LO/9809200.
[She03a] , Not collapsing cardinals in (< )support iterations, Israel Journal of
Mathematics 136 (2003), 29115, math.LO/9707225.
[She03b] , Successor of singulars: combinatorics and not collapsing cardinals
in (< )-support iterations, Israel Journal of Mathematics 134 (2003), 127155,
math.LO/9808140.
8
8
8

r
e
v
i
s
i
o
n
:
2
0
1
0
-
0
7
-
1
6

m
o
d
i
f
i
e
d
:
2
0
1
0
-
0
7
-
1
6

[SS02] Saharon Shelah and Zoran Spasojevic, Cardinal invariants b and t, Publications de
LInstitute Mathematique - Beograd, Nouvelle Serie 72 (2002), 19, math.LO/0003141.
[Zap97] Jindrich Zapletal, Splitting number at uncountable cardinals, Journal of Symbolic Logic
62 (1997), 3542.
Department of Mathematics, University of Nebraska at Omaha, Omaha, NE 68182-
0243, USA
E-mail address: roslanow@member.ams.org
URL: http://www.unomaha.edu/logic
Einstein Institute of Mathematics, Edmond J. Safra Campus, Givat Ram, 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://shelah.logic.at

Andrzej Roslanowski and Saharon Shelah - Lords of The Iteration

Загружено:

Сведения о документе

Исходное описание:

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Andrzej Roslanowski and Saharon Shelah - Lords of The Iteration

Загружено:

Авторское право:

Доступные форматы

8

on ). The main result of the rst section (Theorem

). There will be some

) are total functions

and for p lim(

dom(p) we will let p() =

) be a model such that

tree is a pair T = (T, rk) such

tree for some w ,

is a statement asserting existence of a Tparameter with the lter

implies the existence of a Tparameter (

: < ) such that for each < :

for all < ,

: < ) be an increasing continuous

[ < (for < ). Then

: < ) is an increasing continuous sequence of ordinals below ,

< for each .) Next, consider

D. Let BS be a limit ordinal. Since [P

+1) such that f

rst Generic chooses a set I

of cardinality < and a system p

then Antigeneric answers by picking a system q

Q stronger than p and such that

: < ) is a support iteration,

is reasonably Bbounding over p[P

name for a function from to V, p P

, then there are q p

: < ) such that ( < )([A

(p) the set < : succ

is a complete tree and for each p

) and for < let B

) for each < , we know that B

< : is limit and / S.

for < and T

. Thus we have shown that p

be the condition determined at the end of part (2) and let B =

. Choose an increasing continuous sequence

: < ) and a complete

such that for every < we have

for an ordinal below , there are

: < ) be a sequence of cardinals below .

rst Antigeneric pics

After this, Generic chooses a system p

Antigeneric answers with a system of conditions q

Q stronger than p and such that

bounding over |, p, if for any p Q, Generic has a winning strategy

(while in the related game

: < ) is a sequence of non-zero cardinals below

has no inuence on the answers by

= 1 E (otherwise Generic wins by default). Like in

: < ) is a support iteration, w , [w[ < ,

tree, [T[ < ,

name for an ordinal.

then there are a set A

as the result of 3 stage composition P

(so it belongs to V); the

and then choose a P

< ). For each

for a winning strategy of Complete in the game

name for a condition

rst names we easily

is a standard tree of conditions. Now we use 2.6.1 to nd a set A