Professional Documents
Culture Documents
Andrzej Roslanowski and Saharon Shelah - Sheva-Sheva-Sheva: Large Creatures
Andrzej Roslanowski and Saharon Shelah - Sheva-Sheva-Sheva: Large Creatures
r
X
i
v
:
m
a
t
h
/
0
2
1
0
2
0
5
v
2
[
m
a
t
h
.
L
O
]
2
3
M
a
r
2
0
0
3
SHEVASHEVASHEVA: LARGE CREATURES
ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
Abstract. We develop the theory of the forcing with trees and creatures for
an inaccessible continuing Ros lanowski and Shelah [14], [13]. To make a
real use of these forcing notions (that is to iterate them without collapsing
cardinals) we need suitable iteration theorems, and those are proved as well.
(In this aspect we continue Ros lanowski and Shelah [15] and Shelah [16], [17].)
Contents
0. Introduction 1
A.1. Iterations of complete forcing notions and trees of conditions 4
A.2. Bounding properties 8
A.3. Fuzzy properness over 14
B.4. A creaturefree example 30
B.5. Trees and creatures 33
B.6. Getting completeness and bounding properties 36
B.7. Getting fuzzy properness 40
B.8. More examples and applications 43
References 47
Date: February 2003.
1991 Mathematics Subject Classication. Primary 03E35, 03E40; Secondary: 03E05, 03E55.
The rst author thanks the Hebrew University of Jerusalem for support during his visit to
Jerusalem in Summer2001 and he also acknowledges partial support from University Commit-
tee on Research of the University of Nebraska at Omaha. He also thanks his wife, Ma lgorzata
JankowiakRos lanowska for supporting him when he was preparing the nal version of this paper.
The research of the second author was partially supported by the United States-Israel Bina-
tional Science Foundation. Publication 777.
0
SHEVASHEVASHEVA: LARGE CREATURES 1
0. Introduction
The present paper has two themes.
The rst is related to the quest for the right generalization of properness to higher
cardinals (that is, for a property of forcing notions that would play in iterations with
uncountable supports similar role to that of standard properness in CS iterations).
The evidence that there is no straightforward generalization of properness to larger
cardinals was given already in Shelah [18] (see [19, Appendix 3.6(2)]). Substantial
progress has been achieved in Shelah [16], [17], but the properties there were tailored
for generalizing the case no new reals of [19, Ch. V]. Then Roslanowski and Shelah
[15] gave an iterable condition for not collapsing
+
in support iterations of
complete forcing notions (with possibly adding subsets of ). Very recently Eisworth
[6] has given another property preserved in support iterations (and implying that
+
is not collapsed). At the moment it is not clear if the two properties (the one
of [15] and that of [6]) are equivalent, though they have similar avour. However,
the existing iterable properties still do not cover many examples of natural forcing
notions, specially those which come naturally in the context of reals. This brings
us to the second theme: developing the forcing for reals.
A number of cardinal characteristics related to the Baire space
, the Cantor
space
2 and/or the combinatorial structure of []
2 and []
, giving
simple constraints on the triple of cardinals (b
, d
, 2
>
+
needs large cardinals. One of the sources
of interest in characteristics of the reals is their relevance for our understanding
of the club lter on (or the dual ideal on non-stationary subsets of ) see,
e.g., Balcar and Simon [2, 5], Landver [9], Matet and Pawlikowski [10], Matet,
Roslanowski and Shelah [11]. First steps toward developing forcing for reals has
been done long time ago: in 1980 Kanamori [8] presented a systematic treatment
of the perfectset forcing in products and iterations. Recently, Brown [3], [4]
discussed the superperfect forcing and other treelike forcing notions.
Our aim in this paper is to provide tools for building forcing notions relevant
for reals (continuing in this Roslanowski and Shelah [14], [13]) and give suitable
iteration theorems (thus continuing Roslanowski and Shelah [15]). However, we
restrict our attention to the case when is a strongly inaccessible uncountable
cardinal (after all,
0
is inaccessible), see 0.3 below.
The structure of the paper is as follows. It is divided into two parts, rst one
presents iteration theorems, the second one gives examples and applications. In
Section A.1 we present some basic notions and methods relevant for iterating
2 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
complete forcing notions. The next section, A.2, presents preservation of
analogue of the Sacks property (in Theorem A.2.3) as well as preservation of being
bounding (in Theorem A.2.6). Section A.3 introduces fuzzy properness, a more
complicated variant of properness over semi-diamonds from [15]. Of course, we
prove a suitable iteration theorem (see Theorem A.3.10). Then we give examples
for the properties discussed in Part A. We start with showing that a forcing no-
tion useful for uniformization is fuzzy proper (in Section B.4), and then we turn
to developing forcing notions built with the use of trees and creatures. In Section
B.5 we set the terminology and notation, and in the next section we discuss when
the resulting forcing notions have the two bounding properties discussed in A.2.
Section B.7 shows how our methods result in suitably proper forcing notions, and
the last section introduces some new characteristics of the reals.
Notation Our notation is rather standard and compatible with that of classical
textbooks (like Jech [7]). In forcing we keep the older convention that a stronger
condition is the larger one. Our main conventions are listed below.
Notation 0.1. (1) For a forcing notion P,
P
stands for the canonical Pname
for the generic lter in P. With this one exception, all Pnames for objects
in the extension via P will be denoted with a tilde below (e.g.,
, X
). 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). We will
also assume that all forcing notions under considerations are atomless.
By support iterations we mean iterations in which domains of con-
ditions are of size . However, we will pretend that conditions in a
support iteration
Q = P
, Q
: <
and for
p lim(
Q) and
.
(2) For a lter D on , the family of all Dpositive subsets of is called D
+
.
(So A D
+
if and only if A and A B ,= for all B D.)
The club lter of is denoted by T
.
(3) 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).
Cardinal numbers will be called , , , (with possible sub- and super-
scripts); is a xed inaccessible cardinal (see 0.3).
(4) By we 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().
(5) A bar above a letter denotes that the object considered is a sequence;
usually
X will be X
i
: i < ), where is the length lh(
X) of
X. Sometimes
our sequences will be indexed by a set of ordinals, say S , and then
X
will typically be X
: S).
But also, , and (with possible sub- and superscripts) will denote
sequences (nodes in quasi trees).
For two sequences , we write whenever is a proper initial
segment of , and when either or = .
(6) We will consider several games of two players. One player will be called
Generic or Complete or just I player, and we will refer to this player as
she. Her opponent will be called Antigeneric or Incomplete or just II
player and will be referred to as he.
SHEVASHEVASHEVA: LARGE CREATURES 3
Denition 0.2. (1) A quasi tree is a set T of sequences of length < with
the smallest element denoted by root(T).
(2) A quasi tree T is a tree if it is closed under initial segments longer
then lh(root(T)).
(3) A quasi tree is complete if the union of any increasing sequence of
length less than of members of T is in T.
(4) For a quasi tree T and T we dene the successors of in T, maximal
points of T, the restriction of T to , and the height of T by:
succ
T
() = T : & ( T)( ),
max(T) = T : there is no T such that ,
T
[]
= T : , and ht(T) = suplh() : T.
We put
T = T max(T).
(5) For < and a quasi tree T we let
(T)
(T)
def
= : is a sequence and ( < )( > )( T).
(6) A subset F of a quasi tree T is a front of T if no two distinct members
of F are comparable and
( lim
= ,
(c) for each < ,
<
<
and ( <
)([[
<
).
4 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
Part A
Iteration theorems for support iterations
A.1. Iterations of complete forcing notions and trees of conditions
In this section we recall some basic denitions and facts concerning complete
forcing notions and support iterations.
Denition A.1.1. Let P be a forcing notion.
(1) For a condition r P and a set S , let
0
(P, S, r) be the following game
of two players, Complete and Incomplete:
the game lasts moves and during a play the players con-
struct 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:
if i S, then Complete chooses p
i
and Incomplete chooses
q
i
, and
if i / S, then Incomplete chooses p
i
and Complete chooses
q
i
.
Complete wins if and only if for every i < there are legal moves for both
players.
(2) We say that the forcing notion P is (, S)strategically complete if Complete
has a winning strategy in the game
0
(P, S, r) for each condition r P. We
say that P is strategically (<)complete if it is (, )strategically complete.
(3) We say that P is complete if every
P
increasing chain of length less than
has an upper bound in P.
(4) Let N (H(), , <
N for an ordinal
we have p
N .
(5) 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 A.1.2. (1) On strategic completeness (and variants) see [16, A.1].
Plainly, completeness implies strategic (<)completeness.
(2) Note that if P is strategically (<)complete and D is a normal lter on
, then in V
P
the lter D generates a proper normal lter on . [Abusing
notation, we may call this lter also by D.]
Proposition A.1.3. Suppose that P is a (<)strategically complete (atomless)
forcing notion,
< and q
P (for <
P
(for <
) such that q
<
the conditions p
, p
are incompatible.
SHEVASHEVASHEVA: LARGE CREATURES 5
Proof. For <
let st
0
(P, , q
). By induction on i <
we dene conditions q
i
, p
i
as follows:
p
0
0
= q
0
, q
0
0
is the answer of Complete to p
0
0
) according to st
0
, q
0
= p
0
= q
for
> 0.
Suppose that conditions p
j
, q
j
(where
i <
) so that
() ( <
< i)(q
, q
are incompatible ),
() for each < i, (p
j
, q
j
) : j < i) is a play of
0
(P, , q
) in which
Complete uses the strategy st
, and
() p
j
= q
j
= q
for i > j.
For < i let r
= r
for < i, p
i
= q
= r
for < i, ,=
0
and p
i
= q
is
dened as the answer of Complete according to st
to (p
j
, q
j
) : j < i)
p
i
), and
q
i
= q
for > i.
After the inductive denition is carried out we may pick upper bounds p
to q
j
:
j <
) (for <
, Q
: <
P
Q
is complete.
Then the forcing P
is complete.
Proposition A.1.5. Suppose
Q = P
i
, Q
i
: i < ) is a support iteration and for
each i <
Pi
Q
i
is strategically (<)complete .
Then:
(a) P
is strategically (<)complete.
(b) Moreover, for each and r P
0
(P
0
(P
0
, , r
0
) in which Complete
follows the strategy st(
0
, r
0
),
then (p
i
r[
0
,
1
), q
i
r[
0
,
1
)) : i < ) is a play of
0
(P
1
, , r) in
which Complete uses st(
1
, r);
(ii) if (p
i
, q
i
) : i < ) is a play of
0
(P
1
, , r) in which Complete plays
according to the strategy st(
1
, r),
then (p
i
0
, q
i
0
) : i < ) is a play of
0
(P
0
, , r
0
) in which Com-
plete uses st(
0
, r);
(iii) if (p
i
, q
i
) : i < i
) is a partial play of
0
(P
1
, , r) in which Complete
uses st(
1
, r) and p
P
0
is stronger than all p
i
0
(for i < i
), then
there is p
P
1
such that p
= p
0
and p
p
i
for i < i
.
6 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
Denition A.1.6 (Compare [16, A.3.3, A.3.2]). (1) Let , be ordinals, , =
w . A standard (w, )
: w ), where
each (t)
is a sequence of length ,
(T, ) is a tree with root ) and such that every chain in T has a
upper bound it T.
We will keep the convention that T
x
y
is (T
x
y
, rk
x
y
).
(2) Suppose that w
0
w
1
,
0
1
, and T
1
= (T
1
, rk
1
) is a standard
(w
1
,
1
)
tree T
0
= (T
0
, rk
0
) such that
T
0
= (t)
0
: w
0
rk
1
(t)) : t = (t)
: w
1
rk
1
(t)) T
1
.
The mapping
T
1
(t)
: w
1
rk
1
(t)) (t)
0
: w
0
rk
1
(t)) T
0
will be denoted proj
(w1,1)
(w0,0)
too.
(3) We say that
T = T
: <
: <
) of subsets of we have
(i) T
is a standard (w
, )
),
(ii) if < <
, then T
= proj
(w
,)
(w,)
(T
).
(4) Suppose that
T = T
: <
is a
limit ordinal. Let w
be such that T
is a standard (w
, )
tree (for
<
) and let w =
<
lim(
T ) of
T is a standard
(w,
tree (T
lim
, rk
lim
) such that
() T
lim
consists of all sequences t satisfying
(i) Dom(t) is an initial segment of w (not necessarily proper),
(ii) if Dom(t), then (t)
is a sequence of length
,
(iii) (t)
: w
Dom(t)) T
.
(5) A legal sequence
T = T
: <
, T
lim(
T ).
(6) Let
Q = P
i
, Q
i
: i < ) be a support iteration. A standard tree of
conditions in
Q is a system p = p
t
: t T) such that
(T, rk) is a standard (w, )
w1,1
w0,0
p
1
(or just p
0
p
1
) whenever for each t T
1
, letting t
= proj
(w1,1)
(w0,0)
(t) T
0
,
we have p
0
t
rk
1
(t) p
1
t
.
Remark A.1.7. Concerning Denition A.1.6(4), note that even though T
lim
could
be empty, it does satisfy requirements of A.1.6(1) (so
lim(
T ) is indeed a standard
(w,
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 standard tree of conditions in
Q, [T[ < , and
1 P
0
(P
, , r) as in A.1.5(b). Let
T
max
def
= t T : (t
T)(t t
) = t
: <
(where < is a cardinal). We construct partial plays (p
i
, q
i
) : i ) of
0
(P
rk(pt
)
, , p
t
), p
t
) and
such that
() if < and rk(p
t
) = , then p
1,
() if t t
, t t
, t T, , < , i ,
then p
i
rk(t) = p
i
rk(t) and q
i
rk(t) = q
i
rk(t).
So suppose we have dened p
j
, q
j
for < , j < i < . First we look at (p
i
j
, q
i
j
) :
j < i) it is a play of
0
(P
rk(pt
i
)
, , p
ti
) in which Complete uses st(rk(p
ti
), p
ti
), so
we may nd a condition p
i
i
P
rk(pt
i
)
stronger than all p
i
j
, q
i
j
for j < i, and such
that rk(p
ti
) = p
i
i
1. Next, for < , ,= i, we dene p
i
as follows. Let
t T be such that t t
, t t
i
and rk(t) is the largest possible. We declare that
Dom(p
i
) = (Dom(p
i
i
) rk(t))
_
j<i
Dom(q
j
) Dom(p
t
),
and p
i
rk(t) = p
i
i
rk(t), and for [rk(t), ) we have that p
i
() is the <
rst
P
such that
p
i
P
p
i
() is an upper bound to p
t
() q
j
() : j < i .
The denition of p
i
s is correct by A.1.5(b)(iii+ii). Also, by the choice of the
<
i
s.
Finally we dene q
i
(for < ) as the condition given to Complete by st(rk(t
), p
t
)
in answer to (p
j
, q
j
) : j < i)
i
). (Again, one easily veries (), ().)
The conditions p
, q
, q
.
It should be clear that q = q
t
: t T) is as required.
Let us close this section with recalling an important result on easy ensuring
that support iteration satises the
++
cc. Its proof is a fairly straightforward
modication of the proof of the respective result for CS iterations; see [19, Ch. III,
Thm 4.1], Abraham [1, 2] for the CS case, Eisworth [6, 3] for the general case of
support iterations.
8 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
Theorem A.1.9. Assume 2
=
+
,
<
= . Let
Q = P
i
, Q
i
: i <
++
) be
support iteration such that for all i <
++
we have
P
i
is proper,
Pi
[Q
i
[
+
.
Then the limit P
++ satises the
++
cc.
A.2. Bounding properties
The results on preservation in CS iterations of properties like the Sacks property
and
bounding property were among the earliest in the theory of proper forcing.
Here we introduce relatives of these two properties for reals and we show suitable
iteration theorems. For both properties, the properness is built into the property.
Recall that ,
(i
0
, p, P)
of two players, the Generic player and the Antigeneric player. A play lasts
moves indexed by ordinals from the interval [i
0
, ), and during it the
players construct a sequence (s
i
, q
i
, p
i
) : i
0
i < ) as follows. At stage
i of the play (where i
0
i < ), rst Generic chooses s
i
i+1
and a
system q
i
= q
i
: s
i
i+1
) such that
() s
i
is a complete tree of height i + 1, and
( s
i
)( s
i
)( & lh() = i + 1),
and lh(root(s
i
)) = i
0
,
() for all j such that i
0
j < i we have s
j
= s
i
j+1
,
() q
i
P for all s
i
i+1
, and
() if i
0
j < i, s
i
j+1
and s
i
i+1
, then p
j
q
i
and
p q
i
,
() [s
i
i+1
[ <
i
.
Then Antigeneric answers choosing a system p
i
= p
i
: s
i
i+1
) of
conditions in P such that q
i
p
i
for each s
i
i+1
.
The Generic player wins a play if she always has legal moves (so the play
really lasts steps) and there are a condition q p and a Pname
such
that
() q
P
&
_
i [i
0
,
_
)
_
(i + 1) s
i
& q
i
(i+1)
P
_
.
(2) We say that P has the strong
Sacks property whenever the Generic player
has a winning strategy in the game
Sacks
(i
0
, p, P) for any i
0
< and p P.
(3) We say that P has the
Sacks property if for every p P and a Pname
such that p
:
< ) such that [a
[ <
() a
) .
Remark A.2.2. (1) At a stage i < of a play of
Sacks
(i
0
, p, P), the Antigeneric
player may play stronger conditions, and using A.1.3 we may require that
if p
i
= p
i
: s
i
i+1
) is his move, then the conditions p
i
are pairwise
incompatible.Thus the winning criterion () could be replaced by
()
q
P
_
i [i
0
,
_
)
_
s
i
i+1
__
q
i
P
_
SHEVASHEVASHEVA: LARGE CREATURES 9
(thus eliminating the use of
by an object
).
(2) Note that if Generic has a winning strategy in
Sacks
(i
0
, p, P) for all i
0
< . (Remember: the sequence
is
increasing.) The reason why we have i
0
as a parameter is a notational
convenience.
(3) Plainly, if Generic has a winning strategy in
Sacks
(i
0
, p, P), then she has
one with the following property:
(
nice
) if s
i
, q
i
are given to Generic as a move at a stage i [i
0
, ), then for
every s
i
i
, the set < :
) s
i
is an initial segment of
i
and (j) = 0 for all j < i
0
.
Strategies satisfying the condition (
nice
) will be called nice.
(4) Easily, if P has the strong
Sacks property, then it is strategically (<)
complete and has the
Sacks property.
Theorem A.2.3. Suppose that
Q = P
, Q
P
Q
has the
Sacks property.
(b) If N (H(), , <
), [N[ = ,
<
N N and
, , p,
Q, P
, . . . N,
p P
)generic condition r P
stronger than p.
Proof. (a) First note that each P
name q
for a condition in Q
, let st
(i
0
, q
)
be the <
rst P
(i
0
, q
, Q
).
Let
be a P
. Pick a model N
(H(), , <
) such that
, ,
, p,
Q, P
N is a P
, then
st
(i
0
, q
) N. Also, as
Q is a support iteration of (<)strategically complete
forcing notions, we may use A.1.5 inside N and for each N and r P
N
x a winning strategy st
0
(P
, , r) so that
conditions (i)(iii) of A.1.5(b) hold true.
Fix a list
1 = 1
i<
w
i
= N ).
By induction on i < we dene sequences T
i
: i < ) and p
i
, p
i
: i < is not
a limit ordinal ) such that the following requirements are satised.
() T
i
: i < ) is a continuous legal sequence of trees; T
i
N is a standard
(w
i
, i)
tree, [T
i+1
[ <
i
, and (t T
i
)(t
T
i
)(t t
& rk
i
(t
) = ).
10 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
() For i < and t T
i
such that rk
i
(t) < let
i
(t) = (s)
rki(t)
: t s T
i
.
Then (for each i, t as above) , =
i
(t)
j<i
j
and for each
i
(t) and
i
< (rk
i
(t)) we have (i
) = 0.
() If N , () < i < j < , t T
j
, rk
j
(t) = and t
= proj
wj,j
wi,i
(t) T
i
(so rk
i
(t
) = ), then
i
(t
) = i :
i
(t).
() T
0
= ), p
0
= p
0
), p
0
= p
i+1
,t
: t T
i+1
) are standard trees of conditions in
Q, both
belonging to N and such that p
i+1
p
i+1
.
() If i < j < , then p
i+1
wj+1,j+1
wi+1,i+1
p
j+1
.
() If t
i+1
T
i+1
(for i < ) are such that rk
i+1
(t
i+1
) = and t
i+1
=
proj
wj+1,j+1
wi+1,i+1
(t
j+1
) (for i < j), then p
i+1
,ti+1
, p
i+1
ti+1
: i < ) is a play of
the game
0
(P
(, p).
() If t T
i+1
, rk
i+1
(t) = , then p
i+1
t
1
(i).
() Assume that N, () = i
0
i and t T
i+1
is such that rk
i+1
(t) = .
Let, for j i, t
j
= proj
wi+1,i+1
wj,j
(t) and let r
be the <
rst P
name for a
member of Q
such that
= p() .
Furthermore, for i
0
j i and
j+1
(t
j+1
), x s
j+1
T
j+1
such that
rk
j+1
(s
j+1
) > and (s
j+1
= , t
j+1
s
j+1
, and put r
= p
j+1
s
j+1
().
Then the condition p
i+1
t
forces in P
the following:
there is a partial play s
j
, q
j
, r
j
: i
0
j i) of the game
Sacks
(i
0
, r
, Q
(i
0
, r
) and, for i
0
j i,
s
j
j+1
=
j+1
(t
j+1
) and r
j
= r
: s
j
j+1
).
Let us describe how the construction of T
i
: i < ) and p
i+1
: i < ) is carried
out. We start with letting T
0
= ), p
0
lim(T
j
: j < i)) N ( p
i
, p
i
) = T
i0
, and if
t = (t)
: w
i
rk
(t)) T
, then (t)
(i
0
) <
i0
, and if () = i
0
, then
(t)
i
0
0. (Plainly T
N and [T
we dene a
condition q
t
P
rk
(t)
N and names
t
() for ordinals (for w
i
rk
(t)). For
this let us x t T
and let t
j
= proj
wi,i
wj,j
(t) T
j
for j < i. Put
Dom(q
t
) =
_
w
i
_
Dom(p
j
tj
) : j < i is not a limit
_
rk
(t),
SHEVASHEVASHEVA: LARGE CREATURES 11
and for Dom(q
t
) let q
t
() be a P
chosen as follows.
If Dom(q
t
) w
i
, then q
t
() is the <
rst P
if possible, then q
t
() is an upper bound to p
j
tj
() : j < i is non-limit .
If Dom(q
t
) w
i
, then
t
() N is a P
rst P
be the <
rst P
such that
if possible, then r
is an upper bound to p
j
tj
() : j () is non-limit ,
else r
= p() .
Now, suppose that G
((), r
, Q
((), r
)
G
, and
s
j
j+1
=
j+1
(t
j+1
) and r
j
= r
j
: s
j
j+1
), where r
j
=
_
p
j+1
s
j+1
()
_
G
for s
j+1
T
j+1
such that t
j+1
s
j+1
, (s
j+1
= and rk
j+1
(s
j+1
) = . So we
may look at the answer s
i0
, q
i0
= q
i0
: s
i0
i0+1
) to this play according to
the strategy st
((), r
)
G
. Then, q
t
()
G
for j < i
0
, and such that
if (t
i
)
s
i0
, then q
t
()
G
= q
i0
(ti)
.
Also,
t
() = <
i
: (t
i0
)
) s
i0
.
[If () = i
0
, then we do not have the partial play we started with - the game just
begins and we look at the rst move of Generic, requiring that q
t
()
G
is stronger
than r
and if (t
i
)
s
i0
then q
t
()
G
= q
i0
(ti)
.]
This nishes the denition of q = q
t
: t T
wi,i
wj,j
q for all non-limit j i
0
.
(Note that if () = i
0
, t T
, and rk
= p
t
: t T
) N
such that q p
with rk
t
decides the
values of all names
() for t
, w
i
rk(t
(i
0
) (and
let p
(i
0
) =
t
i0
), and such that p
t
1
for all i
0
. For t T
with
rk
t
() by p
t
. Since
t
()
is a P
& rk
(t
1
) = & w
i
rk
(t
0
) p
t0
t0
() =
t1
().
So we may naturally dene
t
() also for t T
: ( w
i
rk
(t))((t)
(i
0
) <
t
()
and p
i
,t
= p
t
for t T
i
(thus dening p
i
). Plainly, T
i
N is a standard (w
i
, i)
p
i
and
if t T
i
, rk
i
(t) = , t
j
= proj
wi,i
wj,j
(t) for non-limit j i,
then p
j+1
tj+1
: j < i) is a sequence of answers of Complete in some partial play of
0
(P
(, p).
Now one easily veries that T
i
, p
i
, p
i
lim(T
j
: j < )). We will consider this standard (N , )
tree in
universes V
P
. But if (in V
P
) t T
i : w
i
rk
(t)) = proj
N,
wi,i
(t) V.
Also if i < is limit, then the equality T
i
=
lim(T
j
: j < i)) holds in V
P
as well.
We are going to dene a condition r P
for functions in
, and we will put t
= t
r
P
t
,
(ii)
r
P
(i < )
_
(p
i+1
t
|i+1
)
P
_
.
Arriving at a limit stage N ( + 1), we have dened r and t
, and
we should only check that conditions (i)
, (ii)
, (ii)
hold for
< , N).
RE: (i)
: T
: Suppose that G
i
= (t
[i)
GP
T
i
. Then, by (ii)
, we know that p
i+1
t
i+1
G
i+1
= p
i+1
t
i+1
(as t
i+1
t
i+1
), so remembering that
p
i+1
t
i+1
N we conclude p
i+1
t
i+1
G
.
Now suppose that we arrived at stage + 1 N ( + 1) and we have dened
r, t
so that (i)
+ (ii)
hold. Let G
be generic over V, r G
.
For i < let t
i
= (t
[i)
G
T
i
(remember (i)
). Plainly, t
j
= proj
wi,i
wj,j
(t
i
) for
j < i < . By (ii)
we get p
i+1
t
i+1
G
Let i
0
= () and let r
be the <
rst P
such
that (r
V, of course, and)
P
if there is a common upper bound to p
j
t
j
() : j i
0
is non-limit
then r
= p() .
(Note: for each j
j
: j < j
< , j
> i
0
for a moment. In V, for each i [i
0
, j
] and
i+1
(t
i+1
)
let us choose s
i+1
T
i+1
such that t
i+1
s
i+1
, (s
i+1
].
Since p
j
+1
t
+1
G
= s
i
, q
i
, r
i
: i
0
i j
) of the game
Sacks
(i
0
, r
G
, (Q
)
G
) in
SHEVASHEVASHEVA: LARGE CREATURES 13
which Generic uses st
(i
0
, r
G
) and s
i
i+1
=
i+1
(t
i+1
) and r
i
= (p
i+1
s
i+1
())
G
:
s
i
i+1
).
It should be clear that (in V[G
])
j
for i
0
< j
< j
< , so we have
a play =
i0<j
<
j
= s
i
, q
i
, r
i
: i
0
i < ) of the game
Sacks
(i
0
, r
G
, (Q
)
G
)
with the respective properties. This play is won by Generic, so there are a condition
q (Q
)
G
and a (Q
)
G
name
for a member of
such that q r
G
and
() q
(Q
)
G (i [i
0
, ))(
(i+1)
i+1
(t
i+1
) &p
i+1
s
i+1
(i+1)
()
G
(Q
)
G ) .
Let r(), t
name of a member of
Q
and t
is a P
+1
name of a member of
and r forces that they have the
property stated in ()). It follows from our choices that (i)
+1
+ (ii)
+1
hold true,
nishing the inductive construction of r P
and t
s.
For < let a
=
t
: t T
+1
& rk
+1
(t) = (remember:
t
is the value
forced to
() by p
+1
t
). Plainly, [a
[ <
)generic and r
P
( < )(
() a
) .
Proof of the Claim. First, by induction on N ( + 1) we are showing that
p r. There is nothing to do at limit stages, so let us deal with non-limit
ones. Assume we have shown p r.
Suppose that G
is generic over V, r G
. Let t
j
= (t
[j)
G
T
j
, and
let i
0
, r
be dened as in ()
. Since, by (ii)
, p
j
t
j
G
(for non-limit j i
0
)
and by the clause () of the construction, we get
V[G
j
()
G
: j i
0
is non-limit ,
and thus
V[G
] [= (j < i
0
)(p
j+1
t
j+1
()
G
r
G
) .
By the choice of r() we have r()
G
r
G
p()
G
.
Hence r p() r(), as needed.
Now, let G P
[i)
G
T
i
.
By (ii)
we know that p
i+1
ti+1
G. By clause () we have p
i+1
ti+1
1
i
and (by the
denition of a
i
) p
i+1
ti+1
a
i
. The former implies that G intersects 1 N for each
open dense subset of P
G
(i) a
i
.
(b) Included in the proof of (a).
Denition A.2.4. Let P be a forcing notion.
(1) For a condition p P and an ordinal i
0
< we dene a game
bd
(i
0
, p, P)
like
Sacks
(i
0
, p, P), but demand A.2.1(1()) is replaced by
()
[s
i
i+1
[ < .
(2) P has the strong bounding property if Generic has a winning strategy in
the game
bd
(i
0
, p, P) for every i
0
< , p P.
14 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
(3) P has the bounding property if for every p P and a Pname
such that
p
: < )
such that [a
() a
) .
Remark A.2.5. (1) All the remarks stated in A.2.2 have their (obvious) paral-
lels for the bounding properties.
(2) Clearly, (strong)
Sacks property implies (strong, respectively) bounding
property.
Theorem A.2.6. Suppose that
Q = P
, Q
P
Q
), [N[ = ,
<
N N and , p,
Q, P
, . . . N, p P
,
then there is an (N, P
)generic condition r P
stronger than p.
Proof. Basically the same as for A.2.3, just replacing each occurrence of
i
by .
A.3. Fuzzy properness over
A properness-type property preserved in support iterations, so called proper-
ness over semi-diamonds, was introduced in Roslanowski and Shelah [15]. That
property worked for any uncountable regular cardinal satisfying
<
= (not
necessarily strongly inaccessible), so because of the known ZFC limitations a num-
ber of natural forcing notions were not covered. For the context considered in this
paper we may do much better: fuzzy properness introduced in this section captures
more examples. Even though we do not prove a real preservation in support
iterations, our iteration theorem A.3.10 is satisfactory for most applications (see
sections B.4 and B.8 later).
In this section we x
), W [A]
, and
if a W, w [a]
<
, f : w a, then f a (hence also 0 a for a W),
(2) for every x H() there is a model N (H(), , <
: S) is a Dpre-diamond
sequence if
S D
+
contains all successor ordinals below , S is unbounded in
, 0 / S, and
F
for all S.
(2) A convenient Ddiamond is a Dpre-diamond
F = F
: S) such that
(f
)( S : F
f D
+
).
Denition A.3.3. Let P be a forcing notion. A base for P over W is a pair
(R,
Y) such that
(a) R P A is a relation such that
if (p, , x) R and p
P
p
, then (p
, , x) R,
SHEVASHEVASHEVA: LARGE CREATURES 15
(b)
Y = Y
a
: a W) where, for each a W, Y
a
: [a]
<
,
(c) if q P, a W, and < is a limit ordinal,
then there are p
P
q and x Y
a
() such that (p, , x) R.
If R is understood and (p, , x) R, then we may say p obeys x at .
Denition A.3.4. Let P be a forcing notion and let (R,
Y) be a base for P
over W. Also let a model N (H(), , <
: S) be a
Dpre-diamond sequence.
(1) Let
1 = 1
1exact if
( < )( < )(p
).
(2) We say that
F is a quasi Ddiamond sequence for (N, h, P) if for some
(equivalently: all) list
1 = 1
: < ) is
1exact D
(equivalently: p is
1exact) we have
E : ( < )(h F
() = p
) D
+
.
(3) For a limit ordinal S we dene }() = }(N, P, h,
F, R,
Y, ) as the set
_
x Y
a
() : if h F
() : < ) is a
P
increasing sequence
of conditions from P,
then there is a condition p P such that
( < )(h F
()
P
p) and (p, , x) R
_
(4) Let
1 = 1
)(q
,x
1
, and h F
() : < ) is a
P
increasing
1exact sequence of members of P N,
then ( < )(h F
()
P
q
,x
) and (q
,x
, , x) R.
If above X
(r, N,
1, h, P,
F, q) of
two players, the Generic player and the Antigeneric player, as follows. A
play lasts moves, in the i
th
move a condition r
i
P and a set C
i
D
are chosen such that (j < i)(r r
j
r
i
), and Generic chooses r
i
, C
i
if i S = Dom(
F), and Antigeneric chooses r
i
, C
i
if i / S. In the end
Generic wins the play if
() ( < )(i < )(p P N)(p 1
& p r
i
), and
16 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
() if S
i<
C
i
is a limit ordinal, hF
() : < ) is a
P
increasing
() r
i
),
then for some x A
we have q
,x
r
.
(6) Let q be a weak fuzzy candidate over
F for (N, h, P, R,
Y,
1). We say that
a condition r P is (R,
Y)fuzzy generic for q (over (N,
1, h, P,
F)) if
Generic has a winning strategy in the game
fuzzy
(r, N,
1, h, P,
F, q).
Remark A.3.5. (1) For any two lists
1
1
,
1
2
of open dense subsets of P from N,
on a club E of we have
1
1
: < = 1
2
: <
for E. Thus the corresponding notions of exactness agree for E. As
the generic player can choose C
i
E, in A.3.4(4,5,6) we allow not mention
1 as a parameter.
(2) Plainly, every fuzzy candidate is a weak fuzzy candidate.
Denition A.3.6. Let P be a complete forcing notion.
(1) We say that P is fuzzy proper over quasi Ddiamonds for W whenever for
some base (R,
Y) for P over W and for some c H(),
() if N (H(), , <
), [N[ = ,
<
N N, , P, c, R N, and
a
def
= N A W, p P N,
h : N satises P N Rng(h), and
F is a quasi Ddiamond for (N, h, P) and q is a fuzzy candidate
over
F,
then there is r P stronger than p and such that r is (R,
Y)fuzzy generic
for q.
(We may call (R,
Y) and c witnesses for fuzzy properness.)
(2) P is strongly fuzzy proper over quasi Ddiamonds whenever for some base
(R,
Y) for P over W and for some c H(),
()
+
if N (H(), , <
), [N[ = ,
<
N N, , P, c, R N, and
a
def
= N A W, p P N,
h : N satises P N Rng(h),
F is a quasi Ddiamond for (N, h, P) and q is a weak fuzzy
candidate over
F,
then there is a condition r P stronger than p such that r is (R,
Y)fuzzy
generic for q.
(3) P is weakly fuzzy proper over quasi Ddiamonds whenever for some base
(R,
Y) for P over W and for some c H(),
()
if N (H(), , <
), [N[ = ,
<
N N, , P, c, R N, and
a
def
= N A W, p P N,
h : N satises P N Rng(h),
then for some quasi Ddiamond
F for (N, h, P) and a weak fuzzy candidate
q over
F, there is a condition r P stronger than p such that r is
(R,
Y)fuzzy generic for q.
(4) P is fuzzy proper for W if it is fuzzy proper over quasi D
diamonds for
every normal lter D
)
is such that [A
:
S is limit ).) Thus fuzzy properness has a avour of a weaker property. However,
the dierences in technical details of the conditions introduced in this section and
those in [15] and/or [6] make it unclear if there are any implications between the
properness conditions in this section and those in the other two papers.
Proposition A.3.8. Let N, P, h,
1, R,
Y be as in A.3.4,
F = F
: S) be a
Dpre-diamond.
(1) If the forcing notion P is -complete, then there is a fuzzy candidate q over
F for (N, h, P, R,
Y,
1). In fact we can even demand:
(+) for every < , for every large enough S, q
,x
1
for all
x }().
(2) If r is (R,
Y)fuzzy generic for some weak fuzzy candidate q, then r is
(N, P)generic (in the standard sense).
(3) Assume that a condition r is (N, P)generic (in the standard sense),
F is a
quasi Ddiamond and q is a weak fuzzy candidate over (N,
1, h, P,
F). Sup-
pose that Generic has a strategy in the game
fuzzy
(r, N,
1, h, P,
F, q) which
guarantees that the result r
i
, C
i
: i < ) of the play satises A.3.4(5)().
Then she has a winning strategy in
fuzzy
(r, N,
1, h, P,
F, q) (i.e., one en-
suring () + () of A.3.4(5)).
(4) If P is fuzzy proper over quasi Ddiamonds, then it is weakly fuzzy proper
over quasi Ddiamonds. If P is strongly fuzzy proper over quasi Ddiamonds,
then it is fuzzy proper over quasi Ddiamonds.
(5) Assume that P is weakly fuzzy proper over quasi Ddiamonds, , Y
[]
, A
H(), W
[A
(Y, A
, W
V). Then:
(a) forcing with P does not collapse
+
,
(b) forcing with P preserves the following two properties:
(i) Y is a conal subset of []
(under inclusion),
(ii) for every x H() there is N (H(), , <
under
the relevant lter).
Proof. (1) Immediate (by the completeness of P; remember A.3.3(c) and that
R N; note that Y
a
() N).
(2) Remember that 0 / S, so in the game
fuzzy
(r, N,
1, h, P,
F, q) the condition
r
0
is chosen by Antigeneric. So if the conclusion fails, then for some Pname
N
for an ordinal we have r ,
=
0
for some ordinal
0
/ N, what guarantees him to win the play
(remember clause () of A.3.4(5)).
(3) The Generic player modies her original strategy as follows. During the play
she builds aside a
P
increasing sequence of conditions p
i
: i S) P N.
Arriving to stage i + 1, i S, she has two sequences: r
j
, C
j
: j i) (of the
play) and p
j
: j i S) such that p
j
r
j
. Now Generic picks p
i
P N such
that
(j i S)(p
j
p
i
) and ( < i)(p
i
1
),
18 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
and p
i
, r
i
are compatible (remember: we assumed that r is (N, P)generic). Next
she replaces r
i
by a common upper bound of p
i
and r
i
, pretending that that was
the condition played by her opponent, and then she plays according to her original
strategy. One easily veries that this is a winning strategy for the Generic player.
(4) Straightforward (remember that, by A.3.1(3), there is a quasi Ddiamond and
by A.3.8(1) there is a fuzzy candidate over it).
(5) Follows from (2).
Proposition A.3.9.
+
complete forcing notions are strongly fuzzy proper for W.
Proof. This is essentially a variant of [15, 2.5], but since we did not give the proof
there, we will present it fully here.
So suppose that a forcing notion P is
+
complete. Let R
tr
= R
tr
(P) be the
trivial relation consisting of all triples (p, , 0) such that p P and < and let
Y
tr
be such that Y
tr
a
() = 0 (for each < , a W). Assume now that
N (H(), , <
), [N[ = ,
<
N N, , P N, and a
def
= N A W,
p P N, and h : N satises P N Rng(h),
F = F
: S is limit ).
Let
1 = 1
i
: i < )
P N such that
r
0
= p, r
i+1
i
1
,
if there is an upper bound to r
j
: j < i q
i
, then r
i
is such an upper
bound.
Then we pick any upper bound r to the sequence r
i
: i < ) (remember: P is
+
complete). Now we want to argue that Generic has a winning strategy in the game
fuzzy
(r, N,
1, h, P,
F, q). Since r is (N, P)generic it is enough to give a strategy
for the Generic player which ensures that the result of the play satises A.3.4(5)()
(by A.3.8(3)). To this end note that there is a club E
0
of such that
every member of E
0
is a limit of ordinals from S,
for every E
0
and i < ,
q P : q r
i
or q, r
i
are incompatible 1
: < ,
Let Generic play so that arriving to a stage S of the play she puts the <
rst
upper bound to the conditions played so far and E
0
. Why does this strategy work?
Let r
i
, C
i
: i < ) be the result of the play in which Generic plays as described
above and let S
i<
C
i
be a limit ordinal such that h F
() : < ) is a
P
increasing
1exact sequence and
( < )(i < )(h F
() r
i
).
Then no r
i
, h F
() : < ) is
1exact) we have also
(i < )( < )(r
i
h F
()),
SHEVASHEVASHEVA: LARGE CREATURES 19
and hence q
i
(for i < ). Therefore q
.
Theorem A.3.10. Let A, W, D be as in A.3.1 and let
Q = P
, Q
: <
)
be a support iteration of complete forcing notions, and assume that
A.
Suppose also that for each <
we have
Y
and P
names R
, c
such that
,
Y
) and c
.
Then P
= lim(
(for
<
s are meaningful.
Let us x a convenient Ddiamond sequence
F
= F
: S) (so in particular,
S D
+
contains all successors, S is unbounded in and 0 / S). Put
E
0
def
= < : is a limit of points from S , E
1
def
= ( S) E
0
.
Plainly, E
0
, E
1
are clubs of . Let i
= (So E
2
is a club of too).
For each a W x a one-to-one mapping
a
: a
such that
a
(0) = 0
(say,
a
is the <
= (
a
)
1
[i
] (so
a
<
w
a
).
For
let R
[]
consist of all triples (p, , x) P
: w) and ( w)(p
P
(p(), , x
) R
).
Next, for
a
() : w
a
,
thus dening Y
[]
a
and
Y
[]
= Y
[]
a
: a W). If =
, (R
[]
,
Y
[]
) is a base for P
over W.
Proof of the Claim. Immediately by the denition of R
[]
,
Y
[]
we see that clauses
(a), (b) of A.3.3 hold (note: Y
[]
a
() a by A.3.1(1)). Now, to verify A.3.3(c),
suppose q P
let p
(), x
be
P
() q() & x
a
() & (p
(), , x
) R
,
and for Dom(q) w
a
let p
stronger
than q (and names x
). Since P
and
x
a
() (for w
a
) such that p
P
x
= x
(for w
a
). Then,
by A.3.3(a), we have p
P
(p(), , x
) R
(for each w
a
), and hence
(p, , x
: w
a
)) R
[]
.
20 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
Our aim now is to show that P
: <
), R
: <
), R,
Y, S, D,
F
,
Q). So suppose that a model
N (H(), , <
) satises
[N[ = ,
<
N N, , P
, c N, a
def
= N A W,
and p P
= w
a
(for < ).
Let us x a list
1 = 1
from N.
For (
+ 1) N, let
1
[]
= 1
[]
: < ), where 1
[]
= p : p 1
.
(Note that
1
[]
lists all open dense subsets of P
N let
= p P
: p p() ,=
(so
from N)
and let E
3
= E
2
: ( w
)( < )(
= 1
). Clearly, E
3
is a club of .
Now, using the diamond
F
+ 1) 0 : (h F
() : < ) is a
P
increasing
1
[]
exact sequence of members of N P
_
and if Z() ,= then we put () = sup(Z()). Note that Z() N and thus
() N (when dened). Now, the prediamond
F = F
: S) is picked so
that for a limit S:
()
1
if Z() ,= , then h F
() =
_
h F
()
_
() for all < ;
()
2
if Z() = , then h F
() =
P
() : < ) is a
P
N.
Just for notational simplicity, we will identify a sequence =
0
) with its (only)
term
0
. Thus below, when we talk about a standard (w, 1)
tree T , we think of
T as a set of sequences t = (t)
s do not have to be
sequences.
Now we are going to dene sequences p = p
i
: i < ) P
N, T
:
S is limit ), and q
,t
: S is limit & t T
) P
= (T
, rk
) is a standard (w
, 1)
: rk
(t) = Y
[]
a
() for w
,
(ii) q
,t
: t T
> j)(p
j
() = p
j
()),
(v) if t T
, rk
(t) = , then q
,t
P
N is such that
(a)
_
<
Dom(h F
())
i<
Dom(p
i
)
_
Dom(q
,t
),
(b) ( < )
_
(h F
)() q
,t
_
, and
(c) if Dom(q
,t
) w
, then
q
,t
if the set p
i
() : i <
_
h F
()
_
() : <
has an upper bound in Q
,
then q
,t
() is such an upper bound ,
SHEVASHEVASHEVA: LARGE CREATURES 21
(d) if Dom(q
,t
) w
, then
q
,t
if the set p
i
() : i <
_
h F
()
_
() : <
has an upper bound which obeys (t)
at ,
then q
,t
() is such an upper bound,
else q
,t
() is an upper bound of
_
h F
()
_
() : <
which obeys (t)
at ,
(e) q
,t
<
1
[]
,
(vi) if t T
, = rk
(t) <
is the successor of in w
and t
, t
(t) = rk
(t
) =
, t t
, t t
and t
,= t
,
then
q
,t
P
the conditions q
,t
() and q
,t
() are incompatible ,
(vii) if t T
and Dom(q
,t
) w
, then Dom(p
) and
p
if q
,t
P
and p
i
() : i < q
,t
() has an upper bound in Q
,
then p
, rk
(t) = <
, x Y
a
() and
q
,t
,
P
such that t t
and (t
= x.
Assume < and we have dened p
i
, T
i
, q
i,t
for relevant i < and t. If is not a
limit ordinal from S, then only p
be a standard (w
, 1)
: rk
(t) = =
Y
[]
a
() (for w
). For t T
we dene a condition r
t
P
rk
(t)
so that
Dom(r
t
) =
_
_
<
Dom(h F
())
_
i<
Dom(p
i
) w
_
rk
(t),
and for each Dom(r
t
), r
t
() is the <
rst P
such
that:
if w
, then
r
t
P
if the family p
i
() : i <
_
h F
()
_
() : <
has an upper bound which obeys (t)
at ,
then r
t
() is such an upper bound,
if the previous is impossible, but there is an upper bound of
_
h F
()
_
() : < which obeys (t)
at
then r
t
() is such an upper bound,
if neither from the previous two possibilities holds,
then r
t
() is an upper bound of
_
h F
()
_
() : < ,
and if / w
, then
r
t
P
if the family p
i
() : i <
_
h F
()
_
() : <
has an upper bound, then r
t
() is such an upper bound,
if this is not possible, then r
t
() is just an upper bound of
_
h F
()
_
() : < .
22 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
Plainly, [T
[ < and r = r
t
: t T
= r
t
: t T
) N such that r r
and
w
rk
t
decides the truth value of the sentence
r
t
() obeys (t)
at (with respect to R
)
(remember the choice of r
t
() for w
=
_
t T
: for each w
rk
(t), r
t
P
t
() obeys (t)
at
_
,
and notice that T
N is a standard (w
, 1)
there is t
such that t t
and rk
(t
) =
) N which
satises clauses (vi) and (v)(e) and such that r
t
q
,t
(for t T
). It should be
clear that then T
and q
,t
: t T
, t
, Dom(q
,t
) Dom(q
,t
), /
w
+ 1) we let
A
[]
= t : t T
& rk
(t) =
;
if s A
[]
, then q
[]
,s
= q
,t
for some (equivalently: all) t T
such that
s t and rk
(t) =
;
q
[]
= q
[]
,s
: S is limit & s A
[]
);
h
[]
: N is such that h
[]
() = (h()) provided h() is a function,
and h
[]
() =
P
otherwise.
Plainly, , = A
[]
Y
[]
a
() (remember (i)) and h
[]
: N is such that
P
N Rng(h
[]
). Moreover, one easily veries the following claim.
Claim A.3.10.2. Let N(
+1). Then
F is a quasi Ddiamond sequence for
(N, h
[]
, P
) and q
[]
is a weak fuzzy candidate over
F for (N, h
[]
, P
, R
[]
,
Y
[]
,
1
[]
).
We may write q, q
,t
, A
for q
[
]
, q
[
]
,t
, A
[
]
behave the same, so we may
identify them. Of course, we are going to dene an (R,
Y)fuzzy generic condition
r P
for q over
F, but before that we have to introduce more notation used later
and prove some important facts.
For
N we dene a function h
and P
names S
, A
, 1
,
F
,
1
and q
so that:
h
name then h
() = ((h())(), otherwise h
() =
;
P
)(q
[]
,s
P
);
P
if S
is limit, then A
= x Y
a
() : (t A
)(q
,t
& (t)
= x) ;
P
= q
,x
: S
is limit & x A
), where:
P
if S
is limit and x A
, then q
,x
= q
,t
() for some (equiva-
lently: all) t A
such that q
,t
P
and (t)
= x ;
SHEVASHEVASHEVA: LARGE CREATURES 23
P
= F
: S
);
P
= 1
: < ) , where:
P
= p() : p 1
& p
P
.
Naturally, we treat h
as a P
]. Observe
that
P
N[
P
] Q
Rng(h
) , and
from N[
P
] .
Claim A.3.10.3. Assume that N
and r P
is a (R
[]
,
Y
[]
)fuzzy generic
condition for q
[]
over (N,
1
[]
, h
[]
, P
,
F). Then
(1) r
P
D
+
,
(2) r
P
], h
, Q
) , and
(3) r
P
], h
, Q
, R
,
Y
) over
F
.
Proof. (1) Will follow from (2).
(2) Assume that this fails. Then we can nd a condition r
, a P
name
q
= q
N[
P
], and
P
names A
, B
for members of D V such that r
P
and
r
(
<
A
)(q
: < ) is
1
exact ) and
( S
<
B
)(h
() : < ) , = q
) .
Consider a play r
j
, C
j
: j < ) of the game
fuzzy
(r, N,
1
[]
, h
[]
, P
,
F, q
[]
) in
which Generic uses her winning strategy and Antigeneric plays as follows.
Together with choosing r
j
(for j S), the Antigeneric player chooses also
side conditions p
j
N P
, sets A
, B
D and P
names q
N for elements of
Q
(so r
0
r
( j)(A
= A
& B
= B
& q
= q
) , and
if j
, and
p
j
P
(
0
<
1
j)(q
0
q
1
) , and
if < j,
<
A
, then p
j
j
) 1
[+1]
j
s are (forced to be) increasing.] After the play,
the Antigeneric player completes p
j
: j S) to a
P
increasing sequence
p
j
: j < ) N P
letting p
j
= p
min(\S\(j+1))
for j S.
Note that if E
0
is a limit of elements of
<
A
j
) :
j < ) is
1
[+1]
exact and increasing (and p
j
: j < ) is
1
[]
exact). So, as D is
normal and A
, B
, C
j
D and
F is a quasi Ddiamond for (N, h
[+1]
, P
+1
) (by
A.3.10.2), we may nd an ordinal S E
0
<
A
<
B
j<
C
j
(() +1)
which is a limit of members of
<
A
(j) : j < ) =
24 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
p
j
j
) : j < ). By the choice of
F we know that h(F
(j)) is a condition in P
(F
(j)) = q
j
for all j < . Also
(i < )(j S)(h
[]
F
(i)
P
h
[]
F
(j) = p
j
P
r
j
).
Since the play is won by Generic, for some s A
[]
we have q
[]
,s
r
. But then
r
<
B
& h
() : < ) = q
,
a contradiction.
(3) Straightforward (remember the choice of q
,t
s, specically clauses (v)(b,d,e)
and (viii)).
Now we are going to dene an (R,
Y)fuzzy generic condition r P for q over
(N,
1, h, P
,
F) in the most natural way. Its domain is Dom(r) =
N and for
each
N
r r() p
()+1
() is an (R
,
Y
over (N[
P
],
1
, h
, Q
,
F
) .
[So r p
i
for all i < .]
Claim A.3.10.4. For every (
(r , N,
1
[]
, h
[]
, P
,
F, q
[]
).
Proof of the Claim. We will prove the claim by induction on (
+1)N. After
we are done with stage (
+ 1) N, we know that r is (R
[]
,
Y
[]
)fuzzy
generic for q
[]
over (N,
1
[]
, h
[]
, P
,
F). For
is (R
[
]
,
Y
[
]
)fuzzy
generic for q
[
]
over (N,
1
[
]
, h
[
]
, P
,
F) for all
(r, N,
1
[]
, h
[]
, P
,
F, q
[]
).
First, for N x a P
name st
such
r st
is a winning strategy of the Generic player
in the game
fuzzy
(r(), N[
P
],
1
, h
, Q
,
F
, q
) .
We will think of st
as a name for a function from <sequences of members of
Q
D such
that if (
,
C
) Dom(st
) and
,
C
is
won by Generic). (In a play of
fuzzy
(r(), N[
P
],
1
, h
, Q
,
F
, q
) only the
values of st
at legal partial plays according to st
matter, but it is notationally
convenient to have st
giving values for all sequences of elements of Q
D, even if
rst coordinates are not increasing, as well as for sequences after which Antigeneric
should play.)
We will describe the strategy st by giving the answers of Generic on intervals S
[i
, i
+1
) (for < ), where, remember, i
j
() : j
< , N)
and C
j
() : j
, < , N) so that
SHEVASHEVASHEVA: LARGE CREATURES 25
()
1
r
j
() is a P
, C
j
() is a P
, i
+1
)
_
, and w
+1
, then after the -th move
(which is a move of the Generic player) the terms r
j
(), C
j
() : < )
are dened for all j
< i
+1
, and
()
3
if < , w
+1
and p
+1)S
(r
j
are the conditions played in the game), then
p
P
(j (i
+ 1) S)(r
j
() r
i
()) ,
()
4
if < , w
+1
and r
i+1
is the condition played by Generic at stage
i
+ 1 S, then
r
i+1
P
(j
< i
+1
)(r
j
() r
i+1
()) ,
()
5
for each N ,
r
P
r
j
(),
<
C
j
() : j < ) is a play of the game
fuzzy
(r(), N[
P
],
1
, h
, Q
,
F
, q
.
So suppose that < , = min
_
S [i
, i
+1
)
_
and r
j
, C
j
: j < ) is the result
of the play so far. Now Generic looks at ordinals w
+1
. She lets the P
names r
j
(), C
j
() be such that r
j
(),
<
C
j
() : j
< i
+1
) is forced by r
to be a play of
fuzzy
(r(), N[
P
],
1
, h
, Q
,
F
, q
and the
names r
i
(), C
i
() are such that
if i
= (i.e., i
S and r
i
, C
i
have been chosen for i < i
only), then
r
P
(r
i
(),
<
C
i
()) is the value of st
applied to r
j
(),
<
C
j
() : j < i
),
if i
< (i.e., i
/ S so r
i
, C
i
are already chosen for i i
), then
r
P
if (j < i
)(r
j
() r
i
()) then r
i
() = r
i
()
otherwise r
i
() is the rst coordinate of st
applied to
the play so far, and C
i
() =
ji
C
j
for all < .
Then for j (i
, i
+1
) (and w
) the names r
j
(), C
j
() are determined by
applying successively st
, that is
r
P
(r
j
(),
<
C
j
()) is the value of st
applied to r
j
(),
<
C
j
() : j
< j).
If (w
+1
w
j
(), C
j
()
somewhat like above, but starting with subscript j = 0. Thus
26 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
if i
= , then
r
P
r
0
() is the rst coordinate of the value of st
at r
j
(),
i<i
C
i
: j
< i
)
and C
0
() =
i<i
C
i
for all < ,
if i
< , then
r
P
if (j < i
)(r
j
() r
i
()) then r
0
() = r
i
()
otherwise r
0
() is the rst coordinate of the value of st
at r
j
(),
i<i
C
i
: j < i
)
and C
0
() =
ji
C
j
for all < .
Lastly, for 0 < j < i
+1
(and (w
+1
w
) ) the names r
j
(), C
j
() are
determined by applying successively st
(like earlier).
This denes the names r
j
(), C
j
() for j < i
+1
, < and w
+1
. It
is straightforward to check that the requirements ()
1
()
3
and ()
5
restricted to
w
+1
(and with j < replaced by j < i
+1
) are satised.
Next, using the fact that P
is closed and ()
3
of the choice above, Generic
picks a condition r
such that
()
6
r
r
j
for every j < ,
()
7
r
j
() r
() for every j
< i
+1
and w
+1
,
()
8
r
<i+1
1
, and
()
9
for every j
, < i
+1
and w
+1
, the condition r
j
(), say r
j
() = C
j
() , where C
j
() D V.
If i
S (so = i
stronger than r
and (w
) we have:
()
10
either the conditions r
+
and q
[]
,t
are incompatible, or q
[]
,t
P
r
+
,
()
11
if w
and q
[]
,t
P
r
+
, and s A
[]
(i.e., i
, i
+1
)S
she plays
r
j
= r
+
and C
j
= E
3
j
() : j
, < i
+1
, w
+1
.
Plainly, r
i+1
= r
+
satises clause ()
4
.
Why does the strategy described above work?
Suppose that r
j
, C
j
: j < ) is a play of the game (r , N,
1
[]
h
[]
, P
,
F, q
[]
) in
which the Generic player used this strategy and let r
j
() : j
< , N) and
C
j
() : j
() : < ) is
1
[+1]
exact. Now, suppose that a limit
ordinal S
j<
C
j
(so in particular E
3
) is such that
SHEVASHEVASHEVA: LARGE CREATURES 27
()
() r
j
).
(So then ( < )(h
[]
F
() r
() : < ) is
1
[]
exact, then w
().)
We are going to choose t A
[]
. We do this by induction
on ( +1) N, dening t T
(and
for = we get the desired conclusion). Limit stages and the initial stage = 0 are
trivial, so assume that we have dened t and have shown q
,t
= q
,t
r
then t( +1) = t (so it has been already dened). Suppose also that
Dom(q
,t
) (otherwise there is nothing to do). Look at the clause (v)(c) of the
choice of q
,t
at the beginning: r
r p
(and ()
) implies that
r
q
,t
() is an upper bound to p
i
() : i < .
But then also by the clause (vii) there, r
q
,t
() p
() r
() , so we
are done.
Suppose now that w
j
(), C
j
() for j <
and < (remember ()
2
). Moreover, the condition r
() : < ) is
1
[+1]
exact):
the sequence h
() : < ) is
Q
increasing
1
exact, and
r
j
(),
<
C
j
() : j < ) is a play according to st
(by ()
5
), and
S
j,<
C
j
() (remember ()
9
and the choice of C
i+1
for < ),
and hence also S
j<
<
C
j
(),
(j < )(j
< )(r
j
() r
j
()) and (j < )(j
< )(r
j
() r
j
()) (by
()
3
+ ()
4
), so also
( < )(j < )(h
()
Q
j
()).
Since st
is a name for a winning strategy, we may conclude that (by ()
7
)
r
P
(x A
)(q
,x
r
() r
()) .
Now look at ()
11
remembering clause (vi) of the choice of q: by them there is
a unique x Y
a
() such that letting (t)
= x we get t( + 1) T
satisfying
q
,t(+1)
( + 1) r
( + 1).
This completes the inductive proof of A.3.4(5)().
Why does A.3.4(5)() hold? To show this condition, it is enough to prove that
()
_
mod cf(). Now, by induction on
< we choose conditions s
N P
such that
(a)
(j < )(s
r
j
),
(b)
N,
28 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
(c)
) s
),
(d)
<
1
[]
.
So suppose that we have dened s
: either s
s,
or s
, s are incompatible .
Clearly 1
,
N is an open dense subset of P
is
(N, P
from N (by ()
8
), we may nd s
<
1
,
j<
1
[]
j
such that
s
r
j
and s
(d)
are satised.
Now, let conditions s
) =
Dom(s
)
and s
() (for Dom(s
)) is the <
rst P
satisfying
s
P
( < )(s
() s
()) and
if there is a [, ) such that ( < )(s
() s
())
then s
() = s
: < ) is
P
increasing and
( < )( <
)(s
P
s
() = s
() ).
So it follows from (d)
such that s
:
< ) is
1
[]
exact for all C
r
j
), we
may pick a club C
of such that
( C
r
j
).
Now, as
F
u
C
we have s
: < ) = h F
for those we
had clause ()
1
with () and hence s
: < ) = h
[]
F
() : < ).
Therefore ()
).
is a limit ordinal of conality .
Let
0
= 0. By induction on < choose conditions s
such that
(a)
(j < )(s
r
j
),
(b)
N,
(c)
,
(d)
<
1
[]
.
[Possible as r
is (N, P
)generic and by ()
8
.] For each < , for some club
C
of we have
( C
)(s
: < ) is
1
[]
exact).
Also for a club C
of we have ( C
)(w
). Like before, as
F
we have
s
: < ) = h F
and also
s
: < ) = h
[]
F
).
SHEVASHEVASHEVA: LARGE CREATURES 29
is a successor ordinal.
Like before (remember that, letting =
is (N, P
)generic
and it forces that r(
) is (N[G
], Q
)generic).
: S
a
). Then in all
denitions and results we may replace D, S by D
a
, S
a
, where a = N A. In
particular, this way we get the notions of fuzzy properness over quasi
Ddiamonds
which behave nicely in iterations.
30 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
Part B
Building suitably proper forcing notions
B.4. A creaturefree example
In this section we show that a natural forcing notion uniformizing colourings on
ladder systems is fuzzy proper. (This forcing is a relative of Q
) and W [A]
< , S
def
= <
+
: cf() = and for S
:
() B
) = ,
() h
: B
.
Let
B = B
: S
),
h = h
: S
).
The forcing notion Q
= Q
(S
,
B,
h) is dened as follows:
a condition in Q
is a tuple p = (u
p
, v
p
, e
p
, h
p
) such that
(a) u
p
[
+
]
<
, v
p
[S
]
<
,
(b) e
p
= e
p
: v
p
), where each e
p
, and
e
p
u
p
, and
(c) sup(e
p
) = sup(u
p
) (for v
p
), and if
1
<
2
are from v
p
, then
sup(e
p
2
) >
1
and sup(e
p
1
) > sup(B
2
1
),
(d) h
p
: u
p
is such that
( v
p
)( e
p
)(h
p
() = h
());
the order of Q
is an end-extension of e
p
.
A tuple p = (u
p
, v
p
, e
p
, h
p
) satisfying clauses (a), (b) and (d) above will be called
a pre-condition. Note that every pre-condition can be extended to a condition in
Q
.
Proposition B.4.2. (1) The forcing notion Q
+
chain condition and [Q
[ =
+
.
(2) If p Q
, <
+
, S
v
q
)(otp(e
q
) > ).
Proof. (1) Verication of the chain condition is a straightforward application of
the lemma. To check that Q
, j < . Let r = (u
r
, v
r
, e
r
, h
r
) be such
that
v
r
=
i<j
v
pi
, and for v
r
e
r
e
pi
: v
pi
& i < j sup
_
e
pi
: v
pi
& i < j
_
u
r
=
i<j
u
pi
e
r
: v
r
h
r
i<j
h
pi
,
SHEVASHEVASHEVA: LARGE CREATURES 31
and if e
r
e
pi
: v
pi
, i < j, then h
r
() = h
.
(2) Should be clear.
Proposition B.4.3. Q
,
Y
) for Q
over W. We let R
,
and x is a function with Dom(x) u
p
and ( Dom(x))(h
p
() = x()).
Now suppose that a W and let
a
be the <
0
= (
a
)
1
[] <
+
: = sup
_
(
a
)
1
[]
_
,
and then
Y
a
() = x : x is a function from x
0
a to
.
This denes Y
a
and
Y
= Y
a
: a W). It is easy to check that (R
,
Y
) is a
base for Q
), [N[ = ,
<
N N, , Q
,
B,
h, S
, R
N, and
a
def
= N A W, p Q
N,
1 = 1
from N,
h : N satises Q
N Rng(h), and
F = F
) and q is a fuzzy
candidate over
F.
For limit S let }() = }(N, Q
, h,
F, R
,
Y
,
Y
satises the
+
cc, the condition p is (N, Q
(p, N,
1, h, Q
,
F, q) which guarantees that the result r
i
, C
i
: i < ) of
the play satises A.3.4(5)().
Suppose that we arrive to a stage S and r
i
, C
i
: i < ) is the sequence
played so far. First, Generic picks the <
rst condition r
__
q
,x
r & (i < )(r
i
r)
_
,
then q
,x
r
rst condition r
above r
such that
()
1
if v
r
, then otp(e
r
) > , and
()
2
(
a
)
1
[] u
r
and (
a
)
1
[] S
v
r
.
The set C
: (
a
)
1
[] u
q
& (
a
)
1
[] S
v
q
& ( v
q
)(otp(e
q
) > )
(which is an open dense subset of Q
:
< ,
()
4
otp(B
(sup(e
r
,
32 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
()
5
if v
r
and a(sup(e
r
)+1)
with
a
() < .
Why does this strategy work (i.e., why does it ensure A.3.4(5)())?
Let r
i
, C
i
: i < ) be a play according to this strategy, and suppose that
S
i<
C
i
is a limit ordinal such that h F
() : < ) is a
Q
increasing
1
exact sequence of conditions fromQ
() r
i
).
Note that then
()
6
if
i<
v
ri
, then otp(
i<
e
ri
) = and
i<
e
ri
is an unbounded subset of
B
: otp( B
) < , and
()
7
i<
u
ri
N = (
a
)
1
[] =
<
u
hF
()
and
i<
v
ri
N = (
a
)
1
[] S
<
v
hF
()
,
()
8
if (
a
)
1
[] S
, then
_
e
hF
()
: < & v
hF
()
=
_
e
ri
: i < & v
ri
.
We want to show that
() for some x }(), there is a common upper bound to r
i
: i < q
,x
(which, by the denition of our strategy, will nish the proof). For S
let
= (u
r
, v
r
, e
r
, h
r
)
be such that
v
r
i<
v
ri
, u
r
i<
u
ri
: v
r
,
e
r
e
ri
: i < & v
ri
(for v
r
), and
h
r
: u
r
is such that
i<
h
ri
h
r
and h
r
) = h
).
One easily veries that the above conditions indeed dene a pre-condition (remem-
ber ()
6
). Also, note that if v
r
, then
e
r
e
ri
: i < & v
ri
and each e
ri
is a proper subset of
e
ri
: i < & v
ri
. Moreover, if v
r
and
N, then
= sup(u
r
) = sup((
a
)
1
[]
) (by ()
5
+ ()
2
;
remember also ()
7
). Now, extend r
to a pre-condition r
such that u
r
= u
r
0
,
v
r
= v
r
and e
r
= e
r
for v
r
0
(note that
x
0
a). Since r
= u
q
,x
u
r
, v
= v
q
,x
v
r
, h
= h
q
,x
h
r
,
if v
q
,x
, then e
= e
q
,x
, and if v
r
N, then e
= e
r
.
Note that h
is a function from u
to
by ()
7
(remember the choice of x and
that q
,x
N is stronger than all h F
()s). Also, if v
r
N then ( v
q
,x
and) e
q
,x
is an end-extension of e
r
(remember ()
7
+ ()
8
). Hence (u
, v
, e
, h
)
is a pre-condition stronger than both q
,x
and r
. Extending it to a condition in
Q
=
+
,
2
+
=
++
and D is a normal lter on such that there is a Ddiamond. Then
there is a forcing notion P such that:
SHEVASHEVASHEVA: LARGE CREATURES 33
P is complete weakly fuzzy proper over quasi Ddiamonds for W and it
satises the
++
cc,
in V
P
, 2
= 2
+
=
++
and for every
, S
,
B,
h as in B.4.1(2) there is
h :
++
the set
B
: h
() = h()
contains a club.
Proof. The forcing notion P is the limit P
++ of a support iteration
Q = P
, Q
:
<
++
), where each Q
is forced to be Q
(S
,
B
) for some
B
. Then,
by A.1.9, A.3.10 and B.4.3 we are sure that P satises the
++
cc, it is weakly
fuzzy proper over quasi Ddiamonds for W and it has a dense subset of size
++
.
Consequently we may arrange suitable bookkeeping to take care of all P
+
+
names
for objects as in B.4.1(2) the details and the rest should be clear.
B.5. Trees and creatures
Let us introduce the notation used in the forcing notions we want to build. The
terminology here is somewhat parallel to that of [14, 1.2, 1.3], but there are some
dierences as the context is dierent. We start with the tree case.
Denition B.5.1. Let H : H(
+
).
(1) A tree creature for H is a tuple
t = (, dis, pos, nor) = ([t], dis[t], pos[t], nor[t])
such that dis H(
+
), nor + 1,
_
<
<
H(), and ,= pos
_
<
<
H() : .
TCR
<
H() we let TCR
[H] = t TCR
[H] : [t] = .
(2) Let K TCR
:
T) such
that
T is a complete quasi tree of height ht(T) < ,
for each
T, t
K satises succ
T
() = pos[t
],
and
if t (t
:
T), then [t] = root(T) and pos[t] max(T),
if t (t
:
T) and t
(s
:
T
) (for
T), then
t (s
), and
for each t K we have t) Dom() and t (t).
Then (K, ) is called a tree creating pair (for H).
(3) A tree creating pair (K, ) is local if
(t
] nor[t].
34 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
We say that (K, ) is very local if, additionally, for every
<
<
H()
such that K TCR
[H] ,= there is t
K TCR
[H] satisfying (t
K TCR
[H])(t (t
for
(K, ) and the minimal condition p
= T
(K, ) =
<
<
H() : ( < lh())(( + 1) pos[t
])
p
= p
(K, ) = t
: T
).
[Note that, in the general case, T
: T) such that
(a) ,= T
<
<
H() is a complete quasi tree with max(T) = ,
(b) t
TCR
] = succ
T
(),
(c)
1
for every lim
(T), lim(nor[t
] : < , T) = ;
the order be given by:
t
1
: T
1
) t
2
: T
2
) if and only if
T
2
T
1
and for each T
2
there is a complete quasi tree T
0,
(T
1
)
[]
of height ht(T
0,
) < such that t
2
(t
1
:
T
0,
).
If p = t
= t
etc.
(2) Let D
we have
( Y )( (T)
)(nor[t
] [[).
[The set Y
p
above may be called a witness for p Q
tree
D
(K, ).]
(3) The forcing notion Q
tree
cl
(K, ) is dened by replacing the condition (c)
1
by
(c)
cl
() ( T)( T)( & nor[t
] [lh()[), and
() ( T)(nor[t
] = 0 or nor[t
] [lh()[), and
() nor[t
root(p)
] [lh(root(p))[, and
() if < is a limit ordinal and
i
: i < ) T is a increasing
sequence such that nor[t
i
] [lh(
i
)[ for each i < and =
i<
i
, then ( T and) nor[t
] [lh()[.
(4) If p Q
tree
e
(K, ) and T
p
, then we let p
[]
= t
p
: (T
p
)
[]
).
(5) For the sake of notational convenience we dene partial order Q
tree
(K, )
in the same manner as Q
tree
e
(K, ) above but we omit the requirement (c)
e
.
Denition B.5.3. Let (K, ) be a tree creating pair for H, t K. We dene a
relation _
t
on (t) by
t
_
t
if and only if (t
, t
(t) and) t
(t
).
If (K, ) is very local, t
is also denoted by
_
.
Remark B.5.4. (1) Note that the relation _
t
(K, ), then T
p
is a complete tree.
Now we are going to describe the non-tree case of forcing with creatures. For
sake of simplicity we restrict ourselves to what corresponds to forgetful creatures
of [14, 1.2.5].
Denition B.5.5. Let H : H(
+
).
(1) A forgetful creature for H is a tuple
t = (
dn
,
up
, dis, val, nor) = (
dn
[t],
up
[t], dis[t], val[t], nor[t])
such that dis H(
+
), nor + 1,
dn
<
up
< and , = val
dn
<up
H().
CR
up
[t
i
] =
dn
[t
i+1
] for i < i + 1 < j, and
sup
up
[t
i
] : i
< i =
dn
[t
i
] for limit i < j,
and if t (t
i
: i < j), then
=
dn
[t] =
dn
[t
0
],
+
=
up
[t] = sup
up
[t
i
] : i < j, and
val[t]
<
+
H() : (i < j)([
dn
[t
i
],
up
[t
i
]) val[t
i
]),
and
if t
i
(s
i
: <
i
) (for i < j) and t (t
i
: i < j),
then t (s
i
: <
i
, i < j), and
for each t K we have t) Dom() and t (t).
Then (K, ) is called a creating pair (for H).
(3) We say that (K, ) is local if for each t K
up
[t] =
dn
[t] + 1 , and
t
] nor[t].
It is very local if, additionally, for each < there is t
K such that
dn
[t
). The
creature t
at .
(4) For j < , a japproximation for (K, ) is a pair (w, t
i
: i < j)) such that
t
i
K,
up
[t
i
] =
dn
[t
i+1
] for i < i + 1 < j, and
sup
up
[t
i
] : i
< i =
dn
[t
i
] for limit i < j,
and w
<
dn
[t0]
H().
(5) For a japproximation (w, t
i
: i < j)) for (K, ) we let
pos(w, t
i
: i < j)) = v
<
+
H() : w v and for all i < j
v[
dn
[t
i
],
up
[t
i
]) val[t
i
] ,
where
+
= sup
up
[t
i
] : i < j.
Denition B.5.6. Let (K, ) be a creating pair for H.
36 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
(1) We dene the forcing notion Q
1
(K, ):
conditions are pairs p = (w,
t) such that
(a)
t = t
i
: i < ) is a sequence of creatures from K satisfying
up
[t
i
] =
dn
[t
i+1
] for i < i + 1 < , and
sup
up
[t
i
] : i
< i =
dn
[t
i
] for limit i < ,
(b) w
<
dn
[t0]
H()
(c)
1
lim(nor[t
i
] : i < ) =
the order is given by:
(w
1
, t
1
i
: i < )) (w
2
, t
2
i
: i < ) if and only if
for some continuous strictly increasing sequence i
: < ) we have
w
2
pos(w
1
, t
1
i
: i < i
0
)) and ( < )
_
t
2
(t
1
i
: i
i < i
)
_
.
If p = (w, t
i
: i < )), then we write w
p
= w, t
p
i
= t
i
(for i < ).
(2) Let D
D
(K, ) is dened similarly,
replacing the condition (c)
1
by
(c)
D
for some set Y = Y
p
D
we have
(i Y )(nor[t
i
] [
dn
[t
i
][).
[The set Y
p
above may be called a witness for p Q
D
(K, ).]
(3) For the sake of notational convenience we dene partial order Q
(K, ) in
the same manner as Q
e
(K, ) above but we omit the requirement (c)
e
. If
(K, ) is very local, then the minimal condition p
for (K, ) is
p
= p
(K, ) = (), t
: < )) Q
(K, ),
where t
and _
t
=_
be as in 0.3).
(1) We say that a tree creature t K is complete (for (K, )) if
() for every _
t
increasing chain t
] > 0, there is t
_
t
) and
nor[t
] minnor[t
] : < ,
() if t
(t), nor[t
] = 0, then [pos[t
][ = 1 and (t
) = t
,
() if pos[t], then there is t
] = and
nor[t
] = 0.
(2) t K is said to be exactly complete if it is complete and
() if
t = t
<
pos[t
] ,= .
(3) We say that (K, ) is
complete (exactly
complete, respectively) if
SHEVASHEVASHEVA: LARGE CREATURES 37
(a) (K, ) is very local, and
(b)
is
+
lh()
complete (exactly
lh()
complete,
resp.).
We say that (K, ) is just complete if it satises (a) above and
(b)
is complete.
Proposition B.6.2. Assume that (K, ) is a very local tree creating pair for H,
D
0
(P, , r). At stage j <
of the game, after a sequence (p
i
, q
i
) : i < j)
p
j
) P has been constructed (so p
j
is the move of Incomplete), she plays the <
rst condition q
j
P stronger than
p
j
and such that lh(root(q
j
)) > j +.
Why is this a winning strategy? Suppose that the players have arrived at a
limit stage < of the game, Complete has used st and (p
i
, q
i
) : i < ) is
the result of the game so far. Our aim is to show that the (increasing) sequence
q
i
: i < ) has an upper bound in P. To this end we are going to dene a condition
q = t
and t
. First we
declare root(T) =
i<
root(q
i
) and we note that
root(T)
i<
T
qi
and lh(root(T)) <
+
lh(root(T))
.
Now we may choose t
root(T)
TCR
root(T)
[H] so that t
root(T)
(t
qi
root(T)
) (for all
i < ) and nor[t
root(T)
] minnor[t
qi
root(T)
] : i < , and we declare pos[t
root(T)
]
T (thus dening (T)
lh(root(T))+1
). Next we proceed inductively in a similar manner:
suppose that (T)
i<
T
qi
. For each
(T)
choose t
such that
(i < )(t
(t
qi
)) and nor[t
] minnor[t
qi
] : i < ,
and declare pos[t
(for (T)
like above).
This way we build a condition t
: T) Q
tree
)
,
ex(
)
) of (K, ) as follows.
38 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
Let
<
<
H(). We let K
ex(
)
TCR
: ), where t
0
= t
& t
,= t
,
pos[t] = pos[t
],
nor[t] = minnor[t
] : .
Then, for t
, t K
ex(
)
TCR
[H] we let t
ex(
)
(t) if and only if dis[t] dis[t
].
Proposition B.6.4. Assume(K, ) is a very local
complete tree creating pair.
Then (K
ex(
)
,
ex(
)
) is a very local exactly
complete tree creating pair. The
minimal creature for it at is t
] = t
).
Proof. Easy.
Theorem B.6.5. Suppose that
(a) ( < )([H()[ <
), and
(b) (K, ) is a complete very local tree creating pair for H for some strictly
increasing sequence =
<
(for < ), and
(c) D
is a normal lter on .
Then the forcing notion Q
tree
D
(K, ) has the strong
Sacks property
Proof. Let i
0
< and p Q
tree
D
(K, ). Just for notational simplicity we assume
that H()
(i
0
, p, Q
tree
D
(K, )). In the course
of the play she will also choose sets Y
i+1
D
.
First Generic picks T
p
such that lh() > i
0
and she starts the game with
playing s
i0
= (i
0
+ 1) and q
i0
(i0+1)
= p
[]
. She also picks t
i0
(t
p
(i0+1)
)
such that pos[t
(i0+1)
] = (i
0
+ 1) (remember B.6.1(1)()).
Arriving at a successor stage j = i + 1 of the play the players have determined
s
i
, q
i
, p
i
and Y
i
so that, in addition to the demands of the game, for each s
i
i+1
we have root(q
i
T
p
i
strictly
extending and she plays
s
i+1
= s
i
(i + 2) : s
i
, q
i+1
(i+2)
= (p
i
)
[]
for s
i
i+1
.
She also xes a set Y
i+1
D
si
Y
q
i+1
(i+2)
(recall
B.5.2(2)) and for s
i
i+1
she lets t
(t
p
] =
.
Now suppose that the players have arrived to a limit stage < of the game,
and assume that /
i<
Y
i+1
. Generic lets s
T
r
strictly extending . Then
she plays
s
= s
( + 1) : s
, q
(+1)
= (r
)
[]
for s
.
The tree creatures t
(for s
(t
p
), pos[t
] =
.
SHEVASHEVASHEVA: LARGE CREATURES 39
Finally suppose that we are at a limit stage < of the game and
i<
Y
i
.
Let s
let r
) = and nor[t
r
] [[ (there is
one by arguments as in the proof of B.6.2(1) and the choice of the Y
i
s). Then she
plays
s
= s
_
pos[t
r
] : s
, q
= (r
)
[]
for pos[t
r
], s
.
She also lets t
= t
r
(t
p
) (for s
).
It should be clear that the strategy described above always tells Generic to play
legal moves (remember 0.3(c)). It should also be clear that if (s
i
, q
i
, p
i
) : i
0
i < ) is the result of a play of
Sacks
(i
0
, p, Q
tree
D
(K, )) in which Generic uses
that strategy, then letting T =
s
i
: i
0
i < and q = t
: T) we get
a condition in Q
tree
D
(K, ) (as witnessed by
i<
Y
i+1
) stronger than p and forcing
that
(
)(i [i
0
, ))((i + 1) s
i
& q
i
(i+1)
Q
tree
D
(K,)
_
.
satisfy
(b), (c) of B.6.5. Then the forcing notion Q
tree
D
(K, ) has the strong bounding
property.
Proof. Similar to B.6.5.
The above two theorems are applicable to forcing notions of the type Q
tree
cl
(K, )
as they may be treated as a special case (under the assumptions as there):
Proposition B.6.7. Assume that ( < )([H()[ < ) and (K, ) is a
complete very local tree creating pair for H (for some ). Then the forcing
notions Q
tree
cl
(K, ) and Q
tree
D
e
(K, )),
we do easy ways to ensure they are suitably complete (parallel to B.6.1, B.6.2).
However at present we do not know how to get the strong bounding properties of
section A.2 (which were obviously tailored for trees).
Denition B.6.8. (1) For a creating pair (K, ) and t K we dene when
t is complete and exactly complete like in B.6.1(1,2) (but with val
replacing pos).
(2) If (K, ) is very local, then we say that it is
complete (exactly
complete,
resp.) if each minimal creature t
is
+
complete (exactly
complete,
resp.).
Proposition B.6.9. Assume that (K, ) is a very local creating pair for H, D
1
(K, ) or Q
D
(K, ).
(1) If (K, ) is
complete, then P is strategically (<)complete.
(2) If (K, ) is exactly
complete, then P is complete.
40 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
B.7. Getting fuzzy properness
In this section we show that the forcing notions with trees and creatures may
t the fuzzy proper framework. Note that even though the forcing notions covered
by Theorems B.7.1 and B.7.2 below are also covered by Theorem B.6.6, the results
here still have value if we want to iterate that forcing notions with ones which do
not have the strong bounding property.
Theorem B.7.1. Let A, W, D be as in A.3.1 and let D
be a normal lter on
such that for some S
0
D
we have S
0
D. Let be an increasing sequence
of regular cardinals conal in . Assume that (K, ) is an exactly complete very
local tree creating pair for H, and [H()[ < for each < . Then the forcing
notion Q
tree
D
(K, ) is strongly fuzzy proper over quasi Ddiamonds for W.
Proof. By B.6.2 we know that Q
tree
D
(K, ) is complete.
Let R
tr
,
Y
tr
be the trivial base dened as in the proof A.3.9 (but for P =
Q
tree
D
(K, )). We are going to show that for this base and for c = (
, K, )
the condition A.3.6(2)(()
+
) holds. So assume that N, h,
F = F
: S) and
q = q
,x
: S limit & x A
: S limit).
Let
1 = 1
Q
tree
D
(K, ) N and sets Y
such that
(i) p
0
= p, root(p
) = root(p), and p
and Y
S
0
for < < ,
(ii) Y
witnesses p
Q
tree
D
(K, ) (see B.5.2(2)),
(iii) for every < < and (T
p
)
we have T
p
and t
p
= t
p
,
(iv) if < is a successor, < and (T
p
)
)
[]
1
and ( T
p
)( nor[t
p
] = 0),
(v) if
<
Y
for every ,
(vi) if S E S
0
and h F
h F
(i)),
and (T
p
)
[]
, then
(p
)
[]
q
= (p
)
[root(q
)]
and
( T
p
)( root(q
) nor[t
p
] = 0).
[Note that there is at most one as above; remember the choice of E.]
It should be clear that the inductive construction of the p
s and Y
s is possible
(for (v) remember <
S
0
). Now letting root(r) = root(p), T
r
=
<
(T
p
)
, t
r
= t
p
lh()+1
we get a condition r Q
tree
D
(K, ) (as witnessed by
<
Y
s.
Claim B.7.1.1. The condition r is (R
tr
,
Y
tr
)fuzzy generic for q.
SHEVASHEVASHEVA: LARGE CREATURES 41
Proof of the Claim. First note that the condition r is (N, Q
tree
D
(K, ))generic by
clause (iv) above. Therefore we may use A.3.8(3), and it is enough that we show that
the Generic player has a strategy in the game
fuzzy
(r, N,
1, h, Q
tree
D
(K, ),
F, q)
which guarantees that the result r
i
, C
i
: i < ) of the play satises A.3.4(5)().
Let us describe such a strategy.
First, for < let
)( either p
q or p
, q are incompatible ),
and let E
= E : ( < )(
(r, N,
1, h, Q
tree
D
(K, ),
F, q) the players
have arrived at stage i S having constructed a sequence r
j
, C
j
: j < i).
If either i is a successor ordinal or i /
j<i
C
j
, then the Generic player plays the
<
rst condition r
i
Q
tree
D
(K, ) such that (j < i)(r
j
r
i
) and lh(root(r
i
)) > i,
and the set C
i
= E
(S
0
lh(root(r
i
))).
If i
j<i
C
j
is a limit ordinal (so also i E
S
0
), then Generic asks
() is h F
i
() : < i) an increasing
1exact sequence such that
(j < i)( < i)(p
j
h F
i
()) ?
If the answer to () is no, then she plays like at the successor stage.
[Note that if the answer to () is no and h F
i
() : < i) is increasing
1exact,
then for some j < i and < i the conditions p
j
and h F
i
() are incompatible,
and hence r
i
and h F
i
() are incompatible.]
If the answer to () is yes, then Generic looks at clause (vi) (of the choice of
p
s) and =
j<i
root(r
j
) (note that lh() = i). If (p
i
)
[]
is incompatible with some
h F
i
(), < i, then she plays C
i
, r
i
as in the successor case.
[Note that then r
i
, h F
i
() are incompatible.]
Otherwise root(q
i
) T
pi
, q
i
= (p
i
)
[root(qi)]
and ( T
pi
)( root(q
i
)
[pos[t
pi
rst condition r
i
stronger than
all r
j
(for j < i) and q
i
.
It follows immediately from the comments stated during the description of the
strategy that every play according to it satises A.3.4(5)(), nishing the proof of
the claim and that of the theorem.
be a normal lter on
such that for some S
0
D
we have S
0
D. Assume that (K, ) is an exactly
complete very local creating pair for H, [H()[ < for each < . Then the
forcing notion Q
D
(K, ) is strongly fuzzy proper over quasi Ddiamonds for W.
Proof. Like B.7.1
Theorem B.7.3. Let A = H
<
(
),
be as in A.3.1. Let
be an increasing sequence of regular cardinals conal in . Suppose that (K, ) is
an exactly complete very local tree creating pair for H, ( < )([H()[ < ),
and D
<
<
H() a for every a W. Now we let R = R(K, )
consist of all triples (p, , ) such that < ,
<
H() and p Q
tree
D
(K, )
satises root(p). Next, for a W let Y
a
= Y
a
(K, ) : [a]
<
be given
by Y
a
() =
: S) and q = q
,x
: S limit & x A
) be
as in A.3.6(1)(), p Q
tree
D
(K, ) N. Let
1 = 1
Q
tree
D
(K, ) N and sets Y
such that
(i) p
0
= p, root(p
) = root(p), and p
and Y
witnesses p
Q
tree
D
(K, ),
(iii) for every < < and (T
p
)
we have T
p
and t
p
= t
p
,
(iv) if < is a successor ordinal and (T
p
)
)
[]
<
1
and ( T
p
)( nor[t
p
] = 0),
(v) if
<
Y
for every ,
(vi) if S E, h F
i<
root(h F
(i))
and lh() = , and ( < )(i < )(p
h F
(i)),
then ( T
p
]
i<
pos[t
hF
(i)
] we have
(p
)
[]
q
,
= (p
)
[root(q
,
)]
and ( T
p
)( root(q
,
) nor[t
p
] = 0).
[Note that, in the situation as in (vi), A
i<
pos[t
hF
(i)
].]
Plainly, the inductive construction of the p
s and Y
<
(T
p
)
, t
r
= t
p
lh()+1
we get a
condition r Q
tree
D
(K, ) stronger than all p
s.
Claim B.7.3.1. The condition r is (R,
Y)fuzzy generic for q.
Proof of the Claim. It is very much like B.7.1.1. We note that r is (N, Q
tree
D
(K, ))
generic (by clause (iv)), and therefore it is enough to show that the Generic player
has a strategy in the game
fuzzy
(r, N,
1, h, Q
tree
D
(K, ),
F, q) which guarantees that
the result r
i
, C
i
: i < ) of the play satises A.3.4(5)() (remember A.3.8(3)). Let
us describe such a strategy. First, for < let
)( either p
q or p
, q are incompatible ),
and let E
= E : ( < )(
(r, N,
1, h, Q
tree
D
(K, ),
F, q) the players
have arrived at stage i S having constructed a sequence r
j
, C
j
: j < i).
If either i is a successor ordinal or i /
j<i
C
j
, then Generic plays the <
rst
condition r
i
Q
tree
D
(K, ) such that (j < i)(r
j
r
i
) and lh(root(r
i
)) > i and
C
i
= E
lh(root(r
i
)).
If i
j<i
C
j
E
j<i
root(r
j
) and she notes that
lh() = (by the choice of C
j
s at successor stages) and =
i<
root(h F
(i)).
Also, by the exactness and the choice of E
, we have
(j < i)( < i)(p
j
h F
i
()).
So now Generic looks at clause (vi) of the choice of p
rst)
j<i
pos[t
rj
]
j<i
pos[t
hFi()
][ = 1).
Therefore, root(q
i,
) T
rj
and q
i,
(T
rj
)
[root(qi,)]
(for each j < i). So the
Generic player can play C
i
= E
rst condition r
i
stronger than all
r
j
(for j < i) and q
i,
.
Easily, the strategy described above has the required property, and the proof is
completed.
Problem B.7.4. Unlike that was in the case of B.7.1, it is not clear how the proof
of B.7.3 can be modied to get the parallel result for non-tree case. So, assuming
that A, W, H and D
D
(K, ) fuzzy proper for W?
B.8. More examples and applications
Here we are going to present some direct applications of the methods developed
in this paper. Though we do keep our basic assumptions from 0.3, we are going to
introduce more parameters, so let us fully state the context we are working in now.
Context B.8.1. (a) is a strongly inaccessible cardinal, 2
=
+
, and 2
+
=
++
, and
(b) =
: < ),
=
: < ) and =
<
<
<
<
and ( <
)([[
<
),
(d) A = H
<
(
),
are as in A.3.1,
(e) D is a normal lter on such that there is a Ddiamond.
44 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
Let us recall some notions related to cardinal characteristics of reals.
Denition B.8.2. (1) Let o
be the family of all sequences a = a
; < )
such that a
[]
<
)
_
,
c
cl
( ) = min
_
[}[ : } o
& (f
)( a })
_
< : f() a
(T
)
+
__
,
and also
e
cl
( ) = min
_
[([ : (
<
and
(f
<
)(g ()
_
< : f() ,= g() T
__
,
(2) For an ideal of subsets of a set A, the covering number cov() of is
cov() = min[}[ : } &
_
} = A.
Proposition B.8.3. It is consistent that c(
) < e
cl
( ).
Proof. Let H
0
() =
[H
0
] such that:
dis[t]
lh([t])
+ 1,
if dis[t] =
lh([t])
, then pos[t] = [t]
) : <
lh([t])
and nor[t] =
lh([t])
,
if dis[t] <
lh([t])
, then pos[t] = [t]
K
0
: [t
] = [t].
It should be clear that (K
0
,
0
) is a very local exactly
complete tree creating
pair. The forcing notion Q
tree
D
(K
0
,
0
) has the strong
Sacks property (by B.6.5).
Let W
be the canonical Q
tree
D
(K
0
,
0
)name for the generic function in
<
, so
p
Q
tree
D
(K0,0)
root(p) W
.
Then we have
Q
tree
D
(K0,0)
_
f
<
V
__
< : W
() = f() (T
)
+
_
.
Now let P be the limit of a support iteration,
++
in length, of the forcing notions
Q
tree
D
(K
0
,
0
). Then, by A.2.3 + A.1.9 + A.1.4 + A.3.10,
P is complete, proper and satises the
++
cc, and it has a dense
subset of size
++
, thus forcing with P does not collapse cardinals,
P has the
Sacks property, it is weakly fuzzy proper for W,
P
2
= 2
+
=
++
= e
cl
( ) and c(
) =
+
(K
0
,
0
) is a bounded relative of D
from [15,
4.10] (remember B.6.7). It is also a generalization of the forcing notions D
X
from
[12].
SHEVASHEVASHEVA: LARGE CREATURES 45
Proposition B.8.5. It is consistent that c(
) < c
cl
(
+
) = c(
+
), where
+
=
: < ).
Proof. Let H
1
() =
+
1
consist of all tree creatures
t TCR
[H
1
] such that:
dis[t]
+
lh([t])
, either [dis[t][ = 1 or dis[t] is a club of
+
lh([t])
,
pos[t] = [t]
) : dis[t],
if [dis[t][ = 1 then nor[t] = 0, if [dis[t][ > 1 then nor[t] =
lh([t])
.
Let
1
be a local tree-composition operation on K
1
such that
1
(t) = t
1
: [t
] dis[t].
Then (K
1
,
1
) is a very local complete tree creating pair. Let (K
1
,
1
) be the
exactivity of (K
1
,
1
) (see B.6.3); thus (K
1
,
1
) is a very local exactly complete
tree creating pair. The forcing notion Q
tree
D
(K
1
,
1
) is complete fuzzy proper
for W and it has the strong
Sacks property. Also, letting W
be the canonical
name for the generic function in
<
(i.e., p
Q
tree
D
(K1,1)
root(p) W
), we
have
Q
tree
D
(K1,1)
_
a o
+ V
__
< : W
() / a
) T
_
.
Let P be the limit of a support iteration,
++
in length, of the forcing notions
Q
tree
D
(K
1
,
1
). Then (by A.2.3 + A.1.9 + A.1.4 + A.3.10) we have:
P is complete, proper and satises the
++
cc, and it has a dense
subset of size
++
, thus forcing with P does not collapse cardinals,
P has the
Sacks property, it is weakly fuzzy proper for W,
P
2
= 2
+
=
++
= c
cl
(
+
) and c(
) =
+
Remark B.8.6. The result in B.8.5 is of interest as it shows that the versions
of cardinal characteristics of the reals may behave totally dierently from their
ancestors. Recall that if for an increasing function f
we let o
f
consist of
all sequences a = a
n
: n < ) with a
n
[]
f(n) + 1
(for n < ), then
min
_
[}[ : } o
f
& (h
)( a })(n < )(h(n) a
n
)
_
=
min
_
[}[ : } o
g
& (h
)( a })(n < )(h(n) a
n
)
_
for any increasing f, g
<
A
,= .
(2) The family / is a category prebase on if player II has a winning strategy
in the game
on is a category base on
and then a set is Dpresmall if and only if its complement is in D
.
46 ANDRZEJ ROS LANOWSKI AND SAHARON SHELAH
Denition B.8.8. (1) A
smallness base on is a sequence
/ = /
: <
) such that each /
is a
category prebase on
.
Let
/ be a
smallness base on .
(2) Let T
<
<
be a
lter on . We say that
T is
/small if for every (T)
) T
is /
presmall;
T is (D
,
/)small if
_
< : for every (T)
) T is /
presmall
_
D
.
(3) Let
(
/) consist of all subsets X of
<
such that X
<
lim
(T
)
for some
/small trees T
<
<
(for < ).
(D
,
/) is dened similarly, replacing
/small by (D
,
/)small.
Proposition B.8.9. Let
/ be a
smallness base on . Then both
(
/) and
(D
,
/) are proper
+
complete ideals of subsets of
<
,
(
/)
(D
,
/).
They contain singletons and < cov
_
(D
,
/)
_
cov
_
(
/)
_
.
Proposition B.8.10. Let
/ be a
smallness base on and D
be a normal lter
on . It is consistent that cov
_
(D
,
/)
_
>
+
.
Proof. First we dene a tree creating pair (K(
/), (
/)) = (K, ). For < let
H() =
and let st
(/
).
K consists of all tree creatures t TCR
and A
t
i
: i < ) is (an initial
segment of) a play of
(/
,
or dis[t] = ) for some <
;
if dis[t] = ), then pos[t] = [t]
) and nor[t] = 0;
if dis[t] = (, A
t
i
: i < )), then pos[t] = [t]
) :
i<
A
i
and
nor[t] = + 1. (If = 0 then we stipulate pos[t] = [t]
: <
.)
The domain of the tree composition operation consists of singletons only, and
if nor[t] = 0 then (t) = t,
if nor[t] > 0, = lh([t]) and dis[t] = (, A
t
i
: i < )), then (t) consists of those
t
K TCR
[t]
[H] for which:
either nor[t
] = 0 and pos[t
] pos[t],
or nor[t
] > 0, dis[t
] = (
, A
t
i
: i <
)) and A
t
i
: i < ) A
t
i
: i <
).
Claim B.8.10.1. (K, ) is an exactly
complete very local tree creating pair for
H. Hence the forcing notion Q
tree
D
(K, ) is fuzzy proper for W.
Proof of the Claim. Should be clear.
We nish the proof of the proposition in a standard way: we force with support
iteration,
++
in length, of the forcing notion Q
tree
D
(K(
/), (
/)).
SHEVASHEVASHEVA: LARGE CREATURES 47
References
[1] Uri Abraham. Lectures on proper forcing. In M. Foreman A. Kanamori and M. Magidor,
editors, Handbook of Set Theory.
[2] Bohuslav Balcar and Petr Simon. Disjoint renement. In Handbook of Boolean algebras, vol-
ume 2, pages 333388. North-Holland, Amsterdam, 1989.
[3] Elizabeth T. Brown. Tree Forcing at Uncountable Cardinals. Talk at Boise Extravaganza In
Set Theory, March 29-31, 2002.
[4] Elizabeth T. Brown. Tree Forcing at Uncountable Cardinals. Preprint.
[5] James Cummings and Saharon Shelah. Cardinal invariants above the continuum. Annals of
Pure and Applied Logic, 75:251268, 1995. math.LO/9509228.
[6] Todd Eisworth. On iterated forcing for successors of regular cardinals. Preprint,
math.LO/0210162.
[7] Thomas Jech. Set theory. Academic Press, New York, 1978.
[8] Akihiro Kanamori. Perfect-set forcing for uncountable cardinals. Annals of Mathematical
Logic, 19:97114, 1980.
[9] Avner Landver. Singular Baire numbers and related topics. PhD thesis, University of Wis-
consin, Madison, 1990.
[10] Pierre Matet and Janusz Pawlikowski. Qpointness, Ppointness and feebleness of ideals.
Journal of Symbolic Logic, 68:235261, 2003.
[11] Pierre Matet, Andrzej Roslanowski, and Saharon Shelah. Conality of the nonstationary
ideal. Transactions of the American Mathematical Society, submitted.
[12] Ludomir Newelski and Andrzej Ros lanowski. The ideal determined by the unsymmetric game.
Proceedings of the American Mathematical Society, 117:823831, 1993.
[13] Andrzej Roslanowski and Saharon Shelah. Sweet & Sour and other avours of ccc forcing
notions. Archive for Mathematical Logic, accepted. math.LO/9909115.
[14] Andrzej Roslanowski and Saharon Shelah. Norms on possibilities I: forcing with
trees and creatures. Memoirs of the American Mathematical Society, 141(671), 1999.
math.LO/9807172.
[15] Andrzej Roslanowski and Saharon Shelah. Iteration of -complete forcing notions not collaps-
ing
+
. International Journal of Mathematics and Mathematical Sciences, 28:6382, 2001.
math.LO/9906024.
[16] Saharon Shelah. Not collapsing cardinals in (< )support iterations. Israel Journal of
Mathematics, accepted. math.LO/9707225.
[17] Saharon Shelah. Successor of singulars: combinatorics and not collapsing cardinals in
(< )-support iterations. Israel Journal of Mathematics, accepted. math.LO/9808140.
[18] Saharon Shelah. Proper forcing, volume 940 of Lecture Notes in Mathematics. Springer-
Verlag, Berlin-New York, xxix+496 pp, 1982.
[19] Saharon Shelah. Proper and improper forcing. Perspectives in Mathematical Logic. Springer,
1998.
[20] Saharon Shelah and Zoran Spasojevic. Cardinal invariants b and t. Publications de
LInstitute Mathematique - Beograd, Nouvelle Serie, 72, 2002. math.LO/0003141.
[21] Jindrich Zapletal. Splitting number at uncountable cardinals. Journal of Symbolic Logic,
62:3542, 1997.
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/aroslano
Institute of Mathematics, The Hebrew University of Jerusalem, 91904 Jerusalem,
Israel, and Department of Mathematics, Rutgers University, New Brunswick, NJ 08854,
USA
E-mail address: shelah@math.huji.ac.il
URL: http://www.math.rutgers.edu/shelah