9

4
1


r
e
v
i
s
i
o
n
:
2
0
0
9
-
0
5
-
0
7







m
o
d
i
f
i
e
d
:
2
0
0
9
-
0
5
-
0
7


NONPROPER PRODUCTS
ANDRZEJ ROSLANOWSKI, SAHARON SHELAH, AND OTMAR SPINAS
Abstract. We show that there exist two proper creature forcings having a
simple (Borel) denition, whose product is not proper. We also give a new
condition ensuring properness of some forcings with norms.
1. Introduction
In Roslanowski and Shelah [2] a theory of forcings built with the use of norms
was developed and a number of conditions to ensure the properness of the resulting
forcings was given. However it is not clear how sharp those results really are and
this problem was posed in Shelah [4, Question 4.1]. In particular, he asked about
the properness of the forcing notion
Q = w
n
: n < ) : w
n
2
n
, w
n
,= and lim
n
[w
n
[ =
ordered by w w

(n )(w

n
w
n
). In the second section we give a general
criterion for collapsing the continuum to
0
and then in Corollary 2.8 we apply it
to the forcing Q, just showing that it is not proper.
That the property of properness is not productive, i.e. is not preserved under
taking products, has been observed by Shelah long ago (see [3, XVII, 2.12]). How-
ever, his examples are somewhat articial and certainly it would be desirable to
know of some rich enough subclass of proper forcings that is productively closed.
It was a natural conjecture put forth by Zapletal, that the class of denable, say
analytic or Borel, proper forcings would have this property. Actually, it was only
proved recently by Spinas [5] that nite powers of the Miller rational perfect set
forcing and nite powers of the Laver forcing notion are proper. These are two of
the most frequently used forcings in the set theory of the reals. However, in this
paper we shall show that this phenomenon does not extend to all forcing notions
dened in the setting of norms on possibilities. In the fourth section of the paper
we give an example of a forcing notion with norms which, by the theory developed
in the second section, is not proper and yet it can be decomposed as a product of
two proper forcing notions of a very similar type, and both of which have a Borel
denition. The properness of the factors is a consequence of a quite general theo-
rem presented in the third section (Theorem 3.3). It occurs that a strong version
of halving from [2, Section 2.2] implies the properness of forcing notions of the type
Q

(K, ).
The rst two authors acknowledge support from the United States-Israel Binational Science
Foundation (Grant no. 2002323). This is publication 941 of the second author.
The research of the third author was partially supported by DFG grant SP 683/1-2 and the
Landau foundation.
1
9
4
1


r
e
v
i
s
i
o
n
:
2
0
0
9
-
0
5
-
0
7







m
o
d
i
f
i
e
d
:
2
0
0
9
-
0
5
-
0
7


2 ANDRZEJ ROSLANOWSKI, SAHARON SHELAH, AND OTMAR SPINAS
Notation Most of our notation is standard and compatible with that of classi-
cal textbooks on Set Theory (like Bartoszy nski and Judah [1]). However in forcing
we keep the convention that a stronger condition is the larger one.
In this paper H will stand for a function with domain and such that (m
)(2 [H(m)[ < ). We also assume that 0 H(m) (for all m ); if it is not the
case then we x an element of H(m) and we use it whenever appropriate notions
refer to 0.
Creature background: Since our results are stated for creating pairs with
several special properties, below we present a somewhat restricted context of the
creature forcing, introducing good creating pairs.
Denition 1.1. (1) A creature for H is a triple
t = (nor, val, dis) = (nor[t], val[t], dis[t])
such that nor R
0
, dis H(
1
), and for some integers m
t
dn
< m
t
up
<
,= val u, v)

i<m
t
dn
H(i)

i<m
t
up
H(i) : u v.
The family of all creatures for H is denoted by CR[H].
(2) Let K CR[H] and : K T(K). We say that (K, ) is a good creating
pair for H whenever the following conditions are satised for each t K.
(a) [Fullness] dom(val[t]) =

i<m
t
dn
H(i).
(b) t (t) and if s (t), then m
s
dn
= m
t
dn
and m
s
up
= m
t
up
.
(c) [Transitivity] If s (t), then (s) (t).
(3) A good creating pair (K, ) is
local if m
t
up
= m
t
dn
+ 1 for all t K,
forgetful if for every t K, v

i<m
t
up
H(i), and u

i<m
t
dn
H(i) we
have
vm
t
dn
, v) val[t] u, u

v[m
t
dn
, m
t
up
)) val[t],
strongly nitary if for each i < we have
[H(i)[ < and [t K : m
t
dn
= i[ < .
(4) If t
0
, . . . , t
n
K are such that m
ti
up
= m
ti+1
dn
(for i < n) and w

i<m
t
0
dn
H(i),
then we let
pos(w, t
0
, . . . , t
n
)
def
= v

j<m
tn
up
H(j): w v & (i n)(vm
ti
dn
, vm
ti
up
) val[t
i
]).
If K is forgetful and t K, then we also dene
pos(t) =
_
v[m
t
dn
, m
t
up
) : vm
t
dn
, v) val[t]
_
.
Note that if K is forgetful, then to describe a creature in K it is enough to give
pos(t), nor[t] and dis[t]. This is how our examples will be presented (as they all
will be forgetful). Also, if K is additionally local, then we may write pos(t) = A
for some A H(m
t
dn
) with a natural interpretation of this abuse of notation.
9
4
1


r
e
v
i
s
i
o
n
:
2
0
0
9
-
0
5
-
0
7







m
o
d
i
f
i
e
d
:
2
0
0
9
-
0
5
-
0
7


NONPROPER PRODUCTS 3
Denition 1.2. Let (K, ) be a good creating pair for H. We dene a forcing
notion Q

(K, ) as follows.
A condition in Q

(K, ) is a sequence p = (w
p
, t
p
0
, t
p
1
, t
p
2
, . . .) such that
(a) t
p
i
K and m
t
p
i
up
= m
t
p
i+1
dn
(for i < ), and
(b) w

i<m
t
p
0
dn
H(i) and lim
n
nor[t
p
n
] = .
The relation on Q

(K, ) is given by: p q if and only if

for some i < we have w
q
pos(w
p
, t
p
0
, . . . , t
p
i1
) (if i = 0 this means w
q
= w
p
)
and t
q
n
(t
p
n+i
) for all n < .
For a condition p Q

(K, ) we let i(p) = lh(w

p
).
2. Collapsing creatures
We will show here that very natural forcing notions of type Q

(K, ) (for a big

local and nitary creating pair (K, )) collapse c to
0
, in particular answering [4,
Question 4.1]. The main ingredient of the proof is similar to the negative theory
presented in [2, Section 1.4], and Denition 2.1 below should be compared with [2,
Denition 1.4.4] (but the two properties are somewhat incomparable).
Denition 2.1. Let h : R
0
R
0
be a non-decreasing unbounded function and
let (K, ) be a good creating pair for H. We say that (K, ) is suciently h-bad if
there are sequences m = m
i
: i < ),

A = A
i
: i < ) and

F = F
i
: i < ) such
that
() m is a strictly increasing sequence of integers, m
0
= 0, and
(t K)(i < )(m
t
dn
= m
i
& m
t
up
= m
i+1
),
() A
i
are nite non-empty sets,
() F
i
= (F
0
i
, F
1
i
) : A
i

m<mi+1
H(m) A
i+1
2,
() if i < , t K, m
t
dn
= m
i
and nor[t] > 4, then there is a A
i
such that
for every x A
i+1
2, for some s
x
(t) we have
nor[s
x
] minh(nor[t]), h(i) and
(u

m<mi
H(m))(v pos(u, s
x
))(F
i
(a, v) = x).
Proposition 2.2. Suppose that h : R
0
R
0
is a non-decreasing unbounded
function, and (K, ) is a strongly nitary good creating pair for H. Assume also
that (K, ) is suciently h-bad. Then the forcing notion Q

(K, ) collapses c onto

0
.
Proof. The proof is similar to that of [2, Proposition 1.4.5], but for readers conve-
nience we present it fully.
Let m,

A and

F witness that (K, ) is suciently h-bad. For i < and a A
i
we dene Q

(K, )names
i,a
(for a real in 2

) and
i,a
(for an element of

ji
A
j
)
as follows:

(K,)

i,a
(i) = a and
i,a
(j) = F
0
j1
(
i,a
(j 1),

Wm
j
) for j > i ,
and

(K,)

i,a
i 0 and
i,a
(j) = F
1
j
(
i,a
(j),

Wm
j+1
) for j i .
9
4
1


r
e
v
i
s
i
o
n
:
2
0
0
9
-
0
5
-
0
7







m
o
d
i
f
i
e
d
:
2
0
0
9
-
0
5
-
0
7


4 ANDRZEJ ROSLANOWSKI, SAHARON SHELAH, AND OTMAR SPINAS
Above,

W is the canonical name for the generic function in

i<
H(i), i.e., p
Q

(K,)
w
p

i<
H(i) . We are going to show that

(K,)
(r 2

V)(i < )(a A

i
)(j i)(
i,a
(j) = r(j)) .
To this end suppose that p Q

(K, ) and r 2

. Passing to a stronger condition

if needed, we may assume that (j < )(nor[t
p
j
] > 4). Let i < be such that
lh(w
p
) = m
i
; then also m
t
p
j
dn
= m
i+j
for j < (remember 2.1()).
Fix k < for a moment. By downward induction on j k choose s
k
j
(t
p
j
)
and a
k
j
A
i+j
such that
(a) nor[s
k
j
] minh(nor[t
p
j
]), h(i + j) for all j k,
(b) (u

m<m
i+k
H(m))(v pos(u, s
k
k
))(F
1
i+k
(a
k
k
, v) = r(i + k)),
(c) for j < k:
(u

m<mi+j
H(m))(v pos(u, s
k
j
))(F
1
i+j
(a
k
j
, v) = r(i + j) & F
0
i+j
(a
k
j
, v) = a
k
j+1
).
(Plainly it is possible by 2.1().)
Since for each j < both (t
p
j
) and A
i+j
are nite, we may use Konig Lemma
to pick an increasing sequence

k = k() : < ) such that
a
k(+1)
j
= a
k(

)
j
and s
k(+1)
j
= s
k(

)
j
for <

< and j k(). Put w

q
= w
p
and t
q
j
= s
k(j+1)
j
, b
j
= a
k(j+1)
j
for j < .
Easily, q = (w
q
, t
q
0
, t
q
1
, t
q
3
, . . .) is a condition in Q

(K, ) stronger than p. Also, by

clause (c) of the choice of s
k
j
, we clearly have
(j < )(v pos(w
q
, t
q
0
, . . . , t
q
j
))(F
0
i+j
(b
j
, v) = b
j+1
& F
1
i+j
(b
j
, v) = r(i + j)).
Hence
q
Q

(K,)
(j < )(
i,b0
(i + j) = b
j
&
i,b0
(i + j) = r(i + j)) ,
nishing the proof.
Lemma 2.3. Suppose that positive integers N, M, d satisfy N 2
M
< d. Let A, B
be nite sets such that [A[ 2
M
and [B[ N. Then there is a mapping F :
A
d
M B with the property that:
() if 2 M, c
i
: i < d)

i<d
[M]

, then there is a A such that for

every b B, for some c
b
i
[c
i
]
/2
(for i < d) we have
(u

i<d
c
b
i
)(F(a, u) = b).
Proof. Plainly we may assume that [A[ = 2
M
and [B[ = N, and then we may
pretend that A =
M
2 and B = N.
For h A =
M
2 and u
d
M we let F(h, u) < N be such that
F(h, u)

i<d
h(u(i)) mod N.
9
4
1


r
e
v
i
s
i
o
n
:
2
0
0
9
-
0
5
-
0
7







m
o
d
i
f
i
e
d
:
2
0
0
9
-
0
5
-
0
7


NONPROPER PRODUCTS 5
This denes the function F : A
d
M B = N, and we are going to show
that it has the property stated in (). To this end suppose that 2 M and
c
i
: i < d)

i<d
[M]

. For each i < d we may choose h

i
A so that
[(h
i
)
1
[0] c
i
[ /2 and [(h
i
)
1
[1] c
i
[ /2
Then, for some h A and I d we have [I[ d/2
M
and h
i
= h for i I. For
i d I we may pick c

i
[c
i
]
/2
and j
i
< 2 such that hc

i
j
i
.
Now, suppose b B. Take a set J I such that
[J[ +

id\I
j
i
b mod N
(possible as [I[ d/2
M
> N). By our choices, we may pick c
b
i
[c
i
]
/2
(for
i I) such that
if i J, then hc
b
i
1,
if i I J, then hc
b
i
0.
For i d I we let c
b
i
= c

i
(selected earlier). It should be clear that then
(u

i<d
c
b
i
)(F(h, u) = b),
as needed.
Example 2.4. Let m = m
i
: i < ) be an increasing sequence of integers such
m
0
= 0 and m
i+1
m
i
> 4
i+3
. Let h() = /2 for < .
For j < we let H
0
m
(j) = i + 2, where i is such that m
i
j < m
i+1
. Let K
0
m
consist of all (forgetful) creatures t CR[H
0
m
] such that
dis[t] = i
t
, Z
t
j
: m
i
t j < m
i
t
+1
)) for some i
t
< and , = Z
t
j
H
0
m
(j)
(for m
i
t j < m
i
t
+1
),
nor[t] = min[Z
t
j
[ : m
i
t j < m
i
t
+1
,
pos(t) =

j[m
i
t ,m
i
t
+1
)
Z
t
j
.
Finally, for t K
0
m
we let

0
m
(t) = s K
0
m
: i
t
= i
s
& (j [m
i
t , m
i
t
+1
))(Z
s
j
Z
t
j
).
Then (K
0
m
,
0
m
) is a strongly nitary and suciently h-bad good creating pair for
H
0
m
. Consequently, the forcing notion Q

(K
0
m
,
0
m
) collapses c onto
0
.
Proof. It should be clear that (K
0
m
,
0
m
) is a strongly nitary good creating pair for
H
0
m
. To show that it is suciently h-bad let A
i
=
i + 2
2, B
i
= A
i+1
2 =
i + 3
22
and M
i
= i+2. Since [B
i
[2
Mi
= 2
i+4+i+2
< m
i+1
m
i
def
= d
i
, we may apply Lemma
2.3 for A = A
i
, B = B
i
, M = M
i
and d = d
i
to get functions F
i
: A
i

d
i
M
i
B
i
with the property () (for those parameters). For a A
i
and v

j<mi+1
H
0
m
(j)
we interpret F
i
(a, v) as F
i
(a, u) where u
d
i
(i +1) is given by u(j) = v(m
i
+j) for
j < d
i
. It is straightforward to show that m,

A = A
i
: i < ) and

F = F
i
: i < )
witness that (K
0
m
,
0
m
) is h-bad.
9
4
1


r
e
v
i
s
i
o
n
:
2
0
0
9
-
0
5
-
0
7







m
o
d
i
f
i
e
d
:
2
0
0
9
-
0
5
-
0
7


6 ANDRZEJ ROSLANOWSKI, SAHARON SHELAH, AND OTMAR SPINAS
The above example (together with Proposition 2.2) easily gives the answer to [4,
Question 4.1]. To show how our problem reduces to this example, let us recall the
following.
Denition 2.5 (See [2, Denition 4.2.1]). Suppose 0 < m < and for i < m we
have t
i
CR[H] such that m
ti
up
m
ti+1
dn
. Then we dene the sum of the creatures
t
i
as a creature t =
sum
(t
i
: i < m) such that (if well dened then):
(a) m
t
dn
= m
t0
dn
, m
t
up
= m
tm1
up
,
(b) val[t] is the set of all pairs h
1
, h
2
) such that:
lh(h
1
) = m
t
dn
, lh(h
2
) = m
t
up
, h
1
h
2
,
and h
2
m
ti
dn
, h
2
m
ti
up
) val[t
i
] for i < m,
and h
2
[m
ti
up
, m
ti+1
dn
) is identically zero for i < m1,
(c) nor[t] = minnor[t
i
] : i < m,
(d) dis[t] = t
i
: i < m).
If for all i < m1 we have m
ti
up
= m
ti+1
dn
, then we call the sum tight.
Denition 2.6. Let (K, ) be a local good creating pair for H, and let m =
m
i
: i < ) be a strictly increasing sequence with m
0
= 0. We dene an m
summarization (K
m
,
m
, H
m
) of (K, , H) as follows:
H
m
(i) =
mi+11

m=mi
H(m),
K
m
consists of all tight sums
sum
(t

: m
i
< m
i+1
) such that i < ,
t

K, m
t

dn
= ,
if t =
sum
(t

: m
i
< m
i+1
) K
m
, then
m
(t) consists of all creatures
s K
m
such that s =
sum
(s

: m
i
< m
i+1
) for some s

(t

) (for
= m
i
, . . . , m
i+1
1).
Proposition 2.7. Assume that (K, ) is a local good creating pair for H, m =
m
i
: i < ) is a strictly increasing sequence with m
0
= 0. Then:
(1) (K
m
,
m
) is a good creating pair for H
m
;
(2) the forcing notion Q

(K
m
,
m
) can be embedded as a dense subset of the
forcing notion Q

(K, ) (so the two forcing notions are equivalent).

Corollary 2.8. Let H : be increasing, H(0) 2, and let g : R R be an
unbounded non-decreasing function. We dene (K
H
g
,
H
g
) as follows: K
H
g
consists
of all creatures t CR[H] such that
dis[t] = i
t
, A
t
) for some i
t
< and ,= A
t
H(i
t
),
nor[t] = g([A
t
[), m
t
dn
= i
t
, m
t
up
= i
t
+ 1 and pos(t) = A
t
.
For t K
H
g
we let

H
g
(t) = s K
H
g
: i
t
= i
s
& A
s
A
t
.
Then (K
H
g
,
H
g
) is a local strongly nitary good creating pair for H. The forcing
notion Q

(K
H
g
,
H
g
) collapses c onto
0
. In particular, the forcing notion Q dened
in the Introduction is not proper.
Proof. Let p Q

(K
H
g
,
H
g
). Plainly, lim
i
[A
t
p
i
[ = , so we may nd a condition
q p and an increasing sequence m = m
i
: i < ) such that
m
0
= 0, m
1
= lh(w
q
), m
i+1
m
i
> 4
i+3
,
9
4
1


r
e
v
i
s
i
o
n
:
2
0
0
9
-
0
5
-
0
7







m
o
d
i
f
i
e
d
:
2
0
0
9
-
0
5
-
0
7


NONPROPER PRODUCTS 7
if m
i
m
t
q
k
dn
< m
i+1
, then [A
t
q
k
[ = i + 2.
Now we dene a condition q

in Q

((K
H
g
)
m
, (
H
g
)
m
) by
w
q

= w
q
, t
q

i
=
sum
(t
q
k
: m
i+1
k + m
0
< m
i+2
) (for i < ).
The forcing notion Q

(K
H
g
,
H
g
) above the condition q is equivalent to the forcing
notion Q

((K
H
g
)
m
, (
H
g
)
m
) above q

. Plainly, Q

((K
H
g
)
m
, (
H
g
)
m
) above q

is
isomorphic to Q

(K
0
m
,
0
m
) of Example 2.4 above the minimal condition r with
w
r
= w
q

. The assertion follows now by the last sentence of 2.4.

Remark 2.9. (1) If, e.g., g(x) = log
2
(x) then the creating pair (K
H
g
,
H
g
) is big
(see [2, Denition 2.2.1]), and we may even get a lot of bigness. Thus the
bigness itself is not enough to guarantee properness of the resulting forcing
notion.
(2) Forcing notions of the form Q

(K, ) are special cases of Q

f
(K, ) (see
[2, Denition 1.1.10 and Section 2.2]). However if the function f is growing
very fast (much faster than H) then our method do no apply. Let us recall
that if (K, ) is simple, nitary, big and has the Halving Property, and
f : is Hfast (see [2, Denition 1.1.12]), then Q

f
(K, ) is
proper. Thus one may wonder if we may omit halving - can the forcing
notion Q

f
(K
H
g
,
H
g
) be proper for H and f suitably fast?
3. Properness from Halving
It was shown in [2, Theorem 2.2.11] that halving and bigness (see [2, Denitions
2.2.1, 2.2.7]) imply properness of the forcings Q

f
(K, ) (for fast f). It occurs that if
we have a stronger version of halving, then we may get the properness of Q

(K, )
even without any bigness assumptions.
Denition 3.1. Let (K, ) be a forgetful good creating pair
(1) Let t K and > 0. We say that a creature t

(t) is an half of t if
the following hold:
(i) nor[t

] nor[t] , and
(ii) if s (t

) and nor[s] > 1, then we can nd t

0
(t) such that
nor[t
0
] nor[t] and pos(t
0
) pos(s).
(2) Let =
i
: i < ) be a sequence of positive real numbers and m = m
i
:
i < ) be a strictly increasing sequence of integers with m
0
= 0. We say
that the pair (K, ) has the ( , m)halving property if for every t K with
m
i
m
t
dn
and nor[t] 2 there exists an
i
half of t in (t).
Denition 3.2. Let (K, ) be a good creating pair. Suppose that p Q

(K, )
and I Q

(K, ) is open dense. We say that p essentially belongs to I, written

p

I, if there exists i

< such that for every v pos(w

p
, t
p
0
, . . . , t
p
i1
) we have
(v, t
p
i
, t
p
i+1
, t
p
i+2
, . . .) I
9
4
1


r
e
v
i
s
i
o
n
:
2
0
0
9
-
0
5
-
0
7







m
o
d
i
f
i
e
d
:
2
0
0
9
-
0
5
-
0
7


8 ANDRZEJ ROSLANOWSKI, SAHARON SHELAH, AND OTMAR SPINAS
Theorem 3.3. Let =
i
: i < ) be a decreasing sequence of positive numbers
and m = m
i
: i < ) be a strictly increasing sequence of integers with m
0
= 0.
Assume that for each i <
[

n<mi
H(n)[ 1/
i
.
Let (K, ) be a good creating pair for H and suppose that (K, ) is local, forgetful
and has the ( , m)halving property. Then the forcing notion Q

(K, ) is proper.
Proof. The proof will be carried out in a series of claims.
Claim 3.3.1. Let a 2 and I Q

(K, ) be open dense. Furthermore suppose

that p Q

(K, ) and i < is such that nor[t

p
n
] > a for every n m
i
i(p).
Finally let v

n<mi
H(n). Then there exists q Q

(K, ) such that

(a) p q, w
p
= w
q
and t
p
n
= t
q
n
for every n < m
i
i(p);
(b) nor[t
q
n
] a
i
for every n m
i
i(p);
(c) either, letting q
[v]
= (v, t
q
mii(p)
, t
q
mii(p)+1
, t
q
mii(p)+2
, . . .), q
[v]

I or else
there is no r q
[v]
such that r I, w
r
= v and nor[t
r
n
] > 1 for every n.
Proof of the Claim. We know that (K, ) has the ( , m)halving property and
therefore for each n m
i
i(p) we may choose an
i
half t
q0
n
(t
p
n
) of t
p
n
.
For n < m
i
i(p) put t
q0
n
= t
p
n
and let w
q0
= w
p
. This denes a condition
q
0
= (w
q0
, t
q0
0
, t
q0
1
, t
q0
2
, . . .) Q

(K, ). Plainly, (a) and (b) hold for q

0
instead of
q. Now if there is no r q
[v]
0
with r I, w
r
= v and nor[t
r
n
] > 1 for every n < ,
we can let q = q
0
. Hence we may assume that such r = (w
r
, t
r
0
, t
r
1
, t
r
2
, . . .) does exist.
Pick j < large enough such that nor[t
r
n
] a
i
for every n j. Now we
dene q Q

(K, ):
w
q
= w
p
, t
q
n
= t
p
n
for n < m
i
i(p),
t
q
n
= t
r
nmi+i(p)
for n m
i
i(p) + j,
for m
i
i(p) n < m
i
i(p) + j let t
q
n
(t
p
n
) be such that
nor[t
q
n
] nor[t
p
n
]
i
a
i
and pos(t
q
n
) pos(t
r
nmi+i(p)
)
(exists by the halving property).
Clearly p q and (a), (b) hold. Also, for every u pos(v, t
q
mii(p)
, . . . , t
q
mii(p)+j
)
we have q
[u]
r, and hence q
[u]
I, as I is open. Consequently, q
[v]

I.
Claim 3.3.2. Let a 3 and I Q

(K, ) be open dense. Suppose that p

Q

(K, ) and i < is such that nor[t

p
n
] > a for every n m
i
i(p). Then there
exists q Q

(K, ) such that

(a) p q, w
p
= w
q
, and t
p
n
= t
q
n
for n < m
i
i(p);
(b) nor[t
q
n
] a 1 for every n m
i
i(p);
(c) for every v

n<mi
H(n), either q
[v]

I, or else there is no r I such

that r q, w
r
= v and nor[t
r
n
] > 1 for all n.
Proof of the Claim. Let v
l
: l < k) enumerate

n<mi
H(n), thus k 1/
i
. Applying
Claim 3.3.1 k times, it is straightforward to construct a sequence q
l
: l k)
Q

(K, ) such that

q
0
= p, q
l
q
l+1
, w
q
l
= w
p
, and t
q
l
n
= t
p
n
for every n < m
i
i(p),
9
4
1


r
e
v
i
s
i
o
n
:
2
0
0
9
-
0
5
-
0
7







m
o
d
i
f
i
e
d
:
2
0
0
9
-
0
5
-
0
7


NONPROPER PRODUCTS 9
nor[t
q
l
n
] a l
i
for every n m
i
i(p), and
q
l
, q
l+1
, v
l
, a l
i
) are like p, q, v, a) in Claim 3.3.1.
Then clearly q = q
k
is as desired.
Claim 3.3.3. Let a 2 and I
j
Q

(K, ) be open dense for j < . Suppose

that p Q

(K, ) and i

: < ) is an increasing sequence in such that

nor[t
p
n
] > a + 2 + 1 for every n m
i

i(p), for every . Then there exists

q Q

(K, ) such that

(a) p q, w
p
= w
q
, and t
p
n
= t
q
n
for every n < m
i0
i(p);
(b) nor[t
q
n
] a + for every n m
i

i(p), for every ;

(c) for every j < there are innitely many < such that for every sequence
v

n<mi

H(n), if there exists r I

j
such that r q, w
r
= v and nor[t
r
n
] >
1 for every n, then q
[v]

I
j
.
Proof of the Claim. Fix a surjection : such that
1
[j] is innite
for every j. We apply Corollary 3.3.2 innitely many times in order to construct
p

: < ) with p
0
= p such that for every , p

, p
+1
, i

, a + + 1, I
()
) is like
p, q, i, a, I) there. Then we let q be the natural fusion determined by the p

.
Claim 3.3.4. Suppose that I
j
, j < , p, q and i

: < ) are like in Claim 3.3.3.

Let
I
[q]
j
= q
[v]
: v

n<mi
H(n) & i < I
j
.
Then I
[q]
j
is predense above q.
Proof of the Claim. Suppose r

(K, ) is stronger than q. Pick a condition

r
0
= (w
r0
, t
r0
0
, t
r0
1
, t
r0
2
, . . .) r

such that r
0
I
j
. By 3.3.3(c) we may nd such
that
() nor[t
r0
n
] > 1 for every n m
i

i(r
0
), and
() for every sequence v

n<mi

H(n), if there exists r I

j
such that r q,
w
r
= v and nor[t
r
n
] > 1 for every n, then q
[v]

I
j
.
Choose v pos(w
r0
, t
r0
0
, . . . , t
r0
mi

i(r0)1
). Then r
[v]
0
r
0
so r
[v]
0
I
j
(remember I
j
is open) and r
[v]
0
q. Therefore, by (), we see that q
[v]

I
j
. Clearly r
[v]
0
q
[v]
.
Hence we may nd u pos(v, t
r0
mi

i(r0)
, . . . , t
r0
m
k
i(r0)1
) (for some k > i

) such
that q
[u]
I
j
. Then q
[u]
I
[q]
j
and r
[u]
0
is a common upper bound to q
[u]
and r

.
As the sets I
[q]
j
from Claim 3.3.4 are countable, it is now clear how to conclude
the properness of Q

(K, ).
4. A nonproper product
Here, we will give an example of two proper forcing notions Q

(K
1
,
1
) and
Q

(K
2
,
2
) such that their product Q

(K, ) collapses c onto

0
.
Throughout this section we write log instead of log
2
.
Denition 4.1. Let x, i R, x > 0, i 0, and k 0. We let
f
k
(x, i) =
log
_
log(log(x)) i
_
k
9
4
1


r
e
v
i
s
i
o
n
:
2
0
0
9
-
0
5
-
0
7







m
o
d
i
f
i
e
d
:
2
0
0
9
-
0
5
-
0
7


10 ANDRZEJ ROSLANOWSKI, SAHARON SHELAH, AND OTMAR SPINAS
in the case that all three logarithms are well dened and attain a value 1. In all
other cases we dene f
k
(x, i) = 1.
Lemma 4.2. (1) f
k
(
x
2
, i) f
k
(x, i)
1
k
;
(2) letting j =
log(log(x))+i
2
, if f
k
(x, i) 2 then f
k
(x, j) = f
k
(x, i)
1
k
;
(3) letting j as in (2), if minf
k
(x, i), f
k
(y, j) > 1 then f
k
(y, i) f
k
(x, i)
1
k
;
(4) if x 2
2
4+i
and z such that log(log(z)) =
log(log(x))+i
2
, then f
k
(z, i) =
f
k
(x, i)
1
k
.
Proof. (1) Note that for x 2 we have
() log(x 1) log(x) 1.
Indeed, x 2 implies x 1
x
2
. Applying log to both sides we get log(x 1)
log(
x
2
) = log(x) 1.
If x < 2
2
2+i
, then log(log(log(x)) i) < 1 (if at all dened), and f
k
(x, i) = 1 =
f
k
(
x
2
, i). So assume x 2
2
2+i
. Then log(x) 2
2+i
2 and log(log(x)) i 2,
and hence we may apply () with log(x) and log(log(x)) i and obtain
log
_
log(log(
x
2
)) i
_
= log
_
log(log(x) 1) i
_
log
_
log(log(x)) 1 i
_
log
_
log(log(x)) i
_
1.
By dividing both sides by k we arrive at (1).
(2) Note that f
k
(x, i) 2 implies log(log(x)) i 4 and hence log(log(x)) j =
log(log(x))i
2
2. Consequently
f
k
(x, j) =
log
_
log(log(x))j
_
k
=
log
_
(log(log(x))i)/2
_
k
=
log
_
log(log(x))i
_
1
k
= f
k
(x, i)
1
k
.
(3) By assumption we have log(log(y)) j 0. By pluging in the denition of j
and adding j i to both sides we obtain log(log(y)) i
1
2
(log(log(x)) i) and
hence log
_
log(log(y)) i
_
log
_
log(log(x)) i
_
1. After dividing by k we reach
at (3).
(4) Note that
f
k
(z, i) =
1
k
log
_
log(log(z)) i
_
=
1
k
log
_
(log(log(x)) i)/2
_
=
1
k
[log
_
log(log(x)) i
_
1] = f
k
(x, i)
1
k
.

We are going to modify the example in Corollary 2.8 and Example 2.4.
Let m = m
i
: i < ) be an increasing sequence of integers such that m
0
= 0 and
m
i+1
m
i
> 4
i+3
. For j < let H(j) = i +3, where i is such that m
i
j < m
i+1
,
and let g(x) = x. The local good creating pair (K
H
g
,
H
g
) introduced in 2.8 will be
denoted by (K
1
,
1
). By 2.4 we know that ((K
1
)
m
, (
1
)
m
) (see 2.6) is suciently
bad and hence (by 2.8) the forcing Q

(K
1
,
1
) collapses c into
0
.
Recall that for a creature t K
1
we have
dis[t] = i
t
, A
t
) for some i
t
< and ,= A
t
H(i
t
),
nor[t] = [A
t
[ and pos(t) = A
t
.
9
4
1


r
e
v
i
s
i
o
n
:
2
0
0
9
-
0
5
-
0
7







m
o
d
i
f
i
e
d
:
2
0
0
9
-
0
5
-
0
7


NONPROPER PRODUCTS 11
Let l
n
= [H(n)[ and
k
n
=
_
maxk 0 : f
k
(l
n
, 0) > 1 if l
n
> 16,
and k
n
= 1 if l
n
16. Certainly we have lim
n
l
n
= and therefore lim
n
k
n
=
as well. Note also that lim
n
f
kn
(l
n
, 0) = .
Denition 4.3. Let K consist of all creatures t CR[H] such that
dis[t] = m
t
, A
t
, i
t
) for some m
t
< and ,= A
t
H(m
t
), and i
t
,
0 i
t
< log(log(l
m
t )),
nor[t] = f
k
m
t
([A
t
[, i
t
), m
t
dn
= m
t
, m
t
up
= m
t
+ 1 and pos(t) = A
t
.
For t K we let
(t) = s K : m
s
= m
t
& A
s
A
t
& i
s
i
t
.
Lemma 4.4. (K, ) is a local forgetful good creating pair for H. The forcing notion
Q

(K, ) collapses c to
0
.
Proof. It is straightforward to check that (K
m
,
m
) inherits the sucient badness
of ((K
1
)
m
, (
1
)
m
) (remember 4.2(1)).
We are now going to dene the desired factoring Q

(K, ) P
0
P
1
into proper
factors P
0
, P
1
. For this we recursively dene an increasing sequence n = n
i
: i < )
so that n
0
= 0 and n
i+1
is large enough such that
k
ni+1

j<ni
H(j).
We put U
0
=

i<
[n
2i
, n
2i+1
) and U
1
=

i<
[n
2i+1
, n
2i
) and we let
0
: U
0
and
1
: U
1
be the increasing enumerations.
Denition 4.5. Let 0, 1. We dene H

= H

and we introduce K

as follows.
(1) K

consist of all creatures t CR[H

] such that
dis[t] = m
t
, A
t
, i
t
) for some m
t
< and , = A
t
H

(m
t
), and
i
t
, 0 i
t
< log(log(l
n
)), where n =

(m
t
),
m
t
dn
= m
t
, m
t
up
= m
t
+ 1, pos(t) = A
t
and nor[t] = f
kn
([A
t
[, i
t
)
(where again n =

(m
t
)).
(2) For t K

we let

(t) = s K

: m
s
= m
t
& A
s
A
t
& i
s
i
t
.
Lemma 4.6. (1) For 0, 1, (K

) is a local forgetful good creating pair

for H.
(2) Let m = m
i
: i < ) and =
i
: i < ) be such that
0
(m
i
) = n
2i
and

i
= 1/k
n2i
. Then (K
0
,
0
) has the ( , m)halving property.
(3) Let m = m
i
: i < ) and =
i
: i < ) be such that
1
(m
i
) = n
2i+1
and

i
= 1/k
n2i+1
. Then (K
1
,
1
) has the ( , m)halving property.
Proof. (1) Should be clear.
(2) Let t K
0
, nor[t] 2, dis[t] = m, A, i

). Let n =
0
(m) [n
2i
, n
2i+1
) (so
m
i
m = m
t
dn
). Dene j =
log(log(|A|))+i

2
and let z such that log(log(z)) = j.
Certainly i

< j < log(log([A[)) log(log(l

n
)). Let t

K
0
be such that dis[t

] =
9
4
1


r
e
v
i
s
i
o
n
:
2
0
0
9
-
0
5
-
0
7







m
o
d
i
f
i
e
d
:
2
0
0
9
-
0
5
-
0
7


12 ANDRZEJ ROSLANOWSKI, SAHARON SHELAH, AND OTMAR SPINAS
m, A, j). Clearly t

0
(t). We are going to argue that t

is an
i
half of t in

0
(t).
By Lemma 4.2(2), nor[t

] = nor[t]
1
kn
nor[t]
i
. Now let s
0
(t

)
be such that nor[s] > 1. Let dis[s] = m, A

, i

), thus A

A and i

j. Pick
t
0
K
0
such that dis[t
0
] = m, A

, i

). Then t
0

0
(t) and pos(t
0
) = A

= pos(s).
Also, nor[s] > 1 implies log(log([A

[)) > i

j. By the denition of z we conclude

[A

[ > z. Applying this together with Lemma 4.2(4) we obtain

nor[t
0
] = f
kn
([A

[, i

) f
kn
(z, i

) = f
kn
([A[, i

)
1
k
n
nor[t]
i
.
(3) Like (2) above.
Corollary 4.7. (1) The forcing notions Q

(K

) (for = 0, 1) are proper.

(2) Let Q = p Q

(K, ) : i(p) = n
i
, i < . Then Q is a dense suborder
of Q

(K, ) and it is isomorphic with a dense suborder of the product

Q

(K
0
,
0
) Q

(K
1
,
1
). Consequently, the latter forcing collapses c to

0
.
Proof. (1) Let m, be as in 4.6(2). By the choice of n we have
[

n<mi
H
0
(n)[ = [

H(j) : j
_
<i
[n
2
, n
2+1
)[

j<n2i1
H(i) k
n2i
= 1/
i
.
Consequently, Theorem 3.3 and Lemma 4.6(1,2) imply that Q

(K
0
,
0
) is proper.
Similarly for Q

(K
1
,
1
)
(2) Should be clear.
References
[1] Tomek Bartoszy nski and Haim Judah. Set Theory: On the Structure of the Real Line. A K
Peters, Wellesley, Massachusetts, 1995.
[2] Andrzej Roslanowski and Saharon Shelah. Norms on possibilities I: forcing with trees
and creatures. Memoirs of the American Mathematical Society, 141(671):xii + 167, 1999.
math.LO/9807172.
[3] Saharon Shelah. Proper and Improper Forcing. Springer, Berlin, 1998.
[4] Saharon Shelah. On what I do not understand (and have something to say:) Part I. Funda-
menta Mathematicae, 166:182, 2000. math.LO/9906113.
[5] Otmar Spinas. Proper Products. Proceedings of the American Mathematical Society, 137:2767
2772, 2009.
Department of Mathematics, University of Nebraska at Omaha, Omaha NE 68182,
USA
E-mail address: roslanow@member.ams.org
Einstein Institute of Mathematics, 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
Mathematisches Seminar, Christian-Albrechts-Universitt zu Kiel, Ludewig-Meyn-Strae
4, 24098 Kiel, Germany
E-mail address: spinas@math.uni-kiel.de